Abstract
<jats:p>Vitamin D sterols, phosphorus binders and calcimimetics are used to treat chronic kidney disease mineral and bone disorder (CKD-MBD) in hemodialysis. With few randomized trials, providers may titrate agents differently reflecting equipoise and opportunities for clinical trials. We studied patients initiating in-center hemodialysis at Dialysis Clinic, Inc facilities from 2006-2015 and who remained on hemodialysis for ≥90 days (n=23,549). Multinomial logit models assessed titration among users of each medication at the start of the month considering static and dynamic CKD-MBD laboratories. Similarly parameterized logistic models assessed treatment initiation. Differences across facilities were quantified as random effects and corresponding median odds ratios. We observed patterns of titration associated with CKD-MBD laboratories including albumin-corrected serum calcium (Ca), serum phosphorus and parathyroid hormone (PTH) and minimal impact of patient characteristics. Best fit models incorporated 3 months of lagged Ca and phosphorus values and linear splines for current Ca, phosphorus and PTH values. Absolute titration probabilities for vitamin D sterols and calcimimetics were influenced by all three CKD-MBD parameters, such that Ca and phosphorus values altered the threshold PTH at which escalation and de-escalation probabilities crossed. Median odds ratios indicated the greatest facility variation for vitamin D sterol titration. Providers titrate CKD-MBD medications based largely on the full CKD-MBD laboratory phenotype, including the recent serum Ca, phosphorus and PTH history. Facility variation suggests equipoise in titration of vitamin D sterols with opportunities for clinical trials.</jats:p>