Table 2

Overview of PTs validated in independent external validation studies

PT	Source	Population*	Prediction†	Discrimination‡	Calibration§
Start preparation for kidney failure?
No. 2	Retrospective cohort study60	African-American patients with an eGFR between 20 and 65 mL/min/1.73 m²	1-year risk of kidney failure 5-year risk of kidney failure	C-statistic 1-year risk of kidney failure=0.98 C-statistic 5-year risk of kidney failure=0.83	No
No. 2	Retrospective cohort study61	Patients with CKD stage 2–5	3-year risk of kidney failure	C-statistic=0.91	Calibration plot
No. 2	Multicentre retrospective cohort study32	Patients with an eGFR <60 mL/min/1.73 m²	5-year risk of kidney failure	C-statistic=0.95	Calibration plot
No. 2	Retrospective cohort study62	Patients with CKD stage 3–5	5-year risk of kidney failure	C-statistic=0.88	Calibration plot and Hosmer-Lemeshow test (p=0.05)
No. 3	Retrospective cohort study62	Patients with CKD stage 3–5	5-year risk of kidney failure	C-statistic=0.89	Calibration plot and Hosmer-Lemeshow test (p=0.03)
No. 2	Multicentre retrospective cohort study67	Patients with CKD stage 3–5	2-year risk of kidney failure 5-year risk of kidney failure	C-statistic 2-year risk of kidney failure=0.90 C-statistic 5-year risk of kidney failure=0.88	Calibration plot 2-year risk of kidney failure Calibration plot 5-year risk of kidney failure
No. 3	Multicentre retrospective cohort study67	Patients with CKD stage 3–5	2-year risk of kidney failure 5-year risk of kidney failure	C-statistic 2-year risk of kidney failure=0.90 C-statistic 5-year risk of kidney failure=0.88	Calibration plot 2-year risk of kidney failure Calibration plot 5-year risk of kidney failure
No. 3	Retrospective cohort study31	Patients >65 years of age with an eGFR <30 mL/min/1.73 m²	1-year risk of kidney failure	C-statistic development cohort=0.78 C-statistic validation cohort=0.78	No
No. 2	Multicentre prospective cohort study63	Patients >75 years of age with an eGFR <20 mL/min/1.73 m²	2-year risk of kidney failure 5-year risk of kidney failure	C-statistic 2-year risk of kidney failure=0.59 C-statistic 5-year risk of kidney failure=0.57	Calibration plot and Hosmer-Lemeshow test 2-year risk of kidney failure (p=0.57) Calibration plot and Hosmer-Lemeshow test 5-year risk of kidney failure (p=0.70)
No. 4			1-year risk of kidney failure	C-statistic=0.66	Calibration plot and Hosmer-Lemeshow test (p=0.47)
No. 5			5-year risk of kidney failure	C-statistic=0.65	Calibration plot and Hosmer-Lemeshow test (p=0.72)
No. 19			2-year risk of kidney failure 4-year risk of kidney failure	C-statistic 2-year risk of kidney failure=0.64 C-statistic 4-year risk of kidney failure=0.65	Calibration plot and Hosmer-Lemeshow test 2-year risk of kidney failure (p=0.85) Calibration plot and Hosmer-Lemeshow test 4-year risk of kidney failure (p=0.41)
Start with dialysis or not?
No. 25	Multicentre retrospective cohort study64	Patients >67 years of age on dialysis	Death within 6 months of dialysis initiation	C-statistic=0.63	No
No. 25	Multicentre retrospective cohort study46	Patients >67 years of age on dialysis	Death within 6 months of dialysis initiation	C-statistic development cohort=0.66 C-statistic validation cohort=0.67	Calibration plot development cohort Calibration plot validation cohort
No. 25	Retrospective cohort study55	Patients >65 years of age on dialysis (PD/HD)	Death within 6 months of dialysis initiation	C-statistic=0.70	No
No. 26	Multicentre retrospective cohort study65	Patients on dialysis	Death within 3 months of dialysis initiation Death within 6 months of dialysis initiation Death within 12 months of dialysis initiation	C-statistic 3-month mortality=0.74 C-statistic 6-month mortality=0.74 C-statistic 12-month mortality=0.74	No
No. 20	Retrospective cohort study66	Patients >75 years of age on dialysis	Death within 6 months of dialysis initiation	C-statistic=0.67	Calibration plot and Hosmer-Lemeshow test (p=0.004)
No. 25			Death within 6 months of dialysis initiation	C-statistic=0.61	Calibration plot and Hosmer-Lemeshow test (p=0.45)
No. 26			Death within 3 months of dialysis initiation	C-statistic=0.62	Calibration plot and Hosmer-Lemeshow test (p=0.03)
No. 30			Death within 3 months after dialysis initiation Death within 6 months after dialysis initiation	C-statistic 3-month mortality=0.57 C-statistic 6-month mortality=n.a.	Calibration plot and Hosmer-Lemeshow test 3-month mortality (p=0.43) Calibration plot 6-month mortality
No. 33			Death within 6 months of dialysis initiation	C-statistic=0.57	Calibration plot and Hosmer-Lemeshow test (p=0.43)
Accept or decline DDKT offer?
No. 51	Multicentre retrospective cohort study68	DDKT recipients	6-month DDKT graft survival 1-year DDKT graft survival 3-year DDKT graft survival 5-year DDKT graft survival 7.5-year DDKT graft survival	C-statistic 6-month DDKT graft survival=0.62 C-statistic 1-year DDKT graft survival=0.63 C-statistic 3-year DDKT graft survival=0.63 C-statistic 5-year DDKT graft survival=0.63 C-statistic 7.5-year DDKT graft survival=0.63	Calibration plot 6-month DDKT graft survival Calibration plot 1-year DDKT graft survival Calibration plot 3-year DDKT graft survival Calibration plot 5-year DDKT graft survival Calibration plot 7.5-year DDKT graft survival
Accept or decline LDKT offer?
No. 52	Retrospective cohort study69	LDKT recipients	Risk of living donor kidney graft failure	C-statistic=0.55	Calibration plot

*Population formulated as reported in identified records.
†Prediction formulated as reported in identified records.
‡Discrimination describes how accurately a tool identifies a high probability of events in patients with the outcome of interest and is expressed as a slope or C-statistic. A C-statistic of 0.5 represents no predictive discrimination and a C-statistic of 1 represents perfect predictive discrimination. When the C-statistic is >0.7, a score has acceptable discriminatory power.
§Calibration describes the agreement between the observed and predicted outcomes and is generally expressed with a calibration plot, a calibration slope, as calibration-in-the-large or a goodness-of-fit test. A calibration plot compares the predicted risks with observed risks within subgroups of patients and provides the most information on calibration accuracy.
CKD, chronic kidney disease; C-statistic, concordance statistic; DDKT, deceased donor kidney transplantation; eGFR, estimated glomerular filtration rate; HD, haemodialysis; LDKT, living donor kidney transplantation; n.a., not applicable; PD, peritoneal dialysis; PTs, prognostic tools.