The Role of Epidemiology in Equine Practice (AAEP 2011)

The Role of Epidemiology in Equine Practice (AAEP 2011)

Cohen said practitioners rely on evidence-based medicine, which is a combination of the best research evidence with clinical expertise and patients' unique circumstances. Epidemiology is the fundamental underlying science of evidence-based medicine.

Photo: Anne M. Eberhardt/The Horse

"Anecdotal and clinical impressions are appealing but are unreliable sources of information because we often are insufficiently cognizant of the role that chance plays in clinical practice," began Noah Cohen, VMD, MPH, PhD, Dipl. ACVIM, of Texas A&M's College of Veterinary Medicine. He described the importance of integrating the science of epidemiology into equine practice in his delivery of the Milne State-of-the-Art Lecture to a large audience at the 2011 American Association of Equine Practitioners convention, held Nov. 18-22 in San Antonio, Texas. He added, "It is possible to make errors in clinical judgment based on our impressions or the impressions of others. This is because something good or bad can appear to be the effect of some cause (like a treatment) when it is purely due to chance. Moreover, experts are not always correct, and logical reasoning is not always accurate." Thus, veterinarians need to rely on evidence other than just their clinical impressions and the opinions of experts.

Cohen stressed, "As practitioners, we want to be able to rely on the best available clinical evidence, or so-called evidence-based medicine. This integrates the best research evidence with clinical expertise and our patients' unique circumstances. Epidemiology is the fundamental underlying science of evidence-based medicine." Cohen informed his audience that the best evidence comes from studying patients with naturally occurring disease, noting that experimental disease rarely mimics natural or spontaneous disease. "And, when making medical decisions about horses," he continued, "a laboratory mouse is not equal to a horse."

Periodically, Cohen handed off the microphone to his friend and colleague, Bo Brock, DVM, who peppered the talk with entertaining allegories that brought home points Cohen had made. Brock described the task before Cohen as daunting: Within a short few hours, he had to teach a subject from the beginning, starting with the basic terminology of epidemiology. By the end of the session, Cohen had achieved this goal by weaving the language of probability and epidemiology into the context of real-life clinical situations.

The Essentials of Epidemiology

"Why does epidemiology matter to equine practitioners?" asked Cohen. Practitioners generally provide care to communities and populations of horses. A community may be defined based on location, use, breed, or disease status. Epidemiology is the study of disease (and health) and determinants of disease (and health) in populations. Thus, veterinarians rely on principles of epidemiology when considering "community" health. Veterinarians also use epidemiology every day when addressing individual patients. "Things learned from other horses are brought to bear on an individual horse, with epidemiology impacting everything we've seen or done before, putting everything into context with experience," said Cohen. For example, if one hears a heart murmur in a horse the interpretation of that murmur is virtually meaningless without the knowledge derived (by oneself and others) from hearing similar murmurs in other horses. Epidemiology also is used by veterinarians in clinical practice because it is the basis for obtaining and assessing clinical "evidence" from patient-based studies.

Cohen indicated that a hierarchy of clinical evidence has been proposed. Expert opinions, experimental models, editorials, animal research, or in vitro or test tube experiments represent the lowest level of evidence because these are least relevant to patients and might be misleading. Observational studies of patients like case-control studies are in the middle ranks of the hierarchy (e.g., a case-control study of risk factors for pasture-associated laminitis). Randomized controlled trials of patients are the highest form of clinical evidence, as far as individual studies are concerned. The most compelling clinical evidence comes from combining clinical trials for a specific topic in either a systematic review or a meta-analysis (integration of data from a number of independent studies). Because there is a paucity of randomized clinical trials in equine medicine, practitioners must rely heavily on observational epidemiological studies of patients to obtain clinical evidence.


"The concept of causality is critical to the practice of medicine. This is because selecting methods for diagnosis, treatment, and prevention are largely based on observing patients and assessing whether a causal relationship exists. For example, is a new test for EPM better than the old test? Is a new treatment for PPID as effective as the standard treatment?

"Most clinical effects have a set of multiple sufficient causal components-multiple components must come together before an injury or disease can occur," noted Cohen. For example, a racehorse injury has known risk factors such as a pre-existing lesion, track surface type, class of race, accumulation of training intensity and high-speed efforts, and racehorse gender, as well as unknown factors among individual horses. A combination of multiple individual components forms a set with the potential to result in injury. To elucidate this concept, Cohen described an everyday action: Simplistically, we might think that flipping a light switch turns on a light-cause and effect. However, for the light to turn on, he pointed out that there needs to be a paid electric bill, intact wiring, adequate voltage, and a functional, properly installed bulb.

Quantifying: Measuring and Counting

"Epidemiology is a quantitative science for descriptive purposes of counting health outcomes to obtain essential information, or for inferential purposes to measure associations to make inferences about causes," said Cohen.

The NAHMS (National Animal Health Monitoring System reports of 1998 and 2005) outlined national statistics of owner-reported data about frequency of disease, causes of death, geographical distribution, and management practices. Cohen noted that these reports provide valuable inferential information. The 2011 equine herpesvirus outbreak of EHM (equine herpesvirus myeloencephalopathy--meaning it had a neurologic component) further illustrates the value of descriptive epidemiology; providing important information to veterinarians and horse owners helped control the outbreak.

