Community-based health reporting / Wolf Scott

- 26 - COMMUNITY-BASED HEALTH REPORTING Wolf Scotta This article discusses systematic data collection at the local level, the principal purpose of which is to supple- ment national-level information. Another purpose, to supply data for the use of the local population, is dis- cussed in the final section. The Development Monitoring Service at the Local Level (OMS) attempts to highlight a small and as far as possible representative sample of local areas for close analysis, and to do so systematically and continuously to obtain a picture of change over time. While the idea is based on the occasional ad hoe com- munity study, the distinction between it and the OMS is the latter's more consistent approach, including features such as relative representativeness and continuity. Health variables figure prominently, but they take their place within a wider range of data comprising other socioeconomic variables (e.g. income, education, hous- ing, water, sanitation, electricity), as well as the appro- priate economic and social structure. Use of a small area as the unit of analysis makes it possible to study the interrelationships of certain variables that are not readily captured in national surveys, in particular those with a geographic orientation. On the other hand, the sample of local areas is too small to give representative data at the national level. In this way national and local data com- plement each other. Background A pilot study to test the modalities of a OMS at the local level was set up by the United Nations Research Institute for Social Development (UNRISD) in the late 1970s in Kerala (India). The government of Indonesia, jointly with UNICEF, followed suit, as did the state government of Uttar Pradesh in India. The government of the southern region of the Sudan planned a similar activity at the insti- gation of UNICEF. A first phase which consisted of a regionwide survey of population and social infrastruc- ture was completed in Sudan in the early 1980s. Political events have since frustrated implementation of the prin- cipal phase. This article is concerned with the OMS in Kerala, where in spite of a financially-checkered career the structure of the scheme is most clearly seen and first results are available (1, 2). Within the OMS, 12 socio- economic observation areas were set up in 1978/79 by the Kerala Statistical Institute, with financial support from UNRISD. Work in nine of the areas lapsed after two years, but since 1986 the state government of Kerala has supported the scheme financially within its current five-year development plan. Altogether 15 areas are now covered. For three years the World Health Organi- zation financed a special study of factors associated with nutritional status as expressed by weight and height of children under 10 in three of the areas. • The author is a former Senior Research Officer. United Nations Research Institute for Social Development, Palais des Nations, Geneva. The views expressed here are his own and do not necessarily reflect those of the Institute. Structure of the DMS A sample of local areas is selected, as far as possible representative of varying local conditions (socio- economic observation areas). In each selected area data are collected from house- holds in respect of characteristics of households and individuals, from key respondents in respect of the area as a whole (on services, economic and social infrastructure, natural catastrophes likely to affect development, etc.). In addition, studies of specific problems are conducted from time to time in the same areas. The intention is to integrate data from these various sources in analysis and reporting. The system is continuous in the sense that data are collected each year so as to have a consistent record of change over time Paid local staff are used, subject to training, supervi- sion and control as regards data quality, in contrast to systems where interviewers are sent from the centre to the local areas for each survey. Where feasible, participation of the local community is obtained through full consultation with the local councils. The fullest possible coordination between users and producers of the data is sought. The following statistics have been collected in Kerala : • information about each area, collected from key re- spondents; • information by short individual interview from a sam- ple of about 2 OOO households per area, namely: household size, type of house, whether