Myth of substituted judgment. Surrogate decision making regarding life support is unreliable

OBJECTIVE: To identify factors predicting the accuracy of surrogate decision making in life support decisions. DESIGN: Questionnaire. SETTING: Urban Veterans Affairs hospital. PATIENTS AND DESIGN: Fifty hospitalized patients and their chosen surrogates were given questionnaires describing life support modalities and four common medical scenarios in which life support would be contemplated. An additional 50 patients also completed the questionnaire. Patients gave their choices of life support in the different scenarios. Surrogates guessed the patients' answers (substituted judgment). Details of the patient-surrogate relationship were asked. Patients completed a depression inventory. MAIN RESULTS: Surrogates correctly guessed patients' wishes about life support overall on 59.3% of the questions, not better than random chance (kappa = .09). The only predictor of accurate surrogate decision making was specific discussion between patient and surrogate about life support. SECONDARY RESULTS: Patients had an overall low desire for life support (35%), and a majority favored euthanasia under some circumstances (62%). There was no relationship between depression score and desire for life support. CONCLUSIONS: Substituted judgment by surrogates is not more accurate than random chance. Discussion between patient and surrogate about life support correlated with more accurate substituted judgment.