Proton-Pump Inhibitors and Risk for Fractures: Breaking Through the Confusion

David A. Johnson, MD


June 07, 2010

This feature requires the newest version of Flash. You can download it here.

Hello. I am Dr. David Johnson, Professor of Medicine and Chief of Gastroenterology at Eastern Virginia Medical School of Norfolk, Virginia.

New news: The FDA has mandated that proton pump inhibitors (PPIs) will now have product labeling that says that patients should not receive these medications without discussion of bone fracture risks.[1] Incremental risk for osteoporosis and fractures, in particular hip fractures, has been cited from epidemiologic studies, and this is news from the FDA. This will apply to both the 14-day labeling for over-the-counter drugs as well as for prescription of all PPIs.

How do you put this in perspective? Let's start with a little bit of an analysis. This is the kind of analysis that I use when I look at any type of controversy like this one. This has been going on for some time as it relates to epidemiologic studies citing potential hip fracture associations, in particular with PPI use. So let's start with an analysis. How would I do this? I start by asking what is the scientific hypothesis? We will then talk about biologic plausibility, and then about epidemiologic studies with the scientific evidence to look at this. And finally, does this apply to your practice?

Let's start with the scientific hypothesis. The working hypothesis in the analyses to date , which have looked at the "indicting inference" that PPIs cause bone fractures, is that acid reduction leads to decreased calcium absorption, and that may lead to increased risk for bone fractures -- hence, osteoporosis-mediated bone fractures. The scientific premise is that acid reduction leads to decreased calcium absorption, and the premise is somewhat predicated on the biologic plausibility that calcium requires a specific pH for appropriate ionization, and the biologic plausibility is somewhat mixed. If you go to the basic science literature and look at [whether] calcium absorption really relates to the degree of acid suppression, the answer is that it is really very controversial across the board. In fact, if we go to the worst case of pernicious anemia, we have many epidemiologic or biologic studies that have specifically looked at achlorhydria and calcium absorption, and the data there are really nil. So we do not really have a strong biologic premise. The scientific hypothesis is very weighted on the calcium solubility data that are, again, in the specific terms of basic science, very mixed across the board.

We then go to the epidemiology studies. This is where it becomes very interesting. To date, until the last week, no prospective trials have looked at the epidemiologic inference in a prospective way indicting PPI use and calcium absorption [as being] related to bone fractures. In fact, the epidemiologic studies cited by the FDA to justify their unilateral decision to make this product label change were all retrospective, cohort observational-type studies that have had marginal odds ratios inference of associated risk, and the odds ratios for all of these studies to date have been well below 2. These studies have not been very well controlled. With observational retrospective studies, and with odds ratios of well below 2, you start to worry about potential inherent biases, just because of study design.

More recently, a very interesting study from Manitoba was published in Gastroenterology [2]just in the last 2 months. It looked at the longitudinal and cross-sectional analysis of a very significant number of patients who were followed in the Manitoba database. All patients were followed in the same centers, and the investigators reviewed cross-sectional cuts to look at PPI use as associated with the diagnosis of osteoporosis. They did a 3:1 case-controlled match, and there was no association on the cross-sectional analysis. The investigators then did a ratio longitudinal analysis by following these patients over a period of up to 5 years. The patients who were followed had at least 2 bone densitometry studies, and then the investigators looked at associated PPI exposure. There was no detriment in the association of PPI use and deteriorating bone density assessment by bone densitometry. So both the cross-sectional and longitudinal follow-ups of these patients did not show any relative risk or potential harm for PPI exposure.

