Do patient-reported measures of symptoms and health status predict mortality in hemodialysis? An assessment of POS-S Renal and EQ-5D.

ABSTRACT

Introduction Experience with the use of patient-reported outcome measures such as EQ-5D and the symptom module of the Palliative care Outcome Scale-Renal Version (POS-S Renal) as mortality prediction tools in hemodialysis is limited. Methods A prospective survival study of people receiving hemodialysis (N = 362). The EQ-5D and the POS-S Renal were used to assess symptom burden and self-rated health (with a self-rated component). Participants were followed from instrument completion to death or study end. Competing risks survival analysis was used to evaluate associations with time to death, with renal transplant as a competing risk. Findings 32% (N = 116) of participants died over a median (25th-75th centile) of 2.6 (1.41-3.38) years. Factors most notably associated with mortality adjusted hazard ratio (95%CI) included lower EQ VAS score 2.7 (1.4, 5.2) P = 0.004 (lowest tertile), higher POS-S Renal score 2.4 (1.3, 4.3) P = 0.004 (highest tertile), and lower EQ-5D score 2.6 (1.3, 5.3) P = 0.01 (lowest tertile) as well as the presence of "problems with mobility?" 2 (1.1, 3.3) P = 0.01, or "problems with usual activities?" 2.1 (1.4, 3.3), P < 0.001. After age adjustment area under the receiver operating curves (AUC) (95%CI) for mortality were 0.71 (0.62, 0.79) for EQ VAS score, 0.71 (0.63, 0.80) for POS-S Renal-S Renal score, and 0.76 (0.68, 0.84) for EQ-5D score. AUC 95%CI was highest for our fourth model at 0.79 (0.72, 0.86) comprised of individual elements from both instruments and established risk factors. Discussion EQ VAS scores and predictive models based on combinations of elements from the POS-S Renal and EQ-5D instruments may aid in mortality discrimination and possibly in the delivery of supportive care services.