AIM: Risk profile for incident chronic kidney disease (CKD) in Japanese subjects has not been established. Our aim was to identify risk factors for CKD in Japanese. METHODS: Consecutive 171 536 health examinees (median age 49 years and estimated glomerular filtration rate (eGFR) 78.2 mL/min per 1.73 m2 ) without CKD were re-examined after a median period of 6.2 years. Results of Cox proportional hazards models in randomly assigned two thirds (Derivation cohort) were verified in the rest (Validation cohort). CKD was defined as eGFR <60 mL/min per 1.73 m2 or positive dipstick proteinuria. RESULTS: In the Derivation cohort, CKD developed in 1002 (5.8%) subjects. Seven variables such as lower eGFR, male gender, higher uric acid concentration, lower red cell count and higher age and systolic blood pressure were identified as significant risks for CKD, with lowered eGFR being an overwhelmingly strong risk: adjusted hazard ratio for those with the baseline eGFR <70 mL/min per 1.73 m2 was as high as 90.1. Performance of prediction of CKD by the probability on the basis of the seven risk factors combined was only marginally preferable to eGFR alone. The area under the receiver operating characteristic curve (95% CI) for the prediction was 0.846 (0.826-0.864) and 0.822 (0.802-0.840) (P < 0.01), the kappa statistic was 0.263 and 0.250 (n.s.), and the mean absolute difference between "predicted probability" and "observed" CKD was 1.4% and 1.9% (P = 0.14) by the combined model and eGFR alone, respectively. CONCLUSION: Seven risk factors for incident CKD were identified in Japanese health examinees. However, lowered baseline eGFR outweighed other risks to the degree that eGFR alone was suffice for CKD prediction.
AIM: Risk profile for incident chronic kidney disease (CKD) in Japanese subjects has not been established. Our aim was to identify risk factors for CKD in Japanese. METHODS: Consecutive 171 536 health examinees (median age 49 years and estimated glomerular filtration rate (eGFR) 78.2 mL/min per 1.73 m2 ) without CKD were re-examined after a median period of 6.2 years. Results of Cox proportional hazards models in randomly assigned two thirds (Derivation cohort) were verified in the rest (Validation cohort). CKD was defined as eGFR <60 mL/min per 1.73 m2 or positive dipstick proteinuria. RESULTS: In the Derivation cohort, CKD developed in 1002 (5.8%) subjects. Seven variables such as lower eGFR, male gender, higher uric acid concentration, lower red cell count and higher age and systolic blood pressure were identified as significant risks for CKD, with lowered eGFR being an overwhelmingly strong risk: adjusted hazard ratio for those with the baseline eGFR <70 mL/min per 1.73 m2 was as high as 90.1. Performance of prediction of CKD by the probability on the basis of the seven risk factors combined was only marginally preferable to eGFR alone. The area under the receiver operating characteristic curve (95% CI) for the prediction was 0.846 (0.826-0.864) and 0.822 (0.802-0.840) (P < 0.01), the kappa statistic was 0.263 and 0.250 (n.s.), and the mean absolute difference between "predicted probability" and "observed" CKD was 1.4% and 1.9% (P = 0.14) by the combined model and eGFR alone, respectively. CONCLUSION: Seven risk factors for incident CKD were identified in Japanese health examinees. However, lowered baseline eGFR outweighed other risks to the degree that eGFR alone was suffice for CKD prediction.