Click a topic below for an index of articles:

If you would like to submit an article to this website, email us your paper to info@heart-intl.net

We offer a monthly newsletter dealing with the various issues surrounding infectious diseases. To find out more click HERE.

♥

♥

The Impact of HIV/AIDS on Population Growth in Africa

Prepared for The Global Coalition for Africa by John Stover
OPTIONS Project
The Futures Group
March 1993

I. Introduction

When it was first recognized that AIDS would be a critical global health problem and that many nations in Africa would be particularly hard hit by the disease, there was speculation that AIDS might ultimately lead to population decline in many countries. Over the last six years various attempts have been made to model the spread of HIV and the impact of AIDS. In a number of instances these models have been used to assess the impact of AIDS on population growth. The results have been quite diverse. Some researchers have reported that AIDS will lead to negative population growth while others have said that this is not the case. Media reports on this debate have tended to highlight the potential negative impacts with headlines such as "Briton Sees AIDS Halting Africa Population Rise" (New York Times, June 22, 1992). This report takes a critical look at this issue, examines the reasons for the apparent disagreement and summarizes the best evidence to date.

II. Background

Several researchers have examined the issue of the impact of AIDS on population growth. [UN, 1991] The conclusions that they report vary considerably. Consider the following two quotes:

"...new analyses support earlier predictions that in the worst-afflicted areas AIDS is likely to change population growth rates from positive to negative values in a few decades." [Anderson, 1991]

♥

♥

"...we can conclude that population growth rates are unlikely to turn negative in Central Africa. More likely, the population growth rates in Central and East Africa will not drop below half their current values." [Bongaarts, 1989]

Anderson, May and colleagues have presented a number of papers describing the results of modeling work that demonstrates that AIDS could lead to zero or negative population growth. [Garnett and Anderson, 1992; Anderson, 1991; Rowley, 1990; Anderson, 1988]. They use an age-structured model with assumptions based on international data to describe epidemiological processes, rates of partner change and demographic processes. Their results show that AIDS can cause negative population growth in a population with an initial growth rate of 4 percent within 30 to 60 years. HIV prevalence rates rise to 50 percent at the time population growth rates become negative.

Anderson's work shows that AIDS can lead to negative population growth under some circumstances. However, his work has often been misinterpreted and misquoted by the popular press. The most glaring example of this is a New York Times article in June of 1992 that reports on comments made by Anderson at a conference in Nairobi. In this article he is described as stating that Uganda is likely have negative population growth after 2002 and the worst affected countries in Africa will have negative population growth rates within 20 years. Anderson's actual comments were quite different. He said that population growth might become negative in some countries within 20 years and that it is likely to become negative in some regions of Uganda (Rakai, maybe Kampala) within the next decade.

Bongaarts prepared a model of AIDS and population growth that is age-structure and includes both high and low risk groups [Bongaarts, 1989]. He finds that the population growth rate in a high fertility setting would not decline by more than half even in the most severe cases. The major reason for this result is that HIV prevalence rates level off at about 20 percent of adults, never reaching the high levels that are obtained in Anderson's model.

Way and Stanecki [Way 1992, Way 1991] used a different model to reach conclusions similar to Bongaarts. Their findings are based on the iwgAIDS model (Inter-Agency Working Group on AIDS) [Stanley, 1989; Seitz, 1991]. This model disaggregates the population by several dimensions including age, sex, urban/rural residence, marital status, STD status and risk group. In a simulation of a typical African country they report that the population growth rate would decline from about 2.8 percent per year to 2.2. percent per year over a 25 year period due to demographic trends alone. With the impact of AIDS included the population growth rate would decline to 1.8 percent, as a result of HIV prevalence levels reaching about 8 percent of adults after 25 years.

The Bulatao Model used at the World Bank [Bos and Bulatao, 1992] also examines the spread of HIV through a series of behavioral and epidemiological equations. It shows that AIDS will have an impact on population growth but will not lead to negative population growth in any country in Africa. This model has been used to estimate the impact of AIDS on population size in all African countries. Bos and Bulatao do not expect AIDS to reduce population growth rates by more than 0.5 percent in even the most severely affected countries.

