Fine-resolution estimates of HIV prevalence in Blantyre, Malawi: a Bayesian modelling analysis of survey, health facility, and household testing data


BACKGROUND: HIV transmission increasingly occurs in cities, and is highly unevenly distributed by age, sex, and neighbourhood. We leveraged HIV prevalence data from multiple sources to develop fine-resolution estimates of people living with HIV (PLHIV) in Blantyre District, Malawi to better focus diagnosis, care and prevention services.
METHODS: Between April 2019-March 2020, we did an HIV prevalence survey among 11,705 adults in randomly selected households in urban densely-populated areas of Blantyre City in Malawi ('MLW survey'). We combined MLW survey data (age, sex, and location) with Blantyre District data from two national surveys (MPHIA and MDHS 2015-16) and clinic antenatal prevalence data. We fitted a spatially-explicit Bayesian regression model to estimate age-, sex-, and location- (500m x 500m grid) HIV prevalence and uncertainty. We used an offset term to account for differences in prevalence between 2015 (MPHIA and MDHS) and 2019, adjusting for temporal trends using Spectrum.
RESULTS: HIV prevalence within the urban Blantyre City area was 15.2% (95% credible interval [CrI] 14.2'16.3%) among adults age 15-49y. This was higher than Blantyre rural (12.2%, 95%CrI 11.1'13.4%). Within Blantyre City, prevalence was highest in women aged 49 years (44.8% HIV-positive, 95%CrI 40.6'48.9%) and among men prevalence was highest in men aged 54 (44.5% HIV-positive, 95%CrI 38.7'50.1%). HIV prevalence was highest in peri-urban areas around Blantyre City and lower in more central areas of the city. Across neighbourhoods in the MLW survey ' all of which were in Blantyre City - prevalence ranged from 12.1% to 20.1%.

CONCLUSIONS: The age and sex distribution of prevalent HIV reflects heterogenous historical transmission dynamics and current HIV risk exposure across the city, and underscores the ageing population of people living with HIV. Highly spatially and demographically resolved estimates of HIV prevalence are useful for strategic service planning and understanding progression of HIV epidemic.