On this page
Most physicians don't fail at research because they lack the intelligence for it — they fail because no one ever showed them the whole map. A resident starts collecting data before the protocol is finalized. An MD candidate writes the manuscript before deciding which journal it's going to. A PhD candidate spends eight months on a systematic review only to find, at the ethics committee stage, that the study design needs to change.
Each of these is a sequencing problem, not a competence problem. Clinical training teaches diagnosis and treatment extremely well; it rarely teaches research as a project with a defined order of operations. This article lays out that order — the eleven stages every clinical research project moves through, what typically goes wrong at each one, and what a realistic timeline actually looks like.
Why This Happens
Three structural reasons explain why the roadmap is rarely taught explicitly:
Research is squeezed into leftover time
Unlike a funded PhD track, most physician-led research happens in evenings, on-call gaps, and weekends. Without a written sequence to follow, it's natural to work on whatever feels most urgent that week — usually data collection — rather than what the project actually needs next.
Supervision is often implicit
Many supervisors guide by reacting to drafts rather than laying out the stages up front. That works for someone who has already internalized the sequence from a previous project, and works poorly for someone doing it for the first time.
Each stage looks similar but isn't
"Writing the proposal" and "writing the manuscript" both involve writing, but they happen at opposite ends of the project and answer different questions. Conflating them is one of the most common causes of wasted work.
Common Mistakes We See
- Starting data collection before the protocol is finalized. Any change to inclusion criteria, variables, or outcomes after collection has begun usually means partial or full re-collection.
- Skipping the sample size calculation. An underpowered study is one of the most common reasons for rejection at peer review, and it can't be fixed after data collection ends.
- Treating ethics approval as a formality. IRB/ethics committee review can take 4–12 weeks and often requires revisions. Starting this process late is the single most common cause of project delay.
- Deferring the choice of statistical test. The right test depends on the study design decided at the protocol stage — not on what looks reasonable once the data is already in a spreadsheet.
- Writing the manuscript before selecting a target journal. Word count, structure, and reporting checklist (e.g. CONSORT, STROBE) vary by journal and are far easier to follow from the first draft than to retrofit later.
Practical Recommendations
The table below is a working reference, not a rigid formula — timelines vary by study design and institution. What matters most is respecting the order.
| Stage | What it produces | Typical duration |
|---|---|---|
| 1. Research idea | A focused, answerable research question (PICO format for clinical questions) | 1–3 weeks |
| 2. Proposal | Background, objectives, significance — the case for why this study matters | 2–4 weeks |
| 3. Protocol | Design, population, variables, sample size, statistical plan | 3–6 weeks |
| 4. Ethics approval | IRB/ethics committee clearance and informed consent documents | 4–12 weeks |
| 5. Data collection | Completed, cleaned dataset (CRF or REDCap-based) | Varies widely by design |
| 6. Statistical analysis | Analyzed results matching the pre-specified plan | 2–4 weeks |
| 7. Scientific writing | Full manuscript draft (IMRAD structure) | 4–8 weeks |
| 8. Journal formatting | Manuscript formatted to the target journal's guidelines | 1–2 weeks |
| 9. Submission | Submitted manuscript with cover letter | 1 week |
| 10. Revision | Response to reviewer comments and revised manuscript | 2–8 weeks per round |
| 11. Publication | Accepted, published paper | Varies by journal |
Quick Checklist
- Research question is written in a single, specific sentence (PICO format)
- Protocol is finalized and approved before any data is collected
- Sample size is calculated, not estimated
- Ethics application is submitted as soon as the protocol is ready
- Statistical plan is agreed upon before data collection begins
- Target journal is shortlisted before the manuscript is drafted
- Reporting checklist (CONSORT, STROBE, PRISMA) matches your study design
- Timeline includes buffer time for at least one round of revisions
Need Expert Guidance?
Book a free call with PureMed and let our research specialists help you save valuable time, avoid costly research mistakes, and move confidently toward publication.
Frequently Asked Questions
How long does the full process take, from idea to publication?
For a typical single-center study, 12–18 months is a realistic range from a finalized protocol to an accepted paper. Systematic reviews and multi-center studies often take longer, mainly due to ethics approval and data collection timelines.
Can I join the process partway through, for example with data already collected?
Yes. It's common for physicians to start research support after data collection has already begun. The roadmap still applies — the remaining stages simply start from wherever the project currently stands.
Do I need a statistician involved from the beginning?
Ideally yes, even briefly. Sample size and the analysis plan are decided at the protocol stage, and both depend on statistical input. Bringing in a statistician only at the analysis stage is one of the most common causes of avoidable rework.
What's the single biggest reason research projects stall?
Ethics approval delays, closely followed by starting data collection before the protocol is genuinely finalized. Both are sequencing issues, and both are avoidable with the order above.
Should I choose a journal early or wait until the manuscript is finished?
Choose a shortlist early — ideally before writing the first draft. Journal requirements shape structure, word count, and the reporting checklist you should be following, and it's far more efficient to write to those requirements from the start.