We thank Drs. Etminan and Amidzadeh for their interest in our work. The aim of our study was descriptive to inform the discussion on the frequency of adverse jaw outcomes, including osteonecrosis of the jaw (ONJ), and to provide estimates of the relative frequency of this outcome among users of oral and intravenous bisphosphonates.
In several studies we have conducted to date, involving different data sources and study designs, we obtained results that are quite consistent. Our analysis in 2004 and 2005 of the U.S. Food and Drug Administration’s adverse events database found strong safety signals for adverse jaw outcomes only with intravenous forms of bisphosphonates.1 Similarly, in a recent hospital-based case control study of adjudicated ONJ case subjects and randomly selected control subjects, and after controlling for several confounding variables (chemotherapy, steroid use, comorbidities, smoking, obesity), we also found that only intravenous bisphosphonates, especially zoledronate, were significantly associated with an increased ONJ risk.2
Causal inferences are notoriously difficult to draw from observational data, owing to a range of competing explanations for observed associations that are only partially mitigated through careful attention to study design and analysis. Accordingly, reproduction of research findings by different investigators using different data sources and different methods serves an important role in the acceptance of findings from observational research.
We acknowledge that different analytic approaches could have been applied to our research question, but draw comfort from the fact that our findings are consistent with those of several other researchers who have found that use of intravenous bisphosphonates in oncology is significantly associated with an increased risk of developing ONJ.3,4 Most published epidemiologic studies of the association between oral bisphosphonates and ONJ have found no significant associations; the study by Etminan and colleagues5 provides findings related to this question. In their analysis of claims data, Pazianas and colleagues6 report that, after adjustment for confounding variables, receiving at least one oral bisphosphonate prescription did not significantly increase the risk of jaw surgery (OR = 0.91; 95 percent confidence interval = 0.70–1.19), and that a longer duration of therapy was not associated with increased risk of jaw surgery.
Further, no association between oral bisphosphonate use and ONJ was observed in clinical trials involving patients with periodontal disease or implants,7 or in a clinical trial of once-annually zoledronate in 7,714 postmenopausal osteoporotic women with three years of follow up.8 A reason for the difference between our results on oral bisphosphonates and those of Etminan and colleagues might reside in the outcome and its method of identification. While we used codes for dental outcomes, Etminan and colleagues used hospitalizations and codes for aseptic necrosis in any site.
As a caveat to their finding, they state, "We therefore suspect that the majority of the cases identified in this study are osteonecrosis, possibly of the hip, and not ONJ, especially since most of the patients had been prescribed corticosteroids."5 Indeed, the majority of ONJ cases can be managed on an outpatient basis, whereas aseptic osteonecrosis of the femoral head (OFH) is treated in the hospital.
It is possible that differences in the nature of the outcome studied (ONJ versus OFH) and its identification (dental claims versus hospitalizations) account for the differences. Because OFH occurs often in patients with severe osteoporosis who receive oral bisphosphonates to avoid fractures or OFH, more research is needed to study whether the reported association is causal.
TOP
REFERENCES
