How do I begin a medical research study?

How to Begin a Medical Research Study

Start by formulating a clear, focused research question using the PICO framework (Population, Intervention, Comparison, Outcome), then develop a detailed written protocol before collecting any data. 1

Define Your Research Question

Your research question must be clear, focused, succinct, unambiguous, clinically important, and answerable 1. Use the PICOT framework to structure your question systematically 1:

• P (Population): Define the specific patient population or problem you're studying
• I (Intervention): Specify the intervention, test, or marker being evaluated
• C (Comparison): Identify the reference standard or comparison group
• O (Outcome): Determine what you're measuring (mortality, morbidity, quality of life, diagnostic accuracy)
• T (Timing): Establish when the intervention or test occurs

Research questions should meet the FINER criteria: Feasible, Interesting, Novel, Ethical, and Relevant 1. Base your question on past clinical observations, extensive literature review to identify what is already known versus unknown, and awareness of existing gaps in clinical practice 1.

Conduct a Comprehensive Literature Search

Before designing your study, perform a thorough literature search using multiple databases 1. The most commonly used are PubMed, Web of Science, and Google Scholar 1.

Critical steps for literature searching:

• Document everything: Record which search engines you used, specific keywords, date boundaries, number of papers identified, number excluded, and reasons for rejection 1
• Use combinations of keywords restricting the search to specific experiments, study types, and populations 1
• Perform reference tracing in identified articles and review articles 1
• Many journals require PRISMA workflow charts for publications, so keeping detailed records during search and selection eliminates the need to repeat the literature search later 1

Develop a Written Research Protocol

A research protocol is your essential "road map" and must be completed before starting data collection 1, 2. Best practice is to publish or deposit this protocol online where possible 1.

Your protocol must include:

• Abstract and study description 2
• Specific objectives, hypotheses, and aims: Develop by outlining a general research topic (objective), creating a hypothesis from the broad objective, translating it into the null hypothesis, then listing steps to refute or accept the null hypothesis 2
• Study design and methods: Type of study, target population, sample size with statistical power calculations, subject selection criteria, data collection methods, data management plan, and statistical analysis outline 2
• Ethical considerations and institutional review board approval plans 1, 2
• Budget and investigator descriptions 2
• Significance of the study 2

Assemble Your Research Team

Establish a team with appropriate expertise and experience, including a statistician as early as possible in the planning stage 1. The team should include:

• Clinical experts in the relevant field
• Methodologic experts (biostatisticians, epidemiologists)
• Patient/stakeholder representatives to ensure the research addresses clinically relevant questions 1
• Data managers and research coordinators 1

Manage conflicts of interest for all team members 1. All researchers must follow best legal, institutional, medical, and integrity practices and ensure appropriate training 1.

Determine Appropriate Study Design

Select a study design that matches your research question 1. The hierarchy of evidence varies by question type:

• Randomized controlled trials (RCTs) provide maximum-quality evidence on benefits and risks of new interventions or diagnostic tests, with clinical outcomes being the most rigorous evaluation method 1
• Prospective cohort studies ensure higher levels of evidence and should be prioritized over retrospective designs 1
• Registries assess real-world use of interventions and their clinical implications, costs, and effectiveness 1
• Descriptive case series are appropriate for defining tumor characteristics or rare conditions 1

Sample size considerations:

For meta-analyses, include at least 17-20 experiments to have sufficient power to detect smaller effects and ensure results aren't driven by single experiments 1. However, this depends on expected effect size—strong effects may require fewer experiments 1.

Plan Data Collection and Management

Pilot your data collection methods before full implementation 1. Details overlooked at the planning stage very often come to light during piloting 1.

Key planning elements:

• Predefine experimental groups, outcome measures, and stopping points 3
• Establish standardized data collection methods 3
• Define clear inclusion/exclusion criteria 3
• Plan for handling potential biases 3
• Ensure appropriate data storage and confidentiality measures 1
• Document all procedures in detail so the study can be conducted successfully 2

Common Pitfalls to Avoid

• Never start without a detailed protocol, even for small studies 1, 2
• Don't choose topics of marginal interest or ignore important prior work in the field 1
• Avoid projects without clear research questions or methodological rigor 1, 3
• Don't use inadequate reference standards in accuracy studies—choose appropriate gold standards considering patient safety and ethics 1
• Never proceed with insufficient sample sizes for your expected effect size 1
• Don't neglect to involve patients/stakeholders in research design, as this ensures clinical relevance 1
• Avoid poor record keeping and inadequate documentation, which constitute questionable research practices 1

Ensure Ethical Compliance

Before commencing work, ensure all researchers know the study plans and how to conduct the work 1. Obtain institutional review board approval and ensure compliance with:

• National and international data protection requirements 1
• Governance and confidentiality requirements 1
• Conflict of interest disclosure policies 1
• Contractual obligations and intellectual property agreements 1