GLOSSARY – An explanation of words and terms used throughout our website.

If there are any additional words, terms or phrases you would like one of our experts to explain and to be listed here please email us at

Anecdotes (LOWEST)
Anecdotes are the lowest level of evidence. This typically takes the form of someone saying they tried a particular product or supplement or technique for a specific problem and they believe it worked. E.g. “my horse had filled legs and I fed 10g a day of avocado extract and the filling disappeared”. The problem with anecdotes is that the author is biased. The problem being addressed may be poorly defined. It also does not take into account other things that may have changed or the fact that many problems resolve spontaneously.

Case Reports
Case reports are usually conducted by specialists such as veterinary surgeons or physiotherapists and consider a well-defined condition either in a single animal or a group of animals all sharing the same condition or symptoms. Case reports may cover unique cases that cannot be explained by known diseases or syndromes, animals that show an important variation of a disease or condition, unexpected events that may yield new or useful information. A good example is a case report on five horses with a newly recognised condition called Equine multinodular pulmonary fibrosis (EMPF) (Schwarz et al. 2013) At the time of the publication of this paper, very few vets had seen horses with this condition and relatively little was known about it so the description of how it affected these horses was very important for other vets.

Cohort Studies
A cohort study is where a group of horses would be followed prospectively. That is, we start to monitor, for example, a group of 2-year-old racehorses when they enter training and plan to endoscope them every month to follow their respiratory health. When we look back at the data after perhaps 6 months we may try to identify the factors associated with horses having more severe respiratory disease.

Randomised Controlled Trials (RCTs)
This is a study where we take a population of horses and then assign them to a control group who have no intervention (treatment) and one or more groups who have a treatment of some description. Take again a racing yard. We may want to look at the impact of calcium intake on fracture risk. At the start of the season we decide which horses will be included. This could be all 2-year-olds. We would then randomly allocate the horses to either receive no extra calcium or a calcium supplement. Even better, the horses allocated to no extra calcium should be given a PLACEBO. This would be a supplement that looks and smells as similar to the calcium supplement as possible. Furthermore, we should not tell either the people feeding the horse or the vets assessing the fractures which horses are on placebo and which are on the calcium supplement. This is called blinding, or in this case double-blind. The third type of blinding is where the statistician who analyses the results is also blinded as to which is the treatment and which is the placebo group. These studies may also be referred to as Double-Blind, Placebo-Controlled studies.

Meta-Analysis (HIGHEST)
RCTs are excellent, but, if another research group repeats the same type of study but with slightly different parameters, e.g. 3-year-old horses or 2-year-old Standardbreds, or 2-year-old Thoroughbreds who are given a slightly different calcium supplement or which are on a different basal diet, then the results may be conflicting. When there is a large number of research papers in a particular area, such as colic or fractures or respiratory disease, then a meta-analysis can be undertaken. Meta-analysis is a method for systematically combining pertinent qualitative and quantitative study data from several selected studies to develop a single conclusion that has greater statistical power. This conclusion is statistically stronger than the analysis of any single study, due to increased numbers of subjects, greater diversity among subjects, or accumulated effects and results. Meta-analysis can be used to establish statistical significance with studies that have conflicting results, to develop a more correct estimate of how strong a particular effect is (e.g. how important is change in forage as a risk factor for colic), to provide a more complex analysis of harms, safety data, and benefits. It can also be used to combine small groups of subjects from different studies where the numbers from the individual study are too small to draw any conclusions.

Publication and Peer Review
Research that has been published in a peer-reviewed scientific journal is considered to be a higher level than non-peer-reviewed information. This is because in the process of peer review, research that has been submitted to a journal is reviewed by usually 2 or sometimes 3 other experts in the same area. They then make comments, suggestions for modification or ask further questions. They will then make a recommendation to the journal editor, who usually invites the original authors to respond to the reviewers’ comments before making a final decision on whether to accept and publish or reject. Abstracts – short, usually 200-600 word summaries of a piece of research – are often published in association with conferences. These may or may not be peer-reviewed. Authors often present their work at a conference prior to full publication, although many studies do not progress to publication, perhaps due to small numbers of subjects or flawed experimental design. It can be difficult to determine if an abstract has been peer-reviewed. Publication in a journal does not always automatically imply it has been peer-reviewed. Research articles in recognised scientific journals can reasonably be assumed to have been peer-reviewed.

Statistical Significance
You can be forgiven for thinking that scientists are obsessed with statistical significance. Why is this?
Well, scientists use a variety of different statistical tests to determine if their findings could have occurred purely by chance. Statistical significance is by convention normally taken as P<0.05, which stands for probability less than 5%. That means that if you repeated the same study 100 times, 95% of the time you would get the same result. But it would also means that 5% of the time or 1 in 20 experiments, you would find something else, which could be no difference or even the opposite. So effectively when we find P<0.05, there is a 95% probability that this is not just random chance. We may also talk about P<0.01, which is a higher level of significance i.e. we can be even more confident in our findings as the chance of this being random is only 1 in 100.


If there are any additional words, terms or phrases you would like one of our experts to explain and to be listed here please email us at