Medical research follows a structured process: formulate a clear question, design a study that can answer it, get ethical approval, collect and analyze data, then report your findings using established standards. Each step has specific tools and frameworks that the medical community expects you to use. Whether you’re a medical student planning your first project or a clinician exploring a clinical question, here’s how the process works from start to finish.
Start With a Focused Research Question
A vague curiosity won’t get you far. The standard approach in evidence-based medicine is to structure your question using the PICO framework, which breaks any clinical question into four parts: Population (who are you studying?), Intervention (what treatment or exposure are you investigating?), Comparison (what’s the alternative, such as a placebo or standard care?), and Outcome (what result are you measuring?).
For example, instead of asking “Does exercise help with depression?”, a PICO-framed question would be: “In adults with moderate depression (population), does structured aerobic exercise three times per week (intervention), compared to antidepressant medication alone (comparison), reduce symptom severity scores after 12 weeks (outcome)?” This specificity matters because it directly shapes what kind of study you’ll design, what data you’ll collect, and how you’ll search the existing literature to see what’s already been done.
Search the Existing Literature First
Before designing anything, you need to know what research already exists on your question. PubMed, the free database maintained by the National Library of Medicine, is the primary tool for medical literature searches. Learning to use it well will save you enormous time.
The key skills are using MeSH terms and Boolean operators. MeSH (Medical Subject Headings) are standardized labels that PubMed assigns to articles. When you search using a MeSH term, the system automatically “explodes” it to include narrower related concepts underneath it. So searching the MeSH term for blood pressure will also pull up results indexed under more specific terms like ambulatory blood pressure monitoring. You combine MeSH terms and keywords using the Boolean operators AND, OR, and NOT. Using AND narrows your results (returning only articles that match both terms), OR broadens them (returning articles matching either term), and NOT excludes a term. A well-constructed search string using these tools can be the difference between finding 15,000 irrelevant results and finding 80 highly relevant ones.
Read systematically. Pay attention to what study designs were used, what sample sizes were common, what outcomes were measured, and where the authors identified gaps. Those gaps are where your research question lives.
Choose the Right Study Design
Medical evidence is ranked in a hierarchy, and the design you choose determines how much weight your findings will carry. At the top sit systematic reviews and meta-analyses, which pool data from multiple studies. Just below them are randomized controlled trials (RCTs), where participants are randomly assigned to receive the treatment or a comparison. Below RCTs come cohort studies (following groups over time), then case-control studies (comparing people with a condition to those without), and at the bottom, case series and expert opinion.
Your question dictates what’s appropriate. If you’re testing whether a new drug works better than an existing one, an RCT is the gold standard. If you’re investigating whether a workplace exposure increases disease risk over decades, a cohort study makes more sense. If you’re describing an unusual pattern of symptoms you’ve noticed in a handful of patients, a case series is a valid starting point. Not every question requires or permits a randomized trial, and choosing a design that doesn’t match your question is a common beginner mistake.
Practical constraints matter too. RCTs require significant funding, large participant pools, and long timelines. A retrospective chart review uses data that already exists. Be realistic about your resources, especially for a first project.
Get Ethical Approval
Any research involving human participants requires approval from an ethics review committee before you collect a single data point. In the United States, this is an Institutional Review Board (IRB). Other countries have equivalent bodies. The ethical foundation for all modern medical research is the Declaration of Helsinki, first adopted in 1964 and most recently updated in October 2024. It establishes principles like informed consent, minimizing risk to participants, and ensuring that the potential benefits of research justify any harms.
For IRB submission, you’ll typically need to prepare your full study protocol, your informed consent form, any recruitment materials (flyers, emails, scripts), and documentation of your qualifications. The board reviews whether your study design adequately protects participants, whether consent is truly informed and voluntary, and whether you’ve planned for reporting any adverse events. Studies involving vulnerable populations, such as children, prisoners, or pregnant individuals, face additional scrutiny.
The review process takes weeks to months depending on your institution and the level of risk involved. Some minimal-risk studies, like anonymous surveys, may qualify for expedited review. Plan for this timeline early, because it’s one of the most common sources of delay in research projects.
