Introduction to evidence-based medicine (Proceedings)

The principles of evidence-based medicine (EBM) were first developed by clinical epidemiologists during the late 1980's.1 EBM has been described as "the integration of the best research evidence with our clinical expertise and our patient's unique values and circumstances".2 The intent of EBM is to improve clinical decision making by acquiring, assessing, and utilizing evidence that is most clinically relevant: whenever possible, the evidence on which our clinical activities are based should be derived from well designed studies of patients (i.e., those with naturally occurring disease). In its current form, EBM refers not only to this laudable intention but also to a methodology by which this intention is realized. The methodology of EBM consists of the following 5 steps: 1) asking a clinically important question; 2) searching for evidence to answer the clinical question; 3) critically appraising the evidence gathered; 4) applying the results of the first 3 steps to one's patient(s); and, 5) assessing the outcome of applying the results (i.e., how well did the EBM-derived answer work?).3

EBM STEP 1: ASKING THE QUESTION(S) OF CLINICAL IMPORTANCE

The first step in EBM is stating the question(s) of interest about the patient. This is not always as simple as it might seem. Performing a search for too broad a question may yield numerous irrelevant citations, whereas a question that is too focused may yield no results. Any clinical encounter will generate a number of questions related to signalment and history, clinical examination findings, etiology and differential diagnosis, diagnosis, treatment, prognosis, prevention, and client experiences. Attempting to answer all the relevant questions simultaneously, however, will be unrewarding for the clinician, client, and patient. Thus, it is important to prioritize one's efforts. The PICO system for stating questions is recommended by many EBM adherents. PICO is an acronym in which P represents the patient or patient circumstances, I represents the intervention of interest, C represents the comparison of interest, and O represents the outcome to be studies. The most important aspect of posing clinical questions may be that we are reminded to constantly inquire about the evidence on which our clinical actions are based. Breaking away from dominantly relying on authority and empiricism in equine medicine can only occur if we are willing to question what we think we know.

EBM STEP 2: SEARCHING FOR EVIDENCE

Once a clinical question has been posited, it is important to find the best available evidence by reviewing the literature. This review should be as comprehensive as possible. Relying entirely on book chapters or review articles is undesirable: new information is generated faster than it can be incorporated into textbooks, and authors of review articles have their own particular perspectives, beliefs, and perceptions that influence their interpretations and recommendations. Relying primarily on journals to which we subscribe is an inefficient way to acquire comprehensive information. Too many journals with too many articles are published each month that contain information relevant to equine practice for one to keep abreast. Furthermore, journals to which we do not subscribe or routinely access will contain important reports. For example, few equine practitioners subscribe to the journal Genomics but most if not all would have interest in reading about the mutation that causes polysaccharide storage myopathy.4

Academic institutions generally have appropriate licenses for retrieving digital/electronic publications and librarians that may assist with searching for and retrieving documents. If you lack such licenses, it may still be possible to access resources through institutions such as one's alma mater or a local college or university. Access exists to high-quality information using the Internet, using resources such as PubMed (http://wwwncbi.nlm.gov/PubMed/) or BioMed Central (http://www.biomedcentral.com). These (and other) sources are freely available, but access to full articles may be limited. It is important to bear in mind that there is also much low-quality scientific information available via the Internet.

EBM STEP 3: CRITICAL APPRAISAL

The topic of critical appraisal is of crucial importance for practitioners interested in EBM. Consequently, critical appraisal will be the topic of an ensuing presentation at this meeting. A brief overview will be presented for this lecture.

Because the evidence for much of what we do in equine medicine is sparse and primarily derived either from observational (epidemiological) studies or from experiments (using horses or other animal species) from which we must extrapolate results to the clinical setting, critical appraisal is particularly important for equine EBM. These sources of evidence are ranked relatively low in the hierarchy of evidence that has been proposed for EBM.5

Evidence-based medicine places a premium on information derived from patient-based (i.e., epidemiological) studies. Studies derived from experiments are considered weakly relevant to clinical settings, even when they involve in vivo experiments using horses or other species. A full discussion of all possible study designs and their strengths and limitations is beyond the scope of this presentation, but epidemiologic study designs have been summarized elsewhere.6

Epidemiological study designs may be either experimental or observational. In experimental epidemiologic studies, investigators control the assignment of the exposure (often a treatment) to which the patients are assigned. Preferably, treatment assignment is randomized because that process helps to render the treatment groups as similar as possible for both measured and unmeasured factors that might be associated with the outcome of interest. Of course, randomization does NOT guarantee that significant differences between groups won't occur: randomly assigned groups may differ by chance alone, and the potential for differences between groups to occur following randomization is inversely related to sample size. Randomized, controlled clinical trials (RCT) are generally considered the highest form of evidence for an individual study because of the advantages that result from: randomly assigning patients; specifying a priori primary study outcomes; and, blinding of patients and clinicians in regards to the primary study outcomes. The term "randomized clinical trial" or "clinical trial" is often misused in equine medicine to refer to experimental studies involving research horses rather than patients. True RCTs are very rare in equine medicine. Consequently, the bulk of our best clinical evidence is derived from observational epidemiological studies. Observational designs include cohort, case-control, and cross-sectional designs.

