Evidence-based Health Practice

Roberto J. Rodrigues

Information Systems: Enabling Factor for Evidence-Based Health Practice Increasing prominence is being given to the use of best current evidence in decision making, both in clinical practice and healthcare management. A review of the main issues related to Evidence-Based Practice (EBP) and a discussion on how to develop, validate, promote, and use evidence and knowledge in clinical and administrative settings are presented. The role of information systems, and the challenges that must be met in the implementation of Information Systems and Information Technology (IS&T) and knowledge management tools, are examined for six application areas: Reference Databases, Contextual Data, Clinical Data Repositories (Clinical Databases), Administrative Data Repositories, Decision Support Software, and Internet-Based Interactive Health Information. Computerized applications in support of EBP must follow a hierarchy in which systems tasks range in complexity, from reference retrieval and the processing of relatively routine transactions, to complex decision-support systems.

The Evidence-Based Practice Movement Most clinical practice, is based on limited evidence, mostly textbook information, obsolete concepts, nonvalidated case studies, partial or unendorsed reviews and abstracts, and anecdotical or individually accumulated clinical experience. Proven therapies backed by ample evidence are underutilized due to lack of knowledge or grasp of available evidence and, frequently, clinicians do not believe that benefits observed in clinical trials can be translated into clinical practice (1). Evidence-based practice (EBP), which gained impetus in the early 90’s, replaces the traditional paradigm of “medicine by authority”, by a more critical and scientific approach (2,3).

Dr Rodrigues is Regional Advisor in Health Services Information Technology, Essential Drugs and Technology Program, Division of Health Systems and Services Development, Pan American Health Organization / World Health Organization, Washington, D.C.,USA

A major contribution to the call for a more rational approach to clinical decision-making was the growth in clinical, administrative, outcomes research, and investigation into the determinants of healthcare costs. These developments have made it mandatory for healthcare professionals and managers to be aware, appraise, and systematically make use of published data (4). EBP can be seen as a process of turning clinical problems into questions and systematically locating, appraising, and using contemporaneous research and critical reviews, i.e., the best evidence possible and the most appropriate information available as the basis for decisions. The EBP movement is an effort to teach direct care professionals to evaluate research evidence and apply it to clinical activities – it requires that the results of primary research be compiled in a methodical way and made accessible to those involved in the decisionmaking process (5).

2

EBP values, enhances, and builds on clinical expertise, facts about disease mechanisms, and human pathophysiology. Failure to use evidence has been show to result in a considerable lag period before therapies of proven effectiveness are implemented or ineffective therapies are withdrawn from practice. Ultimately, EBP represents the formalization of the care processes that the best clinicians have practiced for generations (3). The introduction of evidence-based clinical practice has had a positive impact on medicine and pharmacy – drug therapy being perhaps one of the most significant areas influenced by EBP. Evidence-based practice concepts have been adopted in physician’s training programs that place emphasis in problem-solving and information search skills (6,7,8).

Information Systems Support to Evidence-Based Practice The role of information systems is to capture, transform, and maintain three levels of facts: raw data, processed data, and knowledge (9). Data in context, as processed data, is traditionally referred as “information”. Information conveys intelligence about a particular topic and can be considered the equivalent of “evidence”. Knowledge is information in context. It represents an intellectual construct of a higher order, where evidence from various fields and sources are linked, validated, and correlated to established scientific truths and thus becoming a generally accepted body of wisdom.

Because EBP is an information- and knowledge-intensive process, focusing attention on evidence-based practice from an informational perspective and the implementation of an information infrastructure (“infostructure”) in the clinical and administrative practice environment constitute the core of the contribution of informatics to EBP.

