Past-year intimate partner violence perpetration among people with and without depression: an individual participant data (IPD) meta-mediation analysis

Purpose To investigate whether (1) depression is associated with increased risk of past-year intimate partner violence (IPV) perpetration, disaggregated by sex, after controlling for potential confounders; (2) observed associations are mediated by alcohol misuse or past-year IPV victimisation. Methods Systematic review and individual participant data (IPD) meta-mediation analysis of general population surveys of participants aged 16 years or older, that were conducted in a high-income country setting, and measured mental disorder and IPV perpetration in the last 12 months. Results Four datasets contributed to meta-mediation analyses, with a combined sample of 12,679 participants. Depression was associated with a 7.4% and 4.8% proportion increase of past-year physical IPV perpetration among women and men, respectively. We found no evidence of mediation by alcohol misuse. Among women, past-year IPV victimisation mediated 45% of the total effect of depression on past-year IPV perpetration. Past-year severe IPV victimisation mediated 60% of the total effect of depression on past-year severe IPV perpetration. We could not investigate IPV victimisation as a mediator among men due to perfect prediction. Conclusions Mental health services, criminal justice services, and domestic violence perpetrator programmes should be aware that depression is associated with increased risk of IPV perpetration. Interventions to reduce IPV victimisation might help prevent IPV perpetration by women. Data collection on mental disorder and IPV perpetration should be strengthened in future population-based surveys, with greater consistency of data collection across surveys, as only four studies were able to contribute to the meta-mediation analysis. Supplementary Information The online version contains supplementary material available at 10.1007/s00127-021-02183-w.


Figure 1 Results of fixed effects meta-analysis of the percentage differences between those with and without depression in past year physical IPV perpetration.
Figure 2 Results of random effects meta-analysis of the percentage differences between those with and without depression in past year physical IPV perpetration.

Figure 3 Results of fixed effects meta-analysis of the mediated percentage differences between men with and without depression in past year physical IPV perpetration.
Figure 4 Results of random effects meta-analysis of the mediated percentage differences between men with and without depression in past year physical IPV perpetration.
Direct effect: In fixed effects meta-analysis, 4.6% (95% CI 2.4-6.7) of the total effect of depression on past year physical IPV perpetration among men was unmediated by alcohol.In random effects meta-analysis, 4.9% (95% CI 0.7%-9.0%) of the total effect of depression on past year physical IPV perpetration among men was unmediated by alcohol.

Figure 5 Results of fixed effects meta-analysis of the unmediated percentage differences between men with and without depression in past year physical IPV perpetration.
Figure 6 Results of random effects meta-analysis of the unmediated percentage differences between men with and without depression in past year physical IPV perpetration.

Alcohol: Women
Total effect: In fixed effects meta-analyses, the proportion of difference in physical IPV perpetration between women with and without depression was 7.2% (95% CI 4.1-10.2).In random effects meta-analyses, the proportion of difference in physical IPV perpetration between women with and without depression was 7.7% (95% CI 3.7%-11.7%).
Figure 7 Results of fixed effects meta-analysis of the percentage differences between women with and without depression in past year physical IPV perpetration.
Figure 8 Results of random effects meta-analysis of the percentage differences between women with and without depression in past year physical IPV perpetration.
Figure 9 Results of fixed effects meta-analysis of the mediated percentage differences between women with and without depression in past year physical IPV perpetration.

Figure 10 Results of random effects meta-analysis of the mediated percentage differences between women with and without depression in past year physical IPV perpetration.
Direct effect: In fixed effects meta-analysis, 6.6% (95% CI 3.5-9.7) of the total effect of depression on past year physical IPV perpetration among women was unmediated by alcohol.In random effects meta-analysis, results were identical.

Figure 11 Results of fixed effects meta-analysis of the unmediated percentage differences between women with and without depression in past year physical IPV perpetration.
Figure 12 Results of random effects meta-analysis of the unmediated percentage differences between women with and without depression in past year physical IPV perpetration.

IPV victimisation: women
Total effect: In fixed effects meta-analysis, the proportion of difference in physical IPV perpetration between women with and without depression was 7.4% (95% CI 4.3-10.6).In random effects meta-analysis, the proportion of difference in physical IPV perpetration between women with and without depression was 8.4% (95% CI 3.6-13.1) Figure 13 Results of fixed effects meta-analysis of the percentage differences between women with and without depression in past year physical IPV perpetration.
Figure 14 Results of random effects meta-analysis of the percentage differences between women with and without depression in past year physical IPV perpetration.
Indirect effect: In fixed effects meta-analysis, 3.3% (95% CI 2.0-4.5%) of the total effect of depression on past year physical IPV perpetration among women was mediated by IPV victimisation.In random effects metaanalysis, 4.7% (95% CI 0.9-8.5) of the total effect of depression on past year physical IPV perpetration among women was mediated by IPV victimisation.

