An integrated approach to computer-based decision support at the point of care
Lack of adequate access to information among clinicians has been identified as prominent source of medical error. The author has investigated this problem systematically over the past few decades. His research has evolved from understanding information needs in general (and how those needs might be resolved through evidence-based practice), to designing applications for clinical information systems that link the clinician to information resources. His applications have evolved from a simple embedded "Medline Button" (renamed to "infobutton" with the advent of the World Wide Web), to the current "Infobutton Manager" (IM), a context-dependent application that attempts to match the clinician's and patient's characteristics, the task being performed, and the information needs with resources. This paper examines the IM technology, as well as the attendant user experiences and lessons learned.
Identifying Clinician Information Needs - methods including focus groups and surveys were abandoned due to discordance among the results. The author then transitioned to using a "portable usability lab", relying on keystroke and videotape transcription analysis to detect information needs. The needs were then coded according to a classification that characterizes: type of information need, the likely resource for resolving the need, and the subjects success in resolving the need.
Each information need discovered the the process noted above was then represented with a question ("what are the guidelines for heparin use") and a Web link that corresponded to the resource chosen to resolve the question. When the link was clicked by the user, the IM was invoked, allowing transmission of data such as patient age & gender, clinician's task, user profession/title, among others.
The IM then matched the contextual information against its database of previously generated data questions, then presented its output (a ranked list of questions) to the user as Web links. The primary of evaluating the use of the IM was through review of the log files. An option for non-context specific information searching was left available on the clinicians's screen as a usage comparison with the Infobutton Manager. The IM application uses subscription information resources such as Harrison's and Up To Date to avoid the issue of outdated information.
Observation of Clinician Information Needs - 251 subjects were observed, generating 250 information needs. Of those requiring resolution through use of an on-line information source, 101 questions have been created with appropriate Web links.
Over 2,700 user accessed information resources 28,519 times. Actual IM usage was defined as user selection of one of the prompted questions, after invoking the IM button. The IM usage was highest while reviewing inpatient drug orders (63-78% of the time), and lowest when reviewing diagnosis lists (10-24%). User feedback: 89% felt the system was easy to use, 74% felt IM has a positive impact on care, and that IM was helpful 77% of the time.
IM represents a context-specific application that provides information resources at the point of care. The system has very positive usability features, and though utilization was found to vary according to the nature of the clinician's activity, an overall increase in information resource usage was seen in the institution.
The author has created a usable and valuable tool to facilitate information retrieval at the point of care. Whereas some clinicians may have the ability to perform a independent search on Medline or other Web service, many lack the skills or time to perform such a task. For those fortunate enough to have access to the Infobutton Manager, quality information retrieval has been greatly facilitated.
Christopher Tessier, MD BMI512 Fall 2008