Article Text

Download PDFPDF

Breaking the Translational Barriers: The Value of Integrating Biomedical Informatics and Translational Research
  1. Philip R. O. Payne,
  2. Stephen B. Johnson,
  3. Justin B. Starren,
  4. Hugh H. Tilson,
  5. David Dowdy
  1. From the Department of Biomedical Informatics (P.R.O.P., S.B.J., J.B.S.), Columbia University, New York, NY; Department of Radiology (J.B.S.), Columbia University, New York, NY; University of North Carolina School of Public Health (H.H.T.), Chapel Hill, NC; Johns Hopkins University School of Medicine (D.D.), Baltimore, MD. The preparation of the manuscript was supported by the National Academies. The views presented in this article are those of the authors and not the Institute of Medicine, the Institute of Medicine's Clinical Research Roundtable, or the Roundtable's sponsoring organizations.
  1. Address correspondence to: Mr. Philip R.O. Payne, Department of Biomedical Informatics, Columbia University, 622 West 168th Street, VC5, New York, NY 10032; e-mail: philip.payne{at}dbmi.columbia.edu.

Abstract

The conduct of translational health research has become a vital national enterprise. However, multiple barriers prevent the effective translation of basic science discoveries into clinical and community practice. New information technology (IT) applications could help address these barriers. Unfortunately, owing to a combination of organizational, technical, and social factors, neither physician-investigators and research staff nor their clinical and community counterparts have harnessed such applications. Recently, at the request of the Institute of Medicine's Clinical Research Roundtable, a qualitative study of these factors was conducted at several leading academic medical centers. We explore the current status of IT in the translational research domain, describe the qualitative results, and conclude with a proposed set of initiatives to further increase the integration of IT into translational research.

Key Words
  • clinical research
  • translational research
  • biomedical informatics
  • information technology

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Key Words

The translation of scientific discoveries into mainstream therapies has been one of the key challenges to our clinical research, medical practice, health care, and public health delivery systems for the past 20 years.1Recently, members of the Institute of Medicine's Clinical Research Roundtable classified these challenges as falling into two different “translational blocks,” one from the bench to the conduct of clinical studies and the other from the evidence derived from studies into implementation in health practice.2The inability to break through these barriers has a significant impact on the ability of health professionals to provide safe, effective patient care and to reduce the cost of health care delivery.3The urgency of finding practicable solutions to these barriers is magnified by the emergence of translational research as a national enterprise. To address these deficiencies, steps must be taken to improve the inherent clinical research capacity, dissemination ability, and information management facilities of the health care system, with the objective of providing additional support, guidance, and infrastructure for both the conduct of translational research and the translation of new findings into patient care or community health.2Numerous efforts in the biomedical informatics domain have attempted to address these areas, most notably the development of electronic medical record (EMR) systems.4-6Unfortunately, these systems have not been widely adopted, and their potential is universally underrealized. To examine the role of biomedical informatics in the context of translational research and assess what steps are necessary to address these challenges, a qualitative study was undertaken to examine the way in which physician-investigators and informaticians interacted. This article discusses the current application of informatics to translational research from the point of view of published studies and from interview data. Based on these observations, recommendations to improve the integration of informatics and translational research are proposed.

BACKGROUND

Physician-investigators are presented with a number of challenges throughout the translational research process. These challenges include the design of hypotheses and protocols, subject identification and recruitment, implementation of data collection instruments, training participating research staff, ensuring regulatory compliance, and generating timely, meaningful reports.2,7,8Information technology (IT) has been applied to every phase of the clinical research enterprise.9The use of IT, especially incorporating Internet technology, has demonstrated benefits in many areas, including initial study design, data collection and analysis, and study monitoring, particularly in multicenter programs.10-16These IT applications have included systems built specifically for research and the application of clinical systems to the research enterprise (Figure 1). An exemplary, rather than exhaustive, set of such applications is discussed below.

