Skip Navigation LinksHome > July 2007 - Volume 82 - Issue 7 > Configuration Challenges: Implementing Translational Researc...
Academic Medicine:
doi: 10.1097/ACM.0b013e318065be8d
Research Issues

Configuration Challenges: Implementing Translational Research Policies in Electronic Medical Records

Kahn, Michael G. MD, PhD; Kaplan, David MD, MPH; Sokol, Ronald J. MD; DiLaura, Robert P. DBA, MBA

Free Access
Article Outline
Collapse Box

Author Information

Dr. Kahn is associate professor, Section of Epidemiology, Department of Pediatrics and Pediatric Clinical Translational Research Center, University of Colorado School of Medicine and The Children’s Hospital, Denver, Colorado.

Dr. Kaplan is professor and head, Section of Adolescent Medicine, Department of Pediatrics, University of Colorado School of Medicine and The Children’s Hospital, Denver, Colorado.

Dr. Sokol is professor and head, Section of Gastroenterology, Hepatology and Nutrition, and program director, Pediatric Clinical Translational Research Center, Department of Pediatrics and Pediatric Clinical Translational Research Center, University of Colorado School of Medicine and The Children’s Hospital, Denver, Colorado.

Dr. DiLaura is director of research informatics and computer systems, Division of Clinical Research, Department of Quantitative Health Sciences, Cleveland Clinic, Cleveland, Ohio.

Correspondence should be addressed to Dr. Kahn, Department of Pediatrics and Pediatric Clinical Translational Research Center, University of Colorado School of Medicine, and The Children’s Hospital, 1056 East 19th Avenue, Denver, CO 80218-1088; telephone: (303) 861-6407; fax: (303) 861-6836; e-mail: (Kahn.Michael@tchden.org).

Collapse Box

Abstract

Prospective clinical trials are a key step in translating bench findings into bedside therapies. Electronic medical records (EMRs) are often cited as a significant new tool for advancing clinical trial capabilities into standard clinical practice. However, combining clinical research and clinical care activities into one unified electronic information system requires integrating a substantial body of regulatory requirements and institutional policies. Differing interpretations of external regulations and internal policies need to be reconciled so that the EMR configuration simultaneously conforms to all requirements.

The authors describe how they used a detailed clinical vignette to help focus discussions about their institution’s current research policies and how regulations and policies might be implemented in a commercial EMR. The vignette highlighted a number of inconsistencies in the institution’s policies and in individual interpretations of regulatory intent.

Attempts to implement potential policies in the EMR system also revealed a number of limitations and inconsistencies in the commercial system. The authors describe a set of compromises that will be implemented at The Children’s Hospital until missing functionality is made available from the commercial vendor. Each institution that implements an EMR will need to resolve similar policy and configuration issues at its own facility. The authors highlight these configuration challenges by presenting a list of questions that must be answered unambiguously before implementing translational research capabilities into an operational EMR.

Prospective trials are a powerful tool for assessing the efficacy of medical innovations. New therapeutic treatments and devices must undergo a series of progressively more stringent prospective trials to gain regulatory approval.1,2 Postapproval prospective trials are often used to compare the effectiveness of alternative care strategies. However, there is substantial evidence that the current approach to organizing and conducting these crucial clinical trials is not meeting the needs of patients, clinical investigators, study sponsors, or regulatory agencies.3,4 Only 7% of eligible patients enroll in clinical trials;5 for cancer studies, only 3% of eligible patients enroll.6 Eighty-six percent of clinical trials fail to complete enrollment on time7; when a trial must take additional days beyond its initial timetable to complete the trial, 85% to 95% of these additional days are a result of investigators not recruiting subjects on schedule.8 Low recruitment rates not only delay study completion times; they also threaten study generalizability, because women, minorities, children, and other vulnerable populations are underrepresented in most studies.9–11 Despite a significant increase in the number of new trials initiated each year, only 3% of all board-certified physicians participate in FDA-approved trials, the number of first-time clinical investigators dropped 11% between 2001 and 2003, and half of all principal investigators never conduct another FDA-regulated clinical trial.12,13

