Studies Comparing Users of Proton Pump Inhibitors With Nonusers
A nested case–control study of the United Kingdom General Practice Research Database (UK GPRD) evaluated 13,556 cases of hip fractures that were matched with 135,386 controls to investigate the risk of hip fracture associated with cumulative exposure to PPIs for more than 1 year. The adjusted odds ratio (AOR) for hip fracture in patients who used a PPI for more than 1 year was 1.44 (95% confidence interval [CI] 1.30–1.59). The AOR increased with each additional year of use, from 1.22 (95% CI 1.15–1.30) with more than 1 year of therapy to 1.59 (95% CI 1.39–1.80) with more than 4 years of therapy (p<0.001). The risk of hip fracture was higher with high-dose PPI therapy (AOR 2.65, 95% CI 1.80–3.90) than low-dose PPI therapy (AOR 1.40, 95% CI 1.26–1.54) when PPI users with exposure of more than 1 year were compared with nonusers.
In contrast, a subsequent case–control study, which also used data from the UK GPRD, did not find an association between any PPI use and hip fracture in patients without major medical risk factors. The investigators first evaluated comorbidities in 48,553 subjects to identify medical conditions that were associated with an odds ratio (OR) for hip fracture of greater than 2.0 independently. The analysis was then restricted to subjects (12,021 subjects) without any of these medical conditions, such as osteoporosis, stroke, falls, and prior nonhip fractures, to minimize bias due to confounders. Regardless of the total number of prescriptions, PPI users did not have an increased risk of hip fracture (relative risk [RR] 0.9, 95% CI 0.7–1.1). However, the study results are not generalizable to the typical patient population that is at risk for fracture since patients with known risk factors for osteoporosis and fracture were excluded. Compared with the other studies, patients included in this analysis likely had a lower fracture risk at baseline, which may have altered the impact of PPIs on fracture risk.
Using the database from Kaiser Permanente Northern California, a study compared 33,752 cases of hip fracture and 130,471 matched controls to assess the risk of hip fracture associated with long-term PPI use. The authors evaluated many potential confounders, but only included smoking in the final model for statistical analyses. The PPI users with cumulative exposure of 2 years or longer had a greater risk for hip fracture compared with nonusers (OR 1.30, 95% CI 1.21–1.39). A saturated model containing all potential confounders yielded a similar, nonsignificant OR of 1.22 (95% CI 0.96–1.54). The strength of association increased with increasing number of pills per day, but the cumulative duration of PPI use did not affect the OR. It should be noted that most PPIs are available in multiple strengths, but the investigators did not define the strength of a standard dose of each PPI. The dose intensity was calculated from the number of pills dispensed divided by the duration of PPI exposure. Patients in the three dose categories (<0.75 pills/day, 0.75–1.49 pills/day, and ≥1.5 pills/day) were likely to have overlapping daily doses, resulting in inaccurate evaluation of the dose-dependent effect of PPIs.
Another large case–control study evaluated the risk of fracture associated with PPI use in the Danish population. A total of 124,655 subjects who sustained a fracture in the year 2000 were identified as cases from a national administrative database and were matched with 373,962 controls. Since this study included trauma-related fractures, the mean age was only 43 years. Any PPI use within the last year was associated with an increased risk of any fracture (AOR 1.18, 95% CI 1.12–1.43), hip fracture (AOR 1.45, 95% CI 1.28–1.65), and spine fracture (AOR 1.60, 95% CI 1.25–2.04). Subjects whose last use of a PPI was more than 1 year ago did not have an increased risk of fracture. The authors did not observe a dose-response relationship with the cumulative dose or the average daily dose, but the cumulative exposure of PPI therapy was much shorter in this population than in other studies. The cumulative PPI exposure was categorized into less than 25 doses, 25–99 doses, or 100 or more doses.
A small case–control study evaluated 1241 cases of hip fracture recorded in a national database in Taiwan from 2005–2006. Exposure to a PPI was defined as any PPI use from 1996 to the index date. Compared with nonusers, patients who received more than 29 cumulative doses of PPIs had an OR adjusted for concurrent drugs or comorbidities of greater than 1.0, but the authors did not report the OR adjusted for all confounders. In addition, the database has not been validated for determining the prevalence of hip fracture. Like the Danish study, the cumulative exposure of PPI therapy was short, with more than 70 cumulative PPI doses as the category of the most exposure.
