CHICAGO – Proton pump inhibitor use was significantly associated with hypomagnesemia in a large, retrospective analysis of 1,317 hospitalized adults.
The potential effects of PPI use on serum magnesium levels did not appear to be dose related, Dr. Jen-Tzer Gau and his colleagues reported at the annual Digestive Disease Week.
In a safety announcement released in earlier this year, the Food and Drug Administration warned that prolonged use of prescription proton pump inhibitors (PPIs) can reduce magnesium levels, possibly leading to serious adverse events including tetany, arrhythmias, and seizures.
The agency recommends that providers obtain serum magnesium levels prior to initiating PPI therapy, and then periodically check levels in patients who are on prolonged PPI treatment or who take PPIs with digoxin or drugs such as diuretics that may cause hypomagnesemia.
The safety alert was based on 38 cases in the Adverse Event Reporting System and 23 cases in the literature (at least 8 of which were included in the 38 in AERS), and it notes that hypomagnesemia is likely underrecognized and underreported.
The current study involved 1,317 patients at a rural community hospital in southeast Ohio, of whom 464 were PPI users and 853 were not. Hypomagnesemia was defined as a magnesium level of less than 1.7 mg/dL.
The mean magnesium level on admission was 1.91 mg/dL among PPI users, compared with 2.00 mg/dL among nonusers (P = .004), said Dr. Gau, chair of geriatric medicine/gerontology at Ohio University College of Osteopathic Medicine in Athens. PPI users and nonusers were similar with regard to age (mean 73 years), gender (62% female), past smoking history (roughly 22%), length of stay (mean 4.5 days), and discharge diagnosis, although PPI users had significantly more GI illness, as would be expected (31% vs. 26%, P = .02).
In logistic regression analyses, PPI therapy was associated with an increased risk of hypomagnesemia occurrence (crude odds ratio 2.41), and the association remained significant after adjustment for all confounders (OR 2.41, 95% confidence interval 1.34-4.34), Dr. Gau said.
The odds ratio was 2.29 after adjustment for age, sex, and serum albumin levels (model 1; 95% CI 1.35-3.89) and was 2.21 after researchers also adjusted for serum levels of potassium, calcium, and creatinine; supplementation with potassium and magnesium; history of diabetes and chronic obstructive pulmonary disease; and diuretic use (model 2; 95% CI 1.27-3.85). The fully adjusted model included all of the above variables plus history of heart failure and coronary artery disease, a diagnosis of acute GI illness, and use of iron supplements, antipsychotics, antihistamines, narcotics, NSAIDs, antidepressants, beta2-agonist bronchodilators, inhaled corticosteroids, and oral laxatives.
In an analysis of variance that evaluated PPI dosage, the mean magnesium level was 2.0 mg/dL with a defined daily dose (DDD) of 0, 1.89 mg/dL with a DDD of 1, and 1.92 mg/dL with a DDD of 2 or more (P = .026), Dr. Gau said.
In a regression analysis, however, no significant association between hypomagnesemia and PPI use was observed for the higher PPI dose compared with no PPI use (DDD 2 or more vs. DDD 0) in the adjusted model 2 (OR 1.80; CI 0.98-3.30). Otherwise, there were significant associations observed in the rest of the models between hypomagnesemia and PPI use.
During a discussion of the study at the meeting, an attendee asked why the association with hypomagnesemia was not more profound at higher levels of PPI use. Dr. Gau replied that the finding that both higher and lower DDD units appeared to significantly increase the risk for hypomagnesemia is consistent with other case reports suggesting that it was a non–dose related or time-related adverse drug reaction.
Patients with lower serum albumin levels had significantly lower magnesium levels and an increased risk for hypomagnesemia (adjusted OR 2.31; CI 1.28-4.16). Dr. Gau said serum potassium levels were also associated with lower serum magnesium levels, but he did not provide specific data.
Limitations of the study include the fact that most of the patients were white, the lack of data on duration of PPI use (although most patients were probably long-term users), he said, and the potential for unmeasured confounders. The FDA noted that PPI-treated patients with reported hypomagnesemia had typically taken PPIs for more than 1 year.
Given the potential for additional confounders in a hospital-based population, one attendee suggested that a community-based cohort would have been more meaningful.
Dr. Gau said that future work should address the potential underlying mechanism for lower magnesium levels among PPI users. The exact mechanism is unknown, although it is thought that PPIs may interfere with magnesium absorption from the GI tract.