Table 1 provides examples of how electronic medical records (EMRs) could accelerate numerous steps in clinical trials, making them more efficient. Further, as EMR pioneers recognized, EMRs have the potential to support clinical care and clinical research simultaneously, streamlining and integrating clinical care and clinical trial systems.14–17 In the United States, the National Institutes of Health (NIH) Roadmap, the NIH National Center for Research Resources (NCRR) 2004–2008 strategic plan, and the Department of Health and Human Services Office of the National Coordinator for Health Information Technology strategic framework all refer to the government’s expectation that the use of health care information technologies will greatly expand the nation’s clinical research capacity.18–21 Numerous private-sector reports also point to the potential of information technology to increase public access to advanced and experimental treatment options available only through clinical trials.22–25 Well-designed studies are now appearing that demonstrate the ability of EMRs to improve aspects of clinical trials.26,27

Table 1
Table 1
Image Tools

Configuring an EMR requires an institution to declare explicitly which users in which specific settings have access to certain system functions and patient data. Little has been written about the types of choices that must be made when an institution attempts to configure an EMR to meet regulatory requirements and institutional policies that apply to translational clinical research. In our opinion, institutional leaders interested in developing an EMR would greatly benefit from stories like ours, of institutions that have already configured and implemented an EMR. Although similar questions and issues around clinical care and clinical trials arise in settings that have not adopted integrated EMRs, the stringent requirements for configuring an EMR bring the many competing interpretations of regulatory and insti-tutional policies into much sharper focus.

Back to Top | Article Outline

Using Clinical Vignettes to Review Institutional Research Policies

Policies and procedures may express an institution’s intentions, but they do not always reflect actual institutional practice. If configured properly, EMR systems could improve compliance with institutional policies by enabling or suppressing users’ access to specific functionalities. However, longstanding institutional policies which may seem acceptable on paper could create impractical or unacceptable workflows when implemented in an EMR. In addition, policy violations could become more visible because of extensive logging and auditing capabilities in EMR systems. Thus, institutional policies and the EMR configuration must be carefully aligned.

In 2003, The Children’s Hospital (TCH) in Denver, Colorado, began the implementation of a comprehensive EMR for all ambulatory clinics and inpatient settings. The rollout and configuration of the EMR will be completed by June 2007. At that time, detailed clinical data from all patient encounters throughout the organization will be captured in the EMR. From the initial conception, a key objective of TCH was to merge prospective clinical research and concurrent clinical care activities and requirements within an integrated EMR. Thus, EMR capabilities needed to be configured to conformto additional policy and regulatory requirements specific to prospective clinical research.

To keep policy discussions grounded to real-world issues, we created a detailed clinical vignette that included all of the key steps in a prospective clinical trial (Appendix 1). In this vignette, the EMR records clinical observations, notes, ancillary reports, and test results generated by all caregivers during routine ambulatory, inpatient, or emergency clinical care. The institution’s goal is to use the same EMR to find potential study subjects and to record study-specific clinical observations and test results on patients enrolled in IRB-approved prospective observational studies and translational clinical trials. With each question raised in response to the vignette, we enumerated a number of competing interpretations which could be implemented in policy and possibly enforced by the EMR software. Discussing policy issues in light of the clinical vignette revealed a surprisingly wide range of opinions regarding which answer(s) best matched existing (pre-EMR) research policies, practices, and regulations. Different parties have markedly different responsibilities to the institution, investigators, study subjects, sponsoring organizations, and regulatory agencies. Configuring an EMR that meets the needs of all these integral parties requires precise definitions of allowable research practices. Inconsistent policies, procedures, and workflows that have developed over time on an ad hoc or case-by-case basis are cast in a glaring light by this analysis process.

Back to Top | Article Outline

Implementing Institutional Policies via User Roles

Defining and implementing carefully constructed user-based roles and permissions for an EMR is an important technical method for ensuring that the system conforms to regulatory and policy restrictions. Distinct roles are assigned to different users who need access to the same subset of system features to perform their jobs. Role-based security manages user access to system functions and patient information. As illustrated in Chart 1, EMR users may play multiple roles in clinical care and clinical research. In settings where patients receive treatment only in the context of a clinical trial or only in standard care, it may be easy to link each EMR user to a specific role. However, in the setting of mixed care—where some part of a patient’s treatment plan is directed by a clinical trial, but other parts of the treatment plan are not—the same user may play different roles for the same patient, even within the same clinical encounter. Thus, it is critical to examine how various functional requirements change the EMR configuration not just for an individual user, but also for a user in a specific role in a specific context.