there is elec- tricity, kinds of sanitation and sources of drinking- water, number of days of illness during the week pre- ceding interview, births, deaths (including infant deaths during the previous year), education, religion, caste of head of household, household income; • for a subsample of about 200 households per area more detailed information is obtained (again by inter- view) on the demographic and socioeconomic charac- teristics of household members, possession of house- hold durables, livestock, land-holding, expenditure on major items of food during the previous week and on items such as clothing, medical and education ex- penses during the past year; • wage rates in selected occupations, obtained from key respondents; • special studies, for example surveys of weight, height and morbidity. Illustration of available data Data are now available on conditions in each area at a point in time and on changes overfive years. Emphasis in analysis has been on the pattern of interrelationships, on seeing the area as a whole. Examples are taken from three panchayats (the smallest rural administrative area in India) for which data have been collected continuously since 1978/79. The brief account that follows should be read in the context of overall conditions in Kerala, the Rapp. trimt1st. statist. sanit. mond .• 41 (1988) - 27 - significant features of which are the highest literacy and lowest mortality rates among states in India, in spite of only a moderate economic level by Indian standards. The relatively high social profile is common to the three areas, which nonetheless each have a unique set of char- acteristics. Community Ab is a fishing village whose inhabitants crowd together on the shore of the Indian Ocean in rela- tively unhygienic conditions (virtually no sanitary facil- ities, 50% drawing their drinking-water from polluted sources) and a large majority with incomes that are low even by Kerala's standards, no agricultural land and dependent on the vagaries of offshore fishing and a declining activity in rope making. On the other hand, communications with nearby towns are fairly good, mainline electricity is available and was used by about a quarter of the population. A primary health care centre and primary and secondary schools also exist. This situation contrasts with that of a highland pan- chayat (community B). Here land is relatively plentiful (for Kerala), incomes are higher, safe water more abundant. The area also benefits from a large government vegeta- ble farm with employment opportunities for both men and women. On the other hand until recently, because of relative inaccessibility, health services have been of poor quality and electricity unavailable. The third area examined here, community C, is again different in that it has ready access to the capital and its facilities, including health services (half an hour by bus at low fares). It is primarily an agricultural community with moderate land-holdings and incomes. Most households have safe water from their own shallow wells, about one-fifth have some kind of sanitary facility, as com- pared with 4-7% in the other two localities. The essential difference between the areas extends also to the change seen over a five-year period, although if the change continues along present lines the end result may well be greater similarity. In line with government policy the most backward of the areas in terms of income and hygienic facilities (the fishing village), benefited more than the others. Fishing was in part motorized, a govern- ment scheme provided employment on a protective sea wall, and even a tidal wave that caused severe damage to houses in the end favoured the local people since funds were attracted that were used to rebuild many houses in a better style. In contrast, incomes did not significantly change in the other two areas. However, the highland village for the first time received mainline elec- tricity for use by private consumers, while road com- munications have been greatly improved. Health condi- tions should be seen against this background. Three kinds of information have been obtained: (1) on infrastructure; (ii) on variables such as infant mortality and morbidity; and (iii) on variables relating to indi- viduals. The distinction is important because only the latter type provides distributional data to identify those most in need or for analytic purposes (with the individual as unit of analysis). Regrettably, health statistics are rarely of this type. b In order to protect the anonymity of the communities concerned, they are identified by initials only in the original report: TKA (community A); ALM (community B); and VPM (community C). c Primary health care centres in Kerala are the principal local health institutions, and are fairly generously equipped. d Only changes in fixed facilities were reported. Additional services (not reported upon) include the various preventive measures such as inoculation, prenatal or school health services. These should eventually be included in the monitoring. Wld hlth statist. quan .