The impact of a web-based prognostic intervention on physicians’ prognostic confidence.

Abstract

12106 Background: Clinicians often hesitate to discuss prognosis with patients because of prognostic uncertainty. The use of validated prognostic models may enhance prognostic confidence and/or prognostic accuracy. Prognostic confidence is a novel concept that has not been well studied and may support prognosis-based decision making. We examined the impact of a web-based prognostic intervention on physicians’ prognostic confidence. Methods: In this prospective study, palliative care specialists estimated the prognosis of patients with advanced cancer seen at an outpatient supportive care clinic using the temporal, surprise and probabilistic questions for 6 m, 3 m, 2 m, 1 m, 2 w, 1 w and 3 d survival. They then reviewed information from a web-based prognostic calculator ( www.predictsurvival.com ) that provided survival predictions from 7 validated prognostic scores, including the Palliative Prognostic Score, Palliative Prognostic Index, and Palliative Performance Status. The clinicians then provided their prognostic estimates post-intervention. The primary outcome was prognostic confidence (0-10 numeric rating scale, where 0 = not at all, 10 = most confident) before vs. after the study intervention. Secondary outcomes included (1) confidence to share the prognosis with patients, (2) confidence to make prognosis-based care recommendations (agreement = strongly agree or agree) and (3) prognostic accuracy. With 220 patients, we had 80% power to detect an effect size of 0.66 with 2-sided α 0.05. We compared the pre-post data using the Wilcoxon signed-rank test for the primary outcome and McNemar test for secondary outcomes. Results: 216 patients with advanced cancer (mean age 61, 50% female) were included and 154 (71%) died. The median (IQR) actual survival was 90 (39, 178) days; the median (IQR) predicted survival before and after intervention were 90 (60, 90) and 80 (60, 90) days, respectively. Prognostic confidence significantly increased after the intervention (pre vs. post: median 6 vs. 7, P < 0.001). A significantly greater proportion of clinicians reported that they felt confident enough about their prognostic estimate to share it with patients (44% vs. 74%, P < 0.001) and to formulate care recommendations (80% vs. 94%, P < 0.001) after the intervention. Prognostic accuracy did not differ significantly before and after the intervention, ranging from 72-100% for the temporal question, 45-97% for the surprise questions and 38%-100% for the probabilistic questions (P > 0.05). Conclusions: Among patients with advanced cancer seen at a supportive care clinic, the web-based prognostic intervention was associated with greater prognostic confidence and willingness to discuss prognosis, despite not significantly altering clinicians’ prognostic estimate or prognostic accuracy. Further research is needed to examine how prognostic tools may be able to augment prognostic discussions and clinical decision making.