Cette page existe uniquement en anglais
Home > Activities > Committees >Emerging Health Threats > Seminar Short Report> Seminar Long Report
Unité d'Enseignement et de Recherche en Démographie, Ouagadougou, Burkina Faso, 12-14 February 2004
Session 1. General perspectives
Session 1 provided a transition from the first seminar on "Determinants of Diverging Trends in Mortality". The first paper by K. Moser, V. Shkolnikov and D. Leon Mortality of the world's population 1950-2000: divergence replaces convergence from the late 1980s demonstrated that life expectancy at birth has converged worldwide up to late 1980s before starting to diverge due to increases in adult-age mortality. This new trend constitutes a major challenge to the general expectation of convergence towards a low mortality regime based on existing concepts of epidemiological transition. J. Vallin and F. Meslé in their paper, Convergences and divergence in mortality: A new approach to health transition, showed that the Abdel Omran's original epidemiologic transition theory can be integrated into a broader scheme. This scheme takes into account all recent changes in mortality patterns, both favourable and unfavourable. The transition includes three stages: vanquishing infectious disease (Omran's epidemiologic transition), the cardiovascular revolution and the fight against aging. Different countries and regions could simultaneously face different stages of the transition or a combination of transitions. While the most developed countries are passing from the second to the third stage, many developing countries have not completed the first stage.
The discussion that followed focused on methodological issues, especially the comparability of the data used. The data are of different quality and were obtained through different methods. For example data on mortality during the 1950s and 1960s from sub-Saharan Africa were obtained through very simple models. The need to look at the urban/rural and sub-regional differential and at the historical context of the decrease of infant mortality in Africa was also stressed. The authors explained that the differential in the quality of data does not affect the big picture portrayed by their papers. The differential in the quality of data was more substantial in the past, at a time when this differential cannot solely account for the observed worldwide divergence in life expectancy at birth. The study of the divergence of mortality actually provides a tool to measure the amount of inter-country diversity in mortality.
Session 2. Evidence from demographic surveillance in Africa - 1
Two papers were presented: by Ian Timæus on Impact of HIV on mortality
in Southern Africa: Evidence from demographic surveillance and by Judith
Glynn on The impact of HIV infection on tuberculosis in Africa.
Using data from several demographic surveillance sites, Timæus found a similarity in age patterns of AIDS mortality in the local populations from Zimbabwe, South Africa, Tanzania, and Namibia. It allowed him to develop a simple one-parameter model for predicting age-specific mortality due to AIDS. Timæus' model can be used to estimate the mortality effects of AIDS, wherever direct mortality data are not available or unreliable. However, it is worth noting that the resulting estimates would depend on contextual characteristics at different settings. Differences may exist in behaviours even though no differences are found in mortality.
Tuberculosis (TB) notification rates have doubled in Africa since the mid-1980s. J. Glynn found that the risk of TB greatly increases due to HIV infection at an individual and population level. Incidence of TB increases with rising HIV-infection rates and will continue to increase for some time even after a fall in HIV incidence due to the long latent period of HIV. Antiretroviral medications can decrease the risk of TB, but would have to be used early and extensively to have a large effect at the population level.
Session 3. Evidence from demographic surveillance in Africa - 2
This session focused on HIV and AIDS related deaths in Zimbabwe (HIV in Zimbabwe
by O.Mugurungi, S. Gregson et al.) and Uganda (Demographic patterns of
AIDS deaths and differences in willingness to have more children in rural southwest
Uganda by L. Muhangi).
In Zimbabwe, sexual networks shaped by cultural and colonial influences and the relatively high but uneven level of development have facilitated the spread of HIV after the mid-1980s. It was rapid and by the mid-1990s about one quarter of the adult population was infected. The epidemic caused crisis-level adult mortality particularly in the most active ages, a reversal of prior gains in child survival, a rapid decline in population growth, and an increase in the number of orphans. A rapid response to the epidemic brought some positive changes in the behaviours since the late 1990s and seems to have levelled off infection rates.
Uganda is one of Africa's success stories. For example, between 1986 and 2001 HIV prevalence dropped from 30 to 6 percent in southwest Uganda due to behavioural changes. Muhangi's paper demonstrated behavioural changes among HIV positive individuals. Demographic surveillance data from the district of Masaka in southwest rural Uganda showed a high frequency of premature deaths from AIDS. HIV positive individuals were more likely than HIV negative individuals to express the intent to have no more children. Despite the stated intent of many HIV positive individuals to have no more children, the desire for more children remained high for a large percentage of the HIV positive informants.