A review of this work leads to two key questions:

1. Why do different researchers, each using sophisticated modeling and analysis techniques, reach different conclusions about the impact of AIDS on population growth?

2. What is this impact of AIDS on population growth likely to be in Africa?

III. Estimating the Demographic Impact of AIDS

Estimating the demographic impact of AIDS requires information about how AIDS affects the natural demographic processes of births and deaths. For AIDS to lead to negative population growth it would have to increase the death rate and/or reduce the birth rate so that the death rate exceeds the birth rate.

Fertility

It is clear that births may be affected if many women die before reaching the end of their childbearing years. However, most births occur to women at young ages. The average age at the time of death from AIDS is ususally around 30 or higher for women. Therefore, the effect of AIDS deaths to potential mothers on the birth rate is not likely to be large if the total fertility rate remains constant. Most published modeling results confirm that this is the case.

However, it is not clear how the total fertility rate might be affected. (The total fertility rate is the average number of children a women would have if she followed the current age pattern of fertility and survived to age 50). Some women may want to have as many children as possible while they can, in order to leave descendants behind. Others may decide to stop childbearing upon learning that they are HIV positive in order to avoid leaving orphans behind. We do not know which response will predominate. However, since most people in Africa do not know if they are infected or not, a large effect on the fertility rate is probably not likely.

Age at marriage may also be affected and could, in turn, affect fertility rates. A lower age at marriage or first union could raise fertility rates since women would be exposed longer to the possibility of pregnancy. Conversely, higher age at marriage would lead to lower fertility rates. Age at marriage could be lowered as men seek younger partners who are less likely to be infected and as parents want to see their children in stable unions as soon as possible. On the other hand, attempts to postpone the beginning of sexual activity as long as possible in order to protect children may lead to later age at first union.

Mortality

It is obvious that AIDS will have a significant effect on mortality. Although our information about the dynamics of infection leading to AIDS and to death is not complete, enough is known to make reasonably accurate projections of AIDS death rates once the prevalence of HIV infection is known. The incubation period (the time from infection with HIV to the development of full-blown AIDS) is about 8-10 years in industrialized countries. In developing countries this period may be even shorter, perhaps as short as 5-6 years, but good data are lacking. The average survival time after the development of AIDS is about one year. Therefore, the total time from infection to death is in the range of 7-11 years. This implies that, once prevalence stabilizes, mortality from AIDS will result in 9-14 percent of the infected adult population dying each year.

For infants the progression to AIDS is much quicker. About 25-40 percent of all children born to infected mothers will themselves be infected with HIV. The average survival time for infants infected at birth is only about two years.

Based on this information it is possible to estimate the adult prevalence of HIV that would be required to cause negative population growth. In fact, there is general agreement among the modelers that adult HIV prevalence would have to increase to 30-40 percent to reduce population growth to zero. An illustration of this calculation is given in Box 1.

♥

♥

Box 1. Illustration of Adult HIV Prevalence Required to Cause Negative Population Growth

Assume that adult HIV prevalence is 40 percent. If the average incubation period is 8 years (from infection to AIDS) and the time from the onset of AIDS to death is one year, then roughly 1/9 of infected adults will die each year from AIDS. Thus, 4.4 percent would die each year. Since adults constitute 55 percent of the total population in Africa, then this translates to a crude AIDS death rate of 2.4 percent. This is enough to reduce the population growth rate from 3 percent to 0.6 percent.

Next, consider perinatal transmission. If 30 percent of babies born to infected mothers are themselves infected, then the percent of new borns that are infected is 40% x 30% or 12%. Some of these babies (about 10%) would die anyway from other causes. So the net result is additional deaths equal to about 11% of births. Since the crude birth rate in Africa is about 45 per thousand population, these additional deaths are equivalent to an increase in the crude death rate of 5 per thousand (45 x 11%). The remaining population growth rate of 0.6% drops to 0.1%.