• 41 (1988) The first category, health-service infrastructure, was briefly referred to in previous paragaphs. Community C had no primary health care centre,c but the inhabitants made use of a centre just outside the panchayat area. In any case, the people in community C who could afford it sought medical treatment in Trivandrum, about half an hour away by bus. This was not a solution readily avail- able for people in community A and even less so in community B, for whom the nearest specialized services are at some considerable distance. However, both com- munities A and B had primary health care centres staffed by a doctor and other qualified medical personnel, as well as various other, partly private, medical facilities. The principal problems were inadequate maternity facil- ities and particularly lack of drugs. The primary health centre in community A initially had no water, though this had been installed by 1984. There were modest changes during the period.d Thus, in community Ca new government Ayurvedic dispensary was created in 1982. On the other hand, a private hos- pital with 15 beds and a full-time doctor closed down. The only change in community A, which has a homeo- pathic dispensary staffed by a doctor apart from its pri- mary health care centre, was the supply of piped water to the centre during the fourth round (1982/83). In com- munity B, the government dispensary was upgraded as a primary health care centre in 1979, with nominally 10 beds and improved staffing. However, no new building has been constructed so far and there has not yet been any inpatient treatment. A new government homeo- pathic dispensary was built in 1980, and the tribal Ayurvedic dispensary upgraded in 1983 to the status of tribal hospital with 10 beds. Expenditure on medica- ments at the primary health care centre, a perennial source of complaint, increased between 1978/79 and 1982 from 50 OOO to 130 OOO rupees, though much of this reflected the rising cost of drugs. Two variables were used within the second category: infant mortality and overall morbidity; weight and height of children under 10 were in the third category. As noted, the former two are area indicators in the sense that the data are accumulated for the area as a whole, and do not give meaningful results for any particular household, for which an infant death in a given year or a case of mor- bidity in a given week is a random event with low prob- ability. With a population of about 12 OOO (the size of the large sample) producing, at current levels of fertility in Kerala, about 260 births annually, with approximately 10 infant deaths on average, a figure which could easily be doubled or halved in any given year as a result of chance events. Table 1 shows infant mortality rates. Because of a change in method (at first only half of the panchayat area was surveyed, after the second round (1980/81) the entire panchayat) rounds 1 and 2 are not comparable with later rounds. Even in the comparable years changes are too great to reflect more than chance events in any one area (for example, the jump from 24 to 42 per 1 OOO live births in community C from round 4 to round 5). The only tentative conclusion is that of the nine annual changes between comparable years (rounds 1-2, 3-4, 4-5 in each of the three areas), seven were negative, reflecting the overall decline in infant mortality in rural Kera la recorded by the government's sample registra- tion (3). The populations per area are too small, and the time span too short, to permit conclusions on the indi- vidual areas. At the prevailing birth and death rates, populations of 100 000-150 OOO per area would be required. Alternatively, trends may be derived from moving averages, but these would require a period of many years. - 28 - TABLE 1. INFANT MORTALITY RATES (PER 1 OOO LIVE BIRTHS) IN THREE PANCHAYATS IN KERALA, INDIA, 1978/80-1983/84 TABLEAU 1. TAUX DE MORTALITE