WORLD HEALTH QUAR I ERLY · RAPPORT TRIMES I RIEL DE SANITAIRES MONDIALES PREVENTABLE MORTALITY LA MORTALITE EVITABLE • Vol. 42, No. 1, 1989 World Health Organization Organisation mondiale de la Santi Geneve The World Health Organization is a specialized agency of the United Nations with pri- mary responsibility for international health matters and public health. Through this organi- zation, which was created in 1948, the health professionals of some 160 countries exchange their knowledge and experience with the aim of making possible the attainment by all citizens of the world by the year 2000 ofa level of health that will permit them to lead a socially and economically productive life. By means of direct technical cooperation with its Member States, and by stimulating such cooperation among them, WHO promotes the development of comprehensive health ser- vices, the prevention and control of diseases, the improvement of environmental conditions, the development of health manpower, the coordination and development of biomedical and health services research, and the planning and implementation of health programmes. These broad fields of endeavour encompass a wide variety ofactivities, such as developing systems of primary health care that reach the whole population of Member countries; pro- moting the health of mothers and children; combating malnutrition; controlling malaria and other communicable diseases including tuberculosis and leprosy; having achieved the era- dication of smallpox, promoting mass immunization against a number of other preventable diseases; improving mental health; providing safe water supplies; and training health per- sonnel of all categories. Progress towards better health throughout the world also demands international cooper- ation in such matters as establishing international standards for biological substances, pes- ticides and pharmaceuticals; formulating environmental health criteria; recommending international nonproprietary names for drugs; administering the International Health Regu- lations; revising the International Classification of Diseases, Injuries, and Causes of Death; and collecting and disseminating health statistical information. Funher information on many aspects ofWHO's work is presented in the Organization's publications. The World Health Statistics Quarterly replaces (since 1978) the monthly World Health Statistics Report (published since 1967) and its forerunner the Epidemiological and Vital Statistics Report (published since 1947). It contains Special Subjects dealing with the detailed analysis of selected health topics of current interest. Annual subscription ( 1989) Sw. fr. 85.- Material from the Quarterly may be reproduced providing due acknowledgement is made. L'Organisation mondiale de la Sante (OMS), creee en 1948, est une institution spccialisce des Nations Unies a qui incombe, sur le plan international, la responsabilitC principale en matierede questions sanitaires et de sante publique. Au sein de l'OMS, les professionnels de la sante de quelque 160 pays echangent des connaissances et des donnees d'experience en vue de faire accede, d'ici l'an 2000 tous les habitants du monde a un niveau de sante qui leur permette de mener une vie socialement et CConomiquement productive. Grace a la cooperation technique qu'elle pratique avec ses Etats Membres ou qu'elle stimule entre eux, !'OMS s'emploie a promouvoir la mise sur pied de services de sante complets, la prevention et l'endiguement des maladies, l'amClioration de l'environnement, le developpement des personnels de sante, la coordination et le progres de la recherche bio- medicale et de la recherche sur les services de santC, ainsi que la planification et l'exCCution des programmes de sante. Le vaste domaine all s'exerce l'action de l'OMS compone des activitCs tre!s di verses: dfveloppement des soins de santf primaires pour que toutes les populations puissent y avoir accCS; promotion de la santt! matemelle et infantile; La lutte contre la malnutrition: lune contre le paludisme et d'autres maladies transmissibles, dont la tuberculose et la li:pre; l'Cradication de la variate Ctant rCaliSCe, promotion de la vaccination de masse contrc un cenain nombre d'autres maladies Cvitables; amClioration de la santC mentale; approvision- nement en eau saine; formation de personnels de santC de toutes catCgories. II est d'autres secteurs encore all une cooperation intemationale s'impose pour assurer un meilleur Ctat de sante a travers le monde et l'OMS collabore notamment aux taches sui- vantes: Ctablissement d'Ctalons intcmationaux pour les produits biologiques, les pesticides et les prCparations phamtaceutiqucs; formulation de criteres de salubritC de l'environnement: recommandations relatives aux denominations communes intemationales pour les subs- tances phannaceutiques; application du RC!glement sanitaire international; revision de la Classification internationale des maladies, traumatismes et causes de dCcCs; rassemblement et diffusion d'informations statistiques sur la sante. On trouvera dans les publications de l'OMS de plus amples renseignements sur de nom- breux aspects des travaux de !'Organisation. Le Rapport trimestriel de Statistiques sanitaires mondiales remplace (depuis 1978) le Rapport de Statistiques sanitaires mondiales(publie depuis 1967) et son precurseur le Rapport t!pidt!miologique et dt!mographique (public depuis 194 7). II contient des Sujets speciaux qui presentent des analyses detaillees sur des sujets specifiques d'interet courant. Prix de l'abonnement annuel (1989) fr. s. 85,- La reproduction d'extraits du Rapport trimestriel est autorisee, sous reserve d'indication de la source. IX ISSN 0043 - 8510 PRINTED IN SWITZERLAND 89/7980 - Atar - 5000 Cover design: Gilbert Aubcrson * T M F N Symbols used in tables Preliminary, approximate or estimated data. Data not available. Nil or magnitude negligible. Category not applicable. Total. Male. Female. Absolute numbers. © World Health Organization 1989 The designations employed and the presentation of material in this publication do not imply the expression of any opinion whatsoever on the part of the Secretariat of the World Health Organization concerning the legal status of any country. terri- tory, city or area or of its authorities, or concerning the delim- itation of its frontiers or boundaries. The statistics which relate to the Federal Republic of Ger- many and the German Democratic Republic include the rele- vant statistics regarding Berlin for which separate data have not been supplied. This is without prejudice to any question of status which may be involved. Where the designation "country or area" appears in the headings of tables. it covers countries. territories, cities or areas. Signed articles express the opinions of the authors and do not necessarily represent the findings or policy of the World Health Organization. Couverture: Gilbert Auberson * T M F N Explication des signes Donnee preliminaire, approximative ou estimative. Donnee non disponible. Zero OU quantile negligeable. Categorie non applicable. Total. Masculin. Feminin. Nombres absolus. © Organisation mondialc de la Sante 1989 Les appellations employees dans cette publication et la pre- sentation des donnees qui y figurent n'impliquent de la part du Secretariat de !'Organisation mondiale de la Sante aucune prise de position quant au statut juridique des pays, territoires, villes ou zones. ou de leurs autorites. ni quant au trace de leurs frontieres ou limites. Les statistiques se rapportant a la Republique democratique allemande et a la Republique federale d' Allemagne compren- nent les statistiques pertinentes concernant Berlin pour lequel des donnees separees n'ont pas ete fournies. Toute question de statut demeure reservee. Lorsque !'appellation «pays ou zone» apparait dans le titre des tableaux. elle couvre les pays. territoires. villes ou zones. Les articles signes expriment les vues de leurs auteurs et ne correspondent pas necessairement aux conclusions ou a la politique adoptee par !'Organisation mondiale de la Sante. ~~ WORLD HEALTH STATISTICS QUARTERLY :'/t!!V;<J RAPPORT TRIMESTRIEL DE STATISTIQUES SANITAIRES MONDIALES Vol. 42, No. 1, 1989 PREVENTABLE MORTALITY CONTENTS [Preventable deaths: an overview) [French only]. Alan D. Lopez ............................. . Preventable mortality : indicator or target? Applications in developing countries. Renate Plaut & Edna Roberts . . . . . . The assessment of preventable infant and child deaths in devel- oping countries: some applications of a new index. Stan D'Souza ............................. . Excess mortality ratio with reference to the lowest age-sex- specific. death rates among countries. Kazuo Uemura . . . . . Geographical variations in mortality from conditions amenable to medical intervention in Europe: the European Community atlas of avoidable death (summary]. Elizabeth A. Paul et al. Years of potential life lost : application of an indicator for assessing premature mortality in Spain and Portugal. Luis Alberto Garcia Rodriguez & Luis Cayolla da Motta . . . . . . Page 2 LA MORTALITE EVITABLE SOMMA IRE Deces evitables: vue d'ensemble. Alan D. Lopez ...... . La mortalite evitable : indicateur ou objectif? Applications dans 4 les pays en developpement [resume]. Renate Plaut & Edna Roberts ............................. . L'evaluation des deces evitables de nourrissons et d'enfants dans les pays en developpement: quelques applications d'un 16 nouvel indice (resume]. Stan D'Souza ............. . Rapport de surmortalite fonde sur les plus bas taux de mortalite 26 des pays par sexe et par groupe d'age [resume]. Kazuo Uemura ............................. . Variations geographiques de la mortalite due a des maladies justiciables d'une intervention medicale en Europe - Commis- 49 sion des Communautes europeennes: atlas des deces evita- bles. Elizabeth A. Paul et al. . . . . . . . . . . . . . . . . . . . Annees de vie potentielle perdues: application d'un indicateur permettant d'evaluer la mortalite prematuree en Espagne et au 50 Portugal [resume]. Luis Alberto Garcia Rodriguez & Luis Cayolla da Motta ............................. . Pages 2 15 25 40 42 56 -2- DECES EVITABLES: VUE D'ENSEMBLE Alan D. Lopeza T oute appreciation de ce qui pourrait etre qualifie de mortalite « evitable » au « prematuree » b suppose de ma- niere implicite que I' on connaisse les forces ·qui agissent sur la mortalite dans une societe donnee et que l'on sache comment les quantifier. II est bien connu par exemple que la survie est etroitement liee a des variables biologiques comme l'ilge et le sexe ainsi qu'a de nom- breux facteurs socio-environnementaux parmi lesquels le niveau d'instruction, la profession et le mode de vie. Les services de sante disponibles et leur utilisation influent egalement sur la mortalite. L'etiologie de plusieurs causes majeures de mortalite reste relativement floue, si bien qu'il est dans ce cas plutot difficile d'evaluer les deces evitables. Cela etant, meme revaluation de !'importance relative, du point de vue de la sante publique, des deces pouvant etre consi- deres comme evitables quel que soit le critere retenu dependra de l'approche statistique adoptee. Ces consi- derations mises a part, la definition des deces evitables pouvant faire l'objet d'interventions sera bien evidem- ment largement fonction de facteurs d'ordre economi- que et logistique, ainsi que du contexte socio-culturel general - ainsi, des deces qui pourraient raisonnable- ment etre consideres comme prematures au evitables dans un contexte donne, pourront, du mains pour un temps, etre largement inevitables dans un autre con- texte, compte tenu du niveau de developpement social et economique. De toute evidence, !'appreciation de la mortalite evitable est un travail extremement complexe et ii n'est done pas surprenant que la plupart des etudes soient axees sur un aspect particulier du probleme, generalement dans un contexte socio-economique donne. Plusieurs des pro- blemes de methodologie que pose la mesure de la mor- talite evitable sont presentes ici avec des exemples illus- trant les moyens d'appliquer les methodes proposees a differentes populations parvenues a differents stades de developpement sanitaire. De toutes les regions en developpement du monde, c'est I' Amerique latine qui est le mieux parvenue a combattre la mortalite. Plusieurs pays de la region sont maintenant bien avances sur la voie de la transition epidemiologique et ant en outre mis au point un systeme d'information sur la mortalite apte a favoriser le developpement sanitaire. L'article de Plaut & Roberts, c qui fait etat de la situation en Argentine et au Mexique, montre tres bien comment peut etre utilise un tel systeme d'information pour eva- a Statisticien, Surveillance epidemiologique et appreciation de la situa- tion sanitaire dans le monde, Organisation mondiale de la Sante, Geneve. b II convient de souligner des le debut qu'il ne s"agit pas de prevenir des deces mais plutot de les retarder, c'est-a-dire d"empecher ceux qui pourraient etre qualifies de « deces prematures ». c La morralite evitable: indicateur ou objectif? Applications dans Jes pays en developpement. p. 4. d Rapport de surmortalite fonde sur Jes plus bas taux de mortalite des pays par sexe et par groupe d'iige, p. 26. • L ·evaluation des deces evitables de nourrissons et d"enfants dans les pays en deve/oppement: quelques applications d'un nouvel indice, p. 16. luer et surveiller les taux de deces evitables dans des pays d'Amerique latine. Deux mesures des tables de mortalite sont proposees dans cette etude, soit les taux-types de mortalite et « les annees de vie potentielle perdues », pour determiner !'importance des deces prematures. Ce dernier indice fournit essentiellement une mesure sommaire de la mor- talite en accordant davantage de poids aux deces surve- nus a un ilge relativement jeune, done largement evita- bles en theorie et qui, parallelement, peuvent etre consi- deres du point de vue des annees perdues de vie active potentielle. Pour cette raison, la limite d'ilge superieure pour le calcul de l'indice est souvent fixee a 65 ans bien que d'autres limites d'ilge puissent etre utilisees selon les buts precis de !'analyse. Cette methode de calcul met en lumiere !'importance des accidents de vehicule a moteur et du cancer, qui frappent de nombreux adultes jeunes. La situation sanitaire actuelle des deux pays etu- dies a ete mise en perspective a l'aide d'une comparai- son entre les cas de deces prematures et des mesures etablies a partir de donnees se rapportant au debut des annees 60 d'une part, et, d'autre part, grilce aux mesu- res obtenues pour un pays au la mortalite est faible (Etats-Unis d' Amerique). Cette idee d'apprecier la mor- talite prematuree a I' aide de taux « cibles » au «minimum» est developpee dans !'article d'Uemurad qui evoque !'ex- perience de pays industrialises. Dans une autre etude sur les deces evitables dans les pays en developpement, D'Souzae fait valoir que les mesures classiques de la mortalite infanto-juvenile ne permettent pas de mesurer la mortalite prematuree compte tenu des changements disproportionnes pou- vant affecter cette mesure a certains niveaux en fonction de !'importance des interventions de sante publique. II propose de transformer le taux de mortalite infantile base sur la fonction logistique en indice de la mortalite infantile avec des valeurs situees entre O et 100 (corres- pondant respectivement a des taux de mortalite infantile de 5 et 300 pour 1 OOO naissances vivantes). Deux appli- cations possibles de cet indice sont exposees dans I' ar- ticle. Dans les pays au la mortalite est faible, ii est possible d'utiliser des methodes plus complexes d'evaluation des deces evitables compte tenu de la disponibilite plus grande des donnees. Dans son article, d Uemura a pro- pose une mesure de la surmortalite (soit la mortalite evi- table) qui repose sur la notion d'un ensemble minimum de taux de deces. II est manifestement extremement important de determiner ce qui peut etre considere comme des taux de deces minimum. Etant donne l'he- terogeneite des risques aussi bien au sein des popula- tions qu'entre elles, ii est clair que ce qui pourrait etre considere comme un taux cible realisable et approprie dans un pays au la mortalite est faible serait impossible a atteindre, du mains a court terme, dans de nombreux pays en developpement. Par ailleurs, si l'on precisait un ensemble de taux cibles de mortalite pour les adultes, du mains pour les pays industrialises, ii se pourrait tres bien que ces taux soient deja enregistres dans de nombreu- ses societes plus pauvres au les facteurs de risque expo- Rapp. trimest. statist. sanit. mond .. 42 ( 1989) - 3 sant aux maladies chroniques sont mains repandus. Ce problem e. ainsi que d 'autres questions liees a I' utilisation d'un ensemble minimum ou cible de taux de deces, est evoque dans cet article. Les taux cibles de mortalite ne constituent bien sur qu'un moyen d'aborder l'etude des deces evitables. Dans un ouvrage tout a fait original, Rutstein et al. (1)ont propose une methode nouvelle de mesure de la qualite des soins medicaux basee sur une classification des cas inutiles de deces et d'incapacite. Les auteurs ont d'abord classe les situations pour lesquelles un cas meme unique de mala- die OU d'incapacite OU de deces premature justifierait que l'on pose la question: «Pourquoi cela s'est-il produit?». lls ont ensuite defini un deuxieme groupe de situations pour lesquelles l'accroissement des taux de morbidite, d'incapacite OU de deces prematures pourrait servir d'indicateur de l'efficacite des prestations de sante. c· est sur cette notion de mortalite evitable prise comme mesure du fonctionnement des services de sante que Paul et al. t ont base leur etude comparative sur les pays membres de la Communaute europeenne. Prenant comme point de depart la classification de Rutstein, ils ont recense plusieurs causes (a l'interieur de groupes d'age determines) de deces entierement OU pour une bonne part evitables compte tenu de la nature des ser- vices de sante en Europe. La mortalite relative due a ces causes a ete calculee pour chacune des 360 zones admi- nistratives de la Communaute europeenne a l'aide des taux comparatifs de mortalite. L'avantage de cette eva- luation au niveau local est bien entendu de determiner de fac;:on plus precise la distribution et la prevalence des deces evitables et de faciliter ainsi sensiblement I' elabo- ration de politiques de sante susceptibles de prevenir les deces «inutiles». L'atlas des deces evitables etabli pour la Communaute europeenne et sur la base duquel a ete redige cet article ne contient pas de donnees sur l'Espagne et le Portugal, ces deux pays n'etant pas encore membres de la Com- munaute lorsqu'a ete faite cette etude. On peut cepen- dant avoir une idee de la nature et de I' etendue de la f Variations geographiques de la mortalite due a des maladies justi- ciables d'une intervention medicate en Europe - Commission des Com- munautes europeennes: atlas des deces evitables. p. 42. g Annees de vie potentielle perdues: application d'un indicateur per- mettant d'evaluer la mortalite prematuree en Espagne et au Portugal, p. 50. mortalite evitable en Espagne et au Portugal d'apres !'ar- ticle de Garcia Rodriguez & da Motta9 qui, comme pour l'etude latino-americaine, utilise la notion d'annees de vie potentielle perdues pour determiner les principales causes de mortalite evitable dans ces pays. Ce qui frappe particulierement dans cette approche est la cotation net- tement differente donnee a certaines causes (accidents de vehicule a moteur par exemple) par rapport a ce que I' on obtient a partir d' autres moyens de ponderation des taux de mortalite par age et par cause (taux bruts de mortalite corriges de l'age). Des mesures telles que les annees perdues sont incontestablement tres utiles pour definir des politiques de sante et accederont on l'espere au rang d'instrument-type d'appreciation de la situation sanitaire. II faudrait enfin mentionner le mode de vie ou les facteurs lies a la prevention primaire. Manifestement, limiter les deces evitables (comme on I' a fait pour I' etude sur la Communaute europeenne) a ceux qui auraient du etre evites moyennant des prestations de sante correctes ne tient pas compte du fait amplement prouve que chacun peut sensiblement reduire ses chances de survie en adoptant des habitudes ou des modes de vie nefastes pour la sante (usage du tabac par exemple). Les deces dus a ce type de cause peuvent etre consideres comme evitables dans un sens plus general. L'inconvenient est qu'il est difficile de mesurer en termes quantitatifs la proportion exacte de ces deces qui tient a des choix individuels et pourrait done etre prevenue. C'est pour cette raison que relativement peu d'etudes ont ete · consacrees a cet aspect de la mortalite evitable, bien que le nombre des travaux consacres ces dernieres annees a la derivation de fractions etiologiques ait augmente. Une etude recente (2) a par exemple etabli, a partir de don- nees recueillies pour les Etats-Unis, des «fractions attri- buables a l'usage du tabac» pour plusieurs causes im- portantes de deces. jusqu'a 80-85% des deces par can- cer du poumon et maladies chroniques des voies respi- ratoires etant attribuables au tabac. La vaste etude de Doll & Peto (3) sur les causes de cancer offre un autre exemple de !'utilisation de donnees epide- miologiques pour determiner en termes quantitatifs le rOle de certains facteurs dus au mode de vie dans les deces par cancer. Ces deces peuvent etre consideres comme evitables et ii appartient done aux agents de sante de continuer a promouvoir des modes de vie sains et la prise de conscience de la necessite de preserver sa propre sante comme l'un des moyens les plus efficaces de reduire la proportion des deces qui pourraient etre prevenus. REFERENCES - REFERENCES 1. RuTSTEIN. D. D. ET AL. Measuring the quality of medical care: a clinical method. New England journal of medi- cine, 294: 582-588 (1976). 2. CENTERS FOR DISEASE CONTROL. Smoking-attributable mortality and years of potential life lost - United Wld hlth statist. quart .. 42 ( 1989) States, 1984. Morbidity and mortality weekly report, 36 (42): 693-697 (1987). 3. DOLL. R. & PETO. R. The causes of cancer: quantitative estimates of avoidable risks of cancer in the United States today. Oxford, Oxford University Press, 1981. -4- PREVENTABLE MORTALITY: INDICATOR OR TARGET? APPLICATIONS IN DEVELOPING COUNTRIES Renate Plauta & Edna Robertsb Death itself cannot be prevented. It can, however, be postponed. The public health importance of this fact has long ago motivated the development of measures for the analysis of mortality statistics, traditionally one of the main tools of public health planners and administrators for assessment of health status, definition of priorities and allocation of resources, and surveillance of specific health problems. It is equally recognized that non-violent death is but the last event in a continuum of progressively worse health; mortality statistics tell a very incomplete story about disease and suffering, and even less about individual and societal determinants of ill health. However, up to now a satisfactory operational definition of "good" health does not appear to exist, neither at the individual nor at the community level. Nor is it clear whether such a defi- nition would be at all feasible, and if so, whether it would be the same for all members of a community and com- munities everywhere (1). Furthermore, those variables that have been accepted as being both sensitive and specific enough to contribute to the assessment of health status are usually difficult to document and much too expensive to obtain for population-wide use. Accordingly, and without giving up the search for appro- priate indicators of positive health, increased efforts are being devoted to the development of indicators based on death statistics, thus acknowledging that the poten- tial information on health status to be extracted from mortality data is still far from exhausted. Mortality rates specific for sex, age, cause, place of residence and other social and economic characteristics of the decedent con- tinue to be the cornerstone of this information, but speci- fic rates are cumbersome to analyse. Crude and age- adjusted (standardized) mortality rates, however, share the major shortcoming of being dominated by mortality at old ages, at which most deaths occur and disease is harder to prevent. Summary measures are needed that, while assessing the impact of mortality as a whole will better reflect changes in those problems that exact their toll at an early age, and highlight the age groups in which this impact is felt the most. Woolsey (2) and Uemurac in their search for achievable target rates for the United States of America and worldwide, respectively, have discussed numerous approaches and have given abun- dant references to this effect. They provided both back- ground and stimulus for the discussion presented here. This renewed interest in mortality statistics can only be welcomed, as, in the words of Shapiro (3): ·· ... they represent the only continuous source of infor- mation on an unequivocal manifestation of health sta- tus that dates back many years and is assured of con- • Epidemiologist/Statistician, Health Situation and Trend Assess- ment, Pan American Health Organization, Washington. b Statistician, Health Situation and Trend Assessment, Pan American Health Organization, Washington. c Uemura, K. Excess mortality ratio with reference to the lowest age- sex-specific death rates among countries. p. 26 of this issue. tinuity into the foreseeable future, and the data can be examined on a geographically disaggregated level often down to subareas within a city, for example, or aggregated across civil subdivisions ... ". According to this author, the challenge of "how to max- imize the utility of this resource" is of special relevance to public health officials in developing countries, who are understandably reluctant to use scarce resources for the gathering of additional information on health problems, rather than for their prevention or alleviation. This article is dedicated to them. Purpose and objectives The purpose of this exercise is to show some applica- tions of mortality statistics based on the concepts of "excess" and "premature" mortality, in the hope that they may become useful components of health situation analyses performed by countries with the aim of contrib- uting to priority setting in the health-services system, and to the surveillance and evaluation of service and programme outcomes. Specifically, there will be a discussion of the scope and limitations of some simple procedures to analyse usually available data for: • estimating gains in mortality from all causes as- sessed against a country's own experience; and • quantifying the gap between the country's current mortality situation and one observed in a more developed country. This latter aspect will be discussed for mortality from all causes, and for cause-specific data using (a) broad cau- sal categories to visualize changes in the overall mor- tality structure, and (b) selected, more specific cause groups that might be useful as sentinel or tracer cate- gories. Procedures For this exercise, excess mortality will be defined empir- ically: mortality will be understood to be preventable if it has shown a sustained reduction over time, either in the country being analysed or in another country being used as reference. To estimate excess mortality two indicators will be used: the standardized mortality ratio (SMR) and the ratio of observed over expected years of potential life lost (RYPLL). Premature mortality will be defined as that occurring under 65 years of age. Both indicators will be computed for each sex; the SMR will be computed for premature mortality and for all ages. To compute age- specific frequencies, age groups are defined as follows: under 1 year, 1-4 years, 10-year groups from 5 to 64 years, and 65 years and above. Rapp. trimest. statist. sanit. mond., 42 (1989) Data from Argentina and Mexico will be used to illustrate the proposed procedures. These two countries have been chosen as examples because their population size prevents excessive instability of observed frequencies in specific categories; mortality data by age and sex are available for more than two decades; the proportion of deaths classified as due to "symptoms, signs and ill- defined conditions" is well under 10%; and their re- spective cause-specific mortality structures are differ- ent. To analyse past experience each country's data for 1982 will be compared to its own data 20 years earlier. As reference for a more favourable situation, the 1982 data for the largest developed country in the Region of the Americas, i.e. the United States, will be used. As the analysis will be centred on the year 1982, both to assess progress against the past as well as to size up the challenges still ahead, the reference population will be the midyear population estimate for 1982 for Argentina and Mexico. To stabilize the data the number of deaths for each of the years to be studied will be estimated to be the three-year average centred in that year. Thus, deaths for 1982 and 1962 will be understood to be the average number of deaths occurring in the years 1981-1982- 1983 and 1961-1962-1963 respectively, as shown in Annex, Table i. Computation of expected mortality will vary according to the purpose of the analysis. For evaluation of gains achieved, expected deaths will be those that would have occurred if the 1982 population had been subjected to the 1962 age- and sex-specific rates of the same coun- try. To compare with a more favourable health situation, expected deaths will be computed applying the three- year 1982-centred age- and sex-specific death rates of the United States to the 1982 population of Argentina and Mexico. Specific rates are shown in Annex, Table ii; expected numbers of deaths are shown in Annex, Table iiifor each sex and both sexes combined, the latter obtained by addition of male and female deaths. The overall SMR is computed by dividing total ob- served by total expected deaths; the SMR for mortality under 65 is restricted to the ratio of observed and expected deaths below that age limit. The last column of Annex, Table iii shows age-specific YPLL per death, i.e. the YPLL for each death in every age group, obtained by subtracting the midpoint of the age interval from 65, the upper limit. Observed and expected YPLL are computed by multiplying (weighting) these age-specific YPLL per death by the observed and expected number of deaths respectively, and adding over all age groups up to but not including 65. The RYPLL is the ratio of the observed YPLL and those expected. Cause-specific analysis For a general view of the changes in the overall mortality structure, causes of death were grounded into 10 broad categories roughly following the chapters of the Ninth Revision of the International Classification of Diseases (ICD-9): 1 . Infectious and parasitic diseases (001-139) ; 2. Neoplasms (140-239); 3. Diseases of the circulatory system (390-459); 4. Diseases of the respiratory system (460-519); 5. Diseases of the digestive system (520-579); 6. Complications of pregnancy, childbirth and the puerperium (630-676); 7. Congenital anomalies (740-759); 8. Certain conditions originating in the perinatal period (760-779); 9. All other diseases (remainder of 001-779); and 10. External causes (E800-E999). Wld hlth statist. quart .• 42 ( 1989) 5 - Deaths due to symptoms, signs and ill-defined condi- tions (780-799) are shown separately as an indicator of data quality; they were not redistributed among defined causes. Annex, Table iv shows observed and expected deaths for these categories, cause-specific proportional mortality and number and percentage of deaths prior to age 65. Some sentinel or tracer categories were defined, to explore their potential use for surveillance and evalu- ation. They will be described in the next section. Results In accordance with the purpose of this article, presenta- tion of results will focus on the indicators rather than on the health situation of the two countries chosen as examples. Monality from all causes The SMR for all ages, the SMR for deaths occurring before age 65, and the RYPLL, also for deaths prior to age 65, are compared in Table 1. The interpretation of these indicators is simple enough to use them for con- veying messages to the general public or authorities not trained in public health. In Argentina the number of male deaths observed in 1982 represents 80.4% of those which would have been expected if the 1962 rates had prevailed, i.e. 19.6% of expected male-and 21.4% of expected female-deaths were avoided due to the reduction in mortality rates experienced since 1962. Similarly, there were savings of 38.5% and 51.2% of expected deaths for men and women in Mexico. Under age 65, the observed savings for each 100 deaths expected were 28.2% for men and 36.8% for women in Argentina and 45.4% and 59. 