Most doctors who want to publish start in the wrong place. They start with a title, or with a journal, or with a deadline. An original research article actually starts about a year earlier, with a question you can answer using patients you will genuinely see.
This guide walks through the whole path: framing the question, writing a protocol, getting ethics committee approval, collecting patient data lawfully, analysing it, and turning the result into a manuscript a journal will take seriously.
The short version: question, protocol, ethics approval, registration if applicable, data collection, analysis, then writing. Every step before writing determines whether the writing is publishable at all. Skipping any of them cannot be fixed later.
Step 1: Turn a Clinical Irritation Into a Research Question
The best research questions come from something that already bothers you in clinic. A test you order that rarely changes management. A complication you suspect is more common in your population than the textbooks suggest. A protocol your department follows for reasons nobody can explain.
Two tools help you sharpen that irritation into a question.
PICO forces you to name the components: Population, Intervention or exposure, Comparison, and Outcome. "Do diabetic patients heal worse?" is not answerable. "Among adults undergoing elective inguinal hernia repair, is HbA1c above 7.5 percent associated with a higher rate of surgical site infection at 30 days compared with HbA1c below 7.5 percent?" is answerable.
FINER tests whether the question is worth asking: is it Feasible, Interesting, Novel, Ethical and Relevant. Feasibility is where most Indian postgraduate projects fail. If your unit sees eight cases of the condition a year, a study needing 200 patients is not a study, it is a fantasy.
Check What Already Exists
Search PubMed, Scopus and Google Scholar before you commit. You are looking for two things: whether your question has already been answered convincingly, and what methods previous authors used. If ten studies exist from Europe and none from India, that gap is itself a legitimate justification.
Step 2: Choose a Study Design You Can Actually Execute
The design follows from the question, not from what sounds impressive.
- Cross sectional study: measures exposure and outcome at one point in time. Good for prevalence. Cannot establish that the exposure came first.
- Case control study: starts with people who have the outcome and looks backwards. Efficient for rare outcomes. Vulnerable to recall and selection bias.
- Cohort study: follows people with and without an exposure forward in time. Stronger evidence, but needs follow up that your timeline may not allow.
- Randomised controlled trial: the strongest design for an intervention, and by far the heaviest in regulation, cost and time. In India a trial of a drug or device brings the New Drugs and Clinical Trials Rules into play and requires prospective registration.
Be realistic about follow up. A postgraduate with 18 months left cannot run a study requiring two year outcomes. Choose a design whose endpoint arrives before your deadline does, or choose a retrospective design using records that already exist.
Step 3: Write the Protocol Before Anything Else
The protocol is the document your ethics committee reviews, and it is also the thing that saves you later when a reviewer asks why you did something. Write it properly once and you will reuse most of it in your methods section.
A workable protocol contains:
- Title and investigators with roles clearly assigned
- Background and the specific gap you are addressing
- Objectives, with one clearly labelled primary objective
- Study design, setting and duration
- Inclusion and exclusion criteria, stated so precisely that a colleague could apply them without asking you
- Sample size calculation with the assumptions and the source of those assumptions
- Sampling method, and randomisation and blinding if applicable
- Outcome variables and exactly how each will be measured
- Data collection tools, including your case record form
- Statistical analysis plan, decided in advance
- Ethical considerations, consent process and data protection measures
- Participant information sheet and consent form, in English and in the local language
The Statistical Analysis Plan Comes First, Not Last
Deciding your analysis after seeing the data is how honest researchers accidentally produce misleading results. Write down, before collection begins, which test you will use for the primary outcome, how you will handle missing data, and what you will do about confounders. If you are unsure, this is the moment to involve a statistician, not after the data is locked.
Step 4: Ethics Committee Approval
In India, research involving human participants or their identifiable data falls under the ICMR National Ethical Guidelines for Biomedical and Health Research Involving Human Participants. Review is carried out by an Institutional Ethics Committee. Committees overseeing regulated clinical trials must additionally be registered with the national regulator.
