MP31: Safely reducing emergency physician admission rate through audit and feedback

Abstract

Background: Most admissions to hospitals occur through the emergency department (ED). The impact of emergency physicians’ decisions to admit a patient to hospital can have wide ranging effects on health care spending, hospital congestion and patient outcomes. A growing body of evidence shows that outpatient management of conditions such as diverticulitis, heart failure and pulmonary embolism is both safe, effective and can reduce costs. Aim Statement: To support emergency staff in making safe, informed decisions to appropriately reduce admission rates without increasing the rate of patients returning and being admitted. Measures & Design: Significant variability in admission rates between emergency physicians exists and no correlation between actual and self-reported admission rates is observed. One means to change behavior is through audit and feedback, however a Cochrane review on this topic concluded that it was only effective if specific conditions were met; findings which were incorporated into this project. An audit tool was created comparing individual physicians’ admission and “bounce back” rates to their peers. The tools contained averages for the individual and site for admission and bounce back rates and were shared with physicians every 2 months. Physicians were divided into three equal groups, low, medium and high admitters and targets established. Department heads met with high admitters. Evaluation/Results: The project was started in September 2016. Admission rates in the three physician groups were compared in the ten months before September 2016 (prior) and after January 2017 (post). September to December 2016 was considered the “rollout” period and not included in the analysis. Significance was tested using a Permutation test and a p-value cut off level of 5%. Nine emergency departments took part. Seven sites experienced a significant decrease in the admission rate of top admitters, three showed a significant increase in the rate of low admitters and two showed a significant increase in the rate of medium admitters. Pooled results showed a decrease in the admission rates of the top admitters and no significant change to the medium or low admitters. Discussion/Impact: Comparing the pre- and post-periods yielded a decrease in admissions of 773 patients on an annualized basis. The impact of the change in the top five highest admitters at the biggest three hospitals estimated an annualized beds savings of 25.3 beds.