ImportanceFew person-centered, scalable models of collaborative intensive care unit (ICU) clinician–palliative care specialist care exist.ObjectiveTo evaluate the effect of a collaborative palliative care intervention compared to usual care among family members of patients in the ICU.Design, Setting, and ParticipantsThis parallel-group randomized clinical trial with patient-level randomization was conducted between April 2021 and September 2023. The study was set at 6 medical and surgical ICUs in 1 academic hospital and 1 community hospital. The study participants included critically ill older adult patients with 1 of 11 poor outcome phenotypes, their family members with elevated palliative care needs, and their attending ICU physicians.InterventionAn automated electronic health record–integrated, mobile application–based communication platform that displayed family-reported needs over 7 days, coached ICU attending physicians on addressing needs, and prompted palliative care consultation if needs were not reduced within 3 study days.Main Outcomes and MeasuresThe primary outcome was change in the family-reported Needs at the End-of-Life Screening Tool (NEST) score between study days 1 and 3. The 13-item NEST score is a number between 0 and 130, with higher scores indicating a greater need. Secondary outcomes included quality of communication and goal of care concordance, as well as 3-month psychological distress.ResultsOf 151 family members, the mean (SD) age was 57.4 (12.9) years, and 110 (72.9%) were female. Of 151 patients, the mean (SD) age was 69.8 (9.7) years, and 86 (57.0%) were male. Thirty-five ICU physicians were male (68.6%). Seventy-six patients were randomized to the intervention group and 75 to the control group. Treatment group differences in estimated mean NEST scores were similar at 3 days between the intervention and control groups (−3.1 vs −2.0, respectively; estimated mean difference in differences, −1.3 points [95% CI, −6.0 to 3.5]) and 7 days (−2.3 vs −2.2, respectively; estimated mean difference in differences, 0 points [95% CI, −6.2 to 6.2]). Median (IQR) need scores were lower among individuals who remained in the ICU at day 3 for intervention participants vs controls (24.5 [16.5-34.5] vs 27.5 [13.0-40.0], respectively); median (IQR) need scores were also lower among those who remained in the ICU at day 7 for intervention vs controls (22.0 [11.0-35.0] vs 28.0 [14.0-35.0], respectively). Goal concordance, quality of communication, and psychological distress symptoms did not differ. Twenty-nine intervention participants (38.2%) had palliative care consultations, compared to only 3 (4.0%) among controls, (P < .001); 66 intervention participants (87.0%) had a family meeting, compared to 48 (64.0%) among controls (P = .001).Conclusions and RelevanceIn this randomized clinical trial, a collaborative, person-centered, ICU-based palliative care intervention had no effect on palliative care needs or psychological distress compared to usual care despite a higher frequency of palliative care consultations and family meetings among intervention participants.Trial RegistrationClinicalTrials.gov Identifier: NCT04414787