Predictors of Better Shared Decision Making in Goals-Of-Care Clinician-Family Meetings for Critically Ill Neurologic Patients: A Multi-Center Mixed Methods Pilot Study
Victoria Fleming1, Abhinav Prasad1, Connie Ge1, Shazeb Meraj1, Carmalin Franco1, Catherine Hough2, Bernard Lo3, Shannon Carson4, Jay Steingrub5, Douglas White6, Susanne Muehlschlegel1
1UMASS Medical School/UMASS Memorial Health Care, 2OHSU, 3Univ of CA - School of Med, 4University of North Carolina Hospitals, 5University of Massachusetts Medical School - Baystate, 6University of Pittsburgh Medical Center
Objective:
- To characterize and quantify SDM in GOC meetings
- To identify factors associated with higher SDM score
Background:
Shared decision making (SDM) is recommended by experts to improve difficult decisions in ICUs; yet for the most difficult decision of all, the goals-of-care (GOC) decision, empirical research in medical-surgical ICUs demonstrated that only 2% of clinician-family meetings contained all SDM elements. Little is known about the extent and predictors of SDM in critically-ill neurological patients (CINPs).
Design/Methods:
Two qualitative coders applied a validated 10-element SDM instrument to 63 GOC clinician-family meetings for CINPs, audio-recorded at seven U.S. medical centers. Clinicians and families completed post-meeting questionnaires assessing their own prognostic estimates. We defined the Clinician-Family-Prognostic-Discordance-Score as the difference between both estimates and prognostic discordance as ≥20% difference. We applied univariate and multivariable longitudinal regression to identify predictors of more SDM.
Results:
The median SDM score (total number of unique SDM elements per meeting) was 7 (range 1-10; IQR 5-8). Only 6% of meetings contained all 10 SDM elements. The most common SDM elements were “discussing uncertainty”(89%) and “assessing family understanding”(86%); the least frequent elements were “assessing the need for input from others”(40%) and “eliciting the context of the decision”(32%). Prognostic discordance was 60% and 45% for hospital-survival and 6-month independent functioning, respectively. In univariate analyses, more SDM was associated with female clinicians, and Cohort 2. Meetings with less clinician-surrogate survival discordance approached significance. After adjustment, there was a trend towards clinician gender being an independent predictor of more SDM(p=0.11), but none of the variables were independent predictors.
Conclusions:
Few clinician-family GOC-meetings for CINPs contained all SDM elements. We uncovered modifiable gaps as opportunities for improvement. While our sample was too small to find independent predictors of SDM, our data supports the urgent need to continue our mixed-method of examination of clinician-family GOC-meetings in larger cohorts.