Secondary multilevel mixed-effects modelling of seroprevalence trends of Crimean–Congo haemorrhagic fever

Abstract

Background: Some review papers and meta-analyses have investigated seroprevalence and fatality trends of the CrimeanCongo hemorrhagic fever (CCHF), but it is not clear if its seroprevalence is increasing. Aim: To investigate the trend in the seroprevalence of CCHF. Methods: We conducted a secondary analysis of the results of a meta-analysis of the seroprevalence of CCHF published in 2019. We used a multilevel mixed effects Poisson regression to find the predictors of seropositivity. To explain the magnitude effect, we reported an incidence rate ratio (IRR) with a 95% confidence interval (CI). We conducted multilevel modeling using Stata 14 for data analysis. Results: In the fixed effects model, time was significantly associated with increased seropositivity (IRR = 1.025, 95% CI = 1.021–1.030), and no significant association was found for local sampling (IRR = 1.026, 95% CI = 0.988–1.065). In the mixed effects model, random intercepts of the country and parallel of latitude were applied as 3 levels of the model (prevalence rate of each study, nested within countries and latitude parallel). Accordingly, time was significantly associated with a reduction of seropositivity (IRR = 0.899, 95% CI = 0.891–0.907), and local sampling was significantly associated with increased seropositivity (IRR = 2.477, 95% CI = 2.316–2.649). Conclusion: Despite reporting increasing trends for seroprevalence of CCHF in previous reviews and the fixed effects model of the present study, the secondary mixed effects modeling showed a decreasing trend. The multilevel generalized model is recommended for such temporal and spatial designs in the future.