7% respec- tively in Mexico. With respect to the RYPLL, for each 100 YPLL expected in the 1982 population if 1962 rates had prevailed, observed data showed a reduction of 41.2% and 47.2%, and 54.3% and 64.8%, for men and women in Argentina and Mexico respectively. In this example it is clear that the SMR under 65 is more sen- sitive to rate changes than the SMR for all ages, and the RYPLL is the most sensitive of all. The greater sensitivity to change of the RYPLL can also be appreciated when a more favourable set of sex- and age-specific rates-such as those of the United States- is used for the comparison. Under these reference rates the YPLL observed exceeded those expected far more than the deaths did, as evidence by the magnitude of the RYPLL in comparison to that of the SMRs. It should be kept in mind that SMRs and RYPLLs of different countries should be compared only to the extent that one would compare crude rates, as the popu- lation of each country is used in both numerator and denominator (4). By the same token this simplifies inter- pretation, since the only difference in numerator and denominator of each ratio derives from the mortality rates used. Cause-specific monality To avoid difficulties arising from changes between dif- ferent ICD Revisions and in data quality, cause-specific mortality will only be analysed with reference to the 1982 United States rates. Table 2 (A & 8) shows the number and percentage dis- tribution of observed and expected YPLL, the YPLL rates -6- TABLE 1. COMPARISON OF STANDARDIZED MORTALITY RATIOS (SMR)a FOR ALL AGES AND UNDER 65 YEARS AND RATIO OF YEARS OF POTENTIAL LIFE LOST (RYPLL), ARGENTINA AND MEXICO, 1982 TABLEAU 1. COMPARAISON DES TAUX COMPARATIFS DE MORTALITE (SMRJ 8 POUR TOUSLES AGES ET POUR MOINS DE 65 ANS ET RAPPORT DES ANNEES DE VIE POTENTIELLE PERDUES (RYPLLJ, ARGENTINE ET MEXIQUE,1982 Indicators - lndicateurs M F T Argentina - Argentine Past experience ( 1962) - SMR all ages - tous ages 80.4 78.6 79.6 Situation passee ( 1962) SMR <65 71.8 63.2 68.4 RYPLL 58.8 52.8 56.2 Future reference (United States) - SMR all ages - tous ages 119.7 128.8 123.5 Reference future (Etats-Unis) SMR <65 132.2 138.0 134.3 RYPLL 157.4 187.7 168.4 Mexico - Mexique Past experience ( 1962) - SMR all ages - tous ages Situation passee ( 1962) SMR <65 RYPLL Future reference (United States) - SMR all ages - tous ages Reference future (Etats-Unis) SMR <65 RYPLL • All ratios multiplied by 100 - Ensemble des rapports multiplie par 100. Source: Annex. Tables i & iii - Annexe, tableaux i et iii. per 100 OOO population under 65 and the RYPLL for broad causal categories. While for all causes mortality under age 65 accounts for 49% and 37% of all deaths in men and women in Argentina, and for 71 % and 60% in Mexico, there is wide cause-specific variation reflecting the age at which different conditions strike (see also Annex, Table iv). Because of space restrictions, the information provided by the SMRs and the RYPLL in comparison to other indi- cators will be illustrated using two cause-of-death cate- gories only: diseases of the circulatory system (390- 459) and infectious and parasitic diseases (001-139). Most countries in the Americas exhibit diseases of the circulatory system among the five leading causes of death while infectious and parasitic diseases rarely ap- pear among the leading five. In Argentina diseases of the circulatory system are the leading cause of death for men in 1982, representing 45% of all male deaths from defined causes, and their mortality rate amounts to 401.4 per 100000 male population. Infectious and pa- rasitic diseases, by comparison, rank eighth for Argen- tine men with 3.3% of all male deaths and a death rate of 29.8 per 100000 male population, 13.5 times smaller than the rate for the leading cause. The observed YPLL rates are closer to each other, with 1 957 YPLL per 100 OOO male population under 65 for diseases of the circulatory system and 901 for infectious and parasitic diseases. The SMRs for infectious and parasitic diseases are 3.89 for all ages and 5.59 for deaths prior to age 65; for diseases of the circulatory system these ratios are 1 . 18 and 1.26 respectively. The most striking piece of infor- mation, however, is provided by the RYPLL: for Argen- tine men this amounts to over 8 observed YPLL from infectious and parasitic diseases for every year of life expected to be lost due to this cause group according to United States rates, while for diseases of the circulatory system the multiplying factor is only about 1.5. This situation is even more pronounced in Mexico, where for infectious and parasitic diseases the YPLL observed for women exceed YPLL expected according to United States rates by a factor of 33. 61.5 48.8 55.4 54.6 40.3 48.0 45.7 35.2 40.8 145.4 162.6 152.3 184.8 204.5 191.9 235.0 263.9 245.8 In some countries deaths due to complications of preg- nancy, childbirth and the puerperium (630-676) are used as sentinel events; any death in this category is deemed excessive, and an investigation of the why and how is set in motion whenever it occurs. However, although important progress has been achieved in regard to improving maternal care, the RYPLL for maternal deaths still exceeds 13 in Argentina and 24 in Mexico. Although originally defined from a different perspective (5, 6), the concept of tracer and sentinel categories can be extended to include cause groups which might be used for surveillance of certain health situations or, at least, to illustrate the road still ahead. Table 3 shows three categories which, among others, might serve this purpose. Diseases preventable by vaccination, such as diphtheria, whooping cough, tetanus, poliomyelitis and measles (032; 033; 037; 045; 055) comprise a cate- gory useful for evaluation of vaccination programmes. The category of intestinal infectious diseases (001-009) is proposed as an indicator of sanitary conditions and the nutritional status of the population. Finally, appendicitis, hernia and intestinal obstruction (540-543; 550-553; 560) is proposed as a category for assessing the quality of abdominal surgery, and even to a certain extent, of general surgery and hospital care. The discrepancies between observed and expected deaths, and in consequence between observed and expected YPLL rates and the RYPLL itself are striking enough and the messages they convey differ sharply from those obtained when examining only leading_ causes of death and mortality rates. While reinforcing the conclusions regarding the importance of the cate- gory of infectious and parasitic diseases as a whole (Table 2), the first category of Table 3-vaccination pro- grammes-evidences a persistent public health chal- lenge, while the second-sanitation and nutrition-may respond more to overall development, public health included. It is notable, and deserves further research, that the gap between developing and developed coun- tries seems smaller once intra-hospital medical care is involved, as for abdominal surgery. Rapp. trimest. statist. sanit. mond .. 42 (1989) -7- TABLE 2. YEARS OF POTENTIAL LIFE LOST (YPLL) BY BROAD CAUSAL CATEGORIES, ARGENTINA AND MEXICO, 1982 TABLEAU 2. ANNEES DE VIE POTENTIELLE PERDUES (YPLLJ PAR GRANDES CATEGORIES DE CAUSES, ARGENTINE ET MEXIQUE, 1982 Causes of death - Causes de deces A. ARGENTINA - ARGENTINE All causes - Toutes causes (001-E999) Symptoms and ill-defined conditions - Symptames, signes et etats morbides mal definis (780-799) . . . ..... Total number of deaths from defined cau- ses - Nombre total de deces dus a des causes definies . . . . . . . . . Infectious and parasitic diseases - Mala- dies infectieuses et parasitaires (001- 139) ..... . Neoplasms - Tumeurs (140-239) ... Diseases of the circulatory system - Mala- dies de I' appareil circulatoire (390- 459) Diseases of the respiratory system - Ma- ladies de I' appareil respiratoire (460- 519) Diseases of the digestive system - Mala- dies de l'appareil digestif (520-579) Complications of pregnancy, childbirth and the puerperium - Complications de la grossesse, de l'accouchement et des suites de couches (630-676) ..... Congenital anomalies - Anomalies conge- nitales (740-759) .. Certain conditions originating in the perina- tal period - Certaines affections dont l'origine se situe dans la periode perina- tale (760-779) . . . . . . . ..... All other diseases (remainder of 001-779) - Toutes autres maladies (reste des rubriques 001-779) . . . . External causes - Causes externes (E800- E999) .......... . All causes - Toutes causes (001-E999) Symptoms and ill-defined conditions - Symptomes, signes et etats morbides mal definis (780-799) .... Total number of deaths from defined cau- ses - Nombre total de deces dus a des causes definies . . . . . . . . . . . . Infectious and parasitic diseases - Mala- dies infectieuses et parasitaires (001- 139) ............. . Neoplasms - Tumeurs (140-239) . Diseases ofthe circulatory system - Mala- dies de l'appareil circulatoire (390- 459) ...... . Diseases of the respiratory system - Ma- ladies de l'appareil respiratoire (460- 519) .......... . Diseases of the digestive system - Mala- dies de l'appareil digestif (520-579) Complications of pregnancy, childbirth and the puerperium - Complications de la grossesse, de l'accouchement et des suites de couches (630-676) ..... Congenital anomalies - Anomalies conge- nitales (740-759) .......... . Certain conditions originating in the perina- tal period - Certaines affections dont I' origine se situe dans la periode perina- tale (760-779) ........... . All other diseases (remainder of 001-779) - Toutes autres maladies (reste des rubriques 001-779) . . . . . . . . . External causes - Causes externes (E800- E999) .......... . Wld hlth statist. quart., 42 ( 1989) YPLL (000) Observed Observees 1 702.4 80.9 1 621.5 121.2 147.9 263.3 125.1 64.2 95.8 357.4 146.5 299.9 1 159.0 61.3 1 097.6 103.8 120.4 145.8 94.7 35.9 15.6 82.6 265.0 127.1 106.9 Expected Prevues 1081.4 59.1 1022.3 14.7 122.6 176.2 35.0 37.0 70.4 142.3 61.6 362.6 617.4 38.0 579.4 11.1 111.0 77.6 23.9 20.5 1.2 61.2 107.5 46.7 118.7 % YPLL rate' - Taux d'YPLL' Observed Observe Expected Prevu Observed Observe Male - Sexe masculin 100.0 100.0 12 655.9 4.8 100.0 7.5 9.1 16.2 7.7 4.0 5.9 22.0 9.0 18.5 5.5 601.7 100.0 12 054.0 1.4 901.1 12.0 1 099.4 17.2 1 957.4 3.4 930.1 3.6 476.9 6.9 712.4 13.9 2 657.2 6.0 1 088.8 35.5 2 229.6 Female - Sexe feminin 100.0 5.3 100.0 9.5 11.0 13.3 8.6 3.3 1.4 7.5 24.1 11.6 9.7 100.0 6.2 100.0 1.9 19.2 13.4 4.1 3.5 0.2 10.6 18.6 8.1 20.5 8 699.8 460.1 8 239.2 778.9 904.1 1094.3 710.8 269.2 116.8 620.0 1 988.9 954.0 802.6 Expected Prevu 8 037.6 439.3 7 598.1 109.3 911.1 1 309.8 260.3 274.9 523.0 1 058.0 457.8 2 695.7 4635.1 285.1 4 349.5 83.0 833.4 582.6 179.3 154.1 8.9 459.4 807.0 350.5 891.1 RYPLL b 1.57 1.37 1.59 8.24 1.21 1.49 3.57 1.74 1.36 2.51 2.38 0.83 1.88 1.61 1.89 9.39 1.08 1.88 3.96 1.75 13.08 1.35 2.46 2.72 0.90 -8- TABLE 2 (continued) TABLEAU 2 (suite) YPLL (000) YPLL rate• - T aux d'YPLL' Causes of death - Causes de deces B. MEXICO - MEXIQUE All causes - Toutes causes (001-E999) Symptoms and ill-defined conditions - Sympt6mes, signes et etats morbides mal definis (780-799) . . . . Total number of deaths from defined cau- ses - Nombre total de deces dus a des causes definies . . . . . . . . . . . . Infectious and parasitic diseases - Mala- dies infectieuses et parasitaires (001- 139) ........... · Neoplasms - Tumeurs ( 140-239) Diseases of the circulatory system - Mala- dies de I' appareil circulatoire (390- 459) ....... . Diseases of the respiratory system - Ma- ladies de l'appareil respiratoire (460- 519) ............ . Diseases of the digestive system - Mala- dies de l'appareil digestif (520-579) . Complications of pregnancy, childbirth and the puerperium - Complications de la grossesse, de l'accouchement et des suites de couches (630-676) ..... Congenital anomalies - Anomalies conge- nitales (740-759) .......... . Certain conditions originating in the perina- tal period - Certaines affections dont l'origine se situe dans la periode perina- tale (760-779) . . . . . . . . . . . . All other diseases (remainder of 001-779) - Toutes autres maladies (reste des rubriques 001-779) External causes - Causes externes (E800- E999) ........... . All causes - Toutes causes (001-E999) Symptoms and ill-defined conditions - Sympt6mes, signes et etats morbides mal definis (780-799) . . . . . . . . . Total number of deaths from defined cau- ses - Nombre total de deces dus a des causes definies . . . . . . Infectious and parasitic diseases - Mala- dies infectieuses et parasitaires (001- 139) ....... · · · · Neoplasms - Tumeurs (140-239) Diseases ofthe circulatory system - Mala- dies de I' appareil circulatoire (390- ~~ ............ . Diseases of the respiratory system - Ma- ladies de l'appareil respiratoire (460- 519) . . . . . . . . . ..... Diseases of the digestive system - Mala- dies de l'appareil digestif (520-579) . Complications of pregnancy, childbirth and the puerperium - Complications de la grossesse, de l'accouchement et des suites de couches (630-676) ..... Congenital anomalies - Anomalies conge- nitales (740-759) Certain conditions originating in the perina- tal period - Certaines affections dont l'origine se situe dans la periode perina- tale (760-779) . . . . . . All other diseases (remainder of 001-779) - Toutes autres maladies (reste des rubriques 001-779) .... External causes - Causes externes (E800- E999) Observed Observees 6 441.9 289.5 6 152.4 1 208.2 143.0 296.9 943.8 309.0 215.8 850.6 526.0 1 659.0 4 320.8 247.6 4073.2 10510.4 1 812.0 2 555.4 7 692.8 1418.1 758.6 185.3 5 911.4 4 255.0 3 964.2 Expected Prevues 2 741.4 183.2 2 558.2 40.2 221.5 287.8 86.7 69.6 237.3 483.1 156.4 975.6 16 370.5 1 208.5 15162.0 317.8 2 033.3 1 414.8 619.5 421.5 31.1 2 070.3 3 651.4 1 211.2 339.1 Observed Observe Expected Pre vu Observed Observe Male - Sexe masculin 100.0 4.5 100.0 19.6 2.3 4.8 15.3 5.0 3.5 13.8 8.6 27.0 100.0 6.7 100.0 1.6 8.7 11.3 3.4 2.7 9.3 18.9 6.1 38.1 18 162.1 816.2 17 345.9 3 406.4 403.3 837.1 2 660.8 871.2 608.4 2 398.2 1 483.1 4677.3 Female - Sexe feminin 100.0 5.7 100.0 25.8 4.4 6.3 18.9 3.5 1.9 4.5 14.5 10.4 9.7 100.0 7.4 100.0 2.1 13.4 9.3 4.1 2.8 0.2 13.7 24.1 8.0 22.4 12 305.2 705.1 11600.1 2 993.3 516.0 727.8 2 190.8 403.9 216.0 527.6 1 683.5 1 211.8 1 129.0 • Rates per 100 OOO population under age 65 - Taux pour 100 OOO habitants de moins de 65 ans. O RYPLL: ratio of years of potential life lost - Rappon des annees de vie potentielle perdues. Source Annex, Tables i & iv, and PAHO technical data base - Annexe, tableaux i & iv, et base de donnees techniques OPS. Expected Pre vu 7728.7 516.5 7 212.8 113.4 624.6 811.5 244.4 196.2 669.1 1 362.0 440.9 2 750.5 4 663.2 344.2 4 318.7 90.5 579.1 402.9 176.4 120.0 8.8 589.6 1 039.9 344.9 965.7 RYPLL 0 2.35 1.58 2.40 30.03 0.65 1.03 10.89 4.44 0.91 1.76 3.36 1.70 2.64 2.05 2.69 33.07 0.89 1.81 12.42 3.36 24.43 0.89 1.62 3.51 1.17 Rapp. trimest. statist. sanit. mond .. 42 ( 1989) -9- TABLE 3. RATIOS OF YEARS OF POTENTIAL LIFE LOST (RYPLL) FOR SENTINEL CATEGORIES RELATED TO DIFFERENT PROBLEMS, ARGENTINA AND MEXICO, 1982 8 TABLEAU 3. RAPPORTS DES ANNEES DE VIE POTENTIELLE PERDUES (RYPLLJ POUR DES CATEGORIES SENTINELLES RELATIVES A CERTAINS PROBLEMES, ARGENTINE ET MEXIQUE, 1982 8 Male - Sexe masculin Female - Sexe feminin Deaths < 65 YPLL ratesb Deaths <65 YPLL ratesb Sentinel categories for problems related to Deces < 65 ans RYPLL Taux d'YPLL b Deces < 65 ans Taux d'YPLL b Categories sentinelles relatives a Observed Expected Observed Expected Observed Expected Observed Expected RYPLL Observes Prevus Observes Prevus Observes Prevus Observes Prevus Argentina - Argentine Vaccination programmes c - Pro- grammes de vaccination c . . . Sanitation and nutrition d - Hygiene 99 37.3 116 39.9 0.5 83.0 et nutrition d 914 23 417.9 9.2 45.4 809 18 375.1 6.6 56.7 Abdominal surgery• - Chirurgie de 3.3 !'abdomen• 261 82 53.4 17.7 3.0 211 65 38.7 11.7 Mexico - Mexique Vaccination programmes c - Pro- grammes de vaccination c 913 140.8 0.2 780.4 1 053 2 155.9 0.4 424.4 Sanitation and nutritiond - Hygiene et nutrition d 16 463 65 2 702.0 11.0 246.4 14 227 52 2 355.0 8.7 269.9 Abdominal surgery• - Chirurgie de !'abdomen• 757 172 79.2 19.7 4.0 675 123 66.5 12.3 5.4 a D1screpanc1es with figures in previous tables are due to rounding - Les chiffres peuvent presenter des ecarts par rapport~ ceux des tableaux precedents du fait qu'ds ont ere arrondis. b Rates per 100 OOO population under age 65 - Taux pour 100 OOO habitants de moins de 65 ans. c Diphtheria (032). whooping cough (033). tetanus (037). acute poliomyelitis (045) and measles (055) - Diphterie (032). coqueluche (033), tetanos (037), poliomyelite aigui! (045) et rougeole (055). d Intestinal infectious diseases (001-009) - Maladies infectieuses intestinales (001-009). • Appendicitis (540-543), hernia of abdominal cavity (550-553) and intestinal obstruction without mention of hernia (560) - Appendicite (540-543). hernie abdominale (550-553) et occlusion intestinale sans mention de hernie (560). Source: PAHO technical data base - Base de donnees techniques OPS. Discussion and conclusions An important consideration relevant to cause-specific analyses is the fact that mortality statistics in their usual form are based solely on the underlying cause of death and therefore tend to underestimate conditions rarely recorded as such, as is the case for malnutrition, among others (7). In addition, it should be kept in mind that the way causal categories are grouped may greatly influence their relative ranking. These issues were discussed at a recent meeting on mortality analysis at which different criteria for cause-groupings, and the construction and use of the YPLL were presented by research teams from several Latin American countries (8-10). One of the recommendations called for improved use of available data and the promotion of simple procedures such as YPLL to highlight preventable mortality. In some countries in Latin America there exists a long tradition in the use of mortality analyses to assess health differentials (11). Only a few specific examples will be mentioned: Behm (12) set a trend in regard to the an- alysis and interpretation of infant mortality in 1962; Taucher' s avoidability criteria do not seem to have been fully exploited so far, in spite of their application in sev- eral countries ( 13); Becker started using YPLL in 1980 (14). Although comparative evaluations exist, there has not been an emphasis on target rates. Nor is there in this article, since it is felt that, with few exceptions, condi- tions responsible for premature death in the developing countries of the Region of the Americas will respond only in part to actions of the health sector: health status appears to be more strongly influenced by the overall living conditions still prevailing for the vast majority of the population in those countries. This aspect, which to a lesser degree also seems to hold in the developed world, as Buck ( 15, 16) and others have pointed out, should be Wld hlth statist. quan .. 42 ( 1989) kept in mind when societal determinants are interpreted as being individual choices and targeted for intervention (17-19). Indicators for excess and premature mortality can be computed for any age-specific subgroup of the popu- lation, and there has been much discussion about how they should be defined. But, as Haenszel (20) says, the problem " ... is not on the mechanics of rate construction but in definition of terms and deciding what is to be meas- ured. The choice of a rate under one criterion would not necessarily preclude the use of another rate under different circumstances ... ". This statement applies equally to age limits and refer- ence rates, the selection of which should be guided by the circumstances of the analysis, and even more so, by its purpose. All three indicators presented here, namely the SMR for all ages and for deaths occurring prior to age 65 and the ratio of observed over expected YPLL, were selected because they are simple to use for the purpose at hand. This is the main reason why YPLL were given preference over indicators derived from life tables; the fact that they use observed data was an added consideration. In their excellent discussion of the main issues involved in the construction and use of the YPLL, the Centers for Disease Control point out that instead of using a com- mon fixed limit, the life expectancy remaining for each age group could be used as that group's upper limit (21). It is felt, however, that this would detract from one of the main appeals of this indicator, namely its simplicity. Cause-specific YPLLs represent the number of years lost - 10 - due to the cause in question under the assumption that- up to the age set as the indicator's upper limit-the decedent would not have died from any other cause. Thus, unlike life-table measures computed for compet- ing risks, YPLL do not address the question of how much life expectancy would be gained should a certain causal category be eliminated or reduced. The 65-year age limit was chosen in this article because, on a population-wide basis, mortality at 65 years and above appears to be more difficult to postpone; it should not be interpreted to imply a limit to economically active or potentially productive life. However, this cut-off point can be varied according to a country's circumstances and the purpose of the analysis. Another important choice involves the reference rates to be used, especially when assessing the gap between what is and what could be. Again, this choice is entirely dependent on the purpose of any given analysis, and the decisions to be based on it. In this context, and also in regard to sentinel categories, it is important to distin- guish between outcome measures and health status indicators. Mortality from diseases preventable by vac- cination can be easily interpreted as failure of the immu- nization programme, while mortality from intestinal in- fectious diseases appears to carry a more general mes- sage about prevailing health and general living condi- tions. It is well documented that some indicators such as infant mortality do not maintain their correlation with overall health or development status after having been subjected to sustained and effective interventions: they then measure-at least in part-the outcome ofthe inter- vention programme. Cause-specific reference rates can be defined based on the knowledge of state-of-the-art technology, and they can be constructed according to prospective planning or scenario techniques. They can also be chosen empirical- ly, as was done here, based on observed mortality reduction in the same country and in a more developed one. In addition, the indicators discussed can be used to high- light differentials and inequalities within a country. Thus, on a subnational level, the reference rates could be those of that region or area in the country exhibiting the least unfavourable sanitary conditions, as Farr proposed over 150 years ago. This idea is especially attractive since in almost all countries of the Americas there exist mortality statistics of sufficient completeness to do this compa- rative exercise, at least for mortality from all causes. The SMR under age 65 or the RYPLL should be excellent planning tools, since they use a country's or area's own population and thus assess the health status from within that area and in regard to itself. Cause-specific analysis will be restricted by data quality. Informative as they are, the ratios discussed are not meant to be used instead of the more traditional indica- tors, but as their complement. No summary measure will substitute the richness of knowledge to be extracted from the analysis of rates specific for sex, age, cause, and known or suspected risk factors. Where summary measures are needed, the level of mortality is still best measured by mortality rates. When comparisons over time or among countries-or different areas within any one country-are desired, rates adjusted for age (by the so-called direct method) will still be the indicator of choice, just as proportional mortality continued to be the measure most appropriate for studying its structure. The RYPLL will be an excellent complement for the assess- ment of differentials and inequalities, of gains achieved and challenges still ahead. But. since ratios only express the relation between two numbers, saying nothing about the size of either one, they should never be used without an indicator providing a yardstick for the size of at least one of the ratio's components. As stated before, the procedures presented here are geared towards analysis to be used by a country or subnational area for its own benefit. Countries in the Americas and elsewhere are invited to replicate this exercise and enrich it with their own perspectives and experience. The quest for better mortality indicators and their application will no doubt contribute new insights to the health-disease processes. However, it should not be forgotten that low mortality is not synonymous with good health. The goals of public health are not-or should not be-restricted to making life longer; they should aim at making life better. Acknowledgements The authors are grateful to Dr Clovis H. Tigre for encour- agement and valuable suggestions; Mrs Aida Perez and Mr Robert Farr for critical programming support; Miss Maria Alicia Heusser and Miss Gabriela Fort for assis- tance in the preparation of the tables; and Mrs Maria Stella Dabancens for referencing bibliographical mate- rial. The secretarial support of Mrs Ana Maria Doria- Medina and Miss Lourdes Noel is also appreciated. Residual errors, however, are the sole responsibility of the authors. Rapp. trimest. statist. sanit. mond., 42 ( 1989) 11 - ANNEX - ANNEXE Table i. Midyear population and number of deaths, Argentina and Mexico, 1982 Tableau i. Population en milieu d'annee et nombre de deces, Argentine et Mexique, 1982 Argentina - Argentine Mexico - Mexique Population • Deaths O - Deces O Population • Deaths O - Deces b Age groups - Groupes d·age M F M F M F M F <1 .. 353 339 11 702 9 127 1 198 1 152 45 548 35 494 1-4 1346 1 308 1954 1 665 4849 4 676 11 897 10898 5-14 2 743 2666 1439 972 10 595 10237 8 148 5 434 15-24 2410 2347 2 784 1 602 7 479 7 295 17 371 6854 25-34 2 170 2 116 3 658 2 376 4823 4836 18 309 7 325 35-44 1 729 1 712 6469 3 825 3070 3 187 18 640 9054 45-54 1 534 1 566 14 227 6 891 2 109 2 239 20 729 12 146 55-64 1 166 1 268 23 787 11 872 1 346 1 492 23 774 16 706 <65 13451 13 322 66020 38 330 35 469 35 114 164416 103 911 65 + 1050 1 333 67 868 64100 1 178 1 424 68 274 69 885 All ages - T ous ages 14 501 14655 133 888 102 430 36 647 36 538 232 690 173 796 • Midyear population estimate for 1982, in thousands - Estimation de la population au milieu de l'annee 1982, en milliers. b Average of the deaths registered for 1981, 1982 and 1983. Excludes a yearly average of 2 541 deaths of unknown sex and 7 64 7 of unknown age in Mexico. as well as 297 deaths of unknown sex and 5 4 79 of unknown age in Argentina - Moyenne des deces enregistres pour 1981, 1982 et 1983. Est exclue une moyenne annuelle de 2 541 deces de personnes de sexe indetermint! et de 7 647 personnes d'Age inconnu au Mexique, et de 297 personnes de sexe indt!termine et de 5 479 personnes d·age inconnu en Argentine. Source : PAHO technical data base - Base de donnees techniques OPS. Table ii. Mortality rates by age and sex, Argentina and Mexico, 1962 and 1982; United States of America, 1982 • Tableau ii. Taux de mortalite par ige et par sexe, Argentine et Mexique, 1962 et 1982; Etats-Unis d'Amerique, 1982 8 Argentina - Argentine Mexico - Mexique United States of America Etats-Unis d. Amt!rique Age groups - Groupes d'Age 1962 1982 1962 1982 1982 M F M M F M F M F <1 6 774.5 5 760.7 3 315.0 2 692.3 8489.1 7 234.2 3 802.0 3081.1 1 271.0 1 018.4 1-4 348.7 346.3 145.2 127.3 1 178.1 1 260.7 245.3 233.0 64.8 51.1 5-14 83.1 62.2 52.5 36.5 201.2 188.8 76.9 53.1 34.0 22.0 15-24 172.0 115.9 115.5 68.3 278.8 227.1 232.3 94.0 149.7 52.3 25-34 243.5 164.4 168.6 112.3 477.7 359.9 379.6 151.5 181.1 71.1 35-44 431.1 269.9 374.1 223.4 740.2 537.0 607.1 284.1 275.7 145.4 45-54 1 014.5 530.0 927.4 440.0 1 165.7 825.2 982.9 542.5 720.2 393.8 55-64 2 452.2 1 245.1 2 040.1 936.3 2 206.5 1 807.4 1 766.3 1 119.7 1 741.4 921.3 65 + 7 112.9 5 242.2 6463.6 4808.7 6 564.0 6 870.2 5 795.8 4 907.7 6 156.4 4 380.0 All ages - Taus ages 1 005.9 706.3 923.3 698.9 1 129.6 1 017.1 634.9 475.7 943.2 777.0 • Rates per 100 OOO population were computed using as numerator one-third of the deaths registered for 1961, 1962, 1963, and for 1981, 1982. 1983 respectively, and as denominator the midyear population for the middle year, i.e. 1962 and 1982 respectively - Les taux pour 100 OOO habitants ont t!tt! calcules en utilisant comme numt!rateur un tiers des deces enregistres pour 1961, 1962, 1963 et pour 1981, 1982, 1983, respectivement, et comme dt!nominateur la population au milieu de l'annt!e mt!diane, soit 1962 et 1982, respectivement. Source: PAHO technical data base - Base de donnt!es techniques OPS. Wld hlth statist. quart., 42 (1989) - 12 - Table iii. Deaths expected in 1982 population of Argentina and Mexico according to country's 1962 rates and United States of America rates for 1982 Tableau iii. Decis prevus en 1982 en Argentine et au Mexique selon les taux du pays en 1962 et les taux des Etats-Unis d'Amerique en 1982 M F Total Age groups - Groupes d·age United States Etats-Unis United States Etats-Unis United States Etats-Unis 1962 1962 1962 Argentina - Argentine <1 .. 23 914 4487 19 529 3452 43443 7 939 1-4 4694 872 4530 668 9224 1 540 5-14 2 279 933 1 658 587 3 937 1 520 15-24 4145 3608 2 720 1 227 6 865 4835 25-34 5 284 3930 3 479 1 504 8 763 5434 35-44 7454 4 767 4621 2489 12075 7 256 45-54 15 562 11048 8300 6 167 23 862 17 215 55-64 28 593 20305 15 788 11 682 44381 31 987 <65 91 925 49950 60625 27 776 152 550 77 726 65+ 74685 61 884 69 879 51 831 144 564 113 715 All ages - Tous ages 166 610 111 834 130 504 79 607 297 114 191 441 Mexico - Mexique <1 .. 101 699 15 227 83338 11 732 185 037 26 959 1-4 57 126 3 142 58950 2 389 116 076 5 531 5-14 21 317 3602 19 327 2 252 40644 5 854 15-24 20851 11 196 16 567 3 815 37 418 15 011 25-34 23039 8 734 17 405 3438 40444 12 172 35-44 22 724 8464 17 114 4634 39 838 13098 45-54 24 585 15 189 18 476 8 817 43061 24006 55-64 29699 23 439 26966 13 746 56 665 37 185 <65 301 040 88993 258 143 50823 559 183 139 816 65+ 77 324 71 017 97 832 56081 175 156 127098 All ages - Tous ages 378 364 160010 355 975 106 904 734 339 266 914 • Years of potential life lost - Annees de vie potentielle perdues. YPLL • for each age group YPLL' pour chaque groupe d'Age 64.5 62.0 55.0 45.0 35.0 25.0 15.0 5.0 64.5 62.0 55.0 45.0 35.0 25.0 15.0 5.0 Table iv. Deaths observed in 1982 and expected according to United States of America 1982 rates, by broad causal categories, Argentina and Mexico• Tableau iv. Decis observes en 1982 et deces prevus en fonction des taux des Etats-Unis d'Amerique en 1982, par grandes categories de causes, Argentine et Mexique a Causes of death - Causes de deces A. ARGENTINA - ARGENTINE All causes - Toutes causes (001-E999) Symptoms and ill-defined conditions - SymptOmes, signes et etats morbides mal definis (780-799) . . . . . . . . . Total number of deaths from defined cau- ses - Nombre total de deces dus a des causes definies . . . . . . . . Infectious and parasitic diseases - Mala- dies infectieuses et parasitaires (001- 139) .......... . Neoplasms - Tumeurs (140-239) .. . Diseases of the circulatory system - Mala- dies de l'appareil circulatoire (390- 459) ................. . Diseases of the respiratory system - Ma- ladies de l'appareil respiratoire (460- 519) .............. . Diseases of the digestive system - Mala- dies de l'appareil digestif (520-579) . Male - Sexe masculin Female - Sexe Mminin All ages - T ous ages <65 All ages - T ous Ages <65 Number % all causes Number % all causes Number % all causes Number % all causes Nombre % de toutes Nombre % de toutes Nombre % de toutes Nombre % de toutes les causes les causes les causes les causes Observed deaths, 1982 - Deces observes, 1982 133 887 100.0 66019 49.3 102 429 100.0 38 329 37.4 3 511 2.6 2090 59.5 2 729 2.7 1405 51.5 130376 100.0 63 929 49.0 99 701 100.0 36 925 37.0 4323 3.3 3 125 72.3 3 388 3.4 2 362 69.7 24098 18.5 11 292 46.9 18 199 18.2 8 108 44.6 58 212 44.6 20 134 34.6 49 135 49.3 9 546 19.4 8 141 6.2 3 751 46.1 5413 5.4 2 283 42.2 7 467 5.7 4128 55.3 4 771 4.8 1 923 40.3 Rapp. trimest. statist. sanit. mond., 42 (1989) - 13 - Male - Sexe masculin Female - Sexe Mminin Causes of death - Causes de dt!ces All ages - Taus ages <65 All ages - T ous ages <65 Number % all causes Number % all causes Number % all causes Number % all causes Nombre % de toutes Nombre % de toutes Nombre % de routes Nombre % de toutes les causes les causes les causes les causes A. Argentina (continued} - Argentine (suite} Observed deaths, 1982 - Deces observes, 1982 Complications of pregnancy, childbirth and the puerperium - Complications de la grossesse, de l'accouchement et des suites de couches (630-676) ..... 443 0.4 443 100.0 Congenital anomalies - Anomalies conge- nitales (740-759) ........... 1 557 1.2 1 543 99.1 1 345 1.3 1 330 98.9 Certain conditions originating in the perina- tal period - Certaines affections dont l'origine se situe dans la periode perina- tale (760-779) ............. 5 543 4.3 5 543 100.0 4108 4.1 4108 100.0 All other diseases (remainder of 001-779) - Toutes autres maladies (reste des rubriques 001-779) .......... 9 550 7.3 4834 50.6 8 591 8.6 3 845 44.8 External causes - Causes externes (E800- E999) .................. 11485 8.8 9 578 83.4 4307 4.3 2 976 69.1 Expected deaths according to 1982 United States rates - Dect!Js prevus selon les taux des Etats-Unis en 1982 All causes - Toutes causes (001-E999) 111 847 100.0 49 955 44.7 79 621 100.0 27 776 34.9 Symptoms and ill-defined conditions - Symptclmes. signes et etats morbides mal definis (780-799) ......... 2094 1.9 1 433 68.4 1 356 1.7 840 61.9 Total number of deaths from defined cau- ses - Nombre total de deces dus a des causes definies ............ 109 753 100.0 48 522 44.2 78 265 100.0 26 936 34.4 Infectious and parasitic diseases - Maia- dies infectieuses et parasitaires (001- 139) ............... 1 110 1.0 559 50.4 888 1.1 393 44.3 18 95 Neoplasms - Tumeurs (140-239) ... 24638 22.4 10396 42.2 9 24.2 8818 46.5 Diseases of the circulatory system - Maia- dies de l'appareil circulatoire (390- 37 49 459) .................. 49456 45.1 15 944 32.2 6 47.9 6 797 18.1 Diseases of the respiratory system - Ma- ladies de l'appareil respiratoire (460- 452 519) ........... 7 801 7.1 2 031 26.0 4 5.8 1 223 27.0 Diseases of the digestive system - Maia- dies de l'appareil digestif (520-579) . 4 271 3.9 2407 56.4 3032 3.9 1 307 43.1 Complications of pregnancy, childbirth and the puerperium - Complications de la grossesse, de l'accouchement et des suites de couches (630-676) ..... 32 0.0 32 100.0 Congenital anomalies - Anomalies conge- nitales (740-759) ........... 1 220 1. 1 1 183 97.0 1 067 1.4 1027 96.3 Certain conditions originating in the perina- tal period - Certaines affections dont l'origine se situe dans la periode perina- 166 tale (760-779) ............. 2 207 2.0 2207 100.0 7 2.1 1 667 100.0 All other diseases (remainder of 001-7 79) - Toutes autres maladies (reste des 2 16 rubriques 001-779) .......... 6 394 5.8 2 747 43.0 6 111 7.8 5 35.4 External causes - Causes externes (E800- E999) .................. 12 656 11.5 11048 87.3 4489 5.7 3 507 78.1 B. MEXICO - MEXIOUE Observed deaths. 1982 - Deces observes, 1982 All causes - Toutes causes (001-E999) 232 691 100.0 164416 70.7 173 795 100.0 103911 60.0 Symptoms and ill-defined conditions - Symptclmes. signes et etats morbides mal definis (780-799) ......... 