Analysis of Study Data
These observational studies suggest a modest increase in the risk of fracture associated with PPIs. Most studies evaluated PPI use for more than 1 year and consistently demonstrated an increased risk of hip fracture by 20–62% and an increased risk of vertebral fracture by 40–60%. The duration of cumulative or continuous exposure did not consistently affect the risk of fracture, but in some studies higher doses of PPIs appeared to be associated with a higher risk of fracture compared with lower doses of PPIs. Only three studies evaluated the effect of short-term PPI therapy of less than 1 month or 30 cumulative doses in subgroup analyses, and two of these studies did not find an association between PPIs and fracture risk in the short-term subgroups.[6,8,9]
Similarly, several meta-analyses recently reported that PPIs were associated with increased risk of hip fracture, vertebral fracture, and any fracture by 20–56%, but the risk of wrist fracture did not reach statistical significance.[19–21] These meta-analyses found that the risk of fracture increased with PPIs regardless of the dose or duration of exposure. Subgroup analyses of studies with different study design, risk of bias, and methodologic quality score did not find that these characteristics influenced the positive association between PPIs and the risk of fractures.
Nonetheless, these meta-analyses reported significant heterogeneity among the studies evaluating the risk of hip fracture. The causes for heterogeneity included differences in patient populations and study design. The studies used databases from Europe, North America, and Asia. The populations from these continents differ in the level of physical activity, dietary intake of calcium and vitamin D, body mass index, and tobacco smoking, which are known risk factors for fracture. Patients taking PPIs have also been shown to have more comorbidities, often have a history of prior fracture, and are involved in limited physical activity. Since each study adjusted for different confounding variables and some known risk factors for fracture were not adjusted, unmeasured confounders or residual confounding may have altered the risk of fracture associated with PPIs, resulting in variable strength of the association among the studies despite a consistently positive association. Another potential cause of heterogeneity is the various definitions of long-term PPI exposure. Whereas most studies evaluated patients with cumulative PPI use over 1 to 2 years, one study found that PPIs were not associated with increased risk of hip fracture until 5 or more years of continuous PPI therapy. Therefore, the duration threshold for the increased fracture risk was variable among the studies and needs to be confirmed with future studies.
In contrast, all five studies evaluating the risk of vertebral fracture showed a positive association with PPIs, and the meta-analyses did not report heterogeneity among these studies. However, the data on vertebral fracture are limited by the inaccuracy in outcome ascertainment. Although the databases used in hip fracture studies were validated and the cases of hip fracture were also adjudicated in the cohort studies,[5,10,11,12,16,24] some studies assessing vertebral fracture were based on self-reports, which were shown to be accurate in only 51% of cases.
Pharmacotherapy. 2012;32(1):67-79. © 2012 Pharmacotherapy Publications