OBJECTIVES: Identification of surgical complications in real-world settings is important to understanding associated burden of illness and unmet needs. Using the linked Clinical Practice Research Datalink (CPRD)/Hospital Episode Statistics (HES) databases, this study evaluated the impact of varying definitions of surgical bleeding complications on incidence and association with hospital length of stay (LOS) in U.K. patients who underwent selected in-hospital surgeries.

METHODS: Patients age≥18y who underwent selected in-hospital surgeries (cholecystectomy, cardiac surgery, hernia repair, hysterectomy, knee/hip arthroplasty, low anterior resection, lung surgery, mastectomy, prostate surgery) between 2010Q1-2016Q1 were identified from the linked CPRD/HES databases. Surgical bleeding complications from date of surgery through discharge were identified using three alternative definitions: (1) primary definition: haemorrhage/haematoma complicating a procedure and/or bleeding reintervention/reoperation; (2) strict definition: reoperation for surgical arrest of bleeding; (3) broad definition: primary definition+broader set of diagnoses and reintervention/reoperation procedures. Multivariable generalized linear models evaluated the impact of surgical bleeding complications on hospital LOS from date of surgery through discharge.

RESULTS: Study included 134,316 eligible surgical patients (51% age≥65yr; 55% female; 91% elective admissions). Incidence proportions of surgical bleeding complications from date of surgery through discharge were: 1.4% primary definition; 0.1% strict definition; 1.9% broad definition; incidence proportions varied substantially among the surgeries, ranging from 6.7% in cardiac surgery to 0.7% in knee/hip arthroplasty (primary definition). In multivariable models adjusting for patient and procedure characteristics, surgical bleeding complications were associated with significantly increased LOS: 10.2d vs. 4.8d primary definition (p<0.0001); 9.8d vs. 4.7d broad definition (p<0.0001); and 15.7d vs. 4.9d strict definition (p<0.0001).

CONCLUSIONS: In the linked CPRD/HES databases, the chosen method of defining surgical bleeding complications can lead to substantial variation in estimated incidence proportions; however, all significantly increase LOS. Improved bleeding management may reduce the clinical and economic burden associated with perioperative bleeding complications.