INFANTILE (POUR 1 OOO NAISSANCES VIVANTES) DANS TROIS PANCHAYATA DU KERALA, INDE, 1978/80-1983/84 Rounds a - T ournilesa Community - Communaute 1 2 3 4 5 1978/80 1980/81 1981/82 1982/83 1983/84 A 91 30 26 24 17 B. 57 14 47 30 20 c 39 25 24 24 42 • The broken line indicates a change in baseline between rounds 2 and 3 (see text) - La ligne en pointille indique un changement des donnees de base entre les deuxieme et troisieme tournees (voir le texte). The morbidity measure used in the OMS is the number of person-days of absence from the person's usual activity due to illness during the week preceding the interview. divided by 7 and expressed as per 1 OOO of the popu- lation at risk. The rate has its problems. It is common experience that in the course of development it is the pattern and not necessarily the frequency of morbidity that changes, and slowly at that. Absence of significant and consistent change in the overall morbidity rate in the three areas over five years thus comes as no surprise (Table 2). The time span is too short, nor would changes in the pattern be captured by an overall morbidity rate. The 1979/80 figures were analysed in detail and symp- toms of those reported ill recorded (Table 3). The symp- toms were grouped in terms of diseases with which they are commonly associated. In communities A and C, the most frequent symptoms were those normally asso- ciated with diseases of the respiratory tract, such as bronchitis and broncho-pneumonia, that account for about half the cases, although in the absence of indi- vidual clinical examination it is impossible to be precise about the relationship of symptoms to disease. In any case, it is the pattern of symptoms (whether or not associated with disease) rather than the overall morbid- ity rate that should be regularly monitored. The third indicator is weight and height of children under 10. Anthropometric measures have been used in the past for at least three distinct purposes: (1) to indicate to • In reference (2) it is stated in error that the mean was used. mothers and to child health staff whether an individual child is growing the way it should; (ii) for community surveillance in conditions where there is gradual rather than drastic deterioration by way of undernutrition or obesity; and (iii) as a comparative analytic device to better understand factors associated with abnormally low or high weight and height. These three purposes by no means coincide. It is not always appreciated for example that the first (clinical) purpose. although essen- tial in its own right, tends not to yield valid statistics, because the population examined is often unrepresenta- tive of the total, or the quality of the measurement is inconsistent. Within the OMS, the second and third purposes were uppermost. Weight and height were used, together with other social indicators, to examine change over time. Table 4 shows an improvement in weight and height over time in community C which is astonishing in its magnitude. The principal interest of WHO was the study of associated factors. The role of the area as a unit of analysis is of interest in this context. Whereas there is clear association within each of the three areas between weight and height on the one hand and parental income as well as some other factors on the other, nonetheless the area with the highest average income (community B) also had the greatest number of children with /ow weight and height (two or more standard deviations below the median of the reference population) (Table 5).e There appear to be area factors influencing weight and height over and above the household factors such as income, source of drinking-water or housing conditions. It was suggested by local spokesmen in community B that the deficiency of health services. until very recently, may TABLE 2. MORBIDITY RATES• IN THREE PANCHAYATA IN KERALA, INDIA, 1978/80-1983/84 TABLEAU 2. TAUX DE MORBIDITE• DANS TROIS PANCHAYATA DU KERALA, INDE, 1978/80-1983/84 A B. c Community - Communaute • Rate - Taux - 1 OOO 2.... 