Cesarean delivery variation across facilities in Uganda

Abstract

Objective: To describe caesarean delivery (CD) rates and their variability across health facilities and regions in Uganda. Study design: Using the 2018-2019 Ugandan Annual Health Sector Report, we extracted data on facility type, location, births, and delivery mode. We restricted analysis to facilities with birth and mode data. We calculated CD rates for each facility and reported the median, interquartile range and 95% confidence interval by facility type (Health Centre IV (HCIV), General Hospital (GH), Regional Referral Hospital (RRH) and Private Not for Profit (PNFP)) and geographic region (Central, Northern, Western and Eastern). We categorized facilities into those with CD>15%, 15-30% and >30% and compared these proportions by facilities type and region using the Kruskal Wallis test. Results: Data was available for 286 facilities representing 509,206 births and 71% of facilities performing CD in Uganda. The overall median CD rate was 17.9% ranging from 0.08% to 78% (95% CI 14.6-21.8). Most births occurred in GH (39.8%, n¼202,895) and HCIV (38.8%, n¼197,620). Most CDs were performed in GH (50.4%, n¼59,871) and RRH (24.5%, n¼28,931). HCIVs and PNFP accounted for only 18.5% (n¼22,005) and 6.7%( n¼7913) of CDs respectively. There was significant variation in CD rates by facility type with RRHs more likely to have CD rates >30% compared to GH and HCIVs (p<0.01; Figure 1).CD rates also varied significantly by region (p¼0.0031; Figure 2) and were highest in the Central region (median CD rate¼24% IQR 37-11.5 95% CI 19.9 -31.5) and lowest in the Eastern region (median CD rate¼12% IQR 23.2-4.3 95% CI 8.9-14). Conclusion: CD rates vary widely across facilities in Uganda with higher rates seen in RRHs and GHs despite a large proportion of births occurring at HCIVs. This unequal distribution suggests overburdening of CD at RRHs and underutilization at HCIVs. Similarly, regional CD rates vary widely. Future research with individual patient-level data to allow for risk-adjustment is needed to further explore drivers and trends of CD variation and implications for patient outcomes.