EP353 - Factors Associated With Use Of Patient Engagement Technology Among Patients Undergoing Colorectal Surgery: A Machine Learning Analysis

Image or Table (Do Not add text here):

Purpose/Background: Patient engagement technologies (PETs) improve surgical outcomes and reduce costs among colorectal surgery patients. However, not all patients have equal access to or engagement with PETs. Therefore, the aim of this study was to identify factors associated with PET use among patients undergoing colorectal surgery.

Methods/Interventions: We conducted a retrospective cohort study utilizing single-institution ACS-NSQIP and PET (SeamlessMD, Toronto, CA) enrollment data for patients undergoing elective colorectal surgery between January 2018 and December 2023. The primary outcome was PET use. PET use was defined by completion of surveys and was categorized into (i) total number of surveys completed and (ii) completion of at least one survey (preoperative, inpatient, postdischarge, or PROMIS). The machine-learning method of classification and regression trees were used to identify factors associated with PET use. All statistical analyses were conducted in R.

Results/Outcomes: A total of 1151 patients were included in the study. Overall, most patients were female (58.8%), White (74.6%), non-smokers (84.4%), non-diabetic (85.4%), lived in urban zip codes (73.2%), and were undergoing an extended (38.8%) or partial colectomy (35.1%). The mean BMI was 29.07 (SD 7.36) kg/m2 and mean age was 55.4 (SD 15.3) years. Overall, the mean PET use was 5.2 surveys completed. In the regression tree for number of surveys completed, hospital length of stay (LOS) of >6 days was the first-level and most important predictor for total surveys completed (Figure). This number increased to 6.3 for patients with a shorter LOS ( < 6 days, 71% of population) but decreased to 2.4 for patients with a long LOS (> 6 days, 29% of population). Age was the second-level predictor for the total number of surveys completed. Among patients who had a short LOS, older patients (age >42 years) completed more surveys (7.1 vs. 3) compared to younger patients ( < 42 years). Among older patients, patients who had at least one readmission completed an average of 3.6 surveys whereas patients who had no readmissions completed more than twice as many surveys (7.5). In the classification tree for completion of at least one preoperative or PROMIS survey, race appeared as the first-level predictor. Overall, 68% of patients completed at least one preoperative survey. This increased to 73% for White patients but decreased to 55% for patients who were Asian, Black, or in another minority racial group. Overall, 51% of patients completed at least one PROMIS survey. This increased to 54% among White patients but decreased to 39% among minority patients. Race was also a second-level predictor for postdischarge health check survey completion and a third-level predictor for inpatient health check survey completion, with minority group patients experiencing consistently low survey completion rates compared to White patients.

Conclusion/Discussion: Race/ethnicity, age, and hospital length of stay are the most important factors associated with PET use among patients undergoing colorectal surgery. Interventions aimed at increasing PET use and enhancing patient engagement should account for these factors.