Predictors of death in patients on peritoneal dialysis: the Missouri Peritoneal Dialysis Study

BACKGROUND: The study was designed to identify predictors of death in subjects on peritoneal dialysis (PD). METHODS: The population consisted of patients initiated on PD at the University of Missouri-Columbia and Dialysis Clinic Incorporated from January 1, 1990, through December 31, 1999. Baseline variables included demographics, clinical data, initial measures of nutritional status, adequacy, and transport characteristics. Co-morbidities were scored using a modified version of the Index of Coexistent Disease. Ongoing (during the course of PD) variables consisted of clinical characteristics and weighted time average of a number of laboratory, adequacy, and nutritional variables. The variables were screened using a univariate procedure, and then analyzed using stepwise logistic regression to evaluate their independent relation to death. RESULTS: There were 105 men and 86 women--180 Caucasians, 10 African-American, 1 Asian, mean age 61 +/- 13 (SD) years, and mean duration of follow-up 21 +/- 18 months. Eighty-two patients suffered the outcome of death. Lean body mass (LBM) at the initiation of PD was negatively associated with the risk of death (p < 0.01). In addition, the need for a partner to perform PD, total morbidity count, and the summated severity score of all co-morbidities were associated with an increased risk of death. The analysis of ongoing variables revealed that serum phosphate (negative association, p = 0.02) and number of hospitalization days per month on PD (p = 0.0006) were associated with an increased risk of death. CONCLUSION: Phosphate levels and LBM are strong negative predictors of death in PD subjects. Further, patients who need the assistance of a partner to perform PD have decreased survival.