11 823 5.1 6949 58.8 11 173 6.4 5 793 51.8 Total number of deaths from defined cau- ses - Nombre total de deces dus a des causes definies ....... 220868 100.0 157 467 71.3 162 622 100.0 98 117 60.3 Infectious and parasitic diseases - Maia- dies infectieuses et parasitaires (001- 139) ............... 27 309 12.4 22 830 85.6 23 727 14.6 19446 81.2 Neoplasms - Tumeurs (140-239) ... 13 929 6.3 6604 47.4 16 913 10.4 9425 55.7 Wld hlrh statist. quan., 42 ( 1989) Table iv {continued) Causes of death - Causes de d<lc~s B. Mexico (continued) - Mexique (suite) Diseases of the circulatory system - Mala- dies de l'appareil circulatoire (390- 459) ....... . Diseases of the respiratory system - Ma- ladies de I' appareil respiratoire (460- 519) ........ . Diseases of the digestive system - Mala- dies de l'appareil digestif (520-579) Complications of pregnancy, childbirth and the puerperium - Complications de la grossesse, de l'accouchement et des suites de couches (630-676) ..... Congenital anomalies - Anomalies conge- nitales (740-759) .......... . Certain conditions originating in the perina- tal period - Certaines affections dont I' origine se situe dans la periode perina- tale (760-779) . . . . . All other diseases (remainder of 001-779) - T outes autres maladies (reste des rubriques 001-779) . . . . . . . . . . External causes - Causes externes (E800- E999) ............ . All causes - Toutes causes (001-E999) Symptoms and ill-defined conditions - Symptomes, signes et etats morbides mal definis (780-799) . . . . . . . . . Total number of deaths from defined cau- ses - Nombre total de deces dus a des causes definies . . . . . . . . Infectious and parasitic diseases - Mala- dies infectieuses et parasitaires (001- 13~ ················ Neoplasms - Tumeurs (140-239) ... Diseases of the circulatory system - Mala- dies de l'appareil circulatoire (390- 459) ................ . Diseases of the respiratory system - Ma- ladies de I' appareil respiratoire (460- 519) ....... . Diseases of the digestive system - Mala- dies de l'appareil digestif (520-579) . Complications of pregnancy, childbirth and the puerperium - Complications de la grossesse, de l'accouchement et des suites de couches (630-676) ..... Congenital anomalies - Anomalies conge- nitales (740-759) .... Certain conditions originating in the perina- tal period - Certaines affections dont I' origine se situe dans la periode perina- tale (760-779) . . . . . . . . . . . . . All other diseases (remainder of 001-779) - Toutes autres maladies (reste des rubriques 001-779) ......... . External causes - Causes externes (E800- E999) . . . . . ......... . - 14 - Male - Sexe masculin All ages - Taus ages Number Nombre % all causes % de toutes les causes Number Nombre <65 % all causes % de toutes les causes Tableau iv {suite) Female - Sexe f6minin All ages - Taus ages Number Nombre % all causes % de toutes les causes Number Nombre <65 % all causes % de toutes les causes Observed deaths, 1982 - Deces observes, 1982 34619 27 183 21 054 3480 13 188 27 928 52 177 15.7 12.3 9.5 1.6 6.0 12.6 23.6 14034 17 917 14 751 3 464 13 188 16 854 47 825 40.5 65.9 70.1 99.5 100.0 60.3 91.7 35 223 22 868 11 122 2 148 2 995 9 165 26 572 11 889 21.7 14.1 6.8 1.3 1.8 5.6 16.3 7.3 11406 14 253 5 983 2 148 2 977 9 165 13 508 9 805 32.4 62.3 53.8 100.0 99.4 100.0 50.8 82.5 Expected deaths according to 1982 United States rates - Deces prevus selon les taux des Etats-Unis en 1982 160012 4 368 155 644 1 725 30492 59 797 9 915 5 681 3904 7 491 9 193 27 446 100.0 2.7 100.0 1. 1 19.6 38.4 6.4 3.6 2.5 4.8 5.9 17.6 88 994 3 610 85 384 1 092 14305 21 221 3 259 3 555 3862 7 491 4988 25 611 55.6 106 867 82.6 2 818 54.9 104049 63.3 1 355 46.9 23 529 35.5 42 739 32.9 5 681 62.6 3 882 81 98.9 3 407 100.0 5 662 54.3 8213 93.3 9 500 100.0 2.6 100.0 1.3 22.6 41.1 5.5 3.7 0.1 3.3 5.4 7.9 9.1 50 816 2 261 48 555 820 12 647 9460 2 111 2020 81 3 365 5 662 3 948 8441 47.6 80.2 46.7 60.5 53.8 22.1 37.2 52.0 100.0 98.8 100.0 48.1 88.9 a Discrepancies with figures in previous tables are due to rounding - Les chiffres peuvent presenter des ecarts per rapport t. ceux des tableaux precedents du fait qu'ils ont ete arrondis. Source PAHO technical data base - Base de donn6es techniques OPS. Preventable mortality, empirically defined, is analysed using the standardized mortality ratio for all ages (SMR), the SMR for deaths occurring before age 65, and the ratio of observed and expected years of potential life lost (RYPLL). Mortality data from Argentina and Mexico are used for illustration. To assess progress made, expected Rapp. trimest. statist. sanit. mond .. 42 ( 1989) - 15 - deaths are computed using data from the country's own past; to show challenges still ahead, expected deaths are computed using the rates of the United States of America. Comparisons with the past are made for mor- tality from all causes only; to assess what lies ahead, cause-specific mortality is also looked at. It is shown that for the data at hand the RYPLL is the indicator most sensitive to discrepancies between number of deaths observed and those expected, and that it provides infor- mation on inequalities beyond and in addition to that obtained from analysing crude and age-adjusted mortal- ity rates and cause-specific proportional mortality. It is hoped that countries and areas will use these procedures for their own benefit to assess their current health situa- tion against their own past and also against a suitable future reference, to assist them in priority setting and surveillance and evaluation of their health-services system. RESUME La mortalite evitable: indicateur ou objectif? Applications dans les pays en developpement La mortalite evitable, definie empiriquement, est analy- see en utilisant le taux comparatif de mortalite pour tous les ilges (standard monality ratio - SMR), le SMR pour les deces intervenant avant l'ilge de 65 ans et le rapport des annees de vie potentielle perdues observees et pre- vues (ratio of observed and expected years of potential life lost - RYPLL). Des statistiques de la mortalite concernant I' Argentine et le Mexique sont presentees a titre d'illustration. Pour evaluer les progres realises, on calcule les deces prevus a partir des statistiques passees du pays en question; pour montrer les difficultes qui restent a resoudre, on les calcule a partir des taux des Etats-Unis d' Amerique. Les comparaisons avec les don- nees passees ne sont faites que pour la mortalite de toutes causes; pour evaluer les perspectives d'avenir, on examine egalement la mortalite par cause. On montre qu'avec les donnees disponibles. le RYPLL est l'indica- teur le plus sensible aux ecarts entre le nombre de deces observes et le nombre de deces prevus et qu'il fournit des informations sur les inegalites qui viennent comple- ter les informations obtenues en analysant les taux de mortalite bruts et par ilge, ainsi que la mortalite propor- tionnelle par cause. On peut esperer que les pays et les zones utiliseront ces procedures dans leur propre inten~t pour comparer leur situation sanitaire actuelle a leur situation passee et egalement a une reference future appropriee, ce qui les aidera a fixer des priorites et a assurer la surveillance et I' evaluation de leur systeme de services de sante. REFERENCES - REFERENCES 1. NAJERA. E. Perspectives and prospects. In: Buck, C. et al. (eds), The challenge of epidemiology-issues and selected readings. Washington, D.C., Pan American Health Organization, 1988. (p. 974). (Scientific Publication No. 505). 2. WOOLSEY. T. Toward an index of preventable mor- tality. Vital and health statistics. Hyattsville, Md, US Department of Health and Human Services. 1981. (Series 2, No. 85). 3. SHAPIRO. S. A tool for health planners. American journal of public health, 67(9): 816-817 (1977). 4. ROTHMAN. K.J. Modern epidemiology. Boston/ Toronto, Little, Brown & Co., 1986. (pp. 45-49). 5. KESSNER. D.M. ET AL. Assessing health quality: the case for tracers. New England journal of medicine, 288(4): 189-194 (1973). 6. AUTSTEIN, D.D. ET AL. Measuring the quality of medi- cal care-a clinical method, New England journal of medicine, 294(11): 582-588 (1976). 7. ISRAEL. A.A. ET AL. Analytical potential for multiple cause-of-death data. American journal of epidemio- logy, 124: 161-179 (1986). 8. P AHO /WHO. Regional meeting on guidelines and procedures for mortality analysis. Epidemiological bulletin, 9(2) : 3-6 ( 1988). 9. BECKER. A.A. ET AL. Health profiles, Brazil, 1984. Epi- demiological bulletin 9(2): 6-12 ( 1988). 10. GIACOMINI. H.F. ET AL. Health profiles, Argentina, 1980-1982. Epidemiological bulletin, 9(3): 2-9 (1988). 11. PAN AMERICAN HEAL TH ORGANIZATION. Health condi- tions in the Americas, 1981-1984. Washington, D.C., PAHO, 1986. (Scientific Publication No. 500). 12. BEHM. H. [Infant monality and standard of living). Santiago, University of Chile, 1962. (In Spanish). Wld hlth starist. quart., 42 ( 1989) BEHM. H. [Mortalite infantile et niveau de vie). Santiago, Universite du Chili, 1962. (En espa- gnol). 13. TAUCHER. E. [Chile: mortality from 1955 to 1975- trends and causes). Santiago, CELADE, 1978. (Series A, No. 162). (In Spanish). TAUCHER. E. [Chili: monalite de 1955 a 19 75 - ten- dances et causes]. Santiago, CELADE, 1978. (Serie A, N° 162). (En espagnol). 14. BECKER. A.A. ET AL. Years of potential life lost-Brazil, 1980. Epidemiological bulletin, 5(5): 3-7 (1984). 15. BucK. C. & BULL. S. Preventable causes of death ver- sus infant mortality as an indicator of the quality of health services. International journal of health ser- vices, 16(4): 553-563 ( 1986). 16. BucK. C. Beyond Lalonde: creating health. Canadian journal of public health, 16 (Supplement 1) : May/June (1985). 17. LALONDE. M. A new perspective on the health of Canadians: a working document. Ottawa, Depart- ment of Health and Welfare, 1974. 18. DEVER. G.E.A. Epidemiological model for health pol- icy analysis. Social indicators research, 2: 453-466 (1976). 19. ROSE. G. Sick individuals and sick populations. Inter- national journal of epidemiology, 14: 32-38 (1985). 20. HAENSZEL. W.A. Standardized rate for mortality de- fined in units of lost years of life. American journal of public health, 40: 17-26 (1950). 21. CENTERS FOR DISEASE CONTROL. Premature mortality in the United States: public health issues in the use of years of potential life lost. Monality and morbidity weekly repon, 35 (2S) : supplement ( 1986). - 16 - THE ASSESSMENT OF PREVENTABLE INFANT AND CHILD DEATHS IN DEVELOPING COUNTRIES: SOME APPLICATIONS OF A NEW INDEX Stan D'Souzaa Those concerned with health policy in developing coun- tries feel a growing need for indicators that can assess the level of "preventable deaths" and provide simple measures of the efficacy of health programmes aimed at reducing such deaths. This is particularly important in view of the problems of scarcity of resources and choice of strategies facing the developing world today (1). The need for sorting out priorities in the area of health inter- ventions has assumed increasing importance with the "structural adjustments" recently imposed on poor countries as a prerequisite for financial assistance (2). Approaches to the development of a preventable death index (PDI). as well as background material regarding various measures used, have been described in detail elsewhere.b It is demonstrated how a statistical trans- formation of the infant mortality rate (IMR) can be utilized to define a new infant mortality index (IMI). In this article the use of the IMI rather than the IMR for health policy is stressed. Applications to developing countries in Asia and Africa are provided to indicate the usefulness of such an index. The experience of developed countries at early stages of their development is of great value in the understanding of mortality transition, as well as the relationship be- tween overall death rates and cause-of-death struc- ture (3). However, improvements in health technology, and the complex social and cultural settings of low- income countries, impose caution in the use of such experience. The mortality declines registered in develop- ing countries such as China, Cuba and Sri Lanka provide important material for the selection of standards. With- in-country mortality differentials, whether based on socioeconomic, ethnic or sex differences, can also serve as appropriate frameworks to determine preventable deaths. The experience of Matlab, a rural field station of the International Centre for Diarrhoeal Disease Research, Bangladesh (ICDDR,B), is important for documenting vital events with a degree of accuracy not otherwise available in the region. The data, in spite of some limita- tions, provide valuable insights into the cause-of-death structure as well as the linkage with mortality levels and the implications of social prejudices resulting, for exam- ple, in higher female mortality (4). The study by Walsh & Warren (5) provides criteria for assessing which causes of death might be considered preventable in terms of establishing country priorities for mortality reduction. They included the following fac- tors: • prevalence/incidence of the disease; • morbidity or severity of disability caused; • risk of mortality; • feasibility of control (including relative efficacy and cost of intervention). a Senior Demographer, United Nations Development Programme. Cotonou, Benin. b D'Souza, S. Measures of preventable deaths in developing coun- tries: some methodological issues and approaches. Paper presented at the International Union for the Scientific Study of Population, seminar on social and biological correlates of mortality, Tokyo, 24-27 November 1984. Clearly, there are few developing countries where data on all these factors are available. However, qualitative approaches are possible using as a standard the expe- rience of the cause-of-death structure prevailing in coun- tries where such data do exist. Historical data sets can also fulfil this role. In Table 1, the major infectious dis- eases endemic in the developing world have been clas- sified into three groups of priorities for disease control based on prevalence of the disease, mortality, morbidity and feasibility of control. Quantitative data on which these classifications have been constructed are to be found in the paper by Walsh & Warren (5). Some epidemiological measures The assessment of the number of preventable deaths through health programmes can be made by small-area controlled studies. Estimates of the deaths averted can be obtained by comparing results against a baseline that includes both treatment and control areas. The use of control areas raises ethical problems in the design of mortality studies. Health care with known and feasible methods of averting deaths should not be deliberately withheld from a population unless there are justifiable constraints of a logistic and financial nature. Alternative- ly, progress in a study area can be assessed by com- paring results with its own baseline and with national population results if available, and the number of deaths averted and person-years of life saved can thus be esti- mated (6). The indicator "person-years of life saved" gives substantial weight to saving children's lives. One may also consider an index which takes into account an age-discounting factor leading to the notion of the dis- counted person-years of life. In this case, a discounting factor is applied that decreases the input of each person- year with time. A measure of the number of "healthy days of life lost" has been developed to assess quantitatively the relative importance of different diseases in developing coun- tries. The impact of the disease in a community is meas- ured by the number of healthy days of life which are lost through illness, disability and death as a consequence of disease (7). A comparison of the number of healthy days saved by various health interventions and their costs provides criteria for the determination of which causes are, in fact, preventable within a realistic set of priorities and the capability of a particular country. The disadvantage of these epidemiological measures as a policy tool is that they can only be developed within the context of small-area studies. Even in the case of the Matlab information system, one of the best in the devel- oping world, data on costs of interventions are usually not available (8). Some demographic indicators used Monality levels The utilization of the neonatal, infant and child ( 1-4 years) mortality rate and the crude death rate as indicators of Rapp. trimest. statist. sanit. mond., 42 (19891 - 17 - TABLE 1. PRIORITIES FOR DISEASE CONTROL IN THE DEVELOPING WORLD BASED ON PREVALENCE, MORTALITY, MORBIDITY AND FEASIBILITY OF CONTROL TABLEAU 1. DEGRES DE PRIORITE DE LA LUTTE CONTRE LA MALADIE DANS LES PAYS EN DEVELOPPEMENT EN FONCTION DE LA PREVALENCE, DE LA MORTALITE, DE LA MORBIDITE ET DES CARACTERISTIQUES DE LA LUTTE Category f infection - Categorie/infection 1. High - Elevee Diarrhoeal diseases - Maladies diarrheiques . Measles - Rougeole . . . . . . . Malaria - Paludisme . . . . . . . . . Whooping cough - Coqueluche Schistosomiasis - Schistosomiase Neonatal tetanus - Tetanos neonatal 2. Medium - Moyenne Respiratory infections - Infections des voies respiratoires . . . . . Poliomyelitis - Poliomyelite Tuberculosis - Tuberculose Onchocerciasis - Onchocercose Meningitis - Meningite . . . . . . Typhoid fever - Fievre typhorde Hookworm - Ankylostomiase Malnutrition . . . . . . . . . . . 3. Low - Faible South American trypanosomiasis (Chagas' disease) - Trypanosomiase americaine (maladie de Chagas) ...................... . African trypanosomiasis Trypanosomiase africaine ........ . Leprosy - Lepre . . . . . Ascariasis - Ascaridiase Diphtheria - Diphterie . . Leishmaniasis - Leishmaniose Giardiasis - Giardiase Filariasis - Filariose Dengue ...... . Source: Reference /7} - Reference /7}. Reasons for assignment to this category - Raisons de I' affectation ~ cette categorie High prevalence, high mortality, high morbidity, effective control possible - Forte prevalence, forte mortalite, forte morbidite, lutte efficace possible High prevalence, high mortality, no effective control - Forte prevalence. forte mortalite, pas de lutte efficace High prevalence, low mortality. no effective control - Forte prevalence, faible mortalite, lutte efficace High prevalence, high mortality, control difficult - Forte prevalence. forte mortalite, lutte difficile Medium prevalence, low mortality, high morbidity, control difficult - Prevalence moyenne. faible mortalite, forte morbidite, lutte difficile Medium prevalence, high mortality, control difficult - Prevalence moyenne, forte mortalite, lutte difficile Medium prevalence, high mortality, control difficult - Prevalence moyenne. forte mortalite, lutte difficile High prevalence, low mortality, control difficult - Forte prevalence, faible mortalite, lutte difficile High prevalence, high morbidity, control complex - Forte prevalence. forte morbidite, lutte complexe Control difficult - Lutte difficile Low prevalence, control difficult- Prevalence faible, lutte difficile Control difficult - Lutte difficile Low mortality, low morbidity, control difficult - Faible mortalite. faible morbidite, lutte difficile Low mortality, low morbidity - Faible mortalite, faible morbidite Control difficult - Lutte difficile changes in mortality levels, can be illustrated from Matlab data (Fig. 1). The data cover the period 1966- 1984 during which time several major events occurred in the country. The two major peaks in the infant mortality rate correspond to periods of stress during the liberation struggle of 1970-1971 and the famine during 1974- 1975. The neonatal mortality rate is generally high, between 60 and 90 per 1 OOO, indicating high levels of tetanus and deaths classified as "certain diseases of infancy'·. It is interesting to note that the peak in the postneonatal rate suggests a large number of deaths due to malnutrition during the period of food shortage of 1974-1975. The child mortality rate also shows two peaks and is thus a more sensitive mortality indicator than the crude death rate which moves more slowly. In most developing countries, medical validation of cause of death is not practicable. Lay reporting systems have been used to obtain information on causes of death (9). Within the Matlab area, data on causes of death have been collected through non-medical person- nel. Problems related to classification have been found to be quite important. Three sources of classification errors may be mentioned - respondent bias, inter- viewer bias, and local versus medical perception of dis- ease (10). Conceptual problems are also inherent in the classifica- tion of cause-of-death data. Death is rarely monocausal and the selection of one underlying cause may not always be easy. Diseases with severe complications, such as malnutrition, measles and malaria, may be over- looked in the classification process. The use of the con- cept "cost per death averted" may be fraught with prob- lems. Averting a death from a particular cause may nev- ertheless result in an individual dying of another cause within a short period of time. For example, oral rehydra- tion therapy is known to be a low-cost treatment for dehydration and deaths from watery diarrhoea. To con- Figs 2A&28 show cause-of-death patterns of child mortality from selected diseases in Matlab from 1975 to 1981 . Apart from the higher mortality rates during the difficult period of 1975, it is worth noting the uniform pattern of diseases as well as the consistently higher female mortality in most categories. This sex differential in mortality would appear to be largely prevent- able (4). Wld hlth statist. quan., 42 ( 1989) - 18 - FIG.1 INFANT (NEONATAL AND POSTNEONATAL) MORTALITY RATE, CHILD (1-4 VEAR) DEATH RATE AND CRUDE DEATH RATE IN MATLAB (BANGLADESH), 1966-1982 TAUX DE MORTALITIE INFANTll:E (NIEONATALE ET POST-NIEONATALE), TAUX DE MORTALITE POST-INFANTILE (1-4 ANS) ET TAUX BRUT DE MORTALITIE A. MATLAB (BANGLADESH), 1966-1982 180 160 140 ~ 120 ::, 0 Q. " i 100 I § ~ 80 a; Q. " la a:: 60 40 "· I \ v /\ / \ I \ ) \ \ ....... \.-•/ ·, - . ,,,. "'" /A. /, ............. __ .,,.. , ..... ..,, ' 20 .···. .... "--------- ····················· ········ ........................... 1966-67 68-69 10-11 72.73 74 75 76 n 78 79 ao 81 1982 Years - Ann6es Source: Reference (4) - Reference (4). - • - • Infant mortality - Mortalit~ infantile --- Neonatal monality - Monalittl n6onatale - Postneonatal monality - MonaliM post-neonatale ___ 1-4 years/ans • ••• ... •, , ... Crude death - Mortalitf§ brute elude that a death due to diarrhoea which has been averted is necessarily a life saved may be incorrect. Mal- nourished children at high risk of death due to diarrhoea are also at high risk of dying from a number of other causes. Infant mortality rate as a health-policy indicator The infant mortality rate (IMR) has been extensively used as an indicator of the overall mortality level for a country. It is sensitive to changes in prevailing levels of public health. Modern medical technology-in particular immu- nization-can reduce infant mortality rates within a rela- tively short period of time. Estimates and projections of the IMR from 1950 to 2025 are routinely prepared by the United Nations Population Division. According to this source, the IMR in less devel- oped countries fell from over 150 per 1 OOO in 1950- 1955 to about 100 per 1 OOO in 1975-1980. During the correspondig periods, the rates in the areas of highest infant mortality-Africa and South Asia-fell from 180 per 1 OOO to around 120 per 1 OOO. Developed country rates for the same periods were about 55 and 19 per 1 OOO respectively. One would expect greater declines in mortality for developing than developed countries since the control of infectious diseases is clearly more easily achieved than a reduction in mortality from congenital defects. There are, however, two major limitations with the use of the IMR as an overall mortality indicator for guiding health policy. First, as just noted, falls in the IMR are not evenly distributed in terms of the difficulty of mortality control. A drop of 10 points from an initial IMR level of 180 per 1 OOO is much more easily achieved through health intervention than an equivalent drop from an initial IMR level of 55 per 1 000. Furthermore, the linkage of the IMR to a particular cause-of-death structure is not clear. A new index of preventable infant and child deaths The necessary link between mortality levels and cause- of-death structure has been the subject of an earlier study by Bourgeois-Pichat (11). He considered causes of death under two broad classes-endogenous and Rapp. trimest. statist. sanit. mond., 42 (1989) - 19 - FIG.2A SELECTED DEATH RATES, BY SEX AND CAUSE, FEMALES, 1-4 YEARS, MATLAB (BANGLADESH), 1975-1981 TAUX DE MORTALld CHOISIS PAR SEXE ET PAR CAUSE, SEXE FEMININ, 1-4 ANS, MATLAB (BANGLADESH), 1975-1981 40 38 36 34 ., .., 32 c c !:. ., .,, ::, 30 j E ::, .. .. 28 c .. ! ., .,, 26 .. ~ ... .. E 24 .. 'E ., § 22 ~ Dysentery - Dysenterie !:i &. .. 20 .., u .., 0 I ! 18 al Cl .. c 16 I ::, Q. 0 Q. 14 :. ., > .,, E 12 § ~ :;; 10 Q. Dropsy - Hydropsie .. .c: .; " 8 0 Respiratory diseases - Maladies respiratoires 6 ~Other - Autres 4 2 Drowning - Noyade 0 Year- 1975 Ann6e ~~:::~~; - 17 293 1976 15947 1977 14652 1978 10652 1979 10413 1980 10651 1981 11 570 Source : Reference (12) - R616rence (12). exogenous. Using the metaphors of "hard rock" and "soft rock" respectively, Bourgeois-Pichat argued that the exogenous causes of death (external) could be con- trolled by public health measures, immunization and anti- biotics. This would be the metaphorical equivalent of erosion of the soft rock. The declining proportion of deaths from infectious diseases to deaths from all causes provided evidence of this phenomenon. Evi- dence on the epidemiological transition in China (12) Wld hlth statist. quan., 42 (1989) reinforces this hypothesis. However, there are growing indications of the need for social development to ensure continued falls in mortality levels, particularly in the con- trol of such diseases as diarrhoea. It is clear from the preceding sections that several indi- cators are required to estimate preventable deaths. A single indicator would be totally insufficient to describe the complex process as a result of which deaths can be - 20 - FIG.2B SELECTED DEATH RATES, BY SEX AND CAUSE, MALES, 1-4 YEARS, MATLAB (BANGLADESH), 1975-1981 TAUX DE MORTALITE CHOISIS PAR SEXE ET PAR CAUSE, SEXE MASCULIN, 1-4 ANS, MATLAB (BANGLADESH), 1975-1981 30 " 28 .., c: c: ~ .. 26 "' :, .!!! e :, 24 ., .. c: .. ! 22 " "' .. .., "' co 20 .. E ~ " § 18 :; 16 0 0. .. ... " ~ 14 ! "' Dysentery - Dysenterie .. 12 "' .. c: 0 ·1; "3 10 0. 0 Diarrhoe~e~ 0. Iii " > "' e § ~ a; 0. .. .t:: ;; ~ 8 6 4 2 Year - An nee Number Nombre ---- 1975 18166 1976 16 910 19n 15 674 Source : Reference ( 12) - Reference ( 12). classified as preventable. In this article, a logistic trans- formation is proposed in order to distribute the degree of difficulty in achieving further mortality declines more uni- formly (13). The infant mortality index (IMI) is defined as: IMI = 100 - A (In {~,qo) + a}1 with a scale from O (IMR = 5 per 1 000) to 100 (IMR = 300 per 1 000). and where A and B are con- stants defined by the high and low points of the scale. It should be noted immediately that values of the IMR higher than 300, which are known to have occurred in the past, would appear after transformation as an IMI greater than 100; IMR values lower than 5 would become negative IMI values. Under these assumptions A= -22.4 and B = 0.8473. First approach In this approach it is assumed that data on causes of death and cost-feasibility of health interventions are not 1978 11 447 Fever - Fievre Other - Autres Drowning Noyade 1979 11 213 Dropsy - ydrops,, I Respiratory diseases - Maladies respiratoires 1980 11 433 1981 12442 available. Since a transformation of mortality measures can better portray the difficulty involved in preventive disease control, an appropriate first approximation of a PDI is taken as : PDI (study population) = IMI (study population) - (standard population). If we assume that in the case of infant mortality a value of the IMI = 30 reflects the beginning of the hard rock of mortality (i.e. this value is taken as an indication of the elimination of most preventable diseases of infectious and parasitic origin). then: PDI (study population) = IMI (study population) - 30. The values of the PDI would range from O to 70 for developing countries (for which the minimum IMI value is set at 30). A set of PDI values is to be found in Table 2. Rapp. trimest. statist. sanit. mond., 42 (1989) - 21 - TABLE 2. CORRESPONDENCE BETWEEN LEVELS OF INFANT MORTALITY AND THE INFANT MORTALITY INDEX (IMI) AND PREVENTABLE DEATH INDEX (PDI) TABLEAU 2. CORRESPONDANCE ENTRE LES NIVEAUX DE MORTALITE INFANTILE, L'INDICE DE MORTALITE INFANTILE (IMIJ ET L'INDICE DES DECES EVITABLES (PDIJ 1 Qo a rounded 1 Qo a arrondi 0.300 0.215 0.150 0.100 0.070 0.045 0.030 0.020 0.010 0.008 0.005 1q0 value 1Qo reel (0.300) (0.215) (0.150) (0.100) (0.068) (0.044) (0.029) (0.019) (0.012) (0.077) (0.005) IMI 100 90 80 70 60 50 40 30 20 10 0 POi 70 60 50 40 30 20 10 0 ' Probability of dying before age 1 - Probabilite de mourir avant rage d·un an. Negative values of the POi could mean erosion of the hard rock and can be utilized for measures of mortality control in developed countries. Clearly other standards can be selected depending on what may be considered feasible for particular regions. Second approach In this case it is assumed that data on mortality patterns are available, at least indirect estimates, and that there is information on the cost-feasibility structure for certain health interventions. In this case, we define: M; death rate due to cause i M death rate due to all causes CFi cost-feasibility factor (taking values be- tween O and 1) for preventing cause of death i. This factor may be a complex function measuring such factors as cost. feasibility, acceptability, etc. of mortality control methods. A value of CFi = 0 corresponds to the situation of prohibitive costs and logistical problems for basic preventive measures; when low-cost feasible ap- proaches to mortality prevention are available, CFi = 1. We may then express the index as: POi (study population) = IMI · ~ (% · CF,) I A simple example may illustrate the construction of the POi. Let us assume that there are only two types of causes of death: type a due to infectious and parasitic diseases and type b due to congenital defects. Also let us assume that: ~ CFa = IMI · l,w M . 1 + ~ OJ =IMI·~ i.e. the IMI multiplied by the proportion of all deaths due to infectious and parasitic diseases. In this simplified case, the POi definition used in the second approach can be related to that given in the first approach and the index may be rewritten as: IMI - POi = IMI - IMI -~ =IMl(1-~) =IMI-~ the proportion of the IMI due to congenital defects (hard rock level) = IMI (standard population) Wld hlth statist. quart .. 42 ( 1989) Bourgeois-Picha! classification - Classification de Bourgeois-Picha! Soft rock - Roche tendre (large percentage of deaths of infectious origin - pourcentage important de deces d'origine infectieuse) Intermediate rock - Roche intermediaire (further declines possible through social development - nouvelles diminutions possibles grace au developpement social) Hard rock - Roche dure (large percentage of deaths due to congenital anomalies and certain perinatal causes - pourcentage important de deces dus aux anomalies congenitales et a certaines causes perinatales) i.e. POi (approach 2) = IMI - IMI (standard population) = POi (approach 1) If one does not have data on cause-of-death structure, one may utilize the indirect estimates of Mi/M obtained from the cause-of-death tabulations of Preston et al. (3). When detailed CFi information is not available, one can utilize, with suitable adjustments, values provided by Walsh & Warren (5). Table 2 presents values of 1q0 (probability of dying before age 1) and corresponding values of the IMI. A line to denote a possible distinction between the hard and soft rock of mortality is set below a value of 1 q0 = 0.030. This implicitly assumes that for an infant mortality rate below 30 per 1 OOO live births, there are few deaths of infectious and parasitic origin. A second dotted line has been set at an infant mortality rate of 100 per 1 OOO, above which substantial mortality declines are unlikely without some improvements in social struc- tures and environmental health. The term "intermediate rock·· has been utilized to describe the cause structure of infant mortality between these two limits. In particular, the control of diarrhoeal diseases is of paramount impor- tance for any further declines in infant mortality. Table 3 presents a transformation of the child mortality measures (4q1), being the probability that a child will die between his or her first and fifth birthdays. Here the high point for 4q 1 has been set at 0. 150 and the low point at 0.002. The division between hard and soft rock is defined as a value of 4q1 = 0.008. Some applications The epidemiological transition in China first began in the cities and then progressed through the rural areas. Apart from infectious diseases, malnutrition and high fertility contributed to the extremely high infant mortality rates. Table 4 provides IMR values from 1950 to 1982. The initial value of 252 per 1 OOO in 1950 had decreased to 200 by the year 1959. However, the gains made prior to the Great Leap Forward of 1958 were to some extent countered by subsequent events including a period of famine, resulting in a sharp rise in infant mortality rates. The IMR rose to 330 per 1 OOO in 1960 before falling back to 193 in 1962. Thereafter the IMR fell rapidly to 100 per 1 OOO in 1972 and was estimated at 45 per - 22 - TABLE 3. VALUES DENOTING LEVELS OF CHILD MORTALITY INDEX (4q 1) AND PREVENTABLE DEATH INDEX TABLEAU 3. VALEURS DENOTANT LES NIVEAUX DE L'INDICE DE MORTALITE POST-INFANTILE (4 q 1) ET DE L'INDICE DES DECES EVITABLES 4 q 1 ' (rounded) Actual 4q 1 value 4 q 1 • (arrondi) 4Q1 (reel) 0.150 (0150) 0.100 (0.100) 0.070 (0.068) 0.045 (0.044) 0.030 (0.029) 0.020 (0.019) 0.010 (0.012) 0.008 (0.008) 0.005 (0.005) 0.003 (0.003) 0.002 (0.002) Index value Valeur de l'indice IMI POi 100 70 90 60 80 50 70 40 60 30 50 20 40 10 30 0 20 10 0 Bourgeois-Picha! classification - Classification de Bourgeois-Picha! Soft rock - Roche tendre (large percentage of deaths of infectious origin - pourcentage important de deces d'origine infectieuse) Hard rock - Roche dure (predominance of congenital anomalies and certain perinatal causes - predominance des anomalies congenitales et de certaines causes perinatales) • This index of child mortality is the probability of dying between the first and fihh birthdays - Cet indice de mortalite post-infantile est la probabilite de mourir entre le premier et le cinquieme anniversaire. TABLE 4. COMPARATIVE INFANT MORTALITY RATE (IMR) AND INFANT MORTALITY INDEX (IMI) VALUES FOR CHINA, 1952-1982 TABLEAU 4. COMPARAISON ENTRE LE TAUX (IMRJ ET L'INDICE (IMIJ DE MORTALITE INFANTILE EN CHINE, 1952-1982 IMR Year - Annee (per 1 OOO live births - pour 1 OOO naissances vivantes) IMI 1950 1952 1954 1955 1957 1959 1960 1962 1964 1965 1967 1969 1970 1972 1974 1975 1977 1979 1980 1982 Source: Reference (14} - Reference (14}. 