Cette page existe uniquement en anglais

Home > Activities > Committees >AIDS (1995-99) > Seminar Report

Seminar for Senior Policy Makers on the Demographic Impact of AIDS organised for the World Bank and uNAIDS

Washington, USA, 12-14 January 1998
organized by the IUSSP Committee on AIDS

Report

As a result of the widespread interest generated by the earlier scientific meeting on this same topic (Durban, 3-6 February 1997) the committee was approached by Debrework Zwedie (Health, Nutrition & Population Section at the World Bank) and Bernhard Schwartlander (uNAIDS) and asked to organize a seminar for senior policy makers at the Bank and in the uN family of organizations, summarizing the presentations from the IUSSP Durban conference and reviewing other published evidence of the demographic impact of HIV. Basia Zaba and Michel Carael spent two days together in Geneva in April 1997 planning the seminar : outlining the programme, listing possible speakers and discussants and preparing briefing documents for the presenters.

The format was finalized in consultation with Bank staff : it was agreed that the first day of the meeting would consist of detailed technical presentations and discussions between demographers and epidemiologists, the second day would be devoted to small group discussions about the main policy implications of these technical presentations, and on the third day the same speakers would present shorter, non-technical versions of their findings, supplemented by two other policy-oriented presentations to a wider, non-specialist audience. This final session was chaired jointly by the Bank’s Vice-president for the Africa region (Callisto Madavo), the Director of the Bank’s Health, Nutrition & Population Division (Richard Feachem) and the Executive Director of uNAIDS (Peter Piot), and recorded on video for dissemination to World Bank staff in national and regional offices.

The topics covered by the technical presentations on the demographic impact of HIV included the following : Mortality - Evidence from Primary Sources in Community Studies (Ties Boerma) ; Mortality - Evidence from Studies of National Level Data from Secondary Sources (Ian Timaeus) ; Evidence of Effects on Fertility (Basia Zaba) ; Impacts on Family and Household Structure (Simon Gregson) ; and Projecting the Impacts of AIDS on Mortality (John Stover). Abstracts summarizing the major findings in these papers follow. These papers will be archived in the uNAIDS 'Best Practice' collection, and it is also hoped to publish them in a special issue of AIDS, and possibly present them again at a round table discussion to be organized by uNAIDS at the Geneva conference on AIDS. The other presentations made on the final day included one by Prof. Roy Anderson (Oxford university) outlining the implications of new findings in the field of molecular epidemiology of HIV for its demographic impact ; a discussion by Desmond Cohen from uNDP about the role of measurement in formulating policy, and an exposition by Peter Piot concerning uNAIDS activities in regions with established & emergent epidemics. There are no plans to publish these three contributions, but Elizabeth Pisani was asked to write up a summary report of the seminar, which will include these contributions, as well as summarising the points made during the discussion sessions.

ABSTRACTS OF MAIN PRESENTATIONS

Mortality Impact of the AIDS Epidemic : Evidence from African Community Based Studies
Ties Boerma

Objectives : To evaluate critically the mortality findings from community based studies in African countries, and to compare selected measures with their equivalents from studies of cohorts selected in clinics and workplaces, and measures from developed country studies.

Methods : Mortality measures reviewed include age-specific mortality of the HIV positive, the relative mortality of HIV+ and HIV-, and indices of impact at the population level. Child and adult mortality effects are considered separately, and interactions with TB are discussed. Most of the data come from the Lake Victoria region.

Results : Community studies indicate a median survival time from infection of about 7 years - higher than early indications from studies of sex workers and blood transfusion patients, lower than estimates based on follow-up of fertility based samples. Mortality among HIV+ adults ranges from 90 to 110 per thousand, with a substantial proportion of HIV infected deaths occurring before the development of clinical AIDS. In communities with prevalence levels between 8% and 20%, the population’s probability of dying between 15 and 60 is 2 or 3 times the same probability for the HIV negative. Women lose more years of life, on average, than men, as they are infected earlier. Excess mortality of children born to HIV+ mothers is most evident in infancy, but is also apparent at ages 1-4.

Conclusions : Clinic-based samples are not suitable for the study of HIV related adult mortality, but may be adequate for studying child mortality impacts. Community based studies are needed urgently in Southern Africa where the epidemic is increasing very rapidly.

Objective : To measure recent trends in all-cause child and adult mortality in national populations in Africa

Methods : The data analysed were collected in the Demographic and Health Survey (DHS) programme of household surveys and in national censuses. The primary index of child mortality used is the probability of dying before age 5. For adult mortality it is the probability of dying between ages 15 and 60. Mortality trends are measured in 3 ways : by comparing mortality data collected in the 1990s with those from the 1980s, by using the birth and sibling survival histories collected in DHS surveys, and by comparing the latter estimates with estimates from data on orphanhood.

