A. M. Khokar1, V. M. Plant1, J. H. DeAntonio1, W. B. Rothstein1, B. C. Ruch1, J. D. Bennett1, L. G. Wolfe1, B. Kaplan1, S. Jayaraman1 1Virginia Commonwealth University,General Surgery,Richmond, VA, USA
Introduction: Readmissions are costly and often preventable. Identifying patients at high risk for readmission can allow for more targeted interventions. Validated readmission risk calculators such as LACE (Length of stay, Acuity, Comorbidity, Emergency department visits) and American College of Surgeons National Surgical Quality Improvement Program (NSQIP) are available but have not been compared in general surgery patients. One key difference between these two calculators is consideration of the type of procedure the patient is undergoing. The NSQIP calculator requires a CPT code, age, gender, comorbidities, and acuity of case whereas LACE only includes more general variables. We hypothesized that NSQIP is better at identifying high risk general surgery patients who are likely to be readmitted than LACE.
Methods: In this retrospective case control study, all patients who had an elective operation (bariatric, GI, and surgical oncology) at a tertiary referral center over 6 months in 2016 were identified. Of these patients, patients who were readmitted within 30 days of discharge and an equivalent number of non-readmitted patients were randomly selected as controls for comparison. LACE and NSQIP were used to determine risk categories and correlated to proportion of patients readmitted. Patients were divided into low (0-5%), moderate (5.1-12.1%), and high risk (greater than 12.1%) based on LACE. Fisher exact test and logistic regression were used to determine statistical significance.
Results: Of 345 patients, 26 patients were readmitted (7.5%) and in this group, LACE identified 7.4%, 52%, and 41% as low, moderate, and high risk, respectively and NSQIP identified 7.4%, 37% and 56% as low, moderate, and high risk, respectively. NSQIP had a linear relationship and identified a greater proportion of readmitted patients as high risk compared to LACE (p=0.011; p<0.05). NSQIP was more likely to identify readmitted patients as high risk than moderate risk compared to LACE (OR 8.4 with 95% CI 1.5 to 46.1, p=0.0008 vs OR 3.0 with 95% CI 0.8 to 11.8, p=0.0488). Of the controls, LACE identified 19%, 65%, and 15% as low, moderate, and high risk, respectively, whereas NSQIP identified 50%, 38%, and 12% as low, moderate, and high risk, respectively. Again, NSQIP had a linear relationship and identified low risk patients better than LACE but this was not statistically significant (p=0.63).
Conclusion: The NSQIP risk calculator appears to identify general surgery patients at high risk for readmissions better than LACE which may be too general for this patient population. Both appear to be similar at identifying low risk patients. Prospective studies comparing these calculators will be essential in the future to address postoperative readmissions.
