Association between alcohol use and femicide in South Africa during the COVID-19 pandemic: a cross-sectional study.

BACKGROUND

Alcohol use has previously been associated with femicide, the most extreme form of gender-based violence, but research on femicide from low-income and middle-income countries is scarce. We aimed to examine the role of alcohol restrictions during South Africa's COVID-19 lockdowns on femicide rates in the country.

METHODS

In this cross-sectional study, we compared estimates of overall femicide, intimate partner femicide, and non-intimate partner femicide from two retrospective national surveys of female individuals aged 14 years and older killed in 2017 and during nine COVID-19 lockdown periods in 2020-21. Both surveys used multistage, stratified cluster sampling with mortuaries or medico-legal laboratories as the primary sampling unit. Sampling frames that listed all mortuaries operating in the country in each study year were developed and mortuaries were stratified by size on the basis of the number of autopsies they performed per year. Mortuary data were supplemented with investigative data extracted from police dockets during interviews with police members. Cases in which age and sex were unknown and in which the cause of death was undetermined were excluded from data collection. Linearisation was used to obtain robust variance estimators for parameter estimates. Female population estimates were derived from the Thembisa mathematical model to calculate age-standardised rates, and the WHO world standard population distribution was used to normalise weightings for age-standardised rate calculations. We used Poisson regression and incidence rate ratios (IRRs) with 95% CIs to compare femicide estimates for periods of complete and partial alcohol sales and curfews with periods of no restrictions and to compare data from corresponding calendar periods in 2017 and 2020-21.

FINDINGS

We found no evidence of a significant change in age-standardised rates of overall femicide between 2017 and 2020-21 (IRR 0·95, 95% CI 0·88-1·03). Our Poisson regression results for 2020-21 showed no evidence of a significant difference in the average number of overall femicide cases between the first two lockdown periods, both periods with no alcohol sales but differing curfew levels (1·16, 0·88-1·54), but we found a significant increase in the average number of cases between periods 2 and 3, when alcohol sale restrictions were relaxed (2·14, 1·70-2·69). When combining periods with similar alcohol and curfew restrictions, we found a significant decrease in overall femicide cases between periods of no restriction and periods of restricted alcohol sales and night and late-night curfews (0·57, 0·49-0·66). Similarly, the average number of femicide cases during the period with a complete ban on both movement and alcohol sales was 63% lower than during periods of no restriction (0·37, 0·30-0·47). A similar significant difference was observed for both cases of intimate partner femicide (0·39, 0·28-0·53) and non-intimate partner femicide (0·39, 0·27-0·55). This pattern was not seen during the corresponding calendar periods in 2017.

INTERPRETATION

The COVID-19 lockdown presented an opportunity to study the association between alcohol use and femicide in South Africa. We found a decrease in overall femicide, intimate partner femicide, and non-intimate partner femicide during periods of complete alcohol prohibition compared with periods of partial or no alcohol sale bans. This analysis supports alcohol use as a risk factor for severe gender-based violence and emphasises the importance of implementing evidence-based alcohol harm reduction interventions and policies as part of gender-based violence prevention strategies.

FUNDING

Ford Foundation, Centers for Disease Control and Prevention Foundation, and South African Medical Research Council.