Abstract
BACKGROUND: Several studies have reported osteomyelitis of the jaw (OMJ) as a side effect of bisphosphonates (BPs), and the risk of oral BPs has been recently clarified. However, other systemic risk factors of OMJ remain unclear. Importantly, the possibility of risk classification based on the clinical characteristics of patients has not been explored. Here, we clarified risk factors of OMJ and evaluate the predictive accuracy of risk indices in osteoporosis patients. METHODS: We performed sub-analysis using a database developed for a retrospective cohort study in patients taking medications for osteoporosis at Kyoto University Hospital. Risk indices for OMJ were constructed using logistic regression analysis, and odds ratios (OR) for OMJ cases and 95% confidence intervals (CI) were estimated. Potential risk factors included in the statistical analysis were age; sex; diabetes; use of oral BPs, corticosteroids, cancer chemotherapy, antirheumatic drugs, and biologic agents; and their interactions. Risk indices were calculated by the sum of potential risk factors of an individual patient multiplied by the regression coefficients. The discriminatory power of the risk indices was assessed by receiver operating characteristic (ROC) analysis. RESULTS: In analysis of all patients, oral BPs (OR: 4.98, 95% CIs: 1.94-12.75), age (OR: 1.28, 95% CI: 1.06-1.60) and sex-chemotherapy interaction (OR: 11.70, 95% CI: 1.46-93.64) were significant risk factors of OMJ. Areas under the ROC curves of these risk indices provided moderate sensitivity or specificity regardless of group (0.683 to 0.718). CONCLUSIONS: Our data suggest that oral BP use, age, and sex-chemotherapy are predictors of OMJ in osteoporosis patients. The risk indices are moderately high, and allow the prediction of OMJ incidence.