The World Hea lth Orga ni zation is a specia lized agency of the United Nations with primary responsibility for inte rnat ional hea lth matte rs a nd public hea lth . Through thi s Organization , which was created in 1948, the hea lth profess ions of over 180 countries exchange the ir knowledge and ex peri ence with the aim of making poss ibl e the attainment by a ll c iti zens of the world of a leve l of hea lth that will permit them to lead a soc ia ll y and economi ca ll y productive life.
The WHO Regional Office for Europe is one of s ix reg ional offices throughout the world , each with it s own progra mme gea red to the particular hea lth probl ems of the countri es it serves. The European Region embraces some 850 million people living in an area stretching from G reenland in the north and the Mediterranean in the south to the Paci fie shores of Russ ia. The European prog ramme of WHO therefore concentra tes both on the problems assoc iated with industrial and post-industri a l soc iety and on those faced by the emerg ing democrac ies of central and eastern Europe and the fo rmer Soviet Union . In its strategy fo r atta ining the goa l of hea lth for a ll the Regiona l Office is arranging its activiti es in three ma in areas: lifestyles conduc ive to hea lth , a hea lthy e nviron ment, and appropri ate se rvices for prevention, treatment and ca re .
T he European Region is characteri zed by the large number of languages spoken by its peoples, and the resu lting difficulties in disseminating infor mation to all who may need it. App lications for rights of translation of Regiona l Office books a re therefo re most we lcome.
Health interview surveys Towards international harmonization
of methods and instruments
WHO Library Cata loguin~' in Publicati on Data
Heal th interview surveys: towa rds 'inte rnational ha1moni zati on o f methods and instruments / edited by ·A.·ae Bruin, I-I .S J . Picavet and A. ossikov
(W HO reg ional publica ti ons. European series; o. 58)
I .Data collect ion 2. l-l ea lth status indicators 3. Hea lth surveys 4 .lnterv iews - methods 5.Europe I. Bru in , A. 11 .Nossikov, A. 111.Picavet, H.S.J . IV .Seri es
ISBN 92 890 1322 2 ISSN 0378-2255
Text editing: Sheila Poole/Frank Theakston Cover design: Sven Lund
(NLM C lass ifi cati on: WA 950)
World Health Organization Regional Office for Europe Copenhagen
Statistics Netherlands Voorburg
Netherlands
Health interview surveys
Towards international harmonization of methods and instruments
Edited by
A. de Bruin H. S.J. Pi cavet
Statistics Netherlands Voorburg, Netherlands
& A. Noss ikov
WHO Regional Office for Europe Copenhagen, Denmark
WHO Reg iona l Publi cations, European Series, No. 58
ISB 92 890 1322 2 ISSN 0378-2255
The Regional Offi ce for Europe of the World Hea lth Organi za tion welcomes requests fo r permi ss ion to reproduce or tran slate its publi ca tions, in part or in full. Applica tions and enquiries should be addressed to the Office of Publi ca ti ons, WH O Regional Office fo r Europe , Scherfigsvej 8, DK-2 100 Copen hagen 0 , Denmark , which will be glad to prov ide the latest in fo rmati on on any changes made to the tex t, pl ans fo r new ed itions, and reprints and translati ons already ava ilable.
© World Health Organization 1996
Publi cations of the World Hea lth Organi zation enj oy copyri ght protec tion in acco rdance with the prov isions of Protoco l 2 of the Uni versa l Copyright Convention. All ri ghts reserved.
The des ignati ons employed and the presentati on of the materi al in thi s publicati on do not impl y the ex press ion of any opinion whatsoever on the part of the Secretariat of the World Hea lth Organi zation concerning the lega l status of any coun try, territory, city or area or of its authoriti es, or concerning the delimitati on of its fronti ers or boundari es. The names of countries or areas used in this publication are those that obtained at the time the original language edition of the book was prepared.
The mention of speci fi e compani es or of certain manufacturers' products does not impl y that they are endorsed or recommended by the World Health Organ ization in preference to others o f a simil ar nature that are not mentioned. Errors and omi ss ions excepted, the names of proprietary products are di s tingui shed by initi al capital letters.
The views expressed in thi s publi cation are those of the contributors and do not necessa ri ly represent the dec isions or the stated pol icy of the World Health Organization or of Statisti cs Netherl ands.
PRINTED IN F INLAN D
CONTENTS
Page
For eword ....... .... .... ...... ... ... ..... .. ..... .. .. .. ..... .. ..... ...... .... ...... .. .. .. ... .. ... 1x
Acknowledgements............................ .. ... .. ....... ..... ................... .. .... x 111
Introducti on ...... .... ......... ... ... .... ....... ........................ .. ..... ... .... .. ....... .
I. Hea lth for all strategy and the ro le of hea lth interview surveys............................ ..... ... .. ....... ......... ..... 3
The European hea lth fo r all monitoring and evaluati on framework ...... ... ........ ........... ..... .. ..... ...... ..... 4
Health for all indica tors - definition and purpose .. .. .. .. . .. .. .. .... .. .. .. .. ... . ........ ..... ..... .. .. ...... ....... .. .. ... . 6
Hea lth fo r all indica tors and hea lth policy ... ... .. ....... .... ...... 8
In formati on sources.... ........ ....... ... ...... ...................... ........... 9 Ex isting stat ist ica l record s. ........ .... ..... ........................... 9 A d hoc inves tigati on or surve illance systems w ithin the hea lth serv ices ..................... .... .... .................. I 0 Populati on surveys ................ .. ..................... ... ....... ..... .. . I I
Health for all indi ca tors that are only or best measured by a hea lth interview survey....................... 12
2. Deve loping common methods and instruments fo r hea lth in terv iew surveys. .... ......... ...... ...... ... .... 15
Coverage of hea lth fo r all indica tors in hea lth interv iew surveys ( 1980- 1990) .................... ........... 16
Hea lth interv iew survey proj ect ....... ............................. ...... 2 1
Consultations to develop common methods and instruments fo r hea lth intervi ew surveys .................... 21
First Consultati on, 1988 ........ ........................ ........ ... ...... 23 Second Consultati on, 1990 ......... ............. ...... ....... ..... ... . 23 Third Consultati on, 1992 ......... ........... ...................... ..... 27 Opportuniti es fo r improv ing the in te rn ati onal comparability of hea l th interv iew survey data.. .. ..... ..... 28
Proj ect rev iew... .. ....... .............. ...... ....................... ............... 29
3. Hea lth interv iew survey methodology........ ......... ... .. ..... .. .... .. .. 3 1
Data co llecti on ............................... .. ...... .. ............................ 32 Constructi on of the questi onnaire .. ........... .. ... ..... .. .. .... .. 32 M emory.................... .. ....................... ..... .... ..................... 34 Language ... ........ ..... ......... ... ............... ..... ... ............... .. ..... 34 Proxy in formants...... .. .. ... ................ .. ...... .. ... .. ... ............. 35 Face-to-face interv iews and se l f-co mpleti on ................ 36 Postal and telephone surveys................. .. .. .... ... ....... .. .... 3 7 Hea lth interv iew and hea lth examinati on surveys........ 37 Interv iewers.. ............................. .. ........ ........................ ... 38
Populati on - sample, size and structure..... .. .... .. .... .. .. ..... .. .. .... 38 Sample...................................................... .. .................... 39 Unit fo r sampling and analys is...................................... 40 Size.. .... ................. ..................... .. .. ... ........................ ... .... 40 Co llecti on peri od ...... .. ........................... .. .. .... ..... ............ 41 Continuous or repea ted surveys............ ..... .. .. .. ... ........... 4 1 Instituti onali zed populati ons............................ .. ............ 42 Chi ldren .................... ..... ... ...... ... ..... .. ...... ... ..... ............ .... 43 Non-response.................................................................. 44
Data process ing and presentati on ................. ... .... .. ...... .. ..... 45
Concluding remark s ...... ........................ ......... .. ............ .. ..... 47
4. Common instruments fo r hea lth fo r all indica tors.................. 49
Recommended instruments......... .. .. .. .. .. ........... .. ........ ... ...... 5 1 Perce ived hea lth ............................................................. 5 1 Temporary di sability...................... .. ... ... ................. .. ..... 53 Long-term di sability (phys ica l )..................................... 55 Di sability-free li fe expectancy ............................... .. .... . 60 Chronic conditions (mental)............................ .. ............ 65 Smoking.......................................................................... 7 1 Phys ica l ac ti v ity.. ..... ...... ...... .... ....... .. ...... .......... .. ... ........ 75 Birth weight.. ......... .. .. .. .. .. ..... .. ............ .. .. ...... ....... ... .. .. .. .. 78 Breast-feeding.............. ....... ............................................ 80 Body mass index .. .. .. ....................... ...................... ......... 83 Soc ioeconomic class ifica ti on .......... .... ... .. ......... .. .. .. .. .... 87
Indicato rs for w hi ch recommended instruments are not yet ava il ab le... .. ... .................... .. .......... 95
Chronic conditi ons (phys ica l )............... ...... .. ........ .. ....... 95 Long-term di sability (mental/soc ial ).. .. ................ .. ....... 98
Food consumption ...... ........ ... ..................... ....... ........ .. ... I 00 Alcohol consumption .... .. ...... .. .. .. .. ........ .. .................... .. . I 06
5. Harmonizing hea lth interv iew surveys: conc lusions and future prospects. .. ................ .... ...................... 111
Implementation of common instruments in hea lth interview surveys...................................................... 11 2
Promotion ................................................................... .... I 12 Eva luation ........................................................... ............ 113 Standard instruments and revisions.............. ................. I 15
Concluding remarks ...... ...... ......... ....... .. ...... ..... .... ............... 115
References..... .... ...... ..... ................................................... ..... ...... .... I 19
Annex I . Li st of participants in the Consultat ions to Deve lop Common Methods and Instruments for Hea lth interview Surveys....................................... 133
Annex 2. Li st of working papers prepared fo r the Consultations to Deve lop Common M ethods and Instruments for Health Intervi ew Surveys that have been used in preparing thi s publication ...... 139
Annex 3. Recommended instruments for chronic mental conditions. ... .......... ... ............... .... .. ... ... 141
Annex 4. Examples of showcard s for use in admini sterin g the instrument for soc ioeconomic class ifi ca ti on ........ 153
Annex 5. 111 ustrati ve se ts of questions on alcohol consumption.......... ......... ... ................... ... ... 157
Foreword
Adequate hea lth information support is essential .for imple111 enring, monitoring and evaluating puhlic health action to achieve health/or all. Althoug h many countries undertake th e routine collection of a large varierv of health data. most of them f eel a need ro improve their health information svsre111s lo /J/ Clke them more effective in supporting health development. Such improvement can have positive i111plications in at least two wavs.