Information and knowledge management constitutes the core activity of the healthcare professional. To put the problem of information, as an essential support component to appropriate decision making, in proper perspective, consider the costly mistakes that stem from incomplete information – of the roughly one trillion dollars spent in healthcare in the United States in 1998, an estimated 25% relates to the delivery of unnecessary or avoidable care, the performance of redundant tests, and excessive administrative costs. Information Systems and Information Technology (IS&T) contributions to EBP is accomplished in six areas of application (Figure 1): • • • • • •

Reference Databases Contextual Data Clinical Data Repositories Administrative Data Repositories Decision Support Software Internet-Based Interactive Health Information

Reference Databases: Literature, Clinical Trials and Reviews, Current Research, and Guidelines Most of the literature on EBP emphasizes the utilization of published evidence, usually restricted to papers, reports, and research recorded in formal technical or scientific sources (7,10,11,12,13,14,15). Basic research and clinical observational data and evidence are represented in the biomedical literature (books and periodicals), in the formal publications of scientifically conducted clinical trials and reviews, and in the formal records of current research initiatives. Jointly, they represent a body of information and knowledge – Biomedical Knowledge – that can be employed in the generation of clinical guidelines, protocols of care, and in the support of evidence-based decisions. Reference databases containing formal scientific and technical literature are helpful in a research environment but are of limited service in the day-today practice of the non-academic environment. What most clinicians require are fast, up-to-date, concise, and structured responses to their queries that address, in a systematic manner, the diagnosis, findings, therapy, and prognosis of health conditions.

3

Reference databases, in the other hand, are useful in researching quality studies, detailing successes as well as shortcomings of health interventions but are of difficult and time-consuming utilization – searches bring up a large number of references that are tricky to sort and the generated references to the formal biomedical literature are not fit to provide the structured answers expected by practical physicians. Randomized clinical trials and systematic reviews of such trials provide the most robust, coherent, and systematic evidence about the effectiveness of health interventions. The Cochrane Collaboration Library (16,17) is an attempt to remedy the inadequacies of formal reference databases, current research findings, informal clinical experience, and the expectations of practitioners. It was established as part of an international effort to facilitate the preparation, maintenance, and dissemination of systematic reviews of the effects of healthcare. It approaches the criteria of an ideal tool for EBP (11).

Librarians play an important function in the spread and support of EBP because of their role in identifying and retrieving appropriate literature from a variety of sources. Understanding how to index and search for diagnosis, etiology, therapy, and prognosis for original studies, systematic reviews, meta-analyses, and development of clinical practice guidelines are an essential set of skills that librarians can provide to support clinicians (13). The traditional search and retrieval methods used in literature search, however, were found to be inadequate and automated tools have been developed to facilitate that process. Literature search software applications (18) and special automated data collection and content databases have been devised (19,20) to help overcome the limitations of the traditional bibliographic retrieval systems.

CLINICAL OBSERVATIONAL DATA

BASIC RESEARCH

BIOMEDICAL LITERATURE CLINICAL TRIALS & REVIEWS CURRENT RESEARCH

CLINICAL DATABASES

UNENDORSED SOURCES OF HEALTH INFORMATION

INDIVIDUAL MEDICAL RECORD (EMR)

BIOMEDICAL KNOWLEDGE GUIDELINES DATABASE

WEB-BASED INTERACTIVE HEALTH INFORMATION

DECISION SUPPORT SOFTWARE ADMINISTRATIVE DATABASES

PATIENT ENCOUNTERS

CONTEXTUAL INFORMATION (ENVIRONMENT, ANTHROPOLOGY, EPIDEMIOLOGY, SOCIO-ECONOMIC)

EVIDENCE-BASED DECISIONS

Figure 1. The Spectrum of IS&T Support to Evidence-Based Practice

Contextual Data Historically and pragmatically, two less controlled and frequently non-validated sources of data and evidence constitute the most frequent foundation for decision making by practitioners: •

Contextual information, related to environmental, socioeconomic, and epidemiological data for a particular site and time, and

•

The accumulated individual past encounters with the healthcare system, present in the individual medical record.

Contextual information is generally short lived, limited in geographic scope, and not subject to the rigorous systematic review process followed by formal literature publications. The central issue here is the fast availability of data that may be only valid to a very specific site, but carrying an extremely high informational value. An example being the diagnostic support importance of the weekly reporting on the incidence of communicable diseases. The enhancement of data definitions and quality controls in the processes of collection and processing of morbidity and mortality data, and advances in the reporting and communication of the incidence and prevalence of communicable diseases and environmental health risk factors, has ascertained the importance of epidemiological and environmental evidence in clinical decision-making. Most countries have established mechanisms for fast access and dissemination of such evidence and informatics and telecommunications have played a central role in the implementation of such mechanisms. Examples are the U.S. Centers for Disease Control and Promotion publications and the Internet sites and epidemiological information dissemination centers maintained by national, state, and municipal health authorities in most countries. Information systems dealing with contextual data and information always played an important role in making such type of evidence available to practitioners. With the more recent developments in telecommunications they gained renewed interest and increased role. The emergence of Internet-based contextual information will be examined in detail latter.