FIGURE 1

Information technology systems supporting the translational research enterprise. CPOE = computerized physician order entry; EMR = electronic medical record.

Hypothesis Development

Perhaps the most underrecognized and ubiquitous tools developed by informaticists are the literature search engines, such as MEDLINE, OVID, and PubMed.17-20These engines use technologies derived from computer and library sciences to allow investigators to execute rapid and complex searches of biomedical literature. Search results allow protocol authors to efficiently review large volumes of pertinent background material.21Recent advances in this domain have led to the development of automated quality assessment metrics that generate scores indicating the relevance of search results to initial criteria.22For the translation of research into practice, these tools likewise allow clinicians unprecedented access to emerging research. Other tools used at this stage in the process include computer-generated molecular and animal models, and tools to extract and aggregate data from disparate sources, such as published literature, genomic databases, and protein databases.23-26

Protocol Development

On completion of preclinical studies, investigators often undertake the composition of clinical protocols via collaboration with multiple, geographically distributed coauthors. The use of Internet technology provides for geographic and communication efficiencies in such cases, as well as access to version control, collaborative editing and commenting, and document distribution tools.21,27Informatics tools have also provided the facility to create and execute novel study designs, such as those leveraging emerging genomic technologies.28,29

Recruitment and Screening

Informatics tools have been shown to increase the efficiency of subject recruitment.30The World Wide Web has become a central tool for informing potential subjects about ongoing research.11Research subjects must often meet a complex corpus of requirements concerning health status, history, and demographic or physiologic characteristics.8IT applications can provide powerful tools to identify potential study participants and assess the availability of such participants in a given population, allowing study planners to assess protocol feasibility and develop recruitment goals.31,32Clinical EMRs can be queried to identify patients matching study criteria. Clinical decision support systems have been used to automatically notify the research staff when eligible patients enter a facility.33Given the current regulatory environment, especially those requirements stipulated by the Health Insurance Portability and Accountability Act (HIPAA), such information systems also provide a means for verifiable masking of identifiable patient data and auditing of access to that data.34,35Finally, applications have been developed that assist the execution of the informed consent process, delivering educational materials to research subjects and recording subject consent to participate in a given protocol.31

Data Collection

Data collection during clinical protocols has historically relied on paper forms, with additional data extrapolated and codified by research staff via a manual review of source documentation. On completion of such forms, recorded data must then be transcribed into databases for use in later analysis, often leading to transcription or coding errors.7,8A number of IT applications have been developed, spanning a spectrum from simple computer-based case report forms that directly convey collected data to databases to complex systems that provide for automated collection of data elements from existing clinical systems. A few systems even incorporate intelligent data collection tools that customize data collection requirements based on subject status, protocol time point, and previously collected data.14,36-38In all such instances, frequent use of Internet technology increases efficiency by providing for multiuser and site access to research records.10,12Furthermore, cross-tabular reporting mechanisms linked to research databases enable investigators to generate reports on research data at both prescribed and ad hoc time points.39Again, it is important to note that the use of IT applications for the collection and storage of data has been demonstrated to increase the facility with which health care institutions are able to comply with privacy standards.35Furthermore, emerging technologies, such as telemedicine, allow for the delivery of an intervention and the collection of data from geographically distributed populations.29,40

Many systems developed for the clinical care environment also have utility in translational research programs. These dual-purpose systems include EMRs, computerized physician order entry (CPOE) systems, guideline delivery systems, and biomedical imaging technologies. EMRs enable structured, normalized collection of health information and hence provide an alternative to the collection of data in an unstructured form, such as that commonly found in paper medical records or through the process of coding data and recording it on paper forms.5,41-44Where such EMR technology has been applied to support translational research, significant decreases in redundant data entry have been realized, accompanied by simultaneous increases in data quality, through the elimination of transcription or omission errors.2Additionally, for postmarketing epidemiologic research, EMRs permit the “banking” of exposed cohorts from trials and long-term prospective follow-up of exposed and comparison populations for ascertainment of subsequent medical events. Similarly, CPOE systems capture provider orders pertaining to therapies and medications given to research subjects,45,46including the interventions being studied. Digital biomedical imaging technologies, such as picture archiving and communications systems (PACS), provide for reduced redundancy of data collection, replacing unstructured data and corresponding physical images with codified data and integral electronic images.47

