Enhanced recovery after surgery components and perioperative outcomes: a nationwide observational study

Enhanced recovery after surgery (ERAS) protocols have been shown to benefit recovery after several operations. However, large-scale data on the association between the level of ERAS use and perioperative complications are scarce, particularly in surgeries with increasing ERAS uptake, including total hip (THA) and knee arthroplasty (TKA). Using US national data, we examined the relationship between the number of ERAS components implemented ('level') and perioperative outcomes.

After ethics approval, we included 1 540 462 elective THA/TKA procedures (2006-2016, as recorded in the Premier Healthcare claims database) in this retrospective cohort study. Main outcomes were any complication, cardiopulmonary complications, mortality, blood transfusions, and length of stay. Eight commonly used ERAS components were included. Mixed-effects models measured associations between ERAS level and outcomes, with odds ratios (OR) and confidence intervals (CI) reported.

ERAS use increased over time; overall, 21.6% (n=324 437), 62.7% (n=965 953), and 18.0% (n=250 072) of cases were classified as 'High', 'Medium', or 'Low' ERAS. 'High ERAS', 'Medium ERAS', and 'Low ERAS' level of use were defined as such if they received either >6, 5-6, or <5 ERAS components, respectively. After adjustment for relevant covariates, higher levels of ERAS use were associated with incremental reductions in 'any complication': 'Medium' vs 'Low' (OR=0.84; CI, 0.82-0.86) and 'High' vs 'Low' (OR=0.71; CI, 0.68-0.74). Similar patterns were found for the other study outcomes. Individual ERAS components with the strongest effect estimates were early physical therapy, avoidance of a urinary catheter, and tranexamic acid administration.

ERAS components were used more frequently over time, and the level of utilisation was independently associated with incrementally improved complication odds and reduced length of stay during the primary admission. Possible indication bias limits the certainty of these findings.