Key points are not available for this paper at this time.
PURPOSE: To assess the accuracy of surrogate decision making for elective cataract surgery. SETTING: Comprehensive Ophthalmology Department of Tertiary Care Hospital. METHODS: Decisions regarding elective cataract surgery of currently competent, elderly patients were compared with the predictions of patient-identified surrogate decision makers in scenarios of current state of mental health and progressive dementia. Patients were identified (age >50 years, Mini-Mental Status Score >20, absence of significant noncataract pathology) consecutively at a scheduled clinic visit to a single provider. Preferences for cataract surgery in the current state of health and hypothetical progressive dementia were assessed on a Likert scale. The same interviewer contacted patient-identified surrogate decision makers within 48 hours. Independently, a survey of community ophthalmologists was performed. The concordance of surrogate predictions with patient preferences (32 pairs) was assessed using percentage agreement, the kappa coefficient with dichotomous Likert scale data, and chi-square analyses (concordance beyond chance). RESULTS: In their current state of health, most patients said they would prefer to have surgery if they were deemed to have a visually significant cataract, which was accurately predicted by surrogates (72% agreement, kappa = 0.65, chi square = 16.5). In a hypothetical dementia scenario, contrary to perceptions of patients and their surrogates, proxies were unable to accurately represent a patient's wishes for elective cataract surgery (34% agreement, kappa = 0.23, chi square = 14.4). Physicians tended to withhold intervention in the dementia scenario. CONCLUSIONS: The findings suggest that current methods of decision making in elective surgery for patients unable to make autonomous decisions may be seriously flawed. This extends previous findings of inaccuracy with the substituted judgment approach to end-of-life issues to reflect elective surgical scenarios.
Mantravadi et al. (Sat,) studied this question.