Discussion following the presentation of papers focused on methodological issues. There was a concern about possible underestimation of the infant mortality increase in Zimbabwe. The reliability of the cause of death classification and of census data used in Zimbabwe raised concern and may account for part of the observed differences between urban and rural areas. With regards to the Uganda paper, the participants expressed some concerns about data quality and pointed to the desirability to consider temporal changes rather than simply cross-sectional comparisons. The author replied that his findings were preliminary and the study will be developed further.
Session 4. Country case studies
In this session three country case studies were presented for Burkina Faso, South Africa, and Madagascar: Population et maladies infectieuses au Burkina Faso by B.Baya, Multiple causes of death statistics and HIV/AIDS in South Africa: their changing profile over the period 1997 to 2001 and utility in the era of HIV/AIDS, by S.Bah and Disparition et émergence des décès infectieux et parasitaires à Tananarive by D. Waltisperger.
Baya showed that infectious diseases constitute a large part of mortality and morbidity in Burkina Faso. They result in heavy economic losses due to long periods of disability and premature deaths. Baya argued that the spread of infectious diseases was facilitated by health-related factors--inefficient health system, underutilization of healthcare facilities by the population--and by non health-related factors--lack of resources, poor nutritional status, poor water supply, environment modification, and lack of education. Discussions following the paper suggested that the implementation of the 1987 Bamako Initiative removing the free-of-charge policy regarding health services and widespread poverty explain the underutilization of healthcare facilities by the population. Baya suggested that one solution would be to revise the Bamako Initiative and reduce health fee charges. He also recommended resorting to the community-based approach, e.g. the training of community health agents to implement the malaria programme. Since health systems-based data represent only one twelfth of all deaths that occurred in Burkina Faso, he was advised to also use population-based data from the Nouna observatory.
Bah studied mortality in South Africa using death records listing multiple causes of death. The analysis revealed that TB, influenza and pneumonia played important roles as contributing causes of death. It could be possible, in the future, to estimate the real impact of HIV/AIDS on mortality and misreporting by using regression models. Participants expressed methodological concerns regarding the completeness of death reporting and the problem of selectivity of the sample used by Bah in South Africa.
In Africa mortality data are usually insufficient for building continuous time series analyses. Waltisperger used a unique collection of mortality data from Tananarive to create a time series analysis of mortality by cause-of-death. Systematic analysis showed epidemiological evolution in Madagascar and revealed exceptional mortality increases in 1987 which he attributed to famine. Further research is however needed to understand what exactly happened in 1987. The author's explanation that the increase in mortality was due to famine was unclear. Drops in life expectancy at birth in Tananarive were smaller than those caused by past famines in Finland and Ukraine. It is not clear also why mortality effect was so much greater for men than for women.
Session 5. Country case studies
Papers presented during this session discussed the relationship between population movement and HIV/AIDS in Africa (Population movement and the AIDS epidemic in Africa by J.Anarfi) and more specifically, the relationships between migration and knowledge and perception of HIV/AIDS in Northern Senegal (L'expérience migratoire face à la connaissance et les perceptions relatives au VIH/SIDA. Le cas de la vallée du fleuve Sénégal by F.Waitzenegger-Lalou, V.Piché and R. Lalou).
Anarfi summarized migratory flows in Africa and showed how these flows contribute to the spread of HIV/AIDS. Migration in Africa is circular and does not lead to permanent removal from one's place of origin. Until migrants finally return, they repeatedly visit their homes creating fertile grounds for the spread of STDs. HIV spreads geographically via people's relocation. Mobile people are more involved in risky sexual behaviors. They are exposed to especially unfavorable conditions such as physical and psychological stress and do not have access to adequate healthcare facilities. Women also increasingly take part in migratory movements though with fewer economic opportunities, many migrant women become involved in commercial sex work.
Waitzenegger-Lalou et al. showed that migrants from Northern Senegal who travelled to cities better covered by anti-aids campaigns or to areas severely affected by HIV did not increase their knowledge about HIV/AIDS. However, migrants who stayed in a country with a high HIV prevalence or who had engaged in risky behaviours during migration did have an increased self-perception of being at risk. The discrepancy between knowledge and perception stems from the fact that AIDS is not only a radio message, but also a reality of every day life, especially in Matam, one of the areas studied.
