Seasonality in hospital admissions of Crimean-Congo hemorrhagic fever and its dependence on ambient temperature—empirical evidence from Pakistan

Crimean-Congo hemorrhagic fever (CCHF) has been reported from all provinces of Pakistan. Little is known about the seasonal variations in the disease and its association with weather conditions. In this study, we explored time-series data about monthly number of CCHF admissions (2007–2010) in three public sector hospitals of Quetta—the capital city of Baluchistan province of Pakistan. Cosinor analysis was carried out to investigate seasonality in the data. To assess the effect of average monthly ambient temperature (°C) on disease, a distributed lag nonlinear model (DLNM) was applied. Cosinor model revealed statistically significant seasonality in monthly number of CCHF patients admitted to the study hospitals. The estimated amplitude was 3.24 cases per month with phase in mid-June and low point in mid-December. DLNM confirmed nonlinear and delayed effect of temperature on hospital admissions. At a lag of 2 months, the cumulative relative risk was more than 1 at temperature at 18.37 °C and above. In addition, relative risk was significantly high at 60th (21.98 °C), 70th (24.50 °C), 80th (27.33 °C), and 90th (29.25 °C) percentiles of temperature (relative to median value, 18.37 °C). Inclusion of Eid-al-Adha as a predictor did not improve the fitness of DLNM. Based on our analysis, we concluded significant seasonality in CCHF hospital admissions. Our findings also suggested average monthly ambient temperature (°C) as a significant predictor of CCHF hospitalizations. DLNM presented in this study may be improved with inclusion of other possible time-varying predictors particularly meteorological conditions of this region.