A cohort study is one in which investigators first define the treatment (or other exposure) status of each group (cohort), and then the experiences of each cohort are followed over time to describe the incidence of disease. When more than 1 cohort is followed, comparisons in disease incidence among groups can be made. The ratio of the risk in a cohort receiving a treatment/exposure of interest (e.g., a group of horses treated with IV lidocaine for post-operative ileus) relative to a reference cohort (e.g., a group of horses not treated post-operatively with lidocaine) is known as the relative risk (RR). The RR is a measure of how many times more (or less) likely the disease outcome (e.g., ileus) is to occur in the exposed cohort (e.g., those that received lidocaine) than in the unexposed cohort. The design of cohort studies may be prospective, retrospective, or both.

In a case-control study, disease status (whether a horse is either a case of the disease of interest or an unaffected member of the control group used for comparison) is determined first, and then the history of the exposure(s) of interest is obtained. Because cases are purposely sampled, the incidence of disease (and the RR) generally cannot be determined from case-control studies. The ratio of the odds of exposure in cases relative to the odds in controls, however, can be calculated (this ratio is mathematically equivalent to the odds of disease among the exposed relative to that among unexposed). This parameter is commonly referred to as the odds ratio (OR). When the disease studied is rare and the cases and controls are representative of the reference population, the OR will approximate the true RR. A case-control study can be prospective, retrospective, or both.

Cross-sectional studies evaluate a study population at a particular point in time such that the status with respect to treatment(or other exposure) and disease are determined simultaneously. This design is best for purely descriptive studies, and is considered particularly weak because concurrently determining the status of outcome and exposure makes it impossible to establish causality (because a cause must precede its effect).

In human medicine, there have been efforts to develop grading systems for evidence.1,3,5 Similar systems are being applied to or developed for veterinary medicine. Careful interpretation of these grading systems is required so as not to confuse grades with quality of the study.

EBM STEP 4: APPLYING THE RESULTS

Although the process of applying results of finding best evidence should be relatively simple, problems exist. There is scant evidence derived from patient-centered studies on which to base our decisions. Thus, we are often left to resort to weaker forms of evidence. Specifications for how to integrate our clinical impressions and expertise with results of searching and appraising the literature are not well defined. Thus, the primary impact of EBM in equine medicine may be in forcing us to recognize and accept the limitations of the evidence on which we base our practice.

EBM STEP 5: ASSESSING THE OUTCOMES

Outcomes assessments in equine medicine are fairly limited. Outcomes assessment by clinicians usually focuses on how the patient fared: did the horse survive? did the disease resolve following treatment? did the horse return to its intended use? Outcomes assessments can also include the perceptions of horse-owners. The perceptions and desires of owners have much to do with their assessment of the clinical outcome for their horse. Owners may request that we implement specific treatments, including ones with which we are unfamiliar or whose effectiveness we doubt. In the absence of strong evidence for much of what we do, it is important to recognize that the basis of their preferences may be as legitimate as our recommendations. Moreover, their perceptions of their interactions may influence their assessment of the case's outcome as much as objective measurements of health or disease. Cumulating and systematically evaluating these various outcomes assessments (patient health, owner perceptions, etc.), can help us to determine the extent to which our efforts benefit our patients and clients.

There are important limitations of EBM, which are both important and beyond the scope of this presentation. Despite these limitations, there are at least 2 major advantages for embracing EBM in equine practice. First, practicing EBM will force us to think more carefully about what we know or believe we know at all stages of case management. Second, awareness of EBM will help to move us away from the traditional approach of using our understanding of disease mechanisms in conjunction with our experiences, the opinions of experts, and logical deductions or extrapolations for basing our clinical decisions, and toward relying more heavily on information derived from clinical research involving patients (rather than experimental animal studies).

REFERENCES

Evidence-based Medicine Working Group (Guyatt G et al): Evidence-based medicine. A new approach to teaching the practice of medicine, JAMA 268:2420-2425, 1992.

Straus SE, Richardson WS, Glasziou P, Haynes RB: Evidence-based medicine, ed 3, Edinburgh, 2005, Elsevier Churchill Livingstone.

Fisher CG, Wood KB: Introduction to and techniques of evidence-based medicine, Spine 32:, S66-S72, 2007.

McCue ME, Valberg SJ, Miller MB, et al: Glycogen synthase (GYS1) mutation causes a novel skeletal muscle glycogenosis, Genomics 2008;91(5):458-466.

Guyatt G, Rennie D: Users' guide to the medical literature. Essentials of evidence-based clinical practice, Chicago, 2002, AMA Press.

Fosgate GT, Cohen ND: Epidemiological study design and the advancement of equine health. Equine Vet J 2008;40:693-700 (also has supplemental on-line materials).

Smith G, Pell JP: Parachute use to prevent death and major trauma related to gravitational challenge: systematic review of randomised, controlled trials, BMJ 327:1459-1461, 2003.