Study Monitoring

Data collected during a protocol must be monitored to ensure completeness, accuracy, and conformance with data entry guidelines.7,8Applications have been developed that automate the review of data, in real time, and provide feedback to necessary research staff. Both prospective and retrospective IT systems have been implemented to address this need. As has been demonstrated with other IT applications, such systems increase efficiency and decrease the resources required to conduct clinical studies.37In addition to quality assurance measures, it is also important to provide investigators with critical information at decision-making time points throughout the course of a protocol. Given the complex nature of treatment and evaluation schemas associated with clinical protocols, the ability to capture such information and subsequently use it to provide patient-specific decision support can result in decreased protocol deviations. Decision support technologies, often used in clinical care environments, have been applied in this context with demonstrable success. Such systems, using methodologies including bayesian logic and knowledge-based heuristics, are able to evaluate data and provide prompts to clinical investigators addressing both protocol- and standard of care-based requirements for the treatment and evaluation of research subjects. Such systems are notable in their ability to prevent protocol deviations proactively, thus reducing subject attrition while also contributing to the provision of increased patient safety throughout the care delivery process.48-52Complex protocol schemata also require that participating research staff be appropriately trained to ensure that guidelines specified in research protocols are accurately executed. Guideline delivery systems developed for clinical applications provide an ideal medium to codify and present such complex information in research studies.53,54

Data Analysis

The most long-standing application of IT in the translational research domain is the use of statistical analysis software, a practice that has revolutionized the biostatistics field over several decades and that provides for rapid, resource-efficient execution of data analysis.55,56Existing temporal expectations that research data will be placed in databases and run through analytic routines in a timely manner are predicated on such applications. Other tools facilitate the extraction of research data from large clinical repositories.57More recent developments in this field have focused on the development of software incorporating artificial intelligence technologies to extrapolate conventional statistical measures from atypical subject populations that may exist in infrequent or highly specific disease states.28

Dissemination

The results of translational research are of little value until they are translated into clinical practice.1Bibliographic tools, such as PubMed, have revolutionized the approach to the clinical research literature but, of course, require educated users with the time and inclination to use them effectively. By providing information at the point of action, decision support and CPOE systems have the potential to greatly accelerate the dissemination of important research findings,58,59facilitating their effective and timely application to current practice.

Summary of the Current Situation

To date, efforts to harness the power of the computer for translational research have largely resulted in isolated, often stand-alone data entry systems and databases. Research-specific systems are rarely integrated with clinical IT systems or with other research systems. In the majority of cases examined, these systems, whether research specific or primarily clinical, have experienced less than wholesale adoption by physician-investigators. Several factors have led to this situation. Chief among these are policy-based objections concerning the adoption of new information technologies, end-user acceptance and training in the use of computing systems, and the absence of required infrastructure components, such as systems and standards for the exchange of data between disparate systems and aggregation for population-based research.60-63These issues are elaborated on in the Discussion section.

