Clinical indications of premenstrual disorders and subsequent risk of injury: a population-based cohort study in Sweden

Premenstrual disorders, including premenstrual syndrome and premenstrual dysphoric disorder, are suggested to be correlated with suicidal behavior and accidents in cross-sectional and retrospective studies. However, prospective data are still lacking.

Methods
We performed a population-based cohort study including 1,472,379 Swedish women of reproductive age who were followed from 2001 to 2012. Within the cohort, we also performed a sibling analysis where we compared the rates of injury between full sisters. By linking to the Patient and the Prescribed Drug Registers, we identified 18,628 women with any clinical indications for premenstrual disorders in the cohort (population analysis) and 7674 women in the sibling analysis. Any injury, primarily suicidal behavior (completed suicide and suicide attempt) or accidents (e.g., fall and transportation accidents), was identified through the Patient and Causes of Death Registers as the primary outcome. We estimated hazard ratios (HRs) and 95% confidence intervals (CIs) of these outcomes among women with premenstrual disorders in both population and sibling analyses using multivariable Cox proportional hazards regression.

Results
During a maximal follow-up of 12 years (mean 9.55 years), we identified 2390 women with premenstrual disorders with any injury; 216 through suicidal behavior and 2191 through accidents. Compared to women without premenstrual disorders, women with premenstrual disorders were at increased risk of any injury (HR 1.37, 95% CI 1.31–1.42), particularly suicidal behavior (HR 2.26, 95% CI 1.97–2.59) and accidents (HR 1.32, 95% CI 1.27–1.38). Such associations somewhat attenuated yet remained significant in the sibling analysis (HRs: 1.31 for any injury, 1.86 for suicidal behavior, and 1.29 for accidents). Additional adjustment for psychiatric comorbidities minimally altered the associations with any injury and accidents in both population and sibling analyses, whereas the association with suicidal behavior was considerably attenuated to non-significance in the sibling analysis. Such risks were particularly strong within 2 years after receiving the diagnosis of premenstrual disorders and were evident among women with premenstrual disorders wit