First. policy-makers and ma110gers can rely on the information they receive to provide a be ff erfoundarionfor decisions on priorities .for action and the most ef/ecrive allocation of resources. This also means that they ha ve he ff er guidance in negotiating with other sectors for investment in health . Second/\', acrive participation ofa/1 people in health de velopment implies that countries. communities and individu als can separarelv 111ake up their mind on what line of action they will rake with respect to their health situation. They can onlv do this if they are appropriate/\! informed. Ir is therefore a basic tenet of the health for all philosophv that people should be g iven the knoil'ledge and injluence necessa1:i7 to ensure their acti ve participation in health development. Such knowledge requires adequate information . of wh ich health statistics are a Fifa/ part .
lnfor111 arion in th e health .field is affected h,· l'Clrious problems. Availahle data are ofien nor used to their .fit!/ potential owing to shortage of rmined staff' and administrati ve prohle111s. Furrh er111ore.
IX
X II EALTII INTERV IEW SURVEYS
there are fi'equent lv defi ciencies in the data produced, in respect of such /actors as relevance, timeliness, reliabilit v, coordination and . .
cooperation in data collection and processing.
From the perspective (~l the health fo r all strategy, however, the quality of" the data also depends on two other factors: international comparabilitv and relevance/or the measurement a/equities in health. Th e latter presupposes that appropriate subgroups o/the population in question are included in the measurement and analvsis a/health levels.
Wh en examined against this background, existing data collection methods and instruments differ in their relative merit. It is not difficu lt to recogni:::e the severa l advantages a/health interview surveys. Th ey are population-based and, therefore, represent all subgroups o/ the population, including the underpri vileged and the non-users a/health services, which is essential/or monitoring health/or all. Th ev are also the on ly (or at least the most important) tool Jar assessing certain dimensions o/health, such as perceived health, quality a/life, patterns o_fpersonal behaviour and th e implications ofhealth problems/or day to-dav/i1nctioning and wellbeing. In addition, through a combination of health variables and socioeconomic and demographic characteris tics, survey data permit an integrated description o/health status and its determinants. Other information Ji·om health interview surveys. e.g. data on consumption o/health care, can complement the in/orma tionfi·om other sources and therebv potentially increase its usefii/ness.
Although many countries employ health interview surveys to pro vide information for a range of health indicators, international com parison of this type ofdata has posed problems owing to differences in the methods and instruments used. Limited international comparabil ity also means limitations in the use of" data at national level, since comparisons with other countries mav be important Jar many health decisions.
Th e need to improve this situation has become more pressing with the adoption al th e health Jar all strategv. and in particular the formulation of" the specific health Jar all targets Jar the European Region of"the World Health Organization (WHO). Th e need has been recognized bv data producers, data users, different levels in national administrations. and the international o,ganizations. More specifically,
FOREWORD XI
this led Statistics Netherlands, t1nder the aegis o/the WHO Regional Office/or Et1rope. to organi::.e a series a/ international const1ltations with those in volved in national health interview st1rveys. So far, three such consultations have been held "to develop common methods and instruments for health interview sul've_vs ". and attempts have been made to evaluate the recommendations, to assess theil'feasihility and to initiate a process ol p!'actical i/1/plementation. Th e outcomes ol these activities al'e summari::ed in this book, which provides a solid foundation for/i1ture wol'k in the a/'ea.
More recentlv. the impol'tance o/ this project has gl'eatlv in creased, as the unde!'ly ing structul'e ol national and intel'national interrelations in the Et1ropean Region has changed pl'o/oundlv. All countl'ies are searchingfol' new or mol'e effective.forms of'integrntion, although the obstacles to. speed of.' and expected outcomes of this pUl'SUil may dif/el'.
Th e main challenge/or the immediate future is the implementation o/the ag!'eed common instruments/or health intel'view su/'ve_vs. Wh en more and more cot1ntries stal'/ us ing the ins ll't1111 ents in theil' national sul'veys, the possibilities fol' comparative analvsis will incl'ease, in pal'ticulal' as related to the monitoring ol health fol' all strategies. Accumulated expel'ience with these instruments will fa cilitate the future development and imprn ve111ent ol measu!'ement methods and instruments, and thus .filJ"ther the ve,y basic goal ol international(v comparnble OJ' standardi::.ed 111 east1!'ement o/ke_v health indicalol's .
A.P.J . Abraharnse Director-General Statistics Nethel'lands
.I.E. Asvall Regional Director WHO Regional Office .fol' Eu,.ope
Acknowledgements
Our gratitude goes to all participants in the Consultations to Develop Common Methods and Inst rum ents for Health Interview Surveys (held in 1988, 1990 and 1992) li sted in Annex I, who provided the bas ic material for this publica ti on.
Furthermore, we wish to ack nowledge the va luable comments on the draft of thi s publica ti on of the fo ll owing: Mr J. va n den Berg and Mr J.T. P. Bonte of Stati stics Netherlands, Dr M.W. de Kleijn-de Vrankrijker of the TNO Institute of Preventi ve Hea lth Ca re, Le iden, Netherlands, and Mr P. Anderson , Dr J.G. Sampaio Faria, Ms E. Helsing and Mr T. Pih a of the WHO Reg ional Office for Europe, Copenhagen, Denmark.
We should also like to acknowledge the help and advice provided by Dr R.J. Butcher and Ms K. Dunnell (Office of Pop ul atio n Censuses and Surveys , London, Un ited Kingdom), Dr T. Spuhler (Federal Sta ti stical Office, Berne, Sw itzerland ), Mr J.M. Robine (lnstitut national de la Sante et de la Recherc he medical e, Montpe ll ier, France), Dr E. M. Badl ey (Arthritis Community Resea rch and Eva luation Unit, Toronto, Canada), Mr G. van de Willige and Dr D. Wiersma (Univer sity of Groningen, Groningen, Netherlands) and Dr H.P.A. van de Water (TNO Institute of Preve nti ve Hea lth Ca re, Leiden, Netherlands).
Introduction
Since the adoption by the World Health Organizat ion in 1977 of the strategy of health for all, countries have become more awa re of the need for adequate information for po licy formulation, im plementation and monitoring . Target 35 of the hea lth for all st rategy of the European Region of WHO states that hea lth information systems in a ll Member States should active ly support the formulation, impl ementation , moni toring and eva luation of hea lth for al l pol icies. To ach ieve this target , hea lth information systems should , inter alia , use appropriate indica tors to measure progress towards health (for all) targe ts and provide for minimum data sets based on internationally agreed standards(/). The adoption of the hea lth for a ll targets and indi cators by the European Region resu lted in the estab li shment of a regu lar monitoring and eva luat ion process, which has he lped cou ntri es to see their information needs more c learly. It appears that information in the health services is often geared on ly to the allocation of resources and the control of spending, not to the need to evaluate se rvices and patient outcomes. Population-based data - on morbidity, disability, the use of serv ices, li festy les and positive health - have not received the attention their importance warrants. It was therefore considered necessa ry to ex plore further the role of population or health intervi ew surveys in the health for a ll monitoring and eva luat ion process, and to deve lop standa rdi zed methods and instruments for such surveys.
To fac ilitate the development of common methods and instruments for health interview surveys, a se ri es of internationa l consultations has
2 II EA LTII INTE RVIE\\' SURVEYS
been organized by Statistics Netherlands under the aegis of the WHO Reg iona l Office for Europe (2- 4). The background , di sc uss ions, proceed ings a nd prospects of the three co nsultati ons he ld so fa r are described in this book.
In 1979, the WHO Reg iona l Office fo r Europe and the Inte rnati o na l Epidemi o logical Association issued a publica tion on the concepts and processes in the measurement of hea lth (5) fo l lowed in 1987 by a nother on the measurement of improvements in hea lth through programmes of disease contro l and hea lth promotion (6) . These two publications provided a n overa ll v iew of measurement in hea lth and some exa mpl es of methods and their use . The foc us of the present book is more speci fic and pract ice-ori ented: it dea ls with the meas urement of hea lth fo r a ll indicators in hea lth surveys, w ith empha sis o n the internationa l compa rability of methods and instruments. By giving concise and structured infonnat ion on survey methodology and recommended measurement instruments, the a im is to provide a reference source for a ll those involved in the planning and implementation of hea lth infonnation systems in general , and of health inte rview surveys in particular.
In C hapter I the role of hea lth inte rv iew su rveys in mo nito ring and eva luat ion of the hea lth for a ll strategy is desc ribed , ending with a se lecti o n of hea lth for a ll indi cators that a re o nly or bes t meas ured by health interv iew surveys. This li st of indica to rs has formed the bas is of discu ss io n in the three consultations.
Th e background, ai ms and ma in di scuss io n topics of the Co nsulta tions to Develop Commo n Methods and Instrum ents for Hea lth Inte r view S urveys are hig hlig hted in Chapter 2. In C hapter 3 the conclu sions of the consultations with respect to methodolog ica l issues in hea lth inte rvi ew surveys a re described . C hapte r 4 forms the core of the book, g iv ing a system ati c and updated descripti o n of the recommended commo n instruments fo r measure ment of hea lth for a ll indi ca tors in hea lth inte rv iew surveys. It should be noted that the recommendations of the consultations are based on ex per iences in the Europea n Reg ion of WHO and in some industri a li zed countries outs ide the Region ; their applicability in other reg io ns and cul tures has not been di sc ussed.
Th e boo k e nds with so me co nc lu sions and future perspec ti ves with respect to the impleme nta ti o n a nd eva luati on of common instrum ents (Chapter 5).
Health for all Strategy and the Role of Health
Interview Surveys
The Member States of WHO, in launching the worldwide movement for hea lth for all in 1977 , decided that the "ma in soc ia l target of governments and WHO in the comi ng decades should be the attai nment by a ll c iti zens of the world by the year 2000 ofa level of hea lth that w ill permit them to lead a soc ia lly and economi ca ll y prod uctive life" (reso luti on WHA30.43). In start ing th is process , Member Sta tes recog ni zed th at it would req uire major efforts by all the co untri es of the wor ld to keep the momentum go ing. At the Internat iona l Co nfere nce on Primary Hea lth Ca re, o rgani zed jointly by UNI CEF a nd WHO in A lma-Ata, USSR in 1978, Me mber States agreed on the minimum content of primary heal th care, which was seen as the key to achieving hea lth for all. In 1979 the World Hea lth Assembly la unc hed a g lobal stra tegy for hea lth fo r all by the year 2000 (reso luti on WHA32 .30) and a lso ag reed to adapt and expa nd the globa l strategy to meet the spec ific needs of regional and nat ional c ircumstances .
At the thirti e th sess ion of the Regiona l Co mmittee, in Fez in September 1980, the Me mber Sta tes of the WHO European Region approved their first common hea lth policy: the Europea n stra tegy for atta ining health for a ll (7). This strategy calls for a fund amental change in nat ional health po lic ies; it urges that hi gh priority be g ive n to health promotion and di sease prevention , that all sectors w ith an impact on hea lth take positi ve steps to ma intai n and improve hea lth , that g reater stress be placed on the role that indi vidua ls, fa milies and comm unities can play in health deve lopment, and that primary hea lth care be the
3
4 IIEA LTII INTERV IE\\ ' SURVEYS
maj or approach to bring ing about these changes. It a lso ca ll s fo r the fo rmul ati on of spec ifi c reg iona l targets to support the impl ementati on of th e stra tegy.
