We showed that a simple intervention based on encrypted digital signature as a trigger of events in an EHR could reduce the time to discharge after medical order around a median of 7 hours by reducing clerical work. Furthermore, this change was immediate and sustained after the intervention. Finally, a slight and persistent decrease in the number of patients diverged for ED waiting for an available bed on the floor.
Reducing boarders in the ED is a complex problem. The phenomenon is associated with several causes that assume different intensities depending on the institution studied. Therefore, identifying the leading causes to solve is paramount. In our institution, unscheduled hospital admission indicated to HOC patients was a workable BPM evaluation problem. The main problem was guaranteeing safety discharge after medical order according to Brazilian legislation regarding clerical work. Before 2017, Brazilian law obliged healthcare personal to ensure that everything was correct before the patient left the hospital. After 2017, Brazilian regulations started to accept the encrypted digital signature as a valid alternative. We implemented this solution to all healthcare personnel and used it as the primary intervention in this project.
This intervention had an immediate and sustained effect, reducing 7 hours of the hospital bed turnover. In 2013, The Advisory Board Company (www.advisory.com) estimated a reduction of 0.25 days in the duration of hospital stay for an institution of 600 beds would provide an extra 25 beds available in 24 hours. This amount of beds is what we should expect to accommodate the unscheduled hospital admissions of HOC patients.
Even though the reduction in time to discharge was immediate, the same did not occur to the number of patients diverted to the ED. There could be several explanations for this finding. One of those is an institutional culture (a "reflex") to diverge patients to ED instead of looking for an available bed on the floor. In our institution, each specialty has some designated beds for their use. The "new" bed that the intervention provided was not always one specialty with a patient needing in-hospital admission. We corrected that by implementing an allocation beds central dashboard and guaranteed the teams that "their" beds would be available again in 24 hours if they "lend" it to the specialty in need. Another reason is that some patients must require intensive care unit, positive or negative isolation beds (for cases of tuberculosis or immune depressed conditions, respectively), which are more scarce in the institution.
Besides the impact of expediting the patient discharge, shifting to the left of the leaving hour is significant. Our patients come from small cities around and depend on working hour transportation for going home after discharge. If the discharge occurs after 17h00, there is no transportation, the patient has to remain in the hospital up to the following day, and we could not use the bed for another patient waiting in the ED. This problem is aggravated during weekends when this transportation is not available. We attributed the bimodal distribution before the intervention to the nurse duty shifty. The shift to the left and assuming a normal distribution after the intervention corroborates this. Other authors showed similar findings[10, 11].
Our approach to avoiding boarder by expediting discharge after medical order is a pull forward strategy since the available bed will "pull" waiting for patients to the floor. Another process recently reviewed in the literature is a push-forward one, represented by the transfer of patients to a lounge on the floor, so the floor staff will try to guarantee an available bed. This push-forward strategy still lacks definitive evidence of effectiveness. Our pull-forward approach seems more logical since it addresses the floor staff's needs instead of putting pressure on them. We acknowledged that our strategy steel requires further studies for a definitive cause-effect relationship. Still, we should consider that differences among institutions, including their culture, should be considered while searching for solutions to avoid boarders in ED.
Our findings are following others. Mustafa et al. demonstrated that the patients diverged to ED from HOC who indicate an unscheduled admission are an essential contributor to crowding. Furthermore, patients with delayed discharge occupied up to 15% of the hospital beds. Muhammad et al. presented similar results, emphasizing the role that elderly patients represent in this situation. Silva et al. presented similar data in Brazil and pointed to the role that teaching hospitals such as ours could have.
Ideally, following patients' flow in a thorough output model, including more active measures such as a dedicated person, would bring more immediate results in reducing boarders. Nevertheless, our institution has limited resources, and the solution implemented is very cost-effective. After the present study's findings, as previously mentioned, we implemented a dedicated central to follow up patient flow, including a hospitalist that can intervene proactively to reduce LOS[16, 17]. Newer strategies such as discrete events simulation could work in tandem with BPM to evaluate an intervention's impact before studying it in the field.