Coaching doctors to improve ethical decision-making in adult hospitalized patients potentially receiving excessive treatment : the CODE stepped-wedge cluster randomized controlled trial

The aim of this study was to assess whether coaching doctors to enhance ethical decision-making in teams improves (1) goal-oriented care operationalized via written do-not-intubate and do-not attempt cardiopulmonary resuscitation (DNI-DNACPR) orders in adult patients potentially receiving excessive treatment (PET) during their first hospital stay and (2) the quality of the ethical climate. Methods We carried out a stepped-wedge cluster randomized controlled trial in the medical intensive care unit (ICU) and 9 referring internal medicine departments of Ghent University Hospital between February 2022 and February 2023. Doctors and nurses in charge of hospitalized patients filled out the ethical decision-making climate questionnaire (ethical decision-making climate questionnaire, EDMCQ) before and after the study, and anonymously identified PET via an electronic alert during the entire study period. All departments were randomly assigned to a 4-month coaching. At least one month of coaching was compared to less than one month coaching and usual care. The first primary endpoint was the incidence of written DNI-DNACPR decisions. The second primary endpoint was the EDMCQ before and after the study period. Because clinicians identified less PET than required to detect a difference in written DNI-DNACPR decisions, a post-hoc analysis on the overall population was performed. To reduce type I errors, we further restricted the analysis to one of our predefined secondary endpoints (mortality up to 1 year). Results Of the 442 and 423 clinicians working before and after the study period, respectively 270 (61%) and 261 (61.7%) filled out the EDMCQ. Fifty of the 93 (53.7%) doctors participated in the coaching for a mean (standard deviation [SD]) of 4.36 (2.55) sessions. Of the 7254 patients, 125 (1.7%) were identified as PET, with 16 missing outcome data. Twenty-six of the PET and 624 of the overall population already had a written DNI-DNACPR decision at study entry, resulting in 83 and 6614 patients who were included in the main and post hoc analysis, respectively. The estimated incidence of written DNI-DNACPR decisions in the intervention vs. control arm was, respectively, 29.7% vs. 19.6% (odds ratio 4.24, 95% confidence interval 4.21–4.27; P