A selection adjusted comparison of hospitalization on continuous ambulatory peritoneal dialysis and haemodialysis

Abstract

A retrospective study was made of the rate of hospitalization (days/annum) in 227 patients established on either continuous ambulatory peritoneal dialysis (CAPD) or haemodialysis (HD) between 1 January 1980 and 31 July 1985. In order to address the biases which may confound simple comparisons of morbidity when treatment allocation is non-random, multiple regression was used to identify and adjust for the effect of pre-treatment variables significantly (p < 0.01) influencing hospitalization rate. Six significant variables were identified out of 86 studied. Adverse factors were (i) diabetes and (ii) atherosclerotic disease. Beneficial parameters were (iii) a living spouse and (iv) a controlled presentation via outpatients. The relationship (v) between age and hospitalization rate was U-shaped; the rate increasing above the age of 60 years and below the age of 20. (vi) Date of commencement of dialysis exhibited a complex relationship with hospitalization rate implying that whilst the rate on HD remained relatively stable over the course of the study, that on CAPD markedly diminished. As a result, the hospitalization ratio (CAPD: HD) fell consistently throughout the study period, at an estimated annual rate of 28.3% (95% confidence limits 15.6–39.1%). For a patient starting therapy on 1 January 1980 the predicted hospitalization rate on CAPD was 7.8 times greater than that on HD (4.4–13.9) but by 1 May 1985 the estimated ratio had fallen to 1.33 (0.85–2.09). Having adjusted for important confounding, it would appear that with growing experience, hospitalization rates on CAPD fall close to those on HD.