rss
Evid Based Med 18:38-39 doi:10.1136/ebmed-2012-100650
  • Aetiology
  • Cohort study

The association between proton pump inhibitor use and hip fracture cannot be explained by differences in dietary and lifestyle choices

  1. Laura E Targownik
  1. Department of Internal Medicine/Gastroenterology, University of Manitoba, Winnipeg, Canada
  1. Correspondence to Laura E Targownik
    805G-715 McDermot Avenue, Winnipeg, Manitoba, Canada, R3E 3P4 Laura.Targownik{at}med.umanitoba.ca

Context

There are currently over a dozen published manuscripts evaluating the relationship between proton pump inhibitors (PPIs) and the occurrence of osteoporotic fractures, in particular, fractures of the hip.1 In the majority of instances, a modest yet significant association between PPI use and hip fracture was detected. However, many of the previous studies were not able to control for potentially relevant confounders, including dietary intake of calcium and vitamin D, smoking and alcohol use, body mass index (BMI) and level of physical activity.

Methods

Khalili et al detail their analysis of 79 899 postmenopausal female participants from the Nurses' Health Study. In this dataset, women were interviewed regarding medical illness, use of medications (including PPIs from 2000 onwards), diet and level of physical activity. Cox proportional hazards models were constructed controlling for age, BMI, calcium and vitamin D intake, smoking, alcohol use, physical activity and other medications known to affect bone mineral density and/or fracture risk.

Findings

PPI use was associated with an increased hazard of developing hip fracture (HR 1.35, 95% CI 1.13 to 1.63), particularly with longer duration of use (HR 1.54, 95% CI 1.03 to 2.31, for those with 6–8 years of use). There was also a significant interaction between PPI use and smoking (HR 1.51, 95% CI 1.20 to 1.91 for current/previous smokers, HR 1.06 (0.77 to 1.46 for never smokers).

Commentary

In this well-designed and performed study, Khalili et al determined that PPIs were associated with an increased risk of hip fracture, even after controlling for dietary and lifestyle factors. There were no significant differences between PPI users and non-users in their reported intake of calcium and vitamin D, and PPI users had higher mean BMI, which theoretically would mitigate against the risk of fracture. These findings are in accordance with the growing body of literature describing a moderate association between PPI use and hip fracture. Furthermore, by controlling for dietary and lifestyle factors, previously raised concerns about possible confounding due to differences in calcium intake, body habitus and physical activity between PPI users and non-users have been appropriately addressed. Overall, this strengthens the case supporting a causal relationship. The authors also report that the correlation between PPI use and fracture is confined to women with a history of smoking. While the suggestion of a synergistic interaction between tobacco use and PPI use is intriguing, the mechanism behind such an effect remains purely speculative.

However, causation cannot be directly inferred from associations detected from epidemiologic analyses. Therefore, complementary lines of evidence, such as evidence supporting a compatible mechanism of action, are required to support the causality of a detected association. While other studies have suggested that PPIs may interfere with intestinal calcium absorption, the actual effect of PPI use on calcium absorption is marginal.2 Other studies have postulated that PPIs have a direct effect on osteoclast proton pumps, but those effects are not well delineated, and appear to promote increased bone mineral density.3 Furthermore, as the effect size of the PPI-fracture association is small, residual confounding may be responsible for the apparent significance of the PPI-fracture association. In particular, this study did not specifically control for concomitant medical comorbidities. The number and severity of concurrent medical illness contributes to overall frailty,4 which increases the risk of falls and fracture. As the prevalence of PPI use is associated with an increased burden of medical illness,5 it remains possible that PPI use may be a marker for overall frailty as opposed to a direct cause of hip fracture.

It remains unlikely that any observational study, no matter how comprehensive and well designed, will provide the level of evidence sufficient to confirm the causality of the PPI-fracture association. As randomised controlled trials to specifically assess the incidence of these adverse outcomes will probably never be performed, further research in this area should focus on determining the mechanism through which PPIs would negatively affect bone health and/or increase fracture risk. In the interim, healthcare practitioners should take care to limit the prescription of PPIs to situations where clear indications exist, and should strive to use the lowest dose for the shortest period of time required to attain a clinically meaningful effect.

Footnotes

  • Competing interests In the last 2 years, I have been on Advisory Boards for Merck Canada and Janssen Canada. I have been on a Speakers Panel for Pfizer Canada.

References