1 OOO in 1982 (aher adjustment for underreporting). Thus during the decade of the 1950s the IMR fell around 50 points if one excludes the difficult years around 1960. During the next decade the fall in the IMR was about 100 points; and finally, in the decade 1972-1982 the fall in the IMR was again about 50 points. If one now reviews the performance of the IMI as an indicator, one notices that during the 1950s the fall in the IMI was about 6 points only. from about 93 to 87. During the next decade (1962-1972) the fall in the IMI was about 17 points from around 87 to around 70. Finally during the decade 1972-1982 the IMI fell 20 points from 70 to about 50. Thus, in terms of reflecting improve- ments in the health status of the country, the IMI would appear to be more relevant than the IMR. Furthermore, if, as is suggested in the first approach, an IMI of 30 is taken as the standard level for the eradication of most 252 240 225 216 200 252 330 193 180 165 139 122 109 100 87 84 64 57 53 45 94.6 93.1 91.2 90.1 87.9 94.6 103.1 86.9 85.0 82.6 78.0 74.7 71.8 69.6 66.2 65.3 58.7 55.9 54.2 50.3 preventable deaths of infectious and parasitic origin, China now has a POi value of 20. The value of utilizing the IMI instead of the IMR for health policy issues is illustated by two examples. In Table 5 a comparison is made between infant mortality levels in Bangladesh and Sri Lanka at each of the five-year periods, 1950-1955 and 1980-1985, utilizing both the IMR and the IMI. If one uses the IMR as a measure of progress in reducing mortality from preventable deaths, the performance of both countries is roughly similar: a decline of 46.9 points for Bangladesh compared with 50. 1 points for Sri Lanka. If one uses IMI values, how- ever, the experience of the two countries is quite differ- ent. In this case, mortality decline has been nearly twice as great in Sri Lanka as in Bangladesh. Using the IMI instead of the IMR for Africa similarly produces interest- ing results. The IMR would indicate that Mauritius and Rapp. trimest. statist. sanit. mond .. 42 ( 1989) - 23 - TABLE 5. COMPARATIVE INFANT MORTALITY RATE (IMR) AND INFANT MORTALITY INDEX (IMI) VALUES FOR SELECTED COUNTRIES, 1950 AND 1980 TABLEAU 5. COMPARAISON ENTRE LE TAUX (IMRJ ET L'INDICE (IMIJ DE MORTALITE INFANTILE DANS CERTAINS PAYS CHOISIS, 1950 ET 1980 IMR (per 1 OOO live births - IMI Region/country - Region/pays pour 1 OOO naissances vivantes) 1950 1980 Difference 1950 1980 Difference Difference Difference South Asia - Asie m~ridiona/e Bangladesh ......... 179.5 132.6 46.9 84.9 76.8 8.1 Sri Lanka .... . . . . . . . . . 90.7 40.6 50.1 67.2 47.9 19.3 Africa - A frique Mauritius - Maurice 98.0 31.7 66.3 69.1 42.4 26.7 Benin - Benin .... 218.8 148.5 70.3 90.3 79.7 10.6 Source: IMR values are estimates prepared by the United Nations Population Division - Les taux de mortalite infantile ont ete estimes par la Division de la population de !"Organisation des Nations Unies. TABLE 6. TRENDS AND PROJECTIONS OF INFANT MORTALITY RATE (IMR) AND INFANT MORTALITY INDEX (IMI) VALUES FOR AFRICA, 1950-2025 TABLEAU 6. TENDANCES ET PROJECTIONS DU TAUX (IMRJ ET DE L'INDICE (IMIJ DE MORTALITE INFANTILE POUR L'AFRIQUE, 1950-2025 Decline compared with Years - Annees IMR previous periods IMI Difference - Difference Diminution par rapport a la periode precedente 1950-1955 184.3 1955-1960 172.3 1960-1965 160.3 1965-1970 ........ 149.4 1970-1975 ......... 138.5 1975-1980 . . . . . . . . 127.1 1980-1985 116.4 1985-1990 ........ 106.0 1990-1995 ......... 96.2 1995-2000 ......... 87.0 2000-2005 ....... 78.3 2005-2010 ...... 70.2 2010-2015 ....... 62.7 2015-2020 55.8 2020-2025 .......... 49.4 12.0 12.0 11.9 11.9 11.4 10.7 10.4 10.2 9.2 8.7 8.1 7.5 6.9 6.4 85.6 83.7 81.8 79.9 77.9 75.7 73.5 71.1 68.7 66.2 63.6 60.9 58.2 55.4 52.5 1.9 1.9 1.9 2.0 2.2 2.2 2.4 2.4 2.5 2.6 2.7 2.7 2.8 2.9 Source: IMR values have been prepared by the United Nations Population Division - Les taux de mortalite infantile ont ete etablis par la Division de la population de !"Organisation des Nations Unies. Benin have been successful in reducing mortality by roughly equivalent amounts, with even a slight a~~an- tage for Benin. The IMI, however, reveals that Mauritius, with a drop of 26. 7 as against only 10.8 for Benin, has been much more successful in reducing mortality. Table 6 provides projected or estimated values of the IMR for Africa from 1950 to 2025. Corresponding IMI values are also shown. The table indicates that infant mortality has steadily declined and is expected to con- tinue to do so in the future, albeit at a slower rate. At face value, this would indicate that the mortality decline is slowly tapering off. On the other hand, from an exami- nation of the trend in the differences in index values (see extreme right-hand column) one notices that these values slowly increase, suggesting that much greater efforts will be needed to reduce preventable mortality in infancy than what is suggested by trends in the infant mortality rate alone. This is depicted graphically in Fig. 3. Discussion The mortality indices developed in this article for assess- ing preventable deaths have been derived from the prob- Wld hlth statist. quart .. 42 ( 19891 ability-of-death (i.e. q) values since such estimates are readily available from the variety of indirect estimation procedures currently in use. However, a knowledge of cause-of-death patterns is basic to a more precise under- standing of the level of mortality. Studies from small areas can be utilized to validate indirect estimates of the cause-of-death structure. More importantly, such study areas can provide information on the various parameters of relevance to the choice of health strategies, such as efficacy of treatment, cost, etc. The mortality index proposed in this article is not only straightforward to calculate, but has been set on a scale of Oto 100 to provide simple percentiles, which greatly facilitates interpretation. Thus, the so-called hard-rock structure of causes of death is considered to correspond to an IMR of less than 30. The particular setting of the low and high points is, of course, arbitrary. However, available experience in the Matlab area of Bangladesh, as well as the work of Preston et al. (3) demonstrate that levels of mortality are clearly related to the structure of causes of death. The analogy of rock types has been used to describe this transition in disease patterns. The mortality index proposed in this article differs from previous measures of preventable death in that it - 24 - FIG. 3 FIVELYEAR DIFFERENCES IN INFANT MORTALITY RATE (IMR) AND INFANT MORTALITY INDEX (IMI) VALUES, AFRICA, 1950-2025 DIFFERENCES QUINQUENNALES ENTRE LES TAUX (IMR} ET INDICES (IMI} DE MORTALITE INFANTILE EN AFRIQUE, 1950-2025 IMR- 12 IMI-- 11 ~ ...,_. , ........ , ....... ,,_ ,,- °' •="• """"" - La difMrence entre les taux traduit un ralentissement de la baisse de la mortalite 10 9 ., :, 0 .. .c 8 .. ., " c !:! ~ 7 I ., " c !:! ~ 6 ,, ., 5 0 .. .c <( 5 4 Difference in IMI values indicates increasing effort required - 3 La difference entre les indices indique I' effort croissant qui deviant necessaire 2 1950 55 60 65 70 75 80 85 90 95 2000 05 10 15 20 2025 Year - Annee Source: see Table 6, col· ,,,ns 2 and 4 - Voir tableau 6, colonnes 2 et 4. integrates the idea of difficulty of control, and is thus a much wider concept than the notion of mortality per se. The objective of the PDI formulation as given here has been to develop a tool that may be useful in developing countries where reliable data are not easily available. Similarly, PDI values over 50 (corresponding to an IMI over 80) can be taken to imply considerable potential for mortality control in a country. Indices are important because of the relative nature of the information they provide. Falls or rises in the cost-of-living index have a policy impact, even though they do not directly desig- nate what constituent has fallen or risen, or what abso- lute value has been utilized as a baseline. Differences in IMI provide more significant information than the use of the IMR alone. The logistic transformation used to derive the index bet- ter illustrates relative changes in mortality through the spectrum of levels of infant mortality and is thus likely to be a more sensitive policy tool. In fact, standardization procedures are commonplace in demography, although standardization of underlying distributions has been used infrequently. The particular choice of distribution logistic (Cauchy or Gaussian) has been shown to make little difference to the interpretation of the results (13). Logarithmic transformations are frequently used in de- mographic analysis. In this case, multiplicative models become additive, a feature which is clearly of consider- able relevance whenever causes of death are being con- sidered (for example, in the calculation of associated multiple decrement tables), It should also be noted that the particular transformation utilized is basically the logarithm of the odds of dying. In fact, since differences in the IMI are mainly utilized, the transformation corres- ponds to the logarithm of the odds ratio commonly used in epidemiology. Finally, the application of the IMI in several different developing country settings has been demonstrated. The case of China is particularly interesting since the decline in the infant mortality rate and the resulting epi- demiological transition has been well documented. These analyses underline the value of the IMI as a useful policy tool in the assessment of preventable mortality. Rapp. trimesr. statist. sanit. mond .• 42 (1989) - 25 - SUMMARY This article briefly reviews epidemiological and demo- graphic measures used to assess preventable deaths in developing country situations. Problems with the use of the infant mortality rate (IMR) as a health-policy indicator are illustrated. The Matlab experience of rural Bangla- desh is used to develop a new index for preventable infant and child deaths. This index, based on a statistical transformation of usual demographic measures, links mortality levels and cause-of-death structure. The use of this index is illustrated with applications to Asia and Africa. It is claimed that the index exhibits more clearly the difficulty involved in the control of particular causes of death, thus providing important information for health policy makers. RESUME L'evaluation des deces evitables de nourrissons et d'enfants dans les pays en developpement: quelques applications d'un nouvel indice Cet article passe brievement en revue les mesures epi- demiologiques et demographiques utilisees pour evaluer les deces evitables dans le cas des pays en developpe- ment. Les problemes lies a !'utilisation du taux de mor- talite des nourrissons (IMR) comme indicateur pour les politiques de sante sont illustres. L'experience Matlab effectuee dans une zone rurale du Bangladesh sert a ela- borer un nouvel indice pour les deces evitables de nour- rissons et d'enfants. Cet indice, fonde sur une transfor- mation stat1st1que de mesures demographiques usuelles, etablit une correlation entre les niveaux de mor- talite et la structure des causes de deces. L'utilisation de cet in dice est ill us tree par des applications pour I' Asie et I' Afrique. On fait valoir que cet indice montre mieux les problemes que pose la lutte contre certaines causes de deces et qu'il fournit ainsi des informations utiles aux responsables des politiques de sante. REFERENCES - REFERENCES 1. EVANS. J. R. ET AL. Health care in the developing world: problems of scarcity and choice. New England journal of medicine, 305 (19): 1117-1127 (1981). 2. MOSLEY. w. H. &JOLLY, R. Politique de sante etchoix des moyens: la compensation des effets negatifs de l'ajustement economique. In: Cornia, B. et al. (eds), L 'ajustement a visage humain. Paris, Econo- mica, 1987. 3. PRESTON. S. ET AL. Causes of death: life tables for national populations, New York, Seminar Press, 1972. 4. D'SouzA. S & CHEN. L. C. Sex differentials in mortality in rural Bangladesh. Population and development review, 6 (2): 257-270 (1980). 5. WALSH. J. & WARREN. K. Selective primary health care: an interim strategy for disease control in developing countries. Social science and medicine 14C: 145-163 (1980). 6. BERGGREN. W. ET AL. Reduction of mortality in rural Haiti through a primary health care program. New England journal of medicine, 304: 1324-1330 (1981). 7. MORROW. R. H. ET AL. A quantitative method of as- sessing the health impact of different diseases in Wld hlth statist. quan .• 42 ( 1989) less developed countries. International journal of epidemiology, 10: 73-82 ( 1981 ). 8. HORTON. S. & CLAOUIN, P. Cost effectiveness of hos- pital and of ambulance services at the Matlab treat- ment centre. Dhaka, International Centre for Diarr- hoeal Disease Research, Bangladesh, 1982. (ICDDR-B Working Paper No. 25). 9. WORLD HEAL TH ORGANIZATION. Lay reporting of health information. Geneva, WHO, 1978. ORGANISATION MONDI ALE DE LA SANTE. Notification d'informations sanitaires par un personnel non me- dical. Geneve, OMS, 1978. 10. ZIMICKI. S. ET AL. A source book of cause-specific mortality rates, Matlab 19 75-1981. Dhaka, Interna- tional Centre for Diarrhoeal Disease Research, Bangladesh, 1985. 11. BOURGEOIS-PICHAT. J. Essai sur la mortalite «biologi- que» de l'homme. Population, 7 (3): 381-394 (1952). 12. WORLD BANK. China: the health sector. Washington, D.C., World Bank, 1984. (World Bank Country Study). 13. MOSTELLER, F. & TUKEY, J., Data analysis and regres- sion. Reading, Mass., Addison Wesley Publishing Co., 1977. - 26 - EXCESS MORTALITY RATIO WITH REFERENCE TO THE LOWEST AGE-SEX-SPECIFIC DEATH RATES AMONG COUNTRIES Kazuo Uemuraa Mortality data have been used widely for assessing the health status of a population, as they reflect the severest negative aspect of health. Mortality indicators have also been used for determining priorities in health-resources allocation and in health intervention in the community. A number of different indicators and different methods of analysis have been devised for the use of mortality data and applied for various purposes. Frequently-used mortality indicators include simple measures of mortality such as crude death rate and age- specific death rates, and adjusted measures of mortality such as age-standardized (or age-adjusted) death rate and ratio, expectation of life at various ages, and other life-table functions such as the number living at various ages. These indicators are often analysed by causes of death and disaggregated by factors such as sex, geo- graphical areas and socioeconomic groups, in addition to age. One of the main reasons for the frequent use of mortality data, as compared with morbidity data, is the availability of reliable data for comparisons among countries. Death is a distinct, objectively recognized and extremely seri- ous event which makes it much easier to record than the occurrence of illness per se. Complete or nearly com- plete death registration systems have been established in industrialized countries and in some of the more advanced developing countries. Improvements in socioeconomic levels in many coun- tries of the world and the application of increasingly more effective medical technologies and public health measures have resulted in remarkable reductions in mor- tality in developed and developing countries alike. Such a mortality decline has been most pronounced in develop- ing countries in terms of absolute numbers of deaths avoided, as the baseline was extremely high. In indus- trialized countries the decline has been less, but a plateau has not yet been reached in most countries. Moreover, a considerable variation exists among them in the mortal- ity level, even after the differences due to the age com- position of the population are taken into account. This suggests that countries with higher mortality have a potential for reducing it further at least to the level attained by those with lower mortality. A study of mortality patterns in different countries is interesting in itself, but it offers a most useful contrib- ution when it concerns the identification of preventable mortality. The notion of preventability depends on the availability and effectiveness of preventive measures in a particular community or a population group against death from a particular disease. It should apply, how- ever, to all countries and all age groups, though the extent of preventability varies among them. Mortality indicators therefore continue to provide a useful guide to • Director. Division of Epidemiological Surveillance and Health Situa- tion and Trend Assessment. World Health Organization. Geneva. b See Geographical variations in mortality from conditions amenable to medical intervention in Europe: the European Community atlas of avoidable deaths, p. 42 of this issue. health planners and research workers in spite of their failure to offer full evidence of the quality of life. This article proposes a new indicator of excess mortality with reference to the lowest age-sex-specific death rates recorded among countries in the world. Two sets of the lowest rates are considered, namely one compris- ing the lowest age-sex-specific rates recorded in each calendar year (the current minimum) and the other com- prising the lowest rates ever recorded since 1950 (the historical minimum). The lowest death rates are applied to the age-sex composition of the population of a given country for a given calendar year to obtain the number of deaths which would have been expected under the low- est mortality pattern; the expected number is then sub- tracted from the actual number of deaths recorded in the country during the calendar year to yield the "excess". The indicator is finally calculated by taking the ratio of the excess to the expected minimum. Various uses of the indicator are described in this article. Recent developments in mortality reduction in most developed countries have been so extensive that life- table estimates currently indicate as high as 90% survi- val at age 5 7 years for males and 66 years for females (e.g. the Abridged Life Table, Japan, 1987), as com- pared with 39 years and 4 7 years, respectively, a quarter of a century ago (The Eleventh Life Table, Japan, 1960). Progress in survival is most likely to continue. Under such circumstances, a death before age 65, or even 70, may be regarded as premature, even if not every such death can be prevented in the foreseeable future. It is for this reason that the age group under 65 is particularly highlighted in this article when estimating excess mor- tality. Comprehensive analysis of mortality should cover causes of death because of their obvious relevance to the prevention and treatment of particular disease con- ditions. Nevertheless, as the first broad indication of the magnitude of mortality, the study of all causes combined has an important role to play. Moreover, for high age groups, the concept of a single, underlying cause of death frequently does not apply, and, hence, cause-of- death statistics traditionally compiled by national statis- tical offices are of little use. The analysis in this article is therefore made on the basis of all causes combined, though major causes of death are also considered for age groups under 65. Approaches to the measurement of preventable deaths Broadly speaking, two approaches have been used by various researchers to identify and measure preventable deaths. The first is based on consideration of the possi- bility for prevention and treatment of specific diseases. Work in this direction covers maternal mortality (1) and lists of "unnecessary" untimely deaths from conditions which could be assumed preventable or manageable (2- 8). A recent project of the European Community has resulted in an atlas of avoidable mortalityb from a number Rapp. trimest. statist. sanit. mond., 42 ( 19891 - 27 - of diseases which are mostly avoidable either by timely and appropriate health-service intervention or through predominantly non health-service intervention (9). The second approach is more statistical. As there is often a considerable degree of heterogeneity in mortality risks among population groups, the mortality level achieved in a low-mortality population might be con- sidered as a feasible target level. Woolsey (10) reviewed work done according to this approach. In relation to a set of the target death rates which have been identified, an appropriate summary index of excess deaths in a given population might be computed. For example, Woolsey proposed an index to be computed as a weighted aver- age of ratios of targeted death rates to observed age- specific or age-cause-specific death rates, or as a simple ratio of preventable to observed deaths. The present article follows this approach in the measurement of excess mortality. The results from an earlier study of excess mortality were given by Uemura.c The concept of the lowest death rates as a target for mortality reduction was first applied to mortality differ- entials within a country. Waaler considered, as the "best" group in Norway, females in the highest social class in the district of Sogan og Fjordane (11). A re- search working group in the United Kingdom used the age-specific mortality prevailing in social class I (i.e. pro- fessional class) as a reference (12). Minimum death rate The idea of using a certain level of mortality as a norm or a standard reference was recognized already over a 100 years ago by William Farr. He observed that the mortality of the English people at that time did not exceed 17 per 1 OOO population; the deaths of 17 persons in 1 OOO might therefore be considered to be natural deaths "in our present imperfect state", and all deaths above that number might be referred to as unnatural deaths (13). Thus by using the crude death rate of 17 per 1 OOO as the threshold level, Farr computed excess mortality in London, other urban populations, unhealthy districts, etc. Following a similar approach, Guralnick & Jackson (14) observed that in the United States of America, around 1960, even the most unfavourable death rate was far below 17 deaths per 1 OOO population and used instead the most favourable rates recorded among the states. These rates were selected as the average of the five lowest rates among the states, and they were deter- mined for each age group, by cause of death. Likewise, Sauer & Brand ( 15) estimated that if death rates (from all causes combined) of the lowest-rate areas had applied to the entire United States from 1959 to 1961, there would have been approximately 100 OOO fewer deaths per year in the under-65 age group alone. The lowest age-sex-specific death rates recorded among countries in the world were used by Japanese researchers. Hishinumad and the Japanese Institute of Population Problems ( 16) used such lowest rates as the eventually attainable goal in connection with the future c Uemura. K. Excess mortality ratio with reference to the lowest age- sex-specific death rates among countries. Abstract paper for the XI Scientific Meeting of the International Epidemiological Association, 8-13 August 1987. Helsinki; and Sex differentials in mortality-application of the excess mortality ratio with reference to the lowest age-sex-specific death rates among countries. Abstract paper for the 46th Session of the International Statistical Institute, 8-16 September 1987. Tokyo (p. 455- 456). d Hishinuma, S History and future prospect of the human life span. Paper for the 50th Japan Medical Congress, 1978. Wld hlth statist. quart., 42 ( 1989) projection of life-table variables. The lowest rates found among countries' data and used by these researchers related to the latest calendar year for which a number of countries had published mortality statistics. These rates may be called the current minimum death rates. The method proposed in this article extends the time horizon so as to search the lowest rates recorded among coun- tries since 1950, which may be called the historical mini- mum death rates. The lowest rates thus found are con- sidered to represent the realistic level to which a coun- try's mortality could be lowered in due course. Excess mortality Excess mortality is defined in relation to a standard or normative mortality pattern by age and sex. As men- tioned above, in the present calculation the minimum age-sex-specific death rates observed among countries of the world since 1950 have been used as the stand- ard. By using symbol D for the total number of deaths in a country during a particular calendar year and symbol 0 5 for the number of deaths expected under the standard mortality pattern, the excess in the number of deaths is: 6.D = D- Ds By denoting the age-sex-specific death rate by miiwhere iindicates the age group andjthe sex, and the population in the age-sex group by Pij, the excess mortality is expressed by : 6.D = Z: Lp .. m. i j lj IJ _ Z: Lp m j j IJ /JS• where m ijs stands for the age-sex-specific death rate in the stanaard mortality pattern. The excess mortality ratio (EMR) may then be defined as: EMR = 6. D/Ds = L L Piimij/(L L Piimiisl - 1. In presenting the results of the application of the method below, the EMR is expressed as a percentage, for con- venience, namely: EMR = 100 (6. D/0 5 ). As implied by the above formulae, excess mortality is computed separately for each sex and added together for both sexes. The EMR is then computed by dividing this excess mortality by the total number of deaths expected in both sexes under the most favourable mor- tality pattern. The EMR, however, may be computed also for part of the population such as a particular age/sex group and a geographical unit of a country. Of special usefulness is the EMR for the age group under 65 (or under 70) for which death may be regarded, to a large extent, as unnecessary and preventable. The indicator thus defined is based on the concept of the standardized mortality ratio. Taking the ratio of the observed numbers to the expected numbers also fol- lows the calculation of the risk ratio and the odds ratio in epidemiological studies. Another way of expressing the amount of excess mortality would be to compute the proportional excess mortality (PEM) among the ob- served number of deaths. The indicator thus defined would then be: PEM = 100 (6. D/D) = 100 · EMR/(100 + EMR). - 28 - The PEM may be more appealing to health policy makers because of its easy interpretation. For the estimation of the EMR one of the approximate formulae derived by Flanders ( 17) for standardized rate ratios will be convenient. The formula given below takes into account the random variability due to the finiteness of the population sizes of both the country concerned and the reference countries which have registered the minimum death rates: Var (EMR) ~ 1002 (0/0~ + (02/0!) I: L m;jsPVPij5 ), where P ijs is the population in the age-sex group of the country which recorded the minimum death rate. For finding the minimum age-specific death rates, the data stored in the WHO mortality data base (18) have been screened. As many of the developing countries do not yet possess civil registration systems of sufficient accuracy for this purpose, and as, in general, their mor- tality level is substantially high, only those data pertain- ing to industrialized countries have been used (see the list in Table 1 in which the country codes used in this article are shown). Data from developing countries have, however, been included in the estimation of excess mor- tality, once the minimum death rates had been identified for the data from the industrialized countries. The computation has been based on age-specific mor- tality data according to the following age grouping: under 1 year; 1-4 years; 5-year groups from 5 to 84 years; 85 years and over. The lowest age-sex-specific death rates in the world The lowest age-sex-specific death rates registered in industrialized countries since 1950 are shown in Table 2. An interesting general feature seen from these rates is that the mortality pattern varies considerably among countries; Nordic and other Northern European countries tend to show very low mortality for children and young adults, while Greece, Israel and Japan display low mortality for the middle-aged. This might be a reflec- tion of the highly effective paediatric services as well as social measures taken for accident prevention in North- ern Europe on the one hand, and relatively healthy dietary habits in Greece, Israel and Japan on the other. The sex differentials in the minimum death rate vary according to age. As noticed from Table 2, the male/ female ratio is about 1.2-1.5 for children, jumps to more than 2.5 for young adults, falls below 2.0 for early middle age, rises to around 2.0 and remains at this level until age 70 years, and then gradually comes down as age advances. The very large sex differential observed in age groups 15-29 years reflects the high level of accident mortality among males in these age groups. Table 3 shows for each country the age groups for which the lowest death rate was found prior to 1970. For males, the lowest rate occurred already during this early period in a number of countries, the most conspicuous ones being Eastern European countries, Denmark, Ireland, Norway and Yugoslavia. For some other coun- tries the lowest rate for age groups 15-24 years as well TABLE 1. COUNTRY CODES USED IN THIS ARTICLE TABLEAU 1. CODES DE PAYS EMPLOYES DANS L'ARTICLE AUS AUT BEL BUL CAN CZE DEN FIN FRA DDR DEU GRE HUN IRE ISR ITA JPN NET NEZ NOR POL POR ROM SPA SSR SWE SWI UNK USA YUG Countries whose data were used in the search for the minimum death rate (period studied) Pays dont les donnees ont ete utilisees pour trouver le taux de monalite le plus bas (pt!riode etudit!e) Australia - Australie . . . . Austria - Autriche . Belgium - Belgique . . . . . Bulgaria - Bulgarie . . . . . . Canada ........... . Czechoslovakia - Tchecoslovaquie ..... . Denmark - Danemark . . . . . . . . . . . . Finland - Finlande . . . . . . . . . . . . . . . . France ...................... . German Dern. Rep. - Rep. dem. allemande .. Germany, Fed. Rep. of - Rep. fed. d'Allemagne Greece - Grece . . . . . . . . . . . . . . Hungary - Hongrie . . . . . . . . . . . . Ireland - lrlande Israel - Israel . . . . . . . . . . . . . . . . Italy - ltalie . . . . . . . . . . . . . . Japan - Japan . . . . . . . . . . . . Netherlands - Pays-Bas . . . . . . . . New Zealand - Nouvelle-Zelande . . . . . . Norway - Norvege . . . . . . . . . . . . . . Poland - Pologne . . . . . . . . Portugal . . . . . . . . . . ..... . Romania - Roumanie . . . . . . . . . . . . . . Spain - Espagne . . . . . . . . . . . . . . . . . USSR - URSS ............... . Sweden - Suede . . . . . . . . . . . . . . . . Switzerland - Suisse . . . . . . . . . . . . . . United Kingdom - Royaume-Uni E&W England & Wales - Angleterre et Galles NIA Northern Ireland - lrlande du Nord . SCO Scotland - Ecosse . . . . . . . . . . United States. of America - Etats-Unis d' Amerique ........ . Yugoslavia - Yougoslavie ........ . (1950-86) (1950-86) (1950-84) (1950-86) (1950-85) (1953-85) (1950-86) (1950-86) (1950-85) (1952-85) (1950-86) (1960-85) (1950-86) (1950-85) (1975-85) (1950-84) (1950-86) (1950-86) (1950-85) (1950-86) (1950-86) (1950-86) (1956-84) (1950-83) (1985-86) (1950-86) (1950-86) (1950-85) (1950-86) (1950-86) (1950-85) (1950-84) ARG CHI COR CUB DOR EGY ECU ELS GUT HON KOR KUW MEX PAN PAR PER PHL SIN SRL SYR THA TRT URU VEN Other countries Autres pays Argentina - Argentine Chile - Chili Costa Rica Cuba Dominican Republic - Republique dominicaine Egypt - Egypte Ecuador - Equateur El Salvador Guatemala Honduras Republic of Korea - Republique de Coree Kuwait - Kowert Mexico Panama Paraguay Peru - Perou Philippines Singapore - Singapour Sri Lanka Syrian Arab Rep. - Rep. arabe syrienne Thailand - Thai1ande Trinidad and Tobago - Trinite-et-Tobago Uruguay Venezuela Rapp. trimest. statist. sanit. mond., 42 (1989) Age - 29 - TABLE 2. LOWEST AGE-SPECIFIC DEATH RATES (PER 100000 POPULATION) REGISTERED IN THE WORLD SINCE 1950 TABLEAU 2. TAUX DE MORTALITE LES PLUS BAS PAR AGE (POUR 100000 HABITANTS) ENREGISTRES DANS LE MONDE DEPUIS 1950 Male Female Sexe masculin Sexe feminin (years/ ans) Death rate Country (year) Death rate Country (year) T aux de monalite Pays (annee) T aux de monalite Pays (annee) <1 564 JPN (1986) 467 FIN (1986) 1-4 22.6 SWE (1984) 20.7 SWE (1981) 5-9 18.6 DEN (1984) 10.1 FIN (1986) 10-14 14.2 SWE (1982) 10.7 FIN (1986) 15-19 53.3 NET (1986) 21.0 IRE (1975) 20-24 76.0 NET (1986) 27.4 NOR (1974) 25-29 73.9 NET (1983) 27.2 IRE (1985) 30-34 81.8 NET (1982) 42.0 NOR (1979) 35-39 113 ISR (1985) 68.8 GRE (1978) 40-44 189 ISR (1984) 104 GRE (1985) 45-49 323 GRE (1984) 169 GRE (1983) 50-54 556 GRE (1985) 261 GRE (1984) 55-59 882 JPN (1982) 397 JPN (1986) 60-64 1 266 JPN (1986) 621 JPN (1986) 65-69 2095 JPN (1986) 1043 JPN (1986) 70-74 3 518 JPN (1986) 1 888 JPN (1986) 75-79 6058 BUL (1950) 3 586 JPN (1986) 80-84 9 976 GRE (1984) 6354 CAN (1985) 85 + 16 133 YUG (1964) 13 788 USA (1979) TABLE 3. YEARS BEFORE 1970 IN WHICH THE LOWEST AGE-SEX-SPECIFIC DEATH RATES WERE RECORDED IN EACH COUNTRY• TABLEAU 3. ANNEES ANTERIEURES A. 1970 POUR LESQUELLES LES TAUX DE MORTALITE LES PLUS BAS PAR AGE ET PAR SEXE ONT ETE ENREGISTRES DANS CHAQUE PAYS• A. Male - Sexe masculin Country Age (years I ans) Pays <1 1-4 5-9 10-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85+ CAN USA JPN BUL 68 64 64 64 61 65 66 65 54 58 50 50 58 58 CZE 58 62 61 64 66 61 60 60 60 60 60 DDR 62 67 55 56 63 63 55 HUN 65 66 66 55 60 66 64 66 66 61 61 66 66 58 POL 68 64 61 66 65 66 65 58 68 58 DEN 62 63 69 56 60 61 68 51 53 53 55 FIN IRE . 