A medical thesis needs a literature review that does three things: proves you know the field, sets a tight question, and shows exactly how you searched and judged the evidence. This guide gives you a clean, repeatable path you can follow today. You’ll get short steps, examples, and templates you can reuse.
Plan Your Scope And Question
Start with a narrow, clinical question that matters to patients or practice. Spell it out using PICO or a close variant. That means the patient or problem, the intervention or exposure, the comparison, and the outcomes that matter. One sentence is enough: write it in plain language.
Write A Tight PICO
Here’s a simple pattern you can copy: “In adults with type 2 diabetes (P), does continuous glucose monitoring (I), compared with finger-stick testing (C), improve HbA1c and hypoglycemia rates (O)?” Swap in your topic, then list any limits you’ll apply, such as study design or setting.
Set Inclusion And Exclusion
Define rules before you search. Typical choices include study types you’ll accept, time window, languages you can read, and minimum sample sizes. Decide on primary outcomes and a few secondary ones. Store these rules in a protocol document so you never guess later.
Pin Down Feasibility
Scan a handful of recent papers to see how much evidence exists and which designs dominate. If the field is tiny, widen the time window. If it’s crowded, focus on a subpopulation, a care setting, or a clearer outcome.
Pick Your Sources And Build The Search
Use at least two biomedical databases, then add trial registries and grey literature. PubMed gives free access to MEDLINE and uses MeSH terms. The Cochrane Library houses controlled trials and systematic reviews. Add a trials registry so you don’t miss ongoing or unpublished work. If your program allows Embase, include it, because drug and device coverage differs from MEDLINE.
| Source | What it covers | Search tips |
|---|---|---|
| PubMed/MEDLINE | Biomedicine across journals | Combine MeSH with keywords; use field tags and filters wisely |
| PubMed User Guide | How PubMed works | Learn automatic term mapping and MeSH explosion |
| Cochrane Handbook | Methods and risk-of-bias guidance | Copy search structure used in high-quality reviews |
| Cochrane Central (CENTRAL) | Randomized and quasi-randomized trials | Filter by condition and intervention |
| ClinicalTrials.gov | Registered and completed trials | Search by condition, intervention, and status |
| WHO ICTRP | Global trial registries | Cross-check trials not listed elsewhere |
| Google Scholar | Broad academic search | Scan the first few pages for citations to chase |
| Institutional repository | Theses, preprints, reports | Use site search and year filters |
Build A Reproducible String
Mix controlled vocabulary and plain terms. In PubMed, pair MeSH like “Diabetes Mellitus, Type 2” with title/abstract keywords such as “type 2 diabetes” OR “T2D”. Connect concepts with AND, synonyms with OR, and use parentheses so the logic behaves. Save the exact string, database name, and date.
Pilot And Refine
Run a few trial searches. Check that landmark papers appear. If not, add synonyms or remove limits that are choking recall. Keep a log of each change and why you made it.
Document As You Go
For every database, store the exported file, the string, the filters, and the hit count. Name files with the date and database so later steps are painless.
Screen Studies Without Guesswork
Export all records, remove duplicates, then screen titles and abstracts against your rules. Move borderline items to full-text review. If you can, have a second screener and settle disagreements by consensus. Record counts at every step so you can draw a flow diagram later.
Use PRISMA To Show Your Trail
Most programs want a PRISMA flow diagram that shows how many records you found, screened, included, and excluded with reasons. Fill it in as you go, not at the end.
Keep Reasons Tight
Set a small list of exclusion reasons and reuse them. Examples include wrong population, wrong design, wrong outcome, duplicate, or no full text. Consistent wording speeds writing later.
Doing A Literature Review For A Medical Thesis: Step-by-step
- Draft the protocol. Capture your PICO, databases, strings, dates, and rules. Store it in your thesis folder.
- Run the searches. Download results in a consistent format. Keep separate files by database.
- De-duplicate. Use a reference manager to merge matches while keeping a backup of raw exports.
- Title/abstract screen. Apply rules fast and record keep/throw decisions.
- Full-text screen. Read carefully. Record a reason for every exclusion.
- Extract data. Use a standard sheet so every study gets the same fields.
- Appraise quality. Pick one tool per study type and rate consistently.
- Synthesize. Group by design, outcome, and direction of effect. If data allow, run a modest meta-analysis with help from your supervisor.
- Write the chapter. Tell the story of the field, then present what your search found and what it means for your question.
- Share your files. Add your protocol, strings, and extraction sheet to an appendix.
Extract The Data That Matters
Create a sheet before you open the first paper. Common fields include study design, setting, sample, eligibility, intervention details, comparator, outcomes, follow-up, effect estimates with precision, funding, and notes on anomalies. Pilot the sheet on three papers and trim fields you don’t use.
Be Consistent With Outcomes
Define a primary outcome and stick to it. If studies report the outcome in different ways, record the raw numbers and the definitions they used. That way you can transform or standardize later if needed.
Record Effect Measures Clearly
Note which measure each study used: risk ratio, odds ratio, hazard ratio, mean difference, or standardized mean difference. Copy the confidence interval and the time point. If a paper gives adjusted and unadjusted values, store both and tag which one you prefer.
