BACKGROUND: The prognostic relevance of health-related quality of life (HRQoL) has been scarcely studied in the dialysis population and the prognostic power for mortality of the HRQoL domains is unknown. METHODS: We tested the prognostic value for mortality of the HRQoL domains included in the 36-item Short Form Health Survey (SF-36) by Cox's regression analysis and by state-of-the-art prognostic techniques {explained variation in mortality [R2], calibration, discrimination [Harrell's C], risk reclassification [Net Reclassification Index (NRI)], Integrated Discrimination Index [IDI]} in a cohort of 951 patients on chronic haemodialysis. RESULTS: In multivariable Cox models, all but two domains (role limitation due to physical health and due to emotional problems) were independently related with mortality. Physical functioning was the domain adding the highest explanatory power (R2+5.3%) to a basic model including established risk factors for mortality in the dialysis population. The same domain improved risk calibration and showed the highest Harrell's C (+1.7%) and the highest reclassification power (categorical NRI + 8.7%, continuous NRI +46%, P ≤ 0.006) and the highest IDI (+3.4%, P < 0.001). These results were fully confirmed in analyses testing the additional prognostic power of SF-36 domains when combined to a well-validated risk score in dialysis patients. CONCLUSIONS: Physical functioning holds the highest prognostic power for mortality among the domains of SF-36. The gain in prognostic ability by this domain is relevant for clinical practice. Physical functioning has the potential for refining the prognosis and for informing exercise programmes in the dialysis population.
BACKGROUND: The prognostic relevance of health-related quality of life (HRQoL) has been scarcely studied in the dialysis population and the prognostic power for mortality of the HRQoL domains is unknown. METHODS: We tested the prognostic value for mortality of the HRQoL domains included in the 36-item Short Form Health Survey (SF-36) by Cox's regression analysis and by state-of-the-art prognostic techniques {explained variation in mortality [R2], calibration, discrimination [Harrell's C], risk reclassification [Net Reclassification Index (NRI)], Integrated Discrimination Index [IDI]} in a cohort of 951 patients on chronic haemodialysis. RESULTS: In multivariable Cox models, all but two domains (role limitation due to physical health and due to emotional problems) were independently related with mortality. Physical functioning was the domain adding the highest explanatory power (R2+5.3%) to a basic model including established risk factors for mortality in the dialysis population. The same domain improved risk calibration and showed the highest Harrell's C (+1.7%) and the highest reclassification power (categorical NRI + 8.7%, continuous NRI +46%, P ≤ 0.006) and the highest IDI (+3.4%, P < 0.001). These results were fully confirmed in analyses testing the additional prognostic power of SF-36 domains when combined to a well-validated risk score in dialysis patients. CONCLUSIONS: Physical functioning holds the highest prognostic power for mortality among the domains of SF-36. The gain in prognostic ability by this domain is relevant for clinical practice. Physical functioning has the potential for refining the prognosis and for informing exercise programmes in the dialysis population.