Monday, October 24, 2011: 5:15 PM
Columbus Hall C-F (Hyatt Regency Chicago)
(BEC) Behavioral Economics

Scott D. Halpern, MD, PhD, MBE1, Kevin Volpp, MD, PhD1, George Lowenstein, PhD2, Elizabeth Cooney, MPH1, Tatiana Silva, MPA1, Robert M. Arnold, MD3, Derek C. Angus, MD, MPH, FRCP3 and Cindy L. Bryce, PhD3, (1)University of Pennsylvania School of Medicine, Philadelphia, PA, (2)Carnegie Mellon University, Pittsburgh, PA, (3)University of Pittsburgh School of Medicine, Pittsburgh, PA

Purpose: To examine how default options affect chronically ill patients’ goals of care and elections to receive specific interventions when completing real advance directives (ADs).

Methods: Randomized trial of patients with non-curable lung diseases recruited from pulmonary and oncology clinics.  Patients were assigned with equal probabilities to complete (1) an opt-out AD (modeled on the Allegheny County Medical Society’s advocated form) in which the default goal of care prioritized extending life “even if that means I may have more pain and suffering,” and patients could opt out individually from 5 interventions (e.g., mechanical ventilation); (2) an opt-in AD in which the default goal of care prioritized comfort “even if that means not living as long” and patients could opt into 5 interventions; or (3) a neutral AD in which patients not making active choices effectively were choosing not to specify a plan of care or intervention preference.  

Results: Among 130 patients enrolled, 38 (29%) completed an AD that was signed by their surrogates and incorporated into their medical records. Non-completion rates were similar across the 3 arms (all p > 0.5), and intention-to-treat analyses produced results similar to the per-protocol analyses reported here. Patients completing opt-in ADs (78%) were the most likely to select the comfort-oriented plan of care, followed by patients completing neutral ADs (57%) and opt-out ADs (20%) (p < 0.001 for trend). Patients completing opt-in rather than opt-out AD’s were more likely to choose to forgo ICU admission, dialysis, and feeding tube insertion (all p < 0.05); corresponding but non-significant findings were noted for mechanical ventilation (p = 0.074) and cardiopulmonary resuscitation (p = 0.088). Patients completing neutral ADs had probabilities of forgoing each service that were intermediate between those for patients completing opt-in and opt-out ADs.

Conclusions: Building on prior research in hypothetical settings, this study provides the first randomized evidence that default options influence real healthcare decisions. Future research is needed to identify methods for increasing AD completion and to quantify how altering the choices patients make in ADs influence their receipt of wanted and unwanted healthcare services, costs of care, and satisfaction with care.