How to choose your MD/MS thesis topic

By Dr. Harshad Ramineni · Updated 1 October 2026

In short

A good MD/MS thesis topic is one you can finish in your own department within the PG timeline, answered with the patients and data you will actually have. It needs your guide’s approval and your institutional ethics committee’s (IEC) clearance before you collect any data. And it passes the FINER test: feasible, interesting, novel, ethical and relevant.

The FINER test

FINER is the standard checklist for judging a research question. It comes from Hulley and colleagues’ Designing Clinical Research; the 2022 fifth edition shortens it to FINE, folding relevance into importance. Put each letter to your idea as a question:

FFeasibleCan I recruit enough participants with the tests, skills and money I have, and finish before my submission date?
IInterestingWill it hold my interest, and my guide’s, through data collection and writing?
NNovelDoes it confirm, refute or extend what is published? Repeating a good study in your own population counts.
EEthicalWould my IEC approve it, with acceptable risk and proper consent?
RRelevantWould every result, including “no difference”, matter to patients, practice or the next study?

The last test is the one residents skip. If a null result would interest nobody, the question is weak.

From idea to question: PICO and PECO

A topic names an area, such as “diabetic foot”. A research question says who you will study, what you will compare and what you will measure. PICO, from evidence-based medicine, gets you there:

PPopulation	Who, where, with what condition: for example, adults with type 2 diabetes in your medicine OPD.
IIntervention	The treatment, test or programme you are studying.
CComparison	Usual care, another drug, or patients without the exposure.
OOutcome	What you measure, how, and at what threshold, defined so that two people would record it the same way.
TTime	When you measure the outcome, or how long you follow patients (PICOT).

Use PECO, with Exposure in place of Intervention, when you observe something patients already have or do. The C is the letter most often missed: “Does drug X lower blood pressure?” has no comparison; “Does it lower blood pressure more than amlodipine?” does. A purely descriptive question, such as “how common is X in our OPD?”, needs no comparison and no hypothesis, but it still needs a sample-size calculation.

Step 1Where good ideas come from

Unanswered questions in your OPD and wards. The patient who did not respond as expected, the test ordered out of habit: careful clinical observation is a classic source of research questions.
Gaps in recent guidelines. Recommendations that rest on limited evidence, or on populations unlike your patients, point to questions worth asking.
Repeating a study in your population. A well-designed study from elsewhere, repeated with the same methods in your patients, confirms or challenges the finding locally.
Your department’s ongoing work. Building on what your guide already studies brings experienced supervision, existing proformas and tested methods.

Before you settle, search PubMed for what is already published and the Clinical Trials Registry – India for studies already registered.

Step 2Check feasibility before you commit

A sample size you can reach. Estimate it now, not after the synopsis (the free sample size calculator gives a first figure), and count the eligible patients your department saw last year. If the numbers fall short, change the question, widen the inclusion criteria or lengthen recruitment.
A duration that fits. Recruitment, follow-up, analysis and writing must finish before your university’s submission date, and recruitment starts only after the synopsis and ethics approval.
Investigations that are available and affordable. Every test should be done reliably in your hospital. If one is not routine care, settle who pays before you commit.
Ethics and consent. Consent in the patient’s own language, and extra protection for children, pregnant women and other vulnerable groups under the ICMR National Ethical Guidelines (2017).
Your guide’s expertise. A question your guide can supervise, with methods and equipment your department already has.
One primary question. A single primary objective drives the sample size; other questions become secondary objectives.

Step 3Choose a design that fits a PG thesis

The question decides the design, not the other way round. Four designs to weigh up:

Cross-sectional

Each participant is studied once, at one point in time.

ForQuick, inexpensive, and the only design that directly measures how common something is.

AgainstExposure and outcome are measured together, so it cannot show which came first or prove cause.

Prospective observational (cohort)

You group patients by exposure, then follow them forward for the outcome.

ForShows that the exposure came first, and measures incidence and relative risk.

AgainstFollow-up must fit inside your timeline, and patients lost to follow-up weaken it. Suits common outcomes that appear within months.

Case-control

You start from patients with the outcome (cases) and without it (controls), and look back for the exposure.

ForEfficient for rare outcomes, with smaller samples than a cohort.

AgainstProne to recall bias; controls must come from the same source population as the cases; it gives odds ratios, not risks.

Interventional, with caution

You assign the treatment, ideally at random.

ForRandomisation gives the strongest evidence that a treatment works.

AgainstHeavier ethics review and safety reporting, and usually registration with the Clinical Trials Registry – India before the first participant is enrolled; your IEC will confirm. A before-and-after study without a control group is not a randomised trial.

A retrospective study from hospital records can also work, but only if the records reliably hold every variable you need.

Step 4Why topics get sent back or stall

It is a topic, not a question. “Drug-induced liver injury” is an area, not an uncertainty you can resolve.
The patients are not there. The sample size was worked out after the topic was fixed, and too few eligible patients come through.
The design label does not match the methods. A single questionnaire visit called a “cohort”, or a before-and-after study called an “RCT”.
There are too many primary objectives. Nobody can tell which one the sample size was calculated for.
It depends on something you do not control. A test the lab does not run, a drug the pharmacy does not stock, or a department that has not agreed to help.

Worked examples: from vague idea to FINER question

These illustrate the method; they are not topics to copy. Your question should come from your own department, patients and guide.

