Skip to main content
NCBI home page
AMIA Annual Symposium Proceedings logoLink to AMIA Annual Symposium Proceedings
. 2017 Feb 10;2016:696–704.

i3b3: Infobuttons for i2b2 as a Mechanism for Investigating the Information Needs of Clinical Researchers

Timothy Kennell Jr 1,2, Donald M Dempsey 3, James J Cimino 2
PMCID: PMC5333310  PMID: 28269866

Abstract

The information needs of clinicians, as they interact with the EHR, are well-studied. Clinical researchers also interact with the EHR and, while they might be expected to have some similar needs, the unique needs that arise due to nature of their work remain largely unstudied. For clinicians, infobuttons (context-aware hyperlinks) provide a mechanism of studying these information needs. Here we describe the integration of infobuttons into i2b2, a popular data warehouse commonly used by clinical researchers, using a plugin. A preliminary survey of i2b2 developers suggests a general interest in infobuttons for i2b2 and indicates good likelihood for their deployment, where they may be used as a tool for further studying these needs in greater detail.

Introduction

Clinical information systems such electronic health records (EHRs), now widely available, address many of the information needs of their users,1,2 yet they simultaneously evoke many new information needs.3,4 The greater availability of EHRs has led to their use by a new constituency: clinical researchers.5,6 Similar to their clinical counterpart, these individuals often provide health care for their subjects as patients and therefore might be expected to have similar information needs. However, this new group of users would also experience different information needs based on their unique use cases, including managing the research workflow of their subject,7 responding to clinical problems that arise because of the research procedures, analyzing the data they collect across their cohort of patients to generate new knowledge,6,8 and using that knowledge to generate new research questions.9,10 It is our hypothesis that these tasks entail information needs that differ from those of routine patient care. However, while the information needs of clinicians have been widely investigated,11–14 an extensive search of PubMed and Google Scholar for studies on those of clinical researches fails to reveal research in this area.

One method for addressing information needs of EHR users is through the context-aware hyperlinks, known as “infobuttons”, that provide patient-specific, just-in-time knowledge at the point of care.15,16 When integrated into EHRs, infobuttons serve not only as a mechanism for decision support, but as a method for studying clinician information needs.17,18 As clinical researchers often access EHR data through clinical data repositories, we believe that integrating infobuttons into these repositories could serve similar dual purposes.

We have therefore developed a method for integrating infobuttons into a popular data warehouse (i2b2). We refer to this version of infobuttons as i3b3. The purpose of this paper is to describe the development of i3b3, explore the interest of the i2b2 community in having such a capability, provide instruction on how i2b2 installations can avail themselves of our software, and discuss the opportunities that i3b3 provides for exploring the varied information needs of clinical researchers.

Background

OpenInfobutton

OpenInfobutton is free-for-use, HL7-compliant infobutton manager provided by the University of Utah (http://www.openinfobutton.org/) that acts as an information broker between infobutton implementations and knowledge resources.14 The EHR infobutton passes relevant, contextual information about the patient, user and user’s status from the originating system using the HL7 standard to OpenInfobutton, which uses that information to select relevant information resources (for example, a drug dosing reference source). It then constructs links to those resources, customized with the contextual information (for example the drug and patient age and gender). The links are presented back to the user, who can then select those that appear to address his/her information need.

LITE

The selection of resources by OpenInfobutton, and the customization of links to those resources, is made possible by a knowledge base used by OpenInfobutton that contains institution-specific information provided by an analyst or librarian responsible for the infobuttons in the institution’s EHR. This customization is provide be a tool called the Librarian Infobutton Manager (LITE),19 which is also freely available from the University of Utah (http://lite.bmi.utah.edu/).

i2b2

The i2b2 (Informatics for Integrating Biology and the Bedside) system (https://www.i2b2.org/) is a popular patient data warehouse consisting of a repository for storing patient data and a user interface for rapidly querying the data for research cohorts that meet pre-selected criteria.20,21 This system was built by separating individual components into “cells” (interoperable modules) of a larger “hive” that communicate with each other via web protocols. This architecture allows the system to be highly extensible through the development of additional cells and plugins in order add desired functionality. In order to interact with the data repository in the cells, i2b2 provides two types of user interfaces (UIs): a workbench and a webclient. The workbench is a Java-based thin client that requires local installation and configuration on each computer that needs to interact with i2b2. The more common webclient is a browser-based UI built with standard web technologies that interacts with i2b2 via a web proxy server.

