Predictive validity of chronic obstructive pulmonary disease phenotypes in inpatient elective surgery: a...

Chronic obstructive pulmonary disease (COPD) is prevalent among surgical patients, yet guidance for its preoperative assessment remains limited. Whether previously defined COPD phenotypes influence outcomes after surgery is unknown.

Population-based retrospective cohort of older adults (≥65 yr) with COPD who underwent inpatient elective surgery in Ontario, Canada. Candidate COPD phenotypes included: advanced COPD with home oxygen; COPD with frailty; COPD with frequent exacerbation; COPD with cardiovascular comorbidity; both asthma and COPD; and COPD alone. Nested Cox proportional hazards models examined the added performance of COPD phenotype when added to a baseline model (age, sex, procedural risk, Surgical Outcome Risk Tool) in predicting survival in the year after surgery using model fit, discrimination, calibration, and net benefit analyses.

A total of 116 757 patients with COPD underwent inpatient elective surgery; the most common phenotypes included: COPD alone (41.8%), COPD with cardiovascular comorbidity (31.6%), and COPD with frailty (21.8%). There were significant differences in survival between phenotypes when added to the baseline model: advanced COPD (adjusted hazard ratio [aHR] 5.59) and COPD with frailty (aHR 3.56) were associated with markedly decreased survival, while COPD with frequent exacerbation (aHR 1.45) and COPD with cardiovascular comorbidity (aHR 1.35) were associated with moderately decreased survival vs COPD alone. Addition of COPD phenotype improved model fit (likelihood ratio test P<0.001), discrimination (C-index 0.775 vs 0.720), calibration (integrated calibration index 0.035 vs 0.043), and net benefit across all decision thresholds.

COPD phenotypes are predictive of postoperative survival and improve perioperative risk stratification. These findings support phenotype-based assessment in the preoperative evaluation of patients with COPD.