Background A recent systematic review has highlighted potential links between depression and fracture risk, but no information regarding possible explanatory health behaviours was available. In this study, we used a well phenotyped cohort of community-dwelling older adults (the Hertfordshire Cohort Study) to examine depression and anxiety in relation to incident fracture risk, and in relation to health behaviours.

Methods Data from 2997 men and women, aged 59-73 years at baseline, were analysed. At baseline, mental health characteristics and health behaviours were ascertained through clinician-administered questionnaires. Anxiety and depression (mild or worse) were identified by scores of >7 on the corresponding Hospital Anxiety and Depression Scale. Health behaviours included current smoking, alcohol consumption, diet quality and physical activity (Dallosso questionnaire). Incident fractures were identified using ICD-10 codes from Hospital Episode Statistics data, available from baseline (1998-2004) until December 2018. Health behaviours in relation to anxiety and depression were examined using logistic regression; anxiety and depression in relation to incident fracture were examined using Cox regression. All analyses were sex-stratified and adjusted for age.

Results Prevalence of anxiety was 15% among men and 25% among women; prevalence of depression was 5% among men and 6% among women. Overall, 9% of men and 22% of women had a fracture during follow-up. Among men and women, poorer diet quality and lower physical activity were associated with depression; these health behaviours were also associated with anxiety among men. Among women, current smoking was associated with anxiety and depression. Anxiety (hazard ratio (95% CI): 1.34 (1.05, 1.72), p=0.019) and depression (1.76 (1.18, 2.64), p=0.006) were related to increased risk of incident fracture in women, while in men, associations for anxiety (p=0.198) and depression (p=0.265) in relation to fracture were weaker.

Conclusion Anxiety and depression were associated with increased risk of incident fracture among women, possibly due to their association with poor health behaviours, which are established risk factors for fracture. Limitations of the study are that participants were all community-dwelling and Caucasian, so these findings may not be generalizable to participants in other settings. These findings suggest that preventive strategies are required to address poor health behaviours and mental health among older adults.