56 60 65 68 52 53 60 67 67 52 NOR 55 51 51 51 51 52 55 SWE UNK: E&W 57 NIA . 62 65 67 59 67 sco 60 GRE 68 69 ITA 61 POR 65 69 60 SPA YUG 68 61 66 66 65 58 63 64 AUT BEL 59 DEU FAA 60 60 NET 50 52 58 60 61 SWI 65 65 AUS NEZ 64 60 53 Wld hlth statist. quan .• 42 ( 1989) - 30 - TABLE 3 (continued) TABLEAU (suite) B. Female - Sexe feminin Age (years/ ans) Country Pays < 1 1-4 5-9 10-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85+ CAN USA JPN BUL CZE DDR HUN POL DEN FIN IRE . NOR SWE UNK: E&W NIR . sea GRE ITA POR SPA YUG AUT BEL DEU FRA NET SWI AUS NEZ 'E.g. 68 - 1968 - Par example 68 - 1968. 58 65 66 61 64 65 69 as high age groups 75 years and over occurred before 1970. For females, such countries are much fewer; in a few Eastern European countries, the lowest rates found prior to 1970 related to adults, while in some countries in other parts of Europe they often related to the age group 15-19 years. As mentioned above, the high mortality in young adults in recent years appears to reflect the ten- dency among youth in many industrialized countries towards acquiring a more risk-taking behavioural pattern predisposing to violent death. It is worth noting also that in countries such as Canada, the United States, Japan, Finland, Sweden, Spain, Austria, the Federal Republic of Germany and Australia the minimum death rate was reg- istered quite recently, mostly since 1980, for every age/sex group, indicating a sustained improvement in survival. Table 4 compares the historical minimum death rates recorded up to 1960, those up to 1970 and those up to 1980, thereby showing how they have been declining over the last two decades. In both males and females the most rapid improvements took place for infants and young children. In females improvements occurred in all other age groups as well. In males, however, the mini- mum rate was not lowered during the two decades for the age groups 15-24 and 60-79. Nevertheless, it should be noted that improvements have also taken place since 1980 for most of these age groups, as can be inferred from Table 2. The cause-of-death structure observed in the lowest rate naturally varies from one age group to the other and 50 50 50 61 66 63 67 60 56 67 68 66 65 67 66 68 58 64 67 67 67 63 63 between the two sexes. Table 5 shows the causes of death for three age groups representing early childhood, youth and middle age. The latest data available from five industrialized countries are also given for comparison with the cause-of-death structure according to disease of the historical minimum death rate. In the two younger age groups, particularly the group 15-19 years, external causes (e.g. accidents) which are mostly avoidable, claim a considerable proportion of all deaths and the largest part of excess mortality. In many developing countries, however, upper respiratory infections and diarrhoeal diseases constitute major causes of death among children, although in the industrialized countries shown in Table 5, mortality from these diseases has practically disappeared. For the middle-aged group in industrialized countries, much of the excess mortality (apart from external causes) is due to cardiovascular dis- eases; the most important component of the latter is ischaemic heart disease but in some countries cerebro- vascular disease also takes a large share. As these dis- eases are considered as being associated with dietary habits and other lifestyle factors, a certain proportion of deaths at middle age may possibly be avoided, or at least postponed, through preventive and therapeutic meas- ures including health education of the public. Life expectancy under the lowest age-specific death rate A life table can be constructed for the mortality pattern represented by the minimum death rates. The life expec- Rapp. trimest. srarisr. sanit. mond .. 42 ( 1989) - 31 - TABLE 4. LOWEST AGE-SPECIFIC DEATH RATES (PER 100 OOO POPULATION) REGISTERED IN THE WORLD SINCE 1950 UP TO 1960, 1970 AND 1980 TABLEAU 4. TAUX DE MORTALITE LES PLUS BAS PAR AGE (POUR 100 OOO HABITANTS) ENREGISTRES DANS LE MONDE DEPUIS 1950 JUSQU'A 1960, 1970 ET 1980 A. Male - Sexe masculin Up to 1960 - Jusqu·a 1960 Up to 1970 - Jusqu·a 1970 Up to 1980 - Jusqu·a 1980 Age Death rate Death rate Death rate (years/ ans) Taux de Country (year) Taux de Country (year) Taux de Country (year) mortalite Pays (annt!e) mortalite Pays (annt!e) mortalite Pays (annee) <1 1810 SWE (1958) 1 258 SWE (1970) 794 FIN (1979) 1-4 90.7 DEN (1958) 59.9 SWE (1967) 36.8 SWE (1978) 5-9 40.1 DEN (1955) 39.1 E&W (1970) 25.3 SWE (1980) 10-14 32.1 DEN (1954) 29.8 IRE (1968) 21.5 SWE (1979) 15-19 58.8 IRE (1956) 58.8 IRE (1956) 58.8 IRE (1956) 20-24 82.2 IRE (1960) 82.2 IRE (1960) 82.2 IRE (1960) 25-29 96.1 NET (1955) 87.3 DEN (1969) 76.3 NET (1978) 30-34 106 NET (1960) 103 E&W (1968) 87.3 NET (1974) 35-39 138 NET (1958) 138 NET (1958) 128 NET (1979) 40-44 230 SWE (1959) 221 GRE (1970) 197 GRE (1978) 45-49 371 SWE (1958) 371 SWE (1958) 342 GRE (1980) 50-54 622 SWE (1960) 617 SWE (1963) 590 GRE (1980) 55-59 1 017 NOR (1951) 1 017 NOR (1951) 915 JPN (1980) 60-64 1470 NOR (1951) 1470 NOR (1951) 1470 NOR (1951) 65-69 2 436 NOR (1951) 2436 NOR (1951) 2436 NOR (1951) 70-74 3961 NOR (1951) 3 961 NOR (1951) 3 961 NOR (1951) 75-79 6058 BUL (1950) 6058 BUL (1950) 6058 BUL (1950) 80-84 10834 GRE (1960) 10374 GRE (1966) 10321 GRE (1978) 85+ 16 408 BUL (1958) 16 133 YUG (1964) 16 133 YUG (1964) B. Female - Sexe feminin Up to 1960 - Jusqu'a 1960 Up to 1970 - Jusqu'a 1970 Up to 1980 - Jusqu'a 1980 Age Death rate Death rate Death rate (years/ ans) Taux de Country (year) Taux de Country (year) Taux de Country (year) mortalite Pays (annt!e) mortalitt! Pays (annee) mortalitt! Pays (annt!e) < 1 1 355 SWE (1958) 933 SWE (1969) 570 SWE (1980) 1-4 67.4 DEN (1958) 40.7 SWE (1970) 25.5 FIN (1979) 5-9 22.8 DEN (1956) 22.5 NEZ (1967) 15.0 FIN (1980) 10-14 19.1 DEN (1955) 16.6 NOR (1968) 12.8 IRE (1979) 15-19 26.7 NOR (1958) 26.7 NOR (1958) 21.0 IRE (1975) 20-24 34.7 DEN (1960) 28.6 NOR (1967) 27.4 NOR (1974) 25-29 42.9 NOR (1959) 38.1 NOR (1969) 36.0 NOR (1977) 30-34 66.5 NOR (1959) 48.8 NOR (1965) 42.0 NOR (1979) 35-39 103 NET (1960) 86.3 NOR (1968) 68.8 GRE (1978) 40-44 143 NOR (1960) 139 NOR (1967) 117 NOR (1980) 45-49 240 NOR (1960) 214 NOR (1967) 183 GRE (1980) 50-54 335 NOR (1960) 335 NOR (1960) 316 GRE (1980) 55-59 611 GRE (1960) 566 GRE (1970) 484 JPN (1980) 60-64 1031 NOR (1960) 894 NOR (1970) 777 FRA (1980) 65-69 1 677 GRE (1960) 1 642 SWE (1970) 1 246 FRA (1980) 70-74 3 187 GRE (1960) 2862 FRA (1970) 2221 FRA (1980) 75-79 5 239 GRE (1960) 4805 CAN (1970) 4072 CAN (1979) 80-84 9 352 GRE (1960) 8499 USA (1970) 6930 CAN (1980) 85+ 14678 POL (1958) 13 992 YUG (1963) 13 788 USA (1979) tancy at birth calculated on the basis of the minimum Life expectancy based on minimum mortality is naturally death rates shown in Table 2 by applying Chiang's greater than the value computed for any individual coun- method for abridged life tables (19) is 76.2 years for try for the corresponding time reference. Nevertheless, males and 82. 1 years for females. For comparison, the the figures shown above for the three periods have now values of life expectancy based on the minimum rates been surpassed by some individual countries. The pat- observed up to 1960, 1970 and 1980 which are shown terns which emerge from the current and historical mini- in Table 4 are: mum mortality, therefore, do not in any way represent an unattainable status for a community. On the contrary, they give fairly modest goals which can be achieved in Male Female the near future, at least by some countries. (years) (years) Up to 1960 .. 73.8 77.6 Excess mortality ratio Up to 1970 . 74.4 78.9 The values of the EMA (in percentage) with reference to Up to 1980 .. 75.2 80.9 the historical minimum death rates computed for indus- Wld hlth statist. quart., 42 ( 1989) - 32 - TABLE 5. DISEASE STRUCTURE OF THE HISTORICAL MINIMUM DEATH RATES (PER 100 OOO POPULATION) FOR AGE GROUPS 1-4, 15-19 AND 50-54 YEARS AND COMPARISON AMONG SELECTED COUNTRIES TABLEAU 5. TABLEAU DES MALADIES CORRESPONDANT AUX TAUX DE MORTALITE MINIMAUX HISTORIQUES (POUR 100 OOO HABITANTS) POUR LES GROUPES D'AGE 1-4, 15-19 ET 50-54 ANS ET COMPARAISON ENTRE CERTAINS PAYS A. 1-4 years/ ans Male - Sexe masculin Female - Sexe feminin Causes 3 Minimum Minimum USA (SWE USA JPN DEU HUN SWE (SWE JPN DEU HUN 1984) 1985 1986 1986 1986 1986 1981) 1985 1986 1986 1986 1. Infectious and parasitic diseases - Maladies infectieuses et parasitaires 1.5 2.2 2.4 3.2 1.5 2.6 3.2 1.7 2.1 2.1 0.8 2. Neoplasms - Tumeurs . .. 3.1 4.9 5.1 4.6 7.3 4.6 1.6 3.6 3.8 3.8 10.4 3. Circulatory diseases - Maladies de l'appareil circulatoire 1.0 2.7 2.9 1.2 0.8 1.0 - 2.5 2.8 1.8 0.4 4. Respiratory diseases - Maladies de l'appareil respiratoire 1.0 3.2 4.6 3.6 6.1 1.0 0.5 3.1 4.0 3.2 4.0 5. Digestive diseases - Maladies de l'appareil digestif . - 1.1 0.6 1.0 0.4 - 0.5 1.0 0.3 0.4 0.8 6. Pregnancy complications - Complications de la grossesse - - - - - - - - - - - 7. Perinatal conditions - Morbidite perinatale ...... 4.1 7.5 9.9 8.0 14.5 6.7 4.8 6.1 9.1 8.6 14.8 8. External causes - Causes externes 8.7 27.2 23.2 17.0 17.6 5.7 6.4 18.6 12.5 11.7 13.2 9. All other diseases - T outes les autres maladies 3.2 9.3 6.1 8.7 6.9 9.8 3.7 7.9 5.4 7.3 5.6 All causes - T outes causes . 22.6 58.1 55.8 47.0 55.1 30.4 20.7 44.5 40.0 38.9 50.0 B. 15-19 years/ans Male - Sexe masculin Female - Sexe feminin Causes 3 Minimum Minimum (NET USA JPN DEU HUN SWE (IRE USA JPN DEU HUN 1986) 1985 1986 1986 1986 1986 1975) 1985 1986 1986 1986 1. Infectious and parasitic diseases - Maladies infectieuses et parasitaires 1.6 0.7 0.3 0.8 0.8 2.0 0.7 0.9 0.6 0.3 0.9 2. Neoplasms - Tumeurs ..... 5.9 5.9 5.6 6.2 9.1 3.7 3.5 3.9 3.9 3.4 5.4 3. Circulatory diseases - Maladies de l'appareil circulatoire 2.4 3.7 4.4 3.4 4.8 3.4 - 2.3 2.4 1.6 2.6 4. Respiratory diseases - Maladies de l'appareil respiratoire 1.0 1.6 1.9 2.1 3.5 1.7 1.4 1.1 0.8 1.4 2.0 5. Digestive diseases - Maladies de l'appareil digestif . 0.6 0.5 0.5 0.5 - - 0.7 0.6 0.4 0.4 1.1 6. Pregnancy complications - Complications de la grossesse - - - - - - 1.4 0.3 - - 0.3 7. Perinatal conditions - Morbidite perinatale ...... 2.4 1.5 0.8 1.1 2.9 0.7 0.7 1.0 0.7 0.7 2.6 8. External causes - Causes externes 31.0 94.2 47.9 59.9 66.4 46.9 9.8 31.7 12.2 21.2 20.8 9. All other diseases - Toutes les autres maladies 8.4 6.6 4.1 7.0 6.8 4.5 2.8 4.7 2.4 4.0 5.2 All causes - Toutes causes . 53.3 114.7 65.5 81.0 94.3 62.9 21.0 46.5 23.4 38.5 40.9 c. 50-54 years/ ans Male - Sexe masculin Female - Sexe feminin Causes 3 Minimum Minimum (GRE USA JPN DEU HUN SWE (GRE USA JPN DEU HUN 1985) 1985 1986 1986 1986 1986 1984) 1985 1986 1986 1986 1. Infectious and parasitic diseases - Maladies infectieuses et parasitaires 5 11 11 7 19 2 2 6 4 3 4 2. Neoplasms - Tumeurs .... 187 236 214 242 401 130 134 208 128 183 239 3. Circulatory diseases - Maladies de l'appareil circulatoire 216 352 161 263 613 237 64 129 70 74 215 4. Respiratory diseases - Maladies de l'appareil respiratoire 15 35 19 24 53 25 5 23 10 11 21 5. Digestive diseases - Maladies de l'appareil digestif 33 54 69 75 176 33 8 27 14 26 65 6. Pregnancy complications - Complications de la grossesse - - - - - - - - - - - 7. Perinatal conditions - Morbidite perinatale ...... - 1 1 1 4 1 - 1 1 1 1 8. External causes - Causes externes 65 84 103 74 221 90 19 29 30 25 60 9. All other diseases - T outes les autres maladies 35 62 31 79 67 58 29 40 18 34 42 All causes - Toutes causes 556 835 609 765 1 554 576 261 463 275 357 647 SWE 1986 1.6 6.5 1.1 3.3 - - 9.8 8.2 3.8 34.3 SWE 1986 0.4 0.7 0.7 0.4 - - 1.4 17.5 3.2 24.3 SWE 1986 4 178 55 14 13 - 2 33 27 326 • Cause groups comprise the following ICD-9 codes (or their equivalent in earlier versions of ICD) : 1 ICD 001-139; 2. ICD 140-239 ; 3. ICD 390-459; 4. ICD 460-519 ; 5. ICD 520-579; 6. ICD 630-676; 7. ICD 740-779; 8. ICD E800-E999; 9. Remainder - Les groupes de causes correspondent aux codes CIM-9 suivants (ou leurs equivalents des vers,onsanterieures de la CIM): 1. ICD001-139; 2. ICD 140-239; 3. ICD 390-459; 4. ICD460-519; 5. ICD 520-579; 6. ICD 630-676; 7. ICD 740-779; 8. ICD E800-E99g; 9. Toutes les causes restantes. Rapp. trimest. statist. sanit. mond .. 42 (1989) - 33 - trialized countries are shown in Table 6 and for some developing countries in Table 7. Two sets of results are tabulated using the latest available data, one for all ages combined and the other for the under-65 age group. The values of the EMR are generally higher for those under 65 than for all ages combined, and thus, by infer- ence, than for those 65 years or over. This reflects the fact that the variation in mortality rates among countries becomes smaller as age advances, which is consistent with the commonly-held view that among the elderly the preventable proportion of mortality diminishes with age. As seen in Table 6, the EMR for all ages and both sexes varies between 5.0% (Japan) and 105.1 % (Hungary) among the industrialized countries, while the EMR for those under 65 and both sexes varies from 6. 7% (Japan) to 172.2% (Hungary). Countries with a low EMR are Japan, Sweden, Switzerland, Greece and the Nether- lands, with the value below 30% for the age group under 65. Those with a high EMR are mostly Eastern European countries, with the value for the under-65 age group often exceeding 100%. Among the developing countries the EMR is generally much higher than in the industrial- ized countries, as shown in Table 7. The value for Guatemala, 552% for both sexes under 65, is the largest among the countries studied, but it should be kept in mind that in many other developing countries which do not have established systematic civil registration and whose mortality has been underreported, the health conditions are poor and the true level of excess mortality may be even higher. Further breakdowns of the EMR by age groups for selec- ted countries are given in Table 8. It is clear that the age pattern of the EMR differs considerably from country to country, even though the general features outlined above still apply. For instance, in Hungary, high values of the EMR occur in age groups 10-14 and 25-7 4; in Japan the value is considerably high in age group 1-4, com- pared with the otherwise low level; and in Sweden the value is somewhat high in age groups 10-14 and 25-39 in contrast to a generally low level in other age groups. In developing countries the EMR is extremely high in the early years of life, with a gradual decline with age. The TABLE 6. EXCESS MORTALITY RATIO (EMR) (%) FOR ALL AGES AND FOR AGE UNDER 65 IN INDUSTRIALIZED COUNTRIES• TABLEAU 6. RAPPORT DE SURMOR'rALITIE {EMRJ (%) POUR TOUS LES AGES ET POUR LES MOINS DE 65 ANS DANS LES PAYS INDUSTRIALISES• EMA for all ages EMA for age under 65 Country EMA pour to.us les ages EMA pour les moins de 65 ans Pays Both sexes M F M F Both sexes Les deux sexes J,.es deux sexes Northern America - Amerique septentrionale CAN (1985) 26.2 15.8 21.4 40.0 47.1 42.4 USA (1985) 37.3 26.1 31.7 72.6 81.5 75.7 Asia - Asie ISA (1985) 22.0 59.0 37.0 33.5 64.1 44.4 JPN (1986) 5.2 4.7 5.0 7.0 6.3 6.7 Eastern Europe - Europe orientale BUL (1986) 72.7 99.1 83.6 97.9 90.1 95.3 CZE (1985) 98.8 93.6 96.2 132.3 101.8 121.5 ODA (1985) 73.7 88.0 81.4 76.8 85.4 80.0 HUN (1986) 110.0 100.0 105.1 183.3 152.5 172.2 POL (1986) 98.3 79.1 88.8 148.7 108.4 134.2 ROM (1984) 82.7 117.3 97.6 126.1 143.9 132.3 SSA (1986) 112.8 78.5 92.7 182.1 148.7 169.1 Northern Europe - Europe septentrionale DEN (1986) 37.8 35.1 36.5 53.1 91.1 65.9 FIN (1986) 55.6 36.4 45.5 80.6 33.6 64.1 IRE (1985) 64.7 67.1 65.8 58.2 75.0 63.8 NOR (1986) 28.6 18.9 24.0 36.1 31.3 34.5 SWE (1986) 20.9 18.0 19.5 18.0 23.9 20.0 UNK: E&W (1985) 46.6 43.5 45.0 43.8 70.9 53.0 NIA (1986) 63.8 59.8 61.8 71.8 83.5 75.9 sco (1986) 65.7 59.4 62.4 77.0 99.7 84.9 Southern Europe - Europe meridionale GAE (1985) 17.7 38.7 26.9 24.8 25.7 25.1 ITA (1984) 37.9 34.9 36.5 46.8 36.3 43.1 POR (1986) 47.0 48.4 47.6 76.6 65.5 72.6 SPA (1983) 25.0 28.4 26.6 35.9 28.0 33.2 YUG (1984) 79.3 99.6 88.3 117.6 125.0 120.2 Western Europe - Europe occidentale AUT (1986) 44.2 45.6 45.0 65.5 52.3 60.7 BEL (1984) 51.6 43.0 47.3 60.3 61.6 60.7 DEU (1986) 41.1 37.4 39.1 47.0 44.8 46.2 FRA (1985) 32.4 15.1 23.5 64.3 34.1 54.0 NET (1986) 31.3 20.4 25.9 24.8 34.6 28.1 SWI (1986) 18.8 12.9 15.9 24.2 20.1 22.8 Oceania - Oceanie AUS (1986) 26.5 21.3 24.1 36.2 43.6 38.6 NEZ (1985) 49.9 50.6 50.2 60.6 90.3 70.5 a Reference; historical minimum death rates - Reference: taux de mortalite minimaux historiques. Wld hlth statist. quan., 42 (1989) - 34 - TABLE 7. EXCESS MORTALITY RATIO (EMR) (%) FOR ALL AGES AND FOR AGE UNDER 65 IN SELECTED DEVELOPING COUNTRIES• TABLEAU 7. RAPPORT DE SURMORTALITE (EMRJ (%) POUR TOUSLES AGES ET POUR LES MOINS DE 65 ANS DANS CERTAINS PAYS EN DEVELOPPEMENT 8 EMA for all ages b EMA for age under 65 Country EMA pour tous les Ages b EMA pour les moins de 65 ans Pays Both sexes Both sexes M F Les deux sexes M Les deux sexes Africa - Afrique EGY (part) - (en partie) ( 1980) ...... 389.0 664.6 483.4 I Latin America - Amerique latine ARG (1983) 76.1 87.0 80.7 126.9 148.2 134.2 CHI (1985) 83.1 84.6 83.8 122.8 134.2 126.8 COR (1983) 18.3 44.5 28.5 65.1 108.8 80.1 CUB (1985) 25.0 68.1 40.8 60.8 136.9 86.0 DOR (1984) 89.0 147.3 110.9 162.9 286.1 205.0 ECU (1986) 121.8 180.1 144.9 214.8 344.0 259.4 ELS (1984) 144.7 133.0 139.8 274.1 288.0 279.0 GUT (1984) 247.0 388.9 300.3 446.3 750.1 552.1 HON (1981) 113.3 184.5 146.7 190.5 297.0 228.3 MEX (1982) 134.6 150.5 141.1 259.0 297.3 272.7 PAN (1984) 35.8 68.7 48.3 65.1 127.5 86.0 PAR (reporting area - zone ayant soumis des rapports) ( 1985) 160.5 317.5 214.5 PER (1983) 87.7 164.6 118.0 155.2 302.7 207.1 TRT (1983) 107.0 161.1 128.5 128.7 190.6 149.3 URU (1985) 60.9 57.9 59.5 100.0 112.2 104.2 VEN (1983) 98.4 122.2 107.9 170.4 213.0 185.3 Asia - Asie KOR (1985) 111.9 64.5 89.7 148.8 115.2 136.9 KUW (1986) 44.5 106.0 63.0 43.6 119.8 63.4 PHL (1983) 325.3 405.3 353.3 SIN (1986) 54.4 68.3 60.1 62.2 88.1 70.7 SRL (1982) 95.3 148.7 114.0 178.6 259.4 205.0 SYR (1985) 11.9 71.7 34.6 86.5 254.7 146.0 THA (1980) 133.2 140.5 136.2 118.3 152.1 130.0 a Reference historical minimum death rates - Reference: taux de monalite minimaux historiques. b ... indicates that EMA cannot be computed in a comparable manner as mortality tabulation by five-year age groups is not available up to 84 years in the respective official statistics - ... signifie que le rapport de surmortalit<! ne peut pas etre calcul,! de fa~on comparable du fait que la pr~sentation en tableaux par groupe de cinq ansjusqu·~ rage de 84 ans n'est pas disponible dans les statistiques officielles respectives. TABLE 8. EXCESS MORTALITY RATIO (EMR) (%) BY AGE GROUPS, BOTH SEXES. SELECTED COUNTRIES• TABLEAU 8. RAPPORT DE SURMORTALITIE (EMRJ (%) PAR GROUPE D'AGE POUR LES DEUX SEXES DANS CERTAINS PAYS 8 Age DEU HUN ITA JPN SWE USA CHI VEN THA (years/ ans) (1986) (1986) (1984) (1986) (1986) (1985) (1985) (1983) (1980) <1 65.3 269.4 121.4 1.4 14.2 106.0 277.8 431.2 156.6 1-4 98.8 144.5 72.5 122.2 50.6 137.3 308.5 714.6 1 546.3 5-9 53.3 93.0 42.7 31.6 8.8 71.5 156.6 249.8 920.1 10-14 61.8 106.7 84.0 29.5 65.6 123.0 215.5 301.6 721.0 15-19 53.4 83.0 45.6 19.7 18.2 116.9 86.3 195.6 367.8 20-24 37.5 67.5 33.5 11. 1 27.0 110.4 118.8 243.6 380.5 25-29 43.2 164.5 30.6 17.6 45.4 124.7 155.0 259.4 426.9 30-34 48.1 228.9 28.8 15.3 37.3 116.4 166.4 209.5 404.3 35-39 50.7 237.4 21.5 14.2 39.6 89.8 129.8 145.0 314.2 40-44 49.8 229.0 25.6 17.8 14.9 72.6 134.1 125.8 259.7 45-49 40.4 203.6 27.1 8.4 20.3 62.7 111.5 92.7 188.1 50-54 37.2 168.6 32.2 8.1 10.4 59.3 90.5 78.2 142.2 55-59 42.6 158.4 38.7 1.0 13.2 61.9 92.5 65.7 105.1 60-64 52.2 153.3 56.3 22.8 69.1 89.3 77.4 111.4 65-60 43.1 119.6 38.4 24.6 51.4 76.7 56.2 80.3 70-74 45.6 108.6 35.3 20.6 36.0 72.1 60.5 59.0 75-79 36.5 82.2 31. 7 0.5 13.7 16.0 55.9 19.0 34.5 80-84 37.4 74.9 33.2 8.4 16.9 9.9 73.4 39.6 28.6 85+ 31.1 52.4 35.7 9.4 23.3 7.0 32.0 31.8 All ages - Taus les ages 39.1 105.1 36.5 5.0 19.5 31.7 83.8 107.9 136.2 a Reference historical minimum death rates - Reference: taux de mortalite minimaux historiques. Rapp. trimest. statist. sanit. mond., 42 ( 1989) - 35 - relatively low figure of the EMR for infants in Thailand has most likely been caused by underregistration of infant deaths. The absence of excess mortality at age 85 and over in that country may also be a statistical artifact caused by inaccurate age recording. In any event, the peculiar features specific to a country point to the need for further study on their etiology and for possible public health and social measures to be taken. The trends in the EMR for selected industrialized coun- tries are shown in Figs 1 & 2, with reference to the his- torical and current minimum death rates, respectively. In Fig. 1 the example of Hungary indicates an increasing mortality trend during the last 20 years, while a sustain- ed decline in the EMR has been the general rule in the other countries. The trend pattern seen in Fig. 2 reveals how each country fares in relation to the most favourable situation existing in each calendar year; Hungary's posi- tion has deteriorated rapidly since the late 1960s after considerable improvements during the preceding pe- riod; Japan has made particularly rapid improvements throughout the three decades; and the United States, Federal Republic of Germany and Sweden have maintain- ed approximately the same relative level of the EMR, though a tendency for a slight increase is seen for Sweden since the 1960s. Table 6 also shows that in many of the industrialized countries the value of the EMR is similar for males and females. Comparison for the age group under 65, shown in Fig. 3, reveals that in Finland, France and some Eastern European countries the EMR is considerably higher for males while in Israel and some Northern European coun- tries it is much higher for females. This suggests the existence in these countries of a substantial sex differen- tial in health-related lifestyles. In the developing countries, on the other hand, the EMR is generally higher for females than for males, as seen in Fig. 4. This may well be indicative of the unfavourable conditions of life to which women are often subjected in these countries, especially in relation to child-bearing and child care. It is interesting to note that the female preponderance in the EMR was also commonly observed earlier in industrialized countries, as seen in Fig. 5 which compares the two sexes around 1955. Besides impro- vements in women's health, deterioration of lifestyles among males may have contributed to the recent rapid change in the sex differentials in EMR. If this trend were to persist, there would be male preponderance in EMR in many developing countries also in the near future. For the rest of the world, although basic data are missing in most of the developing countries, the United Nations has published estimates and projections, including po- pulation by age and sex, and numbers of births and deaths for each country (20). According to these estima- tes, the actual number of deaths for the world as a whole, FIG.1 TRENDS IN EXCESS MORTALITY RATIO (EMR) WITH REFERENCE TO THE HISTORICAL MINIMUM DEATH RATES (AS OF 1986), BOTH SEXES UNDER 65 YEARS, SELECTED COUNTRIES TENDANCES DU RAPPORT DE SURMORTALITE (EMRJ EU EGARD AUX TAUX DE MORTALITE MINIMAUX HISTORIQUES (A PARTIR DE 1986) POUR LES MOINS DE 65 ANS DES DEUX SEXES DANS CERTAINS PAYS EMA % 500 1JPN I \ I 400 I I I \ \ 300 \ \ ' "'\ ·. HUN .... ·· . .................. DEU ••···. USA ' •• DEU ~-.... SWE O-t-~~"'T"""~~-t--~~....-~~t--~--,,--~-+~~-·T"""- JPN 1950 1955 1960 1965 1970 1975 1980 1985 Wld hlth statist. quan .. 42 (1989) - 36 - FIG. 2 TRENDS IN EXCESS MORTALITY RATIO (EMR) WITH REFERENCE TO THE CURRENT MINIMUM DEATH RATES, BOTH SEXES UNDER 65 YEARS, SELECTED COUNTRIES TENDANCES DU RAPPORT DE SURMORTALITIE (ENIR} EU IEGARD AUX TAUX DE MORTALITIE MINIMAUX ACTUELS POUR LES MOINS DE 65 ANS DES DEUX SEXES DANS CERTAINS PAYS EMA % 200 \ JPN I I \ 150 '\ "· 100-+-~-\-~~-=..'--i~~~~~~t--~~~-+~+-~~-1 \ ....... ··· .. .··· .···· 50 DEU ...... "'" ·-·-· ........ _ 0 1950 1955 1960 1965 1970 1975 1980 1985 FIG.3 HUN USA DEU SWE JPN RELATION BETWEEN MALE AND FEMALE EXCESS MORTALITY RATIOS (EMR) IN INDUSTRIALIZED COUNTRIES WITH REFERENCE TO HISTORICAL MINIMUM DEATH RATES, UNDER 65 YEARS, LATEST DATA (AROUND 1985) RELATION ENTRE LES RAPPORTS DE SURMORTALITIE (ENIR} POUR LE SEXE MASCULIN ET POUR LE SEXE FIEMININ DANS LES PAYS INDUSTRIALISES EU IEGARD AUX TAUX DE MORTALITIE MINIMAUX HISTORIQUES POUR LES MOINS DE 65 ANS, DONNIEES LES PLUS RIECENTES (VERS 1985) % Female EMA - EMA pour le sexe feminin 200-.-~~~~~~~~~~~~~~~~~~~~----. 150 100 DEN• ~EZ E&W• 50 . FRA • FIN 0 50 100 • POL 150 HUN . . SSA f; t Q ; I 200% Male EMA - EMA pour le sexe masculin Rapp. trimest. statist. sanit. mond .. 42 (1989) - 37 - FIG.4 RELATION BETWEEN MALE AND FEMALE EXCESS MORTALITY RATIOS (EMR) IN DEVELOPING COUNTRIES WITH REFERENCE TO HISTORICAL MINIMUM DEATH RATES, UNDER 65 YEARS, LATEST DATA (AROUND 1985) RELATION ENTRE LES RAPPORTS DE SURMORTALITIE (EMRJ POUR LE SEXE MASCULIN ET POUR LE SEXE FIEMININ DANS LES PAYS EN DIEVELOPPEMENT EU IEGARD AUX TAUX DE MORTALITi MINIMAUX HISTORIQUES POUR LES MOINS DE 65 ANS, DONNiES LES PLUS RiCENTES (VERS 1985) Female EMR - EMR pour le sexe 1'minin % 600 500 400 300 . I 200 ··. ·::-: 100 0 100 200 300 400 SOO 600 700% Male EMR - EMR pour le sexe masculin FIG. 5 RELATION BETWEEN MALE AND FEMALE EXCESS MORTALITY RATIOS (EMR) IN INDUSTRIALIZED COUNTRIES WITH REFERENCE TO HISTORICAL MINIMUM DEATH RATES, UNDER 65 YEARS, AROUND 1955 RELATION ENTRE LES RAPPORTS DE SURMORTALITi (EMRJ POUR LE SEXE MASCULIN ET POUR LE SEXE FIEMININ DANS LES PAYS INDUSTRIALISES EU IEGARD AUX TAUX DE MORTALITIE MINIMAUX HISTORIQUES POUR LES MOINS DE 65 ANS, VERS 1955 Female EMR - EMR pour le sexe ft!minin % aoo--.-~~~~~~~~~~~~~~~~~~--::1 700 600 500 400 300 . . 200 .. 100 • I . 0 100 200 300 400 500 600 700 800% Male EMR - EMR pour le sexe masculin Wld hlth statist. quan., 42 (1989) - 38 - including both developed and developing countries, amounts to about 50 million per annum. The application of the above method of calculation to the United Nations· data, slightly adjusted to fit them to the age groups used in this article, has resulted in the estimated excess deaths of 33.0 million, corresponding to an EMA of 211 % . For developing countries alone, the EMA is esti- mated at 330%, that is, more than three-quarters of the deaths occurring in these countries constitute excess mortality, in contrast to the value of the EMA of 58% in the developed countries as a whole. The level of the EMA is especially high in Africa; the value is estimated at 838% in Western Africa, 740% in Middle Africa, 707% in Eastern Africa, 396% in Northern Africa and 327% in Southern Africa. Discussion In this study, data since 1950 from 32 industrialized countries were used to find the lowest death rates. This selection of countries was imposed mainly by the avail- ability of reliable data, but, as countries not included in the selection are mostly developing countries which from other evidence are believed to suffer from higher mortality than those selected, it is unlikely that any of the age/sex groups in any of these countries would have experienced a lower mortality pattern than the minimum observed in the 32 countries in question. The use of data from the period since 1950, on the other hand, was mainly due to easy accessibility of the mortality data tabulated according to five-year age groups, even though for some countries earlier data could be found in national reports. Here again, as survival has been gener- ally improving in these countries over recent decades, neglecting the mortality level before 1950 would not have affected the outcome of the search for the lowest age-sex-specific death rates. Caution is required when dealing with the minimum death rate, especially with respect to the influence of random fluctuations. It is important to ensure that the minimum rate detected has not been biased unduly by a random downward deviation. This could be expected to be more pronounced in a small-sized population. Countries with a population under 1 million were therefore not included in the search for the minimum death rate. In order to further reduce such random fluctuations, a modified set of mini- mum death rates was constructed by using only cases involving at least 500 deaths per age-sex group. If the number of deaths occurring in a population is assumed to follow the Poisson distribution, the coefficient of varia- tion around the mean of 500 will be the reciprocal of the square root of 500, namely 0.0447, or about 4.5%. This might be considered as an acceptable degree of preci- sion. In order to ensure the minimum number of 500 deaths, observations over several consecutive years were combined for countries with a relatively small population. The values of the EMA so determined had the effect of reducing the EMA calculated on the basis of the unadjusted minimum rates. However, the difference be- tween the two sets of the EMA was found to be small, namely about 1 percentage point for all ages combined and about 2-6 percentage points for the age groups under 65. The use of the EMA based on the unrestricted minimum death rates therefore does not appear to cause any serious bias from the practical point of view. The second problem relating to the application of the minimum death rates is whether they represent a realis- tic mortality pattern. It may be argued that in a birth cohort lower mortality at younger ages may make less healthy survivors at older ages by simply postponing death; consequently, the world minimum observed for younger ages in the statistics from country A may not be compatible with the world minimum observed for older ages in those from country 8, i.e. the mortality pattern represented by the list of the minimum death rates in the world for different age groups would not be realistic. The controversy which has arisen in relation to the mortality "crossover" effect between the white and black popu- lations in the United States is relevant in this regard. Mortality statistics have consistently shown a higher level of age-specific rate in the blacks than in the whites up to age 75 or so and lower rates thereafter. Some reports and authors (21, 22) consider that this has been caused by a statistical artefact due to exaggerated age reporting both in population censuses and on death cer- tificates for the elderly, while others (23-25) argue that it is a real phenomenon possibly due to some type of selection in the blacks resulting in more robust survivors at high ages. In particular, Nam et al. (23) extended their study to international comparisons of age-specific mor- tality rates to cover 31 cases of mortality crossover for a combination of countries and years and concluded that "cardiovascular" and "other and unknown" diseases appeared to contribute strongly to the crossover effect. The controversy is apparently not settled yet. Such an argument for the existence of strong biological selection, however, seems less plausible in the light of the time-series data obtained in a number of countries. As discussed earlier in conjunction with Table 3, the mortality rate, especially among females, has been im- proving in most age groups including at old age in most of the industrialized countries and also in most of the developing countries. Even in the United States, both the white and the black populations have been improving their survivorship throughout the life span, in spite of the crossover phenomenon referred to above. The sustain- ed rising trends in cohort survival curves for the Japa- nese population provide another good example (26). Furthermore, the age at the black/white mortality cross- over which used to be observed in the United States around 75 years until the 1970s has now shifted to 85 years and above (27), which may be interpreted as being due to the fact that birth cohorts with more accurate age records have taken over the age group 75-84 years. It will be interesting to see whether or not the crossover effect will gradually disappear over the next decade or so. As for cardiovascular and other diseases causing mortality crossovers among countries as studied by Nam et al., one should note that many of these diseases are considered to be linked with dietary habits and other lifestyle factors and, hence, that mortality from these diseases is at least partly avoidable, especially at middle age. It would be a hasty conclusion to associate these deaths with biological selection. It would therefore seem that a reduction of mortality at younger ages has not adversely affected the mortality level of the same popu- lation at higher ages in any appreciable manner, at least at the still-declining level of mortality currently observed in the human population. Instead, one may argue that the minimum death rates, whether current or historic, still contain a considerable proportion due to "avoidable" deaths referred to above (2-8), and, hence, even lower rates may be considered as attainable. The cause-of-death breakdowns of the minimum death rates shown in Table 5 appear to sup- port this argument. Carrying the argument still further, one may consider all mortality under 65 or 70 as poten- tially avoidable, which in fact provides the rationale for indicators of "years of potential life lost". From this angle, the minimum rates used in this article would still appear to be much higher than the ideal level. One may indeed consider an approach by which minimum rates are sought among countries for each disease or dis- ease group in order to estimate the EMA, as done by Woolsey (10) for geographical variations in mortality Rapp. trimest. statist. sanir. mond .. 