If a shorter incubation period is used then the population growth rate would become negative, with a longer one it would be a larger positive value. Box 2 shows the adult HIV prevalence that would cause zero population growth, given typical demographic values for Africa, as a function of the incubation period. The range is from 30-50 percent.

Box 2. Adult HIV Prevalence That Will Produce Zero Population Growth by Length of Incubation Period
Incubation Period	Adult HIV Prevalence	Assumptions:
(Years)	(Percent)	Crude birth rate = 45 Crude death rate = 15 Rate of natural increase = 30 Infant mortality rate = 100 Perinatal transmission (percent) = 30 Life expectancy after AIDS (years) = 1 Percent of population over 14 = 55 Impact on birth rate is neglible
5	30
6	34
7	37
8	41
9	45
10	48

Migration also plays a role in population growth. In this dicussion we have ignored the effect of changing migration rates as a result of AIDS. Some people may move out of regions of high HIV prevalence in order to reduce their risk. However, it is also possible that people would migrate to high AIDS areas, such as cities, because of employment opportunities that might result from increased AIDS deaths.

The differences among modelers about the likelihood of negative population growth are duein part to differences in the simulation models used by each group. However, both sides agree that, in the African setting, negative population growth will not occur unless adult HIV prevalence rates reach 30-50 percent. The Anderson group tends to use pessimistic assumptions that put it at the lower end of the scale and the other groups tends to use more optimistic assumptions that put them at the higher end of the scale. However, there is basic agreement that adult HIV prevalence needs to reach this range before negative population growth occurs.

All researchers agree that, in the African setting, negative population growth will not occur unless adult HIV prevalence reaches 30-50 percent. The big uncertainty lies in projecting future levels of HIV infection.

IV. How Likely Is It That Adult HIV Prevalence Will Reach 30-40 Percent in Africa?

It is difficult to predict future levels of HIV infection in Africa because there is much that is still not known about the epidemic. However, there are several approaches that can help us understand the problem:

1. Simulation modeling

2. Surveys of AIDS-related behavior

3. HIV surveillance studies.

1. Simulation Modeling. Most simulation models of AIDS that differentiate the population by risk factors (multiple partners, condom use, urban/rural residence, etc.) find a plateau effect. That is, in populations with AIDS epidemics, the level of adult seroprevalence will eventually level off well below 100 percent, even if no behavior change tales place. These plateaus occur as certain high risk groups become saturated. In prostitute groups, for example, infection levels may quickly rise to about 80 percent. At that level the prevalence may stabilize as infected people drop out of the group or die and new, uninfected, people join the group. The most susceptible age groups may also saturate quickly, reaching a prevalence level where the entry of new, uninfected members and the exit of infected members due to death, causes an equilibrium to be reached. This does not mean that HIV incidence is zero, simply that new cases are balanced by death or out migration. The level of the plateau will depend on the proportions of the population practicing risky behavior. However, these models all agree that infection will eventually plateau at some level well below 100 percent. Simulation exercises with several different models that disaggregate the population by AIDS-related behaviors that adult prevalence plateaus as high as 30-35 percent may be reached in some population groups as large as major cities, but that for entire countries plateau levels above 15 percent are unlikely.

2. Surveys of AIDS-Related Behavior. Over the past several years the Global Programme on AIDS (GPA) of WHO has sponsored a series of surveys designed to collect information on knowledge, attitudes, beliefs and practices related to AIDS. These surveys have been conducted in over 20 countries. Although full results have not yet been made public, some results have been reported. [Carael, 1990] Of most interest to this subject are the proportions of men and women who report having sex with multiple partners, since these will be the people most at risk for HIV infection. (Of course, even those with a single partner can be at risk if their partner has multiple partners.) The results from six African countries show that 2-17 percent of adult women and 6-53 percent of men report having multiple partners, see Table 1.

