Bayesian geo-additive spatial modelling of HIV prevalence using data from population-based surveys

Samson B. Adebayo, Ezra Gayawan, Adeniyi F. Fagbamigbe, Fatai W. Bello

Research output: Contribution to journalArticle

Abstract

Introduction: Estimates of human immunodeficiency virus (HIV) prevalence in Nigeria have been based on the data from HIV surveillance and sentinel studies among pregnant women attending antenatal clinics at some selected sentinel sites. However, such data overestimate HIV prevalence. This paper explores possible geographical variations in HIV prevalence among the general population of males and females based on two waves of the National HIV/acquired immune deficiency syndrome (AIDS) and Reproductive Health Surveys. Material and methods: Data were extracted from the cross-sectional 2007 and 2012 National HIV/AIDS and Reproductive Health Serological Surveys of men (15-64 years) and women (15-49 years) covering all states of Nigeria. Bayesian geo-additive modelling technique was employed for analysis. Appropriate prior distributions were assigned to the different types of variables in the models and inference was based on the Markov Chain Monte Carlo (MCMC) technique. Models of different specifications were considered. Results: The findings reveal significant spatial variations at a highly disaggregated level of states in Nigeria. The nonlinear effects of respondents' age show a similar pattern of HIV prevalence for male, female and the combined respondents, implying that HIV prevalence is peak among middle-age individuals, from where it declines with age. Also, the results reveal a downward change in HIV prevalence in Nigeria between 2007 and 2012. Conclusions: When these findings are taken into consideration in designing intervention strategies, it is believed that each state can be targeted with the right intervention(s). This can also lead to efficient utilization of the scarce resources witnessed globally and more importantly with the economic recession in Nigeria.

Original languageEnglish
Pages (from-to)247-260
Number of pages14
JournalHIV and AIDS Review
Volume18
Issue number4
DOIs
Publication statusPublished - 2019

All Science Journal Classification (ASJC) codes

  • Epidemiology
  • Infectious Diseases

Fingerprint Dive into the research topics of 'Bayesian geo-additive spatial modelling of HIV prevalence using data from population-based surveys'. Together they form a unique fingerprint.

  • Cite this