Muhib Khan1, Freddie Hildreth3, Diana Haggerty2, Jay Knight3, Malgorzata Miller3, Jenny Tsai4, Victoria Scott3, Tricia Tubergen3, Elizabeth Evans5, Nadeem Khan3, Nabil Wees3, Asad Ahrar3, Jiangyong Min6, Michelle DeJesus Brazitis3, Laurel Packard3
1Neuroscience Institute, Michigan State University, 2Michigan State University, 3Spectrum Health, 4Spectrum Health West Michigan, 5Spectrum Health Medical Group, 6Spectrum Health Neuroscience
Objective:
We aimed to investigate predictors of 30-day same hospital readmission after stroke and incorporate these factors in a predictive model.
Background:
Hospital readmissions after stroke is costly and negatively impact outcomes.
Design/Methods:
Patients discharged with a primary diagnosis of ischemic stroke and intracerebral hemorrhage between 1/1/2018 and 12/31/2020 were included. Transient Ischemic Attack (TIA) patients were excluded from the analysis. LACE+ index, age, smoking status, primary care physician (PCP) status, type of stroke, admission National Institute of Health Stroke Scale (aNIHSS), thrombolysis and thrombectomy use were compared between patients with or without unplanned 30-day readmission. Logistic regression was used to generate readmission predictive models.
Results:
A total of 3319 patients were identified. Of these 196 TIA patients were excluded from the analysis. Mean age was 67.8 ± 15.27 years. Readmission rate was 9.5% for the whole cohort (ischemic stroke 7.7% and hemorrhagic stroke 14%). Readmitted patients had a high LACE+ score (≥ 60; 73%), low aNIHSS (0-5; 50%), were smokers (61%) and did not have an established PCP (94%) at the time of index admission. Readmitted patients were also less likely to have received thrombolysis (88%) or thrombectomy (80%) during index admission. Logistic regression incorporating LACE+, aNIHSS and stroke type (hemorrhage) were mildly predictive of 30-day unplanned readmission (Model A; c-stat 0.60). Addition of smoking status, alcohol status, primary care physician status, thrombolysis and thrombectomy use slightly improved the predictive ability of the model (Model B; c-stat 0.62).
Conclusions:
LACE+ is marginally predictive of 30-day unplanned readmission after stroke. Better scores incorporating socioeconomic, neurological and cerebrovascular variables are needed to predict readmission after stroke.