What the Committee Actually Wants
- A complete protocol, not a summary
- Participant information sheet in language a patient can understand, without jargon
- Informed consent form, translated, with a place for a witness where relevant
- Case record form and any questionnaire you plan to use
- A clear statement of risks, benefits and how confidentiality will be protected
- Investigator CVs and, in many institutions, a conflict of interest declaration
- Details of funding, if any
Three Review Categories
Exemption may apply to research on anonymised data or publicly available records, where participants cannot be identified. Expedited review applies to minimal risk research and to minor amendments. Full committee review applies to everything else, particularly studies involving interventions, vulnerable participants or identifiable data.
You do not decide which category applies. You apply, and the committee decides. Even when you are confident a project is exempt, obtain that determination in writing.
Do not start before the letter arrives. Prospective data collection that begins before written approval is a protocol violation. Journals ask for the approval reference number, and some verify it. There is no way to obtain approval retrospectively for data already collected.
Common Reasons Applications Get Sent Back
- Consent form written at a reading level no patient could follow
- No local language translation
- Sample size stated without any calculation
- Vague plan for storing and protecting identifiable data
- No description of what happens if a participant withdraws
- Inclusion criteria that quietly contradict the exclusion criteria
Step 5: Register the Study Where Required
Clinical trials must be registered prospectively, before enrolling the first participant, in a recognised public registry. In India that is the Clinical Trials Registry India. Journals following the recommendations of the International Committee of Medical Journal Editors will generally refuse to consider an unregistered trial, and registering after enrolment has begun does not repair the problem.
Observational studies are usually not required to register, though some journals view prospective registration favourably. Postgraduate theses have their own registration requirements set by the university or by the National Board of Examinations, which are separate from trial registration.
Step 6: Collect Patient Data Properly
Informed Consent Is a Process, Not a Signature
Consent means the participant understood what you asked. Explain the purpose, the procedures, the risks, the alternatives, the voluntary nature of participation and the right to withdraw at any time without any effect on their clinical care. Do this in the language they speak. Give them time to ask questions. Then take the signature, and give them a copy.
For participants who cannot consent for themselves, follow the specific provisions your committee has approved for legally acceptable representatives, and take assent from older children in addition to parental consent.
Protect Identity From the First Entry
Assign each participant a study identification number and use it everywhere. Keep the single file linking study numbers to names and hospital numbers separately, encrypted, and accessible only to named investigators. Your working dataset should contain no names, no hospital registration numbers, no phone numbers and no full dates of birth.
This is not only good practice. India's Digital Personal Data Protection Act treats health information as personal data, and your institution will have its own record handling policy on top of that. De-identifying at the point of entry is far easier than trying to clean a spreadsheet full of names later.
Build a Case Record Form Worth Using
- One variable per column, one participant per row
- Pre-coded options rather than free text wherever possible, because free text cannot be analysed
- Explicit codes for "not done" and "not applicable", which are different from missing
- Consistent units, decided once and written at the top of the column
- A data dictionary explaining every variable and every code
Enter data as you go. The single most common disaster in postgraduate research is a shoebox of paper forms transcribed in a panic three weeks before the deadline. Enter weekly, run a simple frequency check monthly, and you will catch systematic errors while they can still be corrected.
Step 7: Analyse With the Plan You Wrote
Begin with clean descriptive statistics: how many participants, how many excluded and why, baseline characteristics of the groups. A participant flow diagram accounting for everyone from screening to analysis is expected in most designs and is mandatory for trials.
Then run the analysis you specified in the protocol. Report effect sizes with confidence intervals, not only p values. A p value tells you how surprising the result would be if nothing were happening, and nothing more. A confidence interval tells the reader the range of effects compatible with your data, which is what they actually want to know.
If you run analyses you did not plan, label them as exploratory. That is honest and reviewers respect it. Presenting a post hoc finding as though it were the primary hypothesis is the definition of a misleading paper.
Step 8: Write the Manuscript in IMRaD Order
Almost every original research article follows IMRaD: Introduction, Methods, Results and Discussion. Write them out of order. Methods and Results are the easiest because they describe what you did and what you found.
