A study published in JAMA Oncology recommended a two-pronged approach in order to increase serious illness conversations (SICs) in patients with cancer: machine learning (ML) mortality predictions plus behavioral nudges to clinicians.

“Early discussions about goals and treatment preferences may lead to better perceived quality of life, reduced emotional distress, and decreased health care use near the end of life. However, most patients with cancer die without a documented discussion about goals and treatment preferences,” the study authors observed. Increasing SICs may improve outcomes for cancer patients.

The researchers performed a randomized clinical trial over a 20-week period at nine medical oncology clinics. Seventy-eight SIC-trained oncology clinicians and their patients were evaluated in the trial. The intervention consisted of weekly emails to the clinicians with feedback on SIC performance and peer comparisons; a list of patients at high risk for mortality, identified using a validated ML algorithm; and text messages to the clinicians on the day of a patient encounter suggesting they consider an SIC – the “nudge.” Control clinicians received only usual care (weekly emails with cumulative SIC performance).

The mean (SD) age of the 14,607 patients included in the analysis was 61.9 (14.2) years. The cohort was 53.7% female and 70.4% White.

The proportion of patient encounters that included SICs was significantly lower in the control group than the intervention group (1.3% vs. 4.6%; adjusted difference in percentage points, 3.3; 95% confidence interval [CI], 2.3–4.5; P<0.001).

The difference was even more significant when analyzing the 4,124 high-risk patient encounters; in the control group, 3.6% of encounters included SICs, compared to 15.2% in the intervention group (adjusted difference in percentage points, 11.6; 95% CI, 8.2–12.5; P<0.001).

“An intervention combining ML mortality risk predictions and behavioral nudges led to an increase in SICs among patients with cancer. This study provides guidance for more robust integration of automated risk predictions alongside behavioral interventions in future prospective studies related to and outside of SICs in oncology,” the study authors wrote in their conclusion.