METHODS

To better understand current interaction patterns between physician-investigators, informaticians, and the manner in which IT-based solutions are applied in translational research, a series of interviews were undertaken. For this study, six academic medical centers were selected by the American Association of Medical Colleges (AAMC) committee “Information Technology Enabling Clinical Research”: Brigham and Women's Hospital, Memorial Sloan-Kettering Cancer Center, Mayo Clinic, Regenstrief Institute, University of Pennsylvania, and Vanderbilt University. The selection criterion was that each institution be considered a national leader in both informatics and clinical research. Sixty-two potential subjects were identified through a review of institutional Web sites and published literature. Individuals contacted included those in IT leadership positions, faculty and staff responsible for implementation of IT applications, and physician-investigators. Of the 62 persons contacted, 27 participated in a 30-minute semi-structured interview. A minimum of two clinical researchers were interviewed at each institution, one involved in clinical trials and one in epidemiologic or health services research, and at least one individual responsible for IT support of clinical research. For each interview, the topics addressed included the current IT applications at the institution supporting translational research, current clinical IT applications, the management structure of IT support for translational research, and perceived unmet needs. Interviews were recorded and transcribed.

RESULTS

The participation of key informants varied widely by discipline. Twelve of 17 (71%) individuals in IT leadership and informatics-related positions participated, and 10 of the 20 (50%) health services researchers participated, whereas only 5 of 25 (20%) physicians involved in clinical trials participated. It is unclear whether the low response rate by physicians represents simple scheduling difficulties or whether it bespeaks a general lack of awareness by physicians in clinical trials of the potential role for informatics.

Analysis of the interview transcripts revealed a number of themes that were consistent across multiple institutions. Six themes predominated:

  1. Fragmentation of IT resources frequently hindered IT use and effectiveness. Physician-investigators stated that they accessed IT resources to support research activities through interaction with a number of entities, including General Clinical Research Centers (GCRCs), departmental IT support groups, informatics divisions, statistical support groups, and administrative offices. Communication among these various groups was reported to be uniformly poor. In contrast, at the two institutions in which respondents described effective IT support, there was a separate, identifiable, organized informatics group with expertise in both biomedical informatics and clinical research. These findings suggest the importance of a cohesive informatics group.

  2. Funding for IT is fragmented and sporadic. At the institutions surveyed, IT resource groups depended on multiple, separate funding mechanisms to support their work, including conventional grant funding derived from GCRCs, investigator-initiated clinical research grants, dedicated informatics grants, and institutional funds. However, given the scope of work undertaken by these groups, funding levels were often suboptimal.

  3. Local awareness of IT solutions that could be applied to translational research was often lacking. Many institutions possessed state-of-the-art IT resources, such as literature search tools, Institutional Review Board information systems, data collection and storage systems, research subject tracking systems, and clinical systems that could be used to support research activities. However, such systems were infrequently and inconsistently used. Respondents had varied knowledge concerning these IT resources. A frequent finding was that although IT personnel had knowledge of the structure of systems such as EMRs, they did not necessarily possess an understanding of their use in the clinical or research environment. Similarly, physician-investigators often had an understanding of the proper use of such systems in the clinical or research environment without an understanding of their underlying structure or potential.

  4. Current EMRs are immature and of limited value in translational research. In those institutions in which EMRs existed, use of data from clinical data repositories was usually limited to the initial study design (eg, specification of subject cohorts). Where EMR data were used for other aspects of study design, including the identification of potential subject cohorts, data were usually deidentified and therefore of limited utility. Some creative uses of EMRs were described, including the supplementation of protocol-driven data or the addition of protocol-derived data points in EMR data structures. Another challenge noted was that many EMR data fields involved unstructured free text input, minimizing the applicability for research purposes.

  5. The lack of properly trained IT staff frequently limited the use of IT to support research programs. At almost all institutions surveyed, IT staff played a critical role in data storage and retrieval. However, because the ability to query such data sources was usually dependent on the limited availability of expert staff, use of such resources was inconsistent. Access to clinical data in EMRs was dependent on specialized clinical and/or coding staff that was not always available for research activities.

  6. Web-based subject recruitment has achieved high penetrance. A bright spot in the survey was that all of the institutions provided clinical trials recruitment information via institutional Web sites and intranets. Most institutions limited such systems to the publication of lay-level protocol and contact information rather than providing real-time subject screening. Many respondents indicated that they would be likely to use Web-based subject screening if it were available.

