Journal of Plastic, Reconstructive & Aesthetic Surgery (2014) 67, 693e701

Concurrent panniculectomy with open ventral hernia repair has added risk versus ventral hernia repair: An analysis of the ACS-NSQIP database John P. Fischer a,*, Charles T. Tuggle b, Ari M. Wes a, Stephen J. Kovach a a Division of Plastic Surgery, Perelman School of Medicine at the Hospital of the University of Pennsylvania, Philadelphia, PA, USA b Section of Plastic Surgery, Yale School of Medicine, New Haven, CT, USA

Received 2 December 2013; accepted 19 January 2014

KEYWORDS Concurrent; Panniculectomy; Hernia repair; Outcomes; Complications; ACS-NSQIP; Propensity matching

Summary Background: Recent studies have assessed the risks and benefits of performing concurrent panniculectomy (PAN) in the setting of hernia repair, gynecologic surgery, and oncologic resections with conflicting results. The aim of this study is to assess the added risk of ventral hernia repair and panniculectomy (VHR-PAN) using the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) data sets. Methods: The 2005e2011 ACS-NSQIP databases were queried to identify all patients undergoing VHR alone or VHR-PAN. Propensity scores were used to account for potential selection bias given the non-randomized assignment of concurrent panniculectomy and the retrospective nature of this study. Multivariate logistic regression analyses were used to assess the impact of concurrent PAN on complications. Results: A total of 55,537 patients were identified. Propensity matching yielded two groups of patients: VHR (n Z 1250) and VHR-PAN (n Z 1250). Few statistically significant differences existed between matched cohorts. Overall, wound complications (P < 0.001), venous thromboembolism (P Z 0.044), incidence of reoperation (P < 0.001), and medical morbidity (P < 0.001) were significantly higher in the VHR-PAN group. In an adjusted, fixed-effects analysis, concurrent panniculectomy was associated with wound healing complications (OR Z 1.69, P < 0.001), increased incidence of unplanned reoperations (OR Z 2.08, P  0.001), venous thromboembolism (OR Z 2.48, P Z 0.043), and overall medical morbidity (OR Z 2.08, P < 0.001). Sub-group analysis of wound complications demonstrated that superficial surgical site infections occurred significantly more often in concurrent cases (P Z 0.018).

* Corresponding author. University of Pennsylvania, Division of Plastic Surgery, 3400 Spruce Street, Philadelphia, PA 19104, USA. E-mail address: [email protected] (J.P. Fischer). 1748-6815/$ - see front matter ª 2014 British Association of Plastic, Reconstructive and Aesthetic Surgeons. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.bjps.2014.01.021

694

J.P. Fischer et al. Conclusions: This analysis quantifies the added risk of performing a panniculectomy concurrent with ventral hernia repair, demonstrating higher incidence of wound complications (superficial infections), unplanned reoperations, and VTE. Level of Evidence: Prognostic/risk category, level II. ª 2014 British Association of Plastic, Reconstructive and Aesthetic Surgeons. Published by Elsevier Ltd. All rights reserved.

Introduction An estimated 250,000 ventral hernia repairs are performed each year in the United States (US), where more than one third of the adult population is obese.1,2 The incidence of ventral hernia following midline laparotomy coupled with an obesity epidemic in the US has set the stage for challenging hernia repairs in those patients with excess abdominal adiposity.3e7 The pervasive, multisystemic nature of obesity imparts added surgical risk to an already challenging operation. For patients, the presence of a large pannus can be debilitating both physically and psychosocially, but also confers a significant wound complication risk.8 Plastic surgeons are called upon as ‘soft-tissue experts’ to assist in the management of wound closures and/or soft-tissue management in these complex abdominal wall cases. Recent studies provide conflicting data on the risk-benefit of performing concurrent panniculectomy (VHR-PAN) with hernia repair,9e16 gynecologic surgery,17e25 and bariatric surgery.26e28 Some discuss the benefits of performing a concurrent panniculectomy, and cite improved patient satisfaction, surgical exposure, and decreased tension on the hernia repair.9,10,17,18,20,23,25,29,30 However, the combination of two major operations implies a longer operative time and greater wound healing demands, placing patients at increased risk for VTE and wound complications.24 Several studies have also reported higher incidence of wound complications when hernia repairs are combined with panniculectomy,14,31 but center-specific and institutional/ surgeon preferences limit the generalizability of these data. Logically, the addition of a separate procedure should introduce added risk, but this risk has not been clearly defined in the literature. For practicing reconstructive surgeons a generalizable assessment of the added morbidity of performing a soft tissue intervention, such as a panniculectomy, concurrent with a hernia repair would provide important data which could then be communicated preoperatively to patients and used to improve perioperative decision-making. Thus, the purpose of this study was to assess the added surgical and medical risk of combining panniculectomy with open ventral hernia repair in a singlestage by examining 30-day postoperative outcomes at more than 250 community and academic hospitals that contributing to the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP).