The Reg iona l Co mmittee according ly adopted 38 spec ifi c reg iona l targe ts at its thirty-fourth sess ion in Copenhage n in September 1984. A lso proposed were 65 essentia l reg iona l indi cators o r groups of indi ca tors, incorporat ing the 12 g lobal indi cators, to be used as a means of assess ing progress towards atta inment of t he targets (8). In 199 1, the 38 regiona l hea lth fo r a ll targe ts were updated and , together w ith a li st of indi ca tors, adopted by the Regiona l Committee at its fo rty-first sess ion in Li sbon (!).
The commitments made by each Europea n Me mber State go be yond mere acceptance ofa common hea lth po li cy. To ensure that the ir pl edges are fo llowed by concrete ac ti on, Member Sta tes have under taken to fo ll ow the ir own progress towards hea lth fo r a ll by means of sys temati c and regul ar monitoring and eva luati on. They have a lso ag reed to submit reports peri odi ca lly fo r co ns iderati on by the m all at the Reg iona l Committee and the World Hea lth Assembl y. Thi s coop era ti ve process will prov ide a ll the countries with info rmati on and feedbac k on th e preva iling hea lth and soc ioeconomic s ituation . It should al so make it eas ie r to reach rational dec is ions on any adjust ments a nd modifi ca ti ons that need to be made in nati ona l, regional and internati onal hea lth po li c ies and strategies, the intenti on be ing that, by sharing the ir experi ences, countri es w ill learn fro m the ir indi vidual successes and fa ilures.
THE EUROPEAN HEAL TH FOR ALL MONITORING AND EV ALUA TJON FRAMEWORK
Monitoring is defin ed as the ma intenance or regul ar checking of ongo ing acti v iti es o r programmes w ith respec t to predefin ed obj ec ti ves. The purpose is to reco rd what the system is ac tua ll y doing at prese nt and to detec t poss ibl e dev iati ons from the dec ided course of acti on. There is a difference between monitoring and surveillance : the latter refers to an ongo ing observa ti on of the hea lth statu s of a popul a ti on and the factors that may affect it, and is unde rtaken with the purpose of detec ting poss ibl e changes at an earl y stage and in itiating appropri ate action .
II E/\ l. rl-1 FOR A LL STR A T E(; y /\N D T II E ROI F 0 1· II E/\LTI I INT ERV IEW SU RVEYS 5
Evaluation goes a step further than monitoring in that it attem pts to attac h va lue to acti vi ti es, serv ices, etc. in order to assess how good these are.
The ma in req uireme nt of the monito ring and eva lu ati on process is that it should be useful to policy- and dec ision-makers in their appra isa l of ac tiv ities o r programmes. It should therefore be tim e ly, and answe r a t leas t so me of the questi ons that policy-make rs dea l with.
The European targets have been carefu ll y deve loped by Member States and the WHO Reg iona l Office for Europe to meet the needs of the European Region. Therefore , from the beginning, hea lth for a ll monitoring and eva luat ion reports have followed the structure of the regional hea lth fo r a ll targets. Hea lth for a ll monitoring and eva luati on exerc ises and success ive reporting of progress have taken place a lter nate ly, every three yea rs, s ince 1984- 1985.
Each success ive monitoring and eva luati on has brought about a grea ter understa nding of the responsibilities and purpose of the exer c ises for the Europea n countri es, the Regiona l Office and WHO headq uarters a nd , most importantly, of the va lue of shari ng co ll ec ti ve experience. This has led to continuing improve ments in the monito ring and eva luati on framework , i.e. the process, mechan isms and tools used for monitoring and eva luati on. There follows a short account of the ma in events of the monitoring and evaluation process so far.
The 65 essenti al indi cators, or groups of indicators, adopted by the Reg ional Committee in September 1984 were used in the 1985 eva lu ati on exe rcise. In the li ght of the results of that first evaluat ion of the hea lth for all strategy (9) these indi cators were revised, a process in which many institutions and indi vidua ls partic ipated. On the bas is of the results, the Regiona l Committee in 1987 adopted a rev ised li st of indica tors and procedure fo r monito ring progress (10). Th is se t of indi ca tors (73 essenti a l, 63 supplementary) was used in the 1987- 1988 monitoring exerc ise, the results of w hi ch were presented to the Re g iona l Committee at its thi rty-e ighth sess ion in 1988 .
Fol lowing the 1987- 1988 monitoring exercise, the indi cators and the monitoring procedure were revi sed aga in in o rder to s impli fy the process furthe r and g ive it greater focus in preparation for the second
6 HEA LTH INTl: RVI EW SU RVEYS
hea lth fo r a ll eva luat ion in 199 1. T he changes of substa nce, as adopted by the Reg iona l Committee at its for tieth sess ion in 1990 (1 1), are summarized be low.
A fl ex ibl e "s ituati on assessment" rega rdin g progress towards each ta rget in indi vidua l countries was introdu ced. It requires qua li tati ve descr ipti on and provi des reference po in ts that have repl aced the fo rmer " non-quan titati ve indi ca tors".
A bout one third of the indicators were left unc ha nged and about one s ixth were reformul ated; a sma ll number of new indi cators we re introduced. A ll others were e ither in c luded in the s itua ti on assessment or de leted.
• T he g loba l indi cators and eva luat ion framework were incorpo ra ted into the reg iona l fra mework so as to avo id the need fo r Member States to produce two reports according to two sepa ra te fra meworks (as had been the case in prev ious exerc ises).
• A di st inction was no longer made be tween essent ia l and sup pl ementary indicators, s ince some of the latter had been fo und to be ve ry important.
The purpose of the second eva luati on (12) was not onl y to estimate the leve l of achi eve ment of hea lth fo r a ll but a lso to see w hi ch factors and constra ints influenced progress . It indica ted that in many countri es some info rmati on is s impl y no t avai lab le at centra l leve l because deve lopments at loca l and co mmunity leve ls, suc h as in the areas of lifesty les and hea lth promotion, are not a lways monito red centra ll y. Furthermore, it showed that internati ona l comparability of hea lth in fo rm ati on still needs to be imp roved th ro ug h g rea te r sta nda rdi za ti o n of defi niti ons and meth ods of data co ll ec ti on and process in g (e.g. surveys).
HEALTH FOR ALL INDICATORS - DEFINITION AND PURPOSE
Indicators are usuall y numeri cal (ra ti os, proportions, rates), a lthough they can also be qua litati ve ( e.g. ex istence or absence of a s ign, event, etc. that has been shown to be important) . Qua litati ve indicators, and
HEALT H FOR ALL STRATEGY AND TII E ROLi: OF II EALT H INTERV ll: W SU RVEYS 7
especially meaningful combinations of them, sometimes provide in va luable in sights into the situati on and may be preferred by dec ision makers. In the context of the hea lth for all strategy, indicators are defined as var iab les that he lp to meas ure change in th e leve l of achievement of the hea lth for all targets (13). This defi niti on is not restricted to numerical indicators .
Indica tors concentrate on key measurements that permit a judge ment about the whole process. In other wo rds, they reduce the number of possible meas urements to a few that are necessary and suffi cient for a give n purpose. In that sense, indicato rs already incorporate knowl edge about what is important (pr ioriti es) and also provide a way of dealing with what is frequently a very large amount of information .
Hea lth indicators in genera l, and hea lth for all indicators in particu lar, serve severa l purposes:
•
•
they are an important tool of hea lth policy formulation and implementation , as set out below in the secti on on hea lth for all indi cators and hea lth policy;
they are u ed to measure progress, i.e. they are used fo r mon i tor ing and eva luat ing the hea lth situat ion with respect to spec i fied (health for all) object ives;
they can provide yardst icks whereby countri es can compare their own progress with th at of other countri es, espec iall y those at s imilar leve ls of soc ioeconom ic development;
although they may not be measurable at present because no adequate in fo rmat ion is in place, they are neve rtheless adopted for use because they point to what needs to be done (guidance fo r ac ti on, including in format ion systems development) ; and
they have a communication and coordination fu nct ion: for example, when decided in a proper consultation process they constitute an important message to the community about agreed prioriti es.
Hea lth fo r all indicato rs fo ll ow the same structure and logic as hea lth fo r all targets. They cover the hea lth status of the popul ation and
8 111·.A I TII INTI- R\ ' 11· \\' Sl R\ ' f \'S
the main health determinants, i.e. li festyles, env ironment, hea lth care and genera l support to health development, including hea lth research, poli cies, training, informat ion , etc. Not all the hea lth for a ll targets have stat istical (quantitative) indicators, as progre s towards some of them is difficu lt to measure direc tly in quantitative terms. In suc h cases a textual description of the situation and the progress is used. Most stati stica l indicators are related to hea lth status targets ( 1- 12 ), fewer are formulated for lifestyles (targets 13- 17), environment (targets 18- 25) and hea lth ca re (targets 26- 3 1) and onl y a few are ava ilable fo r the suppo1t targets 32- 38.
HEALTH FOR ALL INDICATORS AND HEALTH POLICY
One aspect of health poli cy is the deve lopment of po li cy objectives. Objectives may be based on compari so ns of concrete data (the empiri ca l approach) or not (the theoretical approach) . Both approaches have their adva ntages, and in reality a policy obj ective may be based on a combinati on of the two. The empirica l approac h may be preferable in instances where hea lth policy is not developed at the national leve l, but rather emerges from decentrali zed negotiations between those suppl y ing and those paying fo r services. Releva nt and differenti ated hea lth indicators are a sine qua 11011 fo r a hea lth in format ion system capab le of supporting the deve lopment of policy object ives.
Indicators are used in health policy for different purposes: to formu late policy objectives; to implement health policies by means of the managerial process; and to eva luate the effect of health policies. Hea lth for all indicators provide the necessary data - fo r example on the distribution of hea lth problems and ri sk fac tors for hea lth prob lems, trends, resource all ocation and outcome of care - to suppo,t these acti vi ties.
The sa me type of information is often required for poli cy-make rs at the local , regional, nati onal and international leve ls, but it may be analysed, interpreted and presented in different ways. Experience shows that information on a number of health for a ll indicators is ava il able at the loca l, regional and national leve ls. Such indicators are useful for hi ghli ghting va ri ati ons and ex tremes that have policy impli cations, which in turn may he lp in understanding better the hea lth needs of populati ons, in targe ting services to identifi ed needs, and in monitoring the outcome of the serv ices. Examples of indicator-based policies at the national level arc hea1t disease campaigns, financing
HLA I TH FOR A LL ST RAT EGY A~ D TIii·. RO Li: CW II E;\ I 111 l~T~ R\ ' IE\\ ' SllR\' EYS 9
services fo r HIV/AIDS, and screening fo r cancer; loca l leve l examples are the reduction of perinatal and in fa nt mortality, and the provi sion of antenatal care fo r res idents of small areas . Apart from data from health services, health interview surveys are an important source of data for such indicators and have proved to be inva luable fo r hea lth management, planning and evaluation. Many survey-based data have contributed to government dec isions (/4). Finland, fo r example, drafted a public hea lth law on the basis of hea lth interview survey data (I 5. 16) and the United States Government used survey data to fo rmulate the legislati on fo r the Medicare and Medica id programmes (17). Fw1hermore, ex isting national statisti cs have been complemented by additi onal data from hea lth inter view surveys, which has increased their usefuln ess substantially ( 14) .
