Electronic Health Records: Incorporating evidence-based information into clinical decision making (2023)

Abstract

The authors have developed two tools aimed at enabling clinicians to review alternative diagnoses and improve access to relevant evidence-based library resources without disrupting established workflows. The Diagnostic Decision Support Tool extracts terms from standard coded fields in electronic health records and sends them to Isabel, who generates a list of possible diagnoses. Doctors select their diagnosis and view "Knowledge Pages," a collection of evidence-based library resources. Each resource automatically populates search results based on the selected diagnosis. Doctors respond positively to knowledge pages.

show

Integrating evidence-based information into patient care requires the right information at the right time1The appropriate time may be during the diagnostic process, during patient care at the bedside, or at another point in the investigation and treatment continuum. Physicians, already under pressure to treat more patients in less time, are less likely to seek information to inform their decisions when the search disrupts their workflow2One way to remedy this situation is to provide a clinical decision support system (CDS) with links to relevant evidence-based information resources that can be accessed directly from the electronic health record (EHR) via the information button. "Research on information buttons shows that doctors find them useful and that specific information about a specific patient condition is more useful than general information."3. CDS systems are "active knowledge systems that use two or more sets of patient data to generate recommendations on a case-by-case basis"4Information buttons are defined as “links between clinical information systems and online knowledge resources”.5.Integrating CDS resources into existing clinical workflows through the information button has the potential to reduce diagnostic or treatment errors by providing evidence-based information specific to the specific clinical scenario.

This report describes a pilot project to integrate a CDS tool that integrates evidence-based and context-sensitive information resources into an EHR, supporting the Emergency Department (ED) of a 250-bed pediatric hospital. As with previously reported library efforts to integrate CDS tools into EHRs, the authors focus on providing library content at the point of care6,7.This project builds on previous work by adding a method to extract patient data specific to ED visits directly from the EHR with the click of an information button. The result is a list of possible diagnoses from which an automatic search can be made for relevant diagnoses.Production of information. Previous attempts did not include a list of possible diagnoses and search queries had to be entered manually by the user. The goals of the project described here were (1) to provide physicians with a simple mechanism to review alternative diagnoses before finalizing a treatment plan and (2) to significantly improve access to relevant evidence-based information resources.

Develop dedicated clinical decision support tools

A new Chief Medical Information Officer at the hospital wants to integrate CDS tools into EHRs in the EU and sees the library as an important part of the project. He appointed a committee of two clinical librarians, the medical librarian and assistant librarian, an information systems coordinator (ISC) who oversees the technical aspects of EHR technology in the emergency department, and two physicians specializing in computer science programs. Create a CDS and integrate with your Emergency Department EHR. The main questions addressed were: (1) finding a vendor willing to work with Wellsoft, the then-current EHR software; (2) decontextualizing data from patient records; (3) providing a mechanism for physicians to review alternative diagnoses; (4) ) provides seamless access to contextual information resources without interrupting the workflow. (5) promote the tool to users and train them in its use; and (6) measure the success of the tool.

EBSCO, UpToDate and Isabel were approached as potential partners for our project. Isabel was selected based on her previous experience and willingness to work with Wellsoft. Isabel is a diagnostic checklist tool that helps physicians expand their differential diagnosis and identify diseases at the point of care8.

In collaboration with Isabel and Wellsoft, ISC created a Diagnostic Decision Support Tool (DDST) and integrated it into the EHR. The tool extracts terms from standard coded fields in records of age, sex, chief complaint, screening impression, positive results and current medical history and sends these terms to Isabel, who then performs the search and provides a possible diagnosis (illustration 1).

Isabel provides an online Knowledge Base (KP) page for each diagnosis, offering introductory textbook chapters, the ability to perform a comprehensive PubMed search, and links to medical library websites. Isabel sends this CP to the librarians on her team and asks for feedback. Because of librarians' extensive training and experience in the clinical application of evidence-based information and in facilitating access to library resources, medical librarians requested additional resources, including clinical abstracts from UpToDate and First Consult, articles classified by level of evidence from TRIP and images from Google Images. Public health information from the Centers for Disease Control and Prevention and consumer health information from MedlinePlus. The librarian did not find the generic link to the library website very useful and requested that it be replaced by a personalized email link to the librarian. In addition, the PubMed search has been modified to focus on recently published research and case studies. All of these changes were made prior to the start of the pilot.

When physicians view patient records in the EHR, they see the DDST tab. When they select the tab, a window appears with a list of possible diagnoses from Isabel (illustration 1). At this point, they can select one of the diagnostics listed and click Diagnostics to launch KP. Words or phrases describing Isabel's diagnosis were incorporated into the Uniform Resource Locator (URL) and used to search all KP resources simultaneously. Doctors will see a new page listing each resource in a column on the left. The full results for each clicked resource are in a larger column on the right that fills the rest of the page. From there, they can select which search result or resource they want to take a closer look at, and that result will populate the column to the right.

to apply

The pilot program started in October 2011 and lasted until April 2012. To introduce and evaluate the new tool, presentations were made to members of the EU department (including attending physicians and researchers) before, during and after the pilot. program. These meetings are agreed with the sector secretaries as part of the regular weekly EU meetings. This includes a demonstration by the DDST and KP through a PowerPoint presentation showing point-and-click screenshots of the user experience of Isabel's EHR launch at each resource listed in KP. Participants are then asked to provide feedback.

