Variability in measures of mineral metabolism in children on hemodialysis: impact on clinical decision-making.

BACKGROUND

Variability in measures of mineral metabolism has not been studied in pediatric end stage kidney disease. We sought to determine the intra-individual variability in measures of mineral metabolism in children on hemodialysis (HD) and its impact on clinical decision-making.

METHODS

We conducted a prospective single-center study of children (3.6-17.3 years old) on chronic HD. Serial twice weekly measures of serum calcium, phosphate and intact parathyroid hormone (PTH), as well as weekly measures of fibroblast growth factor 23 (FGF23) and vitamin D metabolites, were obtained over a 12-week period in 10 children. Samples (n = 226) were assayed in a single batch at the end of the study.

RESULTS

The median intra-individual coefficient of variation (CV) calculated by 4-week blocks was 5.1-6.5% for calcium, 9.5-14.9% for phosphate and 32.7-33.4% for PTH. The median overall CV for FGF23 was 44.4%. Using the first value of each block as a reference, subsequent values would dictate a discrepant management decision 33-56%, 19-28%, and 30-33% of the time for calcium, phosphate, and PTH, respectively. Adjusting for sex and age, most of the variability in phosphate and PTH was attributable to within-participant variability. For calcium, 49% of the variability was attributable to day of blood collection (Monday vs. Friday). The median (range) of an individual participant's values within clinical target ranges was 55% (26-86%) for calcium, 58% (0-96%) for phosphate, and 21% (0-64%) for PTH.

CONCLUSIONS

There is considerable intra-individual variability in measures of mineral metabolism that serve as surrogate markers for bone health in children on HD. Within a 4-week period, at least 20-30% of measures would dictate a discrepant decision from the referent measure of that month. These findings have important implications for clinical decision-making and underscore the need to base therapeutic decisions on trends rather than single measurements.