Comments following the presentations suggested it was necessary to move beyond migration and migrants to introduce a contextual approach, which requires a shift from the individual level to the state level. The situation of Algeria - a country with high population mobility, protracted civil unrest and lack of an AIDS education program, but yet curiously without an HIV outbreak defies the hypotheses that migration itself accelerates the epidemic. Anarfi attributed the Algerian exception to the fact that Algerian migration is directed to a country with low levels of HIV (France), a good healthcare system, less dramatic levels of poverty than elsewhere in the developing world, and the role of Islam, which prevented widespread risky sexual behaviours. However, he emphasized that migrants can import HIV, as was the case of Ghanaian female migrants to Cote d'Ivoire. He also argued that the cause of the escalation of the AIDS epidemic in South Africa is the failure to recognize the seriousness of the disease along with high levels of extra-marital sex. Lalou pointed out that Northern Senegal was not a priority in the anti-AIDS campaign in Senegal, despite a serious problem of sexually transmitted diseases in the area and early research findings that linked international labour migration to HIV spread in the region. Migrants perceive themselves to be at risk of being contaminated but not to be at risk of contaminating others.
Session 6. Socio-economic conditions and infectious disease
The three papers presented during this session challenged widespread wisdom and called into question popular consensus concerning the impact of the HIV/AIDS epidemic. They respectively dealt with Poverty and social exclusion by M. Turshen, Development and infectious disease: the case of HIV in Africa by J. Decosas and Conséquences économiques des maladies infectieuses by J. Brunet-Jailly.
Turshen's paper questions the assumption that HIV/AIDS leads to social exclusion in Africa and the very applicability of the concept of social exclusion as it is currently understood to Sub-Saharan Africa. This paper suggests that it is not HIV/AIDS that leads to social exclusion but rather the current economic, political, and social dislocation that may be driving the HIV/AIDS epidemic and other resurgent infections in Africa.
Decosas' paper considers the complex relationship between the HIV epidemic and development in Africa. Although according to some models, the spread of AIDS could lead to the erosion of human capital; there is, however, no evidence of major effects on macro-economic performance and educational level. Social cohesion, which was been seriously affected at early stages of the epidemic, seems to be much less affected at its later stages. Social capital is being created, though it can be easily broken in societies facing famines, wars or severe political exploitation. In general, there is no reason why HIV would stop the African renaissance in the 21st century.
Brunet-Jailly proposed a cost-benefits approach to ranking priorities for resource allocation. Resource allocation should maximize the benefits for the general health status of the population rather than giving automatic priority to a single disease (e.g. AIDS) and advocating for expensive treatments for its victims. Local communities must participate in the decision-making process and make their own choices concerning the allocation of health resources.
The discussion following the paper presentations, focused on the utility of the concept of social exclusion in the context of sub-Saharan Africa, its definition by African communities themselves and its link to HIV. It was recognized that some social organizations take care of infected people in Africa and that there is more acceptance of HIV/AIDS victims as the epidemic matures. It was argued that using volume of the production rather than GDP would highlight the economic impact of AIDS. Studying the case of malaria would also help to gain insight in the economic impact of infectious diseases.
Criticism of Brunet-Jailly's paper stressed that the cost-benefit approach assumes a tight upper limit of resources. This assumption is questionable. As an example, the Secretary General of the UN has managed to gather two billion dollars for the world fund against AIDS. This demonstrates that the funds do exist and can be raised. The author stressed that it is necessary to have a cost-benefit approach to healthcare even though it is a common good.
Session 7. Development, environmental change and infectious disease
This session dealt with the inter-relations between environmental change, agriculture, population growth, and infectious disease. Three papers were presented: Climate change and infectious disease by A. Githeko, Some considerations on inter-relations between agriculture, infectious diseases and HIV/AIDS by J. du Guerny and Population Growth and infectious disease by G. Garnett and J. Lewis.
According to Githeko, the effects of climate change and variability are expressed in short-term epidemics and could lead to long-term changes in disease epidemiology. Up to 50 percent of the anomalies in hospital based highland malaria cases in Kenya can be connected to changes in meteorological conditions. Meteorological data can be used for a precise forecasting of disease outbreaks. Climate variability has the potential to precipitate simultaneous and multiple disease epidemics. During the discussion, Githeko argued that his model could be used to predict epidemics. Health professionals are, however, usually sceptical about models, especially when they do not use hospital-based data. There is a need to educate policy-makers about the usefulness of this model as a tool for planning vaccination campaigns.
Du Guerny focused on farming systems to demonstrate that HIV/AIDS has an impact on agriculture production by affecting the availability, quality and cost of the labour force. So far, the importance of this inter-relationship has been poorly recognised. Du Guerny also explained that irrigation may have a great impact on some infectious diseases like malaria. Improper use of fertilizers and insecticide may also cause problems.
