Risk of proton pump inhibitor-induced mild hyponatremia in older adults.

Abstract

To the Editor: I read the article by Buon et al.,1 “Risk of Proton Pump Inhibitor–Induced Mild Hyponatremia in Older Adults,” with interest. They reported that 24 of 145 included individuals (16.6%) had moderate hyponatremia, and 48 (33.1%) had been taking proton pump inhibitors (PPIs) for longer than 1 year, 31.3% of whom (95% confidence interval (CI) = 18.7–46.3%) had moderate hyponatremia, versus 9.3% (95% CI = 14.3–16.9%) in the rest of the population (odds ratio (OR) = 4.4, 95% CI = 1.8–11.1, P = .001). Individuals taking PPIs and tramadol had a significantly higher risk of hyponatremia than those taking neither (OR = 7.7, 95% CI = 1.9–31.2).1 They concluded that the association between PPIs and tramadol appears to potentiate the risk of hyponatremia, but when the table in the article was analyzed, the use of two more drugs—namely first-generation neuroleptics (P = .01) and corticosteroids (P = .004) with PPIs—was also associated with higher risk of hyponatremia, even more significantly than use of tramadol and PPIs (P = .04). The authors should have analyzed the association between tramadol and PPI use and hyponatremia using a multivariate regression analysis including use of corticosteroids plus PPIs and first-generation neuroleptics plus PPIs as independent variables to find whether the association between tramadol plus PPIs and hyponatremia would remain significant after excluding the hyponatremic effects of corticosteroids plus PPIs and first-generation neuroleptics plus PPIs. The table clearly demonstrates that all three individuals undergoing corticosteroid therapy were also taking PPIs.1 The hyponatremia might have been related to use of corticosteroids rather than use of PPIs. The hyperglycemic effects of corticosteroids are well known, and physiological calculations suggest that the serum sodium concentration should fall by 1.6 meq/L for every 100-mg/dL (5.5 mmol/L) rise in serum glucose concentration,2 although antipsychotic drugs are also a well-known cause of hyponatremia.3 With their published analysis limited with univariate associations, it cannot be concluded that the association between PPIs and tramadol appears to potentiate the risk of hyponatremia. The authors also concluded that their study demonstrates that the chronic use of PPIs increases the risk of hyponatremia in older adults. The authors should also have analyzed the association between PPI use and hyponatremia using a multivariate regression analysis to determine whether the association between PPI and hyponatremia would remain significant after excluding the hyponatremic effects of corticosteroids. What would have been better would have been if they had made this adjustment using multivariate regression analysis including all drugs with known hyponatremia side effects. Their study should be evaluated in light of this limitation. Conflict of Interest: None. Author Contribution: All work was performed by Gulistan Bahat. Sponsor's Role: No sponsor.