Description and validation of the Postoperative Discharge Recovery State outcome: a patient-partnered population-based cohort study.

Abstract

Older adults prioritise independent return home after surgery. Most discharge outcomes are binary composites that do not incorporate temporal information. We defined and validated a novel ordinal outcome, the Postoperative Discharge Recovery State, and prioritised its temporal measurement, to overcome these limitations. This retrospective cohort study was conducted with patient partnership. Adults ≥65 yr having major, elective, noncardiac, non-orthopaedic surgery were identified from 2012 to 2022 using linked, routinely collected data in Ontario, Canada. Construct, convergent, and predictive validity were estimated. A multivariable ordinal regression model was derived and internally-externally validated. We included 84 422 older adult surgical patients. At the patient-prioritised postoperative day 90, the distribution of patients across Postoperative Discharge Recovery State categories was: (1) dead (2718; 3.2%); (2) hospitalised (1696; 2.0%); (3) in long-term care (179; 0.2%); (4) in rehabilitation (593; 0.7%); and (5) at home (79 236; 93.9%). Directionally expected associations with baseline characteristics supported construct validity. Consistency in associations over time supported reliability. Relationships with days alive and at home supported convergent (ρ=0.373) and predictive (fewer days at home with worse recovery state) validity. A prespecified ordinal logistic regression model had inadequate accuracy (c-statistic 0.700, poor calibration) to support its clinical use. The Postoperative Discharge Recovery State is a 5-level ordinal outcome that can be applied at key time points after surgery to quantify the proportion of patients in patient-prioritised discharge locations. Validity and reliability support utility, but further development will be required to maximise information gain relative to binary outcomes.