Chart 1
Chart 1
Image Tools

Defining which access rights to grant to users in specific roles is one of the early steps in configuring an EMR. To accomplish this task, the EMR implementation team needs precise answers to a number of questions:

1. What tasks are required for each identified role (Chart 1) so that individuals in that role can perform their work? For example, screening, consenting, enrolling, and treating study subjects might all be performed by users in one role, whereas setting up case report forms and data-capture screens and entering patient observations into a database might be performed by a user in a different role.

2. What system functions does a user in a specific role require access to in order to perform each task efficiently and in full regulatory compliance?

3. What tasks have dependencies, such as a specific sequence of activities (“Consent form must be signed before any study procedures can be administered”)? Dependencies may be “hard” (task cannot proceed until the constraint is satisfied) or “soft” (task can proceed but the user must document the need or resolve the constraint before concluding the task).

4. What tasks can be shared amongst users in multiple roles? What tasks can only be done by a user in a specific role?

5. When are specific tasks performed? When should the system functions that support tasks be made accessible to users?

6. If an individual user can change roles, how is a role change identified so that the correct set of system functions and limitations are made available to that user in the right context? How can these changes in role and system functions occur without disrupting workflow?

7. What patient data should be visible to users in each specific role? Conversely, what patient (or study participant) data should not be available to certain users without a change in role?

8. What special terms, code sets, and allowed values are required to capture specialized clinical data, especially if those data are to be shared or exported to other institutions or databases?

9. What special features (e.g., documentation, billing, security, etc.) are required to meet regulatory requirements?

10. How are exceptions to any of the previous questions invoked, what does the exception change, and when is the exception no longer valid?

In developing configuration specifications, our EMR implementation team used the answers to these questions to link required workflows and tasks to specific system functions for each role.

Back to Top | Article Outline

Aligning and Implementing Institutional Policies in a Commercial EMR

To achieve an EMR that truly integrates clinical care and clinical research, unique user roles, system workflows, and functional capabilities must be combined without impeding clinician productivity. Like other large academic centers, our integrated clinical care/clinical trials system needs to meet the needs of a variety of users: a broad array of clinical care generalists and specialists treating tertiary care patients, an active NIH-sponsored general clinical research center, a large regulatory clinical trials office, a multifacility IRB, and a substantial number of investigator-initiated clinical studies. As we worked on specifying EMR implementation requirements, we identified a set of clinical care and clinical trial roles and identified a set of issues—a mixture of institutional policy and regulatory requirements—that require explicit answers associated with these roles (Table 2). Many of the issues contained in Table 2 appear in the clinical vignette (Appendix 1); the vignette was designed to cast the generalized issues enumerated in Table 2 into a tangible, real-world clinical scenario, making the nature of the questions and alternative answers more accessible to the responsible executives and clinicians.

Table 2
Table 2
Image Tools
Table 2
Table 2
Image Tools

Reaching institutional consensus on what constitutes acceptable answers to the questions posed in the clinical vignette and in Table 2 has been difficult and sometimes contentious. Clinical care and clinical trials both have complex workflows with substantial regulatory oversight.28 The EMR team has had to work closely with clinical, administrative, and regulatory leaders to develop creative approaches to resolving differences while satisfying regulatory demands, workflow requirements, and unique information-management approaches within the capabilities of the commercial EMR system.

As we explored alternative policies, the implementation team described or created test implementations to illustrate the resulting workflows using the functional capabilities in our commercial EMR product. For some alternatives, the current product did not have sufficient functionality to implement a proposed policy. In other instances, a policy could be implemented in the ambulatory care application, but not in the more complicated inpatient application, even though both products were created by the same vendor. In still other cases, a policy could be implemented for clinical laboratory results, but not for pathology results. In a number of examples where a test implementation could be created, the changes in workflow required to craft the solution within the product’s existing functionality were clearly onerous and would not be acceptable in actual practice.