In addition, one of the authors (Yaeger), with the support of the chief resident, approached attending physicians and residents at the EU physician workplace to demonstrate DDST and KP at the point of care. Vendor-prepared reference cards and librarian business cards were distributed. Step-by-step instructions on how to use the DDST and KP will be presented at meetings of the Emergency Preparedness Team, the EU Department, the Board of Education and the Resident Luncheon.

A total of 25 EU visits were made, 5 presentations were made and 3 focus group discussions were held. He contacted 150 of the 200 doctors directly.

Measurement

A three-question survey assessing the impact of Isabel's list of possible diagnoses on diagnosis, counseling, and treatment has been included to the right of the list of diagnoses. Findings and visits are documented by the ISC. Processing time from initiation of DDST in the EHR to submission of a potential diagnosis by Isabel was determined. It also recorded the number of times a physician initiated DDST and the number of times a given patient was administered DDST. The number of visits to the KP is not recorded. The librarian (Yaeger) made field observation notes and physician comments during presentations, individual EU visits, and focus group discussions.

Result

From October 2011 to April 2012, 167 DDSTs were initiated for 125 individual patients out of a potential 34,000 patients.

Two users participated in the survey. Both respondents confirmed that Isabel's list of possible diagnoses influenced their differential diagnosis.

The seven physicians who participated in the focus group responded negatively to Isabel's list of possible diagnoses, but expressed interest and enthusiasm for KP. They strongly recommend redesigning DDST to integrate diagnostics from the EHR and send them directly to KP, bypassing Isabel entirely.

Several users outside the focus group also showed great interest in KP, stating that while they were familiar with most of the resources, they had little memory of using them themselves. For most of them, this was also their first experience with TRIP and MedlinePlus.

Processing time from DDST initiation to Isabel's probable diagnosis was 24.5 seconds (+/-3). I see no interest in transferring the list of possible diagnoses to KP, and I am annoyed by the processing time from DDST to Isabel.

wrangle

The doctors in the focus group explained that they did not need Isabel's help to make a correct diagnosis. They like KP because the search results are automatically generated and contain resources they wouldn't normally think they would use.

The user has shown no interest in selecting a diagnosis from Isabel's pooled list of possible diagnoses. Depending on the diagnosis you choose, you will be directed to automatically generated information search results based on data in KP. Doctors hesitate to select a diagnosis from the list because their own diagnosis is often not listed or the suggested diagnosis is too broad. For example, in the case of a patient with a broken elbow, Isabelle offers the possible diagnosis of a broken arm. A possible reason for underdiagnosis and overdiagnosis is that the list was provided at the time of the study when the physician had only partially completed the patient chart. The accuracy of the differences is only as good as the data that DDST retrieves and sends to Isabel. The 24.5 seconds (+/-3) wait time between running DDST and Isabel generating a result was considered a waste of time and prevented use.

The low survey response rate may be due to spatial and temporal placement. In fact, it appears on the bottom right of Isabelle's page of possible diagnostics. When viewing the page, users focus on the list of diagnoses, so many ignore the survey. Temporarily, the survey appears before the user has a chance to access KP.

after all

Integrating evidence-based information resources into bedside workflows requires ongoing effort and time. Physician feedback received as part of this project will help make library resources accessible, mobile, and usable in the clinical setting. By conducting this pilot project, the authors gained a better understanding of the challenges of integrating CDS tools into EHR and physician workflows in the European Union. The goal of providing physicians with an immediate mechanism to review alternative diagnoses before finalizing a treatment plan has been partially achieved. The authors were able to create the DDST, integrate it and link it with the Isabel differential diagnosis tool. However, for the DDST to be effective, the tool needs to be redesigned to extract more patient data and provide results more quickly.

In focus groups and live demonstrations in the EU, physicians were asked to retrieve their diagnoses from the EHR and forward them directly to KP. Unfortunately, their diagnoses often appear in the Notes section of the EHR, which lacks a standard coding field. After the pilot, the authors added a dialog box to the EHR for physicians to enter search terms submitted to each resource in KP via embedded URLs.

Findings from this project show that physicians appreciate one-click access to search results, which are automatically populated by KP with the diagnosis in multiple evidence-based library resources. Bernard Becker Medical Library staff developed a stand-alone KP that could be easily integrated with other EHRs used throughout the medical center. The authors plan to continue this work by testing stand-alone KP with additional EHRs, giving physicians what they want: one-click access to evidence-based resources.

footnote

^*Based on a presentation at MLA '12, the 112th Annual Meeting of the Medical Library Association, Seattle, Washington.

Report