DISCUSSION

Given the reputations of the surveyed institutions for developing and applying best practices in both biomedical informatics and translational research, it is reasonable to infer that the application of IT to translational research at other institutions is equally, if not more, problematic. The low availability and use of IT applications within the translational research domain are, in part, due to a lack of understanding on the part of informaticists as to the requirements of researchers. This is compounded by a lack of awareness among researchers as to the potential resources and capacities afforded to them through the use of IT.

Despite the existence of many examples in which IT applications have been used in translational research and have exhibited benefits in comparison with conventional methods, a major disconnection persists between the two disciplines that prevents investigators in both areas from fully realizing opportunities that such applications provide in overcoming barriers to the translational research process. The current state of affairs can be summarized by the following observations:

  1. The role that biomedical informatics plays within the field of translational research and the contributions IT can make are not well understood by physician-investigators and their clinician and administrative colleagues.

  2. Communication and coordination pathways among investigators in the translational research and biomedical informatics domains are not well formed.

  3. Available technology infrastructures that support the use of IT for translational research are inconsistent.

  4. Funding for the development of IT for translational research is fragmented and often inadequate.

  5. Adequately trained IT staff are scarce, and participation in informatics efforts by physician-investigators is infrequent.

RECOMMENDATIONS

In response to these high-level issues, the following six initiatives are proposed:

  1. Increase the visibility of informatics applications. Existing informatics resources are frequently underused. The interview data suggest that this is, at least partly, due to a lack of awareness. To better inform investigators about solutions and expertise in the integration of translational research and biomedical informatics that are available to them, those resources must be made more visible. At the institutional level, disparate and isolated biomedical informatics infrastructure components, research and development programs, and faculty should be organized into a single functional unit when feasible, or collaborative mechanisms, such as working groups and committees, and shared-technology resources should be established.2,64This would allow for the execution of effective strategies to best use current infrastructure components and adopt future technologies. Furthermore, the existence of a single point of contact will make it easier for physician-investigators to identify informatics resources and determine collaborative relationships with informaticians. At the national level, conferences and journals focused on clinical research should solicit presentations and publications on informatics resources. Finally, easily comprehensible informatics training materials for use by physician-investigators are urgently needed.

  2. Formulate a national research agenda for the development of translational research informatics resources. The significant success enjoyed by investigators involved in development of clinical IT systems has been greatly accelerated by the existence of multiple applicable national research agendas. Notable examples include the National Library of Medicine's (NLM) IAIMS and UMLS initiatives, which have helped drive the development of clinical information systems in academic health care institutions.65Such architectures define critical hardware and software components, electronic data interchange standards, and methodologies for the application of rules and logic to data. No analogous system architecture exists for the translational research domain. Therefore, a national research agenda for the development of an architectural model for translational research information systems should be undertaken.64,66Such an initiative should involve participation by informaticians, physician-investigators, technology professionals, and federal funders including the National Institutes of Health (NIH) and Agency for Healthcare Research and Quality (AHRQ), as well as regulatory agencies and private sector entities.

  3. Revise current research funding mechanisms to provide a means for independent review of informatics proposals. At present, the majority of informatics research is funded via the NLM. The financial resources of the NLM are simply inadequate to develop informatics tools for the national translational research enterprise. Moreover, significant work pertaining to the translational research domain is performed in conjunction with research programs funded by industry, foundations, or government, notably the constituent institutes of the NIH. In the case of the NIH, funding for informatics proposals is contingent on successful review of the scientific merit of associated clinical and basic science research programs without regard to the potential contributions such informatics research and development may provide independently of the specific project. Further, the historical position of most NIH institutes has been that funding of informatics research and development is outside their scope of responsibility. However, given the nature of modern translational research and the increasing importance of information and knowledge management in this process, such distinctions are rapidly becoming anachronistic. To address this problem, measures, including the adoption of grant application modules designed to enumerate informatics components within non-informatics proposals and the inclusion of appropriate reviewers for such applications, should be undertaken. Furthermore, dedicated funds and an award mechanism for informatics research that correlates to specific non-informatics projects funded by the NIH should be established.