Figure 15 Results of fixed effects meta-analysis of the mediated percentage differences between women with and without depression in past year physical IPV perpetration.
Figure 16 Results of random effects meta-analysis of the mediated percentage differences between women with and without depression in past year physical IPV perpetration.
Direct effect: In fixed effects meta-analysis, 3.1% (95% CI 0.3-6.0) of the total effect of depression on past year physical IPV perpetration among women was unmediated by IPV victimisation.In random effects metaanalysis, results were identical.

Figure 17 Results of fixed effects meta-analysis of the unmediated percentage differences between women with and without depression in past year physical IPV perpetration.
Figure 18 Results of random effects meta-analysis of the unmediated percentage differences between women with and without depression in past year physical IPV perpetration.

Severe IPV victimisation: women
Total effect: In fixed effects meta-analysis, the proportion of difference in physical IPV perpetration between women with and without depression was 7.4% (95% CI 4.3%-10.4%).In random effects meta-analysis, the proportion of difference in physical IPV perpetration between women with and without depression was 7.8% (95% CI 3.8-11.8).

Figure 19 Results of fixed effects meta-analysis of the percentage differences between women with and without depression in past year physical IPV perpetration. Figure 20 Results of random effects meta-analysis of the percentage differences between women with and without depression in past year physical IPV perpetration
Indirect effect: In fixed effects meta-analysis, 1.9% (95% CI 0.7-3.2) of the total effect was mediated by severe IPV victimisation.In random effects meta-analysis, the results were identical.

Figure 21 Results of fixed effects meta-analysis of the mediated percentage differences between women with and without depression in past year physical IPV perpetration.
Figure 22 Results of random effects meta-analysis of the mediated percentage differences between women with and without depression in past year physical IPV perpetration.
Direct effect: In fixed effects meta-analysis, 5.4% (95% CI 2.5-8.2) of the total effect of depression on past year physical IPV perpetration among women was unmediated by severe IPV victimisation.In random effects metaanalysis, the results were identical.

Figure 23 Results of fixed effects meta-analysis of the unmediated percentage differences between women with and without depression in past year physical IPV perpetration.
Figure 24 Results of random effects meta-analysis of the unmediated percentage differences between women with and without depression in past year physical IPV perpetration.

Mediation meta-analysis results for past year severe physical IPV perpetration: IPV victimisation: Women
Total effect: In fixed effects meta-analysis, the proportion of difference in severe physical IPV perpetration between women with and without depression was 4.0% (95% CI 1.5-6.6).In random effects meta-analysis, the proportion of difference in severe physical IPV perpetration between women with and without depression was 5.4% (95% CI 0.1%-10.6%).

Figure 25 Results of fixed effects meta-analysis of the percentage differences between women with and without depression in past year severe physical IPV perpetration.
Figure 26 Results of random effects meta-analysis of the percentage differences between women with and without depression in past year severe physical IPV perpetration.
Indirect effect: In fixed effects meta-analysis, 2.1% (95% CI 1.1-3.1) of the total effect of depression on past year severe physical IPV perpetration among women was mediated by IPV victimisation.In random effects meta-analysis, the total effect of depression on past year severe physical IPV perpetration among women was not mediated by IPV victimisation (2.9%, 95% CI -0.3%-6.1%).
Figure 27 Results of fixed effects meta-analysis of the mediated percentage differences between women with and without depression in past year physical IPV perpetration.
Figure 28 Results of random effects meta-analysis of the mediated percentage differences between women with and without depression in past year physical IPV perpetration.
Direct effect: In fixed effects meta-analysis, there was no direct effect of depression on severe IPV perpetration (1.3%, 95% CI -1.03-3.5).In random effects meta-analysis, the results were identical.
Figure 29 Results of fixed effects meta-analysis of the non-mediated percentage differences between women with and without depression in past year severe physical IPV perpetration.
Figure 30 Results of random effects meta-analysis of the non-mediated percentage differences between women with and without depression in past year severe physical IPV perpetration.

Severe IPV victimisation: women
Total effect: In fixed-effects meta-analysis, the proportion of difference in severe physical IPV perpetration between women with and without depression was 2.0% (95% CI 0.6%-3.3%).In random-effects meta-analysis, the proportion of difference in severe physical IPV perpetration between women with and without depression was 3.3% (95% CI 0.2%-6.3%).
Figure 31 Results of fixed effects meta-analysis of the percentage differences between women with and without depression in past year severe physical IPV perpetration.
Figure 32 Results of random effects meta-analysis of the percentage differences between women with and without depression in past year severe physical IPV perpetration.
Indirect effect: In fixed effects meta-analysis, 1.2% (95% CI 0.6-1.8) of the total effect of depression on past year severe physical IPV perpetration among women was mediated by severe IPV victimisation.In random effects meta-analysis, the results were identical.
Figure 33 Results of fixed effects meta-analysis of the mediated percentage differences between women with and without depression in past year severe physical IPV perpetration.
Figure 34 Results of random effects meta-analysis of the mediated percentage differences between women with and without depression in past year severe physical IPV perpetration.
Figure 36 Results of random effects meta-analysis of the unmediated percentage differences between women with and without depression in past year severe physical IPV perpetration.