New prediction tool-LIST-with improved prediction accuracy for 30-day readmission rates in patients with head and neck cancer after major cancer surgery.

Abstract We hope to establish a new readmission prediction score for patients with head and neck cancer (HNC) after major cancer surgery. A retrospective cohort study was conducted from the clinical and cancer registry data at Kaohsiung Veterans General Hospital. We included the data of patients with newly diagnosed HNC who underwent surgical treatment between Nov 2010 and Dec 2017. Multivariate logistic regression was performed to determine independent factors for 30-day readmission rate and establish a new prediction score. We compared the discriminatory ability of our new prediction score, HOSPTIAL score, and LACE index using linear trend chi-square test, the Akaike information criterion (AIC), and c-statistic. The 487 patients with HNC who underwent major surgery were discharged from the medical center. Of these patients, 40 (8.2%) were readmitted to the same hospital within 30 days. Our prediction score, namely LIST (representing leukocytosis, Charlson comorbidity index score of > 0, length of stay of top 33% for the total population, and advanced tumor stage) was derived through multivariate logistic regression. Compared with the HOSPITAL score and LACE index, LIST had a higher linear trend chi-square value (27.8 vs 4.3 and 6.3), higher prediction accuracy (0.743 vs 0.586 and 0.589), and lower AIC value (251 vs 274 and 272). The LIST can estimate 30-day readmission rates in patients with HNC. More intensive discharge planning and transition of care along with patient education can be applied to this high-risk group before discharge.