Methods
Written so that a competent stranger could repeat your study. State the design, setting, dates, eligibility criteria, sample size calculation, how variables were measured, the software used and the statistical tests. Include the ethics approval statement with the committee name and reference number, and the consent statement.
Results
Findings only, with no interpretation. Start with participant flow and baseline characteristics, then the primary outcome, then secondary outcomes. Do not repeat in text what a table already shows. Every table and figure should be understandable without reading the surrounding text.
Introduction
Three movements: what is known, what is missing, and what your study set out to do. Around three to five paragraphs. The final sentence should state your objective in the same terms your results will use.
Discussion
Open with your main finding in plain language. Compare it with previous studies and explain agreements and disagreements. Then state your limitations honestly, because reviewers will find them anyway and a paper that names its own weaknesses reads as competent rather than weak. Close with what your findings mean for practice or for the next study.
Abstract, Title and Keywords
Write the abstract last, once the paper is stable. Follow the structure the journal specifies and respect the word limit exactly. The title should contain the population, the exposure or intervention and the outcome, because that is what makes it findable.
Step 9: Follow the Right Reporting Guideline
Editors expect you to use the reporting checklist matching your design, and many require the completed checklist at submission.
- STROBE for observational studies, including cohort, case control and cross sectional
- CONSORT for randomised controlled trials
- PRISMA for systematic reviews and meta analyses
- CARE for case reports
- STARD for diagnostic accuracy studies
Fill the checklist while writing rather than afterwards. It functions as a free pre review and catches the omissions that generate the most reviewer comments.
Step 10: Choose the Journal Before You Format
Shortlist three journals that have published work similar to yours, verify their indexing in the index own database rather than trusting the journal website, and read their instructions to authors line by line. Formatting for a journal you have not chosen wastes a full day.
For Indian faculty, confirm that the journal is indexed in a database recognised for promotion before you submit. A paper in an unindexed journal costs you the same effort and returns nothing on your academic record. Our guides on choosing the right journal and avoiding predatory journals cover the verification process in detail.
A Realistic Timeline
- Months 1 to 2: question, literature search, protocol drafting
- Months 2 to 4: ethics committee submission, responses, approval
- Months 4 to 12: data collection, with weekly entry
- Month 13: data cleaning and analysis
- Months 14 to 15: manuscript drafting and co author review
- Month 16: journal selection, formatting, submission
- Months 16 to 22: peer review, revision, decision
Peer review timelines are outside your control and vary from six weeks to six months. Build that into your expectations, particularly if a promotion deadline is involved.
Frequently Asked Questions
Do I need ethics committee approval for a retrospective chart review?
In most cases yes, although many institutional ethics committees grant an expedited review or a formal waiver of consent for anonymised retrospective data. The important point is that the decision belongs to the committee, not to you. Apply, describe the study honestly, and keep the letter you receive, whether it grants approval, exemption or a waiver. Journals increasingly ask for the letter number, and a study that was never presented to a committee cannot be retrofitted with approval afterwards.
Can I start collecting data before ethics approval comes through?
No. Prospective data collection must begin only after written approval. Collecting first and applying later is a protocol violation that many journals treat as grounds for rejection, and it cannot be corrected retrospectively. If you are short of time, ask your committee about its expedited review track rather than starting early.
How many patients do I need for an original research article?
There is no universal number. The sample size follows from your primary outcome, the expected effect size, the acceptable error rates and the variability reported in previous studies. A cross sectional study estimating a prevalence needs a different calculation from a study comparing two groups. Calculate it before you start, state the calculation in the methods, and report what you actually achieved.
Do I have to register my study before publication?
Clinical trials must be registered prospectively, before the first participant is enrolled, in a recognised registry such as the Clinical Trials Registry India. Most journals following ICMJE recommendations will refuse to publish an unregistered trial. Observational studies are not usually required to register, although some journals encourage it.
Planning Your First Original Research Article?
We help with study design, sample size, statistical analysis and full manuscript development. You stay the author throughout.
Talk to Us →This article is general educational guidance, not legal advice. Journal policies, regulatory rules and institutional requirements change. Always check the current instructions to authors of your target journal and the current guidance of your institution, the NMC and the ICMR before acting.