In addition to exploring the EMR’s capabilities to support new or revised policies, we had to consider long-established policies in terms of the EMR as well. Traditional institutional practice had established that patient safety concerns for potential duplicative radiation exposure overrode strict clinical trial confidentiality policies. Thus, research-related radiology reports historically have been included in the paper medical record. Research-related clinical laboratory tests traditionally have not been part of the paper medical record, but these results are available in the laboratory information system under a unique subject identifier. Both of these historical practices will continue in the EMR. Traditionally, the remaining research-related diagnostic test results, such as cardiology, pulmonary, and pathology reports, have not been part of the paper medical record. Although the current EMR product allows these reports to be labeled as research results, it does not allow research diagnostic test results to be suppressed in electronic displays. Thus, although these reports were not available in our paper medical records, these research-related findings will appear in the EMR. In the past, research-related orders have been written and processed using study-specific order sheets that were not part of the medical record. The current version of computer-based physician order entry (CPOE) system does not support separating research-related orders from standard-care orders in order entry or order review screens. Until new functionality is available, research-related orders will remain paper-based, and standard-care orders will be entered using the CPOE system. In the ambulatory care setting, the EMR allows physicians to mark specific documents as research notes, and special access restrictions can be placed on them. Similarly, the EMR’s inpatient system allows notes to be marked as research notes, but it does not provide a method for placing special access restrictions on any type of note except for mental-health-related notes.

Despite differences in what can or cannot be suppressed from clinical care EMR users in various settings, all orders, reports, and notes that are marked as research will be removed from the legal medical record when it is printed. References to research results or clinical actions based on research findings that appear in the standard-care documentation notes or dictations, however, will be included in the printed legal medical record. Physicians are encouraged to not make standard-care decisions on the basis of research findings, except for those care decisions related to potential study-related adverse events. However, the limited system functionality does not prevent all research results from being suppressed during standard-care encounters.

The EMR vendor has established a clinical research advisory council to provide input into future product developments to incorporate missing functionality. Over time, as new functionality is released, we will revisit our current approach and remove the current discrepancies so that the research policies and the EMR implementation are consistent across all practice environments.

Back to Top | Article Outline

Moving Forward

Each institution seeking to implement a unified clinical care/clinical research EMR will invest substantial resources in internal discussions, analyses, and compromises to define an internal interpretation of a regulatory- and policy-compliant solution. Different interpretations of acceptable solutions will result in different implementation requirements and system configurations at each site. Unfortunately, without consistent requirements across multiple customers, commercial EMR vendors cannot identify missing functionalities that would support the clinical research market’s needs. Given the relatively small market size for clinical trial software and the endless list of system-enhancement requests from the clinical care marketplace, commercial EMR providers may not be able to respond to the challenges presented by inconsistent requirements.

To achieve the oft-stated goal of expanding clinical trials and clinical research capacity, existing health care information technology efforts must define the functional characteristics of a regulatory-compliant, integrated EMR–clinical trial/research system. Research-advocacy organizations are calling for similar efforts that would allow organizations to share successful clinical care/clinical research implementation strategies.29,30 If common implementation models were developed, institutions could more easily leverage their substantial EMR investments to support prospective clinical trials and translational research. A recent symposium sponsored by FasterCures, the NCRR, and the Agency for Health Care Research and Quality has identified the need to include more focus on clinical research needs in the various national health information programs.31 We urge that one effort within this agenda include the development of vendor-independent model EMR–translational research configuration descriptions that represent “best practices” and meet external regulatory requirements. An organization could then use these model descriptions and tailor them to fit local institutional policies and clinical practices.

If institutional hurdles for executing clinical trials are reduced, the environment for translational clinical trials will improve. But without well-conceived models to guide institutions, the clinical research and clinical care communities will struggle with how best to combine these two worlds. Although each institution may select different answers to issues like those listed in our clinical vignette, having a comprehensive list of questions and alternative responses to consider would accelerate the process significantly.

Back to Top | Article Outline

Acknowledgments

The authors acknowledge the EREX implementation team at The Children’s Hospital, led by Dena Somers, Michelle Akey, Patricia Wimmer, and Chris Poppy, who are responsible for implementing a comprehensive EMR solution that satisfies an endless wish list. They also acknowledge Mary Partin, PhD, who heads the EMR implementation at Cleveland Clinic, and Anil Jain, MD, for their flexibility in exploring new boundaries for increasing the value of institutional systems for eResearch purposes. Steve Ross, MD, John F. Steiner, MD, MPH, and Peter Embi, MD, provided extensive insightful comments.

This work was supported in part by NIH grant MO1-RR00069, General Clinical Research Centers Program, National Center for Research Resources, the National Institutes of Health, and The Children’s Hospital Research Institute.

