*Please see Editorial for members of the Editorial Steering Committee.
Correspondence to Richard Marlink, Harvard School of Public Health, Boston, MA, USA. Tel: +1 617 432 4114; fax: +1 617 432 4545; e-mail: firstname.lastname@example.org
The following papers address the methods and assessment of microeconomic (household) and macroeconomic impacts of HIV/AIDS focusing on Nigeria, Botswana and Cambodia. Quantifying the impacts of HIV/AIDS, even in the absence of treatment, has a number of policy and programmatic benefits. First, it can act as a good advocacy tool for greater prevention efforts. Second, it can usefully inform targeting strategies for HIV prevention and care to mitigate economic effects when impacts are shown to be largest. Third, the economic cost of HIV/AIDS to an individual, household, community, or economy provides a useful baseline from which to assess the economic savings of a prevention or care programme. We now have a wealth of studies that show the economic impacts of HIV/AIDS on households in many countries and settings. Rather than further quantifying impact, however, studies that systematically compare interventions in terms of their impact on the quality of life of families and communities will be of great use to policymakers and programmers.
Possible routes of impact at a micro or household level include direct costs, such as medical and funeral expenses, and indirect costs, such as loss of labour time as a result of the morbidity and mortality of HIV-positive household members and the time spent by others to care for them. Numerous macroeconomic impact pathways exist and include factors ranging from labour sector shortages and reduced productivity to intergenerational impacts from lowered health and educational investment in children. Macroeconomic impact estimates are dependent upon the quality of the underlying microeconomic evidence and, when quality is poor, specific assumptions must be made to underpin model parameters.
Beegle and De Weerdt's article begins this section by addressing key methodological issues that cause variation in micro and macro impact estimates. This paper explores how estimates of the household economic cost of HIV/AIDS can be influenced significantly by different definitions of comparator group, cross-sectional versus longitudinal designs and the time horizon used. If no comparators are used, the household HIV cost analysis will use the assumption that non-HIV-infected households would have no cost of ill health, which will overestimate the impact of HIV/AIDS. When a comparator group is used and there are unmeasured impacts of HIV/AIDS on the comparator group, the household costs of HIV/AIDS can be underestimated.
Other papers in the section that show the results of household impact assessments and macroeconomics in the article by Jefferis et al., address some of these methodological concerns. Mahal et al. found, as the Beegle and De Weerdt paper warns, that a subsample of HIV-negative households matched to HIV-positive case households on key drivers of illness had higher costs of illness than the full HIV-negative sample. This propensity score-matched HIV-negative cohort was, therefore, used as a comparator group in order to avoid overestimating the impact of HIV/AIDS. Even with a possible underestimation resulting from the use of the comparator group, Mahal et al. found a significant impact of HIV/AIDS on Nigerian household expenditures, particularly from out-of-pocket payments for healthcare.
When national-level economic impacts of HIV/AIDS are modest, HIV/AIDS-related mitigation interventions can still be guided by impact analyses to target concentrated areas or groups. Alkenbrack Batteh et al. show that in Cambodia, HIV/AIDS often exacerbates economic and social vulnerability, because households with an HIV-positive member have higher medical and funeral expenses, particularly for children. Pathways of possible intergenerational impact here might be driven by reduced nutritional investment in children rather than by reduced education, as school enrolment rates were found to be similar between households. Or, as Gertler and colleagues have suggested, the lack of parental attention and guidance may matter far more than the income effects of HIV illness (P. Gertler et al., The presence and presents of parents: do parents matter for more than their money? 2003; unpublished). Methodologically, Alkenbrack Batteh et al. attempt to address the potential reverse causality of HIV/AIDS and poverty and overestimation of the impact of HIV/AIDS by identifying households of similar socioeconomic status through a nearest neighbour and fixed asset matching approach for comparison with HIV-positive households.
The impact of HIV/AIDS on households is often assumed rather than quantified in discussions of the economic outlook of countries and regions and in the planning of appropriate national and subnational responses to the epidemic. Communities and sectors within an economy do not always endure high costs uniformly, as stressed in the Beegle and De Weert and Jefferis et al. articles; this assumption may reduce the effectiveness of targeted interventions.
Jefferis et al. combine macro and micro level modelling, incorporating evidence from a variety of sources to inform model parameters. In this way, the study relies on the accuracy and the methods of the underlying analysis from which the parameters are drawn. Despite the high incidence of HIV in Botswana, the authors conclude that the country's economy will continue to grow and poverty will continue to decline. The authors also find that Botswana's treatment provision policy has the potential to relieve some of the macroeconomic impacts of HIV/AIDS with less cost to the health sector than expected as a result of the lowered health service burden of illness and increased government revenue from economic growth. It is worth noting, however, that Botswana already provides universal access to free antiretroviral treatment. It, therefore, remains unclear how these results will influence policymakers both in Botswana and in other countries.
Conflicts of interest: None.