Cohen defined prevalence and incidence and indicated why this distinction matters: Studies based on prevalent cases are less clear about causation than those based on incidence. Prevalence studies consider factors contributing not only to disease development but also to survival and/or duration of disease. Incidence studies examine only factors that contribute to development (and therefore causes) of disease. For example, if researchers conducting a study of "prevalent laminitis" found an association between grazing muzzles and laminitis, we might conclude that grazing muzzles cause laminitis. However, horses with grazing muzzles might survive longer for detection and study relative to horses without muzzles; therefore, the association might reflect survival. In contrast, those without muzzles might die of laminitis earlier in the study period, and the prevalence association is not so much causal as it is a determinant of survival. In contrast, an association of "incident laminitis cases" with endocrine pathology (metabolic disease) is more likely to reflect endocrine disease as causal.

Clinical scientists use inferential epidemiology to investigate causal association. They use this approach to examine diagnosis, treatment, prognosis, and prevention. While the best answers come from controlled studies of natural disease, Cohen cautioned that "noise" Is one disadvantage of these studies. This noise comes from studying naturally occurring cases in a real-world environment where one cannot control all factors; thus, there is "noise" from factors other than the determinant of interest that influence the association scientists see. For example, while diet might affect colic, other factors (exercise, parasite control, pasture management, age) might also influence colic development and should be accounted for. "Filtering out this noise is a major challenge for epidemiologist and clinical scientists," said Cohen.

In epidemiology, Cohen explained that exposure refers to some factor shared by a group of horses being studied, such as treatment, environment (housing, diet, ingested toxins), or attributes (age, gender, or breed). An outcome refers to the endpoint of what is being measured, such as a disease, survival, or performance. Inferential epidemiology is used in studies identifying associations between exposures and outcomes-for example, diet is associated with impaction colic, or track surface is associated with musculoskeletal injury.

The measures of association scientists most commonly use in inferential epidemiological studies are the risk ratios (RR, sometimes termed the relative risk) and the odds ratio (OR). These measures are an estimate of how many more times likely a horse is to develop the disease if he's been exposed to a potential causative agent than one that hasn't been exposed.


Cohen also pointed out that an apparent association between a given risk factor (for instance, breed) and a disease (such as colic) might be explained by other factors, called confounders, that are associated with both the exposure and disease being studied. In the colic example, confounders could include activity level or feeding practices. While there appears to be an association between breed and colic, the apparent association might be because activity level is associated with breed and with colic. Thus, an apparent association between breed and colic might be confounded by activity level. For example, evidence exists that feeding more concentrate and high intensity exercise are an apparent association between Thoroughbred breed and colic might be the result of differences in activity level or diet if, in general, Thoroughbred horses have greater exposure to high intensity group of Arabian horses might differ from non-Arabian horses--other than their breed--in factors such as diet or activity.

"A confounder can make an apparent association (harmful or protective) disappear, can mask an existing association (harmful or protective), or can change a harmful effect to a protective effect or vice versa," Cohen continued. A confounder might be something that investigators have measured and, therefore, can account for, or it might be a factor that has not been measured and so is not possible to consider. Regardless, Cohen remarked, "It is safe to assume that there is some degree of confounding in any observational study because it is virtually impossible for any two groups of horses to be identical for all factors except the one of interest."

Positive Predictive Values

Interpreting test results is essential for medical management of patients. High sensitivity (the probability that actual cases will have a positive test result) and high specificity (the probability that horses without the condition will have a negative test result) don't ensure that a test will perform accurately. Positive predictive value refers to the probability that a positive test means a horse truly has disease (i.e., a positive test is truly positive). Conversely, a negative predictive value looks at the probability that a negative test is truly negative.

A veterinarian's assessment of the probability of disease before running a test influences how he or she might interpret it. For example, consider testing for equine protozoal myeloencephalitis (EPM) in a horse with classical neurologic signs of EPM compared to testing for EPM in a horse with bilateral forelimb (both limbs) lameness localized to the heels. There is a very low chance that the heel-sore horse will test positive for EPM, whereas a positive EPM test in the neurologic horse has a high predictive value for EPM. Cohen noted, "If it looks like a duck, is given a duck test, then it's likely to be a duck." He stressed the importance of incorporating prior knowledge and subjective impressions into test interpretation: "Problems are often not with the test but rather with human expectations and interpretations."

The Value of Epidemiology

Cohen summarized epidemiology as the fundamental science for clinical research and practice, stating that understanding its principles, methods, strengths, and limitations is essential, since veterinarians use it for all clinical activities. He maintained that professional advances occur through scientific inquiry, with the most relevant research focused on patient-based observations. While veterinarians must strive to continually learn, he urged practitioners also to contribute to advancing the profession through patient-based research.

About the Author

Nancy S. Loving, DVM

Nancy S. Loving, DVM, owns Loving Equine Clinic in Boulder, Colorado, and has a special interest in managing the care of sport horses. Her recent book, All Horse Systems Go, is a comprehensive veterinary care and conditioning resource in full color that covers all facets of horse care (available at or by calling 800/582-5604). She has also authored the books Go the Distance as a resource for endurance horse owners, Conformation and Performance, and First Aid for Horse and Rider in addition to many veterinary articles for both horse owner and professional audiences.

Stay on top of the most recent Horse Health news with FREE weekly newsletters from Learn More

Free Newsletters

Sign up for the latest in:

From our partners