Portal de Periódicos da CAPES

The use and abuse of questionnaires in veterinary medicine

2015; Wiley; Volume: 47; Issue: 4 Linguagem: Inglês

10.1111/evj.12429

2042-3306

Rachel Dean,

Animal health and immunology

'It is not the answer that enlightens, but the question.' Decouvertes Research that incorporates questionnaires as part of the methodology is becoming more commonplace in veterinary medicine. The evidence generated by studies involving questionnaires can have significant impact on decision makers if undertaken and reported well. As with all aspects of veterinary medicine, there is an art and science to questionnaire-based research. All too frequently, not enough rigour is given to the planning, design, delivery and analysis of this useful methodology. This can lead to substandard and poorly reported science, with results of dubious quality and therefore of limited use to decision makers. We are all familiar with completing questionnaires because many organisations, such as the government (e.g. the national census), consumer groups and deliverers of healthcare, use them to gather information and opinion. It is possible that because people have experience of completing questionnaires and are familiar with them, the design of the questionnaire is perceived to be simple. However, while many of us have experience of an aeroplane flight this does not mean we can pilot the plane! Delivery of good questionnaire-based research requires the same expertise and methodological rigour as undertaking laboratory-based research or colic surgery. It is important to recognise that a questionnaire is a research tool/method and not a type of study. Questionnaires can be used for a number of different study types to gain information about participants and the animals in their care. Common study types for which questionnaires have been used in equine medicine and surgery are cross-sectional 1-3, case control 4 and cohort studies 5. Questionnaires are also used as a tool to undertake surveys to obtain baseline descriptive information about a population without necessarily testing a hypothesis 6. Questionnaires can also be used to collect data for population census and randomised controlled trials. As with all research, the first step is to identify a clear research question and to identify the correct study design to answer that question. For example, if the research question is about the risk of disease, a cohort, case control or cross-sectional study (if static risk factors are of interest) should be designed and implemented. If the research question concerns treatment efficacy then a randomised controlled trial, or possibly a cohort, must be undertaken. A number of different methods could be used as part of these studies, so the selection of a questionnaire as the method of choice should be made carefully. In some situations, it is not clear why a questionnaire was used, and it may be perceived as an easy or quick option, whereas if done properly, questionnaire-based research is neither of those things. Consistent with all other types of research, ethical approval is required for all studies undertaken using questionnaire methodology 7. In many published veterinary studies, the statement of ethical approval of the study is often not reported. It is therefore unclear whether ethical approval was obtained and not reported or whether it was not obtained at all. All participants must be fully informed about the study prior to participation, and researchers need a good knowledge of the Data Protection Act prior to collecting information. EVJ's requirements for ethical approval and transparent declarations have been discussed in a recent issue 7. Before undertaking a questionnaire, it is important to have an understanding of the basics of survey design, writing questions, piloting, distribution and methods of data analysis. There are textbooks (e.g. Dillman et al. 8) and journals (http://jssam.oxfordjournals.org/ 9) available on the subject of questionnaire design and methodology to educate and ensure quality. Despite this, the biggest flaw in the questionnaire studies reported in veterinary medicine is substandard methodology. Involving an epidemiologist or someone with experience of questionnaire methodology before starting the study can make a significant difference to the quality of the end product. Likewise, if it is possible that help will be needed for the statistical analysis, it is invaluable to engage with an epidemiologist or statistician to plan the analysis as part of the study design phase. At the start of any questionnaire, a clear introduction is needed, stating the purpose of the study. For any jargon or scientific terms, definitions should be used so that the understanding of each respondent is as similar as possible. The layout and 'flow' of a questionnaire is also important to ensure that the questionnaire is in a logical order and questions or sections are not missed out by respondents. A number of different question styles, such as open, closed, semi-closed and Likert, can be used. The types of questions depend on the subject being explored and the type of analysis, e.g. quantitative or qualitative. While open questions can elicit some useful quantitative data, if rich data are needed then other qualitative research methodologies, such as focus groups and interviews, may be more useful. A systematic review about increasing response rates to postal questionnaires found that interesting and user-friendly questions increased response rates 10. One of the most important things is that each question is answerable by respondents and that structured (as opposed to open) questions have mutually exclusive answer categories unless the respondent is permitted to provide more than one answer. Each question should be considered carefully before inclusion to ensure that it can be analysed effectively and provides useful information. Piloting of questionnaires with representatives of the target population is the best way to ensure that a questionnaire is user friendly and will deliver useful information 5, 11, 12. When reporting the study, a description of the types of questions used and the structure of the survey should be given 5, 12, 13. It is important to have a link to the questionnaire or, ideally, to submit the questionnaire as supplementary material when the manuscript is sent for publication 2, 5. EVJ requires that a copy of the questionnaire, anonymised and translated into English if necessary, is submitted with the manuscript so that it can be examined by the peer reviewers, and ultimately, published online if the study is accepted. Questionnaire length is an aspect of questionnaire design that is frequently discussed and speculated upon. Shorter questionnaires may achieve better response rates 10. The length may be affected by the subject, the information needed and the mode of delivery, e.g. postal, via the telephone or online. Whatever the length of the survey, the time taken to complete the survey should be part of the piloting process, and at the point of distribution the time commitment to taking part should be made clear. Some types of incentive (money!) have been shown to improve response rates so should be considered 10. Before distributing any survey, the target population should be clearly defined so that the correct sampling frame and sample are identified. A sample size (power) calculation should be undertaken to establish the size of the sample needed 1, 2, 5, 12 and realistic assessment of likely response rates is critical to ensure enough questionnaires are distributed. This should avoid the situation that often seems to happen of 'suddenly needing more people' halfway through the study. A clear definition of the people or cases of interest is needed so that disease and nondisease states, if relevant, are distinguished 4-6. A number of different methods are available for distribution, including post 2, 11, 12, telephone 4, 5, Internet 6 and through a combination of methods 13. The route of distribution affects who can participate (e.g. not everyone has the Internet/telephone), and the limitations of the route chosen and how this may affect the results should be considered in the discussion. Increasingly, questionnaires are delivered online, and more and more design tools are becoming available. Some of these tools are more flexible than others, and questions may need to be adapted to accommodate this and ensure that the questionnaire still elicits the correct information. It can be tempting to let the tools do the thinking for you, but this is not an excuse for a shabby questionnaire. These types of questionnaires still need validating and testing in the same way as any other research tool. Part of the study design and planning of the route of distribution should include the process for reminding potential respondents about the survey. The timing as well as the method of reminding respondents (e.g. via postcards 11, 12) should be stated clearly. The correct reporting guideline/standard, depending on the study type, should be used to report and publish any study in veterinary medicine (http://www.equator-network.org/). The policies about using guidelines vary between journals 14, 15 but, even if they are not explicitly required by the target journal, reporting guidelines are always a very useful checklist for authors to ensure the accurate and complete report of the work done. A well-reported study is much easier to read, and the strengths and weaknesses of the work are clearer to the reader. As with all publications, a full discussion of the limitations of the study is a vital part of the manuscript. All potential biases should be discussed, including sampling bias, nonresponse bias, recall bias and social acceptability bias. The choice of journal for publishing work is based upon a number of factors. There is an argument that if veterinary surgeons and owners have been used in research then the results should be published in a place where they can be accessed by these people. This could be in the form of a report at the end of the study written specifically for the target group or by ensuring that the article is made open access so that anyone can read it. 'No one is dumb who is curious. The people who don't ask questions remain clueless throughout their lives.' Neil deGrasse Tyson The Centre for Evidence-based Veterinary Medicine is supported by an unrestrictive grant from Elanco and the University of Nottingham.