LATEBREAKER
Methods/Statistics
Improving Global Burden of Disease Study Male Chlamydial and Gonococcal Infection Estimates via Female Data: a Sex-Ratio Exploratory Model Betyna Berice* Betyna Berice Hannah Han Maegan Ashworth Dirac
Background:
Chlamydia and gonorrhea are common sexually transmitted infections (STIs) and major public health issues worldwide. The overall prevalence of chlamydia and gonorrhea in many countries remains uncertain, with many studies conducted in males at high risk, and notable scarcity of data amongst average-risk males. In contrast, female data are typically collected during antenatal visits. Consequently, accurately quantifying the prevalence in males poses a challenge compared to females. We compared two approaches to estimating the prevalence of these diseases in males, one that borrows information for countries without male data from male data from countries in the same region, and one that borrows information from female data in the same country.
Methods:
Point-prevalence data were extracted from studies of chlamydia and gonorrhea reporting for male, female, or both-sex samples. Sex-specific data-points from the same study were paired to calculate log-sex ratios, which were modeled using the Meta-Regression, Bayesian, Regularized, Trimmed (MR-BRT)tool. These ratios were used to split both-sex datapoints into sex-specific datapoints. Sex-specific data-points were then modeled using a Bayesian meta-regression modeling tool, DisMod-MR 2.1 to estimate age, sex, and location specific prevalence of chlamydia and gonorrhea between 1990 and 2021. In approach one, the outputs of DisMod were considered final estimates of prevalence for both males and females. In approach two, the outputs of DisMod for females were combined with modeled sex-ratios to produce estimates for males in the same location, year, and age-group.
Findings:
Twenty-five studies provided male and female-specific data points for chlamydia, and 12 studies provided male and female-specific data points for gonorrhea. To assess performance of our two approaches, we compared the root mean square error (RMSE) of our male DisMod estimates to the male estimates produced by combining female estimates with modeled sex-ratios. RMSE for approaches one and two were similar; 0.035 and 0.040, respectively, for chlamydia, and 0.008 vs. 0.014, respectively, for gonorrhea, meaning that our two approaches performed similarly in fitting to male data.
Interpretation:
Overall, our two approaches to estimating prevalence of chlamydia and gonorrhea in males perform similarly in-sample. Our next step is to assess out-of-sample performance to determine if male estimates for locations with no local data are more reliably produced via DisMod (borrowing strength from males in other countries in the region) or via sex-ratio methods (borrowing strength from females in the same country). In order to enhance chlamydia and gonorrhea estimates and to provide up-to-date policy-relevant information on their respective trends, we need more population-based data for both males and females.