In some countri es, the hea lth fo r a ll stra tegy and the monitorin g process have a lready had spin-offs at the policy leve l, such as the setting-up of resea rch progra mmes on inequiti es in hea lth , determi nants o f hea lth , AIDS and chroni c di seases; the strengthening of preve nti ve poli cy; the deve lopment of new monitoring systems; and the strengthening of epidemi ology at local leve ls (/ 8).
INFORMATION SOURCES
The data sources th at can provide informati on fo r hea lth fo r a ll indica tors can be class ified into three main groups:
comprehensive stati sti ca l records already es tabli shed fo r hea lth or other purposes;
ad hoc (and not necessa ril y comprehensive ) in vesti ga ti on or surve ill ance systems within the hea lth se rvices; and
populati on surveys.
A more deta il ed class ifica ti on and di sc uss ion has been publi shed by WHO ( 13) .
Existing Statistical Records
Examples of stati sti ca l reco rd s a lready establi shed fo r hea lth or oth er purposes in c lude ad mini strati ve reco rd s such as th ose used for
10 I IEA LTl-1 IN'I c RV l c\\l SU RVEYS
demogra phic indicators (including in formati on on topi cs such as abor ti on rates and di stribution of age at deli very), mortality indicators, data on accidents (motor vehi cle traffic accidents in most co untri es , other types of acc idents also included in some), suppl y of tobacco, alcohol and va ri ous nutriti onal elements, literacy and educati on rates , human and other resources for hea lth , and coverage rates for various hea lth services. Registrati on programmes are also used fo r va ri ous hea lth conditi ons (cancer, cardi ovascul ar di seases and other chronic di seases, inc luding mental hea lth problems). They have the adva ntage o f conti nuity, but are usuall y costl y to maintain and often limited in geographi ca l cove rage.
The in formati on ava il able th rough these systems is usuall y pre se nted in a way that is either standard or can be recal cul ated to permit a standardi zed prese ntati on. In the case of reg isters th at are Ii mi ted to certain geographica l areas or otherwise defin ed groups, the informa ti on cannot be ex trapolated to th e to tal popul ati on but time trends can be assessed. Some elements, however, do not provide sa ti sfactory informati on in practi ce:
• reports of occupati onal di seases or acc idents, where prac ti ces may va ry between co untri es and over time to such an extent that even within-country trend analys is may be diffi cult ; and
• reports of notifi able di seases, whi ch present few problems o f identi fica ti on but may invo lve seve re problems of under reporting.
Ad hoc Investigation or Surveillance Systems within the Health Services
The main example of thi s type of system is the "sentine l phys ician" surve ill ance system applied in seve ral countri es, in whi ch a limited number of genera l prac titi oners are asked to report on a defin ed I ist of ca refull y chosen topics (w hi ch change from time to time) to compl e ment th e in fo rm at ion co ll ected routine ly or th ro ugh registries. A senti nel network suppli es regul ar and standardi zed reports on spec ifi c di seases and procedures in primary hea lth care, and usuall y has a coverage of around I% of the nati onal populati on ( 19) .
HEALTH FO R ALL STRATEGY AN D T II E ROLE OF II EALTII INTER VIEW SU RVEYS 11
Population Surveys
T he term " population surveys" is g ive n to a fo rm of data co ll ecti on in which a sufficiently large number of respondents (but usua ll y a sma ll sa mple of the tota l popul at ion), representing the targe t population, is ques tioned in a systemati c and stru ctured way . Popul a ti on surveys can be used to investi ga te li v ing conditi ons, hous ing demands, the labour fo rce, hea lth , nutri ti on, etc. , o r a combinati on of these. Hea lth inter v iew surveys are a type of population survey that inc ludes quest ions on hea lth c haracteri stics (perce ived hea Ith , di seases, di sa b i I ity ), hea lth re la ted be hav io ur (e .g . smok ing, exerc ise), the use of hea lth services (inc luding preventi ve services) a nd a va ri ety of demographi c and soc ioeconomic characteri st ics. T he household is o ften the sa mpl e un it used in surveys and therefore the term "household survey" is a lso commonly used.
The term " hea lth in terv iew survey" as used in thi s book refers to a ll types of popul ati on survey with a hea lth quest ionna ire as a component . Furthe rmore, the te rm is not exc lus ive ly used to denote a survey by mea ns of face-to-face interv iews, but can a lso refer to te lephone interviews and posta l surveys. Hea lth interv iew surveys may a lso conta in a hea lth examinati on compone nt.
Hea lth interview surveys are re levant for those hea lth fo r a ll indi ca tors th at are based on:
in fo rmati on that cannot be co llected routine ly th rough regis tri es;
infor mation outs ide the real m of services a ltogethe r (and can not the refo re be adequate ly co ll ected through hea lth care re lated systems such as sentine l surve ill ance progra mmes); and
in fo rmation that ca n be co ll ected from the genera l popul ati on directl y.
One of the add it iona l benefits of hea lth in te rview survey data is that they can be used to explo re the in terrelat ionships between self-assessed hea lth , hea lth-re la ted behav iour, use of services, and soc ia l, economic and de mographi c va riabl es. For in stance, the area of eq ui ty in hea lth requi res data that can be d iffe rentiated by soc ioeconomi c sta tus. Hea lth
12 IIEALTII INT ER VIEW SU RVE YS
interview surveys are therefore an efficient and rich source of informa tion for many indicators.
Information for many of the health for all indicators can only or can best be collected by sample surveys of the population. This applies mainly in the areas of:
• self-perception of health status and the indicators related to disablement;
• I ifesty le-related indicators (smoking, alcohol consumption pat terns, breast-feeding, physical activity) and indicators related to health promotion;
• (in some countries) indicators related to environmental health (water/sanitation, housing); and
• those aspects of health service provision and use where indi vidual response is a major factor (e.g. family planning).
In summary, the relevance of health interview surveys for the purposes of health for all lies in the provision of data for health for all indicators that cannot sufficiently or efficiently be assessed by other methods, and in the possibility of investigating relationships with other health for all indicators and background characteristics.
HEALTH FOR ALL INDICATORS THAT ARE ONLY OR BEST MEASURED BY A HEAL TH INTERVIEW SURVEY
On the basis of the original list of proposed health for all indicators (8) a list of indicators was selected for which health interview surveys may be relevant (in some cases in addition to other sources). This list was compiled by a working group during the first Consultation to Develop Common Methods and Instruments for Health Interview Surveys in 1988 (2). Because of subsequent revisions in targets and indicators, the list has undergone a number of changes; the most recent list (Table I) is based on the 1991 revision of the health for all targets (I).
The revisions referred to have resulted not only in changes in wording, but also substantial transformations, deletions and additions
111·. A I 111 1·01( A L L ST RATE<;Y AN D TIii · ROI I 0 1 111·. A LT II IN.11·. R\ ' IL\\' SU RV l·YS
Table I . List of health for all indicators for which health interview surveys are relevant
Equity in health (target I)
1.1 Differences in health status between countries
1.2 Differences in health status within countries
Health and quality of life (ta rget 2)
2.2 Assessment of perceived health0
2.6 Assessment of social health and support
2. 7 Assessment of quality of life
2.8 Healthy life expectancy: indices linking life tables with functional aspects of health
Better opportunities for people with disabilities (target 3)
3.2 Percentage of disabled persons of working age engaged in regu lar occupational activities 0
3.3 Assessment of quality of life for people with disabilities
Reducing chronic disease (ta rget 4)
4. 1 Number of days of temporary disability per person per year, by age and sex 0
4.2 Percentage of the population experiencing different levels of long term disability, by age and sex0
4 .5 Disability-free life expectancy at birth and at ages I, I 5, 45 and 65 years , by sex 0
4.6 Incidence and prevalence of selected chronic condition s: all ages, by sex; people aged 65 years an d over, by sex 0
4.9 Long-term incapaci t y fo r work, by age and sex
Healthy aging (target 6)
6.5 Assessment of quality of life for those aged 65 years and over
Reducing mental disorders and suicide (target 12)
12.2 Assessment of quality of life fo r people with mental diso r ders
12.3 Mental health 0
Healthy living (target 16)
16.4 Percentage of neonates having a birth weight of at least 2500 g0
13
14 HEAL T II INTERVIEW SU RVEYS
Table I ( contd)
16.5 Percentage of children with acceptable weight for age and/or weight for height
16.6 Percentage of children breast-fed at six weeks, three months and six months of age 0
16.7 Energy expenditure patterns, by age, sex and socioeconomic groups: total daily energy expenditure, dai ly energy expenditure for physical leisure activit ies, energy expenditure for physical lei sure activities of higher intensity
16. IO Distribution of body mass index by age and sex, including percent age of population with a body mass index (weight/height2) greater than 30 kg/m 2 a
16. 1 I Adequate nutrition
16. 12 Exercise0
Tobacco, alcohol and psychoactive drugs (target I 7)
17.2 Distribution of alcohol consumption by quantity consumed, age and sex
17.3 Consumption of the principal narcotic drugs covered by the Single Convention on Narcotic Drugs, 1961 (as amended 1972)
17.6 Consumption of pharmaceutical psychotropic substances
17. 10 Proportion of population who:
- are nonsmokers0
- are heavy smokers (20 or more cigarettes per day) 0
- have never smoked 0
- have stopped smoking for the past two years0
- have reduced smoking for the past two years 0
Human ecology and settlements (target 24)
24.3 Proportion of population that is homeless and proportion of popula tion that lives in substandard accommodation
a An instrument for this indicator (or part of it) has been recommended by the first, second or third Consultation to Develop Common Methods and Instruments for Health Interview Surveys (see the section on recommended instruments on page 5 I).
of indi ca tors. Several of the improvements a re the res ult of the recom mendations of the Consultation s to Develop Common Methods and In st ruments for Hea lth Inte rvi ew Surveys , which are described further in C hapter 2.
2
Developing Common Methods and Instruments
for Health Interview Surveys
As po inted out in the prev ious chap ter, it has become genera ll y ac cepted that hea lth inte rview surveys a re in va luabl e fo r hea lth in fo rma tio n systems beca use they provide comple te, com pre hens ive and inte gra ted information abo ut hea lth , hea lth-re lated behaviour, medi ca l consumption , and personal and ho usehold characteristics . A na lys is is not fru strated by limitatio ns in va ri abl es, as it is poss ible to inc lude in the questionnai res the necessary background in fo rmat io n about the respondent and hi s or her fa mily. Hea lth interview surveys a re a re lati ve ly cheap and quick too l, espec ia ll y when compared w ith the costs of routine data co ll ec ti on by hea lth care agenc ies.
