Development and prospective validation of a model estimating risk of readmission in cancer patients. Issue 6 (26th February 2018)

• Record Type: Journal Article
• Authors: Schmidt, Carl R.; Hefner, Jennifer; McAlearney, Ann S.; Graham, Lisa; Johnson, Kristen; Moffatt‐Bruce, Susan; Huerta, Timothy; Pawlik, Timothy M.; White, Susan
• Is Part Of: Journal of surgical oncology. Volume 117:Issue 6(2018)
• Page Start: 1113
• Page End: 1118
• Publication Date: 2018-02-26
• Subjects: cancer -- hospital readmission -- statistical model; Cancer -- Surgery -- Periodicals; Neoplasms -- Periodicals; 616
• DOI: 10.1002/jso.24968
• Languages: English
• ISSNs: 0022-4790

Abstract

Abstract : Introduction: Hospital readmissions among cancer patients are common. While several models estimating readmission risk exist, models specific for cancer patients are lacking. Methods: A logistic regression model estimating risk of unplanned 30‐day readmission was developed using inpatient admission data from a 2‐year period ( n = 18 782) at a tertiary cancer hospital. Readmission risk estimates derived from the model were then calculated prospectively over a 10‐month period ( n = 8616 admissions) and compared with actual incidence of readmission. Results: There were 2478 (13.2%) unplanned readmissions. Model factors associated with readmission included: emergency department visit within 30 days, >1 admission within 60 days, non‐surgical admission, solid malignancy, gastrointestinal cancer, emergency admission, length of stay >5 days, abnormal sodium, hemoglobin, or white blood cell count. The c‐statistic for the model was 0.70. During the 10‐month prospective evaluation, estimates of readmission from the model were associated with higher actual readmission incidence from 20.7% for the highest risk category to 9.6% for the lowest. Conclusions: An unplanned readmission risk model developed specifically for cancer patients performs well when validated prospectively. The specificity of the model for cancer patients, EMR incorporation, and prospective validation justify use of the model in future studies designed to reduce and prevent readmissions.