Methods

i2b2 Plugin Development

i2b2 can installed on many different server architectures and it’s functionality can easily be extended through cell or plugin development. Cell development allows the addition of new server-side functionality to i2b2 through communication with other cells in the hive, and can allow additional data to be retrieved directly from the data repository. While this provides extensive flexibility over data manipulation at the database level, cell installation in currently implemented i2b2 instances requires substantial modification to server configuration files. Additionally, user interaction with a custom i2b2 cell typically requires the development of an externally-facing plugin on the webclient as well.

Because we desired to make installation of our i3b3 infobuttons as simple as possible, we chose to develop a plugin. These can be installed into i2b2 by simply adding the plugin’ s directory structure to a standard location in the webclient’s web server and by changing a single configuration file.

In addition to i2b2’s standard plugin structure, we connected two other systems to it in order to establish communication between i2b2 and external knowledge resources. Briefly, i2b2’s data are sent via a HL7-compliant HTTP GET web request to OpenInfobutton. OpenInfobutton then generates a set of links to external resources based on prior settings configured in LITE. The links are then returned to the user as a list of uniform resource locations (URLs) displayed in a browser window for user interaction.

Information Needs Survey

In order to make a preliminary determination of the general interest in the integration of infobuttons into i2b2, we constructed a Google survey (http://goo.gl/forms/XuIaa5s3QW) (see Figure 1). As the survey was a preliminary information needs assessment, we kept the survey short and briefly reviewed the two included questions before finalizing it. The first question was simply, “Would you be interested in infobuttons for i2b2?” (yes/no). The second question was only revealed to participants if they responded “Yes” to the first question. In order to provide participants with a framework for possible locations that infobuttons could be integrated into i2b2, this question began with a screenshot of the i2b2 plugin that we developed. The question following the image, “Where would you like to integrate infobuttons into i2b2 (choose as many as apply):”, then provided both a list of checkboxes of logical locations to integrate infobuttons.

Figure 1.

Figure 1.

Open in a new tab

Information needs survey for assessing infobutton interest in the i2b2 community and desired locations for integrating infobuttons located at (http://goo.gl/forms/XuIqq5s3QW). The first question (Question 1) simply asks the respondent if they are interested in having infobuttons in i2b2 (yes/no). If they answer “yes”, the second question allows the respondent to detail where they would like infobuttons to be integrated. At the beginning, question 2 shows a simple screenshot of the plugin’s UI for reference and then provides several locations for integration including several concept classes present in i2b2 and a fill-in-the-blank response for alternative suggestions.

After completing and reviewing the survey, we emailed generated link to the survey to all recipients on the i2b2 mailing list and posted the link on the “i2b2 Install Help” Google group (https://groups.google.com/forum/#!forum/i2b2-install-help). We collected responses for the following 10 days followed by creating a summary of the data collected.

Results

Plugin Development and Use

Integration of infobuttons into the i2b2 system combines the plugin system, OpenInfobutton, and LITE. The plugin consists of two main components: a UI, built with HTML5 and CSS3, and a controller, built with JavaScript and Prototype (a JavaScript framework), each constructed according to the i2b2 plugin standard (https://community.i2b2.org/wiki/display/webclient/Web+Client+Plug-in+Developers+Guide). Figure 2 provides a schematic view of the components and processes involved.

Figure 2.

Figure 2.

Open in a new tab

i3b3 plugin architecture. Initially, a i2b2 user creates a patient set in i2b2 and loads the i3b3 plugin. The user can then drag the previously created patient set into the i3b3 frame (1) causing a list of patients and associated concepts (medications and diagnoses) to be displayed. Clicking on an infobutton next to any of the concepts opens a secondary browser window (resource window) and sends context-specific information to OpenInfobutton (2). OpenInfobutton queries to its resource knowledge base (created using LITE - the Librarian Infobutton Tailogin Environment) and generates a list of customized hyperlinks (3) that link to selected resources (4). The user can select one or more of the available resource links in the resources window to receive information to their initial question (5).

After selecting the plugin from i2b2’s list (see Figure 3), a user can select a concept class, such as diseases or medications, to investigate in a patient set and then drag a previously created patient set into a text box in the plugin window. This action will cause the controller to generate an AJAX (asynchronous JavaScript and XML) request directed toward the i2b2 data repository, which will respond with an XML message containing the requested information. After receiving the response, the controller parses through it to prepare it for display to the user. Once finished, the controller updates the UI with a list of the patients in the set and the respective information requested (i.e. a list of diseases or medications for each patient; see Figure 4).

Figure 3.

Figure 3.

Open in a new tab

Initial state of the i3b3 plugin after loading. A user can select the plugin from the list of i2b2 plugins in the “analysis tools” section [(A); shown enlarged in (B)]. The plugin is then loaded into the right panel (C) with brief instructions displayed at the top and more detailed instructions in a “Help” tab located at the top of the plugin panel.

