Intergenerational effects of witnessing domestic violence: Health of the witnesses and their children

Abstract

Studies that explore intergenerational effects of witnessing domestic violence during childhood ("witnessing") are lacking. We examined effects of witnessing on general health status for adults who witnessed domestic violence during childhood and their children. Cross-sectional data from population-based phone interviews conducted in Philadelphia during 2012-2013 provided health information for 329 parents and children, and parent's witnessing exposure. We used propensity scores to predict parent's witnessing status using childhood confounders; response models included inverse probability of treatment weighting and population weights for standardization. Separate standardized multivariate logistic regression models provided average treatment effects and 95% CIs for associations between childhood witnessing and below average health for: 1) adults who witnessed and 2) their children. Sensitivity analyses guided interpretation. Standardized models showed no differences in average treatment effects for below average adult health for witnesses vs. non-witnesses [0.04 (-0.12, 0.19)]. Conversely, children whose parents witnessed had considerably higher probability of having below average health than children whose parents did not witness [0.15 (0.02, 0.28)]. An unmeasured confounder would need 3.0-fold associations with both exposure and outcome to completely remove observed effects, indicating a moderate relationship. However, the lower confidence bound could cross 1.0 in the presence of a weaker unmeasured confounder having 1.2-fold associations with both exposure and outcome, while controlling for our same measured confounders. Witnessing during childhood did not affect adult health in our population, but we found moderate evidence supporting harmful intergenerational effects of witnessing on health, with parent's witnessing exposure affecting their child's health.

Keywords: ACE, adverse childhood experience; Adverse childhood experiences; CI, confidence interval; Child health; Domestic violence; Exposure to violence; Family health; Family violence; Health status; Intergenerational relations; Intimate partner violence; OR, odds ratio; RCT, randomized controlled trial; RR, risk ratio; Social determinants of health.

Fig. 1

Sample of Philadelphia adult respondents who completed interviews for themselves and a child under the age of 18 years of age who was living in the home at the time of interview.

Fig. 2

Estimated propensity scores for the witness and non-witness groups before and after balancing with the inverse probability of treatment weight (IPTW).

Fig. 3

Standardized percentage of bias associated with each covariate in the propensity score model before and after weighting with the inverse probability of treatment weight.