42 (1989) - 39 - from 16 groups of diseases within the United States. This however would involve the risk of neglecting the important variations in diagnostic and coding practices which exist among countries, and often among geogra- phical regions within the same country. Various aspects of the validity of mortality statistics. including the pro- blems of medical diagnosis and statistical coding, have been reviewed by Alderson (28). For international com- parisons perhaps only a very broad grouping of causes of death would be appropriate in this regard. Moreover, diseases do not necessarily occur independently of each other, and the direction of their correlation may be posi- tive or negative. In other words, the simple sum total of the disease-specific minimum death rates thus found may not represent a realistic level of mortality which may be achieved. It would be desirable that the "compensa- tions" among disease groups be duly taken into account when adopting this approach. The EMR may of course be computed with reference to the lowest age-sex-specific death rates among subna- tional population groups within a country. The procedure may then be extended to make use of the minimum rates found in all countries of the world in the same way. Pre- caution should be taken however so as to avoid exces- sive disaggregation, as the effects of random fluctuation due to smallness of the population size would become not negligible. As the basic data used in the computation of the EMR are age-specific death rates, it is closely correlated to the usual mortality indicators such as age-standardized death rates. In fact, from the point of view of age-stan- dardization of rates, the EMR is an indicator based on indirect standardization, the standard being the series of the world-lowest age-specific death rates. However, as the standard chosen constitutes a desirable norm at a given point in time, the EMR measures the degree of deviation from the norm and appeals directly to health planners and decision makers. Finally, intercountry comparisons of an indirectly stan- dardized indicator are affected by the age/sex composi- tion of each country's population. An adjustment for this factor may be made by using the same age/sex structure as a standard for all countries. The EMR arrived at will obviously lose a certain degree of comprehensibility, a desirable criterion for an indicator for practical use. Conclusions There are several different ways of establishing a target mortality level for use in health planning. The approach followed in this article is based on the most favourable age-sex-specific death rates observed among coun- tries. With reference to such minimum rates, the suggested measure of excess mortality (EMR) highlights the amount of mortality potentially preventable by the means curren- tly available. With reference to the historical minimum death rates, the EMR in its aggregated form shows an overall level of excess mortality and monitors progress and development, or lack thereof, towards an achievable target. With reference to the current minimum death rates the EMR indicates whether a country is moving ahead in pace with the low mortality countries. In its disaggregated form, the EMR points to the specific population groups in each country which display a high level of EMR and which would require the attention of health and social planners for devising suitable selective intervention, and of researchers for taking further steps in the clarification of the medicosocial problems in- volved. The value of the EMR computed for the age group under 65 appears particularly useful in this re- spect. In industrialized countries the causes of excess mortality are largely due to unhealthy lifestyles and social pathology, and the EMR will provide clues for further study which may be made into excess mortality with a view to developing a health and social intervention policy for its reduction. SUMMARY Indicators based on mortality data have long been used to measure the level of health status and to monitor and evaluate the progress and achievements of health pro- grammes. Their usefulness is particularly obvious when dealing with preventable deaths. This article proposes the use of the lowest death rate recorded among indus- trialized countries for each age/sex group as an achiev- able target and as a reference for assessing the amount of excess mortality. The resulting indicator, excess mor- tality ratio (EMR), reveals some features of the mortality pattern which may not be easily noticed by means of other mortality indicators. Two sets of the lowest age-sex-specific death rates are considered, namely one comprising the lowest rates recorded in each calendar year (the current minimum) and the other comprising the lowest rates ever recorded since 1950 (the historical minimum). The former may be used for monitoring whether a country is moving ahead in mortality reduction in pace with low mortality coun- tries, while the latter may be considered as a realistic goal for a country's mortality reduction. In computing the EMR, the lowest death rates are first applied to the age-sex composition of the population of a given country for a given calendar year to obtain the number of deaths which would have been expected under the lowest mortality pattern; the expected number Wld hlth statist. quart., 42 ( 1989) is then subtracted from the actual number of deaths recorded in the country during the calendar year to yield the ··excess··. The indicator is finally calculated by taking the ratio of the excess to the expected minimum. The historical minimum death rates found from the records maintained in WHO's mortality data base are shown in Table 2 (for country codes used, see Table 1). The minimum rates have themselves declined with time as seen in Table 4, especially in young age groups, and the declining trend has been more marked in females. These trends are seen also in individual countries' data (Table 3). A comparison of the cause-of-death pattern of a country with that of the world· s lowest death rate will reveal the causes to which the country's excess mortality is attrib- utable, as seen in Table 5. At the same time, the table also shows that even the lowest rates are made up of causes which are largely preventable. The world's low- est rates therefore may be regarded as conservative targets for mortality reduction. The most recent values of the EMR (as percentages), with reference to the historical minimum death rates, are shown in Tables 6 & 7 for the industrialized and the developing countries, respectively. Two sets of results, one for all ages combined and the other for age under 65, - 40 - are tabulated. The latter are particularly useful, as most of the excess mortality in these groups is considered preventable. The trends in the EMA for these age groups are shown in Figs 1 & 2, with reference to the historical and current minimum death rates, respectively. The EMA may be disaggregated further such as by age and sex. The disaggregated EMA points to the specific population groups in each country which display a high level of EMA and which would require the attention of health and social planners for devising suitable selective intervention as well as of researchers for taking further steps in the clarification of the medicosocial problems involved. The sex differentials presented graphically in Fig. 3 for industrialized countries do not reveal any ten- dency towards male or female preponderance in the EMA, but Fig. 4 shows a clear female disadvantage in the developing countries. It is interesting to note that female preponderance in the EMA existed also in the industrial- ized countries a few decades ago, as seen from Fig. 5. As for disaggregation by age, Table 8 shows varying patterns from country to country. In developing coun- tries it is infant and child mortality which causes a large part of the excess. In some industrialized countries, how- ever, it is the middle-age mortality which needs to be reduced through the adoption of healthier lifestyles. RESUME Rapport de surmortalite fonde sur les plus bas taux de mortalite des pays par sexe et par groupe d' ige II y a longtemps qu'on utilise des indicateurs fondes sur les donnees relatives a la mortalite pour mesurer la situa- tion sanitaire et pour surveiller et evaluer le degre d' avan- cement et les resultats des programmes de sante. L'in- teret de ces indicateurs est particulierement evident pour les deces evitables. II est propose dans cet article d'uti- liser le taux de mortalite le plus bas enregistre dans les pays industrialises pour chaque groupe d'Age et chaque sexe comme objectif realiste et comme reference pour evaluer !'importance de la surmortalite. L'indicateur qui en resulte, a savoir le rapport de surmortalite (excess mortality ratio-EMR), fait ressortir certains elements du schema de la mortalite que d' autres indicateurs de mor- talite mettraient peut-etre mains bien en evidence. Les plus bas taux de mortalite par age et par sexe consi- deres relevent de deux categories comprenant l'une, les taux les plus bas enregistres pour chaque annee civile (minimum actuel) et l'autre, les taux les plus bas jamais enregistres depuis 1950 (minimum historique). Les pre- miers peuvent servir a verifier si un pays parvient a reduire sa mortalite au meme rythme que des pays ou la mortalite est faible, tandis que les seconds peuvent etre consideres comme un objectif realiste pour un pays qui veut reduire sa mortalite. Pour calculer le rapport de surmortalite, on applique d'abord les taux de mortalite les plus bas a la repartition par groupe d'Age et par sexe de la population d'un pays donne pour une annee civile donnee, afin d'obtenir le nombre de deces auquel on aurait pu s'attendre avec le schema de mortalite le plus bas; ce nombre attendu est alors deduit du nombre effectif de deces enregistres dans le pays durant l'annee civile afin d'obtenir I' «exce- dent». On calcule enfin l'indicateur en prenant le rapport de cet «excedent» au minimum attendu. Les taux de mortalite minimaux historiques enregistres dans la base de donnees sur la mortalite de l'OMS sont presentes dans le tableau 2 (pour les codes de pays employes, voir tableau 1). On peut voir dans le tableau 4 que ces taux minimaux ant eux-memes decline progres- sivement, surtout dans les tranches d'Age inferieures, la tendance au declin ayant ete plus prononcee pour le sexe feminin. Ces tendances se degagent egalement des don- nees des pays pris individuellement (tableau 3). Une comparaison entre le tableau des causes de deces d'un pays et celui du taux de mortalite le plus bas dans le monde revelera les causes auxquelles est attribuable la surmortalite du pays, comme ii ressort du tableau 5. Ce tableau montre egalement que meme les taux les plus bas sont imputables a des causes evitables dans une large mesure. Les taux les plus bas dans le monde peu- vent done etre consideres comme des objectifs modes- tes pour la reduction de la mortalite. Les valeurs les plus recentes du rapport de surmortalite (exprimees en pourcentages), par rapport au taux de mortalite minimaux historiques, sont presentees dans les tableaux 6 & 7 pour les pays industrialises et les pays en developpement, respectivement. Deux categories de resultats, l'une pour !'ensemble des groupes d'Age et l'autre pour les personnes de mains de 65 ans, sont presentees. Les resultats de la seconde categorie sont particulierement utiles dans la mesure ou la surmortalite dans ces groupes d'age est consideree comme evitable en grande partie. Les tendances du taux de surmortalite pour ces groupes d'Age sont presentees dans les figures 1 & 2, eu egard aux taux de mortalite minimaux histori- ques et actuels, respectivement. On peut pousser un peu plus loin la ventilation du rapport de surmortalite, par exemple par age et par sexe. Cette ventilation fait apparaitre les groupes de population spe- cifiques qui presentent dans chaque pays un rapport de surmortalite eleve et auxquels doivent s'interesser les responsables de la planification sanitaire et sociale pour trouver les moyens d'intervention selectifs appropries, ainsi que les chercheurs pour redoubler d'efforts en vue d'elucider les problemes medico-sociaux en cause. Si la ventilation par sexe illustree graphiquement dans la figure 3 pour les pays industrialises ne revele pas de tendance a une preponderance du sexe masculin ou du sexe feminin dans le rapport de surmortalite, la figure 4 montre que le sexe feminin est nettement defavorise dans les pays en developpement. II est interessant de noter que la preponderance du sexe feminin dans le rap- port de surmortalite existait egalement dans les pays industrialises ii y a quelques dizaines d'annees, comme on le voit dans la figure 5. En ce qui concerne la ventila- tion par age, le tableau 8 montre que les profils varient d'un pays a l'autre. Dans les pays en developpement, c'est la mortalite infanto-juvenile qui est a l'origine d'une grande partie de I' «excedent». Dans certains pays in- dustrialises, toutefois, c'est la mortalite des personnes d'Age moyen qui doit etre reduite par !'adoption de modes de vie plus sains. Rapp. trimest. statist. sanit. mond., 42 (1989) - 41 REFERENCES - REFERENCES 1. DEPARTMENT OF HEALTH AND SOCIAL SECURITY. Report on confidential enquiries into maternal deaths in England and Wales 1976-1978. London, HMSO, 1982. (Report on Health and Social Subjects No. 26 - Ninth Report in the series on maternal deaths). 2. RUTSTEIN. D.D. ET AL. Measuring the quality of medi- cal care, a clinical method. New England journal of medicine, 294: 582-588 (1976). 3. RUTSTEIN. D. D. ET AL. Measuring the quality of medi- cal care: revision of tables of indexes- Letter to the Editor. New England journal of medicine, 297: 508 (1977). 4. RUTSTEIN, D. D. ET AL. Measuring the quality of medi- cal care: second revision of tables of indexes- Letter to the Editor. New England journal of medi- cine, 302: 1146 (1980). 5. RuTSTEIN, D. D. ET AL. Sentinel health events (occu- pational) : a basis for physician recognition and pub- lic health surveillance. American journal of public health, 73: 1054-1062 (1983). 6. HETZEL, B. S. The use of mortality and other epide- miological information in the assessment of pre- ventable deaths. In: Hansluwka, H. et al. (eds), New developments in the analysis of mortality and cau- ses of death. Bangkok, World Health Organiza- tion/Mahidol University, 1986. 7. CHARLTON, J. R. H. ET AL. Geographical variation in mortality from conditions amenable to medical in- tervention in England and Wales. Lancet, i: 691- 696 (1983). 8. CHARLTON, J. R. H. & VELEZ, R. Some international comparisons of mortality amenable to medical inter- vention. British medical journal, 292: 295-301 (1986). 9. HOLLAND, W.W. European Community atlas of avoidable death. Oxford, Oxford University Press, 1988. (Commission of the European Communities, Health Services Research Series No. 3). 10. WOOLSEY, T. Toward an index of preventable mor- tality. Vital and health statistics. Hyattsville, Md, US Department of Health and Human Services, 1981. (Series 2, No. 85). 11. WA ALER, H. Inequalities in health: an economic com- ment. In: Smedby, H. (ed.). Inequality in health ser- vices in the Scandinavian countries. The First Scan- dinavian Symposium on Health Services Research, 13-15 November, 1980, Tampere, Finland. Scandi- navian journal of social medicine, Supplement 34: 67-72 (1980). 12. DEPARTMENT OF HEAL TH AND SOCIAL SECURITY. Inequal- ities in health: report of a research working group. London,DHSS, 1980. 13. FARR, W. Vital statistics: a memorial volume of selections of reports and writings. London, E. Stan- ford, 1885. (pp. 146-148). (Edited for the Sanitary Institute of Great Britain by Noel A. Humphreys). Wld hlth statist. quan., 42 (19891 14. GURALNICK, L. & JACKSON, A. An index of unneces- sary deaths. Public health reports, 82: 180-182 (1967). 15. SAUER, H. I. & BRAND, F. R. B. Geographic patterns in the risk of dying. The Geographic Society of America Inc., 1971. (Memoir 123). 16. INSTITUTE OF POPULATION PROBLEMS. {Future popula- tion projections for Japan, by sex and age, 1975- 2050, projected in November 1976]. Tokyo, Insti- tute of Population Problems, 1976. (Research Se- ries No. 213). (In Japanese). 17. FLANDERS, W. D. Approximate variance formulas for standardized rate ratios. Journal of chronic dis- eases, 37: 449-453 (1984). 18. WORLD HEAL TH ORGANIZATION, World health statistics annual. Geneva, WHO, 1988. (Annex I). 0RGANISA TION MONDIALE DE LA SANTE. Annuaire de sta- tistiques sanitaires mondiales. Geneve, OMS, 1988. (Annexe I). 19. CHIANG, C. L. The life table and its applications. Malabar, FI., Robert E. Krieger Publishing Co., 1984. (pp.137-147). 20. UNITED NATIONS. World population prospects: esti- mates and projections as assessed in 1982. New York. United Nations, 1985. (pp. 140-145). 21. U.S. DEPARTMENT OF HEAL TH, EDUCATION AND WELFARE. Mortality trends: age, color, and sex, United States, 1950-69. Vital and health statistics. Hyattsville, Md, US Department of Health and Human Services, 1973. (Series 20, No. 15). 22. RosENWAIKE, I. ET AL. The recent decline in mortality of the extreme aged: an analysis of statistical data. American journal of public health, 10: 107 4-1080 (1080). 23. NAM,C. B. ET AL. Causes of death which contribute to the mortality crossover effect. Social biology, 25: 306-314 (1978). 24. MANTON, K. G. Sex and race specific mortality diffe- rentials in multiple cause-of-death data. The geron- tologist, 20: 480-493 (1980). 25. WING, S. ET AL. The black/white mortality crossover: investigation in a community-based study. Journal of gerontology, 40: 78-84 (1984). 26. NANJO, Z. & KOBAYASHI, K. Cohort life tables based on annual life tables for the Japanese nationals covering the years 1891-1982. Tokyo, Nihon Uni- versity Population Research Institute, 1985. (Re- search Paper Series No. 23). 27. US DEPARTMENT OF HEAL TH AND HUMAN SERVICES. Monthly vital statistics report, 35 (6) : Supplement 2 (1986). 28. ALDERSON, M. International mortality statistics. London/Basingstroke, Macmillan Press Ltd, 1981. - 42 - VARIATIONS GEOGRAPHIQUES DE LA MORTAl,ITE DUE A DES MALADIES JUSTICIABLES D'UNE INTERYENTION ~EDICALE EN EUROPE • ~ COMMISSION DES COMMUNAUTES EUROPEENNES: ATLAS DES DECES EVITABLES Elizabeth A. Paul, Jennifer Evans, Joseph Barry, Marie-Helene Bouvier-Colle, Vera Carstairs, Luis Cayolla da Motta, Diter Hansen-Koenig, Knud Juel, Klaus Kern, Raphael Lagasse, Johan P. Mackenbach, Maria Medrano, Piero L. Morosini & George Papavangeloua Une repartition appropriee des ressources disponibles pour les soins de sante est impossible sans une evalua- tion de l'effet des soins de sante et des services qui les fournissent. Des etudes ont revele d'importantes diffe- rences dans les taux et la duree d'hospitalisation pour des maladies determinees aussi bien entre les pays qu'entre differentes regions du meme pays (1-3). Les liens entre ces differences au niveau des soins et des pratiques et la sante de la population sont difficiles a etablir. Sans moyens de mesurer la morbidite regionale et sans connaitre l'effet de differentes pratiques, ii n'est pas possible de determiner la pratique clinique optimale. Un faible taux d'hospitalisation et de brefs sejours a l'hopital peuvent traduire une incidence faible ou une atteinte mains grave, une utilisation efficace des servi- ces ou une offre de services insuffisante. Pour leur part, des taux eleves peuvent refleter une forte incidence, une atteinte plus grave, des services peu efficaces ou une offre excessive. Pour evaluer dans quelle mesure lesser- vices de soins de sante parviennent a promouvoir la sante et a reduire OU a traiter les maladies, ii faut definir des moyens de mesurer l'effet de ces services. La question de l'effet des soins de sante et des moyens de le mesurer a ete abordee par Rutstein et al. au sein du groupe de travail sur les maladies evitables et maitri- sables (Working group on preventable and manageable diseases) aux Etats-Unis d' Amerique (4). Le groupe a defini une methode permettant de mesurer la qualite des soins medicaux qui consiste a recenser les cas de mala- die et d'incapacite evitables et les deces evitables et prematures. Les soins medicaux ont ete definis au sens le plus large, s'etendant a !'application de toutes les connaissances medicales appropriees, aux services de tous les personnels medicaux et auxiliaires, aux ressour- ces des services officiels, benevoles et sociaux et a la cooperation de l'individu lui-meme. A partir de la, Rutstein et al. ont dresse une liste de maladies justi- ciables d'une intervention medicale qui permette d'eviter le deces ou l'incapacite, en identifiant pour chacune de ces maladies la nature de !'intervention appropriee. Se- lan eux, les deces et incapacites dus a ces maladies pouvaient etre consideres comme des indicateurs nega- tifs des soins de sante. Charlton et al. (5) ont etabli des cartes de la mortalite pour 14 des maladies figurant sur la liste initiale de Rutstein et al. pour les 98 zones sanitaires d'Angleterre et Galles, et ant observe des variations considerables d'une zone a l'autre. Cette experience a ete elargie par le groupe de travail de la Communaute europeenne sur les deces evitables qui a etabli I' Atlas des deces evitables (6). Selan la definition du groupe de travail, on entend par deces evitable tout deces provoque par des maladies determinees (pour des groupes d'age choisis) justicia- bles de soins medicaux qui, s'ils sont demandes et four- nis a temps, permettent d' eviter la totalite ou la quasi- totalite des deces. Evitabilite n'est pas synonyme de prevention; la notion d'evitabilite peut inclure celle de a Department of Community Medicine, United Medical and Dental Schools of Guy·s and St Thomas' Hospitals, Landres, Royaume-Uni. prevention, mais elle est plus large: par exemple on peut prevenir la tuberculose, mais en cas d'echec de la pre- vention, la maladie peut etre soignee chez les plus jeu- nes, ce qui fait que le deces d'un jeune par tuberculose peut etre considere comme evitable soit qu'on aurait pu prevenir la maladie, soit qu' on aurait pu la depister et la traiter a temps. L'enchainement qui aboutit a un deces evitable est parfois complexe. Par exemple, toute une serie de faits peuvent etre a l'origine d'un deces par cancer du col de l'uterus. II se peut que le gouvernement n · ait pas encourage ou finance les programmes de depis- tage; que les administrateurs des services de sante n · aient pas applique les programmes; que le service de depistage ait mal informe le medecin des resultats; que le medecin n'ait pas pris les mesures voulues apres un depistage; que la femme ne se soit pas presentee pour subir l'examen propose. Dans le cas des maladies justi- ciables d'une intervention, laquelle est importante pour eviter le deces, la comparaison geographique de la mor- talite peut indiquer les zones administratives dans les- quelles les services de sante eprouvent des problemes particuliers. L' Atlas des deces evitables contient des cartes et des tableaux montrant les variations geographiques locales de la mortalite par tuberculose, cancer du col de !'uterus, maladie de Hodgkin, cardiopathie rhumatismale chroni- que, maladies hypertensives et cerebro-vasculaires, ma- ladies respiratoires de l'enfant, asthme, hernie abdomi- nale, appendicite, cholelithiase et cholecystite, ainsi que de la mortalite maternelle et perinatale dans 10 pays de la Communaute europeenne. Le classement et !'analyse de ces donnees permet de comparer la mortalite entre les Etats Membres de la Communaute et a l'interieur meme des Etats. Pour trois indicateurs primaires de la prevention (cancer du poumon, cirrhose du foie et accidents de vehicule a moteur), on a presente des taux nationaux plutot que des taux par zones, car la mortalite par ces causes a plus de chances de refleter des politiques nationales que I' action des services de sante locaux. Methodes Donnees Le groupe de travail de la Communaute europeenne a examine la liste des maladies dressee par Rutstein et al. et en a retenu 17. Des Ii mites d · age bien definies ont ete choisies pour chaque maladie afin que la mortalite soit un meilleur indicateur des services de sante. Les codes CIM et les Ii mites d · age pour chaque maladie sont indiques au tableau 1. En raison de problemes de codage et de cer- tification dans certains pays, on a recueilli les donnees aussi bien pour le cancer du col de l'uterus seul que pour le cancer du col et du corps de l'uterus. Chaque groupe de maladies choisi devrait etre justiciable d'interventions bien precises assurees par des prestations de soins determines. Les details des criteres qui ant preside a ce choix sont presentes dans I' Atlas. Rapp. trimest. statist. sanit. mond., 42 (19891 - 43 - TABLE 1. DISTRIBUTION OF FIVE-YEAR STANDARDIZED MORTALITY RATIOS (SMRs) (BASED ON THE EUROPEAN COMMUNITY STANDARD) AMONG 360 ADMINISTRATIVE AREAS OF THE EUROPEAN COMMUNITY TABLEAU 1. REPARTITION DES TAUX COMPARATIFS DE MORTALITE QUINQUENNAUX (SMRJ DE 360 ZONES ADMINISTRATIVES DE LA COMMUNAUTE EUROPIEENNE (INDICE DE LA COMMUNAUTE = 100) EC death rate/ 100000 Cause of death - Cause de deces ICD·8 codes Age group SMR SMR Taux de morta· Codes CIM·8 Groupe d' age minimum maximum lite de la Communaute pour 100 OOO Tuberculosis - Tuberculose 010-019 5-64 9.5 417.7 2.05 Cancer of cervix - Cancer du col de l'uterus 180 15-64 0.0 351.6 4.89 Cancer of cervix and body uterus - Cancer du col et du corps de l'uterus 180, 182 15-54 24.1 254.8 6.23 Hodgkin's disease - Maladie de Hodgkin 201 5-64 17.6 206.5 1.25 Chronic rheumatic heart disease - Cardiopathie rhumatismale chronique 393-398 5-44 0.0 430.6 1.05 All respiratory diseases - Ensemble des maladies de l'appareil respiratoire 460-519 1-14 0.0 461.2 3.01 Asthma - Asthme 493 5-44 0.0 336.7 0.50 Appendicitis - Appendicite . . . . . . . 540-543 5-64 0.0 366.4 0.35 Abdominal hernia - Hernie abdominale . 550-553 5-64 0.0 269.0 0.35 Cholelithiasis and cholecystitis - Cholelithiase et cholecystite 574-575 5-64 0.0 294.8 0.83 Hypertensive and cerebrovascular disease - Maladies 400-404 hypertensives et cerebro-vasculaires 430-438 35-64 48.6 180.6 52.47 Maternal deaths a - Deces maternels a 630-678 all ages - 0.0 106.1 20.49 tous llges confondus Perinatal deaths b - Deces perinatals b 10.0 32.6 18.49 Minimum Maximum Cause of death - Cause de deces ICD-8 codes Age group national rate r national rate c Codes CIM-8 Groupe d'age Taux national T aux national minimumc maximumc Cancer of the trachea. bronchus and lung - Cancer de la trachee, des branches et du poumon 162 5-64 13.4 36.2 Cirrhosis of the liver - Cirrhose du foie 571 15-74 4.4 38.3 Motor-vehicle accidents - Accidents de vehicule a moteur . E810-825 all ages - 12.7 31.3 • Deaths per 100 OOO - Deces pour 100 OOO. b Deaths per 1 OOO - Deces pour 1 OOO. tous llges confondus c These causes of death are presented by country only; the rates are not age-standardized and are expressed per 100 OOO population - Ces causes de deces sont present6es par pays uniquement; les taux ne sont pas corriges de rage et sont exprimes en deces pour 100 OOO habitants. Les participants de chaque pays du groupe de travail - Belgique, Danemark, France, Grece, lrlande, ltalie, Luxembourg, Pays-Bas, Republique federale d' Allema- gne et Royaume-Uni (Angleterre et Galles, lrlande du Nord et Ecosse) - ont obtenu de leur office national de statistique des donnees sur la mortalite et la population pour la periode 1974-1978, sauf la France dont les don- nees portent sur la periode 1973-1977. Ces donnees classees selon I' age, le sexe et la zone administrative ont ete fournies au centre coordonnateur, a savoir le Depart- ment of Community Medicine de I' United Medical and Dental Schools a Landres qui s'est charge de les analy- ser. Tout en cherchant a definir des zones de quelque 500 OOO habitants, ii a fallu respecter les unites adminis- tratives existantes, ce qui fait que la population des zones varie. Celles de la Republique federale d'Allema- gne sont particulierement disproportionnees puis- qu 'elles comptent en moyenne 5 594 OOO habitants. Les chiffres de la population etaient fondes sur les recen- sements et, le cas echeant, sur des estimations offi- cielles entre deux recensements. Pour la France, le seul recensement regional disponible portait sur I' an nee 1975 et pour l'lrlande et l'lrlande du Nord sur l'annee 1976. Les donnees sur la mortalite perinatale et mater- nelle n'etaient pas disponibles pour les zones adminis- tratives en Grece. Wld hlth statist. quart., 42 ( 1989) Analyse des donnees Deux series de taux comparatifs de mortalite (standard- ized mortality ratios-SMR) ont ete calculees pour cha- que zone administrative. La premiere prend comme base Jes taux de mortalite par age de la Communaute euro- peenne; la seconde compare la mortalite de chaque zone au taux de mortalite par age du pays. Pour les tumeurs malignes du col et du corps de !'uterus, le nombre de deces a ete ajuste en fonction de la seule population feminine. On a verifie la variation des SMR pour determi- ner si elle etait statistiquement plus importante qu'une variation purement aleatoire apres avoir d'abord exa- mine si ces taux resumaient les donnees de maniere satisfaisante (7). Les centres participants ont fourni des cartes avec les frontieres administratives, et celles-ci ont ete codees pour ~tre adaptees au systeme de cartographie informa- tise Gimms (8). Trois series de cartes ont ete etablies: deux pour I' ensemble de la Communaute europeenne et pour chaque maladie, et une pour chaque pays et chaque maladie. Pour la premiere serie de cartes de la Commu- naute europeenne, les SMR des 360 zones administrati- ves calcules sur la base de l'indice 100 pour la Commu- naute ont ete ordonnes et repartis en sextiles de 60. La deuxieme serie de cartes presente les differences entre - 44 - les pays et a l'interieur des pays sur la base de l'indice de la Communaute, mais le degrade fait ressortir l'ampleur des ecarts pluteit que le rang. Les intervalles le plus eleve et le plus faible contiennent chacun environ 10% des zones, mais l'intervalle intermediaire a ete divise en quatre sous-intervalles d'egale amplitude qui ne contien- nent pas le m~me nombre de zones. Les intervalles uti- lises pour definir les degrades sont differents pour cha- que cause de deces. Quant aux cartes nationales, elles montrent les SMR calcules sur la base du taux du pays. Pour obtenir une base commune faisant apparaitre les variations a l'interieur des pays, on a regroupe les SMR, les 10% les plus eleves et les plus faibles definissant les deux degrades extr~mes et l'intervalle restant etant sub- divise en quatre sous-intervalles d'egale amplitude. Les travaux se poursuivent actuellement sur une nou- velle edition de I' Atlas pour 1980-1984, ce qui permet- tra d' effectuer des comparaisons dans le temps. L'Espa- gne et le Portugal qui ant adhere au groupe de travail apportent eux aussi leurs donnees au deuxiame Atlas. Resultats Les SMR minimaux et maximaux (sur la base de la norme de la Communaute europeenne) pour les 360 zones administratives de la Communaute et le taux de mortalite global de la Communaute pour chaque maladie sont indi- ques au tableau 1. Pour chacune des 13 causes de decas, les taux dans les zones au la mortalite etait la plus elevee atteignaient 2-4 fois la moyenne de la Commu- naute. Le tableau 2 indique les SMR nationaux sur la base de la norme de la Communaute pour chaque maladie. La mor- talite des differents pays variait selon la maladie. Chaque pays presentait une mortalite nationale superieure a la moyenne communautaire pour certaines maladies et inferieure pour d'autres. Ainsi, pour un pays presentant des taux de mortalite sensiblement plus eleves au plus faibles que la moyenne de la Communaute, la comparai- son entre zones geographiques a l'interieur du pays sur la base nationale donne une image tras differente de la comparaison sur la base de la norme de la Commu- naute. La difference entre les resultats obtenus avec la base nationale et avec la base de la Communaute apparait clairement dans la repartition de la mortalite par cancer du col de l'uterus. Les SMR maximaux et minimaux des zones administratives (selon la base nationale) pour le cancer du col de !'uterus et le SMR national (selon la base de la Communaute) apparaissent au tableau 3 qui indique egalement le pourcentage des zones du pays dont les SMR sont significativement au-dessus du niveau de la norme de la Communaute et le pourcentage de celles dont les SMR sont au-dessus du niveau national. Au niveau de la Communaute, les indices de !'ensemble des zones administratives du Danemark (SMR = 246) et !'ensemble des zones de la Republique federale d'Alle- magne (SMR = 142) sont significativement eleves, alors qu'au niveau national seuls le sont les indices de 13% des zones danoises et de 27% des zones alleman- des. L'ltalie et la Grace presentaient les indices les plus faibles au niveau de la Communaute (28 et 29 respecti- vement) et aucune de leurs zones ne depassait la moyenne de la Communaute. Dans toutes les zones du Danemark et de la Republique federale d' Allemagne la mortalite par cancer du col etait plus elevee que dans toutes les zones de l'ltalie et de la Grace, pourtant l'indice le plus eleve sur la base nationale etait celui d'une zone italienne: le taux de mortalite de l'ltalie etant de 1.4, le taux de mortalite de 4, 7 dans cette zone donnait un indice de 345. Les taux nationaux de mortalite maternelle pour 100 OOO naissances de chaque pays, ainsi que les taux maximaux TABLE 2. NATIONAL STANDARDIZED MORTALITY RATIOS {SMRs) 8 FOR THE CAUSES OF DEATH PRESENTED IN THE EUROPEAN COMMUNITY ATLAS OF AVOIDABLE DEATH {1974-1978)b TABLEAU 2. TAUX COMPARATIFS DE MORTALITE NATIONAUX (SMRJ• POUR LES CAUSES DE DECES FIGURANT DANS L'ATLAS DES DECif:S EVITABLES {1974-1978)b BL Tuberculosis - Tuberculose . . . . . . . . . . . . 91 Malignant neoplasm of cervix uteri - Tumeur maligne du col de l'uterus . . . . 75 Malignant neoplasm of cervix and body uterus - Tumeur maligne du col et du corps de l'uterus 75 Hodgkin's disease - Maladie de Hodgkin . . 103 Chronic rheumatic heart disease - Cardiopathie rhumatismale chronique . . . . . . . . . . 30 All respiratory diseases - Ensemble des maladies de I' appareil respiratoire . . . . . . . . 70 Asthma - Asthme . . . . . . . . . . . . . . 175 Appendicitis - Appendicite . . . . . . . . 53 Abdominal hernia - Hernie abdominale 69 Cholelithiasis and cholecystitis - Cholelithiase et cholecystite . . . . . . . . . . . 60 Hypertensive and cerebrovascular disease - Maladies hypertensives et cerebro-vasculaires 94 All causes (ages 5-64) - Toutes causes (5-64 ans) . . . . . . . . . . . . . . . . . . 