brand logo

Am Fam Physician. 2008;77(3):277-278

Author disclosure: Nothing to disclose.


Each month, three presenters will review an interesting journal article in a conversational manner. These articles will involve “hot topics” that affect family physicians or will “bust” commonly held medical myths. The presenters will give their opinions about the clinical value of the studies discussed. The opinions reflect the views of the presenters, not those of AFP or the AAFP.

This Month's Article

Yang YX, Lewis JD, Epstein S, Metz DC. Long-term proton pump inhibitor therapy and risk of hip fracture. JAMA. 2006;296(24):2947–2953.

Does the long-term use of PPIs (longer than one year) increase the risk of hip fracture?

Mark: Researchers have noted an association between PPI use and increased rates of community-acquired pneumonia1 and Clostridium difficile colitis.2 Because PPIs may reduce calcium absorption, are patients who take PPIs also predisposed to hip fractures?

What does this article say?

Mark: This is a case-control study of all patients in the UK General Practice Research Database who were 50 years or older and who had a hip fracture. The authors compared these 13,556 patients with a control cohort of 135,386 patients without a hip fracture. They then looked at whether patients who were on a PPI for longer than one year had a greater risk of hip fracture. They also tried to control for comorbities and confounding factors, such as Parkinson's disease, steroid use, and thyroxine use.

The rate of hip fractures was four per 1,000 patient-years in patients on PPIs and 1.8 per 1,000 patient-years in those not on PPIs (odds ratio for having a hip fracture was 1.44 [95% CI, 1.30 to 1.59]). But the number needed to harm was 454, meaning that 454 patients have to be treated for one year to cause one hip fracture. Unfortunately, the authors did not give the fracture rate adjusted for comorbidities.

Should we believe this study?

Mark: The data are sound, but there are a couple of problems. First, it is possible that despite all of the effort put into controlling for variables, something other than the PPIs was responsible for the increase in fractures. Maybe those on PPIs were sicker, and thus, fell more often. That was, in fact, the case in this study. The PPI group had more cardiovascular disease and greater use of steroids, tobacco, antipsychotics, and antiparkinsonian drugs. Although the authors tried to control for these factors, a case-control study can only give us an association, not causality.

Andrea: A good example of case-control studies being misleading is the case-control studies that suggested that postmenopausal estrogen was cardioprotective.3,4 This association was disproved in subsequent randomized controlled trials.5,6

Bob: I agree. Case-control studies like this one are problematic because they begin after the patients have developed a disease or complication (in this case, hip fracture). Then the researchers try to go back in time to determine if a medication, intervention, or certain characteristic (in this case, taking a PPI) is more prevalent in those with the disease compared with those without. As such, they are subject to many confounders, as Mark pointed out. Associations can sometimes be determined and can be used as a hypothesis for future studies. But until then, an association does not equate to cause and effect.

Mark: Secondly, and maybe more importantly, the mean age at enrollment was 77. So even though this is a database of patients 50 years or older, the study is really of those with an average age of 77. This is an example of selection bias. While we might apply these results to older patients, we cannot generalize the results to a younger population, even though the database contained data for 50-year-olds.

Bob: This is an important point. When reading an article, make sure the study group is representative of the patient you are going to treat.

Additionally, these results have to be looked at in the appropriate context. For example, while NSAIDs should be avoided in an older population, if you are faced with an older patient who requires an NSAID, I would not hesitate to add a PPI. In this clinical context, PPIs have been shown to decrease the rate of gastrointestinal toxicity.7

Andrea: Also, a lot of PPI use is “as needed.” This study clearly does not apply to those types of situations.

What should the family physician do?

Mark: There are real concerns about the long-term use of PPIs, especially in older persons. These concerns include increases in community-acquired pneumonia,1 C. difficile colitis,2 and, possibly, hip fractures. So physicians should stop PPIs in those patients in whom it is reasonable to do so, especially if their NSAID can also be stopped.

Bob: Not only are these consequences possible, but there are drug-drug interactions involving PPIs that physicians also need to watch for. Because polypharmacy is such a huge problem among older patients, I would suggest using PPIs only when truly necessary.

Andrea: If your patient needs to be on a PPI and also has osteoporosis or is at high risk of falls, consider recommending a calcium citrate supplement. Compared with other calcium formulations, calcium citrate is less dependent on gastric pH for absorption. Patients should also take calcium with meals to improve absorption.

Main Points

  • Long-term PPI use (longer than one year) in older patients may increase the risk of hip fractures.

  • PPIs may also increase the risk of pneumonia and C. difficile colitis.

  • The physician must balance the risks and benefits of PPIs, especially in patients on chronic NSAID therapy or those with a history of ulcer disease.

EBM Points

  • Check to see if the population in the study is similar to the population you see in your practice. For example, in this study, the average age was 77. These findings should only be generalized to younger patients with caution.

  • Case-control studies may not be able to control for all patient variables. This is a potential source of error. Good examples of this are the case-control studies that suggested that postmenopausal estrogen was cardioprotective. Subsequent randomized controlled trials proved that this is not the case.

  • Case-control studies can suggest an association, but not causation. A randomized trial is needed to assess causation.

Continue Reading

More in AFP

Copyright © 2008 by the American Academy of Family Physicians.

This content is owned by the AAFP. A person viewing it online may make one printout of the material and may use that printout only for his or her personal, non-commercial reference. This material may not otherwise be downloaded, copied, printed, stored, transmitted or reproduced in any medium, whether now known or later invented, except as authorized in writing by the AAFP.  See permissions for copyright questions and/or permission requests.