Differences between hospitals in attainment of parathyroid hormone treatment targets in chronic kidney disease do not reflect differences in quality of care

Table 4 Multivariate linear regression (backward) analyses on predictors of plasma (Ln)PTH level

eGFR calculated with	Model with or without treatment center	Explained variance	Independent predictors	Linear regression-coefficient [95% CI]
MDRD-4 before creatinine conversion	Including treatment center	R² = 0.42	Center	0.49 [0.31-0.68]
			Kidney transplant	0.55 [0.33-0.78]
			eGFR (MDRD-4)	−0.04 [−0.05 - -0.03]
	Without treatment center	R² = 0.34	Kidney transplant	0.72 [0.48-0.95]
			eGFR (MDRD-4)	−0.04 [−0.04 - -0.03]
MDRD-6 before creatinine conversion	Including treatment center	R² = 0.45	Center	0.36 [0.19-0.54]
			Kidney transplant	0.54 [0.32-0.76]
			eGFR (MDRD-6)	−0.04 [−0.05 - -0.03]
	Without treatment center	R² = 0.40	Kidney transplant	0.67 [0.46-0.89]
			eGFR (MDRD-6)	−0.04 [−0.05 - -0.03]
MDRD-6 after creatinine conversion	Including treatment center	R² = 0.45	Center	0.13 [−0.05-0.31]
			Kidney transplant	0.54 [0.32-0.76]
			eGFR (MDRD-6)	−0.03 [−0.04 - -0.03]
	Without treatment center	R² = 0.44	Kidney transplant	0.59 [0.38-0.79]
			eGFR (MDRD-6)	−0.03 [−0.04 - -0.03]

The difference in R² between the model with and without treatment center illustrates the contribution of the treatment center to the explained variance. The linear regression coefficient reflects the association between the independent predictor and (Ln)PTH. eGFR is a very important determinant, since for every extra ml filtration per minute, PTH decreases by three percent (based on the final model using MDRD-6 after creatinine conversion).
Ln: natural logarithm; PTH: parathyroid hormone (pmol/l); eGFR: estimated glomerular filtration rate (ml/min/1.73 m²); MDRD: modification of diet in renal disease; CI: confidence interval.

ISSN: 1471-2369