Statistical analysis

Descriptive statistics were performed and results were presented with frequency tabulations and percentages for categorical variables. Quantitative variables were presented as medians with their interquartile range (IQR). HIV prevalence was estimated with corresponding 95% confidence interval (95%CI) based on estimation of confidence interval for a proportion [27]. Comparisons of medians were performed using Mood’s Median test as it is more resistant to outliers and is more appropriate when distributions in the sub-populations being compared differ (one symmetrical and the other asymmetrical). Proportions were compared using Chi-squared test when sample size of all cells was greater than 5 or Fisher’s exact test when at least one cell was below 5 [28]. Binary logistic regression analyses were performed to identify factors associated with HIV infection among older adults. In the univariable logistic regression, variables with a p-value < 0.20 were fitted into the multivariable analyses. A backward procedure approach was performed for selection of variables and adjusted odds ratio (aOR) were reported with their 95% CI. All analyses were performed using R® software. The significance level was set at 5%.