An increas ing number of countri es in the Euro pea n Region have already had ex peri ence with hea lth interv iew surveys within the ir nat iona l health informati on framework. Statistics Netherlands, fo r exa mpl e, has conducted a continuous hea lth interview survey s ince 198 1. Outside the European Region we ll known exampl es include the hea lth interv iew survey in the United States, undertaken continuous ly since 1957, hea lth interview surveys in Japan, in progress si nce 1953, and the impress ive disability surveys conducted by Statistics Canada si nce 1983.
ln the 1980s, increas ing contac ts between stati s tica l burea ux , min istries of health and indivi dual s in various countries pointed to the need for an inte rnational exchange of knowledge about the methods and instruments for heal th inte rv iew surveys. This need was ev ident from two observations.
15
16 111; /\LT II INTE RVIEW SU RVEYS
I . Despite the undi sputed re leva nce of hea lth intervi ew surveys for hea lth inform ation syste ms and the ex istence of continuous hea lth interv iew surveys in some countri es, many countri es ca rry them out on an ad hoc bas is if at all. These surveys seem to ha ve developed large ly independentl y, and decisions about them and the questions they ask are made in the spec ifi c context of hea lth problems, hea lth care systems and policy issues in ex istence at a ny particular time in the different countries. Not surpri s ing ly, thi s gives ri se to a variety of tec hnical so lutions. However, many of the differences in survey methods and instruments seem to be unnecessary.
2. A great number of hea lth for a ll indi cators appear to be completely dependent on the ava ilability of data from hea lth interview surveys (see Chapter I). The hea lth for all monitoring and eva luation exercises showed that information on only a small proportion of these indicators was available in a sufficient numbe r of countries. Unless significant improvements are made in this situation, the lack of information will hamper monitoring of progress towards the targets that depend on survey indicato rs. WHO therefore has a great deal of interest in ensuring a high leve l of comparability between countries with respect to the measurement of health for all indicators . Other well known endeavours to attain international comparability of health sta ti stics are the use of the International C lass ification of Diseases (20), most importantly to code cause of death , and the more recently developed International C lassification oflmpairments, Disabilities, and Handicaps (ICIDH) (2 1).
Although many countries have been conducting health interview surveys for many years, international comparison of survey data is a re latively new issue. The lack of availability and comparability of these data has been found to be a major constraint in thi s rega rd.
COVERAGE OF HEALTH FOR ALL INDICATORS IN HEAL TH INTERVIEW SURVEYS (1980-1990)
To ga in more insight into the coverage of health for all indicators in health interview surveys, an inventory study was carried out in 1990 by Statistics Netherlands, at the request of the WHO Regional Office for Europe (22,23), by means of a questionnaire to all statistical bureaux and rel evant research institutes of the countries in the European Region and some se lected countries outside the Region (Australia, Canada,
DE VELO l'I NC, COMMO N MFT II OIJS AN D INS I RUMEN IS FOR 111 :ALTII IN"l l: RVII. W SU RVl· YS I 7
Japa n and the United States). The total response was 59 questionnaires from 33 bureaux, persons and/or institution s in 26 cou ntri es.
Thi s inve ntory provided information on 17 of the hea lth for a ll indicators for which hea lth interview surveys a re re levant (see Ta ble I ); the results are summarized in Table 2 and Fi g. I . The data for each of these 17 indicators were co ll ected by at least o ne co untry. The study showed that there is cons iderab le va ri ation in the number of indicators that countries cover by mea ns of a survey; the greatest number covered by any country was 15 and the smallest 2. Indicators such as perceived health (2.2), long-term disability (4 .2) and smoking ( 17. 10) were covered by surveys by a lmost a ll countries, whil e others such as birth weight ( 16.4) were covered in onl y a few countries . In genera l it can be conc luded that coverage by survey is most frequent for the selected indicators relating to diffe rent aspects of health status (targets 1- 12), followed by the indicators on healthy lifestyles (targets 13- 17).
The study inc luded an in ve ntory of the actual survey methods and question s used in the differe nt countries in collecting information on hea lth for al l indicators. Differences were found in the method of sampling, the inclusion or exc lus io n of certain institutionalized groups (e.g. in nursing homes, pri sons, military barracks), the method of data collection and the actual wording of quest ions. The study also showed that many question s, which in principle apply to the same indicator, cannot provide comparabl e results because they lack uniformity . All these differences limit the co mparab ility of data from different coun tries, and confirm the need to deve lop and use agreed standards in surveys.
The study also showed that the ava i !able inte rnat ional data from hea lth interview surveys in the period from 1980 to 1990 are not very suitable for health for a ll monitoring . Some health for a ll indicato rs for which information can only or best be collected by means of a survey are on ly covered in a few countries; but a lso when the indicators are covered, the results are often not comparable. International and interregional comparisons of population-based data on hea lth condi tions and determinan ts , which are of paramount importance for setting hea lth objectives and for good management of resources, are therefore considerably restri cted or even impossible. However, it was also conc luded that most of the differences in health interview surveys
Table 2. Coverage of health for all indicators in health interview surveys in various countries, 1990° I oo
Country Indicator
2.2 3.2 4.1 4.2 4.5 4.6 4.9 16.4 16.6 16.7b 16. 10 16.11 17.2 17.3 17.6 17.1 oc 24.3d Totale
Austria X X X X X X X X X 9
Belgium X X X X X X X 7
Bulgaria X X X X X X X X X X X X X 13
Denmark X X X X X X X X X X X X 12
Finland X X X X X X X X X X X X X X 14
France X X X X X X X X X X X X X 13
Germanyf X X X X X 5
Hungary X X X X X X X X X X 10
Iceland X X X X X X X X X 9
Ireland X X X X X 5
Israel X X X X X 5
Italy X X X X X X X X X X X X X X 14
Netherlands X X X X X X X X X X X X X X X 15
Norway X X X X X X X X X X X X 12
Poland X X X X X X X X X X 10
Portugal X X X X X X 6
Spain X X X X X X X X X X X X 12
Sweden X X X X X X X X X X X X X 13
Swiuerland X X X X X X X X X X X X X 13
Turkey X X 2
Table 2 ( contd)
Indicator Country
2.2 3.2 4.1 4.2 4.5 4.6 4.9 16.4 16.6 16.7b 16. 10 16.1 1 17.2 17.3 17.6 17.1 o c 24.3d Total e
United Kingdom X X X X X X X X X X X X X X X 15
Subtotal, Eu ropean Region 19 18 15 18 7 15 16 5 5 12 14 13 15 4 10 19 9
Australia X X X X X X X X X X X X X 13
Canada X X X X X X X X X X X X X 13
Japan X X X X X X X X X X 10
United States X X X X X X X X X X X 11
Totalg 23 21 19 21 7 19 20 5 7 16 18 16 19 6 12 23 9
0 Adapted from Evers (22); this table includes only the indicators of this study that are also included in the 1993 list of indicators (see Table I of the source).
b Data based on 1987 definition, "amount of time spent daily on voluntary physical activities" ( I 0).
c "x" means that one or more of the five components of this indicator (Evers /22). Table I) are covered.
d "x·· means that the component "proportion of population that lives in substandard accommodation" of this indicator (Evers /22), T able I) is covered.
e Out of 17 indicators.
f Information relates to the Federal Republi c of Germany prior to reunification.
g Out of 25 countries.
0 ,n < m r 0 .., z cl
~ ~ 7
~ :,:: m ~ :,: 0 0 V,
> z 0
z J,
~
"' r :,::
~ ~ V, .,, 0
"' r,; > r
z ~
"' "' < m s:: V,
C ,0
< ~ V,
'°
20 II FALTI I INTERl ' l lcll SU RVEYS
Fig. I . Coverage of health for all indicators through health interview surveys
in countries of the European Region, 1990 (total number of indicators = 17)0
0 Adapted from Evers (22).
b Information relates to the Federal Republic of Germany prior to reunification .
The designation and the presentation of material on this map of the WHO European Region Member States (as at 31 August 1995) do not imply the expression of any opinion whatsoever on the part of the World Health Organization concerning the lega l status of any country, territo ry, city or area or of its au thor ities or concerning the delimitation of its frontiers or boundaries.
lll· \ "ELOPIN<i CO~t ~to, \ll: T IIODS ,\1\ll "SI Rl •\ 11· 1\ ·t S fOR 111:Al Tit I '. I ER\"l b.\\ ' SL R\ ' l· YS 2 J
could be harmoni zed by using common methods and questionn aires, cut-off points and classifications.
HEAL TH INTERVIEW SURVEY PROJECT
It was felt that the situati on rega rding the lac k of compara bility of survey data would be improved by encouraging those involved in hea lth interview surveys at nati ona l leve l to benefit from each other's ex peri ence and to lea rn from each other ' s successes and fa ilures by means of in te rnati onal meet ings and the creat ion of a netwo rk of ex perts in the area. The major long- term obj ec ti ve was twofold:
to deve lop common meth ods and instruments (questions) for at least a core set of hea lth for all indicators; and
to get these used by co untri es in order to achi eve better inter nati onal compara bility and enhance the va lue and use of survey res ults.
Following an intensifi ed exchange of views between the WHO Reg ional Office fo r Europe and Stati stics Netherlands, the Reg ional Office requested Statistics Netherl ands to orga ni ze the first Consulta ti on to Deve lop Common Methods and Instruments fo r Hea lth Inter view Surveys. Sponsored by the Regional Office, thi s took place in June 1988, fo ll owed by th e second Consultati on in September 1990 and the third in September 1992.
Common know ledge and col laborati ve resea rch accumulated in connec ti on with the consultations. With the respecti ve rounds of hea lth for all monitoring and eva luati on, the proceedin gs of the consultati ons were di sseminated and data on the health for a ll indicators collected. However, it was also necessary to obtain more complete feedback from Member States on the progress that could be ex pected in co ll ecting internati onall y comparable data, both immed iately and in the longer run . These aspects of the project are desc ribed on page 28.
CONSULTATIONS TO DEVELOP COMMON METHODS AND INSTRUMENTS FOR HEALTH INTERVIEW SURVEYS
For each consultation, a number of ex perts were invited who were in vo lved in hea lth interview surveys in the European Reg ion or in
22 11 1·. J\ I rll INTl: R\ ' 11· 11 ' SU R\ ' l· YS
selected countries outside th e Reg ion (Australia, Canada, Japan and the United States) where continuous or at least regular surveys have been carried out. The participants in the three consultations are li sted in An nex I.
The aims of the three consultations can be summari zed as fo ll ows:
•
•
•
to rev iew the reg ional health for al l indicators, including re vised and new ones, for which information can best be provided through hea lth interview surveys;
to consider whether the li st of hea lth for a ll indicators lacks relevant indicators normally covered by hea lth intervi ew sur veys and , if appropriate, to recommend add itional indicators;
to deve lop common methods and instruments for co ll ecti ng informati on for the above-mentioned indicators (thi s relates to rev iew and coordination of existing inst ruments and the devel opment of new inst ruments) ;
to provide an overview of and to discuss current and future hea lth interv iew surveys in the European Region and se lected countries outside the Reg ion , in order to assess th e implica ti ons for collecting information on hea lth for all indicators in these countries;
• to di scuss the possibilities of including recommended common instruments in current and future hea lth interview surveys.