Point-of-Care Data Capture and Data Wharehousing: Clinical and Administrative Data Repositories In the manual non-automated medical and administrative record paradigm, data fragmentation, lack of structure, terminology incompatibilities, the separation of clinical from financial and administrative data, and episode fragmentation – the breakup of patient data over time and geographic space, all residing in different physical sites under different patient identifiers – disallowed a systematic utilization of the wealth of data residing in patient records. Moreover, record integration at the individual level requires a unique identifier and uniform clinical encounter data sets and classification terminologies. Such standardization is a necessary pre-requisite for the upstream consolidation of individual patient clinical and administrative data. With the possibility of capturing data at the point of care and the emergence of computer-based structured health records (Electronic Medical Record), a whole new spectrum of possibilities are opened to improve access to individual and collective patient data residing in data repositories, independently of the site of care. Data related standards are a major issue in the operationalization of the electronic medical record. Inside a number of organizations most of the standardization issues have been already addressed – notable examples being the Kayser-Permanente and the Mayo Clinic models of EMR. There is a mounting effort to reach consensus at national and international levels. Data “warehousing” applications permits digital data from various sources to be merged and accessed by any authorized user and, when used in conjunction with knowledge engineering techniques and products, the digitalized clinical, administrative, and financial data can be mined and exploited to support outcomebased clinical processes and health interventions. From a population perspective, collections of structured records can be maintained in clinical and administrative data repositories and selectively extracted to assist in the construction of clinical guidelines and evidence-based clinical and administrative decisions processes. “Intelligent agents” is a recent development in software

5

technology that allow selective extraction of data from large data repositories. The use of clinical and administrative data repositories is rapidly becoming a viable and affordable method of supporting EBP, although there are a number of problems to be solved, the most significant being: data capture, consolidation, and reliability; standards; and confidentiality issues. Evidence-based decision making is also making inroads in management and are rapidly gaining acceptance in areas like purchasing, contracting, and resource allocation taking advantage of cumulative regional databases of past activities to support administrative decisions. A number of support informational tools have been developed to manage human, material, and financial resources, product and supplier selection, etc. The privacy issue regarding access privileges and data control is particularly serious – the transformation of data residing in paper in different sites into integrated digital data, easily retrieved through a variety of access points, and the recovery of data about individually identified persons, creates the possibility of misuse. Although, many of the possible uses of detailed patient data, such as population-based studies of the efficacy of alternate care interventions, can be done without individual source identification, a variety of possible uses require patient identification. On the other hand, the demands of many care-related tasks such as reimbursement, composite individual record, auditing for fraud and abuse, “push” techniques in healthcare delivery, and support to direct patient care require the ability to access patientspecific data outside the care setting. From a technical perspective alone, to merge patient data from diverse sources requires the identification and transport of patient-specific data with all the attendant privacy issues that such process involve. “Push” techniques, whereby health managers, using patient-specific information identify candidates for potential interventions and reach out health programs, even though considered, in most circumstances, justifiable from the preventative public health and health promotion standpoint, can trigger strong opposition. In this regard, consider the controversy

generated in the U.S. when several chain pharmacies disclosed patient-specific data to third parties (21).

Clinical Decision Support Software Clinical decision software applications are designed to support the analysis of patient data and to automate aspects of clinical decision making that can be expressed as rules. Such rules can be built and maintained by a database of guidelines and the implications of the mainstream introduction and use of clinical decision-support software are many and farreaching. Concerns raised involve questions such as: •

When the amount of automated “clinical thinking” done by the software application constitutes “practice of medicine” with all its associated ethical and legal aspects?

•

Is it to be considered a medical device passive of regulation?

