Receiver operating characteristic analysis to determine optimal fluid management during open colorectal surgery

No Thumbnail Available
Pache, Basile
Hübner, Martin
Solà, Josep
Hahnloser, Dieter
Demartines, Nicolas
Grass, Fabian
Aim The present study aimed to analyse fluid management and to define optimal fluid-related thresholds for elective open colorectal surgery. Method A retrospective analysis was made of all consecutive elective open colorectal resections performed in our tertiary centre between May 2011 and May 2017. The main outcomes were postoperative complications [overall (I-V) and severe (IIIB-V) according to the Clavien classification], respiratory complications and postoperative ileus (POI). Critical thresholds regarding perioperative fluid management and postoperative weight gain were identified by using receiver operator characteristic (ROC) analysis. Independent risk factors for overall complications were identified by multivariable logistic regression analysis. Results Of 121 patients who had open operations, 84 (69%) had some complication and 26 (21%) had severe complications. Respiratory complications and POI occurred in 15 (12%) and 46 patients (38%), respectively. The thresholds for intravenous fluids were 3.5 l at postoperative day (POD) 0 [area under ROC curve (AUROC) 0.7 for any 0.69 for respiratory complications] and 3.5 kg weight gain at POD 2 (AUROC 0.82 for respiratory complications). Multivariable analysis revealed weight gain of > 3.5 kg at POD 2 (OR 5.9; 95% CI 1.3-16.6) as a significant risk factor for overall complications. Acute kidney injury was observed in five patients (4%), three (5%) in the group with > 3.5 l at POD 0 and two (3%) in the group with POD 0 (P = 0.64). Creatinine increase was transitory and all patients regained baseline levels before discharge. Conclusion A weight gain of > 3.5 kg at POD 2 has been identified as the critical threshold for overall and respiratory complications and prolonged length of stay after open elective colorectal surgery.
Publication Reference
Colorectal Disease, vol. 21 (2), pp. 234-240