Jn each consultation a number of general subj ects related to th ese aims were di sc ussed. However, most of the time was reserved for the deve lopment of common instruments for spec ific hea lth for all indica tors. The main di sc ussion topics were introduced by means of working papers prepared by the participants (see Annex 2). An overview of the main discuss ion items in the three Consultations is given in Table 3. Reports have been produced for all three consultations (2- 4). For 11 indicators, common instruments could be recommended during the consu ltations (see the section on recommended instruments beginning on page 51 ). For four other indicators for which common instruments are still in development, the current situation is described in the section
Ill \ 'LLOl'I N(; CO~ l ,'vlO~ Ml · I IIOllS AN ll INS rR ll~ ll ·. N I S I OR 111-.111 111 IN 11 R\ ' II II SL'R\ ' l · \' S 23
on indica tors for which reco mmended instruments arc not ye t avai l ab le, beg inning on page 95. Where poss ibl e, comparabi lity with the ICD and ICI DH classifications was cons idered (in particular for long term disa bility and chroni c co nditi ons).
First Consultation, 1988
At the fir st Consultation , agree ment was reac hed on the li st of hea lth fo r all indica tors for whi ch hea lth interview surveys arc re leva nt. An updated overview of th ese indicators, including the modificati ons that we re ca rried th rough in later rev isions of th e hea lth fo r a ll indicators, is present ed in Tab le I (sec page 13 ).
In struments were recommended for six of these indica tors: per ceived hea lth , temporary disability, long- term disability , smok ing, birth weight and breast-feeding. It was also recommended that an indicator on adult weight and he ight should be added to the li st. For breast-feeding, the hea lth lor a ll indicator was broadened to inc lude breast-feeding at six weeks of age.
The methodolog ica l issues that need to be co nsidered when des ign ing hea lth interview surveys were a lso a main topic or discuss ion.
To faci litate the monitoring of progress towards hea lth fo r all , it was reco mmended that regul ar hea lth interview surveys should be impl eme nted in all coun tr ies of the European Reg ion . It was further advocated that WI 10 shou ld di stribut e the in strument s recommended by the cons ultations to all co untri es of the Europea n Region, so as to promote their impl ementati on.
Seco nd Consultation, 1990
The report of the 1990 in ve ntory study of coverage or health fo r all ind ica tors in hea lth interview surveys was discussed , and it was recom mended th at th e exe rcise shoul d be repeated after about five yea rs, taking into acco unt the reco mmended instruments.
The participant s lea rned that th ere was some evidence of an in crease in th e number of hea lth interview surveys being carried out , and
Consultation
First
Second
Table 3. Overview of three Consultations to Develop Common Methods and Instruments for Health Interview Surveys
Period
21 - 23 June 1988
I S-20 September 1990
Participants (countries)
19 ( 13)
22 (13)
Working papers
6
5
Main discussion topics
Selection of health for all indicators to be covered by health interview surveys
Methodological issues in health interview surveys
Measurement of - perceived health - temporary disability - long-term disability - smoking - birth weight - breast-feeding
Coverage of health for all indicators in health interview surveys Promotion of health interview surveys and use of recommended
instruments
Measurement of - socioeconomic classification
- disability-free life expectancy
- long-term disabil ity
- body mass index
- chronic physical conditions
- physical activity
N .j:,.
Table 3 ( contd)
Consultation Period
Third 22-24 September 1992
Participants (countries)
22 (12)
7
Working papers
Main discussion topics
Internacional comparisons of socioeconomic inequalities in self reported health
Promot ing the use of recommended instruments
Measurement of - physical activity - mental health (conditions and disability) - chronic physical conditions - disability-free life expectancy - food consumption - alcohol consumption
~ < rn r 0 ~ 'l
~
0 ;': ::::: ~ ;':
'.:; :r 0 v V,
> z v 7'. er
"" C ::::: 1/ -l V,
2 ,,
> :!
"% .,. "' r
::: V,
C
"' r ,,; -< V,
N u,
26 I l l ,\ I 111 1, T l R\ 11-. 11 SL R\ ' I YS
that th e Co unc il o f Europe (24) had adopted the long-term di sability in strum ent reco mmended by the first Co nsultation as th e minimum set of questi ons that should be used in hea lth intervi ew surveys, with a fe w additions (see below) .
With respect to spec ifi c hea lth fo r a ll indi cators, recommendati ons were made for measuring di sa bility-free life expectancy, long-term di sa bility, body mass index (on th e bas is of se lf-reported adult we ight and height) and soc ioeconomi c class ifi ca ti on_ f or long- term di sa bility, three opti onal items were added to th e instrum ent recommended by the fir st Consultati on, as suggested by the Co unc il of Europe (24), In additi on, two summary sco res were reco mmended, one fo r handi ca p and one fo r di sa bi I ity. A broadening o f the ti ti e of the indicator on body mass index was reco mmended : apart fro m th e percentage o f th e popu lati on with a body mass index grea ter than 30 kg/m~, the di stributi on of the body mass index was to be inc luded - thi s has since bee n adopted for use by the European Region o f WHO.
The meas urement of chro ni c physica l co nditi ons was also di s cussed and a short li st o f those that could be in cluded in hea lth int erview surveys was proposed (see the secti on on chroni c conditi ons (phys ica l) on page 95) .
Regardin g phys ical ac tivity it was co nc luded that th e ques ti onnaire based on the Quebec Hea lth Survey, suggested by WH O in 1990 ( II ),
co uld not be used in it s current fo rm in other cultura l se ttings. Thi s subject was fo ll owed up in the third Consultati on.
The participants di scussed the proposa l of th e WH O Reg ional Offi ce for Europe that a common hea lth interview sc hedul e should be deve loped. They co ncluded th at the process of coo rdinating ongo ing hea lth interview surveys in th e Europea n Region and stimulating new ones should be encouraged and supported. Thi s is parti cul arl y impor tant for reori enting hea lth informati on systems in support o f the hea lth for all strateg ics. The process should be a iming at:
enco urag ing co untri es to co nduct hea lth inler l' i e11 · surveys regul arl y;
coordinating with hea lth examination surveys;
improving the methodolog ies used and the effi ciency of sur veys; and
faci litating the use of common methods and instruments in order to im prove the ava ilability of comparable populati on-based data.
It was recommended that the Reg ional Office use future hea lth fo r all mon itorin g exe rcises to enco urage co untri es to use the recom mended instruments and , if poss ib le, to coordinate tim etabl es for fo rth coming surveys . It was also reco mmended that a WH O co ll aborat ing centre for popul ati on-based hea lth surveys be des ignated, and that the poss ibiliti es for co ll aborat ion with other internati ona l orga ni za ti ons be ex plored.
Third Consul tat ion, 1992
At the third Consultati on, in strum ent s we re reco mmended fo r th e measurement o f ph ys ica l ac ti vity and fo r chroni c mental hea lth condi ti ons. For menta l hea lth , the in stru ments we re based on a study under taken by th e Department of Soc ia l Psyc hi at ry of the Uni ve rsity of Gronin gen (a WH O Co ll aborat ing Ce nt re) {]5). For th e measure ment of disa biliti es beca use of mental hea lth problems, an in strument was recommended for te mporary di sa biliti es onl y.
Regarding the measurement o f di sa bility-free li fe ex pec tancy, some furth er recommendati ons we re made in addition to those on thi s subj ect made by th e second Consultati on.
For th e meas urement of food consumpti on th ro ugh hea lth inter view surveys, it was recommended th at pri ority be give n to meas uring the intake of fa ts (total and saturated) re lati ve to tota l energy intake. It was a lso reco mm end ed th at trend s in th e ava il abilit y o l' macro nutri ents should co ntinue to be monitored .
With res pect to a lco hol consumpti on, it was recommended that th e information co ll ected should in clude th e quantity co nsum ed during a "typical " or "average" peri od o r tim e, and th at th e ques ti ons on thi s topic should , where poss ib le, be se lf-completed. No spec ifi c instru ment was recommended, but so me illustra ti ve sets of ques ti ons we re prese nted that co uld be appli ed in tri al studi es.
28 II EA I TII INT I- RV II · \\ " SU RVEYS
Two important items that need furth er study were touched upon : th e relationship between health interview and hea lth exa mination surveys, and ways of obtaining information on the hea lth of people res ident in in stitutions.
The results of a study on internati ona l va riati on in soc ioeco nomic inequaliti es in se lf-reported hea lth , undertaken by the Department of Public Hea lth and Socia l Medicine, Erasmus Un ivers ity, Rotterdam and Statistics Netherlands (26), were a lso di scussed. Despite the limited comparability of the data on soc ioeco nomic and hea lth indica tors requested from the participating co untries, which res tricted the possibilities for comparati ve analysis, the study yielded some impor tant findings. For men, the smallest inequ aliti es were found in Norway, Spain , Sweden and the United Kingdo m and the largest in Canada , Germany, Italy and , in particular, the United States. Denmark, Finland, Japan and the Netherlands occupied an intermedi ate position. For women, the international pattern was less clear.
With respect to the hea lth indicators used, obviously incomparable data were exc luded from anal ys is, but data res ulting from survey questions with onl y minor differences in wording or structure were included. Thi s probabl y did not cause a major bi as in the study 's findings, because the sa me pattern of international variation in health ineq ualiti es was found when the anal ys is was restricted to optimall y comparable hea lth indicators. Howeve r, in order to improve the possi bilities for reliable international comparisons, it was recogni zed that more co untri es should use comparable questions to meas ure hea lth and should also include education and at least one of the oth er soc ioeco nomic variables (income, occupation) in the ir surveys.
It was recommended that th e study be repeated 111 the future, preferably every fiv e years.
Opportunities for Improving the International Comparability of Health Interview Survey Data
One of the conclusions of the 1990 study on the coverage of hea lth for a ll indicators in hea lth interview surveys was that, despite a large var iat ion in the design of questionn aires for and tec hniques of health interview surveys, most of the differences could be harmoni zed by
DEVE LOl'I N( , cm l MON METHODS AN D INSTRUM I-.N TS HlR 11 1:ALTH INT l: RV II-. \\' SU RVEYS 29
us ing comm o n measureme nt in strum ents. The Regio na l O ffice the re fo re undertoo k a fo ll ow- up ma iling exerc ise in 1992, requesting the opini o n of survey practiti oners o n o pportuniti es for ha rmo ni za ti on (27) . Mo re spec ifi ca ll y, the a im s of thi s in ves ti ga ti o n we re to obta in a n indicati o n of how fa r the hea lth inte rv iew surveys that were already planned for 1993- 1994 we re like ly to produce data tha t a re compatibl e w ith the recommendati ons of the fi rs t and second Consultati o ns, a nd to take stoc k of the oppo rtuni t ies and/or probl ems that a re anticipated with rega rd to th e ha rmo ni zati on of survey da ta in future.