7(P) 1 1978/80 9 29 9 2 1980/81 20 10 22 Rounds• - T ournees• 3 4 5 1981/82 1982/83 1983/84 17 19 18 9 8 9 18 23 19 Where Sis the number of person-days of illness causing restricted activity during the reference week, and P the number of persons in the sample. The definition is likely to locate only fairly serious illness. Thus children with worm infestation, a widespread problem in Kerala, are not likely to be reported as ill unless other more urgent symptoms also occur - Ou S est le nombre de jours/homme de maladie entrainant une baisse d·activite pendant la semaine de reference et P le nombre de personnes dans l'echantillon. La definition ne permet vraisemblablement de recenser que les maladies relativement graves. C'est ainsi que les enfants atteints d'infestation vermineuse, tres repandue au Kerala, risquent de ne pas Atre signales comme etant malades s'ils n·accusent pas aussi d'autres symptomes rev~tant un plus grave caractere d'urgence. • The broken line indicates a change in baseline between round 2 and 3 (see text) - La ligne en pointille indique un changement des donnees de base entre les deuxieme et troisi0me tourn,es (voir le texte). Rapp. trimest. statist. sanit. mond., 41 (1988) - 29 - TABLE 3. NUMBER OF SYMPTOMS REPORTED AND SELECTED ASSOCIATED DISEASES DURING THE WEEK PRECEDING THE INTERVIEW BY AGE, SEX AND SYMPTOM, KERALA, INDIA, 1979-1980 TABLEAU 3. NOMBRE DE SYMPT6MES SIGNALES ET CERTAINES MALADIES RETENUES PENDANT LA SEMAINE PRECEDANT L'ENTREVUE PAR AGE, SEXE ET SYMPT6MES, KERALA, INDE, 1979-1980 Number of cases - Nombre de cas Community A - Communaut6 A Community B - Communaute B Community C - Communaut6 C 0-4 5-14 15+ 0-4 5-14 15+ 0-4 5-14 15+ years years years years years years years years years ans ans ans ans ans ans ans ans ans M F M F M F Principal symptoms - Sympt6mes principaux Cold - Rhume ............... 40 22 77 63 5 3 15 8 40 14 29 25 Cough (no phlegm) - Toux (sans pituite) 16 5 14 14 6 2 1 1 8 6 11 15 Cough (with phlegm/blood) - Toux (avec 35 pituite/sang) ....... 27 13 79 61 - 2 32 17 37 19 47 Fever - Fievre ...... 113 45 151 151 18 14 22 24 138 82 130 96 Diarrhoea - Diarrhee ... 26 - 16 17 2 - 2 2 16 10 12 8 Palpitation - Palpitations 17 5 33 41 1 1 7 6 - 4 17 14 Breathlessness- Dyspnee 39 12 61 70 22 6 18 13 28 11 40 33 Abdominal pain - Douleurs abdominales 5 3 14 26 1 2 14 7 6 6 16 16 Passing worms - Expulsion de vers ... 29 4 3 10 1 - - - 7 3 3 - Vomiting - Vomissements ........ 24 7 20 21 - 1 1 4 4 6 4 10 Nodular swelling - Tumefaction nodulaire - 13 10 12 - 4 1 - - 2 9 1 Headache - Cephalees ...... 4 13 53 65 2 4 16 22 6 24 43 47 Chest pain - Douleurs thoraciques .... 6 3 83 77 - 6 19 18 - 9 58 44 Giddiness - Etourdissements ....... - 3 43 58 - 1 4 6 1 9 29 24 Difficulty in passing urine - Difficulte de mic- tion ....................... 17 7 30 42 - - - 2 - 2 18 10 Pains in joints - Arthralgie .......... 1 4 26 43 1 2 7 2 - - 33 23 Selected diseases commonly associated with groups of symptoms - Maladies couram- ment associees cl des groupes de sympt6- mes Bronchitis - Bronchite ............ 5 5 34 30 1 1 15 16 7 6 32 20 Influenza - Grippe . . . . . . . . . . . . . . - 2 15 8 - 1 - - 1 4 7 11 Broncho-pneumonia - Broncho-pneumonie 9 11 9 6 3 1 - 4 25 3 3 1 Diseases of the upper respiratory tract - Maia- dies des voies respiratoires superieures .. 5 - 3 8 2 2 1 - 5 4 6 2 Population at risk - Population ii risques .. 1 232 2 656 3 534 3 812 1 896 4 110 4 955 4970 1 390 2 922 3 653 3 818 Number of persons ill - Nombre de personnes malades . . . . . . . . . . . . . . . . . . . . 57 28 115 104 16 22 79 50 68 37 91 84 TABLE 4. STANDARDIZED WEIGHT AND HEIGHT OF CHILDREN: CHANGE IN COMMUNITY C, 1981/82-1983/84 TABLEAU 4. POIDS ET TAILLE NORMALISES DES ENFANTS: CHANGEMENTS DANS LA COMMUNAUTE C, 1981/82-1983/84 Round - T ournee 3 (1981/82) 4 (1982/83) 5 (1983/84) Percentage of children with 2 or more standard deviations below the median of a reference population• Pourcentage d'enfants avec 2 6cans types ou plus au-dessous de la m6diane d'une population de r616rence• Height-for-age T aille pour I' age 41.6 39.1 33.2 Weight-for-age Poids pour I' age 58.5 63.2 44.7 Weight-for-height Poids pour la taille 28.5 21.5 16.4 • For details see reference (4) - Pour plus de pr6cisions, voir la ref6rence (4). TABLE 5. COMPARISON OF STANDARDIZED HEIGHT AND WEIGHT IN ALL THREE COMMUNITIES, 1982/83 TABLEAU 5. COMPARAISON POUR LATAILLE ET LE POIDS NORMALISES, POUR LES TROIS COMMUNAUTES, 1982/83 Height-for-age - Taille pour l'ilge Weight-for-age - Poids pour l'ilge .. Weight-for-height Poids pour la taille Percentage of children with 2 or more standard deviations below the median of a reference populationa Pourcentage d'enfants avec 2 6cans types ou plus au-dessous de la mMiane d'une population de ref6rence• Community A CommunityB CommunityC Communaute A Communaute B Communaute C 43.7 49.2 39.1 56.0 63.8b 63.2b 26.1 31.6 21.5 • For details see reference (4) - Pour plus de pr6cisions, voir la reference (4). b All differences are significant at p .a;0.01 except between these two - Toutes les differences sont significatives, p . .a;0,01 sauf pour ces deux chiffres. Wld hlth statist. quan .. 