Multidimensional Nutritional Phenotypes and Anemia/Metabolic Complications in Chronic Kidney Disease: Protocol for an EHR-Based Retrospective Baseline Cross-Sectional Study with a Prospective Continuous-Enrollment Extension

Background: In Chronic Kidney Disease (CKD), malnutrition and inflammation often coexist and are intertwined. Reliance on single indicators fails to capture the heterogeneity of the "nutrition-inflammation" complex, thereby limiting the interpretability and clinical translatability of risk stratification for anemia and metabolic complications. Objective: To construct multidimensional nutritional phenotypes in patients with CKD, evaluate their associations with anemia and metabolic complications, and further explore the stratifying and modifying roles of inflammation within these associations. Study Design: A baseline cross-sectional analysis based on a retrospective medical record review, supplemented by a prospective consecutive enrollment extension (restricted to baseline measurements without longitudinal follow-up outcomes). Setting and Participants: This study was conducted at a single-center tertiary care hospital, enrolling patients aged ≥18 years who met the KDIGO diagnostic criteria for CKD. Exposures (Phenotype Construction): Anthropometric measurements (BMI, waist circumference, body fat), protein/nutritional reserves (albumin, prealbumin, etc.), micronutrients (e.g., 25-hydroxyvitamin D, ferritin), and inflammatory markers (e.g., CRP or NLR) were integrated. Bayesian latent profile analysis (LPA)/mixture modeling was employed to identify nutritional phenotypes, with class uncertainty addressed using posterior probabilities. Outcomes: The primary outcome was anemia (defined by sex-specific thresholds). Secondary outcomes included metabolic complications (abnormal glucose metabolism, dyslipidemia) and their composite endpoints. Statistical Analysis: Within a Bayesian framework (independent of P-values), Bayesian regression models were utilized to estimate the associations between nutritional phenotypes and outcomes (reporting posterior means/medians, 95% credible intervals [CrI], and probability of direction). The modifying effects of inflammation were evaluated through stratification and interaction terms. Missing data were handled via multiple imputation, and sensitivity analyses were conducted (varying prior strengths, number of phenotypes, and excluding end-stage renal disease or acute infections). Ethics and Dissemination: This study utilized de-identified data with a waiver of informed consent. Findings will be disseminated through peer-reviewed publications and academic conferences. Reproducible analysis code and a data dictionary will be made available, subject to privacy preservation. Discussion: By characterizing the heterogeneity of the nutrition-inflammation complex in the CKD population through multidimensional nutritional phenotypes, this study provides an interpretable evidentiary basis for risk stratification of anemia and metabolic complications. Furthermore, it offers a scalable data interface for future mechanistic validation and prospective follow-up studies.