A prospective study of risk factors and decision making in discontinuation of dialysis

Discontinuation of dialysis is a common cause of death. We performed a prospective study to determine if a detailed description of patients could be used to identify those at risk for discontinuing dialysis.

Two hundred thirty-five dialysis patients were studied for 3 1/2 years. Using questionnaires, medical chart analyses, and interviews, 300 sociodemographic, quality-of-life, medical and dialysis variables were prospectively obtained. The association of these variables with death from discontinuation of dialysis was studied in univariate and multivariate analyses.

Seventy-six patients (32%) died during the follow-up period, 31 (41%) of cardiovascular problems, 13 (17%) of discontinuation of dialysis, and 32(42%) of other causes. Patients who discontinued dialysis were older (mean age, 66 vs 54 years), were more likely to be divorced or widowed (46% vs 16%), were more likely to live in nursing homes (31% vs 6%), spent less time outdoors (23% vs 53%), had twice as much comorbidity as other patients (major comorbidity score, 1.2 vs 0.6), had the same psychological quality-of-life score (60 vs 62) but a lower score on the Karnovsky Scale, and a higher physical discomfort index. In multivariate analysis, comorbidity, widowed or divorced status, and severe pain were independent predictors of discontinuation of dialysis, but no reliable predictive model could be created. A detailed analysis of the final event indicated that a relentless succession of problems often led nursing home patients to discontinue dialysis. These problems are common to all dialysis patients, especially those who have no one to support them in their suffering.

Discontinuation of dialysis is common, and can been seen as a failure of and a relief from long-term dialysis. We believe mortality caused by discontinuation of dialysis could be decreased by better psychological support and improved management of ischemic problems and pain. We found no predictable model for such patients, but many successive acute problems in widowed and divorced patients seem to trigger the decision.