The difficulty in measuring and estimating the effects of HIV/AIDS on the number of orphans

Until 2002, various organisations and international institutions yielded estimates of children orphaned by AIDS around the world United Nations, 1995, Hunter and Williamson, 2000, UNAIDS, 2000.

Differing estimates are principally due to the use of various definitions as seen in the previous section but also to the use of different geographical scales regarding statistics and eventually differences in demographic and epidemiological assumptions WHO and UNAIDS, 2002. For example, 56% of the differences concerning maternal orphan estimates between the United Nations and the U.S. Bureau of Census are due to differing HIV prevalence estimates, 14% result from differences in the estimated time lag between HIV infection and death, and 7% to the different assumptions on mother-to-child HIV transmission Hunter and Williamson, 2000.

Consequently, these gaps persist between the projections of the number of orphans based on models of HIV/AIDS and estimations resulting from Demographic and Health Surveys (DHS) and Multiple Indicator Cluster Surveys (MICS) WHO and UNAIDS, 2002.

These discrepancies are significant most of the time as shown by Hunter and Williamson who compared orphan estimates from different sources.

Table 1 : Estimates of the number of orphans due to HIV/AIDS according to different sources.

Source	Accepted Definition	Number of covered countries	Year	Number (thousand)	Projections
					Years of forecast	Number (thousand)
UNAIDS	Children under 15 years old, maternal or double orphans	78 (worldwide)	2001	14 000
U.S. Census Bureau	Children under 15 years old, maternal or double orphans	87 (in sub-Sahara Africa, Asia, Latin America, and Caribbean	1990 1995 2001	903 4523 13 440	2005 2010	20 106 25 296
Neil Monk/AFXB	Children under 18 years old, maternal orphans, nearly orphans, children affected by AIDS	Extrapolations from data published in “Children on the Brink 2000”	2001	40 000	2010	100 000
U.S. Census Bureau (2004)	Children under 18 years, maternal or double orphans	Includes 39 countries in sub-Sahara Africa	1990 1995 2000 2003	550 3000 8500 12 300	2010	18 400

(Source : Appaix et Dekens, 2005 )

In order to sort out this statistical and methodological imbroglio, members of UNAIDS (Reference group on Estimates, Modelling, and Projections) met in July 2001 to adopt a common definition of an orphan due to AIDS on the one hand, and to elaborate standardized and widely accepted methods for estimates and projections of orphan numbers WHO and UNAIDS, 2002.

Thus a new approach to estimate maternal, paternal, and double orphans derived by Grassly and Timaeus was adopted by UNAIDS in 2002  Grassly and Timaeus, 2003. This method was already used by the U.S. Census Bureau to estimate the number of orphans in Africa, Asia, and Latin America; it has henceforth been adopted by UNAIDS, UNICEF, and USAID WHO and UNAIDS, 2002.

The estimation of the number of maternal, paternal, and double orphans due to AIDS or other causes takes into account many indicators presented in the following paragraphs.

 Maternal orphans

According to UNAIDS and UNICEF, « maternal orphans are those children whose mother has died, and where the survival status of the father is unknown (alive, dead from AIDS, or dead from other causes). The number of children born to women who have died from AIDS over the preceding 17 years is estimated using country- and age-specific fertility rates, and the number of these who are still alive and under 18 years old is calculated using a country-specific life table. These calculations take account of the impact of HIV infection on fertility, as well as the probability of the virus being transmitted from mother to child, resulting reduced chances for the child’s survival. The HIV status of the mother in the years prior to death from AIDS must be back-calculated, using estimates of the rate of disease progression (…) The calculations also account for the impact of maternal death on child survival in the year before and after birth, which occurs irrespective of the HIV status of the child Crampin, Floyd et al., 2003, Nakiyingi, Bracher et al., 2003, Ng'weshemi, Urassa et al., 2003 »  UNICEF, UNAIDS et al., 2004.

 Paternal orphans

Since 1990, Hunter and Williamson have worked at developing a method to estimate paternal orphans due to any cause by applying a ratio empirically computed from maternal and double orphans to paternal orphans Hunter and Williamson, 2000. These estimates are based on the male fertility rate. According to UNAIDS and UNICEF, « the population projections based on female fertility schedules imply a total fertility rate for men that, together with standard male fertility schedules, can be used to estimate age-specific fertility for men (…) Male fertility can then be used to estimate the number of children whose fathers died from AIDS in the preceding 17 years in the same way as estimating maternal orphans due to AIDS» UNICEF, UNAIDS et al., 2004. However paternal orphan estimates are calculated with more difficulty given the lack of data on male fertility WHO and UNAIDS, 2002. « To account for the impact of HIV on the fertility of a man’s partner, and the impact of mother-to-child HIV transmission on child survival, additional information on concordance of parents’ HIV status is required» UNICEF, UNAIDS et al., 2004.