106 All causes (ages 1+) - Toutes causes confondues (lanet+) ............. 108 Number of small areas - Nombre de zones administratives 9 LX FRG DK 59 106 40 110 142 246 98 109 160 58 94 75 60 79 40 71 95 49 0 173 53 158 159 59 19 115 38 81 136 45 127 90 66 119 104 96 114 104 95 11 15 FR 131 67 100 83 65 67 66 98 105 84 92 105 96 95 GR 140 29 77 118 115 126 32 37 47 44 93 74 88 10 IT 125 28 102 135 187 117 11 98 110 173 116 93 96 89 IR 165 81 65 95 126 176 150 111 83 36 134 110 117 8 NL 23 115 78 84 54 64 58 66 78 65 65 84 91 11 EW 54 151 98 91 87 131 151 69 94 37 107 101 104 98 NI 63 107 75 90 107 131 104 53 105 49 135 123 118 SC 86 153 96 88 134 103 181 70 99 37 151 126 115 12 ' Calculated using the age-specific death rates of the whole European Community - Calcules sur la base des taux de mortalitol corrigols de rage de 1·ensemble de la Communaute europeenne. b BL: Belgium - Belgique; LX Luxembourg; FAG: Federal Republic of Germany - Republique fedolrale d' Allemagne; DK: Denmark - Danemark; FR: France; GR: Greece - Grece; IT: Italy - ltalie; IR: Ireland - lrlande; NL: Netherlands - Pays-Bas; EW: England & Wales (UK) - Angleterre et Galles (R.-U.); NI: Northern Ireland (UK) - lrlande du Nord (R.-U.); SC Scotland (UK) - Ecosse (R.-U.). Rapp. trimest. statist. sanit. mond., 42 (1989) - 45 - TABLE 3. CERVICAL CANCER - MINIMUM AND MAXIMUM STANDARDIZED MORTALITY RATIOS (SMRs) AND AGE-STANDARDIZED DEATH RATES IN EACH COUNTRY BASED ON OWN COUNTRY STANDARD, NATIONAL SMRs BASED ON EUROPEAN COMMUNITY STANDARD AND PERCENTAGE OF AREAS SIGNIFICANTLY HIGH COMPARED WITH EACH STANDARD• TABLEAU 3. CANCER DU COL DE L'UTERUS - TAUX COMPARATIFS DE MORTALITE (SMR} MINIMUM ET MAXIMUM ET TAUX DE MORTALITE CORRIGES DE L'AGE DANS CHAQUE PAYS SUR LA BASE NATIONALE, SMR NATIONAUX SUR LA BASE DE LA COMMUNAUTE ET POURCENTAGE DES ZONES SIGNIFICATIVEMENT AU-DESSUS DES NORMES DE LA COMMUNAUTE EUROPEENNE ET NATIONALE BL LX FRG DK FR GR IT IR NL EW NI SC SMR Age-standardized rates - Taux corrige de Minimum Maximum National 56 122 75 Minimum 2.0 Maximum 4.5 110 86 197 142 73 143 246 5.9 8.8 13.7 17.2 27 177 67 21 181 29 0 345 28 70 189 81 0.0 2.8 4.7 7.5 74 134 115 40 162 151 107 73 165 153 5.5 12.4 l'age ....... . National 3. 7 5.4 7.0 12.0 0.9 5.8 3.3 0.3 2.6 1.4 1.4 4.0 4.1 2.9 7.5 12.0 5.6 7.4 5.2 7.5 % areas significantly high - Pourcentage des zones significati- vement eleve . . . Number of areas Nombre de zones European Community - Communaute europeenne National 0 22 9 100 100 27 13 11 15 0 20 95 0 10 10 0 13 89 13 13 8 27 27 11 73 28 98 67 17 12 • Calculated using the age-specific death rates of the whole European Community - Calcules sur la base des taux de monalit6 corriges de rage de !'ensemble de la Communaute europeenne. b BL : Belgium - Belgique ; LX : Luxembourg ; FRG Federal Republic of Germany - Republique federale d' Allemagne; DK : Denmark - Danemark; FR: France; GR : Greece - Grece; IT: Italy - ltalie; IR: Ireland - lrlande ;NL: Netherlands - Pays-Bas; EW: England& Wales (UK) -AngleterreetGalles (R.-U.); NI: Nonhern Ireland (UK) - lrlandedu Nord (R.-U.); SC: Scotland (UK) - Ecosse (R.-U.). TABLE 4. MATERNAL DEATH RATES - MINIMUM, MAXIMUM AND NATIONAL RATES IN EACH COUNTRY, AND PERCENTAGE OF AREAS SIGNIFICANTLY HIGHER THAN THE EUROPEAN COMMUNITY DEATH RATE AND THAN THE NATIONAL DEATH RATE• TABLEAU 4. TAUX DE MORTALITE MATERNELLE - TAUX MINIMAUX, MAXIMAUX ET NATIONAUX DANS CHAQUE PAYS ET POURCENTAGE DES ZONES SIGNIFICATIVEMENT AU-DESSUS DU TAUX DE MORTALITE DE LA COMMUNAUTE EUROPEENNE ET DU TAUX DE MORTALITE NATIONAL• BL LX FRG DK FR GR IT IR NL EW NI SC Death rates per 100 OOO - Minimum 5.8 20.6 0.0 0.0 0.0 9.8 5.2 0.0 0.0 Taux de mortalite pour Maximum 19.2 43.2 12.3 106.1 57.7 36.2 16.1 34.3 40.1 100000 ........ National 11.8 59.7 34.0 5.7 20.2 22.8 14.1 10.6 12.3 8.3 14.7 % areas significantly high European Pourcentage des Community- zones ayant un taux Communaute significativement eleve europeenne 0 73 0 15 16 0 0 1 8 National 0 18 0 15 10 25 9 8 17 Number of areas - Nombre de zones ........ 9 11 15 95 10 89 8 11 98 12 • Calculated using the age-specific death rates of the whole European Community - Calcules sur la base des taux de monalite corriges de rage de !'ensemble de la Communaute europeenne. b BL: Belgium - Belgique; LX: Luxembourg; FRG: Federal Republic of Germany - Republique federaled'Allemagne; DK: Denmark - Danemark; FR: France ;GR Greece - Grece; IT: Italy - ltalie; IR: Ireland - lrlande; NL: Netherlands - Pays-Bas; EW: England & Wales (UK) - Angleterre et Galles (R.-U.); NI: Nonhern Ireland (UK) - lrlande du Nord (R.-U.); SC: Scotland (UK) - Ecosse (R.-U.). et minimaux des zones administratives, font l'objet du tableau 4. La France presentait un taux national voisin de la moyenne communautaire alors que 15% des zones franc,:aises avaient un taux significativement superieur au taux de la Communaute et au taux national. En Republi- que federale d' Allemagne, 73% des zones etaient au- dessus du taux de la Communaute de 20 .4 pour 100 OOO alors que 18% des zones seulement depassaient le taux national de 34 pour 100 OOO. Aucune des zones irlan- daises n'avait un taux superieur au taux de la Commu- naute mais 25% depassaient le taux national. Les echelles utilisees pour le degrade des cartes de I' Atlas visent a refleter les differences selon les deux systemes. Les differents indices concernant le cancer du col de l'uterus sur la base de la Communaute sont mon- tres par la cane 1. Le degrade est fonde sur une repar- tition par sextile. Les taux comparatifs de mortalite vont Wld hlth statist. quan .. 42 (1989) de mains de 1 pour 100 OOO a 17 pour 100 OOO. Ces differences etaient statistiquement significatives et une variation significative au niveau europeen a ete consta- tee pour toutes les maladies visees. On n'a pas cepen- dant releve de variation a l'interieur de chaque pays pour toutes les maladies. Pour le cancer du col, on retrouvait des grappes de zones a taux de mortalite eleve dans certains pays et ii fallait done utiliser une autre echelle pour le degrade afin d'identifier les zones presentant un taux de mortalite particulierement eleve par rapport au reste du pays. La carte d' Angleterre et Galles (Cane 2) illustre la repartition des taux comparatifs (SMR) sur la base nationale, le degrade refletant le niveau de ces taux. Les taux comparatifs de mortalite s'inscrivaient entre 4 et 8 pour 100 000. La carte europeenne indique que la plupart des zones d' Angleterre et Galles presentaient des taux eleves de cancer du col sur la base europeenne, alors que la carte d' Angleterre et Galles fait ressortir les - 46 - MAP 1. MALIGNANT NEOPLASM OF CERVIX UTERI MORTALITY, AGES 15- 64, EUROPEAN ATLAS OF AVOIDABLE DEATH, 1974- 1978 CARTE 1. MORTALITE PAR TUMEUR MALIGNE DU COL DE L'UTERUS (15-64 ANS), ATLAS DES DECES EVITABLES, 1974-1978 ·f Greater London Greater Paris Grand Landres Grand Paris SMR a (sextile intervals - sextiles) 1 165 - 352 117-<165 83 -< l 17 52 -< 83 25 -< 52 0 -< 25 a SMR : standardized mortality ratios - taux comparatifs de mortalite . zones ou les taux sont particulierement eleves par rap- port au reste du pays . Par consequent si les cartes euro- peennes permettent d" effectuer des comparaisons entre les taux de mortalite des differents pays, les cartes natio- nales comparent la mortalite dans les differentes zones d 'un pays aux taux de mortalite pour le pays tout entier. Pour permettre aux differents pays d' etudier de plus pres leurs propres niveaux de mortalite, I' Atlas comprend, outre les cartes , des tableaux des taux comparatifs et des taux bruts de mortalite de chaque zone administra- tive . Une analyse preliminaire de !'evolution de la mortalite evitable entre 1974- 1978 et 1980-1984 fait apparaTtre pour la quasi-totalite des maladies etudiees une baisse de la mortalite dans la Communaute. Toutefois, en com- parant le rang des zones de la Communaute en 197 4- 1978 au classement de 1980- 1984, on constate que les zones a taux eleves au cours de la premiere periode ont tendance a presenter des taux eleves au cours de la seconde . C'est ce qu'illustre pour le cancer du col de !'uterus la figure 1 ou apparaissent les taux pour 1974- 1978 et pour 1980-1984 (tous deux calcules en utilisant la base de la norme de la Communaute de 1974- 1978). Le coefficient de correlation du rang des taux pour les deux periodes etait de 0 ,91. Discussion La collaboration dont les 12 pays ont fait preuve pour etablir I' Atlas des deces evitables traduit le souci inter- national d 'elaborer des moyens de mesurer l'effet des services de sante. L'idee qui consiste a considerer la mortalite comme un indice negatif de la qualite de la sante ou des soins de sante des populations n'est pas nouvelle et la mortalite perinatale sert depuis longtemps d ' indice de la qualite des soins medicaux . Les maladies figurant dans I' Atlas ont ete choisies parce qu' elles sont supposees justiciables d 'une prevention primaire, d 'une prevention secondaire ou d'un traitement medical, et la mortalite par ces maladies chez les jeunes est totalement ou en grande partie evitable. Ainsi, les taux eleves de mortalite par ces maladies peuvent indiquer d'eventuel- les carences au niveau des soins medicaux. On n'a pas retenu toutes les maladies justiciables d'une intervention medicale susceptible d'eviter le deces. II ne s 'agit pas dans I' Atlas de quantifier I' ensemble de la mortal ite evi- table toutes causes confondues mais de s'attacher a une serie de maladies qui serviront d 'indicateurs . On constate des variations frappantes entre les taux de mortalite par ces maladies en Europe . Pour la mortalite maternelle, un sixieme des zones de la Communaute presentait un taux inferieur a 8 pour 100 OOO alors que le sextile le plus eleve presentait des taux de 29 ou plus . La comparaison entre les taux de mortalite des differentes zones administratives du meme pays peut indiquer celles ou les prestateurs des soins de sante sont en mesure d 'ameliorer leurs services . Les comparaisons internatio- nales peuvent servir a indiquer des cibles specifiques, sous forme de taux de mortalite minimaux, que les ser- vices de sante peuvent s' efforcer d' atteindre. La Suede a par exemple utilise ses taux de mortalite maternelle et perinatale a cette fin . Rapp. trimest. s tatis t. sanit. mond., 42 ( 1989) - 47 - MAP 2. MALIGNANT NEOPLASM OF CERVIX UTERI MORTALITY, ENGLAND & WALES, AGES 15-64, EUROPEAN COMMUNITY ATLAS OF AVOIDABLE DEATH, 1974-1978 CARTE 2. MORTALITE PAR TUMEUR MALIGNE DU COL DE L'UTERUS, ANGLETERRE ET GALLES, 15-64 ANS, ATLAS DES DECES EVITABLES, 1974-1978 Middle of interval Milieu de Number of observations l'intervalle Nombre d'observations 0 .0 20 .0 40.0 60.0 80.0 100.0 120.0 140.0 160.0 0 0 3 -10- 21 26 18 12- a- a SMR standardized mortality ratios - taux comparatifs de mortalite . L' evolution de la mortalite evitable dans le temps peut contribuer a identifier non seulement les autorites sani- taires qui continuent d'eprouver des problemes, c'est- a-dire les zones OU la mortalite etait elevee et I' est restee, mais aussi les zones ou la mortalite baisse . Beaucoup de zones a forte mortalite par cancer du col de l'uterus en 1974-1978 ont continue de presenter des taux eleves en 1980-1984 (Fig. 1) mais dans un pet it nombre de cas la mortalite a sensiblement diminue. On a egalement constate divers types d'evolution pour les taux de mor- talite nationaux. Le Danemark qui presentait le taux de mortalite corrige de l'age le plus eleve en 1974-1978 (12 pour 100 000) I' avait ramene a 9 pour 100 OOO en 1980- 1984. L' Angleterre et Galles, et l'Ecosse, qui venaient au deuxieme rang en 1970-197 4 avec un taux de 7 pour 100 OOO, ont enregistre une reduction inferieure a 1 pour 100 OOO en 1980-1984. Si les pays ou les zones ayant enregistre une baisse de la mortalite d'une periode a l'autre ont apporte des modifications a !'organisation des soins de sante, aux politiques ou a !'affectation des ressources, leur experience pourrait indiquer des strate- gies potentielles a suivre pour reduire la mortalite . L'efficacite des soins de sante au niveau local a fait l'ob- jet d' audits cliniques et d' enquetes confidentielles dans de nombreux pays. Tout deces maternel ou perinatal fait regulierement l'objet d'une enquete dans plusieurs pays de la Communaute. Une recente enquete sur les deces consecutifs a une intervention chirurgicale en Angleterre a indique plusieurs zones ou !'organisation et la pratique pouvaient etre ameliorees (9). Wld hlth statist. quart., 42 (1989) SMR ' ! 145 - 345 121 - < 145 . 98 -< 121 74 - < 98 50 -< 74 0 - < 50 Greater London Grand Landres • Lorsque les indicateurs des deces evitables ont ete publies pour la premiere fois en Angleterre en 1983, plusieurs des zones a forte mortalite ont mene des enquetes confidentielles sur les deces par ces causes. Le groupe de travail de la Communaute sur la mortalite evi- table envisage d 'elaborer des formules types pour les enquetes confidentielles qui pourraient etre utilisees dans differents pays. La collaboration de la Communaute en vue de normaliser certaines des informations fournies par ces audits faciliterait les comparaisons des resultats entre les pays . L'information contenue dans I' Atlas peut etre utilisee par les autorites sanitaires pour surveiller leurs resultats mais elle peut aussi servir de base pour aborder certaines questions epidemiologiques fondamentales sur les ma- ladies concernees. Les ecarts de mortalite enregistres d'une zone a l'autre peuvent etre dus a des differences d'incidence OU de prevalence OU a des differences liees a la letalite. Le lien entre !'incidence et l' issue de la maladie pour certaines causes de deces evitables (par exemple la tuberculose, le cancer du col de l'uterus ou I' appendicite) pourrait peut-etre faire l'objet d'une enquete au niveau europeen car dans certains pays de la Communaute on peut obtenir des donnees relativement fiables sur la mor- bidite grace aux registres et aupres des hopitaux . Les questions de l'efficacite de differents programmes de traitement de l'asthme et le lien entre la mortalite, !'inci- dence et le traitement de l'asthme retiennent actuelle- ment !'attention et un groupe de la Communaute a recemment ete constitue dans ce domaine. - 48 - FIG.1 CERVICAL CANCER - STANDARDIZED MORTALITY RATIOS (SMRs) FOR 1974-1978 AND 1980-1984 (BASED ON THE EUROPEAN COMMUNITY 1974 STANDARD) FOR 360 HEALTH ADMINISTRATION AREAS OF THE EUROPEAN COMMUNITY (1974 BOUNDARIES) CANCER DU COL DE L'UTERUS - TAUX COMPARATIFS DE MORTALITE (SMRJ POUR 1974-1978 ET 1980-1984 (SUR LA BASE DE LA NORME DE LA COMMUNAUTE EUROPEENNE POUR 1974) DANS 360 ZONES D'ADMINISTRATION SANITAIRE DE LA COMMUNAUTE (FRONTIERES DE 1974) I,] I,] I,] I,] {/:J I,] iP I,] I,] I,] "1:1 I,] 200 " 00 "' 6 I,] 00 ~ a: :;;; I,] I,] en 100 I,] I,] 0 100 200 300 400 SMR 1974- 1978 Rank correlation= 0.91 - Coefficient de correlation du rang= 0,91 II est bien etabli que les differentes pratiques de certifi- cation des deces et de codage ont des repercussions sur les statistiques de mortalite (10-14), mais ii est malaise d'en quantifier les effets . Si d'autres etudes sont neces- saires pour quantifier les effets de fac;:on plus detaillee, l'ampleur actuellement connue du phenomene des diffe- rentes pratiques de certification et de codage ne saurait expliquer les differences de mortalite du simple au quin- tuple ou au decuple que l'on constate dans I' Atlas pour la plupart des maladies . Le groupe de travail de la Commu- naute en collaboration avec l'OMS poursuit des recher- ches sur les pratiques de codage et la certification des causes de deces evitables. Le groupe de travail de la Communaute a toujours consi- dere l'etablissement d'un Atlas comme une etape de l'etude des raisons expliquant les variations de la mor- talite et non comme une fin en soi . II ne s'agit pas d'uti- liser !'information pour critiquer tel ou tel organisme res- ponsable des soins de sante. Ce qu'on propose c'est d'utiliser les taux de mortalite eleves par certaines cau- ses de maniere constructive afin de sensibiliser les auto- rites sanitaires locales aux aspects de leurs services qui pourraient etre ameliores. II convient d' elaborer une methodologie type pour des enquetes confidentielles afin de fournir un moyen immediat et peu coOteux d'ap- profondir le probleme. RESUME Sensible a la necessite d' elaborer d'urgence des moyens de mesurer l'effet des services de soins de sante, 10 pays de la Communaute europeenne ont collabore a l'etablissement de l'Atlas des deces evitables pour la periode 1974-1978. On a retenu 17 groupes de mala- dies justiciables d 'une intervention medicale qui, a condi- tion d'etre demandee et fournie en temps utile, doit per- mettre d' eviter la totalite ou une grande partie des deces. La mortalite par ces causes a ete comparee dans 360 zones administratives de services de sante des pays participants. Pour toutes les maladies, on a constate des variations importantes de la mortalite tant a l'interieur des pays de la Communaute qu'entre eux et les autorites sanitaires sont encouragees a considerer des taux ele- ves de mortalite par ces causes comme le signe de carences eventuelles des services de soins de sante. Les travaux en vue d'une nouvelle edition de I' Atlas pour la periode 1980-1984 sont en cours . L'evolution de la mortalite evitable dans le temps pourrait indiquer les autorites sanitaires qui continuent d' eprouver des pro- blemes et celles qui ont reussi a reduire la mortalite evi- table . Rapp. trimest. statist. sanit. mond.. 42 ( 1989) - 49 - SUMMARY Geographical variations in mortality from conditions amenable to medical intervention in Europe: the European Community atlas of avoidable death The urgent need to develop measures of the outcome of health-care services has led to the collaboration of 10 countries of the European Community in the produc- tion of the European Community atlas of avoidable death (1974-1978). Seventeen disease groups were chosen for which it was considered that death within specified age groups should be either wholly or substantially avoidable when appropriate medical care is sought and provided in good time. Mortality from these causes was compared across 360 health-service administrative ar- eas in the participating countries. For all diseases there was considerable variation in mortality both within and between the countries of the European Community and it is suggested that high levels of mortality from these causes should be viewed by health authorities as warn- ing signals of potential failures of health-care services. Work is in progress on a further edition of the Atlas for the years 1980-1984. Changes in avoidable mortality over time could indicate which health authorities have persistent problems and which authorities are succeed- ing in reducing avoidable mortality. REFERENCES - REFERENCES 1. CHASSIN, M. R. ET AL. Variations in the use of medical and surgical services by the Medicare population. New England journal of medicine, 314: 285-290 (1982). 2. MCPHERSON, K. ET AL. Small-area variations in the use of common surgical procedures: an international comparison of New England, England and Norway. New England journal of medicine, 307: 1310-1314 (1982). 3. MORGAN, M. M. ET AL. Length of stay for common surgical procedures: variation among districts. British journal of surgery, 74: 884-889 (1987). 4. RuTSTEtN, D. D. ET AL. Measuring the quality of medi- cal care: a clinical method. New England journal of medicine, 294: 582-588 (1976). 5. CHARLTON, J. R. H. ET AL. Geographical variation in mortality from conditions amenable to medical in- tervention in England and Wales. Lancet, i: 691- 696 (1983). 6. HOLLAND, W.W. (ed). European Community atlas of avoidable death. Oxford, Oxford University Press, 1988. (Commission of the European Communities, Health Services Research Series No. 3). 7. GAIL, M. The analysis of heterogeneity for indirect standardized mortality ratios. Journal of the Royal Statistical Society, 141: 224-234 (1978). Wld hlth statist. quan .. 42 ( 1989) 8. WAUGH, T. C. & McCALDEN, J. Gimms reference ma- nual. Edinburgh, Gimms Ltd, 1983. 9. BUCK. N. ET AL. The report of the confidential enquiry into perioperative deaths. London, Nuffield Provin- cial Hospitals Trust, King's Fund, 1987. 10. REID. D. D. & RosE, G. A. Assessing the comparabi- lity of death statistics. British medical journal, 2: 1437-1439 (1964). 11. PERCY, C. & DOLMAN, A. Comparison of the coding of death certificates related to cancer in seven coun- tries. Public health reports, 93: 335-350 (1978). 12. PERCY, C. The international comparability of coding cancer data-present state and possible improve- ment by ICD-10. Bethesda, !ARC/National Cancer Institute, 1986. (Working paper). 13. KELSON, M. C. & HELLER, R. F. The effect of death certification and coding practices on observed dif- ferences in respiratory disease mortality in 8 EEC countries. Revue d'epidemiologie et de sante publi- que, 31: 423-432 (1983). 14. KELSON, MC. & FAREBROTHER, M. The effect of inac- curacies in death certification and coding practices in the European Economic Community (EEC) on the international cancer mortality statistics. Internatio- nal journal of epidemiology, 16 : 411-414 ( 1987). - 50 - YEARS OF POTENTIAL LIFE LOST: APPLICATION OF AN INDICATOR FOR ASSESSING PREMATURE MORTALITY IN SPAIN AND PORTUGAL Luis Alberto Garcia Rodrigueza & Luis Cayolla da Mottab Despite their known limitations, mortality rates and other indicators of mortality data are still one of the main sources of information for the planning and evaluation of national health services and activities. Even the most developed countries do not yet possess morbidity sta- tistics which are sufficiently accurate and compre- hensive at a national level. Mortality data are thus the most accessible source for the study of the relative importance of the major causes of death at both national and international levels (1-3). It is well known that the comparability of mortality data, particularly at an international level, is affected by differ- ences in diagnostic practices, coding procedures, etc. and hence one must exercise prudence when interpret- ing mortality statistics, particularly in the context of health planning. It was precisely for the purpose of improving the deci- sion-making process in health planning and evaluation that a certain number of investigators (4-9), including the authors of the present article, came to propose the determination of "premature mortality". This is the cal- culation of the number of "years of potential life lost" (YPLL), an analysis which incorporates what many call "early" or "premature" mortality in its interpretation. The up-to-date bibliography for YPLL is to be found in the first work of Garcia Rodriguez recently published on the subject (8). The method has been regularly applied over the last three or four years by the Centers for Disease Control (CDC) of the Public Health Service of the United States of America. Since 1984, CDC regularly publishes the number of years of potential life lost, between the first and the 65th birthdays, by main groups of causes of death. Methods There is no agreement yet among the investigators as regards the age or age limits to be considered for the determination of the YPLL. Some authors calculate it for the age group 1-64 years, as in the United States, while others, preferring as upper limit the "life expectancy at birth" even though this varies from country to country, propose 1-69, as the 70th birthday is nearest to the expectation of life in most of the developed coun- tries (8). For the present study, it was decided to adopt the age group 1-69 years, as the 70th birthday (the upper limit for exclusion) is close to the life expectancy at birth at the beginning of the current decade both in Spain and in Portugal. • Medical Department, Pharmacoepidemiology, Ciba-Geigy Ltd, Basel, Switzerland. b Professor of Epidemiology, National School of Public Health, Lisbon, Portugal. c Although there were more recent data available from Portugal. this was the last year with available detailed mortality data from Spain ; this year may be considered as representative of the middle 1980s in the Iberian peninsula. It was also decided to maintain as lower limit the first birthday, as in earlier studies (6-9). We believe that the probability of death in the group of < 1 year is more readily calculated and better appraised, directly and in comparative terms, by means of infant mortality rates, easily calculated and already well known. Thus in the present study, in so far as the determination of classic mortality rates is based on the total population, the YPLL are determined on the basis of the population aged 1-69 years, that is, from the first birthday to the day before the 70th birthday. Calculations are sex specific, given the well-known sex differences in mortality for several leading causes of death. The year chosen for this comparison of mortality between the two Iberian countries was 1984. c Selection of causes or groups of causes of death The causes or groups of causes of death which were chosen for this study include essentially two types of cause, classified according to the Basic Tabulation List of the International Classification of Diseases - Ninth Revision (ICD-9). The first and most important causes or groups of causes which constitute, in both absolute and relative numbers, the main causes of death in both countries are as follows (numbers in parentheses refer to items in the Basic Tabu- lation List of ICD-9) : Malignant neoplasms (08-16) Diabetes ( 181) lschaemic heart disease (27) Cerebrovascular disease (29) Pneumonia and influenza (321-322) Bronchitis and other chronic obstructive diseases of the lower respiratory tract (323) Cirrhosis of the liver (34 7) Congenital abnormalities (44) Motor-vehicle accidents (E 4 71) Other accidents (E 4 7-52/E 4 71 excluded) Suicide (E 54). The causes have been grouped, as far as possible, according to the list of causes used by the CDC in its studies of mortality (5). Some additional causes of death were also chosen which, while indisputably less frequent, were felt to be indicative of possible important deficiencies in the health systems of the countries of the Iberian peninsula at the beginning of the current decade. These causes include: Tuberculosis (02) - Meningitis and meningococcal infections (036-220) - Chronic rheumatic heart disease (251). High mortality from tuberculosis reflects deficiencies in primary health care (preventive policy). whereas Rapp. trimest. statist. sanit. mond., 42 (1989) - 51 unusually high death rates from the other two causes may be indicative of deficiencies in secondary and even in tertiary health care, i.e. clinical activities particularly. For the 14 causes of death referred to above, we have calculated the following indicators for Spain and Portugal in 1984, for each sex separately and for both sexes together: • mortality rates per 100 OOO population ; • years of potential life lost (YPLL) (crude and standard- ized); • YPLL rates per 1 OOO population (crude and standard- ized). The calculation of the years of potential life lost is obtained by multiplying the number of deaths from each cause in each of the age groups (1-4, 5-9, 10-14 ... 60- 64, 65-69). by the number of years remaining up to the 70th birthday. This number of "remaining years" is determined in each age group as the difference between the midpoint of the age group and the 70th birthday. Finally all the YPLL are summed over each age group to obtain the total value of the YPLL, for each sex and for each cause. The rationale underlying this approach is to give more prominence to deaths that occur earlier in life. The gen- eral formula for the calculation of the number of YPLL in a given population in relation to each chosen cause is as follows: YPLL = 1 D; x A; i D; = number of deaths in age group i A; = number of remaining years of life between the age group i and 70 years. Adjustment for differences in the age distribution be- tween Spain and Portugal was carried out by the method of direct standardization, yielding an age-adjusted YPLL as follows: YPLL5 = 1 A; x D; x C; i where C;, the age-correction factor, is defined for age group i as: N; = number of persons aged iyears in the actual popu- lation N;5 = number of persons aged i years in the standard population N5 = number of persons in the age group 1-69 years in the standard population. The total population of Spain and Portugal in 1984 was chosen as the standard population. Finally the rates of YPLL and YPLL5 were computed per 1 OOO population in the age group 1-69. Results The results are shown in Tables I & 2. For all causes of death, as well as for each one of the specific cause-of-death groups, Portugal shows higher values of this indicator of premature mortality than Spain. The excess of YPLL rates of Portugal over Spain is especially marked for the following causes: Wld hlth srarisr. quart .. 42 ( 1989) bronchitis and other chronic obstructive lung dis- eases (153%); tuberculosis ( 100%); pneumonia and influenza (92%); motor-vehicle accidents (86%); cerebrovascular disease (79%); cirrhosis of the liver (64%); all causes (45%). The ranking order of the main causes of premature death is different for each country, the first positions being occupied: • in Spain by: 1. malignant neoplasms; 2. accidents (total); 3. ischaemic heart disease; 4. cerebrovascular disease; and 5. cirrhosis of the liver; • in Portugal by: 1. accidents (total); 2. malignant neo- plasms; 3. cerebrovascular disease; 4. ischaemic heart disease; and 5. cirrhosis of the liver. Causes or groups of causes of death (ICD-9. Basic Tabulation L,st) 1 . Malignant neoplasms (08-16) 2. Motor-vehicle accidents (E 4 71) 3. Other accidents (E4 7-E52/E4 71 excl.) 4. lschaemic heart disease (27) 5. Cerebrovascular disease (29) 6. Cirrhosis of the liver (34 7) 7. Suicide and self-inflicted injury (E 54) 8. Pneumonia and influenza (321- 322) 9. Congenital abnormalities (44) 10. Chronic obstructive lung disease (323) Standardized rate of YPLL (per 1 OOO population) Spain (rank) Portugal (rank) 12.6 (1) 4.7 (2) 4.3 (3) 3.4 (4) 2.6 (5) 2 2 (6) 1.3 (7) 0.7 (8) 0.6 (9) 0.4 (10) 12.9 (1) 8.8 (2) 5.3 (3) 3.7 (5) 4.7 (4) 3.6 (6) 2.5 (7) 1.4 (8) 0.8 (10) 0.9 (9) Together, malignant tumours and accidents (traffic and others) are the leading causes of premature death in both Portugal and Spain. Overall, the extent of premature mortality from several other causes, including bronchitis and chronic obstruc- tive lung disease, tuberculosis, pneumonia and influenza, accidents (particularly motor-vehicle accidents). suicide, cerebrovascular diseases and congenital abnormalities is markedly lower in Spain than in Portugal. These differ- ences are depicted graphically in Fig. 1. The value of the YPLL as a policy-relevant mortality indi- cator is emphasized from a comparison of the ranking of causes of death calculated on the basis of mortality indi- ces, such as crude or age-standardized death rates. This comparison is shown in Table 3. On the basis of the YPLL (index of premature death). motor-vehicle accidents are the second cause of prema- ture death in Portugal, and if they are added to other accidents, they represent the major cause of premature death and of years of potential life lost. However, accor- ding to the usual indicator of specific mortality, motor- vehicle accidents are ranked fifth as a cause of death. The prominence of the YPLL indicator for this cause emphasizes the disproportionate number of young peo- ple among the victims of traffic accidents in Portugal. With regard to these comparisons, we must point out that in the year used for this comparative study ( 1984). the percentage of deaths ascribed to "unknown cau- ses" (unknown, unexplained or "senility") was only 3.3% in Spain but 12.3% in Portugal (that is almost four times higher than in Spain). Such a difference may affect - 52 - TABLE 1. CRUDE AND STANDARDIZED YEARS OF POTENTIAL LIFE LOST (YPLL) AND RATES, 1-69 YEARS, PORTUGAL, 1984 TABLEAU 1. ANNEES DE VIE POTENTIELLE PERDUES (VPLLJ ET TAUX CORRESPONDANTS, DONNEES BRUTES ET NORMALISEES, 1-69 ANS, PORTUGAL, 1984 ICD-9 (Bas,c Tabulation list) CIM-9 (Liste de base pour la mise en tableaux) All causes - Toutes causes . . . . . . . . . T M F Malignant neoplasms - Tumeurs malignes (08-16) T M F Motor-vehicle accidents - Accidents de vehicule a moteur (E 4 71) T M F Other accidents - Autres accidents (E47-52/E471 excl.) T M F lschaemic heart disease - Cardiopathies ischemiques (27) T M F Cerebrovascular disease - Maladies cerebro-vasculaires (29) . . T M F Cirrhosis of the liver - Cirrhose du foie (34 7) T M F Pneumonia and influenza - Pneumonie et grippe (32 1-322) T Meningitis and meningococcal infections - Meningites et infections a M F meningocoques (036. 