Medical Research Topics for Students: Narrow Ideas You Can Finish and Defend

Before I settle on a medical research topic, I need answers to a few questions.

Do I have enough patients, records or samples within my rotation or posting to reach a workable sample size?
Is my question observational, or would it need trial-level approval I cannot realistically get?
Has someone in my hospital or department already studied the same question?
Should I collect my own data, use a registry or national survey, or run a systematic review?
Which reporting guideline will my paper follow, and does my supervisor know it well?

Medical research covers clinical medicine, laboratory science and public health, so no single thesis can cover it. These topics focus on clinical and population-health questions for final-year MBBS and BDS projects, MS and MPhil theses, and FCPS trainees. Nursing practice questions have their own resource in nursing thesis topics.

What to settle before you pick a medical research topic

Match the design to what you can access

Most student projects are observational: a cross-sectional survey, a retrospective chart review, a case-control study or a small cohort. Clinical trials need stricter approval, funding and follow-up, so they rarely fit a student timeline. As a result, a well-designed observational study usually finishes sooner and defends better.

Ethics comes first

Any study involving patients, their records or their samples needs approval, usually from your hospital or university ethical review committee. Ask early which committee applies and how often it meets. Retrospective record reviews may still need approval, so do not assume they are exempt.

If you are training for FCPS

The College of Physicians and Surgeons Pakistan requires FCPS trainees to submit a synopsis, have it approved, and then complete a dissertation before the FCPS-II examination. According to its published FAQs, trainees should choose the topic with their supervisor, check that it has not been over-studied or duplicated in their institute, and attend the research methodology workshop before writing the synopsis. Read the full guidance in the CPSP synopsis and dissertation FAQs.

Position correct as of September 2026. Check CPSP and your own university for the current rules.

If you have no patient access

Not every student can work in a ward. Survey-based studies among students, registry or survey data, laboratory record reviews and systematic reviews all avoid direct patient contact. Several topics below suit this route, including the headache study, the vitamin D study, the AI attitudes survey and the review-based options.

Testing a topic before you commit

Search recent Pakistani and regional studies first, then narrow your question until your contribution is clear. A research gap is something you confirm by reading, not something you assume. Then check that your planned sample size matches what your unit sees in a year. Finally, ask whether you can explain the question, the design and the main limitation in two minutes, because your panel will expect exactly that.

Medical research topics by specialty

Cardiology

1. Delay between symptom onset and arrival in ST-elevation myocardial infarction

Interview patients at one tertiary cardiac centre about what happened first: a clinic visit, a wait at home or a transfer from another hospital. Compare delay times across these routes. Recall bias is the main limit, so record times from triage notes where you can.

2. Adherence to statins and antiplatelets six months after coronary stenting

Combine self-report with prescription refill records at one outpatient clinic. The comparison of what patients say against what they collect is the useful finding.

3. Blood pressure control among treated hypertensive patients in one outpatient clinic

Look at which factors separate controlled from uncontrolled patients, such as number of drugs, visit gaps and cost. Measure blood pressure with a standard protocol so your data can be trusted.

4. Echocardiographic screening for latent rheumatic heart disease in school-age children

A screening yield and feasibility study in one area. It needs a cardiology partner with echo access and careful parental consent, so treat it as a more demanding project.

Oncology

5. Stage at presentation and time to biopsy in breast cancer at one oncology centre

Review records for stage and the interval from first symptom to tissue diagnosis. Add a short patient interview to explain the delays that records cannot show.

6. Smokeless tobacco and oral potentially malignant disorders

A case-control study at one oral surgery or dental outpatient clinic, comparing use of naswar, gutka and paan between patients with lesions and matched controls. Measure duration and frequency of use, not just yes or no.

7. Out-of-pocket cost and treatment abandonment among cancer patients at one public hospital

A mixed design pairing a short cost questionnaire with interviews. Focus on one cancer type so the sample stays comparable, and take care with sensitive financial questions.

Neurology

8. Door-to-needle time for stroke thrombolysis at one emergency department

Audit each step from arrival to treatment and find where time is lost, such as scanning, consent or transfers. A retrospective audit is quick, but records must be complete.

9. Epilepsy treatment gap and stigma among adults at one neurology clinic

Measure how many patients are untreated or on irregular treatment, and pair this with a validated stigma scale. Confirm the scale’s translation before use.

10. Depression at three months after stroke and functional outcome

A small prospective cohort using the modified Rankin scale and a validated depression tool. Loss to follow-up is the main risk, so plan reminders from day one. Students interested in the mental health side can also browse psychology research topics.

11. Headache burden and disability among medical students in one university

A survey using the MIDAS disability scale. It needs no patient access, which makes it practical for a student without ward exposure.

Paediatrics

12. Serum ferritin control and chelation adherence in transfusion-dependent beta-thalassaemia

Follow children at one thalassaemia centre and compare ferritin trends with adherence to chelation. Adherence needs more than one measure, for example caregiver report plus pharmacy collection.

13. Recovery and default rates in an outpatient therapeutic programme for severe acute malnutrition

A retrospective review of programme records at one site, with a follow-up interview for children who dropped out. The 2018 National Nutrition Survey can give background for your introduction.

