Abnormal mineral metabolism and mortality in hemodialysis patients with secondary hyperparathyroidism: evidence from marginal structural models used to adjust for time-dependent confounding

Background: Hemodialysis patients with mineral and bone disorders (MBDs) have an abnormally high relative risk of death, but their absolute risk of death is unknown. Further, previous studies have not accounted for possible time-dependent confounding of the association between MBD markers and death due to the effect of markers of MBD on treatments, which subsequently may affect MBD markers.

Study design: Multicenter, 3-year, prospective, case-cohort study.

Setting & participants: 8,229 hemodialysis patients with secondary hyperparathyroidism (parathyroid hormone level ≥180 pg/mL and/or receiving vitamin D receptor activators) at 86 facilities in Japan.

Predictors: Serum phosphorus, calcium, and parathyroid hormone levels.

Outcome: All-cause mortality.

Measurements: Marginal structural models were used to compute absolute differences in all-cause mortality associated with different levels of predictors while accounting for time-dependent confounding.

Results: The association between phosphorus level and mortality appeared U-shaped, although only higher phosphorus level categories reached statistical significance: compared to those with phosphorus levels of 5.0-5.9 mg/dL (1.61-1.93 mmol/L), patients with the highest (≥9.0 mg/dL [≥2.90 mmol/L]) phosphorus levels had 9.4 excess deaths/100 person-years (rate ratio, 2.79 [95% CI, 1.26-6.15]), whereas no association was found for the lowest phosphorus category (<3.0 mg/dL [<0.97 mmol/L]; rate ratio, 1.54 [95% CI, 0.87-2.71]). Similarly, hypercalcemia (≥10.0 mg/dL [≥2.50 mmol/L]) was associated with excess deaths, and the highest level of hypercalcemia (≥11.0 mg/dL [≥2.75 mmol/L]) was associated with 5.8 excess deaths/100 person-years (rate ratio, 2.38 [95% CI, 1.77-3.21]) compared to those with levels of 9.0-9.4 mg/dL (2.25-2.37 mmol/L). Abnormally high parathyroid hormone levels were not associated with excess deaths.

Limitations: Possible residual confounding.

Conclusions: These results reinforce the idea that serum calcium (in addition to phosphorus) level is an important predictor of the absolute risk of death in hemodialysis patients with secondary hyperparathyroidism.