The Trial of Ascertaining Individual Preferences for Loved Ones' Role in End-of-Life Decisions (TAILORED) Study: A Randomized Controlled Trial to Improve Surrogate Decision Making

Abstract

Context: Patients with terminal illnesses often require surrogate decision makers. Prior research has demonstrated high surrogate stress, and that despite standards promoting substituted judgment, most patients do not want their surrogates to make pure substituted judgments for them. It is not known how best to help loved ones fulfill the surrogate role.

Objectives: To test the effectiveness of an intervention to help surrogate decision makers.

Methods: One hundred sixty-six patients (41% with amyotrophic lateral sclerosis and 59% with gastrointestinal cancers) and their surrogates at two university medical centers were randomized to an intensive nurse-directed discussion of the end-of-life decision control preferences of the patient (TAILORED) or a discussion of nutrition (CONTROL); 163 completed baseline interviews and underwent the intervention.

Results: Twelve patients died during follow-up and 137 dyads completed the study. Post-intervention, using all available data, TAILORED patients and surrogates became more likely to endorse mutual surrogate decision making, that is, a balance of their own wishes and what the surrogate thinks best (adjusted odds compared with baseline for patients = 1.78, P = 0.04; adjusted odds for surrogates = 2.05, P = 0.03). CONTROL patients became 40% less likely to endorse mutual surrogate decision making (P = 0.08), and CONTROL surrogates did not change significantly from baseline (adjusted odds = 1.44, P = 0.28). Stress levels decreased for TAILORED surrogates (impact of events scale = 23.1 ± 14.6 baseline, 20.8 ± 15.3 f/u, P = 0.046), but not for CONTROL (P = 0.85), and post-intervention stress was lower for TAILORED than CONTROL (P = 0.04). Surrogates' confidence was uniformly high at baseline and did not change. Caregiver burden (Zarit) increased from 12.5 ± 6.5 to 14.7 ± 8.1 for TAILORED (P < 0.01), while not changing for CONTROL, yet satisfaction with involvement in decision making was higher at follow-up for TAILORED than for CONTROL (71% vs. 52%, P = 0.03).

Conclusion: TAILORED patients and surrogates who completed the study adopted a more mutual decision-making style, balancing their own wishes with what the surrogate thinks would be best for them. Surrogates reported less stress and more satisfaction. Confidence was high at baseline and did not change. There was a modest increase in caregiver burden. These findings suggest that interventions like TAILORED might positively impact surrogate decision making.

Trial registration: ClinicalTrials.gov NCT01160367.

Keywords: Ethics; decision-making; end-of-life care; surrogates.

Figure 1

TAILORED Study Subject Recruitment, Participation, and Retention

Figure 2. Baseline distribution of patient Decision Control Preferences

Substituted Judgment = “I prefer that my loved ones tell my doctor which tests or treatments to order for me based on my own personal wishes” Shared/autonomy = I prefer that my loved ones tell my doctor which tests or treatments to order for me based on my own personal wishes after having seriously considered what they think is best” Shared = “I prefer that my loved ones tell my doctor which tests or treatments to order by equally weighing both my personal wishes and what my loved ones think is best” Shared/deferring = “I prefer that my loved ones order tests or treatments for me based on what my loved ones think is best, after having seriously considered my personal wishes” Deferring = “I prefer that my loved ones tell my doctor which tests or treatments to order for me based on what my loved ones think is best”

Figure 3. Changes in Patient Decision Control Preferences from baseline to follow-up at eight weeks

Figures represent the proportion of patients changing their Decision Control Preferences after the intervention towards more autonomy (“I prefer that my loved ones tell my doctor which tests or treatments to order for me based on my own personal wishes”) or towards more deference to the family (“I prefer that my loved ones tell my doctor which tests or treatments to order for me based on what my loved ones think is best”), on the supposition that they have lost consciousness and are unable to speak for themselves. The figure depicts a generalized estimated linear model with binomial distribution and logit link estimated using generalized estimating equations and exchangeable working correlation structure, and adjusting for study site, patient age, gender, importance of religion and surrogate relationship to patient (spouse vs. other) to test the significance of these changes. Compared with baseline, the odds after the intervention of the patient choosing shared decision making (“I prefer that my loved ones tell my doctor which tests or treatments to order for me by equally weighing both personal wishes and what my loved ones think is best”) were 0.59 (p = .08) for the CONTROLS and 1.78 (p=.04) for the TAILORED group.