We read the January JADA article by Dr. Vassiliki Cartsos and colleagues, "Bisphosphonate Use and the Risk of Adverse Jaw Outcomes: A Medical Claims Study of 714,217 People" (JADA 2008; 139[1]:23–30), on the risk of osteonecrosis of the jaw (ONJ) and bisphosphonates with great interest.
We believe the protective effect of oral bisphosphonates and ONJ seen in the study is a spurious finding that is contrary to other research.1 We have recently shown,2 although not specific to jaw necrosis, that these drugs may indeed increase the risk of bone necrosis.
In their study, the authors state that since exposed and unexposed person-time were not available in the database, they took a case-control approach in the data analysis. In fact, an unbiased case-control study requires exposed and unexposed person-time information.3 Thus, the authors’ study is merely a cross-sectional study. Cross-sectional studies cannot show a temporal relationship between the exposure and outcome.
Another limitation of the study are a number of biases that may have influenced the results. The more obvious source of bias may arise from the lack of statistical adjustment for potential confounders, including disease severity and use of drugs such as corticosteroids.
This may be the main reason why an increase in the risk of developing ONJ was seen with intravenous (IV) bisphosphonates among patients with cancer. Since this group may be more prone to developing ONJ, lack of adjustment for the severity of the cancer will make it difficult to know whether ONJ was caused by the disease or IV bisphosphonates—hence, confounding by indication.
In addition, another type of bias, mainly channeling bias, may have occurred as patients with cancer who have the more severe form of the disease are more prone to developing hypercalcemia or bone pain secondary to metastasis. This group is therefore more likely to receive IV bisphosphonates, considered mainstay therapy for both conditions.
In the osteoporosis cohort, there may be several explanations as to why more control subjects (those with no diagnosis of ONJ) received bisphosphonates than the case subjects. First, healthier patients with a higher socioeconomic status may be more likely to receive a bisphosphonate and less likely to be diagnosed with the more severe case of the disease. These patients may be treated in a less aggressive manner by a dentist and subsequently not be registered in the database.
Finally, since no information on timing of the study is given, the study may be subject to a time bias. Case subjects may be less likely to receive oral bisphosphonates simply due to restrictions from their health insurance plans. On the other hand, control subjects may have been more likely to receive a prescription at particular periods when the number of prescriptions for these drugs may have been exceptionally high. For example, physicians may be more likely to prescribe a bisphosphonate because of publication of a landmark trial or due to heavy marketing from the industry.
We believe that the authors should have discussed potential biases and confounding issues that may have influenced their results given the paucity of the information provided in their database.
TOP
REFERENCES