41 ( 19881 - 30 have played a role (but note in Table 2 the relatively low morbidity rate in that community); also that more wo- men than in the other two areas leave their home for work, at a state farm some distance away, possibly without suitable provision for their children's meals. Data for local use The data were intended for local as well as national (i.e. state) use, but other than the involvement of the local people in data collection, local participation has been virtually nil. The reasons are related in turn to each of the three stages: measurement, analysis (or interpretation) and remedial action. (1) Although it is the simplest of the three stages, accu- rate measurement is not as easy as may appear. For example, even one of the trained DMS interviewers read the height on the wrong side of the measuring rod for some time. The error was found when the data were compared with the previous year's fig- ures. Problems of obtaining valid morbidity data in surveys are well known. (ii) Interpretation poses a more serious problem. As noted earlier, items such as infant mortality or mor- bidity rates are either invalid at this level or, at the very least, require expert analysis. Even if a moder- ate change (e.g. in infant mortality) were statistically valid, careful analysis of causes would be required, unless the reasons are obvious-a measles epidemic for instance-in which case the statistics would be superfluous. As regards low height and weight, interrelations are often complex. Slow growth may be the result of malnutrition or ill-health or the inter- action of both; each may be linked to a variety of social and environmental factors but, as we found here, the association is by no means straightfor- ward. (iii) As regards remedies, there is an understanding locally that the health service is a matter for the state, and indeed there is no provision for health in the panchayat budgets. Nor is the potential role of local people by way of stimulated self-help well under- stood in a society that in earlier days was accus- tomed to patriarchal intervention by the maharajas, and more recently to action of a political kind. The people would be more inclined to mount a lorry and demonstrate for effective services outside Govern- ment House in Trivandrum than set up self-help schemes. Indeed, in the DMS' contacts with the elected panchayat councils the demand most fre- quently made has been for comparative information on the area, relative to other areas, by way of ammu- nition for the local councils to prod the central authorities into action. This does not mean that remedial action involving the local people is not possible. However, even if significant causes are known and a remedy-within reasonable financial and other limits-were feasible, the institutional arrangements would have to be initiated by an external agency, government or private. The logic evidently ap- plies with equal force to information for use at the central level. The links between producers and users of data at this level are by no means clear as regards the DMS, or indeed in general. But there may be an institutional framework at the central level, potentially able to utilize information, which usually has no counterpart at the local level. SUMMARY A development monitoring service at the local level was established in 1978/79 in several communities in Kerala State (India) to collect socioeconomic and health-related data. These were collected from key community mem- bers and by household surveys using paid local staff as interviewers. Selected results for five years from three geographically distinct areas are presented, for three categories of data: health service infrastructure, infant mortality and overall morbidity, and weight and height of children under 10 years old. The three areas were a fishing village (A), a relatively remote highland commu- nity (B) and an agricultural area close to the state capital (C).b During the five-year period there were rather modest improvements in the delivery of health and other servi- ces. Infant mortality rates fluctuated between 14 and 91 per 1 OOO live births, reflecting the fact that the sample size ( 10 OOO to around 25 OOO for each area) was ina- dequate for calculating that rate. Morbidity rates also fluctuated markedly, without showing any definite trend. However, morbidity was defined as illness resulting in the inability to carry out normal activities, thus excluding cases of worm infestation and much diarrhoea, for example. A detailed analysis of reported symptoms for one year showed that respiratory tract diseases ac- counted for about half the cases of illness reported. The percentage of children with abnormally low weight- or height-for-age fell markedly in one area over three successive years (height-for-age from 41.6% to 33.2%, weight-for-age from 58.5% to 44. 