Preventing Hospital Readmissions: Healthcare Providers’ Perspectives on “Impactibility” Beyond EHR 30-Day Readmission Risk Prediction

Abstract

BACKGROUND: Predictive models based on electronic health records (EHRs) are used to identify patients at high risk for 30-day hospital readmission. However, these models' ability to accurately detect who could benefit from inclusion in prevention interventions, also termed "perceived impactibility", has yet to be realized. OBJECTIVE: We aimed to explore healthcare providers' perspectives of patient characteristics associated with decisions about which patients should be referred to readmission prevention programs (RPPs) beyond the EHR preadmission readmission detection model (PREADM). DESIGN: This cross-sectional study employed a multisource mixed-method design, combining EHR data with nurses' and physicians' self-reported surveys from 15 internal medicine units in three general hospitals in Israel between May 2016 and June 2017, using a mini-Delphi approach. PARTICIPANTS: Nurses and physicians were asked to provide information about patients 65 years or older who were hospitalized at least one night. MAIN MEASURES: We performed a decision-tree analysis to identify characteristics for consideration when deciding whether a patient should be included in an RPP. KEY RESULTS: We collected 817 questionnaires on 435 patients. PREADM score and RPP inclusion were congruent in 65% of patients, whereas 19% had a high PREADM score but were not referred to an RPP, and 16% had a lowmedium PREADM score but were referred to an RPP. The decision-tree analysis identified five patient characteristics that were statistically associated with RPP referral: high PREADM score, eligibility for a nursing home, having a condition not under control, need for social-services support, and need for special equipment at home. CONCLUSIONS: Our study provides empirical evidence for the partial congruence between classifications of a high PREADM score and perceived impactibility. Findings emphasize the need for additional research to understand the extent to which combining EHR data with provider insights leads to better selection of patients for RPP inclusion.

Predicting preventable hospital readmissions with causal machine learning

“…8,[14][15][16] Ostensibly, this assumption has intuitive appeal, given that higher-risk patients appear to have "more room to move the needle," but it is not guaranteed to hold in practice 17,18 , especially in the context of readmissions 7,19 and other settings where treatment effect heterogeneity may exist. 18 The need for analytical approaches that estimate patient-level benefit-referred to in some contexts as impactibility [20][21][22][23] and falling under the umbrella of precision medicine more generally-is beginning to be recognized, particularly for readmission reduction programs. 22 However, the distinction between benefit and risk may currently be overlooked in the development and application of risk assessment tools.…”

“…[26][27][28] Few, if any, analytical approaches to identify "care-sensitive" patients, or those whose outcomes may be most "impactible," currently exist, despite a clear need for such methods. 20,23 Existing approaches based on off-the-shelf supervised machine learning methods, despite their flexibility and potential predictive power, cannot meet…”

“…The need for analytical approaches that estimate patient‐level benefit—referred to in some contexts as impactibility 20‐23 and falling under the umbrella of precision medicine more generally—is beginning to be recognized, particularly for readmission reduction programs 22 . However, the distinction between benefit and risk may currently be overlooked in the development and application of risk assessment tools.…”

LACE Score-Based Risk Management Tool for Long-Term Home Care Patients: A Proof-of-Concept Study in Taiwan

“…There is an emerging trend of developing predictive models. By introducing abundant variables in EMR database [43], collecting social determinants of health [44], and using sophisticated machine learning methods can build a surprisingly accurate but complex readmission prediction model [2,16,45,46]. There is always a huge information gap between the transition from hospital to home, and missing data are not uncommon for home care patients.…”

Can targeting high-risk patients reduce readmission rates? Evidence from Israel

“…Structured follow-up by the primary care clinics' staffs (physicians and/or nurses) included telephone, clinic-based, or home care needs assessments, self-management support, medication review, and referral to services. Details of this intervention are also discussed in Flaks-Manov et al (2020).…”