Related Experiment Videos
Effect of an Intersectoral Telemedicine Intervention on Hospitalization in Nursing Home Patients: Specification Curve
John Grosser1, Sophie Pauge1, Birthe Aufenberg1
1Department of Health Economics and Health Care Management, School of Public Health, Bielefeld University, Universitätsstraße 25, Bielefeld, 33615, Germany, +49 52110686319.
Background:
The use of emergency medical services in Germany has significantly increased, particularly for nonemergency cases. This issue becomes especially relevant for nursing homes, where staff is confronted with resource shortages and organizational challenges, making emergency medical services involvement a common course of action. This leads to potentially avoidable hospitalizations for nursing home residents and exposes them to severe health risks, such as nosocomial infections. To address this, we developed the Optimal@NRW project, implementing an innovative intersectoral telemedicine intervention that includes an early warning system and mobile nonphysician medical assistants in order to enable outpatient treatment in acute medical cases for nursing home residents. We hypothesize that hospitalizations could be significantly reduced, as patients could be treated on-site in the absence of a life-threatening emergency.
Objective:
This study aims to determine the effect of an intersectoral telemedicine-based intervention for acute care situations on hospitalization among residents in nursing homes.
Methods:
We conducted a prospective multicenter cluster-randomized stepped wedge trial from May 2021 to April 2023 across 24 nursing homes in western Germany. Nursing homes were randomly assigned to one of four sequences and followed over eight 3-month periods. All sequences crossed stepwise from the control to the intervention phase. We collected primary data in an electronic patient record and claims data from statutory health insurance funds. In a specification curve analysis, we evaluated the intervention effect using multiple mixed-effects regression model specifications with varying modeling approaches, datasets, dependent and independent variables, and operationalizations.
Results:
The number of participants ranged from 158 to 285 nursing home residents per sequence and period, totaling 1151 participants in the control phase and 1271 participants in the intervention phase. A total of 1260 regression models were analyzed. Approximately two-thirds of all models exhibited a negative (ie, protective) intervention effect. However, no models exhibited a statistically significant (ie, P<.05) intervention effect. The distribution of effect direction and size differed by modeling approach, choice of outcome, and dataset used.
Conclusions:
Our results do not support the hypothesis of a significant reduction of hospitalizations among nursing home residents. The effectiveness of the intervention may have been compromised by external factors such as a flood disaster in the study region and the COVID-19 pandemic, which introduced unforeseen challenges. Future studies should prioritize targeted outcomes (potentially avoidable hospitalizations) that align with the specific intervention mechanism, while also allowing for extended implementation periods to ensure that interventions reach their full potential.
