AIM: Most of the existing risk scores for identifying people with reduced estimated glomerular filtration rate (eGFR) involve laboratory-based factors, which are not convenient and cost-effective to use in a large population-based screening programme. We aimed at using non-invasive variables to identify subjects with reduced eGFR in a Chinese working population. METHODS: Two study populations were recruited in 2012 and 2015, respectively. The 2012 study population (n = 14 374) was randomly separated as the training dataset (n = 9621) or the internal testing dataset (n = 4753) at a ratio of 2:1, and the 2015 study population (n = 4371) was used as the external testing dataset. Stepwise logistic regression analysis with age, gender, hypertension and body mass index (BMI) status were first performed in the training dataset and then validated in both internal and external testing dataset. A nomogram was further developed based on the final model. RESULTS: Results showed that older females with higher BMI status were more likely to have reduced eGFR. The model had excellent discrimination (AUC: 0.887 [95%CI: 0.865, 0.909] in the internal validation and 0.880 [95%CI: 0.829, 0.931] in the external validation) and calibration (Hosmer-Lemeshow test, P = 0.798 and 0.397 for internal and external dataset, respectively). The probability of having reduced eGFR increased gradually from <0.1% at a total score of 0 to 26% at a total score of 58 shown in the nomogram. CONCLUSION: Non-invasive variables could help identify individuals at high risk of reduced eGFR for further kidney function testing or intervention, aiding in decision-making and resource allocation in large population screening.
AIM: Most of the existing risk scores for identifying people with reduced estimated glomerular filtration rate (eGFR) involve laboratory-based factors, which are not convenient and cost-effective to use in a large population-based screening programme. We aimed at using non-invasive variables to identify subjects with reduced eGFR in a Chinese working population. METHODS: Two study populations were recruited in 2012 and 2015, respectively. The 2012 study population (n = 14 374) was randomly separated as the training dataset (n = 9621) or the internal testing dataset (n = 4753) at a ratio of 2:1, and the 2015 study population (n = 4371) was used as the external testing dataset. Stepwise logistic regression analysis with age, gender, hypertension and body mass index (BMI) status were first performed in the training dataset and then validated in both internal and external testing dataset. A nomogram was further developed based on the final model. RESULTS: Results showed that older females with higher BMI status were more likely to have reduced eGFR. The model had excellent discrimination (AUC: 0.887 [95%CI: 0.865, 0.909] in the internal validation and 0.880 [95%CI: 0.829, 0.931] in the external validation) and calibration (Hosmer-Lemeshow test, P = 0.798 and 0.397 for internal and external dataset, respectively). The probability of having reduced eGFR increased gradually from <0.1% at a total score of 0 to 26% at a total score of 58 shown in the nomogram. CONCLUSION: Non-invasive variables could help identify individuals at high risk of reduced eGFR for further kidney function testing or intervention, aiding in decision-making and resource allocation in large population screening.