EBM Terms & Other FAQ'S

What is evidence-based medicine?
All or most medical decisions are based on evidence of some type. Some evidence is very strong (rigorously tested in the target-species under natural conditions in experiments designed to prove a theory to be false), some is very weak (not tested), and some is intermediate. Evidence-based medicine is the effort to place all medical decisions on the strongest scientific proof (evidence) available.

What are levels of evidence?
Evidence can be ranked on the basis of its scientific soundness. Sources regarded as strong evidence include randomized controlled clinical trials in the target species under a similar husbandry environment or systematic reviews of more than one trial that meet these criteria. These are followed by cohort studies, experimental challenge studies (unique to veterinary medicine), case-control studies, and case-series. The weakest evidence includes opinions based on in vitro research, pathophysiologic rationale, uncontrolled observations, and descriptive studies (e.g., case reports).

An explanation of the various types of studies (starting from the strongest type of evidence to the weakest):

Randomized controlled clinical study (Experimental study)
Investigator assigns subjects to study groups randomly (i.e., one group gets treatment or preventive measure, the other does not) in a setting that approximates the natural setting and allows natural exposure to the risk being investigated.
Comment: This is the best experimental design to study interventions such as treatments.

Cohort study (Observational study)
A “cohort” is a group of individuals with some characteristic(s) in common. Two or more cohorts may be compared for research purposes. A cohort study may be:

1) prospective in which the investigator defines the study cohorts without the outcome (disease) in question. The study groups are alike in many ways, but differ in one observable factor, the factor to be studied. The groups are followed forward in time and compared for the outcome of interest. Difference in the occurrence measured (i.e. incidence rates, cumulative incidence, etc.) are then thought to be associated with the cohort. However, because allocation to the cohort was not random, it is also possible the observed association is confounded by another unobserved factor that is not evenly distributed across the groups – this is why these studies provide less evidence then randomized controlled studies. For example, all puppies from single-pet households that receive a standard immunization and parasite control program at one or more participating veterinary hospitals that either do or do not annual dental prophylaxis and are evaluated for age at death and cause of death over the following 20 years. The study groups are alike in many ways (i.e. enter the study as puppies, receive base-line veterinary care, and see a veterinarian) but differ in one factor (annual dental prophylaxis) and are followed forward in time.
Comment: This study type provides good estimations of risk and rate for the groups studies, but they are long, costly studies that can only evaluate the exposure factors identified at the start of the trial when data collection began. These types of trials are best suited for common outcomes (diseases).

2) retrospective (non-concurrent or historical) in which the investigator uses a database to look at cohorts where both the exposure and outcome have occurred before the study begins and then asks specific questions about the nature of the cohorts. For example, a veterinarian reviews the records collected over the past 10 years from a swine operation with a detailed database to compare the morbidity and mortality risk of pigs born to dams that were or were not from litters with high mortality (< 2 standard deviation above mean).
Comment: This type of study is fast and inexpensive compared to a prospective cohort study and is useful to study rare exposures. However, because the data collection is not controlled by the investigator, selection bias, misclassification, and confounding are significant risks.

Challenge study (Experimental study)
Investigator randomly assigns subjects to study groups that either receive or do not receive the preventive measure or treatment and then all the study groups are challenged with a predetermined amount of risk (i.e. disease-causing agent) by a predetermined route and at a predetermined time.
Comment: This type of study is uniquely available to veterinary medicine and allows for the control of many bias risks, but relies on exposure that does not mimic natural exposure.

Case-control study (Observational study)
Investigator chooses a case group and a control group out of a large population, based on a known outcome (i.e., those having the disease and those without the disease), then the groups are compared as to frequency of exposure to past risk.
Comment: This type of study is fast and inexpensive and is useful to study the association of many risk factors to rare diseases. However, selection bias and misclassification bias are significant risks.

Cross-sectional study (Observational study)
Investigator divides a population into two groups (e.g., with and without disease), then compares and assesses causal factors.
Comment: This fast and inexpensive study type can determine prevalence, but cannot determine incidence or prove causation.

Case series study
A report on a series of patients with an outcome of interest. No control group is involved.
Comment: This type of study is very inexpensive but provides no evidence about the commonness of the condition, possible causation, or efficacy of intervention.

What is a systematic review?
A systematic review is an overview of primary studies (i.e., original research). Systematic reviews contain specific components, including an explicit statement of objectives, the materials and methods that have been conducted according to an explicit and reproducible methodology. The search for relevant articles must be thorough and objective, and the criteria used to reject articles as flawed must be explicit and independent of the results of those trials. Once the studies for inclusion in the review have been determined, the goal is to synthesize the findings of the body of work, using a qualitative approach, or, if the data are available, a quantitative approach such as meta-analysis. The summation approach depends upon the question asked; for example, questions of treatment efficacy that are amenable to randomized controlled studies may use meta-analysis. Reviews addressing disease causation may not provide data suitable for meta-analysis because these questions tend to rely more heavily of observation studies such as cohort, cross section and case control and therefore may use a qualitative summation.

What is meta-analysis?
Meta-analysis is a mathematical synthesis of the results of two or more primary studies that addressed the same hypothesis in the same way. It is an attempt to overcome the problem of reduced statistical power in studies with small sample sizes. Prior to calculating a summary effect measure, it is necessary to assess if the studies included in the review are homogeneous, and multiple tests are available for this – Cochrans Q is a common methods of assessing homogeneity. If the tests for homogeneity suggest the studies are homogenous then met-analysis (either fixed or random) can be used to calculate the summary effect measure which can be thought of as the overall best estimate of the association. If the studies appear to be different – it is possible to do meta- regression to identify sources of variation and conduct subgroup- meta-analysis – however in veterinary science there is often insufficient data for meta- regression analysis.

What is a CAT (Critically Appraised Topic)
According to the Oxford-Centre for Evidence Based Medicine, a CAT (critically appraised topic) is a document that a practitioner can create on his/her own. A CAT (critically-appraised topic) is the statement of a clinical question and its answer based on the best (strongest) evidence available. It also includes a summary of the steps taken in the ensuing search for the evidence to support the question. A CAT is generally brief, intended for rapid transfer of the information, and typically contains the following three components: 1) An answerable question that includes a description of the patient or problem, an evaluation of one or more interventions or risk factors, and the desired clinical outcome measurement; 2) The search strategy used to track down the best available evidence, including the literature database searched and the search terms used; and 3) An appraisal of the evidence based on three considerations: i) the internal validity of the research as determined by the study method and appropriate use of controls for bias; ii) the population used in the research and its appropriateness as a model for the population that generated the clinical question; and iii) the clinical relevance of the outcomes of the research.