The above studies defined PPI exposure by duration of exposure or cumulative doses; hence, some subjects might have had distant, intermittent PPI exposure. A Canadian case–control study assessed the risk of osteoporosis-related fractures associated with continuous PPI therapy by comparing 15,792 cases with 47,289 matched controls. Continuous PPI exposure was defined as PPI exposure in greater than 70% person-time before the index date. The combined risk of hip, vertebral, or wrist fracture in PPI users was not different from that of nonusers until they had 7 or more years of exposure (AOR 1.92, 95% CI 1.16–3.18). Likewise, the risk of hip fracture was increased after continuous PPI use for 5 or more years (AOR 1.62, 95% CI 1.02–2.59), and the strength of association increased in the subsequent years.
A prospective cohort study evaluated the risk of fracture in 130,487 postmenopausal women from the Women's Health Initiative (WHI) observation study and clinical trials. Fractures were reported by patients with central adjudication for all hip fractures and for nonhip fractures in the WHI clinical trials. Current PPI exposure was reported by patients at baseline and at a 3-year visit. Participants were followed for a mean duration of 7.8 years. The results were adjusted for many confounders such as physical activity levels, which would not be available in administrative databases used in case–control studies. Calcium and vitamin D intake were also evenly distributed among the PPI users and nonusers, due to the concurrent enrollment to the nutrition supplementation arms of the study. Use of PPIs at baseline was not associated with hip fracture (hazard ratio [HR] 1.00, 95% CI 0.71–1.40) but increased the risk for clinical spine fracture (HR 1.47, 95% CI 1.18–1.82), forearm or wrist fracture (HR 1.26, 95% CI 1.05–1.51), and total fracture (HR 1.25, 95% CI 1.15–1.36). Incorporating the change in PPI use over time as a time-dependent variable had minimal effect on the HRs. Increasing duration of PPI therapy did not increase the risk of any fractures.
Data from 5339 female participants of the Study of Osteoporotic Fractures (SOF) and 5755 male participants of the Osteoporotic Fractures in Men Study (MrOS) were retrospectively reviewed to assess the fracture risk associated with current PPI use. Use of a PPI was determined at clinic visits, and fractures were reported to the investigators periodically with confirmation by radiographic reviews. Trauma-related fractures were included in the analyses. The average duration of follow-up for fractures was 7.6 years in SOF and 5.6 years in MrOS; however, only 2856 women (54%) completed the second visit when the repeated hip BMD measurement and updated drug use information were obtained in the SOF trial. In the SOF trial, use of PPIs was associated with an increased risk in nonspine fracture (relative hazard [RH] 1.34, 95% CI 1.10–1.64) but did not increase the risk of hip fracture. In the MrOS trial, neither nonspine fracture nor hip fracture was associated with PPI use.
Researchers reviewed data of 1211 postmenopausal women who had participated in the Osteoporosis and Ultrasound Study to evaluate the risk of lumbar and thoracic vertebral fracture. The investigators collected data on osteoporosis risk factors from a questionnaire. Omeprazole use was identified at baseline as part of the initial patient interview, which identified 61 omeprazole current users within the previous year. Patients were followed for a mean duration of 6.1 years. Forty-nine patients experienced at least one radiographically confirmed vertebral fracture. The relative risk (RR) for vertebral fracture was 3.10 (95% CI 1.14–8.44). Several methodology issues limited the validity of the results, such as incomplete information on 1178 patients who were excluded from the analysis due to lack of baseline data on omeprazole use or were lost to radiographic follow-up.
In a prospective cohort study of Japanese postmenopausal women who presented to two hospitals for the evaluation and treatment of both osteoporosis and upper gastrointestinal disorders, 18 patients who required PPIs for more than 6 months for refractory reflux esophagitis were compared with 57 patients without reflux esophagitis. Patients taking PPIs for refractory reflux esophagitis had a larger number of vertebral fractures, assessed by spine radiographs, than did patients without reflux esophagitis (2.50 ± 2.60 vs 1.11 ± 1.68 episodes, p=0.009). Nonetheless, this study had major methodologic limitations, including the lack of information on the baseline characteristics of patients and the frequency of vertebral fracture assessments. The authors also did not report an estimate of the effect size of vertebral fractures comparing patients taking PPIs for refractory reflux esophagitis and patients without reflux esophagitis.