  4. Develop methods to identify and disseminate best practices in translational research informatics. As has been the case with clinical informatics, few, if any, institutions will have the breadth of expertise necessary to develop optimal systems in isolation. To take advantage of the institutional, structural, and personnel changes proposed, the process of developing, deploying, and managing informatics applications must undergo fundamental changes. Specifically, strategies derived from empiric evidence and best practices co-opted from the computer science, management, and health care domains must be developed to provide a means for cost-, time-, and resource- efficient execution of informatics research and development.60,67-70Specific, critical needs in this area include the development of electronic data interchange standards for research data,71,72standards for human-computer interface components used to access such information systems, methods for adequate ethnographic evaluation, tools for the iterative design of IT systems targeting the translational research domain,73and systems for the aggregation for population-based research and the translation of those findings into practice.

  5. Initiate educational and professional programs to prepare biomedical informatics professionals and physician-investigators for the IT-based translational research environment. The majority of biomedical informatics training programs focus on the patient care and basic science application areas.74Although these areas are extremely pertinent to translational research, there are certain confounding factors, such as the unique organizational and functional structure of clinical research programs, that lend an additional layer of complexity to the integration of informatics and translational research. Therefore, training and education programs addressing this area should be developed.2In the same way that the AAMC has educational informatics objectives for the training of physicians,75the next generation of clinical researchers must understand informatics and be trained to expect informatics expertise in the research team.

  6. Provide incentives to increase integration of informatics and translational research. The integration of informatics and translational research will require scientists on both sides of the divide to take significant personal career risks to bridge the divide. Given the difficulty of engaging the clinical community in the development and use of IT systems, incentive programs must be developed to provide an impetus to do so. These incentives could include protected funding and time to allow physician-investigators to participate in the development of IT systems. The cross-training of informaticians and translational researchers will be ineffective unless jobs and career paths are available that leverage their new skills. Institutions that wish to be leaders in this area should consider the creation of new faculty positions and centers focused on the application of informatics to translational research. To encourage institutional action, national evaluation groups should include measures of IT use as an evaluation metric. For example, during reviews of GCRC grants, National Cancer Institute cancer center grants, and other large center grants, NIH, AHRQ, and other funding agencies interested in advancing the translation of research into practice could explicitly address the use of appropriate IT resources at the applicant site.

CONCLUSIONS

In the modern health and health care environment, the complexity, volume, scope, and nature of information needed to deliver patient and population-based care are rapidly driving the adoption of informatics-derived technologies. The need for IT solutions for translational research is equally great. Existing work has already demonstrated the potential for improvements in the efficiency, quality, and productivity of the research enterprise. Numerous applications addressing a variety of needs in ways never before deemed possible—from the rapid harnessing of preclinical studies, to the multicenter drafting of clinical protocol documents, to their execution integrated into the clinical or community context and their subsequent analysis and dissemination—have been developed for the translation research domain. Unfortunately, the promise of such applications is unrealized. For a variety of reasons, translational researchers are often unaware of or unable to adopt existing technology solutions in spite of their proven benefits. At the same time, the technology solutions themselves are still in their infancy. Significant improvements in those applications, their integration, and national standards will be required. A series of efforts must be undertaken in policy, education, and funding to ensure that physician-investigators have ready access to appropriate information technologies, in concert with other measures to improve the effectiveness of translational research programs and the delivery of the fruits of these labors to the patients and communities whose better health awaits.

ACKNOWLEDGMENTS

We would like to acknowledge the contributions of Alex Ommaya and the members and staff of the Institute of Medicine's Clinical Research Roundtable, as well as those respondents who participated in the qualitative study reported on.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
View Abstract