Uncertainty of current algorithm for bisphosphonate-related osteonecrosis of the jaw in population-based studies: a...

To assess the relevance of previous epidemiologic studies on bisphosphonate-related osteonecrosis of the jaw (BRONJ), we first conducted a systematic review of large population-based observational studies and evaluated the validity of claims-based algorithms for the identification of BRONJ. Studies containing primary observational epidemiologic data regarding bisphosphonate (BP) exposure and outcomes of osteonecrosis of the jaw were systematically reviewed. Using surrogates for identifying potential BRONJ cases from a population-based hospital registry, validation was performed through medical chart review. Positive predictive value (PPV) was estimated for each diagnostic code and for the overall algorithm utilized. Various strategies to increase PPV were also performed. Seventeen studies were systematically reviewed and presented with variations in study quality as well as inconsistent findings. Moreover, there was a high level of methodological heterogeneity. A total of 1920 patients were identified through the ICD-10 algorithm with potential BRONJ, although only 109 cases were confirmed, corresponding to an overall PPV of 5.68% (95% confidence interval [CI] 4.68-6.81). Only K10.2 (inflammatory conditions of the jaw) exhibited a relatively high PPV of 26.18%, which increased to 74.47% after confinement to BP users. Other strategies to increase PPV value were not effective. Our findings showed that the overall PPV for BRONJ identification was very low, indicating low validity of the current algorithm and possible overestimation of ONJ occurrence. There is an urgent need to develop more reliable and specific operational definitions for the identification of BRONJ cases in large population databases. © 2016 American Society for Bone and Mineral Research.