Proton Pump Inhibitors and Hypomagnesemia in the General Population

A Population-Based Cohort Study

Brenda C.T. Kieboom, MD; Jessica C. Kiefte–de Jong, RD, PhD; Mark Eijgelsheim, MD, PhD; Oscar H. Franco, MD, PhD; Ernst J. Kuipers, MD, PhD; Albert Hofman, MD, PhD; Robert Zietse, MD, PhD; Bruno H. Stricker, MMed, PhD; Ewout J. Hoorn, MD, PhD

Disclosures

Am J Kidney Dis. 2015;66(5):775-782. 

In This Article

Results

Cohort Characteristics

The study population consisted of 9,818 participants, of whom 96.0% were of European ancestry. Population characteristics are shown in Table 1. A total of 724 participants used PPIs (7.4%), with 1.2 ± 0.6 defined daily doses. Thirty-six PPI users (5.0%) had hypomagnesemia (lowest serum magnesium, 0.68 mEq/L). There were 250 participants who used H2RAs (2.5%), with 0.9 ± 0.4 defined daily doses. Twelve H2RA users (4.8%) had hypomagnesemia (lowest serum magnesium, 1.34 mEq/L). Figure 1 shows how serum magnesium concentrations were distributed between participants without acid-suppressive medication and participants who used PPIs or H2RAs. This figure shows that participants using PPIs or H2RAs more often had a serum magnesium level in the lowest quartile.

Figure 1.

Distribution of serum magnesium levels among participants not using acid-suppressive medication, and participants who used proton pump inhibitors (PPIs) or H2 receptor antagonists (H2RAs). PPI users and H2RA users more often had a serum magnesium level in the lowest quartile.

Relationship of PPI use to Serum Magnesium

Use of a PPI was associated with significantly lower serum magnesium levels (−0.024 [95% CI, −0.034 to −0.016] mEq/L) compared to no use and this relationship persisted after multivariable adjustment (−0.022 [95% CI, −0.032 to −0.014] mEq/L; Table 2). We found significant interaction between PPI use and loop diuretic use (P for interaction = 0.03). A greater reduction in serum magnesium levels was observed in participants who concomitantly used PPIs and loop diuretics (−0.070 [95% CI, −0.120 to −0.020] mEq/L) compared to participants who used none of these drugs. No significant interaction between PPI use and thiazide diuretic use or the presence of diabetes mellitus was identified (P for interaction = 0.9 for both). After multivariable adjustment, lower serum magnesium levels were also observed in participants who used H2RAs (−0.016 [95% CI, −0.032 to −0.002] mEq/L) compared to those with no use. No significant interaction was observed between H2RA use and loop diuretic use (P for interaction = 0.09). This also pertained to H2RA use and thiazide diuretic use, as well as the presence of diabetes mellitus (P for interaction = 0.09 and = 0.5, respectively).

Relationship of PPI use to Hypomagnesemia

Use of a PPI was associated with higher risk of hypomagnesemia compared to no use, before and after adjustment for potential confounders (odds ratios [ORs] of 2.27 [95% CI, 1.58–3.27] and 2.00 [95% CI, 1.36–2.93], respectively; Table 3). H2RA use was also associated with increased risk of hypomagnesemia, before and after adjustment for potential confounders (ORs of 2.19 [95% CI, 1.21–3.98] and 2.00 [95% CI, 1.08–3.72], respectively) compared to no use. A duration-of-use analysis showed that the increased risk of hypomagnesemia was mainly present in participants in the highest tertile of duration of PPI use (>182 days; P for trend < 0.001; Table 4). In these participants, the risk of hypomagnesemia nearly tripled (OR, 2.99; 95% CI, 1.73–5.15) compared to no use. In H2RA use, the same trend was observed, with increased risk of hypomagnesemia in the highest tertile of duration of use (>111 days; OR, 2.55; 95% CI, 0.99–6.58; P for trend = 0.02) compared to no use.

Sensitivity Analyses

To analyze whether PPI-induced hypomagnesemia could be explained by confounding by indication, we performed sensitivity analyses with serum phosphate level, dietary magnesium intake, and use of vitamin and mineral supplementation. No significant association was found between PPI use and serum phosphate level (0.006 [95% CI, −0.028 to 0.040] mg/dL) compared to no use. This suggests that the indication for PPIs did not result in poor dietary intake and therefore lower serum magnesium and phosphate levels. Furthermore, the association between PPI use and serum magnesium level was not altered by the addition of energy-adjusted magnesium intake to the adjusted model (before adjustment, −0.030 [95% CI, −0.046 to −0.014] mEq/L; after adjustment, −0.030 [95% CI, −0.044 to −0.014] mEq/L). Similarly, the addition of self-reported use of vitamin and mineral supplementation in the model did not change the association between PPI use and serum magnesium level (−0.022 [95% CI, −0.030 to −0.012] mEq/L before and after adjustment) compared to no use.

processing....