Risk Calculator for Surgical Complications: Analysis of the Mastectomy Reconstruction Outcomes Consortium Data

Abstract

INTRODUCTION: Patient education regarding mastectomy reconstruction can be enhanced by individualized assessment of risk vis-a-vis patient factors. However, large-scale modeling of individualized risk can be hampered by intrinsic limitations of the databases used. The Mastectomy Reconstruction Outcomes Consortium (MROC) created a large scale, robust, plastic surgery-specific database from which we tried to generate an individualized risk calculator for breast reconstruction patients. METHOD(S): Patients with a minimum of 1-year follow-up were included. Incidences of surgical complications, including surgical site infection (SSI), seroma, mastectomy flap necrosis (MFN), wound healing complications, secondary implant complications, prosthesis explantation/replacement, and flap loss, were examined. Additionally, multivariable logistic regression was used to assess independent risk factors for overall complications. Resultant data were used to generate a risk calculator for overall complications. RESULT(S): There were 1,647 patients who met inclusion criteria. Outcome incidence is shown in Table 1. Risk factors assessed and their independent odds ratios were as follows: age (1.01 per year), BMI (1.01 per point), current/past smoking (1.29), diabetes (1.27), bilateral (1.32), and preoperative radiation (1.42). This risk model demonstrated Hosmer-Lemeshow and c-statistics of 0.747 and 0.677, respectively. Range of predicted probability was 5.70%- 48.76%. This model was added to the online BRAScore interface, an open access risk estimation tool available at BRAScore.org. CONCLUSION(S): Regression analysis of the MROC data yielded a well-calibrated model that demonstrated expected trends in odds ratios. The model was used to generate a calculator of individual patient risk. The wide spread in predicted risk highlights the usefulness of an individualized risk calculator like this in clinical practice. (Table presented).