Validation of two prediction models of undiagnosed chronic kidney disease in mixed-ancestry South Africans

BMC Nephrology

Table 2 Overview of the tested models of prevalent chronic kidney disease (CKD) prediction and their performance for the original model and the intercept adjusted model, based on MDRD equation defined CKD

Characteristics	Korean risk score [10]				Thai risk score [9]				Bellville South
Authors	Kwon et al.				Thakkinstian et al.
Year published	2011				2011
Country	South Korea				Thailand
Validation	Internal, External				Internal
Sample size	6565				3459
Design	Cross-sectional				Cross-sectional
Age range	≥30 years				≥18 years
Population	Korean				Thai
Definition of of CKD	eGFR (MDRD) ≤ 60 ml/min per 1.73 m²				stage I-V based on eGFR (MDRD)
Development c-statistic	0.83				0.77
Predictors
Age	Yes				Yes				Yes
Sex	Yes				No				Yes
Anemia	Yes				No				No
Hypertension	Yes				Yes				Yes
Diabetes	Yes				Yes				Yes
History of CVD	Yes				No				No (statin use)
Proteinuria	Yes				No				Yes
Kidney stone	No				Yes				No
Performance	Original		Adjusted		Original		Adjusted
Outcome	eGFR < 60	Any CKD	eGFR < 60	Any CKD	eGFR < 60	Any CKD	eGFR < 60	Any CKD
E/O (95 % CI)	0.07 (0.07-0.08)	0.07 (0.06-0.08)	0.76 (0.67-0.86)	0.76 (0.67-0.85)	0.90 (0.79-1.01)	0.87 (0.77-0.98)	0.98 (0.87-1.10)	0.97 (0.86-1.09)
Brier score	0.265	0.274	0.164	0.161	0.166	0.166	0.165	0.164
Yates slope	0.026	0.029	0.208	0.225	0.191	0.199	0.200	0.211
C-statistic (95 % CI)	0.797 (0.765-0.829)	0.811 (0.780-0.842)	0.797 (0.765-0.829)	0.811 (0.780-0.842)	0.760 (0.726-0.793)	0.772 (0.739-0.805)	0.760 (0.726-0.793)	0.772 (0.739-0.805)
Best threshold	0.03	0.02	0.30	0.31	0.27	0.27	0.31	0.32
Sensitivity (%)	82	84	82	84	73	74	73	74
Specificity (%)	67	68	67	68	72	73	72	73

CI: confidence interval, CKD; chronic kidney disease, CVD: cardiovascular disease, E/O: expected/observed

ISSN: 1471-2369