The conduct of meaningful clinical research is challenging. First, it is sometimes difficult to create clinically relevant and testable hypotheses. Additionally, clinically generated data include many uncontrolled variables and a lack of uniformity of test subjects, resulting in a large degree of variability in data. These characteristics make it necessary to have more subjects, so that study power is adequate to provide meaningful results. Further, there is often limited funding for such studies; for these reasons, academic clinicians, who produce most of the equine research, find it simpler and better for their careers in academic departments to conduct prospective benchtop-type research. However, properly conducted clinical studies are those with the most direct applicability to clinical medicine, at least in the relative short term.
Dr Peter Rossdale, the long-time and respected editor of the Equine Veterinary Journal, was consistent in his call for clinical research from the practitioner, reflecting that the best place to resolve the real problems faced by clinicians, to the betterment of equine health care, was to use ‘the clinic as laboratory’. Albeit an admirable goal, the realities of the demands of clinical practice and associated difficulties in conducting quality research, communicating with other investigators in a timely manner, and the shortage of financial support for equine research have historically made such efforts difficult, although notable exceptions can be identified.
There are a number of study designs that are appropriate to address clinical questions that can have a direct impact on the health and welfare of horses, and for which primary care clinicians can serve a very important role. Case reports and case series are the historical purview of the clinician–scientist, and they remain important means to advance veterinary medical knowledge. However, they are descriptive in nature and do not answer questions about diagnosis or treatment. As such, they represent the least robust of the investigative methods for clinical medicine. Randomised controlled clinical trials are widely considered the gold standard of clinical investigative science, yet are unfortunately less common than desired in veterinary medicine. There are a number of observational study designs, however, that can be powerful in investigating clinical illness and which are well suited for participation by clinicians. These include case–control, cohort and cross-sectional studies. Cohort studies begin with a patient with exposure risk (or a specific treatment, for example), and follow them over time, comparing them to patients without the exposure risk of interest. Cross-sectional studies are very similar, although the outcome is assessed at one point in time. Case–control studies generally compare patients with the disease and those without, and then examine historically for differences in exposure risk. Such studies can be conducted from a single centre, or as a multicentre trial.
A multicentre trial is a collaborative effort that involves more than one independent centre in the tasks of enrolling and collecting data on study participants. The benefits of a multicentre study include the opportunity for an increased number of study subjects, representation of different geographic regions, different breeds or types of horses, comparison across study centres, and also a reduced amount of time necessary to accrue adequate data. These factors result in a more heterogeneous study population and results that are more ‘generalisable’ to the overall population. In addition, it affords people the opportunity to participate in science and research when they otherwise might not have the chance.
However, multicentre studies are more difficult and more expensive to perform than single centre studies, and there are problems related to quality control of data collection and interpretation. A uniform definition of the disease condition and consistent application and interpretation of diagnostic test results and end points are of critical importance, but are more difficult to establish across multiple centres than just one. Individuals must carefully consider their role in the study and evaluate their availability to comply with the study requirements. In addition, does the clinical investigator have the technical ability, equipment and facilities to contribute meaningfully to the study? Does he/she have cases within their practice area that fit the selection criteria? If the answer to any of these questions is ‘no’, then it is inadvisable for the individual to begin.
Planning and prestudy organisation are required to address these concerns. This usually takes the form of prestudy meetings of all participants. This can be done by email or conference calls for small studies – but may require a formal ‘on-site’ meeting for all participants to have adequate input and to comply with regulatory issues if necessary. These regulatory issues may include training in good clinical practice methods, which are required for regulated studies in the USA; similar regulatory agencies and guidelines exist in other countries. In addition, after study initiation, the study organiser often needs to have on-site visits to each study participant to validate data entry and ensure consistent adherence to study protocol.
Additional challenges to performing the multicentre study are the identification of study collaborators and the recruitment of cases. In the past, recruitment of study collaborators depended upon identification of people working and publishing in the field, prominent members of professional organisations, and people that you already knew. This approach may limit participants to a geographic area or exclude people who not known to the study organisers or who are actively working or publishing in the field – both of which might induce a bias into the study results, as well as exclude potentially valuable contributors to the study.
The birth of the information age, however, has removed at least some of these obstacles by making communication more convenient and efficient. Many veterinary specialty organisations have an email LISTSERV, which is used by the membership to discuss common topics and concerns of the organisation. These are valuable tools with which to recruit collaborators as well as cases that may be seen by the list members. Such LISTSERVs are usually national, if not international, in scope and allow for the identification and accumulation of rare or uncommon cases. Further, breed associations usually maintain a website presence, and there are disease-specific websites maintained by laymen and veterinarians with a specific interest in the condition. Communicating with such groups is a powerful means to recruit cases, but the investigator must be careful about introducing case selection bias in this process.
The traditional method of data collection in multicentre trials is to create a data capture form, which is completed by the investigator then returned to the study coordinator for entry into the appropriate database for analysis. This method, albeit effective, is cumbersome and time consuming and adds the opportunity for transcription errors. Use of online resources minimises these problems; such resources are widely available.
Once the required data have been identified, Internet-based survey instruments (SelectSurveya for one example) can be used to provide an interface for entering data. Alternatively one can use commercial database software (such as Access)b, which can be linked to a website, distributed on the Internet and then completed by the study participants as the data are collected. Using these types of data collection method, the investigator can list potential responses, allowing the person entering data to select the response from a pull-down list. Although it takes some time to create these types of spreadsheet, it is time well spent as it speeds up data entry substantially and reduces the number of data entry errors. This standardisation of terminology adds greatly to the integrity of the data collected among different study centres, and reduces the burden of subsequent data validation.
A simpler alternative is to create a spreadsheet and distribute it electronically to the participants to fill in. Once collected, the spreadsheets can be combined centrally into a single database for analysis. Accuracy and standardisation of data entry are more difficult to obtain using this method.
Although clinical research remains challenging, use of the Internet to aid communication among investigators, enrol study subjects and collect and enter data is revolutionising this type of veterinary research. Owing to the penetration of the Internet into all communities, opportunities for clinicians in practice to participate in clinical research are enhanced. I believe that we can expect many more multicentre clinical research studies in the future using the tools described above. The Internet is indeed a welcome addition to our efforts to improve equine health.