Abstract Background Symptom burden in chronic kidney disease (CKD) patients has a negative impact on functional status and quality of life. Despite the high prevalence among hemodialysis (HD) patients, symptoms are often underestimated. We investigated the determinants of symptoms and their impact on mortality in a large cohort of HD patients. Methods We analyzed 1825 HD patients for whom at least one questionnaire about symptoms was available. Machine learning-based algorithms were used to identify longitudinal variables associated with symptoms. Univariate and multivariate Cox regression analyses were used to investigate the association between symptoms and mortality. The additional prognostic value of symptoms was explored by using C-statistics and machine learning classification analyses. Results Appetite Rating, Non-vascular Nervous System Score and Difficulty Following Diet were the variables most associated with symptoms, both at baseline and in longitudinal analyses. Survival analyses showed an independent association between symptoms and all the considered outcomes (Death, Hazard Ratio HRfper 10-unit increase: 0.91, 95% confidence interval CI 0.87–0.96, P 0.001; Cardiovascular hospitalizations, HRper 10-unit increase: 0.89, 95% CI 0.84–0.94, P 0.001; infectious hospitalizations, HRper 10-unit increase: 0.91, 95% CI 0.87–0.96, P 0.001). However, classification analyses performed by machine learning showed that adding a symptom score to a base model did not significantly improve the performance of the models. Similarly, the improvement in R2, in discrimination power, and in reclassification capability was almost null. Discussion In the large dataset of the Hemodialysis (HEMO) trial, we found that, among the analyzed variables, those related to eating habits and previous comorbidities were the most closely associated with the symptoms score, both at baseline and during follow-up. Although symptoms were associated with severe clinical outcomes, their prognostic power was limited. However, considering symptom burden in dialysis patients, strict monitoring of eating habits may help improve quality of life in these patients.
Mezzatesta et al. (Tue,) studied this question.