External validity of a model to predict postoperative atrial fibrillation after thoracic surgery

Abstract

OBJECTIVES: A prediction model developed by Passman et al. stratifies patients' risk of postoperative atrial fibrillation (POAF) after major non-cardiac thoracic surgery using 3 simple factors (sex, age and preoperative resting heart rate). The model has neither undergone external validation nor proven to be relevant in current thoracic surgery practice. METHODS: A retrospective single-centre analysis of all patients who underwent major non-cardiac thoracic surgery (2008-2017) with prospective documentation of incidence and severity of POAF was used for external validation of Passman's derivation sample (published in 2005 with 856 patients). The model calibration was assessed by evaluating the incidence of POAF and patients' risk scores (0-6). RESULTS: A total of 2054 patients were included. Among them, POAF occurred in 164 (7.9%), compared to 147 (17.2%) in Passman's study. Differences in our sample compared to Passman's sample included mean heart rate (75.7 vs 73.7 bpm, P < 0.001), proportion of patients with hypertension (46.1 vs 29.4%, P < 0.001), proportion of extensive lung resections, particularly pneumonectomy (6.1 vs 21%, P < 0.001) and proportion of minimally invasive surgeries (56.6% vs 0%). The model demonstrated a positive correlation between risk scores and POAF incidence (risk score 1.2% vs 6.16%). CONCLUSIONS: The POAF model demonstrated good calibration in our population, despite a lower overall incidence of POAF compared to the derivation study. POAF rates were higher among patients with a higher risk score and undergoing procedures with greater intrathoracic dissection. This tool may be useful in identifying patients who are at risk of POAF when undergoing major thoracic surgery and may, therefore, benefit from targeted prophylactic therapy.