Communicating Narrative Concerns Entered by RNs (CONCERN)

项目总结/文摘

PROJECT SUMMARY/ABSTRACT The Communicating Narrative Concerns Entered by RNs (CONCERN) system is an expert-knowledge, machine learning, predictive model (CONCERN-PM) that produces an early warning score (CONCERN-EWS) and is implemented as clinical decision support (CONCERN-CDS) to alert nurses and physicians about hospitalized patients who are at risk of decompensation. CONCERN-PM leverages nursing documentation patterns as a proxy measure of nurses’ decisions to increase surveillance and related interventions which are an indicator of nurse concern about hospitalized patients – a key early indicator of decompensation. This five- site observational implementation study is a renewal to our successful National Institute for Nursing Research (NINR) funded R01, in which we developed and evaluated the CONCERN (Communicating Narrative Concerns Entered by RNs) predictive model and clinical decision support (CDS) for hospitalized patients in a multi-site randomized controlled clinical trial, with interim results of a 27% reduced likelihood of inpatient mortality for patients whose clinicians received the CONCERN CDS intervention versus those who did not. CONCERN-CDS alerts clinicians that a patient is at risk 42 hours earlier than leading EWS1. In this renewal R01, we will conduct a five-site observational implementation study to evaluate fairness and population validity of the CONCERN-PM, ecological validity of the CONCERN-CDS, and explore the contribution of caregiver generated data to the performance of the CONCERN-PM as the next phase in our multi-site nursing data science and implementation science program. EWS are in broad use at health systems nationally, however, we have identified a lack of research findings demonstrating external validity and fairness in other leading EWS.2 We will leverage 5 diverse study sites to test fairness, population validity, and ecological validity across rural, suburban, and urban settings. The evaluation will be informed by the RE-AIM (Reach, Effectiveness, Adoption, Implementation and Maintenance) and CFIR ( Consolidated Framework for Implementation Research) implementation science frameworks. The specific aims of this R01 Renewal are to: Aim 1. Translate the CONCERN-PM to 2 additional specialty patient populations (emergency department and inpatient pediatrics) and with the addition of social determinants of health (SDOH) data to evaluate model calibration and performance across different patient populations and health systems; Aim 2. Evaluate CONCERN-CDS implementation across all adult inpatient units at 5 health systems, in inpatient pediatric units at 3 health systems and in the emergency department at 2 health systems to understand health system and patient population characteristics associated with a successful implementation using RE-AIM and CFIR Frameworks, and Aim 3. Using a patient-centered design approach, explore the creation of a patient-facing CONCERN-PM and CDS for caregivers of inpatient pediatric patients, to compare predictive model performance with additional caregiver feedback to the clinician-only data-driven model.