In the most recent literature just published in the Archives of Internal Medicine, a study from the Women's Health Initiative looked at over 160,000 patients followed up to 7 years; this was culled down to 130,000 or so patients who were in the complete database. These people were followed up to 7 years, and the study looked at self-reported associated risk for bone density-related fractures of the hip, spine, and forearm. No associated risk was demonstrated for hip fractures, which has really been the challenge for the epidemiologic studies to date. There is marginal risk as far as odds ratio for cervical spine (the odds ratio was 1.25) and for forearm fracture (the odds ratio was 1.26). I think we are talking about a lot of noise here, potentially, as it relates to realistic expectations. In this type of study, the odds ratios are nominal and, again, very much subject to potential design flaws in cross-sectional bias. But the important part was that hip fractures were not associated with an incremental risk. The investigators did look at bone density, and there was a marginal statistical difference in bone density for hips (the P value was .05), but again, these findings do not equate to a bone density that translates into a hip fracture.

So how would I put this together? I start again with the scientific premise and whether the scientific hypothesis makes sense. We are talking about the risk for PPIs, not achlorhydria. We are talking about marginal reductions in acid potentially making the patient's pH greater than 4.0 for maybe16 -- and in the best of worlds -- 18 hours, and that is about all we can do. If we talk about the biologic plausibility, we talk about the application as it relates to marginal reductions in acid maybe interfering with calcium. But, in fact, if you look at some of the urine calcium studies on people that were put on PPIs, their urine calcium excretion was actually diminished. That doesn't make sense, does it? Take it a step further: If you look at the proton pump, it lives on the osteoclasts which drive bone resorption, so theoretically if you inhibited the proton pump you would inhibit osteoclastic activity and thereby may also change the bone resorption potential. That somehow doesn't make sense with the biologic plausibility [hypothesis for PPIs and bone fracture].

We then go to the epidemiologic studies to date, and with the exception of the recent study in the Archives of Internal Medicine, these were retrospective studies. Perhaps the best studies that were done were from the Manitoba database, which looked at the longitudinal cross-sectional studies and found that there was no associated risk. Then we go to the prospective study that was just reported in the Archives of Internal Medicine by Dr. Gray and her colleagues[3] (the Women's Health Initiative). This looked at 130,000 patients (over 1 million patient-years of exposure) and found no associated risk as it relates to hip fracture, and marginal odds ratios for cervical spine and forearm fractures: 1.25 and 1.25, respectively.

I then come back to the final premise: What do you see in your clinical practice? I have been in medicine for over 30 years. Do you see bone density problems and fractures appearing in your patients with achlorhydria? Think about all of your vagotomy patients and think about your gastrectomy patients. These are people with achlorhydria. There should be a dominant perspective of associated risk in those patients in particular.

I think we have to come back to the overall summary that this is really just "news on the noise" -- there is really not a whole lot here. The observational retrospective studies lend themselves to inherent study biases, especially with low odds ratios.[4] We have excellent data as it relates to more of the prospective data and the cross-sectional and longitudinal database that we see from the excellent study from Manitoba. I think you just have to ask whether this applies to your patient and what your perspective is from your individual patient database. As always, you should look at the exposure for any medication, and I do think the PPIs are incredibly overused in the majority of the patients that I see. But in the ones that you see, you should always ask whether they need the medication. If so, certainly the bone fracture data -- in my mind and from the evidence that I presented to you today -- should not justify unilateral exclusion of these patients [from treatment with PPIs]. I think the FDA language certainly fosters a conversation with the patient, but I would reassure patients -- based on the evidence, the data, and the scientific plausibility as well as the observational evidence based on the prospective studies -- that we really do not have the data to suggest that they would be at increased risk. Look at the risk-[benefit] assessment and what would happen if these patients did not take their medications. Many of these cases are being taken for nonsteroidal prophylaxis and severe reflux disease. Again, I do think we overuse PPIs. It should be a healthy conversation -- eg, "Can you stop the medication? And if so, by all means do it; but if you need the medication, by all means continue it," and be assured that the data are pretty strong that we are really not doing any harm here.

Again, with future studies and longer prospective trials we await further confirmation of this, but my wife takes a PPI and I tell her continue to do so. I think it is appropriate, and I love my wife. Given that, I will leave that to your discretion; apply it to your patients and apply it to your clinical practice, working with those who you are seeing in your practice.

I'm Dr. David Johnson. Thanks for listening. I look forward to future conversation with you.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.