Figure 4.

Figure 4.

Open in a new tab

Example of user interaction with i3b3 and the response provided. (1) A user selects a class of concepts that he/she desires to know more about from an available list (example depicts “Medications” chosen). (2) The user then selects a patient set from a previous query (shown enlarged in “A”) and (3) drags and drops it into the available space in in the plugin panel. (4) i3b3 will then return a list of the patients in the set and the associated concepts initially chosen in step (1). Each concept (in this case, medications, shown enlarged in “B”) has an associated infobutton that links to OpenInfobutton.

Next to each of the concepts listed below each patient, the controller places an infobutton icon that contains a URL with HTTP GET parameters to be passed to OpenInfobutton (see Figure 4). The parameters in the URL, which include the displayed concept, the patient’s age, and the patient’s gender as well as the institution (“UAB”) and the task (“i2b2”), are context-aware and translated from i2b2’s terminology into HL7-compliant terms. A user’s click on one of the infobutton icons sends a request to Openlnfobutton containing the previously mentioned information. The infobutton manager will then utilize a pre-configured, XML settings file (created with LITE) to select appropriate resources and construct links to those resources that include the appropriate parameter values obtained from the plugin. The links to each resource are combined into a final output in a secondary window. Each resource returns a result set on the original concept automatically refined by the concept’s original context. The result is context-specific links to knowledge resources based on the original concept that was selected (see Figure 5).

Figure 5.

Figure 5.

Open in a new tab

Context-aware links returned by OpenInfobutton after an i3b3 infobutton request is initiated. The response shown was generated by selecting the “albuterol” medication listed underneath a 19 year-old female after placing a patient set in the i3b3 plugin. The links (on left) shown above are for knowledge resources chosen in an example account in LITE. In addition to the list of links, OpenInfobutton initiates the first link (in this case, to PubMed Clinical Queries) and displays the results in the large frame on the right by simultaneously searching for the primary concept (“albuterol” in this example) and limiting the results to information relevant to the context passed to OpenInfobutton.

The finalized infobutton plugin has been distributed via our institution’s GitHub page found at https://github.com/uab-informatics-institute/i3b3.This release method, combined with the design of the system as an i2b2 plugin, allows for rapid installation in any i2b2 instance requiring only minimal instructions that are provided on the GitHub page. Additionally, GitHub provides a simple method for community contributions to future development of the plugin. Following construction and distribution, the plugin was easily installed by a system administrator in our institution’s i2b2 development environment with only the GitHub URL and confirmation of the i2b2 concept pathway structure provided. This highlights the fact that the i3b3 plugin remains independent of the actual ontology used in an i2b2 installation and needs only a minor setting modification in order to provide the location specific classes of concepts desired to be present for user selection.

Survey Responses

The distribution method of the survey allowed it to reach a wide audience from new to experienced researchers and clinicians along with several system analysts responsible for installing and maintaining i2b2 at their institution. The survey reached approximately 285 individuals through the “i2b2 Install Help” Google group and an unknown number through the mailing list. In total, the survey received 19 responses with all answering “Yes” to the question about interest in infobuttons for i2b2 (see Figure 6). In response to the question regarding desired infobuttons domains, participants requested infobutton integration for Diseases, Medications, Laboratory Results, and Procedures most frequently while others received a much lower frequency of requests. This trend may indicate the greatest, unmet information needs that arise in clinical researchers during interaction with i2b2.

Figure 6.

Figure 6.

Open in a new tab

Results of the i2b2 infobutton survey. There were 19 total responses to the survey. All respondents answered “Yes” to the first question asking about interest in infobutton for i2b2. The figure above shows the responses to the second question asking about location integration. General concepts such as diseases, medications, laboratory results, and procedures received the highest recommendations with most other options chosen much less frequently.

Most respondents did not take advantage of the “Other” category, suggesting that the majority of potential locations for infobutton integration were already present in the survey options. Of those participants that did choose the “Other” category, one simply suggested that all of the categories listed as answer choices would be useful and added that an infobutton for genotypic data would also be useful. Taken together, these results suggest the presence of unmet information needs in the i2b2 community, potentially extending to clinical investigators. Additionally, the survey responses indicate that these information needs may be further studied and potentially met with the use of infobuttons in applications that clinical investigators frequently interact with during the course of their studies.

Discussion

This paper describes an easy-to-adopt solution to the integration of infobuttons into any i2b2 installation. Coupled with the freely available OpenInfobutton and LITE, any i2b2 instance in the world can now provide its users with context-aware knowledge resource links and, based on our survey, it seems there is significant interest in doing so.

