The study population was drawn from the Providence and University of California Los Angeles Health (UCLA) electronic health records (EHRs) based Center for Kidney Disease Research, Education, and Hope (CURE-CKD) Registry that includes persons with or at risk for CKD from diabetes mellitus, hypertension and prediabetes.⁸ The study was approved by Providence and UCLA Health institutional review boards and conducted according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines for cohort studies. Demographic data were abstracted from the EHR; race categories include only non-Hispanic individuals.

Between 1 January 2013, and 31 December 2020, adults (≥18 years) were identified with prediabetes by the American Diabetes Association definition-based criteria: HbA1c of 5.7%–6.4%, two fasting blood glucose measures of 100–125 mg/dL or two random blood glucose measures of 140–199 mg/dL. Individuals were included in the analysis at the first identification of prediabetes any time during 2013–2020, regardless of the date of entry, if they were tested for CKD and diabetes at baseline and during follow-up. Laboratory evidence of CKD, an administrative CKD code, or kidney failure were exclusionary (Table S1). Baseline demographic and clinical characteristics were obtained during the 6 months following the identification of prediabetes. The index date for follow-up started 6 months after entry into the cohort and ended at the onset of CKD or diabetes, or the last encounter before 31 December2020, whichever came first (Figure S1). Individuals who developed diabetes before CKD were removed from the risk set at the time of diabetes identification. Those included in the study had more frequent hypertension but were otherwise similar to those excluded from the study for missing measures (Table S2).

The primary outcome was CKD onset identified by two estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m² (CKD Epidemiology Collaboration 2021 equation with serum creatinine), urine albumin-to-creatinine ratio (UACR) ≥ 30 mg/g, urine protein-to-creatinine ratio (UPCR) ≥ 0.15 g/g for ≥90 days, or by a new administrative code for CKD. Diabetes was identified by HbA1c ≥ 6.5%, two blood glucose measures (≥126 mg/dL fasting/≥200 mg/dL random), administrative codes or a prescription for blood glucose-lowering medication. Hypertension was identified by two blood pressures ≥140/90 mmHg or administrative codes (Table S1).

Baseline characteristics were reported overall and by hypertension or atherosclerotic CVD (ASCVD) status. Descriptive measures are reported as mean ± standard deviation (SD) for normally distributed variables, median and interquartile range (IQR) for skewed variables or frequencies for categorical variables.

The overall incidence of CKD was calculated as the number of CKD cases/1000 person-years at risk and directly standardized by age (reference: 18–39), race and ethnicity (reference: non-Hispanic [NH] White), and sex (reference: men) to the 2020 U.S. Census Bureau population. Estimates by demography were standardized by the remaining variables, for example, age estimates were standardized by sex, race and ethnicity. CKD incidence was also assessed by hypertension or ASCVD status at cohort entry. Incidence rate ratios (IRRs) and 95% confidence intervals (CIs) were estimated relative to the reference group for each demographic category. The threshold for statistical significance was selected at p < 0.05. Bonferroni correction for demographic categories resulted in significance threshold adjustments for age (p < 0.02) and race and ethnicity (p < 0.01). Incidence was computed using the procedure PROC STDRATE in SAS version 9.4. All other analyses were completed in R version 4.2.2.

The study included 281 933 individuals with prediabetes. Mean ± SD age was 57 ± 15 years, and 55% were women. Baseline HbA1c was 5.9 ± 0.2%, BMI was 30 ± 7 kg/m² and eGFR was 93.3 ± 15.7 mL/min/1.73 m². Persons with hypertension or ASCVD had a similar level of HbA1c but higher BMI and lower eGFR than those without (Table 1).

Over a median follow-up of 2.5 (IQR 1.3–4.2) years, 3.6% (n = 10 104) of persons with prediabetes developed CKD, mostly ascertained by eGFR <60 mL/min/1.73 m² (81.4%, n = 8221; Figure S2). The overall age-, sex- and race and ethnicity-standardized CKD incidence was 10.9 (95% CI 10.5–11.2) cases/1000 person-years. The incidence of CKD was lower for women compared with men (IRR 0.92, [95% CI: 0.87–0.98], p = 0.009). It increased exponentially with age and was higher among Native Hawaiian/Pacific Islander (7.94, [1.66–37.95], p = 0.01) and NH Black (1.56, [1.36–1.78], p < 0.001) populations compared with the NH White population (Figure 1). Incidence was higher in participants with hypertension or ASCVD at baseline, 11.9 (95% CI 11.5–12.3) cases/1000 person-years. In persons without hypertension or ASCVD, CKD incidence was 9.4 (8.9–9.9) cases/1000 person-years (Table S3).