•

To what extent does the application software allows clinicians to examine the underlying logic and to independently evaluate how the software arrived at particular conclusions?

•

What is the role of such products in critical clinical decisions?

•

How the consequences of eventual errors are to be dealt with?

It has been pointed out that there is a vast array of problems that have not yet been addressed. The issues are important in light of the fact that the health information technology industry is investing millions of dollars in developing such applications. Developments on a number of policy, legislation, and regulation issues are badly needed in this area (22).

Internet-based Interactive Health Information Advances in telecommunications, computer applications, and user interfaces are changing the nature of interactions between health professionals and the population and the Internet (World Wide

6

Web), a low cost, ubiquitous, and flexible technology, is the fastest growing medium in history. The growing area of Interactive Health Communication (IHC) is defined as (23): “ the interaction of an individual, consumer, patient, caregiver, or professional, with or through and electronic device or communication technology to access or transmit health information or to receive guidance and support on a health-related issue.” The impact of these technologies are found in the relay of data and information, in enabling informed decision making, in the promotion of healthy behaviors, in the facilitation of peer information exchange, self-care, and in the management of the demand for health services. IHC applications are changing the nature of health communications and health practice, they are accelerating the empowerment of providers and patients, and making easier for consolidated entities to operate within increasingly large and complex organizations. “Old” media, including radio, television and print have effectively contributed to knowledge dissemination and health promotion – the new opportunities offered by Internet-based IHC have the following characteristics (23): •

Increased access to health information and support on demand in any site via a plain telephone line or any variety of wireless communication options.

•

Increased opportunity for interactions among users using electronic messaging.

•

Enhanced ability for widespread dissemination and for keeping contents and functions current.

•

Improved opportunity to tailor information to the specific needs of individuals or user groups.

•

Multimedia resources, combining text, audio and visuals.

•

Increased opportunity for users to remain anonymous, thus facilitating access to sensitive information and frank discussions about health status, behavioral risks, and fears and uncertainties.

The main health-related concern of the widespread use of interactive Web-based technologies is the enormous volume of unendorsed, nonvalidated, misleading, and potentially harmful health information available on the Internet (24). Clearly, there is enormous interest of the general population in retrieving health information – according to one study, of the approximate 50 million Internet adult users in the U.S., over 17 million searched the Internet for health information for the year ending July 1998, and this number is estimated to double in the next two years (25).

IS&T and Knowledge Management Challenges How to approach the utilization of clinical and management health information is still a complex, chaotic, and controversial subject. It is not surprising that many expectations regarding the contribution of health information systems to clinical practice have not been fulfilled. The history of failures to deliver products that are acceptable to practitioners, the difficulties of automating clinical decision-making processes, the industry failure to appreciate the peculiar economics and culture of healthcare delivery systems, the nagging lack of policies, and the improperly understood and incompletely addressed information systems technological issues, consist the core problems of most experiences with health information systems development and implementation. Patient medical records and encounter data as source of evidence data requires the redesign of clinical records, methods to capture and organize data, computerization of records, methods for data search, and data interpretation. Many investigators have attempted to extract meaningful data out of textual material recorded in clinical encounters, but the handling of medical record text is fraught with serious difficulties because of the idiosyncrasies of medical records. An example is HEALNet, a project directed at information extraction from medical text in support of EBP using statistical concept representation and the

7

construction of a core information retrieval engine (26). Although information technology has the potential to dramatically sharpen the focus of the healthcare systems on patients’ needs and preferences, this will only be reached by a concerted effort by professionals, governments and the industry to address the above problems (27,28). The great majority of innovative developments in informatics products for the health sector and their implementation will arise from the motive for profit. While the private sector is promoting the rapid development of new health information technologies, the public sector has been mostly passive and may even create roadblocks, intentionally or otherwise, to the spread of technology. In special, privacy concerns, software regulatory policies, and ethical and legal issues regarding telehealth may be the source of conflicts in the near future [21]. The dissemination of guidelines present special problems. Segmentation of the target audience is necessary for effective dissemination of guidelines, a study in the Netherlands conducted among family physicians to evaluate the adoption of evidence-based practice guidelines indicated that for some practitioners it is desirable to make evidence quickly available through publications, for other spreading the guidelines through local networks was found effective, while for another group a more active, personal approach was found to be necessary [29]. It has been suggested that the most important transformation needed today in the teaching of medicine is to foster the information management aptitude of future professionals. Training has been considered a major component in the implementation of EBP. Evidence-based medicine is an increasingly important concept in medical school curricula and continuing medical education [7]. Getting evidence into practice requires resources and organization [30,31,32] and to use evidence-based decision-making requires organizations that have systems for finding and appraising evidence and professionals who are skilled in searching, evaluating, storing, and using information and knowledge.