Table 1. Percent of Adults Reporting Multiple Partners

in Last Twelve Months

Country	Men	Women
Central African Republic	14	5
Côte d'Ivoire	53	16
Lesotho	30	17
Togo	20	2
Kenya*	24	8
Rwanda**	8	3

* Adjusted for all adults from reported figures on sexually active adults.

** In last six months

Source: Based on Carael, 1990.

These results indicate that, in at least some countries, a significant proportion of people are at risk for HIV infection. Translating these figures into ultimate plateau levels is difficult for several reasons. Plateau levels could be much lower than the percent having multiple partners at any one time if the number of partners and frequency of contact is low, the average HIV prevalence is low, or condom use is high for casual contacts. On the other hand, plateau levels could be higher than these reported figures if the figures understate the actual levels because of under-reporting, if the number of contacts is high, HIV prevalence is high, or condom use is low.

3. Surveillance of HIV Infection. A large number of studies have been performed at hospitals and clinics and special sites to determine HIV prevalence among client populations. The most useful of these for estimating general prevalence levels are usually reports by ante-natal clinics since these studies report HIV prevalence among pregnant women. These results will be somewhat higher than overall adult prevalence levels (since pregnant women are all sexually active) but can give a good idea of levels and trends.

Figure 1 presents graphs showing trends in HIV prevalence from population groups in six different countries. These graphs are all examples of population groups where HIV prevalence is still increasing. They illustrate the fact that we are still in the earlier stages of the epidemic. In many cases we do not have a good idea of just how high prevalence levels may reach. In four of the populations shown in these graphs, prevalence is significantly below the levels required to cause negative population growth. In urban Malawi and Lusaka, however, prevalence is approaching the 30 percent level. These graphs confirm that prevalence is still increasing in many areas but do not indicate whether prevalence levels in the range of 30-50 percent are likely.

Figure 2 presents graphs showing trends in HIV prevalence in five different population groups. These graphs are all examples of population groups where HIV prevalence has apparently reached a plateau or is increasing very slowly. These graphs illustrate several points. First, apparent plateaus are reached in some population groups. This conclusion must be qualified because it is possible that these reflect temporary plateaus and not permanent ones. Second, note that the highest plateau levels occur among prostitute groups, as expected. The other graphs are all for urban areas. Since infection rates are usually much higher in urban areas than in rural areas, the plateau levels for the entire country would likely be considerably below these levels. Third, in Abidjan and Kinshasa apparent plateau levels are considerably below the levels required to produce negative population growth. In Kampala, however, prevalence is approaching 30 percent among pregnant women.

Please contact author for copies of the graphs in Figures 1 and 2 (j.stover@tfgi.com).

V. Conclusion

What conclusions can we draw from the evidence presented above?

- Anderson et al argue that we are still at the early stages of the epidemic. We simply do not know where prevalence will plateau. We do not have enough knowledge to state confidently that it will plateau below 30 percent. They agree that it seems unlikely that HIV prevalence would get that high in entire countries but argue that we really do not know enough to rule it out. Therefore, it is possible negative population growth will occur.

- Way and others argue that there is evidence from both simulation modeling and epidemiological data that HIV prevalence does plateau. In some specific areas, such as the worst affected areas in Uganda (Rakai) and some cities (Kampala, Lilongwe) these plateau levels might well approach 30-40 percent. But the rural areas are likely to plateau at much lower levels. Therefore, it is very unlikely that levels of 30-50 percent will be reached for any area as large as a country.

There is no reason for complacency. Worldwide the epidemic is still in the early stages. Even if AIDS does not cause negative population growth in any county, it will have severe consequences. Millions of people will be affected by AIDS either directly or indirectly.

Of course, this entire discussion has left out the effects of AIDS interventions and spontaneous behavior change. Certainly some behavior change has taken place. AIDS interventions have proven successful in pilot studies and condom use is certainly increasing dramatically in many African countries. To the extent that these changes continue, negative population growth becomes more unlikely. However, we have not yet seen epidemiological evidence that either spontaneous behavior change or interventions can stop prevalence from rising on a national scale.

