Validation of preoperative cardiopulmonary exercise testing‐derived variables to predict in‐hospital morbidity after major colorectal surgery. BJS 2016; 103: 744-752.

Results

Seven hundred and three patients from six centres in the UK were available for analysis (428 men, 275 women). ROC curve analysis of oxygen uptake at estimated lactate threshold (o₂ at _L) and at peak exercise (o_2peak) gave an area under the ROC curve (AUROC) of 0·79 (95 per cent c.i. 0·76 to 0·83; P < 0·001; cut‐off 11·1 ml per kg per min) and 0·77 (0·72 to 0·82; P < 0·001; cut‐off 18·2 ml per kg per min) respectively, indicating that they can identify patients at risk of postoperative morbidity. In a multivariable logistic regression model, selected CPET variables and body mass index (BMI) were associated significantly with increased odds of in‐hospital morbidity (o₂ at _L 11·1 ml per kg per min or less: odds ratio (OR) 7·56, 95 per cent c.i. 4·44 to 12·86, P < 0·001; o_2peak 18·2 ml per kg per min or less: OR 2·15, 1·01 to 4·57, P = 0·047; ventilatory equivalents for carbon dioxide at estimated lactate threshold (_E/co₂ at _L) more than 30·9: OR 1·38, 1·00 to 1·89, P = 0·047); BMI exceeding 27 kg/m²: OR 1·05, 1·03 to 1·08, P < 0·001). A laparoscopic procedure was associated with a decreased odds of complications (OR 0·30, 0·02 to 0·44; P = 0·033). This model was able to discriminate between patients with, and without in‐hospital morbidity (AUROC 0·83, 95 per cent c.i. 0·79 to 0·87). No adverse clinical events occurred during CPET across the six centres.