Back to Top | Article Outline

References

1 Spilker B. Guide to Clinical Trials. New York, NY: Raven Press; 1991.

2 Mathieu M. New Drug Development: A Regulatory Overview. Waltham, Mass: Parexel International; 2000.

3 Getz K. Informing and engaging stakeholders on the periphery of the clinical research enterprise. Res Pract. 2005;6:4–8.

4 Food and Drug Administration. Innovation stagnation: challenge and opportunity on the critical path to new medical products. Available at: (http://www.fda.gov/oc/initiatives/criticalpath/whitepaper.pdf). Accessed March 19, 2007.

5 Sung NS, Crowley WF Jr, Genel M, et al. Central challenges facing the national clinical research enterprise. JAMA. 2003;289:1278–1287.

6 Clinical Trials Program Review Group. Report of the National Cancer Institute Trials Program Review Group. Available at: (http://deainfo.nci.nih.gov/advisory/bsa/bsa_program/bsactprgmin.htm). Accessed March 19, 2007.

7 Lamberti MJ. Enrollment delays are getting worse. In: Lamberti MJ, ed. An Industry in Evolution. 4th ed. Boston, Mass: Thomson CenterWatch; 2003:79.

8 Cruz Rowe J, Elling ME, Hazlewood JG, Zakhary R. A cure for clinical trials. McKinsey Q. 2002;2:134–141.

9 Vidaver RM, Lafleur B, Tong C, Bradshaw R, Marts SA. Women subjects in NIH-funded clinical research literature: lack of progress in both representation and analysis by sex. J Womens Health Gend Based Med. 2000;9:495–504.

10 Heiat A, Gross CP, Krumholz HM. Representation of the elderly, women, and minorities in heart failure clinical trials. Arch Intern Med. 2002;162:1682–1688.

11 Field MJ, Behrman RE. Institute of Medicine (U.S.). Committee on Clinical Research Involving Children. Ethical conduct of clinical research involving children. Washington, DC: National Academies Press; 2004.

12 Getz K. Numbers of Active Investigators in FDA-Regulated Clinical Trial Drop. Boston, Mass: Tufts Center for the Study of Drug Development, Tufts University; 2005.

13 Mathieu MP, ed. PAREXEL’s Pharmaceutical R&D Statistical Sourcebook 2005/2006. Waltham, Mass: PAREXEL International Corp.; 2005.

14 King C, Manire L, Strong RM, Goldstein L. Data management systems in clinical research. In: Blum BI, ed. Information Systems for Patient Care. Computers and Medicine. New York, NY: Springer-Verlag; 1984:404–415.

15 Ross Davies A. Health care researchers’ needs for computer-based patient records. In: Ball MJ, Collen MF, eds. Aspects of the Computer-Based Patient Record. Computers in Health Care. New York, NY: Springer-Verlag; 1992:46–56.

16 Institute of Medicine (U.S.). Committee on Improving the Patient Record; Dick RS, Steen EB. The Computer-Based Patient Record: An Essential Technology for Health Care. Rev ed. Washington, DC: National Academy Press; 1997.

17 Rindfleisch TC, Brutlag DL. Directions for clinical research and genomic research into the next decade: implications for informatics. J Am Med Inform Assoc. 1998;5:404–411.

18 Zerhouni E. Medicine. The NIH Roadmap. Science. 2003;302:63–72.

19 National Center for Research Resources. NCRR: A catalyst for discovery. 2004–2008 strategic plan: challenges and critical choices. Available at: (http://www.ncrr.nih.gov/about_ncrr/StrategicPlan2004-08.asp). Accessed January 9, 2006.

20 Clinical Trials Working Group of the National Cancer Advisory Board. Restructuring the national cancer clinical trials enterprise.Available at: (http://integratedtrials.nci.nih.gov/ict/CTWG_report_June2005.pdf). Accessed March 19, 2007.

21 Thompson TG, Brailer DJ. The decade of health information technology: delivering consumer-centric and information-rich health care. Framework for strategic action. Available at: (http://www.hhs.gov/healthit/documents/hitframework.pdf). Accessed March 19, 2007.

22 Hanna KE. Think research: using electronic medical records to bridge patient care and research. Available at: (http://www.fastercures.org/pdf/emr_whitepaper.pdf). Accessed January 9, 2006.

