Several steps have been outlined that are necessary for the implementation of evidence based medical therapy. These can be applied in the veterinary setting. The four main steps to practicing evidence based medicine have been summarized as follows:1
1. Formulate a clear clinical question from a patient's problem
2. Search the literature for relevant clinical articles
3. Evaluate (critically appraise) the evidence for its validity and usefulness
4. Implement useful findings in clinical practice
The Clinical Question
When faced with a patient in the clinic that has been identified with a particular cardiac disease, the question with which a practitioner is faced can roughly be stated as follows: "Is there evidence to suggest that administering a particular treatment to this patient, known to have this disease at this stage in its progression will result in an improved outcome?"
Finding the Evidence
In the electronic era, finding evidence is often quite easy for both practitioners and owners. The challenge for practitioners is to evaluate the quality of the evidence with which they are faced and the relevance of the study to the patients that they treat.
How to Evaluate the Evidence for its Validity and Usefulness?
All clinical evidence can be considered to occupy a place in what has been termed the "hierarchy of evidence". The better the methodological quality of a study, the higher it can be considered to be placed in the hierarchy. The highest order of primary information (i.e., information that is generated on its own rather than by reviewing or combining other information) derives from randomized controlled clinical trials. Evidence from other designs of study should be interpreted more cautiously, because it is more open to potential bias. We are fortunate in veterinary cardiology that a number of well conducted randomized studies have been undertaken. For the two most common acquired cardiac diseases of dogs, mitral valve disease and dilated cardiomyopathy, there is increasingly good quality evidence to guide our decision making when treating individual patients.
A well designed clinical trial should be designed to address a single question and somewhere in the resulting article (or presentation) there should be a clear statement of what that question was. There is a conventional design for clinical trials and this can be reduced to five constituent parts which are sometimes referred to by the initials PICOT.2 These initials stand for:
1. Patients (or persons for studies involving human patients)
3. Comparative intervention
It is very important that these aspects of a study are made clear when describing the design and analyzing the results. For instance, taking the QUEST study3; the patients were dogs with congestive heart failure secondary to myxomatous mitral valve disease (MMVD), the intervention was pimobendan plus standard therapy, the comparative intervention was benazepril plus standard therapy, the outcome of interest was a composite survival outcome and the time over which patients were monitored was a maximum of three years (the duration of the study).
The bases for the conclusions of any study are the results that were found, the way they were analyzed and the inferences that were subsequently drawn. As stated above, most well conducted clinical studies set out to answer a single question, often stated as the "primary outcome". The conclusions of a study that pertain to the primary outcome are usually the most valid. Conclusions that are drawn regarding other findings, sometimes termed secondary outcomes or discovered incidentally or by sub-analysis of the data, are always a little more open to question.
The validity of the conclusions relies on the validity of the statistical analyses that have been undertaken. Statistical analysis of clinical studies can be regarded as a "necessary evil". Statistical analysis is necessary to be able to conclude that any apparent difference between treatment groups is a genuine effect of therapy rather than a random effect within the data. Statistics can become confusing and are often the area of studies in which readers become lost. A common-sense approach to statistics is to try to reduce to words, without reference to the statistical test being used, the purpose of the test. By way of an example, the purpose of the multivariable analysis in the QUEST study3 is to answer the question "taking into account all of the things that were known about the animals at the time they were enrolled into the study, is there independent evidence that treatment with pimobendan will lead to an improved outcome compared to benazepril?" In the same study, the purpose of the Log-Rank test is to answer the question "over the entire duration of the treatment period for all animals in the study, was the risk at any individual point in time of having reached the endpoint lower for dogs receiving pimobendan compared to those receiving benazepril".