Results : under 5 mortality is stagnant or rising in several African countries. In some countries, however, an adverse trend developed too early in the 1980s to be attributable to HIV. The 3 approaches to monitoring increases in adult mortality yield largely consistent results. Adult death rates doubled or tripled between the 1980s and mid 1990s in uganda, Zambia and Zimbabwe. They also rose substantially in the other East African countries for which data exist, but not in those in Western Africa. By 1996, 60% of those Zambians surviving to age 15 could expect to die before age 60. The rise in mortality is concentrated among young adults. In most countries, men have been affected worse so far, but in uganda adult women’s mortality has risen most.

Conclusions : Data can be collected in household surveys and censuses to monitor the mortality impact of HIV in African populations. The available data begin to document the differential impact of the epidemic. In countries where HIV became prevalent by the late 1980s and for which data exist, massive rises in mortality occurred by the mid 1990s. Support needs to be provided for the collection of further such data.

Objectives : To review evidence for the impact of HIV on fertility from empirical sources pertaining to African populations and discuss the implications for surveillance based on ante-natal clinic populations.

Methods : We demonstrate the theoretical equivalence between the rate ratio of fertility in HIV+ and HIV- women and the odds ratio of being infected for pregnant women and the general female population. using this result, we compare fertility rate ratios observed in cohort studies with indirectly calculated fertility rate ratios based on comparisons of cross-sectional HIV prevalence in pregnant women and the general female population in the same area. The plausibility of reported differentials is examined in the light of the statistically significant results obtained from case-control studies and the theoretical predictions from a model of the proximate determinants of fertility and HIV incidence. Estimates of population attributable change are made.

Results : Fertility of HIV+ is lower than that of HIV-, in all but the youngest age group, and the differential increases with women’s age and epidemic duration. Selection for early start of unprotected sexual activity explains the reverse pattern at young ages. Lower fertility amongst the HIV+ causes a population attributable decline in total fertility of the order of 0.4% for each percentage point HIV prevalence in the general female population.

Conclusions : In non-contracepting populations, HIV+ women have lower fertility as a result of infection with HIV and co-infection with other STDs ; behavioural factors tend to enhance this differential. Other factors being equal, HIV prevalence estimates based on ante-natal surveillance underestimate true prevalence in women in the childbearing ages.

Objectives : To review the evidence for the impact of HIV on orphanhood and other indices of family structure.

Methods : Orphanhood measures from community studies of families and households affected by HIV are compared with evidence of general population trends in orphanhood from recent DHS surveys and model predictions made in recent projections. We also use projection models to examine some of the implications of current findings concerning the impact of HIV on fertility and the age-sex distributions of adult deaths for future trends in orphanhood. The more limited evidence concerning trends in widowhood, headship rates and household size is also reviewed.

Results : Paternal orphanhood rates are generally higher than maternal orphanhood rates in the initial phases of the epidemic. Orphanhood prevalence levels lag HIV prevalence by 10 years or more, so current rates are not a reliable guide to future impact. Lower fertility among HIV+ women will lower the orphanhood impact, though this will remain substantial. Changes in other indices of family structure are less amenable to modelling or to generalization across populations.

Conclusions : The relationship between HIV prevalence, orphanhood and other impacts on family structure will change as the epidemic develops. The mechanisms underlying such changes include the saturation of the epidemic in high risk groups, and changing mortality and fertility among the infected.

Objectives : to review national level projected impacts of AIDS on mortality made by different forecasting groups and explain the reasons for the differences ; to assess which indicators of mortality have the most stable and easily described relationship to HIV prevalence ; and to introduce methods for easily making estimates of mortality impacts based on a few key assumptions.

Methods : Indicators compared for different national projections include : net effects on population size, life expectancy, crude death rate and child mortality. To investigate the sensitivity of projected impact on life expectancy of different assumptions about underlying epidemiological and demographic parameters, a thousand projections were made for a notional population using different input parameters, and then the relationship of the outcome measures was compared with the inputs using regression analysis. Simple algebraic relationships were developed to describe the relationship of crude death rate and under 5 mortality rate to HIV prevalence.

Results : uS Bureau of the Census projections predict life expectancy impacts some 20% lower than those predicted by the uN for the same African countries. The regression analysis showed that the crucial assumptions from the point of view of forecasting mortality were : the future prevalence of HIV, non-AIDS mortality, mean adult survival time from HIV infection, and the male-female ratio of AIDS cases.

Conclusions : The key areas of uncertainty in the prediction of HIV impact on mortality are current estimates of HIV prevalence, and whether prevalence can be expected to decline significantly in the next 20 years as a result of prevention programmes, spontaneous behaviour change, successful treatments or vaccines.

Basia Zaba