Healthcare under fire: quantifying the impact of violence on medical services in facilities supported by International Medical Corps in three prefectures of Central African Republic, 2016-2020.

Background: Attacks on healthcare in the Central African Republic (CAR) are widespread, under-reported, and inadequately addressed. This study examined the impact of attacks on healthcare in three conflict-affected prefectures-Ouaka, Haute-Kotto, and Vakaga-between 2016 and 2020, assessing both immediate and longer-term disruptions.

Methods: Disruptions at primary and referral facilities were evaluated using program data from International Medical Corps (IMC) and attack data from IMC, key informants, secondary datasets, and local media. Key indicators-outpatient consultations, first antenatal care visits (ANC1), facility deliveries, measles vaccination (MCV1), and hospitalizations-were analyzed using visual trend analysis, estimation of immediate change, and interrupted time series (ITS). Survival analysis assessed the association between attacks and time to the first reported measles case during a nationwide outbreak.

Results: A total of 127 individual attacks were identified over five years, primarily from IMC security reports. The most common incidents were asset removal or attempted removal (27%), threat (18%), and pillage (17%). At least one form of physical or sexual violence was documented in 23.2% of attacks, with 13 instances of murder. Visual analysis showed three impact patterns: facility closure, disruption of specific services, and minimal or short-term changes. Immediate changes also varied, with service changes ranging from - 100 to 655%. In ITS analysis four of eight facilities showed > 25% deficit in outpatient consultations while three showed > 25% surplus. Survival analysis demonstrated significant difference in time to first measles case between attacked and facilities without attacks (p < 0.001), though findings are limited by small sample size. Overall, maternal health services had fewer fluctuations while vaccination services ceased altogether in some facilities.

Conclusions: This study shows improved identification and profiling of attacks is possible in low-resource settings and presents several approaches to quantify their impact. It also highlights challenges conducting analysis with significant limitations in data quality and availability. Findings reinforce the urgent need for systematic data collection, real-time monitoring, tailored, context-specific mitigation strategies, and support to local actors who maintain services when external support is limited. Future research should build on these findings to provide more effective protection, mitigation, and recovery strategies for healthcare systems in conflict zones.