Kidney-Related Screening Indicator Trends in Korea: 2010–2024 (eGFR, 2012–2024) and 2025–2028 Forecasts Using National Health Insurance Service Health Screening Data

Background and Objectives: Chronic kidney disease (CKD) risk classification depends on reduced estimated glomerular filtration rate (eGFR) and kidney-damage markers such as albuminuria or proteinuria. National screening statistics can identify population-level signals—referred to here as screening-based kidney risk indicators and abnormal screening prevalence—that serve as proxy indicators rather than confirmed CKD diagnoses, and require confirmatory testing when abnormal. Materials and Methods: This retrospective repeated cross-sectional ecological trend study analyzed National Health Insurance Service (NHIS) health-screening aggregate tables distributed through the Korean Statistical Information Service (KOSIS), representing individuals who underwent NHIS health screening rather than all insured individuals. Age- and region-stratified tables for eGFR, serum creatinine, urine protein, triglycerides, and hemoglobin were used. Non-eGFR indicators covered 2010–2024 and eGFR covered 2012–2024. Operational endpoints were eGFR < 60 mL/min/1.73 m2, serum creatinine ≥ 1.5 mg/dL, dipstick proteinuria ≥1+, triglycerides ≥ 150 mg/dL, and sex-specific anemia proxy; triglycerides and the anemia proxy were treated as associated metabolic/hematological indicators rather than kidney-specific endpoints. Crude national trends were supplemented with direct age-standardization (to the 2024 screening population) for eGFR, creatinine, proteinuria, and triglycerides, and with sex-stratified trends for all indicators. Binomial logit generalized linear models with calendar year as the predictor generated 2025–2028 forecasts; forecast reliability was assessed by temporal hold-out validation (fitting through 2020, predicting 2021–2024) and a recent-window (2018–2024) sensitivity analysis, and 95% bootstrap prediction intervals were derived by resampling model residuals and parameter uncertainty. Regional ecological correlations were assessed by Pearson and Spearman methods. Results: Dipstick proteinuria ≥1+ increased from 2.20% (age-standardized 2.34%) in 2012 to 3.71% in 2024 and was forecast to reach 4.64% (95% bootstrap prediction interval [PI], 4.50–4.83%) in 2028. eGFR < 60 decreased from 4.13% (age-standardized 5.20%) in 2012 to 3.43% in 2024, with a 2028 forecast of 3.06% (95% PI, 2.88–3.27%); the age-standardized decline was steeper than the crude decline, indicating the finding is not an artifact of population aging. Hold-out validation supported the eGFR, creatinine, and proteinuria forecasts (mean absolute error 0.05–0.14 percentage points) but showed poor fit for triglycerides and the anemia proxy, whose forecasts are reported as exploratory only. In 2024, eGFR < 60 reached 41.61% among persons aged ≥ 85 years. Jeonnam had the highest crude regional eGFR < 60 prevalence (5.09%), though regional comparisons could not be age-standardized. Regional eGFR < 60 was strongly correlated with elevated serum creatinine (Pearson r = 0.954, p < 0.001), reflecting their mathematical dependence rather than independent validation, and moderately correlated with the anemia proxy (Pearson r = 0.611, p = 0.009). Conclusions: NHIS/KOSIS screening data showed an age-standardization-robust upward proteinuria signal and persistent age-concentrated kidney function abnormalities. The findings support confirmatory testing after abnormal screening results and targeted kidney-risk follow-up, but they should not be interpreted as confirmed CKD incidence or prevalence.