Are the results of the study relevant to the patients that I am treating? This question gets to the heart of what the results of trials mean for practitioners. I stated above the question with which practitioners are faced as follows; "Is there evidence to suggest that administering a particular treatment to this patient, known to have this disease at this stage in its progression will result in an improved outcome?" Here we need to ask carefully in what way were the patients in the study similar to the patient to which you wish to apply the results and in what way were they different. Central to answering this question are the inclusion criteria and exclusion criteria for the study. For instance, if a patient being treated is not suffering from mitral valve disease but rather is suffering from dilated cardiomyopathy, then the results of the QUEST trial may not be helpful and it may be more relevant to turn to studies more specifically designed to evaluate the effect of therapy in this disease.4,5 Similarly, if the patient in question has mitral valve disease, but has yet to reach the stage of developing heart failure, then it is probably not safe to extrapolate the findings from the QUEST study to such a patient. Previous experience would suggest that treatment shown to be of benefit in dogs once they have developed heart failure, may not be of clear benefit in those prior to the onset of failure.6 Applying this "purist" approach to all evidence can be problematic and sometimes extrapolation may be necessary; particularly when faced with animals with rare diseases. For instance, if I am faced with a young dog that has heart failure secondary to congenital mitral dysplasia, one might argue that we have no basis on which to support the use of pimobendan, but by extrapolation since the cause of the heart failure is similar (but not identical) we have good reason to believe pimobendan may be of benefit. Applying a sliding scale, one could argue that the more the patient you are trying to treat is like the ones enrolled in the study, the more the results of the study are applicable. Several large scale studies involving human patients have been criticized for having very restrictive inclusion and exclusion criteria resulting in limited applicability of their conclusions to more general populations.7,8
Do significant differences in outcome in the study translate into clinically meaningful differences in outcome that will be of benefit to my patients and clients?--It is important to consider the difference between clinical and statistical significance. A very large clinical trial with many thousand participants could potentially demonstrate a difference in survival of only a few days in response to a very expensive therapy. One needs to weigh up many factors, as well as just the findings of the study, when deciding whether the results of a study should lead to a change in current practices.
Implementing Useful Findings in Clinical Practice
The implementation of results relies on understanding the implications of the results and deciding how practice should be changed as a consequence. If we aim to provide optimal care that enables us to prolong the period of good quality life that patients with debilitating and life-threatening conditions enjoy, we should implement convincing new evidence in practice as soon as is practicable. We should still not lose sight of the fact that despite the benefits we have obtained with our new treatments, the majority of patients still die within a year of diagnosis. We should continue to strive to develop still better ways of managing this condition.
1. Rosenberg W, Donald A. Evidence based medicine: an approach to clinical problem-solving. BMJ, 1995. 310(6987): p. 1122-6.
2. Sackett DL. The tactics of performing therapeutic trials, in Clinical epidemiology: How to do clinical practice research, B.R. Haynes, et al., Editors. 2006, Lippincott Williams & Wilkins: Philadelphia. p. 66-172.
3. Haggstrom J, et al. Effect of pimobendan or benazepril hydrochloride on survival times in dogs with congestive heart failure caused by naturally occurring myxomatous mitral valve disease: The QUEST Study. J Vet Intern Med, 2008. 22(5): p. 1124-1135.
4. Luis Fuentes V, et al. A double-blind, randomized, placebo-controlled study of pimobendan in dogs with dilated cardiomyopathy. Journal of Veterinary Internal Medicine, 2002. 16: p. 255-261.
5. O'Grady MR, et al. Effect of pimobendan on case fatality rate in Doberman pinschers with congestive heart failure caused by dilated cardiomyopathy. J Vet Intern Med, 2008. 22(4): p. 897-904.
6. Kvart C, et al. Efficacy of enalapril for prevention of congestive heart failure in dogs with myxomatous valve disease and asymptomatic mitral regurgitation. J Vet Intern Med, 2002. 16(1): p. 80-8.
7. Concato J, Shah N, Horwitz RI. Randomized, controlled trials, observational studies, and the hierarchy of research designs. N Engl J Med, 2000. 342(25): p. 1887-92.
8. Cowie MR, et al. Survival of patients with a new diagnosis of heart failure: a population based study. Heart, 2000. 83(5): p. 505-10.