Trends in serum albumin concentration over time provide a better prediction of clinical outcome in CAPD patients than a single mean value. This was a retrospective review of outcome at 36 months in 225 adult CAPD patients. Mean serum albumin was determined for the first (SA1) and second (SA2) 6 months of treatment and patients grouped according to SA1 (group I, > 37; group II, 34-37; group III, < 34 g/l) and according to the change in serum albumin (delta SA) between the first and second 6 months (increased/static or decreased). Patient (PS) and technique (TS) survival were determined by Kaplan-Meier survival analysis. The effect of SA1 and delta SA on survival were determined in a multivariate Cox regression analysis model that included age and presence or absence of a systemic disease. By SA1 group, PS and TS survival at 36 months were 94 and 76% (group I), 64 and 53% (group II) and 70 and 52% (group III). If delta SA increased/remained static, then SA1 did not predict PS (group I, 100%; group II, 96%; group III, 74%; P = n.s.) or TS (group I, 72%; group II, 63%; group III, 65%; P = n.s.). If delta SA decreased, PS was worse in groups II and III, both as compared to group I (PS group I, 88%; group II, 52%; group III, 34%; P = 0.02) and as compared to the groups II and III when delta SA increased (PS group II, 74 vs 52%, P = 0.05; group III, 82 vs 34%, P = 0.005) The same trend was seen for TS. In the multivariate Cox regression model, age, direction of change in serum albumin, and presence of a multisystem disease were significant predictors of survival, whereas SA1 was not. Early hypoalbuminaemia in CAPD only predicts a worse patient and technique survival if mean serum albumin decreases further from the first to second 6 months of dialysis therapy. Change in serum albumin between the first and second 6 months of CAPD and the mean serum albumin over the first 6 months together offer better discrimination of outcome than either alone.