Methods Analytic cohort and database The 2005e2011 ACS-NSQIP participant use data files were accessed on December 1, 2012 and queried to identify all

patients undergoing VHR or VHR-PAN. Current Procedural Terminology (CPT) codes were used to identify open hernia repairs: 49560, 49561, 49565, and 49566. Both hernia repairs with (49568) and without mesh were included. Concurrent panniculectomy was identified using CPT codes 15830 and 15847. Patients undergoing hernia repair alone (VHR) and hernia repair with panniculectomy (VHR-PAN) were included. Patients undergoing emergency surgery and those who had surgery 30 days prior to VHR were excluded (Figure 1). Intra-abdominal procedures were identified in cases of enterolysis (44005/44180), exploratory laparotomy (49000), omentectomy (49255), enterectomy (44120), enterorrhaphy (44602), cholecystectomy (47600/47562), appendectomy (44955), closure enterostomy (44620), revision colostomy (44346), total abdominal hysterectomy and oophorectomy (58150), and oophorectomy (58940). Concurrent bowel procedures were identified if enterectomy (44120), enterorrhaphy (44602), closure enterostomy (44620), or revision colostomy (44346) were performed. The use of acellular dermal matrix (ADM) mesh (15330, 15331, 15430, and 15431) as an adjunct in complex reconstructions was also identified.

Outcomes Complications examined were wound (superficial/deep surgical site infection and wound dehiscence), unplanned return to operating room, venous thromboembolism (symptomatic deep venous thrombosis or pulmonary embolism), and medical complications (Table 1). Medical complications included any defined NSQIP non-surgical endpoint, such as pneumonia, pulmonary embolism, postoperative renal insufficiency (Creatinine > 2 mg/dl), urinary tract infection (UTI), stroke, myocardial infarction (MI), symptomatic deep venous thrombosis (DVT), or sepsis. A surgical complication was defined as either a wound complication or an unplanned return to the operating room. Complications were treated as a dichotomous variable (none vs. one or more). Information regarding severity was not available. All complications were identified within 30 days of the index procedure.

Independent variables Variables for patient demographics, comorbidities, and perioperative risk factors were selected. These included baseline health characteristics, past medical and recent surgical history, and American Society of Anesthesiologists (ASA) physical status. The World Health Organization (WHO) definition of obesity was used to stratify patients into class I obesity (BMI Z 30e34.9 kg/m2), class II obesity

Concurrent PAN with open VHR has added risk versus ventral hernia repair

Figure 1

695

Attrition diagram of exclusion parameters to final analytic cohort.

(BMI Z 34.9e39.9 kg/m2), and class III obesity groups (BMI  40 kg/m2).32 Comorbidities were examined individually, by system, and in aggregate. System-based definitions of comorbid conditions were defined as follows: cardiovascular included a history of hypertension, angina, congestive heart failure, myocardial infarction, percutaneous coronary intervention, cardiac surgery, rest pain, or peripheral vascular disease; pulmonary included recent dyspnea, chronic obstructive pulmonary disease, recent pneumonia, or a recent need for ventilator-assisted respiration; neurologic included paralysis, recent coma, recent delirium, cerebrovascular accident, or transient ischemic attack; and renal included preoperative renal insufficiency and dialysis.

Propensity score matching Propensity scores were incorporated into this analysis to account for potential selection bias given the nonrandomized assignment of concurrent panniculectomy and the retrospective nature of this study.33e35 A propensity score for undergoing VHR-PAN was calculated by assessing perioperative factors associated with selection of PAN (yes or no) in a multivariate logistic regression. The propensity

score ranged from 0, representing the lowest probability of undergoing concurrent panniculectomy, to 1, representing the highest probability of undergoing concurrent panniculectomy. In brief, patients undergoing VHR-PAN were selected via a sophisticated matching process to have similar characteristics to VHR patients. VHR-PAN cases were then matched with VHR cases by propensity score, using a caliper of 0.005. A sub-group analysis was then performed on the matched groups to examine the effect of BMI on surgical complications defined as any wound or any complication requiring an unplanned return to the operating room.

Statistical analysis Bivariate analyses of independent variables by our outcomes of interest were performed using Pearson Chi Square and Fisher’s Exact tests for categorical variables and the Wilcoxon rank-sum test for continuous variables. All tests were two-sided, with statistical significance set at a probability value of 0.05. Propensity-matched encounters were compared with respect to demographics, comorbidities, and outcomes. Individual-adjusted risk models were created for each complication type. Conditional logistic

696 Table 1 Summary complications.

J.P. Fischer et al. of

Wound complication

Unplanned return to operating room Any surgical complication

Venous thromboembolism

Medical complications

30-day

ACS-NSQIP-defined

Superficial wound infection Deep wound infection Wound dehiscence

Wound complication or unplanned return to operating room Symptomatic deep venous thrombosis Symptomatic pulmonary embolism Pneumonia Venous thromboembolism Renal insufficiency (Creatinine > 2 mg/dl) Urinary tract infection Stroke Myocardial infarction Sepsis

regression analyses were conducted on the two matched cohorts to assess the impact of PAN on outcomes. Perioperative characteristics which remained significant after propensity-matching were included in all the risk models of specific outcomes. The significance level for inclusion in each of the final models was set at P  0.10. Data management was performed using SAS (SAS Institute Inc., Cary, North Carolina) and data analysis using STATA IC 10.0 (StataCorp, College Station, Texas).

