Familial confounding is confounding due to genetics or environmental exposures shared by family members. We aimed to study whether familial confounding explains the association between body mass index (BMI) and severe hip osteoarthritis (OA).Methods:
We linked data from the Norwegian Arthroplasty Registry with the Norwegian Twin Registry on the National ID-number in 2014, generating a population-based prospective cohort study of same-sex twins born between 1915 and 1960 (53.4% females). BMI was calculated from self-reported height/weight. The outcome was incident hip arthroplasty due to OA (follow-up time, 1987–2014; 424 914 person-years). We performed sex-specific co-twin control analyses of dizygotic (N = 5,226) and monozygotic (MZ, N = 3,803) twin pairs using Cox regression models and explored reasons for any familial confounding using bivariate twin models.Results:
The mean (SD) BMI was 22.6 (2.96), peak lifetime BMI 25.6 (2.61), and N = 614 had hip surgery due to OA. In cohort analyses, BMI was associated with hip OA for women and men (hazard ratio [HR] = 1.09, 95% confidence intervals [CIs] = 1.06 to 1.11 and HR = 1.08, 95% CI = 1.04 to 1.12, respectively). When adjusting for familial confounding within MZ twins, the association got stronger for women (HR = 1.19; 95% CI = 1.05 to 1.36) but weaker for men (HR = 0.93; 95% CI = 0.75 to 1.16). There was no genetic overlap between BMI and hip OA, yet the familial confounding in men provides suggestive evidence that the association could be explained by shared environmental factors.Conclusion:
The association between BMI and hip OA may be explained by familial confounding for men. For women, there was no evidence for familial confounding, consistent with a causal association. See video abstract at, http://links.lww.com/EDE/B336.