Platelet-to-HDL-Cholesterol Ratio as a Predictor of New-Onset Diabetes in Rheumatoid Arthritis: A Prospective Cohort Study

Abstract Objectives Patients with rheumatoid arthritis (RA) are at an increased risk of new-onset diabetes due to chronic inflammation and metabolic dysregulation. The platelet-to-high-density lipoprotein cholesterol ratio (PHR), which integrates thrombotic propensity and lipid metabolism, may serve as a novel biomarker for predicting diabetes, yet its utility in RA populations remains uninvestigated. Methods This national prospective cohort study included 3,029 patients with rheumatoid arthritis (RA) aged ≥45 years without baseline diabetes from the China Health and Retirement Longitudinal Study (CHARLS). Participants were categorized based on their diabetes status during follow-up. Multivariable logistic regression models were used to assess the association between PHR and diabetes, while restricted cubic splines (RCS) were employed to explore non-linear relationships. Subgroup analyses were performed to examine effect modifications by age, body mass index (BMI), and gender. Random forest (RF) algorithms were utilized to evaluate the predictive importance of PHR, and receiver operating characteristic (ROC) curves were used to assess model performance, with a 70% training/30% validation data split. Results A total of 534 new-onset diabetes cases were identified in our study. Elevated PHR independently predicted the risk of diabetes: each 5-unit increase in PHR was associated with a 5% higher adjusted odds ratio (OR=1.05, 95% CI: 1.02–1.08). Restricted cubic spline (RCS) analysis revealed a threshold effect at PHR=0.48: exceeding this threshold significantly increased the risk (OR=1.12 per unit increase, 95% CI: 1.01–2.16, p=0.01), whereas values below this threshold showed no significant association (OR=0.78, p=0.54). Subgroup analyses revealed greater susceptibility, particularly in individuals aged ≥60 years (OR=1.07, p=0.008) and those with BMI≥30 kg/m² (OR=1.24, p=0.01). Random forest (RF) analysis confirmed the discriminative ability of PHR, with area under the curve (AUC) values of 0.77 (training set) and 0.62 (validation set). Conclusions In middle-aged and older patients with rheumatoid arthritis (RA), the platelet-to-HDL-C ratio (PHR) exhibits a non-linear, threshold-dependent association with new-onset diabetes. These findings support the integration of PHR into metabolic surveillance protocols for RA patients to guide targeted diabetes prevention strategies.