Garnett and Lewis proposed a general model mapping the relationship between population density, the spread of infectious disease and its consequences (e.g. mortality). They argued that population growth has three major consequences for infectious disease: (1) the sheer scale of cities provides more opportunities for disease to persist by providing a large supply of susceptible individuals. (2) Growing population and poverty produce increasing contact rates creating conditions for epidemics. (3) Travel and migration increase global contacts turning epidemics into pandemics. Infectious diseases, in turn, increase death rates amongst young people with potential to reverse the epidemiologic transition. In respect to the communication by Lewis, it was noted that it would be interesting if the model were directly applied to sub-Saharan Africa. However, data availability could be a problem. For example, modelling the spread of HIV would require data on sexual networks.
Session 8. Socio-cultural factors and AIDS
Paul Nkwi et al. examine the reasons why many cultural practices conducive to HIV infection still persist in sub-Saharan Africa despite education and awareness campaigns in their paper The impact of socio-cultural practices on the spread of HIV/AIDS in selected countries. The study is based on contextual analysis of descriptions of various social-cultural practices from Côte d'Ivoire, Togo, Cameroon, Kenya, and Malawi within the framework of an African Population Advisory Council study. The following practices were considered: initiation rituals, widow inheritance (levirate), bodily scarification, female genital cutting, male circumcision, funeral rites including indiscriminate sex. They noted that behavioural change can only occur if norms and values at the societal level are understood as well as the way individuals internalize them and act on them in their daily life. The discussion pointed out that cultural approaches prevalent during late 1980s and early 1990s had serious limitations. The cultural changes they promoted were difficult to achieve. Another problem is how to measure cultural change. Nkwi emphasised that one has to go to the community in order to understand the rationale behind the cultural practices and then let the community find and propose the solutions to the threat posed by the HIV epidemic and initiate changes in their cultural practices.
E. Slaymaker analysed data from nationally representative surveys of 1996 and 2001-2 in the paper: Sexual Behaviour and HIV prevalence in Zambia 1996 to 2002. This paper showed that sexual risk behaviours have become less common and marital dynamics are changing. Never married male informants reported decreases in sex with more than one partner and sex at age under 15. Re-marriage of men has become more common. At the same time, prevalence of HIV was stable between 1996 and 2001. She suggested that HIV prevention efforts should pay more attention to married people.
Session 9. Successes and failures in disease control. What lessons can be learned?
Two papers were presented during this session: Pourquoi la mortalité des enfants ne baissent-t-elle plus depuis une quinzaine d'années en Afrique au sud du Sahara ? by G. Pison and Determinants and dynamics of the sexual transmission of HIV in sub-Saharan Africa: implications for HIV control by M. Caraël and K. B. Nsarhaza.
Pison presented results of a systematic analysis of demographic surveillance data on mortality trends by cause of death from rural Senegal. He showed that during the last fifteen years child mortality has no longer decreased and that this reversal in mortality trends is not solely due to HIV/AIDS. The study attributes the reversal in rural Senegal to a slackening of the immunization campaign and to an increase of malaria prevalence due to chloroquine resistance. In the discussion, participants pointed to the experiences of African countries other than Senegal and biases in structural adjustment programs were identified as being responsible for the worsening in child mortality. In Niger for example, this bias led to an increase in the antenatal care visits in the rural areas and a decrease in the capital city Niamey.
The study by Caraël and Nsarhaza of HIV/AIDS in four cities in sub-Saharan Africa demonstrated that the course of the epidemic depends on numerous inter-related factors. A broad explanatory framework was considered. It included structural determinants such as demographic trends and urbanization, political and economic change, cultural factors (education, male circumcision, attitudes towards the disease, and gender inequality) and proximate determinants connected with certain characteristics of sexual networks. The discussion of Caraël's presentation focused on sexual initiation and its relation to schooling, the role of poverty and the anti-AIDS success stories in Uganda and Senegal. Programmes designed exclusively for adolescents were criticised for not being realistic. Programmes should integrate other groups (adults and women) that are involved in the adolescents' sexuality.
The idea of modifying the characteristics that cause the sexual behaviours was judged as promising. For example, increasing the age of first sex might influence subsequent sexuality and reduce risk factors such as the total number of partners because early sexual debut is associated with greater life-time number of sexual partners in many settings worldwide. However, it is difficult to implement because age at first sex and the age gap between partners depend on factors that are not easy to alter or may have differential and contradictory effects. Women with secondary education delay their sexual debut and first marriage while educated men have their first sexual act earlier than the less educated. In the beginning of the epidemic, those with some level of schooling were most infected. Later on, they started using condoms and now are less infected. Schooling does not reduce risky sexual behaviours; it is sexual education that matters. As for poverty, models are unable to explain its role in the epidemic due to difficulties conceptualizing and measuring poverty.
