A 12-year longitudinal study reveals nonlinear associations between dietary intake and outcomes in peritoneal dialysis patients
image: Researchers collected long-term high-quality data to address dietary nutrition management issues in ESRD patients and developed methods to explore the nonlinear relationship between dietary nutrition and mortality risk.
Credit: [Yueying Wu, Peking University]
End-stage renal disease (ESRD) poses a substantial burden on patients’ quality of life and healthcare systems worldwide. A collaborative research team from Peking University, Peking University Third Hospital, Xuzhou Municipal Hospital affiliated with Xuzhou Medical University, Peking University People’s Hospital, Tsinghua University, and the University of Edinburgh has published a new study in Health Data Science, systematically examining the relationship between dietary intake and clinical outcomes in peritoneal dialysis patients. The study analyzed real-world longitudinal data from 656 patients over more than 12 years, with follow-ups conducted every three months on average. By integrating expert-supervised 3-day dietary records with high-resolution electronic health records, the researchers explored how different levels of nutrient intake are associated with mortality risk.
Dietary intervention has long been considered a cornerstone of ESRD management. However, high-quality evidence directly linking dietary intake to hard clinical outcomes, such as mortality, remains limited. This challenge is particularly pronounced in peritoneal dialysis populations, where rapid disease progression, complex comorbidities, and fluctuating dietary behaviors complicate long-term, high-precision data collection. Moreover, the relationship between diet and outcomes is not necessarily linear—both insufficient and excessive intake may carry risks—making it difficult for conventional analytical approaches to generate actionable nutritional recommendations.
To address these challenges, the research team developed a two-stage analytical framework. In the first stage, multivariable Cox proportional hazards models were used to evaluate mortality risk while adjusting for potential confounders. In the second stage, restricted cubic spline models were applied to identify nonlinear associations between nutrient intake and mortality, allowing the estimation of intake ranges associated with lower risk. Patients were further stratified by baseline serum albumin levels, a key indicator of nutritional status and prognosis, enabling a more refined assessment of how dietary patterns interact with clinical conditions.
The study found that among 26 commonly analyzed nutrients, 14 currently lack clear clinical guidelines for ESRD or peritoneal dialysis populations, and 13 were significantly associated with mortality risk. Notably, approximately 69% of the diet–outcome relationships exhibited nonlinear patterns. This finding underscores that nutritional management should move beyond simplistic “more or less” approaches and instead focus on optimal intake ranges. For example, while existing guidelines typically recommend protein intake of 1.0–1.2 g/kg/day, the study identified a range of 0.88–1.13 g/kg/day associated with better outcomes. Similarly, energy intake recommendations were refined from the conventional 25–35 kcal/kg/day to a broader range of 26–42 kcal/kg/day.
The analysis also revealed that in patients with higher serum albumin levels, a greater number of nutrients showed sensitivity to mortality risk, whereas in patients with lower or moderate albumin levels, fewer key nutrients were strongly associated with outcomes. These findings highlight the importance of tailoring nutritional strategies to individual patient conditions, rather than applying uniform dietary recommendations.
This study helps address a critical gap in understanding the link between dietary intake and long-term outcomes in ESRD patients. Beyond providing high-quality, long-term, fine-grained data, it introduces a methodological framework adaptable to complex clinical scenarios, offering a foundation for more precise nutritional interventions.
Looking ahead, the research team plans to expand the study to multi-center and multi-regional datasets to validate and refine the model across broader populations. Future work will also explore advanced approaches, such as Bayesian inference and causal analysis, to further improve the robustness and clinical applicability of dietary recommendations. As real-world data and precision medicine continue to evolve, this research is expected to contribute to more individualized nutritional management strategies and improved long-term care for patients with chronic diseases.