Results A total of 55,537 patients who met inclusion criteria were identified in the 2005e2011 ACS-NSQIP databases. Comparison of unmatched VHR and VHR-PAN patient characteristics revealed differences for year of surgery (P < 0.001), race (P < 0.001), age (P < 0.0001), smoking status (P < 0.001), alcohol use (P Z 0.001), ASA physical status (P < 0.001), and BMI (P < 0.0001). Compared to VHR alone, VHR-PAN procedures were less often outpatient (P < 0.001), had longer operative times (P < 0.0001), more often involved component separation (P < 0.001) and biologic mesh (P < 0.001), and more often involved concurrent intra-abdominal procedures (P < 0.001), bowel procedures (P Z 0.004), and contaminated wounds (P < 0.001). Overall, VHR-PAN patients were more often obese (P < 0.001), with a greater pulmonary comorbidity burden (P < 0.001). Bivariate analysis of the unmatched VHR and VHR-PAN groups showed that wound complications (P < 0.001), VTE (P < 0.001), unplanned return to the operating room (P < 0.001), and medical complications (P < 0.001) were significantly higher in the VHR-PAN group. Multivariate logistic regression analysis of patient and procedure characteristics identified several variables that were independently associated with undergoing VHR-PAN

(Table 2). VHR-PAN procedures were more often performed in years 2009e2011 (P < 0.001), in higher BMI patients (P < 0.001), in contaminated wounds (P < 0.001), and less often as outpatient surgery (P < 0.001). These characteristics along with age, race, active smoking status, alcohol use, and ASA physical status were used to match the two cohorts. VHR-PAN was associated with a longer operative times (mean: 68 min) (P < 0.0001) and hospital length of stay (1.3 days) (P < 0.0001) (Table 3). VHR-PAN patients more frequently underwent component separation (P  0.001) and biologic mesh reconstruction (P Z 0.003), usually in the setting of contaminated wounds (P < 0.001). Matched analysis revealed that wound complications were significantly higher when VHR-PAN was performed (P < 0.001) with an absolute additional risk of 4.4% (Table 4). Also, a 0.9% increased risk of VTE was associated with VHR-PAN compared to VHR alone (P Z 0.044). Overall, a 2-fold difference in medical complications was observed in patients undergoing concurrent panniculectomy (11.9% vs. 6.1%, P < 0.001). VHR-PAN patients, also, more frequently experienced complications requiring an unplanned return to the operating room within 30 days (P < 0.001). When controlling for cohort differences which persisted after matching, panniculectomy was independently associated with added risk of wound healing complications (OR Z 1.69, P < 0.001), with unplanned return to the operating room (OR Z 2.08, P  0.001) and with VTE (OR Z 2.48, P Z 0.043) (Table 5). Risk factors for medical complications included undergoing panniculectomy (P < 0.001) and the presence of clean-contaminated (P Z 0.036) or infected (P Z 0.014) wounds. Sub-group analysis demonstrated that of all wound complications panniculectomy was associated with only a significant added risk of superficial infections (5.2% vs. 7.6%, P Z 0.018) (Table 6). VHR-PAN was not associated with significant added risk of deep infection (P Z 0.053),

Table 2 Multivariate regression analysis of unmatched variables associated with concurrent panniculectomy. Variable

Odds ratio

Confidence interval

P value

Year Black race Hispanic race Other race Age Smoking Alcohol ASA physical status Body mass index Wound class Pulmonary comorbidity Diabetes Concurrent intra-abdominal procedure Outpatient surgery

1.42 0.67 0.63 0.35 0.97 0.61 0.46 0.75 1.02 1.22 1.12 1.11 1.08

1.25e1.60 0.55e0.81 0.49e0.82 0.18e0.67 0.97e0.98 0.52e0.72 0.25e0.87 0.66e0.85 1.02e1.03 1.13e1.33 0.99e1.28 0.96e1.30 0.93e1.24

Concurrent panniculectomy with open ventral hernia repair has added risk versus ventral hernia repair: an analysis of the ACS-NSQIP database.

Recent studies have assessed the risks and benefits of performing concurrent panniculectomy (PAN) in the setting of hernia repair, gynecologic surgery...
393KB Sizes 0 Downloads 0 Views