According to Pison, the Senegal success story may be explained by the relatively high age at first sex, lower prevalence of sexually transmitted diseases, small migrant population and an early and comprehensive response to the epidemic. The few success stories in the continent have demonstrated that success is multi-dimensional and requires the involvement of the community. It is the prerogative of the community to define what is acceptable to them or not.
Session 10. Modelling and predicting the future
In his paper Impact of infectious disease on future demographic trends N.
Brouard proposed a model of the HIV/AIDS epidemic based on an analogy between
reproduction of the epidemic and the reproduction of the human population. Sustainability
of the virus in the population is related to relatively long periods between
being infected by HIV, the onset of AIDS and death. The epidemic in Africa is
also fuelled by a higher rotation of sexual partners due to a higher competition
among men to find a partner. This imbalance in the sexual market stems from
the combination of high African population growth rates during the early 1970s
and leads to a large age gap between sexual partners. In the discussion, Brouard's
purely demographic explanation of the AIDS epidemic in Africa was challenged
by Vallin's comments that stressed the very cultural nature of the age gap between
partners. Brouard acknowledged that culture has an impact but he maintained
that the demographic factors play a key role in the heterosexual transmission
of HIV. He argued that if there was equilibrium in age between partners, there
would be less competition and partner rotation. Homosexual transmission for
example has nothing to do with demography.
In his paper The AIDS epidemic in Botswana: Optimal treatment and prevention, G. Feichtinger applied an age-structured epidemiological model to data from Botswana. This country freely provides antiretroviral medication to all those who need it. However, the model shows that the medication-only approach is inefficient. In some cases it decreases the country's welfare even more than doing nothing at all. Prevention through a change in sexual behaviour appears to be more efficient. This should include school-based programs and mass media education programs to fight AIDS.
The paper Forecasting AIDS impact nationally and in sub-populations by R. Dorrington compared the macro-micro model developed by the Actuarial Society of South Africa (ASSA) with the Spectrum and EPP combination model, upon which UNAIDS and WHO estimates are derived, to assess the impact of HIV on mortality rates. The ASSA model allows one to graph the epidemic at the level of population sub-groups (for example, by region). It makes for a more realistic model and provides a better understanding of the general evolution of the epidemic. On the other hand, it requires more assumptions about parameters for which there is usually little information.
The two last papers draw on mathematical models and raise concerns about methodological issues. Shkolnikov recommended avoiding drawing conclusions before verifying the assumptions of the models.
Final session. Synthesis of the seminar and General discussion
M. Caraël and I. Timæus summarized the papers presented during the seminar using a five- level nested framework. The deepest level represents the interaction between people and disease. This level is nested within the healthcare system, which in turn is nested within the economic, social and cultural context.. The next level is the climate and environment, which is nested in the global context.
Caraël and Timæus remarked that the presentations and discussions during the seminar focused on determinants of morbidity and mortality and their reverse influence on population change. Other topics such as fertility, the mother-child transmission of HIV, the impact of malaria on pregnant women, the interaction between malnutrition and respiratory infections, diarrheic diseases, the role of water supply, the privatization of health services received less attention. There was no presentation of a specific control program. This latter point stems from the fact that no HIV/AIDS control program has been adequately evaluated in the continent so far. Indeed, the results of the existing attempts are disappointing and do not allow for drawing general conclusions. The known success stories so far, Senegal and Uganda, are exceptions that do not provide answers to all questions.
The seminar lent itself to the development of a framework that will be useful for the research and policy-making. There remains the challenge for demographers and sociologists to operationalise this framework. The seminar also demonstrated that the medical approach is not sufficient to address infectious diseases and HIV/AIDS in Africa. Many factors at different levels need to be included into the explanations to understand for instance why countries that are most economically developed and most advanced in their transition in Africa are the hardest hit by the epidemic. One obstacle to overcome is the lack of adequate data, e.g. nationally representative data on adult mortality. It would be interesting for the time being to use demographic and health surveys (DHS) data to look at other factors like health facility attendance. There are also efforts among demographers and anthropologists to link qualitative and quantitative data in order to increase the quality of data.
This report was based on a summary report written by Macoumba Thiam.