Is malnutrition an independent predictor of mortality in peritoneal dialysis patients?

It has been established that malnutrition (MN) is a strong predictor of mortality in peritoneal dialysis (PD) patients. However, MN is often the consequence of co-morbid diseases (CMD), and the confounding effect of CMD on mortality in malnourished PD patients has not been clearly defined. In this study, we tested the hypothesis that MN without CMD may not be associated with significant mortality. This study was, therefore, designed to dissociate the influence of CMD on mortality in PD patients from that of MN.

A total of 153 consecutive PD patients (88 males, mean age 53.3 +/- 12.3 years) were included in this study. All underwent initial assessment of nutrition, CMD survey and peritoneal equilibration test at a mean of 7 days (range 3-24 days) after beginning PD. Nutritional status was assessed by subjective global assessment (SGA) and other methods. CMD surveyed included diabetes, cardiovascular disease, liver disease and respiratory disease, and co-morbidity was graded by Davies index. Based on the nutritional status as assessed by SGA and presence of CMD, patients were divided into four groups; MN with (n = 50) or without (n = 14) CMD, and normal nutrition (NN) with (n = 53) or without (n = 36) CMD.

Of 153 patients, 64 (41.8%) were malnourished and 103 (67.3%) had one or more CMD. Of the 103 patients with CMD, 48.5% had MN, and 78% of the 64 patients with MN had CMD. Patients with MN and CMD were older and had lower initial serum albumin (sAlb), serum creatinine, fat-free oedema-free body mass, percentage lean body mass and SGA score and higher initial dialysate/plasma creatinine concentration ratio at 4 h dwell (D4/P4 Cr) and co-morbidity score. On Kaplan-Meier analysis, 2-year patient survival was significantly lower in patients with MN and CMD than in the other groups (63.1, 90.9, 87.5 and 96.4% for subgroups with both MN and CMD, MN without CMD, NN with CMD and NN without CMD, respectively, P = 0.001). On Cox proportional hazards analysis, age, co-morbidity score and D4/P4 Cr, but not SGA score or sAlb concentration, were found to be independent risk factors for mortality. After adjustment for age, gender, sAlb, residual renal function and D4/P4 Cr, patients with both MN and CMD had a risk of mortality that was 3.3 times that of patients with MN but without CMD (risk ratio 9.01 vs 2.72). Patients with MN without CMD had a risk ratio of 2.72 compared with NN without CMD, but this difference was not statistically significant. In patients with NN and CMD, the risk ratio for mortality was five times that of patients with NN without CMD.

This study demonstrates that there is a high prevalence of MN and CMD at the start of PD and that the combined presence of CMD and MN is associated with high mortality. MN alone is associated with a statistically insignificant increase in mortality. This underlines the importance of CMD as a cause of poor clinical outcome in malnourished PD patients. However, in the present study, a relatively limited number of patients with MN but without CMD were analysed and a type two error therefore cannot be excluded.