It was co nc luded that mo re or less comparabl e da ta wo uld probabl y be produced by a s ignifi ca nt number of countri es but tha t, in many cases, min or mod ifi ca tio ns of th e in struments used would be neces sa ry, as these in st ruments we re s li g htl y di ffe rent from the recom mended in strume nts. It may be useful to cons ider the surveys ca rri ed out in Europe in three g roups:
those that can imm edi ate ly prov ide compara bl e results fo r so me of the hea lth fo r a ll indi ca tors;
those for which , with littl e effort , survey ques ti ons can be c hanged to produce co mpa rabl e results or, poss ibl y, even ex ist ing res ult s can be adapted to make compara bl e da ta sets; and
those fo r w hi ch new o r additi ona l questi o ns a re required.
For s ix of the e leven hea lth for a ll indicators for whi c h commo n instruments have been recommended by the cons ulta ti o ns, it appea red th at no o r onl y mino r di ffe rences in measurement ex ist between coun tri es. In o rde r to eva lu ate th e ex peri ences of countri es w ith measure ments for these s ix indi ca to rs, and to show th e potenti a l fo r progress, a furth e r study was initi ated by Stati s ti cs Nethe rl ands and the Reg iona l Office in 1993. Thi s study is described in C hapte r 5.
PROJECT REVIEW
In bri ef, the hea lth inte rvi ew survey proj ect has so far bro ug ht about :
three inte rnati o na l consultati o ns during the pe ri od 1988- 1992 , w hi c h prov ided a fo rum for exc ha nge of ex pe ri ence and w hi ch
30 111 \I 111 I' 11 R\ ' I1-11 ' SL R\ ' I· \ S
resulted in the formulation o r recommended instruments for 11 hea lth for a ll indicators;
• a study of hea lth interview surveys in 1990 lo assess the coverage of health for all indica tors in such surveys, followed by a second study in 1992 to obtain an overv iew of the oppor tuniti es for countries to harmoni ze measurements for health for all indicators; and
• a more in-depth follow-up study carried out in 1993- 1994 to evaluate the ex periences with six (recommended) measure ment in struments.
Ex perience after some five yea rs of co nsultati ons and studi es has shown that in most European co untri es th ere is a widespread interest in the devel opment of hea lth interview surveys. It is now genera ll y recogni zed that the information coming from surveys (cove rage of the general population, int egra ted data on hea lth, lifesty le, use of se rvi ces, bac kground charac teris ti cs) is ex treme ly useful. Surveys have proved to be an attract ive in strument fo r th e monitoring of hea lth and the hea lth behaviour of populations: their po li cy re leva nce at the national and int ern ati onal level is well estab li shed and grow ing, and the number of Europea n co untri es performing them is increasing.
Ex perience has also shown, however, that difTcrcnces between questions in existi ng health interview surveys in var ious co untri es, although often unn ecessary, are difficult to overcome. Once a country has reali zed a survey, it is adva nt ageo us to repeal it at least parti a ll y unchanged so as to permit trend eva luati on at the nati onal le vel. Unfortun ately, in many cases .. the areas co untri es wis h to reta in un changed often include the hea lth for all indicators lor whi ch the co nsu ltati ons have advised the use of common instruments.
A I though reaching understanding and agreement on making changes is a tim e-cons umin g process that appears to res ist any attempt at acce lerat ion, the healt h interview survey project has been successfu l with respect to its first objective: the development of' com mon instru ments for some hea lth for a ll indica tors. Whi le that process needs lo be extended, the second object ive of the project - the ac tual implementa ti on of th ese instruments - now has lo be ac ti ve ly pursued. Chapter 5 highlights the activit ies and prospects in this regard.
3
Health Interview Survey Methodology
This chapter hi g hli g hts the most important components of hea lth interv iew survey methodology , with respect to measurement of hea lth and hea lth-related indi cators in general and health for a ll indicators in particular. It is not, however, a complete gu ide to the design and co nduct of a health interview survey. For this purpose the reader is referred to literature on research methodology in genera l (28) and health interview surveys in particular (29- 31).
Methodolog ica l aspects of surveys, such as data co ll ect io n meth ods an d sa mpl e design , are important because of their effect o n data quality. As Cartwri g ht (29) has pointed out, "survey data can on ly be as good as the weakest process in their co ll ection". The quality of data col lected is determined by the va lidity and reliability of the question naire and by a number of ot her factors assoc iated with population coverage, non-response a nd interviewer and respondent characteris tics. The va lidity and re li ab ility of survey data should be regula rly assessed, and it is the refore desirable that all surveys should include re li abi lity and validity c hecks (28,32). Internat iona l co ll aborat ion and exchange of knowl edge o n these issues is particularly important.
Some methodo logica l issues were discussed during the consulta tions, resulting in so me genera l recommendations. A number of others that a re important for comparability be twee n co untries, such as age standard izat ion and weighting for non-response, have not yet been discussed and are on ly mentioned briefly in thi s chapter. Exchange of
31
32 111 ,A LT II INTER\' l c\\ ' SU RVl·YS
ex peri e nce on these topi cs is a lso important in orde r to continua ll y improve the qua lity of survey stati sti cs.
The foll owing sec ti o ns cover data co ll ec ti o n, po pul ation (sampl e, s ize, structure ) and a summary of cons ide rati o ns re la ted to data process ing and presentati o n. T he conc lus io ns from the co nsultati o ns a re inco rpo rated in th ese sec ti ons .
DATA COLLECTION
Before data co ll ec ti o n can beg in , many dec is io ns and cho ices have to be made conce rning the constructi o n and co ntent o f the questionna ire fo r a hea lth inte rv iew survey a nd re lated issues (reca ll pe ri od, lan g uage, mode of data co ll ec ti o n), a lte rnat ives to persona l inte rviews ( i.e. pos t, te lepho ne), use of proxy info rmants, inte rv iewer cha racte ri s ti cs a nd the use of hea lth examinati ons.
Construction of the Questionnaire
The ques tion s to be in c luded in th e ques ti onna ire a re de te rmined by the purpose of the survey a nd th e analys is pl an th at has bee n dev ised in adva nce. For most of the hea lth fo r all indi ca tors th at ca n be covered by health inte rview surveys, the meas urement obj ec ti ves and a na lyti ca l requireme nts have been defin ed and ag reed (see Chapte rs I and 4 ). Whil e the questi onna ire should conta in a ll the ques ti ons that a re necessary fo r the purpose of the survey and the ana lys is of th e result s, they sho uld be as s impl e and sho rt as poss ibl e, and redundant questi ons should be carefull y avo ided in orde r to minimi ze the burden on th e respondents and to increase the cost- effecti veness of the survey in gene ral. Ques tions mu st be c lear and prec ise and a t a le ve l that th e leas t g ifted person ca n answe r.
The ques ti ons ca n have precoded answers o r a ll ow open responses. Most questi ons in hea lth inte rview surveys have precoded, fi xed re spo nse categori es from whi ch the respondent se lec ts one or mo re answers. The des ign o f thi s type of questi on is more difficult because al l poss ibl e responses should be inc luded. The advantage is th at responde nts ' an swers a re immediately recorded in appropriate catego ri es . With open ended question s the respondent formulates hi s or her own answers, whi ch a re then copi ed ve rba tim on the questionnaire .
111:A LTII INTER\' IE\\' SU RVEY METII ODOLO(;Y 33
Althoug h such ques ti o ns a re somet im es un avo idabl e, inte rpre ting and class ify ing the answers require more sk ill and additi o na l effort o n the part of both the inte rviewe rs a nd the in vesti ga tors.
It is des ira bl e to ask the responde nts onl y those ques ti o ns in th e inte rview schedul e tha t a re re levant fo r them. For exa mple, a man sho uld not be asked about d iseases that o nl y affec t women a nd vice ve rsa, a nd a co ll ege student sho ul d not be asked w hat profess ion he or she has. Bes ides the use of comple te ly di ffe rent ques ti o nna ires for different g roups of respondents, spec ia l routing of ques ti ons is often used. Such routing is a lso necessa ry w hen in some sec ti ons of th e questionna ire a procedure invo lv ing two or mo re stages is used . Thi s procedure is a lso ca ll ed conditi ona l questi oning: a respo nde nt onl y proceeds to the second stage if the answers on the "fi lte r" o r screening ques ti o ns in the first stage have certa in spec ifi c va lues . T hi s procedure is, fo r in stance, often used in the assess ment of menta l hea lth pro bl ems (see the sec ti o n o n chro ni c cond itio ns (menta l) on page 65, a nd An nex 3).
Compl ex ro uting of quest io ns makes hi gh demand s on the des ign and layout of the ques ti onna ire and inc reases the burden on the inte r v iewer o r, in the case of se lf-admini stered questi onna ires, the respond ent. For face -to- face and te leph one inte rv iews, pro bl ems ca used by co mpl ex routings ca n be so lved by us ing the tec hnique of co mputer ass isted inte rvi ew ing (CA I), whi c h is a lready practi sed in some coun tri es. O f course, CA I requires a heavy input of skill ed reso urces in the preparato ry phase, but o nce o pera ti ona l it reaps eno rmo us benefit s fo r the co ll ecti on a nd ed itin g of da ta in large surveys (see the secti o n o n data process ing and presentati o n, beginning o n page 45).
Beca use hea lth in te rv iew surveys usua ll y inc lude a la rge number of questi o ns, some log ica l orde ring of ques ti ons should be used, deter mined on the bas is of psyc ho log ica l and behav io ura l know ledge. For examp le, a ge ne ra l questi on on a certa in topi c ca n come before or after spec ific o nes on the same topi c . T he ad vantage of asking the specifi c question s first is tha t they may stimul a te peopl e to think about the to pic, afte r whi ch the ge nera l quest ion can summari ze the ir vi ews. Howeve r, sometimes a genera l a ttitude (e .g. perce ived hea lth) may be bi ased when spec ifi c items (e .g. di seases) have a lready been men ti o ned; th is is not therefore the appropri a te approach and th e order
34 II F/\ 1 T II INTI-R Vll :W SU RVl· YS
should be reversed. The influence of quest ion order should a lso be considered when designing repeated surveys: differences may affec t comparab ility from yea r to year.
Survey questionnaires should preferab ly include instruments of proven va lidity and re li ab ility.
Memory
The extent to whi ch questions have to rely on the memory of the respondent sho uld be minimized, because recall from memory can be a source of bias . The magnitude of the recall bi as depends on the length of the recall period and the sa li ency of the events to be reca ll ed (besides res pondent atte nti on or moti va ti on in genera l). For exampl e, for the co ll ection of data on temporary restricted ac ti vity, the peri od for which an indi vidual is requested to report restri cted ac ti vity should not exceed two weeks because under- or ove rreporting as a result of memory gaps has been demonstrated when longer periods are used. Similarl y, there can be problems in getting informants to remember accura te ly th e num ber of alcoholi c dri nks they have consum ed in the past seven days. It is poss ible to design di ari es and other aids to memory to minimi ze such difficulti es. Furthermore, it is necessary to exa mine the va ri ability in reporting over the peri od concerned to assess reca ll effects and , possibly, to make adjustments to account for them.