Watch For Multiplicity
Many studies report many time points or overlapping outcomes. Pick a consistent time point that matters clinically and note any sensitivity checks you try.
Judge Study Quality And Bias
Quality appraisal is not a checkbox. It shapes how you read the results and how much weight you give them. Choose one validated tool per design and keep the criteria in front of you while you rate. If two people rate, record both scores and how you resolved differences.
| Study type | Common tool | What to check |
|---|---|---|
| Randomized trial | RoB 2 (Cochrane) | Sequence, concealment, blinding, missing data, selective reporting |
| Cohort or case-control | Newcastle-Ottawa or JBI | Selection, comparability, outcome/exposure measurement |
| Diagnostic accuracy | QUADAS-2 | Patient selection, index test, reference standard, flow and timing |
| Systematic review | AMSTAR 2 | Protocol, search, study selection, synthesis, bias appraisal |
| Qualitative study | JBI qualitative checklist | Sampling, data collection, reflexivity, credibility |
When uncertainty is high, say so. You can down-weight weak evidence in narrative synthesis and explain why.
Map Risk To Your Narrative
Create a small summary table that lists each study with its risk rating and the main outcome. Place it before any pooled figure so readers see the context first.
Synthesize Without Losing The Plot
Start with a narrative overview. Group studies by design and outcome, then by patient or setting. State where findings agree and where they don’t. If numbers are compatible, you can pool effect sizes with a fixed or random approach, but only when studies measure the same construct in a similar way.
When A Meta-analysis Makes Sense
You need comparable participants, interventions, and outcomes, plus enough data to calculate an effect and its variance. Report the model, the measure you used, and any heterogeneity checks. Keep code and inputs so someone else can rerun your result.
Handle Heterogeneity
Differences in methods or patients can pull estimates apart. Before you pool, write down the clinical reasons studies might vary. If the spread is wide, narrate instead of forcing a number.
Draw Clear Figures
If you create a forest plot, label axes plainly and match the order of studies to your tables. Keep captions short and explain any symbols. Place the PRISMA flow figure before the results section so the path is obvious.
How To Conduct The Literature Review For Your Medical Thesis: Write It Up
Use A Layout That Readers Expect
Open with a short purpose paragraph and the exact question. Follow with Methods that mirror your protocol: databases, dates, strings in an appendix, rules, screening process, extraction fields, and tools for bias. Results should start with the PRISMA flow, then the study table and figures. Discussion should cover what the field knows, what your review shows, where the limits are, and what your thesis will tackle next.
Keep Reporting Transparent
Follow the PRISMA 2020 checklist item by item. When you draw on methods for risk of bias or synthesis, cite the online Cochrane Handbook chapter you used so examiners can see your rulebook.
Reference Style And Tools
Pick one citation style and stick to it. Make sure every in-text citation has a matching entry. A manager like Zotero, EndNote, or Mendeley saves time and keeps duplicates down. For medical journals, the ICMJE Recommendations outline authorship and reporting norms that your thesis can mirror.
Place Tables Where They Help
Put the study characteristics table near the start of Results so readers grasp the field at a glance. If you have a quality table, keep it close to the narrative paragraph where you weigh the strength of evidence.
Common Mistakes And Fast Fixes
- Scope creep. If your question keeps growing, freeze the protocol and move spinoff ideas to a later paper.
- Vague rules. Write inclusion and exclusion so a stranger would screen the same way you do.
- Search blind spots. Scan reference lists of key papers and trial registries before you stop.
- One-person screening. If a second screener isn’t available, rescreen a random slice a week later and compare your own decisions.
- Cherry picking. Report all outcomes you listed, not only the ones with tidy effects.
- Messy files. Keep raw exports, deduplicated sets, and final sets in dated folders.
- Thin synthesis. Summaries that list study after study without comparing them read flat. Group, contrast, and explain.
- Unclear figures. Captions that don’t state what the viewer should notice waste space. Tell readers the punchline.
- Missing appendices. Without strings, dates, and forms, the work can’t be checked. Add them.
Files And Templates You Can Reuse
Protocol Outline
One page is enough: title, PICO, objectives, databases and dates, draft strings, inclusion and exclusion, outcomes, screening flow, extraction fields, and planned appraisal tool.
Extraction Sheet
Create a table with rows as studies and columns for design, setting, sample, intervention, comparator, outcomes, follow-up, numbers, and notes. Add a column for bias rating so you can sort.
Search Log
Make a small log with date, database, string, hits, changes you made, and why you made them. Your Methods section almost writes itself from this file.
Study Table Essentials
For a crisp study table, include citation, country, design, sample size, main eligibility points, intervention and comparator, primary outcome, follow-up length, and any remarks that affect interpretation.
Finish Strong And Link To Your Thesis Aim
End the chapter by tying results back to your research aim. Spell out where the evidence is thin, where it is solid, and how your study will add clarity. Thank your reader with a clean summary figure and a short paragraph that states the take-home message from the evidence you gathered.