Long-Term Proton Pump Inhibitor Therapy and Risk of Hip Fracture

Abstract & Commentary

By Malcolm Robinson, MD, FACP, FACG, Emeritus Clinical Professor of Medicine; University of Oklahoma College of Medicine, Oklahoma City. Robinson serves as a consultant for TAP, Pfizer, Janssen, Eisai, J&J-Merck, and Procter & Gamble, is on the speaker's bureau of Janssen, Eli Lilly, Solvay, TAP, and Aventis, and does research for Forest Labs, Wyeth-Ayerst, AstraZeneca, and Centocor.

Synopsis: Evaluation of a large United Kingdom (UK) database supports a dose-related association with hip fracture. There appears to be a correlation between PPI use, including dosage and chronicity, and occurrence of hip fractures in patients over age 50.

Source: Yu-Xiao Yang, et al. JAMA. 2006;296:2947-2953

More than 47,000 hip fractures occur annually in the UK, usually related to osteoporosis. Known factors predisposing to osteoporosis include low calcium consumption and calcium malabsorption. Calcium absorption decreases with age and urinary calcium excretion increases. One-year mortality following hip fracture is 20%, and a similar percentage of hip fracture patients require nursing home care. Health care costs associated with hip fracture are immense. Proton pump inhibitors (PPIs) are very widely utilized, often chronically.

The authors speculate that elderly patients may already have hypochlorhydria, possibly exacerbated by H. pylori infection. Elderly patients also might exhibit decreased PPI clearance. Low gastric acidity might impair calcium absorption, a phenomenon seen in a few animal and human studies. Conversely, PPIs also may decrease bone resorption (thereby potentially averting osteoporosis) by inhibiting osteoclastic vacuolar hydrogen-potassium-ATPase. Data from the UK General Practice Research Database were utilized in this study. There are 9.4 million patients in the database. Exclusions for this study included less than 365 days of follow-up (2.3 million) and age less than 50 (6.9 million). Other exclusions were very short term PPI use and prevalent hip fractures. A total of 1.8 million patients remained in the final study group. Of these, 192,028 patients received at least 1 PPI prescription, 187,686 patients received H2RA prescriptions but no PPIs, and 1.4 million patients received neither form of acid suppressive therapy. Hip fracture cases were defined as occurring at least 1 year after beginning acceptable follow-up. Ten controls were selected for each case, matched for age, index date, year of birth, and duration of follow-up. Periods of PPI or H2RA use were cumulated along with an estimate of "high dosing" (defined as > 1.75 x the standard dose/day). GERD was the most prominent diagnosis within this selected patient group. A variety of pertinent co-morbidities were tabulated including congestive heart failure, CVA's, celiac sprue, and inflammatory bowel disease (and many more). Multiple additional drug exposures were considered including corticosteroids, thyroxine, thiazide diuretics, and bisphosphonates (among other exposures such as smoking).

As would be expected, most of the drugs and co-morbidities expected to have an association with hip fractures were indeed more commonly present in the hip fracture patient group vs. controls. PPIs were found to be associated with an increased risk of hip fracture with statistically significant adjusted odd ratios of 1.2 for one year ranging up to about 1.6 for 4 years of PPI therapy. Higher doses (> 1.75 times the normal dose per day) of either PPIs or H2RAs seemed to increase the relative risk of hip fracture. The authors ultimately speculated that calcium malabsorption could be the mechanism for this effect. In a previously reported but shorter duration Danish observational study, similar increases in hip fractures associated with PPIs were observed although no dose effect or effect of H2RAs were demonstrated in Denmark. The authors admit that their study may not have captured all possible co-morbidities, and they had no information regarding the utilization of over-the-counter calcium by either the patients or the control population. Despite these admitted deficiencies, the authors recommend that PPI users should increase calcium intake in the form of dairy products or co-ingestion of insoluble calcium supplements with presumably relatively acidic meals.


PPIs have had extensive pre-clinical testing including evaluation of their effects on nutrient absorption (including calcium). The degree of acid inhibition attained with oral PPIs is actually quite modest. Indeed, even when dosing PPIs multiple times daily, gastric pH cannot be held above 6.0. With smaller PPI doses, acid will be present in the stomach during large segments of the 24-hour day in most individuals. As the authors point out in this study, the hip fracture patients are suffering from many other medical conditions and taking many other medications that are associated with hip fracture. Although they have attempted to adjust for all of these variables, it seems likely that many additional unmonitored medications and conditions also varied between groups. This article has already raised great concern among patients, and it is likely that many patients have independently discontinued their use of PPIs based on these speculative results. In the mind of this reviewer, this article and its data do not substantiate the modification of PPI use for otherwise appropriate indications. Although adequate calcium and vitamin D intake are known to be important in prophylaxis against osteoporosis and fractures, there are insufficient data to recommend any new additional dietary maneuvers in our patients who are taking acid suppressive medication. Patients should be advised to continue needed anti-secretory medications despite potential alarm raised by publications like this one. Statistical associations are often interesting, but we must always remember that association does not prove causality.