 Double orphans

According to UNAIDS and UNICEF, «the number of double orphans due to AIDS as defined can be estimated by calculating the total number of children whose parents have both died from any cause and subtracting those children where both deaths were not due to AIDS (…) The death of the father and the mother are linked due to shared risk factors, such as socio-economic status and also due to heterosexual transmission of disease (…) The number of double orphans is therefore higher than would be expected if deaths were independent. This excess risk of being a double orphan was estimated by fitting a multi-level regression model to data on maternal, paternal, and double orphan numbers from Demographic and Health Survey (DHS) carried out in 31 countries; these analyses reveal that the excess risk, and hence the ratio of double to maternal and paternal orphan numbers is dependent on a child’s age, HIV prevalence five years before the survey, and marriage patterns in the population (proportion of 15- to19-year old unmarried women and prevalence of polygamy) » UNICEF, UNAIDS et al., 2004.

 Validation of estimates

To check the validity, according UNAIDS and UNICEF, «estimates of orphan numbers, based on the methods described above, were compared to estimates of orphans in countries in sub-Saharan Africa derived from household surveys Grassly and Timaeus, 2005 »  UNICEF, UNAIDS et al., 2004.

«Estimates of total orphans ages 0 to 14 from the DHS and MICS surveys were found to be in fairly close agreement with estimates derived from the demographic models, after accounting for an overestimate of adult mortality due to causes other than AIDS»  UNICEF, UNAIDS et al., 2004. In conclusion, these estimates are accurate.

«Estimates of orphan numbers will only be as accurate as the demographic and epidemiological data on which they are based» UNICEF, UNAIDS et al., 2004.

Differences in demographic and epidemiological data and assumptions in the past have led to differing estimates of the numbers of orphans due to AIDS by different organisations United Nations, 1995, UNAIDS, 2000, Hunter and Williamson, 2000. The data and assumptions have changed over time and methods are standardized. Thus, a consensus has been reached on global estimates of numbers of orphans and of the impact of HIV/AIDS.

 Biased data

Despite the standardized projections and estimating methods, discrepancies persist and some data seem to be missing. Orphan crisis in Africa is assessed using demographic projections and also from household surveys which raises the question of the survival of parents. “While the two methodologies—modelling exercises and household surveys—give broadly similar results for the overall number of orphans, there are some significant country-specific differences. These differences spring from limitations with both methodologies” UNICEF and UNAIDS, 1999.

Thus, according Grassly and Timaeus, “since 1990, most of the household surveys collect data on the survival status of the parents of all the children living in the household at the time of the survey, which enables orphan identification (maternal, paternal, double). These studies have 6.5% of non response regarding the parents’ survival status” Grassly and Timaeus, 2003. The estimates based on household surveys do not include children outside of family care.
Indeed, a certain number of orphans live on the streets or work as domestic workers. It is difficult to evaluate the scale of AIDS orphans by accounting for all the children who are not supported by the community or the extended family. This entails an underestimation of orphans UNICEF, 2003.

Furthermore, as explained previously, orphans in infancy who are cared for by the extended family are often not recognised as orphans and are taken in by the family as biological children. Timaeus, 1998. Likewise, when the father of maternal orphans remarries, it can happen that the step-mother considers her husband’s children as her own Bicego, Rutstein et al., 2003, Urassa, Ng'weshemi et al., 1997.

Finally, it is difficult for surveys to collect data on the survival status of parents who do not live with their children. This problem entails an overestimation of the number of orphans when a parent’s prolonged absence is considered as a death. By contrast, there is an underestimation of the number of orphans when the death of the parents has not been registered or when the household has not been informed. Respondents may also misreport the survival status of the biological parent. There is a concern about the consequences but also the extent of these missing data Grassly and Timaeus, 2003.

Besides, the number of children living with parents infected with HIV/AIDS is unknown. In most African countries, there exists no estimation of the number of children living with parents infected with HIV. However, these could shed light on the scale of the future orphan crisis World Bank, 1997, Sengendo and Nambi, 1997.

As concludes the World Food Program:

Carrying out a census of the number of orphans is at best an inaccurate exercise. Figures vary considerably whether we include those who lost their mothers, their fathers or both; or those who are under 15 or 18 years of age. They also depend on whether parents who died of AIDS or of other causes are taken into account Landis, 2002.

^ Top