Illustration 1 · Descriptive

Vague idea: “Elderly patients are on too many medicines.”
Question: What proportion of patients aged 65 and above in our medicine OPD are prescribed at least one potentially inappropriate medication by the 2023 AGS Beers Criteria?
Design: P and O only: it describes, it does not compare. Descriptive cross-sectional, so no hypothesis; the sample size comes from the expected proportion and the precision you want.
FINER note: Feasible from prescriptions seen every day. Add “and its association with the number of drugs prescribed” and it becomes analytic, needing a hypothesis and a comparison.

Illustration 2 · Exposure (PECO)

Vague idea: “Something on antibiotic resistance in the ICU.”
Question: Among adults admitted to our ICU (P), is fluoroquinolone use in the 90 days before admission (E), compared with none (C), associated with carbapenem-resistant Enterobacterales colonisation on surveillance culture by day 7 (O)?
Design: Prospective cohort. Null hypothesis: colonisation is equally likely in exposed and unexposed patients.
FINER note: Feasible only if surveillance cultures are routine or funded, and earlier antibiotic use can be documented reliably.

Illustration 3 · Intervention (PICOT)

Vague idea: “Counselling might help patients keep taking their statins.”
Question: In adults newly started on a statin at our OPD (P), does structured counselling (I), compared with usual care (C), increase the proportion who are adherent, defined as a proportion of days covered of at least 0.8 from refill records (O), at 6 months (T)?
Design: Randomised controlled trial, with caution: allow time for IEC review and trial registration, and check that 6 months of follow-up fits your timeline.

What comes next

Once your guide agrees the question, the work usually runs in this order, with dates set by your university:

The synopsis. Your protocol in your university’s format: aim, objectives, methods and sample size. Synopsis support →
Sample size and analysis plan. Justify the number with a formula and stated assumptions, and fix the statistical tests before data collection (which statistical test to use). Statistics support →
Ethics approval. IEC clearance before you enrol the first participant.
Data collection, analysis and writing. Full thesis support →

What the NMC requires

Checked on 1 October 2026 against the NMC’s Post-Graduate Medical Education Regulations, 2023 (PGMER-2023), dated 29 December 2023:

All broad-speciality and super-speciality students do thesis-related research and write a thesis (Regulation 5.2(iii)).
Every postgraduate student completes an online course in research methodology, expected in the first year; the certificate is required to be eligible for the final examination (5.2(xi)).
A course in ethics, including Good Clinical Practice or Good Laboratory Practice as relevant, is also expected in the first year (5.2(xii)).
For MD/MS, 20 marks, 5% of the clinical/practical and viva voce marks, are for the thesis. An external examiner from outside the state evaluates it and takes a viva voce on it.

Everything else, including topic and synopsis deadlines, the synopsis format, word limits and when you submit, is set by your university. Follow your university’s dissertation guidelines and check them with your guide.

Source: PGMER-2023 (PDF), listed on the NMC’s Rules & Regulations page.

Want checked topic ideas for your department?

Three to five topics that suit your department, patient load and time, each with a PubMed summary of what is already published and a feasibility note. You and your guide choose.

See Topic ideas →

Want support with the whole thesis?

Design, literature review, statistics, drafting support, Vancouver referencing, a plagiarism check and formatting to your university’s template. You remain the author.

See thesis support →

Questions

How do I choose an MD thesis topic?

Start from a real, unanswered question in your OPD or wards, your department’s ongoing work, or a good study worth repeating in your population. Write it as a PICO or PECO question, test it against FINER, and check early that you can reach the sample size in the time you have. Then agree it with your guide and get ethics committee approval before collecting any data.

What is the FINER criteria?

FINER stands for Feasible, Interesting, Novel, Ethical and Relevant: a checklist for judging a research question. The 2022 fifth edition of Designing Clinical Research uses the shorter FINE (Feasible, Important, Novel, Ethical), folding relevance into importance.

Which study design is best for an MD thesis?

The one your question needs and your setting can deliver in time. Cross-sectional studies measure how common something is; prospective cohorts follow exposed and unexposed patients forward; case-control studies suit rare outcomes; interventional studies give the strongest evidence on treatments but need more ethics work and usually trial registration.

Can I change my thesis topic later?

It depends on your university’s rules. Speak to your guide first and follow your university’s dissertation guidelines. A new topic also needs its own ethics committee approval before you collect any data for it.

Sources

Browner WS, Newman TB, Cummings SR, Grady DG, et al. Designing Clinical Research. 5th ed. Philadelphia: Wolters Kluwer; 2022. Chapter 2 (the research question; FINE) and Chapter 5 (hypotheses).
Farrugia P, Petrisor BA, Farrokhyar F, Bhandari M. Practical tips for surgical research: research questions, hypotheses and objectives. Can J Surg. 2010;53(4):278–81. PubMed 20646403
Richardson WS, Wilson MC, Nishikawa J, Hayward RS. The well-built clinical question: a key to evidence-based decisions. ACP J Club. 1995;123(3):A12–3. doi:10.7326/ACPJC-1995-123-3-A12
American Geriatrics Society 2023 updated AGS Beers Criteria for potentially inappropriate medication use in older adults. J Am Geriatr Soc. 2023;71(7):2052–81. doi:10.1111/jgs.18372
Indian Council of Medical Research. National Ethical Guidelines for Biomedical and Health Research Involving Human Participants. New Delhi: ICMR; 2017.
National Medical Commission. Post-Graduate Medical Education Regulations, 2023. nmc.org.in (accessed 1 October 2026).