Abstract
IMPORTANCE: The optimal dietary protein intake (DPI) for patients with chronic kidney disease (CKD) remains uncertain. Long-term, clinical practice data using objective DPI measures are limited, and concerns about nutritional risk persist. OBJECTIVE: To evaluate the association of objectively measured DPI with long-term kidney and clinical outcomes in adults with CKD stages 3 and 4. DESIGN, SETTING, AND PARTICIPANTS: This retrospective cohort study included adults with CKD stages 3 and 4 receiving care in Clalit Health Services, Israel, with index entry from January 1, 2007, through December 31, 2022, and follow-up of up to 15 years. DPI was assessed using 24-hour urinary nitrogen excretion and normalized to adjusted body weight (normalized DPI [nDPI]). Participants were stratified by an nDPI threshold of 1.0 g/kg/d (selected using time-dependent receiver operating characteristic analysis). Data were analyzed from August to October 2025. EXPOSURE: nDPI calculated from 24-hour urinary nitrogen excretion, analyzed dichotomously using the 1.0 g/kg/d threshold. MAIN OUTCOMES AND MEASURES: The primary outcome was a composite of 50% or more decline in estimated glomerular filtration rate (eGFR), initiation of long-term dialysis, or all-cause mortality. Secondary outcomes included individual components and trajectories of eGFR and albuminuria. Propensity score matching was used to balance baseline characteristics between lower (<1.0 g/kg/d) and higher (≥1.0 g/kg/d) nDPI groups. Kidney function was evaluated using mixed-effects models and joint models linking eGFR trajectories with time-to-event outcomes. RESULTS: Of 1441 included patients (mean [SD] age, 67.20 [11.26] years; 507 [35.2%] women), 530 were matched (265 per group). During follow-up, the lower-nDPI group had a lower risk of the composite outcome (hazard ratio [HR], 0.77; 95% CI, 0.62-0.97; log-rank P = .03), mainly related to fewer dialysis initiations (HR, 0.65; 95% CI, 0.42-0.99). In adjusted Cox models, low nDPI remained associated with lower composite risk (HR, 0.75; 95% CI, 0.60-0.93). Longitudinal models showed no significant between-group differences in eGFR or albuminuria slopes; eGFR decline was numerically slower in the low-nDPI group (slope difference, 0.152 mL/min/1.73 m2/y). No differences in nutritional markers were observed. CONCLUSIONS AND RELEVANCE: In this retrospective cohort study of adults with CKD stages 3 and 4, lower nDPI (<1.0 g/kg/d) was associated with lower dialysis risk without nutritional harm. These findings support moderate protein restriction with routine DPI monitoring in CKD care.