220) . . . . . . . . T Chronic rheumatic heart disease - Cardiopathies rhumatismales chro- M F niques (251) . . . . . . T M F Congenital abnormalities - Anomalies congenitales (44) . . . . T Bronchitis and other chronic obstructive diseases of the airways - Bronchite chronique et autres maladies pulmonaires chroniques, M F obstructives (323) T Diabetes - Diabete ( 181) Tuberculosis - Tuberculose (02) .. Suicide and self-inflicted injuries - Suicide et traumatismes provoques M F T M F T M F par soi-meme (E 54) . . . . . . . T YPLL - Years of potential life lost - Annees de vie potentielle perdues. M F YPLL 590866 396 268 194 598 117 826 64426 53 401 80 554 65 393 15 161 48612 38 633 9 979 34240 26 615 7 625 42 977 25 207 17 771 32 608 23 375 9 234 13 052 8 509 4 543 3489 2 245 1 245 4 563 1 730 2 833 7 580 4100 3 394 8 388 5 578 2810 5 995 2 868 3 128 4914 3 396 918 22 698 15 595 7 103 YPLL, - Standardized years of potential life lost - Annees de vie potentielle perdues (donnees normalisees). Rate / Taux YPLL - Rate of YPLL per 1 OOO population - Taux d"YPLL pour 1 OOO habitants. YPLL, 591 004 398 343 193 846 118 219 65 200 53 124 80463 65 225 15 182 48468 38 524 9 940 34320 27 023 7 571 42 987 25424 17 657 32 741 23 838 9 163 12 999 8 492 4 537 3469 2 223 1 242 4562 1 740 2 810 7 507 4111 3 387 8 372 5 605 2 791 5 998 2 892 3 106 4931 4060 915 22 700 15 643 7 101 Rate / Taux YPLL, - Standardized rate of YPLL per 1 OOO population - Taux d"YPLL pour 1 OOO habitants (donnees normalisees). Rate - Taux YPLL 64.41 88.08 41.62 12.84 14.32 11.42 8.78 14.54 3.24 5.30 8.59 2.13 3.73 5.92 1.63 4.68 5.60 3.80 3.55 5.20 1.97 1.42 1.89 0.97 0.38 0.50 0.27 0.50 0.38 0.61 0.83 0.93 0.73 0.91 1.24 0.60 0.65 0.64 0.67 0.54 0.89 0.20 2.47 3.47 1.52 Rate - Taux YPLL, 64.43 88.55 41.46 12.89 14.49 11.36 8.77 14.50 3.25 5.28 8.56 2.13 3.74 6.01 1.62 4.69 5.65 3.78 3.57 5.30 1.96 1.42 1.89 0.97 0.38 0.49 0.27 0.50 0.39 0.60 0.82 0.91 0.72 0.91 1.25 0.60 0.65 0.64 0.66 0.45 0.82 0.10 2.47 3.48 1.52 Rapp. trimest. statist. sanit. mond., 42 ( 1989) - 53 - TABLE 2. CRUDE AND STANDARDIZED YEARS OF POTENTIAL LIFE LOST (YPLL) AND RATES, 1-69 YEARS, SPAIN, 1984 TABLEAU 2. ANNEES DE VIE POTENTIELLE PERDUES (VPLLJ ET TAUX CORRESPONDANTS, DONNEES BRUTES ET NORMALISEES, 1-69 ANS, ESPAGNE 1984 ICD-9 (Basic Tabulation List) CIM-9 (Liste de base pour la mise en tableaux) All causes - T outes causes . . . . . . . . . . . . . . T M F Malignant tumours - Tumeurs malignes (08-16) . . . . . . . . . . . T M F Motor-vehicle accidents - Accidents de vehicule a moteur (E 4 71) T M F Other accidents - Autres accidents (E4 7-52/E4 71 excl.) . . . . . . T M F lschaemic heart disease - Cardiopathies ischemiques (27) T M F Cerebrovascular disease - Maladies cerebro-vasculaires (29) . . . . T M F Cirrhosis of the liver - Cirrhose du foie (347) . . . . . . . . . . . . . T M F Pneumonia and influenza - Pneumonie et grippe (321-322) . . . T Meningitis and meningococcal infections - Meningites et infections a M F meningocoques (036, 220) . . . . . . . . . . . . . . . . . . . . . . . T Chronic rheumatic heart disease - Cardiopathies rhumatismales chro- M F niques (251) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . T M F Congenital abnormalities - Anomalies congenitales (44) . . . . . . . T Bronchitis and other chronic obstructive diseases of the airways - Bronchite chronique et autres maladies pulmonaires chroniques, M F obstructives (323) . . . . . . . . . . . . . . . . . . . . . . . . . . . . T M F Diabetes - Diabete ( 181) . . . . . . . . . . . . . . . . . . . . . . . . T M F Tuberculosis - Tuberculose (02) . . . . . . . . . . . . . . . . . . . . T Suicide and self-inflicted injuries - Suicide et traumatismes provoques M F par soi-m~me (E 54) . . . . . . . . . . . . . . . . . . . . . . . . . . . T YPLL - Years of potential life lost - Annees de vie potentielle perdues. M F YPLL 1543920 1056254 487 667 438 705 265 112 173 593 164104 125 552 38 553 149 744 118642 31 102 116 751 97 113 19 638 90833 55 846 34 987 75 471 58 573 16 898 25 811 21 246 8 787 11 823 5 889 5 934 14198 5 198 9000 21439 10680 10 760 12 508 8 357 4 152 20 762 10998 9 764 9 380 7 407 1 973 45893 35600 10 293 YPLL, - Standardized years of potential life lost - Annees de vie potentielle perdues (donnees normalisees). Rate I Taux YPLL - Rate of YPLL per 1 OOO population - Taux d'YPLL pour 1 OOO habitants. YPLL, 1 543 701 1054529 487 933 438 386 264 283 173 811 164 165 125 649 38 540 149 830 118 720 31 116 116 691 96 757 19 676 90 825 55 731 35 036 75400 58 310 16 924 25 818 17 012 8 790 11 849 5914 5 935 14191 5 178 9018 21 497 10740 10763 12 503 8 339 4156 20775 10 975 9 787 9372 7 376 1 975 45889 35 564 10301 Rate I Taux YPLL, - Standardized rate of YPLL per 1 OOO population - Taux d'YPLL pour 1 OOO habitants (donnees normalisees). Rate - Taux YPLL 44.39 60.76 28.02 12.61 15.25 9.98 4.72 7.22 2.22 4.31 6.82 1.79 3.36 5.59 1.13 2.61 3.21 2.01 2.17 3.37 0.97 0.74 0.98 0.50 0.34 0.34 0.34 0.41 0.30 0.52 0.62 0.61 0.62 0.36 0.48 0.24 0.60 0.63 0.56 0.27 0.43 0.11 1.32 2.05 0.59 Rate - Taux YPLL, 44.39 60.66 28.04 12.61 15.20 9.99 4.72 7.23 2.21 4.31 6.83 1.79 3.36 5.57 1.13 2.61 3.21 2.01 2.17 3.35 0.97 0.74 0.98 0.51 0.34 0.34 0.34 0.41 0.30 0.52 0.62 0.62 0.62 0.36 0.48 0.24 0.60 0.63 0.56 0.27 0.42 0.11 1.32 2.05 0.59 some of the specific mortality rates and YPLL rates and the above comments on the results found in this study. Therefore, it must be borne in mind that this factor can also affect comparisons between the two countries. Despite this, it is believed that the differences found and reported and the main conclusions reached will not be appreciably altered by the reduction, or even the elimi- nation, of this cause of error. Wld hlth statist. quan .• 42 (1989) - 54 - FIG.1.A STANDARDIZED YPLL RATES• IN PORTUGAL AND SPAIN, 1984 TOTAL TAUX D'YPLL NORMALISES• AU PORTUGAL ET EN ESPAGNE, TOTAL EN 1984 15 ,. 10 i ;:: " e ...J ~ > 5 12.9 12.6 • Ponugal !23 Spain - Espagne 0 Malignant neoplasms Tumeurs malignes Motor-vehicle accidents Accidents de vehicule Other accidents Autres accidents lschaemic hean disease Cardiopathies ischemiques Cerebro- vascular disease Maladies cerebro- Cirrhosis of the liver Cirrhose du foie Suicide a moteur vasculaires • Years of potential life lost per 1 OOO population - Annees de vie potentielle perdues pour 1 OOO habitants. FIG.1.B STANDARDIZED YPLL RATES• IN PORTUGAL AND SPAIN, 1984 TOTAL TAUX D'YPLL NORMALISES• AU PORTUGAL ET EN ESPAGNE, TOTAL EN 1984 " :, ., 'i e ...J ...J .. > 2 1.5 0.5 0 1.4 • Ponugal ~ Spain - Espagne Pneumonia and influenza Pneumonie et grippe Meningitis and meningococal infections Mt!ningite et infections a meningocoques Chronic rheumatic hean disease Cardiopathies rhumatismales chroniques Congenital abnormalities Anomalies congenitales Chronic obstructive diseases of the airways Diabetes Oiabete Tuberculosis T uberculose Maladies pulmonaires chroniques. obstructives • Years of potential life lost per 1 OOO population - Annt!es de vie potentielle perdues pour 1 OOO habitants. Discussion The traditional indicators of mortality (ratios or propor- tions of deaths, specific rates and even life expectancy) are becoming, with the progressive improvement in sur- vival prospects in the developed countries, less and less discriminatory. On the other hand, if the age at death is taken into account and its "prematurity" in relation to an acceptable standard emphasized (70 years in this study, to correspond with the average life expectancy in the two countries of the Iberian peninsula) one can estimate the number of potential years lost each year. This study revealed that the differences in mortality between Por- tugal and Spain become more accentuated when the specific rates are based on the YPLL measure calculated between the first and 70th birthdays. The results of this study show that, in relation to the major causes of death considered, as well as for all causes combined, Portugal occupies a distinctly less favourable position than Spain with regard to the number of years of potential life lost ( 1-69 years). This indicator, which naturally emphasizes the weight of deaths occurring at younger ages (an indicator of prema- ture mortality) is of particular interest to those responsi- ble for the establishment of policies, and the planning and evaluation of health services. Beyond this general utility, its particular interest should be emphasized with respect to the primary, or basic, health-care services, because the standardized rate of YPLL shows, better than the traditional indicators, which causes mainly contribute to premature death. For this reason the YPLL method is perhaps more applicable as information sup- port to the planning and implementation of basic health services at community level, and the selection of those which deserve priority. As demonstrated, the major cau- ses of premature death in both Spain and Portugal (and presumably in other developed countries as well) are amenable to primary prevention. Rapp. trimest. statist. sanit. mond., 42 (1989) - 55 - TABLE 3. VARIATION BETWEEN SPAIN AND PORTUGAL ACCORDING TO VPLL 8 RATES AND MORTALITY RATES TABLEAU 3. VARIATION DES TAUX D'YPLL8 ET DES TAUX DE MORTALITE ENTRE L'ESPAGNE ET LE PORTUGAL IC0-9 (Basic Tabulation List) CIM-9 (Liste de base pour la mise en tableaux) Standardized rates Taux normalises YPLL Mortality rate Taux de mortalite Portugal Spain Espagne % variation Portugal/ Spain Espagne Portugal Spain Espagne % variation Portugal/ Spain Espagne All causes - Toutes causes . . . . ........ . Malignant neoplasms - Tumeurs malignes (08-16) Motor-vehicle accidents - Accidents de vehicule a moteur (E471) .. . .. . .. .. .. ............ . Other accidents - Autres accidents (E4 7-52/E4 71 excl.) . . lschaemic heart disease - Cardiopathies ischemiques (27) Cerebrovascular disease - Maladies cerebro-vasculaires (29) Cirrhosis of the liver - Cirrhose du foie (347) ....... . Suicide and self-inflicted injuries - Suicide et traumatismes provoques par soi-meme (E 54) . . . . . . . . . . . . . . . . . Pneumonia and influenza - Pneumonie et grippe (321-322) Meningitis and meningococcal infections - Meningites et infec- 64.43 12.89 8.77 5.28 3.74 4.69 3.57 2.47 1.42 44.39 12.61 4.72 4.31 3.36 2.61 2.17 1.32 0.74 45% 2% 86% 23% 11% 79% 64% 87% 92% 9.53 157 27 24 84 239 30 10 26 7.81 174 15 18 81 124 21 7 21 21% -10% 80% 33% 4% 93% 43% 43% 24% tions a meningocoques (036, 220) . . . . . . . . . . . . . . . 0.38 0.34 10% 0% Chronic rheumatic heart disease - Cardiopathies rhumatisma- les chroniques (251) . . . . . . . . . . . . . . . . . . . . . . 0.50 0.41 0.62 22% 34% 3 6 5 5 -40% 20% Congenital abnormalities - Anomalies congenitales (44) . . . 0.82 Bronchitis and other chronic obstructive diseases of the airways - Bronchite chronique et autres maladies pulmonaires chroni- ques obstructives (323) . . . . . 0. 91 0.36 0.60 0.27 153% 8% 100% 21 14 4 15 22 3 40% -37% 33% Diabetes - Diabete ( 181) 0.65 Tuberculosis - Tuberculose (02) 0.45 • YPLL = years of potential life lost - annees de vie potentielle perdues. Some of the differences between specific mortality rates in both countries are accentuated, or even revealed, by this new analytical approach. This is the case, for exam- ple, with meningococcal infections, which show identi- cal specific rates of mortality in Spain and Portugal, but are somewhat more important in Portugal than in Spain (about 12% higher) when evaluated on the basis of YPLL. More marked differences are apparent for tuber- culosis, for which the specific mortality rate in Portugal in 1984 was 33% higher than in Spain, whereas the diffe- rential increased to 100% when the comparison was made on the basis of the YPLL rate. The relatively unfavourable position of Portugal compar- ed to Spain with regard to premature death, is also appa- rent for other causes of death, such as cirrhosis of the liver (64% higher in Portugal), chronic bronchitis and other chronic obstructive lung diseases (153% higher), cerebrovascular diseases (79% higher), suicides (87% higher) and accidents, especially motor-vehicle acci- dents for which the YPLL rate is 86% higher in Portugal than in Spain. Although many other factors- non-medical and outside the direct responsibility of the health sector-contribute to the predominance of "young" deaths for such causes of death as accidents, suicide, and even cirrhosis of the liver, one must not lose sight of the basic role of the primary health care network in the prevention and reduction of premature mortality. The use of indices such as the YPLL to evaluate the importance of these causes of death is a vital information support to the implementation of primary health care strategies. SUMMARY In this article, the authors present an analysis of causes of death in Spain and Portugal in 1984 based on a cal- culation of the "years of potential life lost" (YPLL) between the first and the 70th birthdays, the latter age corresponding approximately to the average life expec- tancy in both countries. This analysis of the YPLL led to a substantially different ranking of the main causes of death, based on what might be termed "premature mortality" compared with that obtained from more conventional mortality indices. According to this criterion, which is especially appro- priate for the planning and evaluation of health interven- tions, the main causes of premature death (1-69 years) in the two countries of the Iberian peninsula are malignant tumours and, particularly in Portugal, violent deaths (es- pecially motor-vehicle accidents, but also suicides). This is in contrast to the predominance of cardiovascular dis- Wld hlth statist. quart., 42 ( 1989) eases indicated by other weightings of age-specific mor- tality rates. Portugal shows significantly worse YPLL rates than Spain not only for general mortality (45% higher than in Spain), but also for several major groups of causes. In Spain only malignant neoplasms, diabetes and chronic rheumatic heart diseases show higher speci- fic mortality rates than in Portugal, based on traditional mortality indicators. The authors comment on these results and discuss the value and significance of the differences found, taking into account possible causes of error, such as the pro- portion of deaths due to "senility or unknown causes", which, in the mid-1980s and at the end of the last decade, were responsible for 3.3% of deaths in Spain and 12.3% in Portugal. Despite this and other limitations, the authors consider that the differences revealed by the analysis of the YPLL are largely real and emphasize the - 56 - role of primary health care interventions, since the main causes of premature death, especially in Portugal, are susceptible to better health promotion and health ser- vices, targeted, for example, at tuberculosis, cirrhosis of the liver. motor-vehicle accidents, suicide and chronic obstructive diseases of the lungs. Interventions designed to control these causes of death have implications beyond the health sector and must be integrated into a multidisciplinary approach to primary health care as the most efficient strategy for the preser- vation and promotion of individual, family and commu- nity health. RESUME Annees de vie potentielle perdues: application d'un indicateur permettant d' evaluer la mortalite prematuree en Espagne et au Portugal Les auteurs de I' article procedent a une analyse des cau- ses de deces en Espagne et au Portugal en 1984, sur la base du calcul des « annees de vie potentielle perdues » (years of potential life lost- YPLL) entre le premier et le 70° anniversaire, ce dernier correspondant approximati- vement a I' esperance de vie moyenne dans les deux pays. L'analyse des YPLL a entraine un classement sensible- ment different des principales causes de deces, sur la base de ce que I' on pourrait appeler la « mortalite prema- turee », par rapport au classement obtenu sur la base des indices de mortalite traditionnels. Selan ce critere, qui est particulierement adapte a la planification et a I' eva- luation des interventions sanitaires, les principales cau- ses de deces prematures ( 1-69 ans) dans les deux pays de la peninsule iberique sont les tumeurs malignes et, en particulier au Portugal, les marts violentes (notamment les accidents de vehicule a moteur, mais aussi les suici- des). Cela contraste avec la predominance des maladies cardio-vasculaires etablie d'apres d'autres ponderations des taux de mortalite en fonction de l'ilge. Les taux d'YPLL pour le Portugal sont sensiblement plus eleves que pour l'Espagne, non seulement en ce qui concerne la mortalite generale (de 45% plus elevee qu'en Espagne). mais aussi pour les principaux groupes de causes de deces. En Espagne, seuls les taux de mortalite par tumeurs malignes, diabete et cardiopathies rhumatisma- les chroniques sont plus eleves qu'au Portugal, si l'on utilise les indicateurs de mortalite traditionnels. Les auteurs commentent ces resultats et etudient la signification des differences qu'ils ont mises en eviden- ce, en tenant compte des causes d'erreur possibles, tel- les que la proportion de deces dus a « la senilite OU des causes inconnues » qui, au milieu des annees 80 et a la fin de la derniere decennie, etaient responsables de 3,3% des deces en Espagne et de 12,3% au Portugal. Cette restriction mise a part, les auteurs considerent que les differences revelees par I' analyse des YPLL sont dans une large mesure reelles et soulignent le role des inter- ventions de soins de sante primaires, puisque les prin- cipales causes de deces premature, notamment au Portugal, sont justiciables d'une meilleure promotion de la sante et de meilleurs services de sante diriges, par exemple, contre la tuberculose, la cirrhose du foie, les accidents de vehicule a moteur. le suicide et les maladies pulmonaires chroniques obstructives. Les interventions destinees a prevenir ces causes de deces ont des repercussions au-dela du secteur de la sante et doivent etre integrees a une approche multidis- ciplinaire des soins de sante primaires comme etant la strategie la plus efficace pour proteger et promouvoir la sante de l'individu, de la famille et de la communaute. REFERENCES - REFERENCES 1 . COMMISSION OF THE EUROPEAN COMMUNITIES. European Community atlas of avoidable death 1980-1984. (In preparation). 2. KELSON. M. C. & HELLER. R. F. The effect of death cer- tifications and coding practices on observed differen- ces in respiratory disease mortality in 8 EEC coun- tries. Revue d'epidemiologie et de sante publique, 31 (4): 423-432 ( 1983). 3. REID, G. A. & ROSE. D. D. Assessing the comparability of mortality statistics. British medical journal, 2: 1437-1439 (1964). 4. RoMEDER.J. M. &McWHINNIE.J. R. Ledeveloppement des annees potentielles de vie perdues comme indi- cateur de mortalite prematuree. Revue d'epidemiolo- gie et de sante publique, 26: 97-115 (1978). 5. CENTERS FOR DISEASE CONTROL. Premature mortality in the United States: public health issues in the use of years of potential life lost. Morbidity and mortality weekly report, 35 (supplement 250) : 1-11 ( 1986). 6. CA YOLLA DA MOTT A, L. [Premature deaths and preven- tive intervention strategies]. 0 medico, 112 (1723): 54 (1985). (In Portuguese). CA YOLLA DA MOTT A, l. [Les deces prematures et les strategies d'intervention preventive]. 0 medico, 112 (1723): 54 (1985). (En portugais). 7. CAYOLLA DA MOTTA. L. & SEOUEIRA, M. L. [Premature deaths in Portugal by main causes (1971-1983)]. Lisbon, Ministry of Health, 1986 (Saude 2000 No. 1). (In Portuguese). CAYOLLA DA MOTTA. L. & SEQUEIRA. M. L. {Deces pre- matures au Portugal par causes principales ( 19 71- 1983) ]. Lisbonne, Ministere de la sante, 1986. (Saude 2000 N° 1). (En portugais). 8. GARCIA RODRIGUEZ. L. A. {Evolution of potential years of life lost in Spain in 1951-1978]. Alicante, Univer- sity of Alicante, 1986. (Faculty of Medicine thesis). (In Spanish). GARCIA RODRIGUEZ. L. A. {Evolution des annees poten- tielles de vie perdues en Espagne en 1951-1978]. Alicante. Universite d • Alicante. 1986. (These de la Faculte de medecine). (En espagnol). 9. GARCIA RODRIGUEZ. L. A. ET AL. [Potential years of life lost: a way of evaluating premature deaths]. Medicina clinica, 81: 55-57 (1986). (In Spanish). GARCIA RODRIGUEZ. L. A. ET AL. [Les annees potentielles de vie perdues: une fac;:on d' evaluer les deces prema- tures ]. Medicina clinica, 81: 55-57 (1986). (En espa- gnol). Rapp. trimest. statist. sanit. mond .. 42 (1989) ' PUBLICATIONS OF THE WORLD HEALTH ORGANIZATION 1989 WORLD HEAL TH FORUM An international journal of health development (Separate editions in English, French, Spanish, Arabic, Chinese and Russian) U-'orld health forum is a quarterly journal for policy-makers, health planners, adminis- trators, health educators, and public health workers ofall kinds. It provides a medium for the presentation and discussion of new concepts in public health and new approaches to health problems, and it is devoted to the improvement of health through the promotion of health services covering the entire population and the undertaking of a wide variety of public health measures, whether or not they are supported by WHO. The Forum is the main organ of WHO's Member States for the international e.changc of health infonnation and an active instrument in technical cooperation among developing countries. Subscription (4 numbers) Price per copy BULLETIN Sw. fr. 55.- Sw. fr. 16.- OF THE WORLD HEALTH ORGANIZATION The Bulletin is the principal scientific organ of WHO; its role is to review progress in medical and related sciences ("update" articles) and to bring to light new knowledge by publishing original papers on scientific research in the laboratory and field. The Bulletin contains original articles in either English or French. with a summary in the other language. One volume is published annually, consisting of six numbers. Subscription (6 numbers) Price per copy .. WEEKLY EPIDEMIOLOGICAL RECORD (Bilingual : English and French) Sw.fr.130.- Sw. fr. 22.- Prepared for the guidance of health administrations and health authorities, the Weekl.v epidemiological record contains notifications made under the International Health Regula- tions and information concerning their application. The Record also contains epi- demiological information on communicable diseases of international importance. Annual subscription Sw. fr. 150. - WORLD HEAL TH STATISTICS QUARTERLY The H/or!d health .'ltat1s111 , ,,: .. 1r1('r/\,rcplaces (since 1978) the ivorld health statL!itics report (published since 1967) and 11, l<1rcrunner the f.'p1dem10/ogital and ,,.ital statatic.'I report (published since 1947). It dcah ,.,,h the detailed analysis of selected health topics of current interest. The Quarterly contain~ articles in etther French or English with a summary in both languages. Annual subscription Price per copy . WORLD HEALTH STATISTICS ANNUAL (Bilingual: English and French) Sw. fr. 85.- Sw. fr. 23.- Tht~ forerunner of this series was the Annual epidemwlngical report of the League ,f Na11ons. It wa~ followed by the Annual epidem10/ngica/ and ~·aa/ statLstLcs issued by the World Health Organization. Latest publications: 1984. Vital statistics and life tables. morbidit~. causes of death (single volume). 402 pages . Sw. fr. 40. - 1985. Vital statistics and life tables. environmental health, causes of death (single volume), 531 pages Sw. fr. 80.- 1986. V,tal statistics and life tables, evaluation or the Global Strategy for Health for All, causes of death (single volume). 692 pages . Sw. fr. 90.- 1987. Vital statistics and life tables, oral health and care. causes of death (single volume). 455 pages Sw. fr. 90. - 1988. Vital statistics and life tables, the health professions in the 1980s. causes of death (single volume), 513 pages Sw. fr. 90. - PUBLICATIONS DE L'ORGANISATION MONDIALE DE LASANTE 1989 FORUM MONDIAL DE LA SANTE Revue internationale de developpement sanitaire (Editions sepan!es en franc;ais. anglais, espagnol, arabe, chinois et russe) Forum mondia/ de la santt! est une revue trimestrielle destinCe aux responsables des politiques sanitaires. aux planificateurs, administrateurs et educateurs sanitaires, enfin aux travailleurs de la santC publique de toutes catCgories. Tribune pour la presentation et la discussion de nouveaux concepts en santC publiquc et de nouvelles approches des problemes de sante, Forum se consacrc a l'amClioration de la sante par la promotion de services de santC couvrant la population tout entiCre et d'une vaste gamme de mesures de sante publique, qu'elles soient ou non soutenues par l'OMS. II est le principal organe a la disposition des Etats Membres de l'OMS pour l'Cchange international d'infonnations sanitaires en mCme temps qu'un instrument de cooperation technique entrc pays en developpement. Abonnement (4 numeros) Le numero BULLETIN Fr. s. 55.- Fr. s. 16.- DE L'ORGANISATION MONDIALE DE LA SANTE Le Bulletin est le principal organe scientifique de l'OMS; ii a pour role de passer en revue les progri:s des sciences medicales et apparentees (articles de la rubrique « Le point») et de mettre en lumiCre les connaissances nouvelles en prCscntant dans des aniclcs originaux les rCsultats de recherches scientifiques au laboratoire et sur le terrain. Le Bulletin contient des articles originaux en fran<;ais ou en anglais, accompagnes d'un resume dans l'autre langue. Un volume est publiC annuellement, Ctant formC de six numCros. Abonnement (6 numeros) Le numCro ..... . RELEVE EPIDEMIOLOGIQUE HEBDOMADAIRE (Bilingue: franc;ais et anglais) Fr. s. 130.- Fr. s. 22.- Destine aux administrations sanitaires et aux services de santC, le Rele~·e ep;demio/ogique hebdomadairecontient lcs notifications exigtts par le RCglement sanitaire international, ainsi que d'autres renseignements concernant l'application de ce rCglement. Le Re/et,'e contient Cgalement des informations CpidCmiologiques concemant les maladies transmissibles d'une importance intemationale. Prix de l'abonnement annuel . Fr. s. 150.- RAPPORT TRIMESTRIEL DE ST A TISTIQUES SANIT AIRES MONDIALES Le Rappor1 trimestriel de .statisllques sanitaires mondw.le.s, rem place (depuis L 978) le Rapport de JtatHtiques sanitaires mondiales (publif depuis 196 7) et son precurseur le Rapport epidhmo/og1quc•et di!mographique(publie depuis 194 7). II presente des analyses detaillees sur des sujets sJ)Cciliques d'intfrCt courant. Le Trinu•strie/ prCsente des articles originaux en fran~ais ou en anglais, accompagnCs d'un resume dans les deux langucs. Prix de l'abonnement annucl Le numCro ANNUAIRE Fr. s. 85.- Fr. s. 23.- DE ST A TISTIQUES SANIT AIRES MONDIALES (Bilingue: fran~ais et anglais) Cet annuain.· rcmplace les Statistiques l;fudt!mio/ogiques et demugraphiques annuel/es publifes par !'Organisation mondiale de la Santi! et qui avaient. elles-memes, remplacC le Rapport £:p,demwlogique annue/ publiC par la Societe des Nations. Publications rCccntes: 1984, Mouvcments de la population et tables de survie, morhiditC. causes de dCcCs ( I scul volume), 402 pages Fr. s. 40. - 1985. Mou'vemcnt de la population et tables de sunic, salubritC de l'environncment, causes de deces (I seul volume), 531 pages Fr. s. 80.- L 986. Mouvcment de la population et tables de survie. evaluation de la Strategic mondialc de la sante pour tous, causes de deces (I scul volume), 692 pages . Fr. s. 90.- 1987, Mouvcment de la population et tables de survie. santi: et !i.oins bucco-dentaircs, causes de deces ( I seul volume), 455 pages Fr. s. 90. - 1988. Mouvcment de la population et tables de survie. la situation des professions de sante dans les annccs 80. causes de dcces ( I seul volume), 513 pages Fr. s. 90. - WHO publications may be obtained, direct or through booksellers, from: Les publications de l'OMS peuvent etre commandees, soit directement, soit par l'intermediaire d'un libraire, aux adresses suivantes: A LG ERIE: Entreprise nationale du Livre (ENAL). 3 bd Zirout Youcef, ALGER ARGENTINA: Carlos Hirsch SRL, Florida 165, Galcrias Gucmes. Escritorio 453/465, BUENOS AIRES AUSTRALIA: Hunter Publications, SSA Gipps Street, COLLINGWOOD. VIC 3066 AUSTRIA: Gerold & Co., Graben 31, 1011 WIEN I BAHRAIN: United Schools International, Arab Region Office, P.O. Box 726, BAHRAIN BANGLADESH: The WHO Programme Coordinator, G.P.0. Box 250, DHAKA 5 BELG[QUE: Pour toute commande hors abonnement: Office International de Libra1rie s.a., avenue Marn ix 30, 1050 BRUXELLES. Abonnements: Office International des PCriodiqucs, avenue Louise 485, 1050 BRUXELLES BHUTAN: see India. WHO Regional Office BOTSWANA: Botsalo Books I Pty) Ltd., P.O. Box 1532. GABORONE BRAZIL: Centro Latinoamericano de lnforma<;:3.o em Cicnciasde Saude (BIREME), Organizaciio Panamerica- na de Saude, Sector de Publicacoes, C.P. 20381 - Rua Botucatu 862, 04023 Sii.o PAULO, SP BURMA: see India, WHO Regional Office CAMEROON: Cameroon Book Centre, P.O. Box 123, South West Province, VICTORIA CANADA: Canadian Public Health Association, 1565 Carling Avenue, Suite 400, OTTAWA, Ont. KIZ SRI. (Tel: (613) 725-3769. Telex: 21-053-3841) CHINA: China National Publications lmpon & Expon Corporation, P.O. Box 88, BEIJING (PEKIN) DEMOCRATIC PEOPLE'S REPUBLIC OF KOREA: see India, WHO Regional Office DENMARK: Munksgaard Book and Subscription Ser- vice, P.O. Box 2148, 16!0 K0BFNHAVN K (Tel: +45 I 12 85 70) ESP AGNE: Comercial Atheneum S.A., Consejo de Ciento 130-136, 08015 BARCELONF Genera! Moscard6 29, MADRID 20 - Librcria D1a1 de Santos, P.0. ~--:. ~,0.:'.J 28006 MADRID: Balmes 417 y 419, 08022 BARL'E- LONE FIJI: The WHO Representati.e, P.O. Box 113, SUVA FINLAND: Akateeminen Kirjakauppa, Keskuskatu 2, 00101 HELSINKI 10 •·RANCE: Arnette, 2, rue Casimir-Delavigne, PARIS 75006 GERMAN DEMOCRATIC REPUBLIC: Buchhaus Leip- zig, Postfach 140, 701 LEIPZIG GERMANY, FEDERAL REPUBLIC OF: Govi-Vcrlag GmbH, Ginnheimerstrasse 20, Postfach 5360, 6236 ESCHBORN - Buchhandlung Alexander Horn, Kirch- gasse 22, Postfach 3340, 6200 WIESBADEN GRECE: G.C. Eleftheroudakis S.A.. Librairie inter- nationale, rue Nikis 4, 105-63 ATHENES HONG KONG: Hong Kong Government Information Ser- vices, Publication (Sales) Office, Information Services Depanment, No. I, Battery Path, Central, HONG KONG HUNGARY: Kultura P.0.B. 149, BUDAPEST 62 ICELAND: Snaebjorn Jonsson & Co., Hafnarstraeti 9, P.O. Box 1131, IS-IOI REYKJAVIK INDIA: WHO Regional Office for South-East Asia, World Health House, lndraprastha Estate, Mahatma Gandhi Road, NEW DELHI 110002 !RAN (ISLAMIC REPUBLIC OF): Iran University Press, 85 Park Avenue, P.O. Box 54/551, TEHERAN IRELAND: TDC Publishers, 12 Nonh Frederick Street, DUBLIN I (Tel: 744835-749677) ISRAEL: Heiliger & Co., 3 Nathan Strauss Street, JERU- SALEM 94227 ITALIE: Edizioni Minerva Medica, Corso Bramante 83- 85, 10126 TURIN: Via Lamarrnora 3, 20100 MILAN: Via Spallanzani 9, 00161 ROME JAPAN: Maruzen Co. Ltd, P.O. Box 5050. TOKYO Inter- national 100-31 JORDAN: Jordan Book Centre Co. Ltd., University Street, P.O. Box 301, (AI-Jubeiha), AMMAN KENYA: Text Book Centre Ltd, P.O. Box 47540, NAIROBI KUWAIT: The Kuwait Bookshops Co. Ltd, Thunayan Al- Ghanem Bldg, P.O. Box 2942, KUWAIT LAO PEOPLE'S DEMOCRATIC REPUBLIC: The WHO Representative, P.O. Box 343, VIENTIANE LUXEMBOURG: Librairie du Centre, 49 bd Royal, LUXEMBOURG MALAYSIA: The WHO Representative, Room I 004, I Oth Floor, Wisma Lim Foo Yong (formerly Fitzpatrick's Building), Jalan Raja Chulan, KUALA LUMPUR 05-10; P.O. Box 2250, KUALA LUMPUR 01-02 - Parry's Book Center, 124-1 Jalan Tun Sambanthan, P.O. Box 10960, 5fi'd0 KUALA LUMPUR MALDIVES: see India. WHO Regional Office MAROC: Editions La Pone, 281 avenue Mohammed V, R•.JAT ",'IEXICO: Libreria Interacademica S.A., Av. Sonora 206, 06100-MEXICO, D.F. MONGOLIA: see India, WHO Regional Office NEPAL: see India, WHO Regional Office NETHERLANDS: InOr-Publikatics, P.O. Box 14, 7240 BA LOCHEM NEW ZEALAND: New Zealand Government Printing Office, Publishing Administration, Private Bag, WEL- LINGTON: Walter Street, WELLINGTON: World Trade Building, Cubacade, Cuba Street, WELLINGTON. Go,- ernment Bookshops at: Hannaford Burton Building, Rutland Street, Private Bag, AUCKLAND: 159 Hereford Street, Private Bag, CHRISTCHURCH: Alexandra Street, P.O. Box 857, HAMILTON: T & G Building, Princes Street, P.O. Box 1104, DUNEDIN - R. Hill & Son Ltd. Ideal House, Cnr Gillies A venue & Eden St., New-Market. AUCKLAND I NORWAY: Tanum - Karl Johan A.S., P.O. Box 1177, Sentrum, N-0 I 07 OSLO I PAKISTAN: Mirza Book Agency, 65 Shahrah- E-Quaid E-Azam, P.O. Box 729, LAHORE 3 PAPUA NEW GlllNEA: The WHO Representative, P.O. Box 646, KONEDOBU PHILIPPINES: World Health Organization, Regional Office for the Western Pacific, P.O. Box 2932, MANILA: National Book Store Inc., 701 Rizal Avenue, P.O. Box 1934, MANILA PORTUGAL: Livraria Rodrigues, 186 Rua do Ouro, LISBOA 2 REPUBLIC OF KOREA: The WHO Representative, Cen- tral P.O. Box 540, SEOUL SAUDI ARABIA: World of Knowledge for Publishing and Distribution, P.O. Box 576, JEDDAH SINGAPORE: The WHO Representative, 144 Moulmein Road, SINGAPORE 1130; Newton P.O. Box 31, SINGA- PORE 9122 SOUTH AFRICA: Contact ma1or book stores SRI LANKA: see India, WHO Regional Office SUISSE: Medizinischer Verlag Hans Huber, Uinggass- strasse 76, 3012 BERN 9 SWEDEN: For books: Aktiebolaget C.E. Fritzes Kung!. Hovbokhandel, Regeringsgatan 12, 103 27 STOCK- HOLM. For periodicals: Wennergren-Williams AB, Box 30004, !04 25 STOCKHOLM THAILAND: see India, WHO Regional Office UNITED KINGDOM: H.M. Stationery Office: 49 High Holborn, LONDON WCI V 6HB; 71 Lothian Road, EDINBURGH EH3 9AZ: 80ChichesterStreet, BELFAST BTI 4JY: Brazennosc Street, MANCHESTER M60 8AS; 258 Broad Street, BIRMINGHAM BI 2HE; Southey House, Wine Street, BRISTOL BS I 2BQ. All mail orders should be sent t~: HMSO Publications Centre. 51 Nine Elms Lane, LONDON SW8 SDR UNITED STATES OF AMERICA: Copies ofindi,idual publicatfrms (not subscnptwns): WHO Publications Cen- ter USA, 49 Sheridan Avenue, ALBANY, NY 12210. Subscrtptwn orders and correspondence concermng sub- scriptions should he addressed to the World Health Or- ganization. Distribution and Sales Service, 12 L I GENE- VA 27, Switzerland. Publications are also a,ai/ab/e from the United Nations Bookshop, NEW YORK. NY IOOl 7 (retail only) USSR: For readers in the USSR requlring Russlan editwns: Komsomolskij prospekt 18, Mcdicinskaja Kniga. MOS- COW - For readen outslde the USSR requiring Russian edllions: Kuzncckij most 18, Mei:dunarodnaja Kniga, MOSCOW G-200 VENEZUELA: Libreria Medica Paris, Apanado 60.681, CARACAS 106 YOUGOSLAVIE: Jugoslovenska Knjiga, Terazije 27/ll, 11000 BEOGR~D ZIMBABWE: Textbook Sales (PVT) Ltd, I Norwich Union Centre, \!UTARE Special terms for developing countries are obtainable on application to the WHO Representatives or WHO Regional Offices listed above or to the World Health Organization, Distribution and Sales Service, 1211 Geneva 27. Switzerland. Orders from countries where sales agents have not yet been appointed may also be sent to the Geneva address, but must be paid for in pounds sterling, US dollars, or Swiss francs. Unesco book coupons may also be used. Prices are subject to change without notice. Des conditions speciales sent consenties pour les pays en developpement sur demande adressees aux Representants de l'OMS ou aux Bureaux regionaux de l'OMS enumeres ci-dessus ou bien a !'Organisation mondiale de la Sante, Service de Distribution et de Vente, 1211 Geneve 27, Suisse. Dans les pays ou un depositaire n'a pas encore ete designe, les commandes peuvent etre adressees egalement a Geneve, mais le paiement doit alors etre effectue en francs suisses, en livres sterling ou en dollars des Etats-Unis. On peut egalement utiliser les bans de livres de l'Unesco. Prix sujets a modification sans preavis. 8/1/89 Prix: Fr. s. 23. -
Всемирная организация здравоохранения (ВОЗ / WHO) · Journal articles
Health for all: situation assessment = Santé pour tous : le bilan [full issue]
Открыть оригинал документа
Полный текст размещён на сайте публикующей организации. lawenc.com индексирует метаданные и ведёт на официальный источник.
Полный текст