The impact of concurrent PPI therapy on the antifracture efficacy of alendronate at the hip, spine, humerus, and forearm was evaluated in 38,088 patients, by using national databases in Denmark. The mean duration of follow-up was 3.5 years, and the mean age was 70 years. A base model, incorporating alendronate refill compliance, age, sex, comorbidities, prior fracture, and concurrent drugs, was established to calculate the fracture risk reduction associated with alendronate therapy. Concurrent PPI use had significant interaction with hip fracture risk reduction with alendronate (p<0.05). The adjusted HR for hip fracture was 0.61 (95% CI 0.52–0.71) with complete alendronate compliance in PPI nonusers, but the benefit of alendronate was attenuated with concurrent PPI use (adjusted HR 0.81, 95% CI 0.64–1.01). The risk reduction of hip fracture by alendronate was progressively attenuated by increasing doses of PPIs, but PPIs did not affect the risk reduction of spine, humerus, or forearm fracture by alendronate. The interaction between alendronate and concurrent PPI use on hip fracture risk was also significant only in patients aged 70 years or older (p<0.001) but not in patients younger than 70 years (p=0.90).
Studies Comparing Current Users With Past Users of Proton Pump Inhibitors
The risk of fracture after the discontinuation of PPIs was evaluated in three studies, the results of which showed reduction in fracture risk in 1 month–1 year after PPIs were discontinued. The case–control study of the Kaiser Permanente Northern California database found that the risk of hip fracture associated with prolonged PPI exposure was not statistically significant if the last use of a PPI was more than 1 year ago. Compared with PPI nonusers, prolonged PPI users who discontinued their PPIs in less than 1 year, in 1–1.9 years, and in 2–2.9 years had ORs for hip fracture of 1.30 (95% CI 1.21–1.41), 1.24 (95% CI 0.90–1.72), and 1.09 (95% CI 0.62–1.85), respectively.
Similarly, a case–control study using the Dutch PHARMO Record Linkage System evaluated 6763 cases of hip fracture that were matched to 26,341 controls by age and sex. The authors defined current, recent, past, and distant past uses by the last date a PPI was dispensed: within 30 days, 31–91 days, 92–365 days, and more than 1 year before the index date, respectively. The current PPI users had an increased risk of hip fracture compared with nonusers (AOR 1.20, 95% CI 1.04–1.40); however, after discontinuation of the PPI, the risk returned to baseline as demonstrated by AORs of 0.96 (95% CI 0.83–1.12), 0.97 (95% CI 0.74–1.26), and 1.24 (95% CI 1.08–1.43) associated with recent, past, and distant PPI use, respectively. In contrast to the Canadian case–control study, increasing the duration of continuous PPI therapy from 3 months or less to more than 36 months was associated with reducing AORs from 1.26 (95% CI 0.94–1.68) to 1.09 (95% CI 0.81–1.47).
In another study that used the UK GPRD, data from patients who started PPI therapy between January 1988 and February 2007 were evaluated in a retrospective cohort study to compare the effect of current PPI use and past PPI use. The period of current use was defined as the period from the start date of a PPI prescription to 91 days after the end date of the prescription, and the period of past use was the period after discontinuation of PPIs for more than 91 days. Patients could move between the two exposure categories. Compared with past use of a PPI, current use increased the risk of any fractures (ARR 1.15, 95% CI 1.10–1.20), hip fracture (ARR 1.22, 95% CI 1.10–1.37), and vertebral fracture (ARR 1.40, 95% CI 1.11–1.78), but it did not increase the risk of radial or ulnar fracture. The risk of hip fracture increased with increasing daily dose during the current use period, but longer duration of current use was not associated with increasing risk of fracture.
Pharmacotherapy. 2012;32(1):67-79. © 2012 Pharmacotherapy Publications