HIV fall in Zimbabwe due to deaths and behaviour change, not migration

Behaviour change and high AIDS death rates contributed to the substantial decline in HIV prevalence in Zimbabwe from 29% in 1997 to close to 16% in 2007 according to findings published by Simon Gregson and colleagues in the April 20th advance online edition of the International Journal of Epidemiology.

While there is convincing evidence of sexual behaviour change, the authors caution “the decline in HIV prevalence also reflects a continuing period of sustained crisis-level mortality and in 2008 it is still the case that one in every seven adults is infected with a life-threatening virus.”

While it is difficult to determine the actual numbers, recent international migration from Zimbabwe is considerable. However, the authors note, limited data as well as mathematical modelling did not provide evidence of this being a significant contributing factor to the decline or that HIV-infected individuals were any more likely to immigrate than those not infected.

Data from antenatal care surveillance and general population surveys suggest that Zimbabwe is the first country in the southern African region to provide evidence of credible declines of people living with HIV.

A decrease in national HIV prevalence can result in significant declines in the death rate over a 5-10 year period. The natural history of a disease can explain a decrease in both new infections and existing infections as the authors note. Once the infection has reached it saturation point in those populations most vulnerable to the illness the transmission rate will slow down to eventually be less than the numbers who die. So, the authors stress, such decreases cannot necessarily be explained by changes in sexual behaviour.

However, understanding the effect behaviour change may have had on the decline, is critical for programming.

The authors undertook a comprehensive review of all epidemiological data for Zimbabwe to

• confirm the existence of a decline in HIV prevalence
• establish what effect the decreasing numbers of people newly infected with HIV and the rising death rates had on the falling numbers of people living with HIV, and
• determine if changes in sexual behaviour contributed to the decline in HIV infection.

The authors analysed all published as well as unpublished data on HIV prevalence in Zimbabwe from the mid-eighties onwards. National estimates show prevalence increased in the early to mid 1990s, to stabilise in the late 1990s at 29.3% before falling to 15.6% in 2007.

The authors presented for the first time comparisons of data on HIV incidence in cohorts of pregnant women and male factory workers at different time points. Among pregnant women HIV incidence fell from 4.8 % per 100 person-years at risk in 1992 to 3.4% in 1999 and among male factory workers from 3% in 1994-5 to 1.3% in 2002-3.

Rates of HIV prevalence among 15-24 year old is considered a useful indicator of HIV incidence in this age group. Using the Genscreen test HIV prevalence among 15-24 year old women attending antenatal care clinics fell from 30% in 2000 to 20% in 2004. However, because of the low specificity of this test under field conditions and a resulting overestimation of prevalence a new combined algorithm test was introduced. With this test prevalence among this age group fell from 21% in 2002 to 13% in 2006.

These declines were seen from 2002 and 2006 in all socio-economic areas (urban, rural).

In the Manicaland study (rural area) from 1998-2003 prevalence in the general population fell from 5.4 % (95%CI: 4.5-6.5%) to 3.3% (95% CI: 2.6-4.1%) in men aged 17-24 years and from 16.5% (95%CI 14.9-18.1%) to 8.8% (95%CI: 7.9-9.7%) in women aged 15-24 years.

These findings, note the authors, together with back-calculation estimates based on vital registration data from Harare show that HIV incidence may have peaked in the early 90s and fallen in the late 90s. The trends in incidence and mortality show that the decline in HIV prevalence was driven by a combination of falling HIV incidence and high mortality the authors add.

The authors note that HIV incidence fell after 1990 in line with the natural disease history among a heterogeneous population leading to HIV prevalence levelling by 1997-2002. Yet they point to mathematical modelling of the HIV epidemic in Zimbabwe that showed that the “pace of the decline in HIV prevalence could not have occurred without changes in behaviour or other risk factors, probably between 1999 and 2004.”

Further analyses revealed evidence of significant reductions in the proportions of individuals, notably men, reporting casual partners during this time. Condom use in this population did not increase after 1999. Yet, the authors note that the high level of pre-existing condom use probably contributed to the low proportion of men not using condoms with casual partners.

The authors also highlight the potential impact of early condom use on HIV incidence.

Inconsistencies in trends in age at first sex did not provide evidence of having contributed to a decline in HIV prevalence the authors note. Even where there is evidence of delayed sexual debut the numbers are too small to have had any effect.

While the authors point to the strength of their findings because of the consistency of the wide range of data, they note the potential for time-related biases in estimates of trends. For example, HIV prevalence trends may reflect improved testing capabilities and administration. Or, differences in survey procedures or a bias in what is socially desirable may be reflected in trends in HIV incidence and sexual behaviour.

The authors conclude that these “findings provide the first convincing evidence of an HIV decline accelerated by changes in sexual behaviour in a southern African country.” Further analysis and investigation is needed to understand the reasons why behaviour changed they add. The decline in HIV prevalence nonetheless, the authors stress, also highlights AIDS-related mortality and prevalence remaining at crisis-level.