Researchers: Natalie Flaks-Manov¹, Einav Srulovici^1,2, Calanit Key¹, Henia Peri-Mazra¹, Moshe Hoshen¹

Background: Increasingly, big- data electronic health record (EHR) warehouses are used for developing and implementing high-risk identification algorithms for targeted hospital readmission prevention programs (RPPs).
However, the ability of these computerized readmission risk score (RRS) to accurately detect the "appropriate" patients for RPP according to personal and clinical characteristics (termed "care-sensitivity") has not yet been established.

Objectives: To evaluate the concordance between RRS and team scoring with regards to patient selection for RPP.

Method: This was a mixed-methods qualitative-quantitative design. Three focus groups comprised of managers, medical and nursing teams from community and hospital settings were conducted. A quantitative survey was conducted on 15 internal medicine wards' teams in hospitals and roughly 150 community clinics. Teams were asked to assess the degree to which each patient’s RRS was care-sensitive and which patient should be included in readmission prevention programs. These associations were examined with multivariable logistic regression models.

Findings: Four themes were identified. Questionnaires were filled for 373 patients, of whom 24.4% had a readmission within 30 days following their hospital discharge. In 34% of patients, the computerized readmission risk scores differed from the team's readmission risk assessment. Twelve percent of patients, who had a low RRS, were assessed by hospital nursing teams as being at high risk for readmission; 33% of these had a readmission. In all models, patients' behavioral characteristics were found significantly associated with selection for RPP (OR=2.4-3.3, 95%CI: 1.24-6.98).

Conclusions: Combining EHR with insight from teams regarding patients' clinical and personal characteristics provides more "care-sensitive" information, which allows better adaptability and synchronization across different healthcare providers, along with a better selection of patients for inclusion in RPPs.

Recommendations: The clinical and personal data collected from the teams' assessments will be used to improve the predictive tools for improving the ability to identify patients at high risk for readmissions.

Research end date: 09/2018