Language
Many countri es have signifi ca nt minoriti es in the population who may not be flu ent in the main language and are th erefore unable to partici pate in the survey if the ques ti onnaire is only ava ilable in thi s language. The use of questionnaires in different languages, interviewers with adequate knowledge of alternati ve languages, or interprete rs should be considered when the ex pected effect of exc luding these groups is not marginal. In genera l there is some loss of standardi zation in questions when different languages are used: it is not always poss ible to find words or phrases with prec ise ly the same meaning. Furthermore, the signifi cance of hea lth and hea lth-related problems can differ substan tially between different cultures. These limitations are even more serious when interpreters are used, because the researcher has no control over the quality of interpretation .
IIEA LTII I, TERV IEW SU RVEY 1ETII ODOLOGY 35
Proxy Informants
The bas ic des ign of many surveys is the co ll ec ti on of interview data from a ll occ upants of sa mples of house ho lds o r addresses. Some information about the househo ld ca n be coll ected from one person who acts in thi s way as proxy info rma nt fo r the othe rs. Thi s is a useful approach and proxy in for mants are therefo re a lso frequentl y used fo r obta ining informati on about children, and in some cases about peopl e w ith menta l and se nsory di sabiliti es. The use of a proxy in fo rmant is, however, less app ropri ate fo r some areas in the like ly content of hea lth surveys, for exa mple questions on fee lings . There is ev idence that deta il s o f spec ifi c impa irments o r di sabiliti es can be di ffic ult to assess with a proxy in formant (33). Thi s appli es in pa rti cul ar to those in the menta l and se nsory areas and to topi cs, such as incontinence, that peopl e may w ish to keep priva te even fro m other members of the ir fa mil y. For perce ived hea lth , however, Van Sonsbee k (34) did not fi nd a bias - at any aggregated leve l in the gene ral popul ati on - as a result of proxy interv iew ing.
The use of proxy info rmants is most justifi ed w hen the true inform ant is in some senses a dependent of the proxy. T hi s is the case wi th young children and the ve ry ill or di sabled . Not onl y w ill the proxy be like ly to be abl e to an swer full y and accurately but, from an ethica l viewpo int, such an app roac h is acceptable . In househo ld-based surveys there are two other groups who tend to be more di f fi cult to contac t: marri ed men and young s ing le adults. Proxy interviews are often take n fo r these.
The te rm " proxy effec t" is used to desc ribe the bia s that occ urs when the report of the true in fo rmant dev iates fro m the report of the proxy informant. Some a reas where proxy effec ts are like ly have a lready been mentioned. A numbe r ofrece nt studi es co uld no t demon strate a proxy effect w ith res pec t to ra th e r seve re hea lth probl e ms (3 4- 36).
Opinions on the rol e of proxy interview ing in re lati on to indicato rs such as a lcohol intake and smoking di ffe r substa nti a lly. On the one hand it ca n be argued that proxy questioning may lead to inc reased, and possibl y truer, estimates of consumpti on than se lf-reporting in areas where these habits are co nsidered socia ll y undes irabl e (37). On the other hand, proxy ques ti oning may not be suffi c iently informati ve
36 111 ,ALTII INTl, RVll · W SU RVl' YS
because many people do not necessa rily disc lose their behaviour to others - particularly spouses and parents - in a detail ed and prec ise manner.
The philosophy of proxy use also differs by country. Whil e in the Un ited Kingdom proxy interviews are onl y accepted as a la st resort , in France and the Netherlands they are used as a standard practice . In order to prevent non-response, proxy intervi ewing should be se riously considered, at least for those parts of the interview where proxy effects are unlikely. For other parts of the intervi ew, it may be poss ible for the interviewer to leave a questionnaire to be se lf-administered by the true informant, and co ll ect it later.
Face-to-face Interviews and Self-completion
The prefe rred mode of data collection in health surveys is th e personal or face-to-face intervi ew, in which th e interviewer asks questions and assesses th e answers, usua ll y by mea ns of a stru c tured qu es tionn aire. Th e advantages are that th e respon se rates are ge nerally hi gher and th e questi onnaires are usu ally fill ed in more compl ete ly th an with oth er methods. For so me topic s, however, it may be useful to introduce some type of "se lf-completion " by the informant. Self-completion meth ods are usually prefe rred when th e subject matter is se nsiti ve (e.g. a lcohol, drugs, contraception or sex ual behaviour) and/or it is difficult to ensure complete privacy fo r an interview. One method is to introduce a se lf-admini stered questionnaire during th e interview and to a llow th e informant time to complete it befo re carrying on with th e intervi ew. Another is to leave th e questi onnaire behind after th e interview and to co llect it late r. With thi s method a ll the qu estion s and answers are dea lt with in complete privacy, although it should be noted that peo ple will tole rate re lat ive ly long intervi ews much bette r than very lon g questionna ires (38). In a further method, th e interviewer asks question s in th e usual way, whil e th e informant replies according to a set of answers presented on a showcard . Thi s procedure can be used for two reaso ns: to inform th e res pondent about th e response possibi lit ies and, when answers are coded , to encourage the re spondent to give the correct rather than a soc ially des irable answer (it is possibl e to carry out interviews in thi s way without th e interviewer being aware of th e answers).
I l l .A l Ill I \/IU{\/ 11· \\ ' SURI '!:.\' ~IElllODOIO(;y 37
Postal and Telephone Surveys
Large-sca le na ti o na l surveys a re ex pensive, and there is continuous pressure to explore a lternati ve methods of interviewing. Posta l surveys and inte rvi ewing by te lephone are cheaper than face- to-face interv iews.
Th e te leph one inte rview is an a ttracti ve meth od beca use it is re lative ly c heap and fl ex ibl e. At present , howeve r, access to te lepho nes is not suffi c ie ntl y w idespread in many co untri es; te lepho ne ownership is like ly to be b iased towa rds those g roups who tend to be in hi gher hea lth sta tus g roups. Other di sad va ntages of thi s method are: the mo re ra pid pace of communi ca ti on ofte n res ults in less co mpl ete in fo rmati on be ing g iven and more " don ' t knows" (39); it is less suitabl e fo r obta ining data o n se nsiti ve issues; and peopl e ove r 65 years o ld may be overreprese nted because th ey are more like ly to be at home to answer th e te lepho ne (32). For hea lth inte rv iew surveys, te lepho ne methods a re not recommended as the so le source of in fo rmati on, but they may be used fo r suppl eme ntati on or va lidati o n purposes.
Postal methods a lso have we ll kn own adva ntages a nd di sadva n tages. Two important di sadva ntages a re low respo nse ra tes , and the difficulti es of identi fy ing indi v idua ls fro m wide ly used sampling frames such as pos ta l fil es. T he response rate is a lso dependent o n the length of the ques ti o nna ire: the lo nge r the quest io nn a ire the lowe r the re spo nse. Whil e te lepho ne inte rviews genera ll y y ie ld hi g her response rates , ma il inte rv iews ca n prov ide data of hi gher qua lity, fo r instance w ith respec t to se nsitive issues (30).
Posta l meth ods can be used effec tive ly fo r parti cular popul ati o ns o r purposes. One exampl e is in fo rmati o n o n in fa nt feeding prac ti ces. Ex peri ence has shown that it is easy to ga in access to wome n who have recently had a baby a nd these women, be ing ve ry in vo lved in the subj ec t matte r of the survey, produce hi g h leve ls of respo nse . Posta l sc ree ning meth ods a re a lso cos t-effec ti ve fo r identify ing sa mpl es of sma ll sub-groups for furth er study, e.g. peopl e w ith di sa biliti es (40) or wo me n of child -bearing age.
Health Interview and Health Examination Surveys
Surveys that cons ist of phys ica l exa minati o ns, functi ona l assessment of lungs and heart, laboratory measure ment of blood and urine, e tc . are
38 HEALTH INTERVIEW SU RVEYS
genera ll y ca ll ed hea lth examin ation surveys . A lthough most of the topics that are norma ll y inc luded in a health survey can be invest iga ted using traditi ona l structured questions in a perso na l interv iew (hea lth interview survey), the scope for inc luding addi ti ona l meas urements and tests is increas ing. For certa in types of disorder (e.g. ca rdiovascu lar di sease) hea lth exam in at ions are essenti a l in providing obj ecti ve info rmati on about the di sease a nd/or its ri sk factors. In such cases, hea lth exa mination and hea lth inte rview data should idea lly be co l lec ted as part of the same survey, because they are complementary. Where thi s is not possible, effo rts should be made to link the data in some way.
Interviewers
Besides the tasks of ask ing the questions as la id down in the question na ire and ga thering the responses, the interviewer a lso has to identify and track down appropria te peopl e to be interviewed, to pers uade them to cooperate in the survey, to clarify certa in issues, to ask suppl emen tary questions if necessary, and to co nfo rm to divers ethical standards. Interv iewers for hea lth interview surveys are not norma ll y hea lth professionals and, g iven the requirements mentioned, considerable attenti on has to be paid to the ir se lection, training and supervis ion. When examinat ions are included in the interview, profess iona l staff may need to be recruited and trained. Exper ience in the Office of Popul ati on Ce nsuses and Surveys in London has demonstra ted , how eve r, that much is possible with carefu l trai ning and monitoring of lay interviewers and standardi z ing of eq uipment. For example, lay inter viewers have carried out meas ureme nts of he ight , weight, v is ion and , more recently , blood pressure using e lectronic sphygmoma nomete rs, and have collected 24-hour urine sa mpl es.
The demographic characteristics of the interviewer - race, age or sex - can produce measurement bi ases when the measurements are re lated to these characteristics ; these bi ases are not present when other top ics a re di scussed (28).
POPULATION - SAMPLE, SIZE AND STRUCTURE
Many of the decis ions about sampl e des ign will depend on the re sources ava ilable and the need for info rmation relating to different
II EALT II INTERV IEW SU RVEY MET HODOLOGY 39
groups. A numbe r of methodologica l aspects, such as determinat ion of sa mpl e size and method of sampl e se lection , a re c lose ly re lated to the main resea rc h questions and the soc iocultural context in which the study is conducted.
Sample
Th e sa mpl e fo r a hea lth inte rv iew survey sho uld rep rese nt the genera l population of a ll or part of the cou ntry. The sa mpl e ca n be drawn by seve ra l methods, for exa mpl e, by us ing:
•
address o r posta l files
e lectora l registers
popul ation reg iste rs
telephone directories.
Some of these sa mpling method s may be biased. As a rul e, e lec tora l reg iste rs only li s t people who are aged 18 years and over, and have the na ti o na lity of the co untry. As me nti o ned before, telephone ownership may be bi ased towards groups w ith hi g her soc ioeconomi c status. A lth o ugh ad dress and posta l files mostly conta in a ll of th e buildings in a country, they a re also bi ased beca use they exc lude those not li ving in pri va te househo lds, such as home less people and residents in institu tions . Furthermore, when on ly one person in the househo ld is inte r viewed, address/postal files a