27.04 A Preoperative Prediction Model for Risk of Multiple Admissions after Colon Cancer Surgery

J. H. Fieber1, C. E. Sharoky1, K. Collier1, R. L. Hoffman1, C. Wirtalla1, E. C. Paulson1, G. C. Karakousis1, R. R. Kelz1  1University Of Pennsylvania,Department Of Surgery,Philadelphia, PA, USA

Introduction: Colon cancer treatment commonly has a profound impact on patients’ and caregivers’ ability to maintain their involvement in the work force, potentially leading to loss of insurance and income. The use of medical services, including multiple hospital admissions [MuAdmin], contributes to time lost at work. We developed a simplified model to predict preoperative risk of MuAdmin amongst patients undergoing colon resection to help patients prepare for treatment and to guide improvement efforts.

Methods: Patients ≥18 years-old with colon cancer that underwent elective surgical resection without postoperative complications identified in discharge claims from California and New York (2008-2011) were included. The primary outcome factor, MuAdmin, was defined as the 90th percentile for admissions following resection. Logistic regression models were developed to identify factors predictive of MuAdmin. A weighted point system was developed using beta-coefficients (β) (p<0.05). A point value of 1 was assigned to β<0.5, 2 was assigned to 0.5≤β<1, and 3 was assigned to β≥1.   A random sample of 75% of the data was used for model development leaving a 25% sample for validation.

Results: A total of 14,805 patients underwent colon surgery with 27.3% requiring at least 1 admission. MuAdmin, defined as ≥2 admissions following resection, impacted 9.7% of patients.  The statistically significant predictors of MuAdmin were Elixhauser comorbidity index ≥3 (β=0.30), metastasis (β=0.96), payer system (Medicare β=0.25, Medicaid β=0.58), and the number of prior admissions in the year before resection (1: β=0.43, 2: β=0.54, 3: β=1.45). Scores ranged from 0-8. Scores ≤1 had <7% risk of MuAdmin, scores between 2-5 had at 10-21% risk of MuAdmin and scores ≥6 had a >30% risk of MuAdmin. Our prediction model accurately stratified patients by the likelihood of MuAdmin. [Table 1: Observed and Predicted Rates of MuAdmin following Colon Cancer Resection by Risk Score]

Conclusion: Following discharge after resection of colon cancer, almost a third of patients are admitted at least once and nearly 10% require 2 or more admissions in the year following surgery.  A simple, preoperative clinical model can predict the likelihood of multiple admissions in patients anticipating resection.  This information can assist patients and caregivers in managing time off from work to minimize the threat of unemployment and financial hardship.