Response to: Gregson S, Gonese E, Hallett TB et al. HIV decline in Zimbabwe due to reductions in risky sex? Evidence from a comprehensive epidemiological review

Abstract

Listen

Translate article

The authors of this article1 review several sexual behaviour surveys from Zimbabwe and ask whether the declines in HIV prevalence seen in that country between 1997 and 2007 are attributable to behavioural change. They conclude that declines in the fraction of men and women with casual—or ‘non-regular’—partners, combined with high rates of condom use in casual sex are at least partly responsible for this favourable trend. We wonder why there is no mention of the dramatic decline in the fraction of people with ‘concurrent’ or overlapping relationships observed in the Manicaland study—one of the main studies discussed in this review. Manicaland, Eastern Zimbabwe, is home to one of the best studied longitudinal open HIV cohorts in Southern Africa. Gregson and colleagues have been following this cohort for many years, and in 2006, they reported that between the late 1990s and the early 2000s, the fraction of men and women with concurrent partnerships on the day they were interviewed fell from 18.6 to 11.0% and from 2.2 to 1.3%, respectively.2 This represents a decline of ~41% for both sexes over a ~3-year period. If concurrent partnerships were more frequent before the late 1990s, the decline since the start of the epidemic would be even greater. This is important because there is growing empirical and theoretical evidence that HIV rates are high in Southern Africa not because casual sex is particularly common there, but because people in this region are more likely than people in other world regions to have a small number of (typically two) regular ongoing or ‘concurrent’ partnerships at a time.3 Studies show that small changes in the fraction of people with concurrent partnerships can have powerful effects on the transmission and prevalence of sexually transmitted infections like HIV.4–6 This makes the steep decline in concurrency seen in the Manicaland cohort a strong candidate for explaining the decline in HIV incidence that occurred in this region during that time. Ultimately, a network modelling study would be needed to determine the relative contributions of reductions in concurrency, reductions in casual sex and increases in condom use to the Zimbabwe HIV decline. The remarkable decline in concurrency in Manicaland is reported only in the supplementary online material of Gregson et al.’s 2006 paper. Neither that article nor this one mentions it in the main text. This is odd given the growing interest in examining the empirical evidence for the concurrency hypothesis. We are curious whether the other studies reviewed in this article also have data on concurrency that has not been reported.