Quantifying the gender gap in the HIV care cascade in southern Mozambique: We are missing the men

Abstract

Background: HIV-infected men have higher rates of delayed diagnosis, reduced antiretroviral treatment (ART) retention and mortality than women. We aimed to assess, by gender, the first two UNAIDS 90 targets in rural southern Mozambique.

Methods: This analysis was embedded in a larger prospective cohort enrolling individuals with new HIV diagnosis between May 2014-June 2015 from clinic and home-based testing (HBT). We assessed gender differences between steps of the HIV-cascade. Adjusted HIV-community prevalence was estimated using multiple imputation (MI).

Results: Among 11,773 adults randomized in HBT (7084 female and 4689 male), the response rate before HIV testing was 48.7% among eligible men and 62.0% among women (p<0.001). MI did not significantly modify all-age HIV-prevalence for men but did decrease prevalence estimates in women from 36.4%to 33.0%. Estimated proportion of HIV-infected individuals aware of their status was 75.9% for men and 88.9% for women. In individuals <25 years, we observed up to 22.2% disparity in awareness of serostatus between genders. Among individuals eligible for ART, similar proportions of men and women initiated treatment (81.2% and 85.9%, respectively). Fourfold more men than womenwere in WHO stage III/IV AIDS at first clinical visit. Once on ART, men had a twofold higher 18-month loss to follow-up rate than women.

Conclusion: The contribution of missing HIV-serostatus data differentially impacted indicators of HIV prevalence and of achievement of UNAIDS targets by age and gender and men were missing long before the second 90. Increased efforts to characterize missing men and their needs will and their needs will allow us to urgently address the barriers to men accessing care and ensure men are not left behind in the UNAIDS 90-90-90 targets achievement.

Fig 1. Study profile for HIV home-based testing in rural southern Mozambique.

Percentage are calculated over the previous step. The shaded boxes refer to those participants included in each step and the arrows to the left give the reasons for non-inclusion. The number of female and male participants at each step are shown in brackets.

Fig 2. Individuals reached for HIV testing during a home-based testing campaign in rural southern Mozambique.

Proportions were calculated as the number of people achieving each step divided by the numerator of the previous step.

Fig 3

Age-specific HIV prevalence according to sex (A, men; B, women; C, total) and population distribution in rural southern Mozambique. HIV prevalence is shown unadjusted and excluding missing values (white bars) and after adjustment for missing values by multiple imputation (MI) and by inverse probability weighting (IPW). Fold-change in prevalence between unadjusted and adjusted values was calculated by dividing the adjusted prevalence estimate by the unadjusted prevalence estimate.

Fig 4. Gap in knowledge of HIV positive serostatus between men and women adjusted by multiple imputation method.

(A) Age-specific and sex-specific proportion of HIV-infected individuals who were aware of their serostatus at the time of the survey; (B) the proportional difference in serostatus awareness between the two sexes. The graph shows the point estimates plus 95% confidence intervals (CI) for the percent difference between female and male awareness of their HIV status. The confidence interval for the proportion difference was calculated from the two-sample unpaired z test for each comparison.