What links to provide remains an open question. The literature has few studies investigating the information needs of clinical scientists, and managers of i2b2 systems should not be quick to assume that the needs of clinical researchers will be the same as those of clinicians. More studies of clinical researchers’ needs will be required as studies of clinicians’ needs consistently show that those needs are task-dependent.22 While clinical researchers may have similar needs because they may have similar tasks (that is, caring for patients in their studies), they have additional tasks related to research, including adherence to study protocol, analysis and publication of data, and new hypothesis generation, all of which may be triggered by the same EHR data that trigger clinician needs. Those studies also consistently show that information needs cannot always be predicted without empiric observation. We conclude, therefore, that the information needs of clinical researchers should be a target of future study as part of the deployment of infobuttons. As with studies of clinician needs, infobuttons themselves (in this case, i3b3) can serve as a valuable tool in such studies.

The plugin we constructed was only developed and tested within the purview of a single hospital. However, we do not believe that our hospital environment and clinical research environment differ from other hospitals in ways that would impact i2b2 implementation, and we believe that our experience is applicable to many other medical institutions. Although our integration of infobuttons into i2b2 through a plugin was limited in the number of contexts in which information needs might be addressed, the results of our survey show that the initial scope chosen for the plugin covers some of the most important contexts, according to the i2b2 developer community. Furthermore, the design of the plugin easily allows for the scope to be extended in order to cover or exclude those information needs deemed important or irrelevant for future study respectively. While the plugin was developed for installment into a single data management system, i2b2, and might not be able to be easily installed in a different data warehouse, i2b2 is a popular option for installation at many hospital systems21 and i2b2 is not dissimilar to other data management systems that are available. Additionally, while the plugin does not incorporate group metrics for its context-aware searches, we believe that i3b3 provides an initial step in studying the clinical scientists information needs with group metrics being an additional feature that could be added as needed. Therefore, the results of this study, including the infobutton plugin, should be applicable to research institutions without i2b2 instances.

Conclusion

We have developed software that enables interested i2b2 developers (of which there are at least 19) to provide infobuttons to their users. OpenInfobuttton and LITE, also freely available, provide a comprehensive infobutton management strategy. What to do with infobuttons in i2b2 contexts is currently unknown, but the installation of i3b3 provides a mechanism for studying this question, much as infobuttons have been used to study information needs in clinical settings.

Acknowledgments

Mr. Kennell is supported by an NIH Medcial Student Training Program grant to the University of Alabama at Birmingham under grant 5T32GM008361-23 and a TL1 grant from the University of Alabama School of Medicine (UASOM) Center for Clinical and Translational Sciences (CCTS) under grant 1TL1TR001418-01 from the National Center for the Advancement of Translational Science (NCATS). Dr. Cimino is also supported by the CCTS NCATS grant and by research funds from the UASOM Informatics Institute. The authors thank R. Dale Johnson, BS and Matt C. Wyatt, MS for their advice on plugin development and Guilherme Del Fiol.

