Healthcare responding to violence and abuse in Brazil: a quasi-experimental difference-in-differences analysis.

Background: Domestic violence against women (DVAW) is a public health issue and a breach of human rights, yet evidence on effective interventions remains limited, particularly in low-income and middle-income countries. This study aimed to evaluate changes in identification and referral to specialist support associated with system-level strategies implemented within Brazilian primary healthcare (PHC) to strengthen the response to DVAW. The strategies comprised an intervention called Healthcare Responding to Violence and Abuse (HERA). Methods: Using a quasi-experimental design, HERA was implemented in eight PHC clinics, while 33 served as controls. Data on DVAW identification and referral were obtained from the national Epidemiological Surveillance System. Difference-in-differences analysis, using negative binomial regression, assessed HERA's impact, controlling for patient inflow, clinical supervision, COVID-19 lockdown, region, and clinic. Results are reported as marginal effects with 95% confidence intervals (CI). Findings: There was an increase in the probability of DVAW identification (0.47; 95% CI 0.18-0.77) and referral to support services (0.38; 95% CI 0.03-0.73), when adjusting for panel effects and time. The results were even larger when further controlling for additional variables (0.82 for identification [95% CI 0.44-1.21] and 0.87 for referrals [95% CI 0.47-1.29]). Interpretation: HERA strategies increased DVAW identification and referral in PHC settings. Clinics implementing HERA were already more likely to identify and refer cases before the implementation, suggesting that HERA's strategies may be more effective in clinics that find DVAW interventions more acceptable, at least in Brazil. Funding: NIHR Global Health Research Group Award.