|
|
Back to 2016 Annual Meeting
ACS-NSQIP-Based Risk Score Predicts Readmission After Gastrectomy
Anh-Thu Le*, Jeremiah T. Martin, Ching-Wei Tzeng, Sean Dineen
General Surgery, University of Kentucky, Lexington, KY
Introduction: Unplanned readmissions are an economically and clinically relevant problem in patients who have undergone high-risk gastrointestinal surgery, including gastrectomy. Risk factors such as postoperative complications can qualitatively predict early readmission. However, no quantitative method exists to direct clinical care at the time of discharge. This study used a national database to examine predictors of early readmission and to create a scoring system to direct clinical decisions at discharge.
Methods: We identified all patients with readmission data who had undergone gastrectomies from the 2012-2014 ACS-NSQIP database. A multivariate cox-regression model for readmission identified independent predictors of readmission. These factors were converted into a quantifiable readmission risk score (RRS) based on their hazard ratios. The RRS was calculated for each patient. Readmission rates within 30 days of surgery were compared across the spectrum of RRS.
Results: Of 4,064 patients (3,096 partial; 968 total gastrectomies; 49.7% for cancer), 475 (11.9%) were readmitted within 30 days of surgery, at a median interval of 6 days after discharge. 174 of the readmissions were for gastrointestinal/failure-to-thrive (FTT) issues (147 from partial vs. 27 from total, p=0.025). 163 readmissions were due to infection (121 from partial vs. 42 from total, p = 0.035). Extent of resection (total vs. partial), cancer diagnosis, operative time, age, and sex did not affect readmission rates (all p>0.05).
Independent factors associated with early readmission which were used to create the RRS included organ site infection (HR 4.61, p<0.001, 4 points in RRS), deep surgical site infection (HR 4.33, p<0.001, 3 points), postoperative myocardial infarction (HR 3.09, p=0.002, 2 points), superficial surgical site infection (HR 2.39, p<0.001, 2 points), postoperative venous thromboembolism (HR 2.46, p<0.001, 2 points), sepsis (HR 1.86, p<0.001, 1 point), and diabetes (HR 1.29, p=0.025, 1 point).
Of 2,734 patients with the lowest RRS 0, only 8.3 % were readmitted. 12.3% of 457 patients with RRS 1 were readmitted. 25.1% of 203 patients with RRS 2 were readmitted. 34.1% of 85 patients with RRS 3 were readmitted. 41.3% of 240 patients with RRS ≥4 were readmitted (p<0.001 across all RRS).
Conclusions: Post-gastrectomy infectious complications are the primary drivers of the readmission risk score. RRS 0-1 was associated with readmission rates close to the baseline level of 11.9%. However, RRS 2, 3, and ≥4, nearly doubled, tripled, and quadrupled the readmission rates from baseline, respectively. In the current healthcare climate which requires balancing the need to reduce readmissions while responsibly using limited resources, this novel scoring system can be used to focus post-discharge care on higher-risk gastrectomy patients with RRS ≥2.
Back to 2016 Annual Meeting
|