References

  • 1.Chaudhry B, et al. Systematic Review: Impact of Health Information Technology on Quality, Efficiency, and Costs of Medical Care. Ann. Intern. Med. 2006;144:742–752. doi: 10.7326/0003-4819-144-10-200605160-00125. [DOI] [PubMed] [Google Scholar]
  • 2.Using Diffusion of Innovation Concepts to Enhance Implementa…: Nursing Administration Quarterly. LWW. Available at: http://journals.lww.com/naqjournal/Fulltext/2006/07000/Using_Diffusion_of_Innovation_Concepts_to_Enhance.4.aspx(Accessed: 7th March 2016) [DOI] [PubMed]
  • 3.Koopman R. J, et al. Physician Information Needs and Electronic Health Records (EHRs): Time to Reengineer the Clinic Note. J. Am. Board Fam. Med. 2015;28:316–323. doi: 10.3122/jabfm.2015.03.140244. [DOI] [PubMed] [Google Scholar]
  • 4.Ajami S, Bagheri-Tadi T. Barriers for Adopting Electronic Health Records (EHRs) by Physicians. Acta Inform. Medica. 2013;21:129–134. doi: 10.5455/aim.2013.21.129-134. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Jensen P. B, Jensen L. J, Brunak S. Vol. 13. Nat. Rev. Genet: 2012. Mining electronic health records: towards better research applications and clinical care; pp. 395–405. [DOI] [PubMed] [Google Scholar]
  • 6.Hripcsak G, Albers D. J. Next-generation phenotyping of electronic health records. J. Am. Med. Inform. Assoc. JAMIA. 2013;20:117–121. doi: 10.1136/amiajnl-2012-001145. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Coorevits P, et al. Electronic health records: new opportunities for clinical research. J. Intern. Med. 2013;274:547–560. doi: 10.1111/joim.12119. [DOI] [PubMed] [Google Scholar]
  • 8.Kohane I. S. Vol. 12. Nat. Rev. Genet: 2011. Using electronic health records to drive discovery in disease genomics; pp. 417–428. [DOI] [PubMed] [Google Scholar]
  • 9.Wang T. D, et al. Aligning Temporal Data by Sentinel Events: Discovering Patterns in Electronic Health Records. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems; ACM; 2008. pp. 457–466. [Google Scholar]
  • 10.Murdoch TB , Detsky AS. THe inevitable application of big data to health care. JAMA. 2013;309:1351–1352. doi: 10.1001/jama.2013.393. [DOI] [PubMed] [Google Scholar]
  • 11.Jensen L. G, Bossen C. Factors affecting physicians’ use of a dedicated overview interface in an electronic health record: The importance of standard information and standard documentation. Int. J. Med. Inf. 2016;87:44–53. doi: 10.1016/j.ijmedinf.2015.12.009. [DOI] [PubMed] [Google Scholar]
  • 12.Ruiz-Fernández D, Monsalve Torra A, Soriano-Paya A, Marin-Alonso O, Triana Palencia E. Aid decision algorithms to estimate the risk in congenital heart surgery. Comput. Methods Programs Biomed. 2016;126:118–127. doi: 10.1016/j.cmpb.2015.12.021. [DOI] [PubMed] [Google Scholar]
  • 13.Danial-Saad A, Kuflik T, Weiss P. L. T, Schreuer N. Vol. 11. Disabil. Rehabil. Assist. Technol: 2016. Usability of clinical decision support system as a facilitator for learning the assistive technology adaptation process; pp. 188–194. [DOI] [PubMed] [Google Scholar]
  • 14.Del Fiol G, et al. Disseminating context-specific access to online knowledge resources within electronic health record systems. Stud. Health Technol. Inform. 2013;192:672–676. [PMC free article] [PubMed] [Google Scholar]
  • 15.Cimino J. J, Johnson S. B, Aguirre A, Roderer N, Clayton P. D. The MEDLINE Button; Proc. Annu. Symp. Comput. Appl. Med. Care; 1992. pp. 81–85. [PMC free article] [PubMed] [Google Scholar]
  • 16.Maviglia S. M, Yoon C. S, Bates D. W, Kuperman G. KnowledgeLink: Impact of Context-Sensitive Information Retrieval on Clinicians’ Information Needs. J. Am. Med. Inform. Assoc. 2006;13:67–73. doi: 10.1197/jamia.M1861. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Cimino J. J, Li J, Bakken S, Patel V. L. Theoretical, empirical and practical approaches to resolving the unmet information needs of clinical information system users; Proc. AMIA Symp; 2002. pp. 170–174. [PMC free article] [PubMed] [Google Scholar]
  • 18.Del Fiol G, Rocha R. A, Clayton P. D. Infobuttons at Intermountain Healthcare: Utilization and Infrastructure; AMIA. Annu. Symp. Proc. 2006; 2006. pp. 180–184. [PMC free article] [PubMed] [Google Scholar]
  • 19.Cimino J. J, Jing X, Del Fiol G. Meeting the Electronic Health Record; ‘Meaningful Use’ Criterion for the HL7 Infobutton Standard Using OpenInfobutton and the Librarian Infobutton Tailoring Environment (LITE); AMIA. Annu. Symp. Proc. 2012; 2012. pp. 112–120. [PMC free article] [PubMed] [Google Scholar]
  • 20.Murphy S. N, et al. Architecture of the open-source clinical research chart from Informatics for Integrating Biology and the Bedside; AMIA Annu. Symp. Proc. AMIA Symp. AMIA Symp; 2007. pp. 548–552. [PMC free article] [PubMed] [Google Scholar]
  • 21.Murphy S. N, et al. Serving the enterprise and beyond with informatics for integrating biology and the bedside (i2b2) J. Am. Med. Inform. Assoc. JAMIA. 2010;17:124–130. doi: 10.1136/jamia.2009.000893. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Currie L. M, et al. Clinical Information Needs in Context: An Observational Study of Clinicians While Using a Clinical Information System; AMIA. Annu. Symp. Proc. 2003; 2003. pp. 190–194. [PMC free article] [PubMed] [Google Scholar]

Articles from AMIA Annual Symposium Proceedings are provided here courtesy of American Medical Informatics Association

RESOURCES