Abstract Objectives This study aims to explore the biological variability of calcium homeostasis biomarkers, including calcium, phosphate, magnesium, 25-OH vitamin D, 1,25-OH vitamin D, and parathyroid hormone (PTH), and to develop a model that predicts PTH concentrations based on these parameters. Methods Thirty healthy volunteers were sampled on a weekly basis over a period of 10 weeks. Each sample was analyzed in duplicate. Statistical methods included estimating within- and between-subject biological variation (CV I , CV G ) using ANOVA with outlier removal and a Bayesian model. A linear mixed-effects model (LMM) was used to estimate 95 % prediction intervals for PTH as a function of calcium, phosphate, and 25-OH vitamin D in healthy participants. The effect of applying this prediction model in routine practice was estimated using data from the Laboratory Information System. Results Estimated CV I from ANOVA and the Bayesian model was: PTH, 18.0 % and 17.4 %; calcium, 1.7 % and 1.6 %; phosphate, 9.3 % and 9.3 %; 25-OH vitamin D, 4.7 % and 7.2 %; and 1,25-OH vitamin D, 16.2 % and 17.1 %. The LMM indicated the best 95 % prediction interval for PTH included calcium (PTH (pmol/L)=exp(3.46−2.67·ln(Ca (mmol/L)) ± Z·0.29)) and did not improve with phosphate and/or 25-OH vitamin D. Compared with conventional reference intervals, this model flagged 56 % vs. 41.6 % of routine PTH/calcium results, respectively. Conclusions Similar CV I data were obtained using two different methods. A predictive model can be used to predict concentrations for co-regulated biomarkers, potentially enhancing sensitivity in identifying non-physiological results and facilitating clinicians’ awareness of conditions affecting calcium homeostasis.
Guldhaug et al. (Fri,) studied this question.