HIV incidence – labour intensive to measure but key to inform effective HIV prevention programmes

Editor’s notes: It is increasingly clear that in order to control the HIV epidemic, we need to invest not only in the scale up of treatment but also in effective and evidence based prevention programming.  UNAIDS has set ambitious targets to reduce the number of new HIV infections to 500 000 by the end of 2020. A major, ongoing, challenge is that we must use mathematical models to estimate the number of new infections.  We do not have enough data on the actual number of new infections and the characteristics of the people newly infected.  So the household surveys conducted in Rwanda by Nsanzimana and colleagues are an important study.  The authors conducted two nationally representative surveys in 2013 and 2014, and were able to repeat HIV tests in 12 593 people out of 13 728 whose initial test had been negative.  They found 35 people who had seroconverted, which reminds us how large this sort of study needs to be in a setting where HIV prevalence is around 3%.  The incidence rate of 0.27 per 100 person-years (95% CI: 0.18 – 0.35), is higher than the authors had anticipated from previous modelled estimates.  Part of the reason for this was that they found several villages and households with multiple seroconversions, which suggests multiple outbreaks.  This heterogeneity in the pattern of new HIV infections could have important implications for prevention activities and approaches, as well as for methodologies used in designing surveys.  In Rwanda, the risks of HIV infection were higher among older adults (aged 36-45 years) compared to the younger participants (aged 16-25 years); higher in the West of the country and higher in urban areas.  However, the small number of seroconversion events means that the confidence intervals for these comparisons are wide, particularly given the observed heterogeneity between villages in the sample.


Household survey of HIV incidence in Rwanda: a national observational cohort study

Nsanzimana S, Remera E, Kanters S, Mulindabigwi A, Suthar AB, Uwizihiwe JP, Mwumvaneza M, Mills EJ, Bucher HC. Lancet HIV. 2017 Oct;4(10):e457-e464. doi: 10.1016/S2352-3018(17)30124-8. Epub 2017 Aug 8.

Background: In Rwanda, HIV prevalence among adults aged 15-49 years has been stable at 3% since 2005. The aim of this study was to characterise HIV incidence across Rwanda.

Methods: We did a nationally representative, prospective HIV incidence survey for the period of 2013-14, which used two-stage sampling. We randomly selected 492 villages in the first sampling stage and 14 households per village in the second stage. Participants completed a questionnaire and 14 140 people were tested for HIV. 13 728 participants were HIV negative, and were enrolled in the incidence cohort. Participants were retested and surveyed again after 12 months. Weights were calculated as the inverse of the probability to select the villages and the households.

Findings: The study period was from Nov 5, 2013, to Nov 15, 2014. Among 14 222 respondents from 6792 households, 14 140 were tested for HIV and 13 728 were HIV negative. Of 12 593 people who participated in the endpoint data collection activities, 5965 (47·4%) were men and the mean age was 30 years (SD 10·8). 11 237 (89·2%) participants lived in rural areas, 4826 (38·3%) were single, and 7140 (56·7%) were married or cohabitating. During the year, 35 participants had seroconversion, including 13 men and 22 women, resulting in an overall incidence of 0·27 per 100 person-years (95% CI 0·18-0·35). Incidence was 0·21 per 100 person-years (0·10-0·32) in men and 0·32 per 100 person-years (0·19-0·45) in women. Our findings suggested multiple breakouts, with multiple seroconversions occurring in three villages and two households. Incidence was higher in adults aged 36-45 years (0·37 per 100 person-years, 0·12-0·62; adjusted hazard ratio [aHR] 4·49, 95% CI 1·30-14·70) relative to those aged 16-25, higher in western province (0·57 per 100 person-years, 0·31-0·87; aHR 5·90, 1·33-25·28) relative to the northern province, and higher in urban areas (0·65 per 100 person-years, 0·23-1·07; aHR 3·10, 1·28-6·99) than in rural areas.

Interpretation: The incidence of HIV in Rwanda was higher than that previously estimated from models, with outbreaks seeming to contribute to the ongoing epidemic. Characterisation of incident infections can help the national HIV programmes to plan for preventive interventions tailored to the most at-risk populations.

Abstract access [1]

Epidemiology [3], HIV [4]
Africa [5]
Rwanda [6]
  • [7]