Besides the need for economic and organizational discipline, true healthcare reform will require a health information revolution. A key symptom of the absence of rationalization in the sector is the virtual lack of systematic and uniform operationsmeasurement techniques and tools and quality and control data, which were not required under the public cost-plus or private fee-for-service models of care. Health reform “industrialization” processes involve statistical process control, operations research and reengineering techniques, line-employee (direct care provider) performance measurement, benchmarking, and outcomes measurement. New techniques include program budgeting and marginal analysis studies, an economic evaluation combining practice data for the before period and literature data to model the after period [33]. These initiatives require vast quantities of healthcare data and a wealth of data processing and analytical intelligence. The ultimate goal of healthcare industrialization is to provide every clinical situation, no matter how unique, with automated digital support that will generate quantitative predictors, optimized decision-making, and maximized results, be they lower costs, improved quality, or a combination of both. The computerized applications to support such objectives must follow a hierarchy in which systems tasks range in complexity from reference retrieval and the processing of transactions to more complex decision-support systems. The functions required by EBP in management and clinical service delivery involves both ends of the spectrum. They include static and dynamic modeling routines, driven by rules, normative knowledge databases, and vast stores of clinical and administrative data. Their outputs consist of quantitative comparisons of outcomes associated with alternative managerial and medical decisions. Those factors are major challenges for medical informatics and argue for significant public investments in technology development through collaborative research efforts among the industry, academic centers, research centers, the health establishment, and the government.

8

References 1. McMurray, J.J. Failure to practice evidence-based medicine: why do physicians not treat patients with heart failure with angiotensin-converting enzyme inhibitors?. Eur Heart J 19 (Suppl L):15-21 (1998). 2. Evidence-Based Medicine Working Group. Evidence-based medicine: a new approach to teaching the practice of medicine. JAMA 268 (17):2420-2425 (1992).

13. Marshall, J.G. The impact of hospital library on clinical decision making: the Rochester study. Bull Med Libr Assoc 80 (2):169-178 (1992). 14. Johnston, M.E.; Langton, K.B.; Haynes, R.B. and Mathieu, A. Effects of computer-based clinical decision support systems on clinician performance and patient outcome. A critical appraisal of research. Ann Intern Med 120 (2):135-142 (1994).

3. Sackett, D.L.; Rosenberg, W.; Gray, J.A.; Haynes, R.B. and Richardson, W.S. Evidencebased medicine: what it is and what it isn’t. BMJ 312 (7023):71-72 (1995).

15. Haynes, R.B.; McKibbon, K.A.; Walker, C.J.; Ryan, N.; Fitzgerald, D. and Ramsden, M.F. Online access to MEDLINE in clinical settings. A study of use and usefulness. Ann Intern Med 112 (1):78-84 (1990).

4. Berkvits, M. From practice to research: the case for criticism in an age of evidence. Soc Sci Med 47 (10):1539-1545 (1998).

16. Becker, L. The Cochrane Collaboration. Valuable resource for family physicians. Can Fam Physician 43:403-404, 412-414 (1997).

5. Rosenberg, W. and Donald, A. Evidence-based medicine: an approach to clinical problem-solving. BMJ 310 (6987):1122-1126 (1995).

17. Silagy, C. and Lancaster, T. The Cochrane Collaboration in Primary Care: and international resource for evidence-based practice of family medicine. Fam Med 27 (5):302-305 (1995).

