Generating evidence involves three steps: asking the right research questions, finding or collecting the fit-for-purpose data, and conducting the appropriate analyses. The first step in any research is to formulate the research question clearly and accurately. The research question should specify the problem or gap in knowledge to be addressed and how it will be investigated in the proposed research. The research question should formulate the ‘why’ (main justification for starting the research), the ‘what’ (exposure and endpoints), the ‘who’ (target population), the ‘how’ (main study design) and the ‘when’ (time period of the study) of the research in a way that helps understanding the choice of study objectives and methods. It should make it clear whether a hypothesis will be tested and in this case whether the hypothesis is pre-specified or data driven. The research question should also state who will be the primary end-users of the study results (e.g. regulators or public health authorities, health technology assessment organisations, payers, pharmaceutical company, research community).
Previous findings are useful for the methodological planning of the current study and may support the background, research question, hypotheses and design of the proposed study. They may also serve to determine the expected effect size and, if available in the target population, to characterise risk factors for the event, to identify relevant outcomes and measures and to assess the feasibility of the proposed study. A critical and thorough review of the literature forms the basis for the theoretical framework of the research question and should usually be included in the background section of a protocol. Such a review aims to evaluate current evidence around the question at hand and identify gaps in knowledge that a study is intended to fill. How to formulate research recommendations (BMJ 2006;333:804-6) proposes the EPICOT format with 5 core elements (Evidence, Population, Intervention, Comparison, Outcome and Time stamp) for research recommendations on the effects of treatments following systematic reviews. This format was adopted by the European network of Health Technology Assessment (EUnetHTA) in its Position paper on how to formulate research recommendations. Chapter 5.7 and Annex 1 present methods for reviewing and synthesising findings from the literature through the means of systematic review and meta-analysis.
When the study data source is not well characterized or known, a feasibility study should be considered. The aim of a feasibility study is not to answer the research question directly but to determine whether the data source could answer it within the expected timelines and what is the required statistical power for the proposed study design. Feasibility studies can provide information on the number of people with a specific exposure or outcome, the availability of covariates and the follow up period needed. A feasibility study can also provide insights into the potential difficulties which may be encountered in the conduct of the study or which may introduce bias. The ISPE Good pharmacoepidemiology practice (GPP) explains how a data collection method or data source can answer a research question with justifications coming from the feasibility study when relevant.