Concordance between self-report and six commonly used clinical estimates or serological measures: Insights from a...

To assess the concordance between self-reported and clinically assessed prevalence of selected chronic conditions and latent viral infections among women living with and without HIV.

Women (aged ≥ 16 years residing in British Columbia) enrolled in the BCC3 Study, a prospective cohort, between 2020 and 2024 were included in the cross-sectional analysis. Self-reported prevalence of six conditions/viruses (chronic kidney disease, liver disease, depression, post-traumatic stress disorder, and hepatitis B and C viruses (HBV, HCV)), were compared to clinical estimates based on screening tools and serology. Agreement was assessed via Cohen's kappa.

In both women with (n = 220) and without HIV (n = 309), clinical estimate-based prevalence of depression and PTSD was higher than self-reported prevalence (all p < 0.001). Among women with HIV, clinical estimate-based prevalence of HBV was higher than self-report-based prevalence (p < 0.001). For both groups, there was no difference between the two prevalence estimates for chronic kidney disease. Among women without HIV, clinical estimate-based prevalence of liver disease was lower than self-report-based prevalence (p < 0.001), but this was not the case for women with HIV. In both groups, agreement between self-report and clinical estimate of prevalence was fair to poor for all conditions/viruses (all κ < 0.4), except for HCV, for which the agreement was near perfect (κ > 0.8).

Self-reported HCV history shows high concordance with serology, but the same is not true for HBV. The prevalence of liver disease, kidney disease, depression, and post-traumatic stress disorder as reported by participants may differ from clinical estimates. Our findings highlight the complexity of aligning self-report data with clinical estimates and suggest that both types of data should be used for a comprehensive assessment of prevalence in research.