Proton pump inhibitors and osteoporosis-related fractures

Two of the authors respond:

We thank Mathieu Forster for his interest in our article.1 He expressed concern about the possible influence of unmeasured variables, specifically smoking status, alcohol use and body mass index, on the association that we observed between use of proton pump inhibitors and the development of osteoporosis-related fractures. For an unmeasured variable to influence the results of the study, it would have to be associated both with the outcome of interest in the unexposed cohort and with the exposure of interest. It has previously been established that smoking, excessive alcohol use and low body mass index are associated with higher risk for osteoporosis and osteoporosis-related fractures,2^–4 but for these factors to influence (positively or negatively) the relation between proton pump inhibitor use and osteoporosis-related fractures, they would also have to be independently associated with the use of proton pump inhibitors.

There are currently very little data available to confirm a relation between proton pump inhibitor use and any of the above factors. A recently published analysis of long-term use of proton pump inhibitors suggests a lack of a relation between use of proton pump inhibitors and both smoking and body mass index.5 However, smoking and alcohol use are both positively associated with gastresophageal reflux disease, the most common indication for proton pump inhibitor therapy. Conversely, gastresophageal reflux disease is also strongly associated with obesity,6 which is in fact protective against osteoporosis and osteoporosis-related fractures. Given the likelihood that these influences have opposing effects, it is very likely that the overall effect of these unmeasured variables on the detected association is minimal. We agree that further studies are required to determine whether the association between use of proton pump inhibitors and the development of osteoporosis and osteoporosis-related fractures is truly causal.