We appreciate the interest of Dr. Wakasugi and Dr. Sakaguchi in our recent publication about the association of serum mineral parameters with mortality in hemodialysis patients, an analysis of the data from the nationwide Korean Society of Nephrology (KSN) End-Stage Renal Disease (ESRD) Registry. As they have commented, our results have some discrepancy with other studies on the association between serum phosphorus level and mortality, including one from Japanese Renal Data Registry data [
1], in that hyperphosphatemia was not significantly associated with increased mortality. Although limited, data from one study showed results similar to ours. In a Portuguese observational cohort study, patients with hyperphosphatemia did not show difference in mortality (hazard ratio [HR], 1.00; 95% confidence interval [CI], 0.74–1.35), whereas patients with hypophosphatemia showed increased mortality (HR, 1.29; 95% CI, 1.07–1.55) [
2].
We speculated several explanations for our results. Favorable nutritional status may have attenuated the hazardous effect of hyperphosphatemia. In our
post-hoc analysis of the KSN ESRD Registry data, increased risk of mortality in the high phosphorus group was exhibited only if combined with concurrent low serum albumin level, suggesting that the nutritional status attenuated the harmful effect of high phosphorus level (
Table 1). Nutritional indices other than serum albumin could provide further information. However, we were unable to examine the relevance of normalized protein nitrogen appearance (nPNA) in the relationship between phosphorus and mortality, because data on nPNA were missing in many subjects.
Second, we performed analysis based on one-time measurement of phosphorus at baseline. Serum phosphorus is known to have a wide diurnal variation. The Kidney Disease Improving Global Outcomes Guidelines recommends that therapeutic choice of phosphate-lowering treatment be based on repeated measures of serum phosphorus level. The increased mortality risk of high phosphorus was accentuated in a time-dependent model compared to the baseline model [
1,
3].
Third, Koreans tend to ingest phosphorus less from milk products and more from legumes and vegetables compared with the United States population [
4,
5]. Differences in food sources of phosphorus intake may partly account for the non-significant risk of hyperphosphatemia.
As was suggested, performance status and magnesium can also affect the outcome of hyperphosphatemic patients, information on which is not available in the current KSN ESRD Registry.
Further research to investigate possible confounding factors is needed using analysis of repeated measurement of phosphorus level or more comprehensive analysis including diet and nutritional variables.