Protecting Patient Data
If your research involves health records or patient information in the U.S., you’ll need to comply with HIPAA’s privacy rules. When publishing or sharing data, patient information must be de-identified. The Safe Harbor method requires removing 18 specific categories of identifiers: names, geographic details smaller than a state (with limited exceptions for ZIP codes covering populations over 20,000), all date elements except year (including birth dates, admission dates, and discharge dates), phone numbers, fax numbers, email addresses, Social Security numbers, medical record numbers, health plan numbers, account numbers, certificate and license numbers, vehicle identifiers, device serial numbers, URLs, IP addresses, biometric identifiers like fingerprints, full-face photographs, and any other unique identifying number or code.
Even if you’re not publishing the data, your IRB will want to see a plan for how participant information is stored, who has access, and how it will be destroyed after the study ends. Encrypted storage and restricted access logs are standard expectations.
Collecting and Analyzing Your Data
How you collect data should be determined before the study begins, not figured out along the way. Your protocol should specify exactly what variables you’re measuring, how you’re measuring them, and at what time points. Consistency matters: if two research assistants are collecting data differently, your results become unreliable.
When it’s time to analyze, your choice of statistical test depends on the type of data you have and how it’s distributed. If your data follows a normal, bell-curve distribution, you’ll use parametric tests and report means and standard deviations. If it doesn’t (which is common with smaller samples or skewed measurements), you’ll use nonparametric tests and report medians and interquartile ranges. Getting this wrong is one of the most frequent statistical errors in published medical research. If statistics aren’t your strength, partnering with a biostatistician early in the design phase is far better than trying to fix analysis problems after data collection is complete.
Decide your sample size before you start recruiting. A study with too few participants may not have enough statistical power to detect a real effect, and a study with too many wastes resources and exposes more people to research procedures than necessary. Power calculations, which estimate the sample size needed to detect a meaningful difference, are a standard part of study planning and something your IRB will expect to see.
Register Your Trial
If you’re running a clinical trial, registration is legally required in many jurisdictions. In the U.S., applicable clinical trials must be registered on ClinicalTrials.gov no later than 21 days after enrolling the first participant. Registration requires detailed information about your study protocol, including who the responsible party is and how to contact them. Most major medical journals will not publish trial results that weren’t registered before enrollment began, so even if registration isn’t legally required for your specific study type, doing it strengthens your credibility and publication prospects.
Writing and Reporting Your Results
Medical research papers follow a standard structure (introduction, methods, results, discussion), but the specific details you’re expected to report depend on your study type. The EQUATOR Network maintains reporting guidelines for each major design. CONSORT covers randomized trials, requiring you to document things like how randomization was performed and how many participants dropped out at each stage. STROBE covers observational studies like cohort and case-control designs. PRISMA covers systematic reviews and meta-analyses. Following the correct checklist isn’t optional for most journals; editors will return your manuscript if it doesn’t comply.
Authorship on the paper should reflect genuine contribution. The International Committee of Medical Journal Editors defines four criteria that must all be met: making substantial contributions to the study’s conception, design, or data work; helping draft or critically revise the manuscript; approving the final version; and agreeing to be accountable for the integrity of the work. Simply providing funding, collecting data without involvement in analysis, or holding a senior position in the department doesn’t qualify someone for authorship. These conversations are best had at the start of a project, not at the end when feelings and politics get involved.
Common Pitfalls for New Researchers
The biggest mistake beginners make is choosing too broad a question. A focused, answerable question with a realistic design will produce a publishable paper. An ambitious question that can’t be adequately studied with your available resources will produce nothing.
Other frequent problems: underestimating how long IRB approval takes, failing to keep meticulous records of every protocol deviation, not planning data management before collection begins, and choosing statistical tests after looking at the data (which introduces bias). Medical research rewards careful planning far more than it rewards speed. Build twice as much time into your timeline as you think you’ll need, especially for your first project, and find a mentor who has published in your area of interest. Their experience navigating the process is as valuable as any textbook.