Abstract: A Validated, Risk Assessment Model for Predicting Morbidity After Breast Surgery (Plastic Surgery 2013)

Monday, October 14, 2013: 10:15 AM

Karim A Sarhane, MD, MSc , Plastic and Reconstructive Surgery, Johns Hopkins University, Baltimore, MD

José M Flores, MPH , Epidemiology, Johns Hopkins University, Baltimore, MD

Andrew D Shore, PhD , General Surgery, Johns Hopkins University, Baltimore, MD

Francis M Abreu, BSc , Biostatistics, Johns Hopkins University, Baltimore, MD

Zuhaib Ibrahim, MD , Plastic and Reconstructive Surgery, Johns Hopkins University, Baltimore, MD

Mohammed Alrakan, MD , Plastic and Reconstructive Surgery, Johns Hopkins University, Baltimore, MD

Carisa M Cooney, MPH , Plastic and Reconstructive Surgery, Johns Hopkins University, Baltimore, MD

Pablo Baltodano, MD , Plastic and Reconstructive Surgery, Johns Hopkins University, Baltimore, MD

Carolyn Drogt, BSc candidate , School of Medicine, Johns Hopkins University, Baltimore, MD

Martin A Makary, MD , General Surgery, Johns Hopkins University, Baltimore, MD

Gerald Brandacher, MD , Plastic and Reconstructive Surgery, Johns Hopkins University, Baltimore, MD

Gedge D Rosson, MD , Dept. of Plastic and Reconstructive Surgery, Johns Hopkins University School of Medicine, Baltimore, MD

PDF file

OBJECTIVE We introduce a validated model for identifying patients at increased risk for morbidity after breast surgery.

BACKGROUND With increased mastectomy rates across all ages,¹ it is crucial to devise a validated model to reliably predicts postoperative morbidity. This allows better identification of high-risk patients, provides evidence-based strategy for preoperative safety optimization, and contributes to hospital cost reduction. We reviewed a large prospective risk-adjusted registry to build a risk assessment tool that predicts morbidity after mastectomy.

METHODS All females undergoing mastectomy-only were identified from ACS-NSQIP²2008-2010 databases. Patients experiencing 30-day postoperative morbidity were compared to controls. Morbidity included events affecting the wound, heart, CNS, respiratory and urinary tracts, sepsis and venous thromboembolism. Exploratory univariate analyses and forward and stepwise logistic regression were used to develop a clinical predictive model of postoperative morbidity based on patients' specific baseline characteristics. A clinical risk score was derived to simplify the data. Goodness-of-fit was verified using Hosmer-Lemeshow X².

RESULTS 36,900 patients were analyzed; 1,292 (3.5%) experienced postoperative morbidity. Significant independent predictors of morbidity included: myocardial infarction (OR=6.01), congestive heart failure (OR=2.33), infected surgical wound (OR=2.17), angina (OR=2.15), sepsis (OR=2.06), inpatient-status (OR=1.99), general anesthesia (OR=1.95), impaired functional status (OR=1.79), bleeding disorder (OR=1.58), high INR (OR=1.57), obese diabetic (OR=1.53), smoking (OR=1.52), elevated SGOT (OR=1.51), stroke history (OR=1.47), ASA class III-V (OR=1.41), dyspnea (OR=1.39), hypertension (OR=1.17). Calibration of this model (Figure) was acceptable by Hosmer-Lemeshow's X²(6.246, 7-df, p=0.5116). We then derived a simplified Clinical Risk Score (CRS), assigning numerical point values to significant predictors of morbidity based on logistic coefficients rounded to the next integer (minimum of 1). Overall score is obtained by summation, and ranges from 0-18: each predictor receives one point except for MI which receives two (Table). The simplified system preserved the ability to accurately predict morbidity; it showed low-risk patients (SRS 0-2, n=20,205) had a morbidity rate of 2%, intermediate-risk patients (SRS 3-5, n=14,692) a rate of 5%, while high-risk patients (SRS 7+, n=957) 15%.

CONCLUSION We present a simple model that prospectively predicts morbidity after breast surgery using data readily available to clinicians. This may allow targeted intervention in high-risk patients, improving health outcomes and decreasing overall cost. Different strategies for high and low-risk groups should be implemented in patient safety optimization and cost reduction efforts.

22412 A Validated, Risk Assessment Model for Predicting Morbidity After Breast Surgery