23 Bleicher P. Along the road to electronic clinical development. BioExecutive Int. 2005:48–54.

24 Uehling MD. Merging EHR and EDC: Couldpharma, hospital IT converge? Health–IT World News December 6, 2005. Available at: (http://www.health-itworld.com/newsletter/index_12062005.html). Accessed May 6, 2007.

25 Committee on Data Standards for Patient Safety. Key capabilities of an electronic health record systems: letter report. Available at: (http://books.nap.edu/html/ehr/NI000427.pdf). Accessed March 19, 2007.

26 Embi PJ, Jain A, Clark J, Bizjack S, Hornung R, Harris CM. Effect of a clinical trial alert system on physician participation in trial recruitment. Arch Intern Med. 2005;165:2272–2277.

27 Mosis G, Vlug AE, Mosseveld M, et al. A technical infrastructure to conduct randomized database studies facilitated by a general practice research database. J Am Med Inform Assoc. 2005;12:602–607.

28 Bleicher P. Integrating EHR with EDC: whentwo worlds collide. Appl Clin Trials. 2006;March:6–14.

29 FasterCures. Accelerating research through the national health information network: meeting report. Available at: (http://www.fastercures.org/pdf/NHINReport.pdf). Accessed March 19, 2007.

30 Task Force II on Clinical Research. Promoting translational and clinical science: the critical role of medical schools and teaching hospitals. Washington, DC: Association of American Medical Colleges; 2006.

31 Highnam P. Ensuring the inclusion of clinical research in the national health information network: workshop summary. Available at: (http://www.esi-bethesda.com/ncrrworkshops/clinicalResearch/index.aspx). Accessed March 19, 2007.

Back to Top | Article Outline
Cited Here...
Appendix 1
Appendix 1
Image Tools
Appendix 1
Appendix 1
Image Tools

Cited By:

This article has been cited 6 time(s).

Bmc Medical Informatics and Decision Making
Text data extraction for a prospective, research-focused data mart: implementation and validation
Hinchcliff, M; Just, E; Podlusky, S; Varga, J; Chang, RW; Kibbe, WA
Bmc Medical Informatics and Decision Making, 12(): -.
ARTN 106
CrossRef
Presse Medicale
Electronic health records and biomedical research
Daniel, C; Jais, JP; El Fadly, N; Landais, P
Presse Medicale, 38(): 1468-1475.
10.1016/j.lpm.2009.06.009
CrossRef
Methods of Information in Medicine
Perspectives for Medical Informatics Reusing the Electronic Medical Record for Clinical Research
Prokosch, HU; Ganslandt, T
Methods of Information in Medicine, 48(1): 38-44.
10.3414/ME9132
CrossRef
Journal of Biomedical Informatics
An Integrated Model for Patient Care and Clinical Trials (IMPACT) to support clinical research visit scheduling workflow for future learning health systems
Weng, CH; Li, Y; Berhe, S; Boland, MR; Gao, JF; Hruby, GW; Steinman, RC; Lopez-Jimenez, C; Busacca, L; Hripcsak, G; Bakken, S; Bigger, JT
Journal of Biomedical Informatics, 46(4): 642-652.
10.1016/j.jbi.2013.05.001
CrossRef
Current Opinion in Anesthesiology
Can anesthesia information management systems improve quality in the surgical suite?
Balust, J; Macario, A
Current Opinion in Anesthesiology, 22(2): 215-222.
10.1097/ACO.0b013e328324b9e6
PDF (120) | CrossRef
Academic Medicine
Synergies and Distinctions Between Computational Disciplines in Biomedical Research: Perspective From the Clinical and Translational Science Award Programs
Bernstam, EV; Hersh, WR; Johnson, SB; Chute, CG; Nguyen, H; Sim, I; Nahm, M; Weiner, MG; Miller, P; DiLaura, RP; Overcash, M; Lehmann, HP; Eichmann, D; Athey, BD; Scheuermann, RH; Anderson, N; Starren, J; Harris, PA; Smith, JW; Barbour, E; Silverstein, JC; Krusch, DA; Nagarajan, R; Becich, MJ; on behalf of the CTSA Biomedical Informatics Key Function Committee,
Academic Medicine, 84(7): 964-970.
10.1097/ACM.0b013e3181a8144d
PDF (67) | CrossRef
Back to Top | Article Outline

© 2007 Association of American Medical Colleges

Login

Article Tools

Images

Share