7%), but were signifi- cantly greater in another area with a higher average in- come, indicating that economic factors alone were not responsible for the improvement. Problems of data col- lection, interpretation and use are discussed. RESUME Notification de la situation sanitaire fondee sur la communaute Un service de surveillance du developpement au niveau local a ete cree en 1978/79 dans plusieurs communau- tes de l'Etat de Kerala (lnde) pour rassembler des don- nees socio-economiques ou en rapport avec la sante. Ces donnees ont ete recueillies aupres de membres influents de la communaute et au moyen d'enquetes dans les menages, l'anamnese etant effectuee par un personnel local remunere. Certains resultats pour cinq annees provenant de trois lieux geographiquement dis- tincts sont presentes pour trois categories de donnees: Rapp. trimest. statist. sanit. mond., 41 (1988) - 31 - infrastructure des services de sante, mortalite et morbi- dite globales infantiles, et poids et taille des enfants de mains de 10 ans. Les trois lieux etaient un village de p~cheurs (A), une communaute montagnarde relative- ment isolee (B) et une zone agricole proche de la capitale de rEtat (C).f Pendant la periode quinquennale consideree, on a enre- gistre des ameliorations assez modestes dans la distri- bution des prestations sanitaires et autres. Les taux de mortalite infantile variaient entre 14 et 91 pour 1 OOO naissances vivantes, ce qui reflate le fait que la taille de r echantillon (de 10 OOO ii environ 25 OOO pour chaque lieu) etait insuffisante pour calculer ce taux. Les taux de morbidite accusaient eux aussi des fluctuations conside- rables, sans qu'il se degage une tendance bien nette. 'Pour proteger ranonymat des communautes en cause, on les a desi- gnees dans le rapport original par leurs seules initiales: TKA (commu- naute Al; ALM (communaute BI; et VPM (communaute q. Toutefois, la morbidite etait definie comme etant une maladie conduisant ii l'incapacite d' exercer des activites normales, si bien que les cas d'infestation vermineuse et une grande partie des cas de diarrhee, par exemple, etaient exclus. Une analyse detaillee des symptOmes signales pendant un an a revele que la moitie environ des cas notifies concernaient des maladies des voies respi- ratoires. Le pourcentage des enfants ayant un poids ou une taille anormalement faibles pour leur Age a baisse notable- ment dans un des lieux consideres pendant trois annees consecutives (le pourcentage est tombe de 41,6% ii 33,2% pour la taille en fonction de rage et de 58,5% ii 44,7% pour le poids en fonction de rage), mais ii etait nettement superieur dans un autre lieu ou les revenus moyens etaient plus eleves, ce qui indique que !'amelio- ration n'etait pas imputable aux seuls facteurs economi- ques. Les problemes que posent la collecte, !'interpreta- tion et I' exploitation des donnees sont passes en revue. REFERENCES - REFERENCES 1. SCOTT, w. & MATHEW, N. T. Levels of living and poverty in Kera/a. A development monitoring service at the local level-Vo/. II. Geneva, United Nations, 1983. (UNRISD Report No. 83.2). 2. SCOTT. W. & MATHEW, N. T. Monitoring change in Kera la: the first five years. A development monitor- ing service at the local level-Vo/. Ill. Geneva, United Nations, 1985. (UNRISD Report No. 85. 7). 3. GOVERNMENT OF KERALA. Sample registration: annual report 1980. Trivandrum, Government of Kerala, Wld hlth statist. quart., 41 ( 19881 1980. (Unpublished figures for 1981-1983 were also used). 4. WORLD HEAL TH ORGANIZATION. Development of indica- tors for monitoring progress towards health for all by the year 2000. Geneva, WHO, 1981. (Health for All Series No. 4). (Annexes 1-3). ORGANISATION MONDIALE DE LA SANTE. Elaboration d'in- dicateurs pour la surveillance continue des progres realises dans la voie de la sante pour tous d'ici /'an 2000. Geneve, OMS, 1981. (Serie sante pour tous N° 4). (Annexes 1-3).