Putting the science into science-based medicine

Texas A&M University professor Dr. Noah D. Cohen shared with equine practitioners how they can use epidemiologic principles and findings in day-to-day practice during his Frank J. Milne State-of-the-Art Lecture. The honorary lecture was held Nov. 20, 2011, during the American Association of Equine Practitioners' 57th Annual Convention in San Antonio, Texas.

The common perception of epidemiology among equine practitioners is that it is concerned with areas such as public health, food safety, and regulatory medicine. In actuality, Dr. Cohen said, epidemiology also is directly relevant to individual patient care. 

Equine practitioners often deal with populations of horses, either at settings such as racetracks, showgrounds, and barns or in groupings by activity level or disease. Also, when examining an individual horse, a veterinarian brings to bear information from the population of similar patients he or she has examined or that have been examined by others. Plus, solid clinical evidence provides the basis for everything a veterinarian does, including collecting a history, obtaining a diagnosis, performing a physical examination, and prescribing preventive treatment.

According to Dr. Cohen, epidemiology is the fundamental science of evidence-based medicine; the most clinically relevant evidence is derived from epidemiological studies of patients.

Practitioners should keep in mind, though, that there is a hierarchy of clinical evidence, he said. At the top of this pyramid are systematic reviews and meta-analyses. Following that are, in order of decreasing reliability, randomized, controlled, double-blind studies; cohort studies; case-control studies; case series; case reports; ideas, editorials, and opinions; animal research; and, finally, in vitro research.

Epidemiology can help veterinarians to consider and identify a cause of disease or other health-related outcome—be it a diagnostic test result, a response to treatment, or a prognosis—by studying associations.

But it's important for veterinarians to remember that single causes that are both necessary and sufficient are rare, Dr. Cohen said. Instead, health outcomes often have many component causes, and disease occurs when all components of a sufficient component set are accumulated. With a racing injury, for example, the practitioner must take into consideration that there are many components that may contribute to an injury event, such as the presence of pre-existing lesions, the type and condition of the track surface, the class of the race, the accumulation of high-speed exercise, and the sex of the horse. Many of these multiple components must come together before an injury occurs.

After introducing basic epidemiologic concepts, including causation, prevalence and incidence, and measures of association such as odds ratios and risk ratios, Dr. Cohen selected five clinically relevant "facts" from published reports and illustrated how only one of the five was likely to be valid.

These examples were as follows:

• That certain breeds are at greater risk for colic.
• That risk factors for strangulating intestinal obstruction, such as a high peritoneal fluid lactate concentration, can be used to classify horses as having a strangulating versus a nonstrangulating obstruction.
• That tests for detecting antibodies against the causal agent of equine protozoal myeloencephalitis are highly sensitive and specific and, therefore, highly accurate for diagnosing the disease.
• That heparin is ineffective for preventing laminitis among horses with forms of colic that require surgery.
• That furosemide reduces the severity of exercise-induced pulmonary hemorrhage.

Of these five, only the validity of the fifth item was considered to be sound, he said.

For example, with regard to the possible association between certain breeds and the risk of colic, Dr. Cohen said these associations might be valid, but they also might be caused by confounding.

Confounding is the apparent association between a given risk factor and disease explained by some other factor for which one hasn't accounted. So, the observed differences in risk of colic between groups might be due to breed, chance, or other factors not taken into account.

"If activity level and breed are associated, maybe breed isn't a cause of colic, but instead, the activity level explains the risk of colic," Dr. Cohen said. "For example, if Thoroughbreds are more active and activity level is known as a risk factor for colic, then the association between Thoroughbreds and colic might be attributable to activity level."

Epidemiology can help veterinarians to consider and identify a cause of disease or other health-related outcome—be it a diagnostic test result, a response to treatment, or a prognosis—by studying associations.

Confounding, then, can result in an association appearing to be present when none exists, or vice versa.

"It's safe to assume there is always some degree of confounding, because two groups compared are rarely ever completely the same. Associations of breed and colic may be confounded, so we may question validity," Dr. Cohen said, and thus, practitioners should carefully consider potential confounders when reviewing the literature.

A recognized expert in equine infectious disease, epidemiology, and internal medicine, Dr. Cohen is the director of the Equine Infectious Disease Laboratory at Texas A&M's College of Veterinary Medicine & Biomedical Sciences. He received his DVM degree from the University of Pennsylvania in 1983. In addition, he holds a master's in public health and a doctorate in epidemiology from the Johns Hopkins Bloomberg School of Public Health. Dr. Cohen is a diplomate of the American College of Veterinary Internal Medicine.