Back to Annual Meeting Program
VALIDATED MODEL TO PREDICT POSTOPERATIVE RESPIRATORY FAILURE (PRF) FOLLOWING ABDOMINAL WALL RECONSTRUCTION: ANALYSIS OF 1,706 PATIENTS FROM NSQIP DATABASES
Presenter: Jonas A Nelson, Resident1
Co-Authors: John P. Fischer MD1, Eric K. Shang MD2, Charles E. Butler MD3, Jonas A. Nelson MD1, Benjamin M. Braslow MD2, Joseph M. Serletti MD1, and Stephen J. Kovach MD1
1Division of Plastic Surgery, Hospital of the University of Pennsylvania, Philadelphia, PA, 2Department of Surgery, Hospital of the University of Pennsylvania, Philadelphia, PA, 3Department of Plastic Surgery, The University of Texas MD Anderson Cancer Center, Houston, TX
Purpose: Complex abdominal wall reconstruction (AWR) often requires component separation to achieve fascial approximation. These reconstructions can be associated severe postoperative respiratory events. There is a significant need to quantify these occurrences and identify patients at risk for post-operative respiratory failure (PRF), so that proactive strategies can be employed to reduce pulmonary morbidity following AWR. In this current study, we aim to characterize factors associated with PRF and derive a model to predict PRF following AWR using the ACS-NSQIP database.
Methods: We reviewed the ACS-NSQIP databases (2005-2010) and identified encounters for CPT codes including both hernia repair (49560, 49561, 49565, 49566, 49568) and component separation (15734). Patient co-morbidities, laboratory values, and intraoperative details were extracted from the NSQIP datasets (http://www.acsnsqip.org/). We defined PRF as either an unplanned intubation or prolonged ventilation (>48 hours). Exploratory univariate analyses were performed and multivariate regression analyses were used to identify potential independent predictive or protective factors for PRF. A PRF risk model was created and validated using bootstrap methodology.
Results: A total of 1,706 AWRs were performed during the study period. Patients’ age was 55.9 ± SD years and 57% were female. The vast majority were inpatient (94.1%) and 56.9% included a mesh repair. Approximately 1/3 involved a concurrent intra-abdominal procedure. Operative time was 212 ± SD minutes. The incidence of PRF was 6.0% and the length of stay was 6.8 ± SD days. The 30-day, all-cause mortality was 0.9% for the study cohort.
Multivariate logistic regression analysis revealed that COPD (p<0.001), dyspnea at rest (p=0.032), functional status (p=0.032), malnutrition (p<0.001), recurrent, incarcerated hernia (p=0.006), concurrent intra-abdominal procedure (p=0.041), ASA >3 (p<0.001), and prolonged operative time (p<0.001) were independent predictive factors for development of PRF (Table 1).
A validated model and a simplified, weighted risk assessment tool (scaled from 0 to 10) (Table1 and Figure 2) were created incorporates these risk factors with excellent discrimination (C-statistic, Figure 1) and sensitivity. A respiratory risk score (RRS) was derived using the following simplified equation:
RRS = 2*COPD + 2*Dyspnea + Functional status + Malnutrition + Recurrent hernia + Intra-abdominal procedure + ASA>3 + Prolonged operative time
Patient with a RRS >4 experienced a PRF of more than 50% compared (Figure 2). Patients experience PRF stayed on average 15 days longer (P<0.001) and experienced a mortality rate of 14.8% (P<0.001)
Conclusions: This study quantifies the incidence of PRF following AWR, and describes a novel validated, predictive model and clinical, risk assessment tool for the occurrence of PRF. This simplified risk score system divide patients unto low (2.9%), intermediate (14.8%), and high (50%) for experiencing PRF. Data derived from this large cohort can be used to risk-stratify patients and to enhance peri-operative decision-making.
Back to Annual Meeting Program