Discussion
In this study we found evidence of an association between child sexual abuse and hypertension among women but noted no associations among other forms of maltreatment and hypertension or among any of the forms of child maltreatment examined and hypertension among men. A positive relation between child maltreatment and hypertension in adulthood has been previously demonstrated. For example, Riley and colleagues, using the Nurses Health Study II data, were able to characterize abuse that occurred prior to age 11 and later experiences of abuse noting that those who experience abuse in both childhood and adolescence, classified as more severe and frequent, had a higher risk of self-reported hypertension. Our results, using actual blood pressure data, validate and extend these prior results using self-reported data.
Sex differences in the association between maltreatment and hypertension have, to our knowledge, only been examined in one other study; an analysis using the National Comorbidity Survey showed that experiencing physical abuse during childhood is associated with self-reported hypertension among adult men but not women, and sexual abuse was associated with self-reported hypertension among women but not men. Prior studies have highlighted sex-specific responses in blood pressure to job strain with negative consequences for men. In contrast we note no relation between physical abuse and hypertension among men or women but note a significant association between sexual abuse and hypertension among women. Differential coping mechanisms or differential HPA axis dysregulation in response to stress between men and women may account for the differences noted.
To our knowledge, all studies examining child maltreatment and hypertension in adulthood, which have noted significant associations, have used self-reported hypertension. In contrast we use measured blood pressure to define hypertension thus reducing the potential for self-report bias. A recent study of the relation between intimate partner violence and cardiovascular endpoints in the Nurses Health Study, noted stronger associations between non-verified cardiovascular outcomes and violence exposure compared to associations that relied on verified cardiovascular outcomes suggesting self-report bias.
The effects of chronic stressors, such as violence, on hypertension can be the result of modified health behaviors known to be affected by chronic stressors; smoking, drinking, and poor dietary habits have been shown to be more prevalent in stressful environments and have also been shown to be associated with cardiovascular disease, including hypertension. Adjusting for obesity, physical activity, smoking and alcohol consumption in our analyses did not change the effect estimates noted though the effect of sexual abuse on hypertension among women was no longer statistically significant. A more direct pathway through which stress can affect health is through its activation of the autonomic nervous system. Animal studies have demonstrated that hypertension following social stress is associated with increased norepinephrine turnover, with other mechanistic studies providing evidence for the role of inflammatory responses and renal mechanisms. Chronic stress has also been shown to induce a chronic and systemic state of mild inflammation (i.e., C-reactive protein and interleukin-6) a key mechanism in the development of cardiovascular disease.
There are a number of limitations worth mentioning. First, as is typical with longitudinal studies, there is a reduction in the sample available from the original Add Health cohort over time; it is likely the loss to follow-up is non-differential with respect to outcome, potentially biasing our results towards the null. However, we use longitudinal sampling weights, which adjust the sample to be representative of sample characteristics at baseline. Second, while the retrospective assessment of maltreatment was collected prior to our outcome assessment it consisted of only three questions that did not account for the severity of the exposures. Thus we were limited to a yes/no characterization of maltreatment exposure. A more detailed assessment that would query about different experiences of neglect, emotional, sexual and physical abuse separately would allow for better classification of severity of exposure. However, the fact that we were able to detect associations between these crude measures of maltreatment and hypertension suggest the dynamic is robust. Third, our assessment of blood pressure was conducted at only one point in time; multiple assessments of blood pressure would provide a more reliable measure of hypertension in this sample. However, we improve upon other studies that relied solely on self-reported hypertension. Finally, as noted in prior studies, the prevalence of hypertension in this national sample of young adults is higher than in other nationally representative samples. If the prevalence of hypertension is associated with inaccurate measurement then this would have likely led to underestimation of the association between maltreatment and hypertension in our analysis.