AIDS is likely to cause negative population growth in some sub-national regions in Africa and might do so in a small number of major cities.

AIDS is unlikely to cause negative population growth in any entire country in Africa.

The actions required as a result of these conclusions seem clear.

- Effective AIDS control programs need to be implemented in order to control the spread of AIDS to the maximum extent possible. These need to include efforts to develop vaccines and drugs as well as efforts to inform people and encourage behavior change, condom use and STD control.

- Effective family planning programs are still needed. Most of these programs are designed to provide couples with the information and means to have the number of children they want, when they want them. The use of family planning also reduces maternal and child mortality by spacing births and limiting births to older women and those who have had many children. These needs will still exist no matter what the effect of AIDS on population growth rates. Family planning programs can also serve to reach the sexually active population with messages, counseling and services to combat the spread of AIDS.

- Governments should plan for the impact of AIDS on their national development. In some regions of a country, there may be pockets of very high prevalence that could lead to zero or negative population growth. This is unlikely for the entire country. However, the impact of AIDS-related mortality on special populations (capital city, key occupations, tourism potential) should be recognized and addressed by planners and policy makers.

References

Anderson, R.M., R.M. May, M.C. Boily, G.P. Garnett and J.T. Rowley, "The Spread of HIV-1 in Africa: sexual contact patterns and the predicted demographic impact of AIDS", Nature, Vol 352, August 15, 1991, pps. 581-589.

Anderson, R.M., R.M. May and A.R. McLean, "Possible Demographic Impact of AIDS in Developing Countries," Nature, 1988, 332:228-234.

Bongaarts, John. "A Model of the Spread of HIV Infection and the Demographic Impact of AIDS", Statistics in Medicine, Vol 8, (1090), pps. 103-120.

Bos, Eduard, Rodolfo A. Bulatao, 'The Demographic Impact of AIDS insub-Saharan Africa: Short- and Long-Term Projections", International Journal of Forecasting, Vol. 8, 1992, pps. 367-384.

Garnett, Geoff P., Roy M. Anderson, "No Reason for Complacency about the Potential Demographic Impact of AIDS", Transactions of the Royal Society of Tropical Medicine and Hygiene, (in press) 1993.

Piot, P., F.A, Plummer, F.S. Mhalu, J.L. Lamboray, J. Chin and J.M. Mann (1988), AIDS: An International Perspective, Science 239: 573-579.

Rowley, Jane T., R.M. Anderson and T. Wan Ng, "Reducing the Spread of HIV Infection in Sub-Saharan Africa: Some Demographic and Economic Consequences", AIDS, 1990, 4:47-56.

Seitz, Steven Thomas, iwgAIDS User's Manual, Version 3.0, Merriam Laboratory for Analytic Political Science, University of Illinois, December 1991.

Stanley, E. A., S. T. Seitz, P. D. Johnson, P. O. Way and T. F. Curry (1989), "The United States Interagency Working Group Approach: The IWG Model for the Heterosexual Spread of HIV and the Demographic Impact of the AIDS Epidemic," The AIDS Epidemic and Its Demographic Consequences, Proceedings of the United Nations/World Health Organization Workshop on Modelling the Demographic Impact of the AIDS Epidemic in Pattern II Countries, New York, December 13-15, 1989.

Way, Peter O., Demographic Impact of HIV in Less-Developed Countries, presented at the VIII International Conference on AIDS, Amsterdam, July 1992.

United Nations, The AIDS Epidemic and Its Demographic Consequences, United Nations Department of International Economic and Social Affairs, New York, 1991.

Way, Peter O. and Karen Stanecki, The Demographic Impact of and AIDS Epidemic on an African Country: Application of the iwgAIDS Model, Center for International Research, U.S. Bureau of the Census, January 1991.