Background: Achieving viral suppression is associated with better health outcomes for people living with HIV. Studies in some countries have revealed socioeconomic disparities in viral suppression. Due to lack of valid measures of socioeconomic status, most studies construct wealth indices based on household asset variables often collected in population-based surveys. There are, however, multiple techniques for constructing a wealth index and there is lack of consensus regarding which technique is superior. This study examined the socioeconomic gradient in viral suppression in Malawi, checking robustness of results to choice of method for constructing the wealth index.
Methods: The Malawi Population-based HIV Impact Assessment (MPHIA) was a nationally representative survey conducted in Malawi in 2015-2016. Among 17,187 adults aged 15-64 who were interviewed and tested for HIV using the national algorithm, 2227 (13%) tested HIV positive, and 2220 (99.7%) of these had a viral load test performed. The Erreygers Index (EI) for income-related health inequality was computed for viral suppression under three scenarios based on statistical procedures for deriving the wealth index: Standard Principal Components Analysis (PCA), Multiple Correspondence Analysis (MCA) and Uncentered Principal Components Analysis (UPCA). All analyses were adjusted for the survey''s complex design using jackknife replicated weights. These weights were also applied to calculation of viral suppression rates within the quintiles.
Results: Across the PCA and MCA scenarios, viral suppression rates appear to slightly increase monotonically from lowest to the highest wealth quintile while in the UPCA scenario, viral suppression rates vary only marginally and non-monotonically across the quintiles. On the other hand, the EI for viral suppression using the standard PCA wealth index was statistically insignificant (EI value= 0.0120; p-value= 0.2860). The EI for the MCA and UPCA indices were also not statistically significant (EI values= .00743 and -0.0204; p-values of 0.5128 and 0.0860 respectively).
Conclusions: Based on the inequality index used in this study, there were no substantial differences with respect to the socioeconomic gradient in viral suppression across the three scenarios, which implies there are no socioeconomic disparities in viral suppression among PLHIV in Malawi. This result is robust to choice of method for constructing the asset index.