Reader Comments

Post a new comment on this article

Income Inequality and HIV Prevalence

Posted by plosmedicine on 31 Mar 2009 at 00:17 GMT

Author: John Hargrove
Position: Professor/Director
Institution: SACEMA (South African Centre for Epidemiological Modelling and Analysis)
Additional Authors: Brian Williams, WHO
Submitted Date: December 06, 2007
Published Date: December 7, 2007
This comment was originally posted as a “Reader Response” on the publication date indicated above. All Reader Responses are now available as comments.

Two considerations cast doubt on the claim by Piot et al. [1] that income inequality plays an important part in determining the variability in the prevalence of HIV in Africa.

Firstly, using current HIV prevalence is not appropriate since the epidemic has peaked at different times in different countries and this confounds the association. In 2000, for example, the prevalence of HIV in ante-natal clinics in urban Uganda and Swaziland was about 8% and 37%, respectively [2]; in 1992, the corresponding figures were about 30% and 5%, almost exactly the reverse. It would be better to compare data from similar stages of the epidemic in each country; below we use the estimated peak prevalence of HIV.

Secondly, both the proportion of men that are circumcised and people’s religious affiliations have significant impacts on HIV prevalence and the effect of the Gini coefficient should be viewed, in multivariate analysis, in the context of these and other factors [3].

We estimated the peak prevalence of HIV in 28 sub-Saharan African countries, for which Gini coefficients were available, by fitting a double logistic function to the time series of country specific ANC data for urban settings [3]. To guard against outliers we took as the peak value the smaller of the observed and predicted peaks of prevalence.

In univariate analysis, the Gini coefficient was significantly associated with peak HIV prevalence levels (R2 = 0.28; P = 0.004). However, there were stronger associations with the percentage of Muslims (R2 = 0.63; P less than 0.001) and the proportion of circumcised males (R2 = 0.77; P less than 0.001) in a country. In bivariate analyses, using the Gini coefficient and either of these factors, the Gini coefficient was no longer significant (P more than 0.1 in both analyses). On currently available evidence it is thus hard to argue convincingly for a causal link between a country’s HIV prevalence and the degree of income inequality.

After adjusting for the proportion of the population that are Muslim and the proportion of men that are circumcised, eleven countries in southern Africa still had HIV peak prevalences 5 and 14 percentage points higher than countries in the east and west/central regions, respectively [4]. Muslim, and male circumcision, percentages and region accounted for 89% of the variance in peak HIV prevalence in the 28 countries analysed above, and 85 per cent among all 42 continental sub-Saharan African countries.

While the Gini coefficients are significantly higher in southern African than in all other regions of sub-Saharan Africa (P = 0.002), they do not remove any further significant proportion of variance when added to the above model (P > 0.9). Nonetheless, there are clearly unidentified factors particular to southern and, to a lesser extent, east Africa which ensure that these regions have had the most severe HIV epidemics in the world. Elsewhere we argue that these factors are intimately related to the high levels of oscillating migratory labour, associated with large scale mining and farming, which lead to severe disruption of family structure [4].

Further analysis of the kind carried out by Piot et al. [1], but including more potential explanatory variables and examining data on a finer geographical scale, will help us to answer the pressing question: what determines the differences in the magnitude of the HIV epidemic in different countries not only in Africa but throughout the world?


1. Piot P et al. (2007) Squaring the circle: AIDS, poverty, and human development. PLoS Med 4(10): e314. doi:10.1371/journal.pmed.0040314
2. UNAIDS (2006) Epidemiology Fact Sheets. UNAIDS Geneva Switzerland
3. Drain PK et al. (2004) J Acquir Immune Defic Syndr 35, 407-420.
4. Hargrove JW (2007) Migration, mines and mores: The HIV epidemic in southern Africa. November 2007: SACEMA, University of Stellenbosch, South Africa.

No competing interests declared.