6. Kenney, A.F.; Hill, J.E. and Mcrary, C.L. Introducing evidence-based medicine into a community family medicine residency. J Miss State Med Assoc 39 (12):441-443 (1998). 7. Michaud, G.C.; McGowan, J.L.; van derJagt, R.H.; Dugan, A.K. and Tugwell, P. The introduction of evidence-based medicine as a component of daily practice. Bull Med Libr Assoc 84 (4):478-81(1996). 8. Shin, J.H; Haynes, R.B. and Johntson, M.E. Effect of problem-based, self directed undergraduate education on life-long learning. CMAJ 148 (6):969-976 (1993). 9. Zvarova, J. On the medical informatics structure. Int J Med Inf 44 (1):75-81 (1997). 10. McKibbon, K.A. Evidence-based practice. Bull Med Libr Assoc 86 (3):396-401(1998). 11. Odunsi, K.O.; Cooke, I.E. and Olive, D.L. Evidence-based medicine: how do we find the evidence?. J Am Assoc Gynecol Laparosc 5 (3):313-318 (1998). 12. Lindberg, D.A. and Humphreys, B.L. Medical informatics. JAMA 277 (23):1870-1872 (1997).

18. Michaud, G.C.; McGowan, J.L.; van der Jagt, R.H.; Dugan, A.K. and Tugwell, P. The introduction of evidence-based medicine as a component of daily practice. Bull Med Libr Assoc 84 (4):478-81(1996). 19. Langton, K.B.; Horsman, J.; Hayward, R.S. and Ross, S.A. A Clinical Informatics Network (CLINT) to support the practice of evidence-based health care. Proc AMIA Annu Fall Symp pg 428432 (1996). 20. Nicoll, L.H. Don’t forget GratefulMed when conducting literature searches. ANNNA J 19 (6):521 (1992). 21. The Washington Post. Prescription Sales, Privacy Fears: CVS, Giant, Share Customer Records with Drug Marketing Firm. February 15, page A1 (1998). 22. Moran, D.W. Health information policy: on preparing for the next war. Health Affairs 17 (6):922 (1998). 23. Robinson, T.N.; Patrick, K.; Eng, T.; Gustafson, D. and the Science Panel on Interactive Communication and Health An evidence-based approach to interactive health communication: a

9

challenge to medicine in the information age. JAMA 280 (14):1264-1269 (1998). 24. Health Summit Working Group. Criteria for Assessing the Quality of Health Information on the Internet – Policy Paper. Health Information technology Institute, Mitretek Systems, Virginia (1998). Available online at http://hitiweb.mitretek.org/hswg 25. Fitzgibbons, S.M. and Lee, R. The health.net Industry: The Convergence of Healthcare and The Internet. Industry Report Jan 8, 1999. Hambrecht & Quist Institutional Research (1999). 26. Kagolovsky, Y.; Freese, D.; Miller, M.; Walrod, T. and Moehr, J. Towards improved information retrieval from medical sources. Int J Med Inf 51(23):181-195 (1998). 27. Kendall, D.B. and Levine, S.R. Pursuing the promise of an information-age health care system. Health Affairs 17 (6):41-43 (1998). 28. Pan American Health Organization Information Systems and Information Technology in Health – Challenges and Solutions for Latin America and the Caribbean. Health Services Information Systems Program/Division of Health Systems and Services Development, Washington, D.C., ISBN 92 75 12246 6 (1998). 29. Grol, R.; Zwaard, A.; Mokkink, H.; Dalhuijsen, J. and Casparie, A. Dissemination of guidelines: which sources do physicians use in order to be informed?. Int J Qual Health Care 10 (2)135-140 (1998). 30. Moehr, J.R. Informatics in the service of health, a look to the future. Methods Inf Med 37 (2):165-170 (1998). 31. Godlee, F. Getting evidence into practice. Needs the right resources and the right organization. BMJ 317 (7150):6 (1998). 32. Scally, G. and Donaldson, L.J. The NHS’s 50 anniversary. Clinical governance and the drive for quality improvement in the new NHS in England. BMJ 317 (7150):61-65 (1998). 33. Scott, A.; Currie, N. and Donaldson, C. Evaluating innovation in general practice: a pragmatic framework using programme budgeting and marginal analysis. Fam Pract 15 (3)216-222 (1998).