14. Resistance patterns in neonatal sepsis isolates at one NICU

Use laboratory records over a fixed period to describe organisms and antibiotic resistance. Check that the laboratory follows a consistent testing standard across the period you choose.

15. Parental knowledge and practice on managing childhood fever

A cross-sectional survey in one paediatric outpatient clinic. Compare stated practice with current guidance, and note misuse of antipyretics or antibiotics.

Infectious disease

16. Antimicrobial susceptibility of Salmonella Typhi at one diagnostic laboratory

Extensively drug-resistant typhoid emerged in Sindh in 2016, so resistance data is relevant to local practice. Report susceptibility to third-generation cephalosporins and change across years if your laboratory keeps records.

17. Non-response to direct-acting antiviral treatment in hepatitis C at one liver clinic

Compare responders and non-responders by fibrosis stage, prior treatment and adherence. The main limitation is small numbers of non-responders, so plan the analysis accordingly.

18. Antibiotic prescribing for acute respiratory infections in outpatient clinics

Audit prescriptions against a named guideline and record how often antibiotics are given without a clear indication. Ethics wording matters, because prescribers may feel judged.

19. Patient and health-system delay in tuberculosis diagnosis at one urban treatment centre

Separate the time patients wait before seeking care from the time the system takes to diagnose. Interviews add the reasons behind each delay.

Endocrinology and metabolic disease

20. Secondary analysis of the National Diabetes Survey of Pakistan

The second NDSP ran across all four provinces from 2016 to 2017 and is described in a BMJ Open paper. Published sub-studies already cover tobacco use, HbA1c cut-offs and obesity, so choose a different pair of variables and confirm nothing similar exists before you start. Ask the research group about data access first.

21. Risk factors for amputation in diabetic foot ulcers at one hospital

A retrospective cohort comparing patients who kept the limb with those who did not. Record ulcer grade, infection, vascular status and time to first review.

22. Vitamin D status and sun exposure habits among female medical students

Combine a laboratory value with a short questionnaire on sun exposure, clothing and diet. Fix the cut-off you use and justify it, because definitions of deficiency differ.

Public health and health systems

23. Air quality and respiratory outpatient visits in one city over one winter

An ecological study linking daily air quality readings to daily visit counts at one hospital. Weather and viral seasons can confuse the result, so include them in your model.

A descriptive audit of cases, outcomes and risk groups such as outdoor workers and older adults. It is quick to plan and useful for local preparedness.

25. Dropout between the first and last dose of routine childhood vaccines in one union council

Use vaccination card review and short caregiver interviews. Ask about distance, cost, health worker contact and fear of side effects.

26. Outpatient waiting time and patient satisfaction at one public hospital

Measure real waiting times with a simple log and compare them with satisfaction scores. Students who lean towards service management can also use management thesis topics for framework ideas.

27. Burnout among house officers and postgraduate trainees in one teaching hospital

Relate burnout scores to weekly hours, night duties and supervision. Check the licence terms of your scale, and protect anonymity, because trainees may fear their supervisors.

Medical AI and diagnostics

28. Chest X-ray AI software for tuberculosis against radiologist reading and laboratory confirmation

A diagnostic accuracy study following the STARD guideline. You need permission for image use and a reference standard, so agree access with radiology before proposing it.

29. Agreement between AI grading and ophthalmologist grading of diabetic retinopathy

Compare grading of fundus photographs at one eye clinic. Report agreement with a suitable statistic and describe cases where they disagree.

30. Attitudes of doctors and medical students towards AI tools in clinical decisions

A survey across two medical colleges. It avoids patient data, so it suits students who want an AI topic without clinical access.

31. Performance of a skin lesion classification model on darker skin tones

Many imported models were trained on images from lighter skin. Testing one on consented local dermatology images could show whether performance changes, but image ownership and consent make this demanding.

Review-based options

Some departments accept a review as a thesis or paper. Confirm this with your supervisor before starting.

32. Barriers to breast cancer screening among South Asian women

A systematic review following PRISMA with a clear question and a documented search strategy. Register or record your protocol before you search.

33. Effect of mobile phone reminders on childhood vaccination completion in low- and middle-income countries

A systematic review or meta-analysis, if enough comparable trials exist. Check this before committing, because too few studies would leave you with a narrative synthesis.

34. Antibiotic resistance in typhoid isolates across South Asia

A scoping review that maps what has been reported and where gaps remain. It works well as groundwork for a later primary study.

Common mistakes when choosing a medical research topic

The first mistake is choosing a whole disease instead of one setting and one question. “Diabetes in Pakistan” cannot be finished. “Amputation risk in diabetic foot ulcers at one hospital” can.

Second, students often choose a sample size that their unit cannot supply in time. Count last year’s cases before you commit.

Third, many overlook approval timelines and data access. Write to the data holder or committee first, not last.

Finally, a topic copied from an online list may already be done in your hospital, so check recent local studies.

Where to go next

Pick two or three ideas, test each against the checks above, and take your shortlist to your supervisor. Once you have a direction, you can send your topic, level and deadline through the order form to get support with the proposal or analysis plan. We have spent several years on the problems you are facing now: an unclear topic, a weak literature review, results that will not add up. The support is meant for reference and learning, so use it according to your institution’s policy.

Students often want to see how others found the support, so you can also read what students have said about working with us.