External validation of six clinical models for prediction of chronic kidney disease in a German population

BMC Nephrology

Table 1 Identified prediction models for CKD: included parameters and their coefficients

	Bang SCORED	Bang „modified SCORED “	Kearns	Kshirsagar	Kwon	Thakkinstian
No of parameters	9	7	5	8	7	4
Intercept	-5.40^a	-5.38^a	-3.63	-3.30	-6.53^a	-2.8^b
Age (yrs)	1.55 [50–59] 2.31 [60–69] 3.23 [≥ 70]	1.55 [50–59] 2.29 [60–69] 3.29 [≥ 70]	1.075 [per 10 yrs^c] -0.01 [age²/10 yrs^c] 0.104 [age < 50]	0.63 [50–59] 1.33 [60–69] 1.46 [≥ 70]	1.16 [50–59] 1.91 [60–69] 2.71 [≥ 70]	0.6 [50–59] 1.4 [60–69] 2.1 [≥ 70]
Sex – Female	0.29	0.34	0.73	0.13	0.40
Anemia	0.93			0.48	0.94
Hypertension	0.45	0.47	0.74 + age < 50: 0.56	0.55	0.48	0.80
Diabetes	0.44	0.47		0.33	0.73	0.90
Ischemic heart disease or stroke (Hx)	0.59	0.67		0.26	0.60
Heart failure (hx)	0.45	0.51	0.86 CHF + age < 50: 0.29	0.50
Ischemic heart disease (Hx)			0.51 + age < 50: 0.13
Peripheral vascular disease (Hx)	0.74			0.41
Proteinuria	0.83	0.88			0.48
Kidney stones (Hx)						1

Intercept: the baseline prevalence of CKD in the cohort if all other predictors of a model are not present, meaning their coefficients are zero
Coefficient: The impact with which a predictor effects the estimation of CKD probability. The larger the coefficient, the stronger the impact; e.g. anemia (coefficient 0.93) has a higher impact in the SCORED model on the estimation of CKD probability than diabetes (0.44)
^a Intercept according to personal information by H. Bang
^b Intercept estimated using the prevalence of CKD in the validation population
^c age per 10 yrs = (age -46.72) / 10yrs

ISSN: 1471-2369