Causal Inference
Frailty and adverse surgical outcomes: exploring the mediating pathway through Enhanced Recovery After Surgery protocols Sophia Fuller* Sophia Fuller Anlan Cao Sidney Le Elizabeth Cespedes Feliciano
Complications of surgery in frail adults are common and costly; perioperative care pathways such as Enhanced Recovery After Surgery (ERAS) protocols are designed to improve outcomes for patients after surgery. However, adherence to ERAS protocols is low and the extent to which achievement of ERAS milestones, such as early ambulation, can improve outcomes for frail patients is unclear. Causal mediation analysis can help clarify whether additional interventions should be mounted to support these patients in adhering to ERAS protocols. A cohort of 11,115 patients who had abdominal surgery between 2015 and 2020 at Kaiser Permanente were evaluated using the Hospital Frailty Risk Score and categorized as frail or non-frail. We modeled the causal relationship between frailty and 30-day morbidity, defined as mortality and major surgical complications, mediated by early ambulation. Among the 23% of patients classified as frail, 42% achieved ambulation within 12 hours of surgery compared to 65% of non-frail patients. We found that, had all patients been frail, morbidity would have been 15.3% (95%CI: 13.4% to 17.2%) higher than had all patients been non-frail. Most of the total effect is explained by the direct effect of frailty on 30-day morbidity (13.8%; 95%CI: 11.9% to 15.7%), however, the average causal mediation effect among frail and non-frail patients differed slightly. Had all patients been frail, there is a slight decrease in morbidity (-1.9%; 95%CI: -2.4% to -1.5%) if patients had been set to achieve early ambulation under what it would have been for non-frail patients than for frail patients. If all patients had been set to non-frail, this decrease is 1.1% (95%CI: -1.4% to -0.01%). This indicates a small benefit to patients, with more improvement among frail patients, of early ambulation. Using a causal mediation framework, we were able to quantify the relationship between frailty, early ambulation, and morbidity to understand additional benefit to frail patients.