KONSİLMost ethics applications that bounce back do not bounce for being unethical. They bounce for mundane reasons: the sample size has no justification, the consent form is written in clinical jargon, the methods don't match the stated aim — or the file simply went to the wrong committee. This guide walks through what to write, in what order and at what level of detail, to clear review on the first pass.
Scope note: Committee names and procedures vary by country and institution — IRB in the US, REC in the UK/EU, and national variants elsewhere. This guide describes the shared skeleton; always request your own institution's current template before submitting, since most boards mandate their own form.
1. Find the right committee first
| Study type | Typical committee | Extra approval |
|---|---|---|
| Drug / biologic / medical-device trial (interventional) | Clinical research ethics committee (IRB/REC) | National regulator (FDA IND, EMA/national CTA, etc.) |
| Interventional but non-drug (extra imaging, additional blood draw, randomized behavioural intervention) | Clinical research ethics committee | Usually none; ask the regulator if a device/drug component appears |
| Non-interventional: retrospective chart review, survey, scale validation, leftover samples | Non-interventional / minimal-risk board (often expedited review) | Institutional data-access permission is commonly required |
| Animal research | IACUC (US) / AWERB (UK) / national animal-ethics board (e.g. HADYEK in Türkiye) | Laboratory-animal certification for the people performing procedures |
| Social sciences & humanities with human participants | University social/behavioural sciences ethics board | Site permission when collecting data in schools or institutions |
"Does a retrospective study need ethics approval?" Yes — this is the most common misconception. Never meeting the patient does not mean never using the patient's data. Chart reviews need (usually expedited) board approval, and nearly every journal asks for the approval number at submission. "It was retrospective, so we skipped it" is a desk rejection.
2. The skeleton of the file: 17 sections
Forms differ; the content boards expect is remarkably consistent. The third column is the question a board member silently asks at each section — if your text answers it in one sentence, you're on track:
| # | Section | What the board looks for |
|---|---|---|
| 1 | Background & rationale | Why does this question matter, and why now? |
| 2 | Literature summary | What is known; the gap your study fills |
| 3 | Aim & hypothesis | One-sentence primary aim; a testable hypothesis |
| 4 | Study design | Prospective/retrospective, randomization, blinding, arms |
| 5 | Population & setting | Who, where, over which date range |
| 6 | Sample size justification | Power analysis with a SOURCED effect size (details below) |
| 7 | Inclusion / exclusion criteria | Numbered, unambiguous lists |
| 8 | Methods & data collection | Who collects which variables with which instrument |
| 9 | Primary & secondary outcomes | ONE measurable primary outcome |
| 10 | Statistical analysis plan | Which test for which comparison; missing-data strategy |
| 11 | Ethical considerations & consent process | Who obtains consent, when, how; vulnerable groups |
| 12 | Data protection | Coding/anonymization, retention period, access list (GDPR/local law) |
| 13 | Timeline | A realistic month-by-month plan |
| 14 | Budget & funding source | Costs and who covers them; even "no cost" must be stated |
| 15 | Team & role allocation | Each name with a concrete role in THIS study |
| 16 | References | Everything cited in the text |
| 17 | Appendices | Consent form, CRF/questionnaire, CVs, site permissions |
3. The section that gets files rejected: sample size
✗ "We plan to enrol 100 patients." — That is an intention, not a justification. Why 100? Why not 50?
✓ "X et al. (2024) reported a between-group difference of 4.2 points (pooled SD 8.1; Cohen's d≈0.52). Detecting this effect with 80% power at α=0.05 requires 60 per arm, 120 in total (two-sided independent-samples t-test). Allowing 10% attrition, we target 132."
The good version has four ingredients: an effect size, its source (a citation or your own pilot data — never an invention), power and alpha, and an attrition allowance. If no reliable effect size exists, say so: a descriptive aim with a precision-based (confidence-interval-width) calculation is always more credible than a fabricated Cohen's d.
4. Informed-consent checklist
- Purpose of the study — in plain language, no clinical jargon
- Exactly what participation involves and how long it takes
- Foreseeable risks and discomforts (state explicitly if none are expected)
- Potential benefits — unexaggerated; say so if there is no direct benefit
- Voluntariness and the right to withdraw at any time, without reason, without affecting care
- How data are stored, who can access them, when they are anonymized
- Contact details of the responsible investigator for questions and complaints
- Signature blocks: participant + person obtaining consent + witness/legal representative where required
5. The most common rejection / revision reasons
| Reason | Prevention |
|---|---|
| Unjustified or unsourced sample size | All four ingredients from section 3 |
| Consent form written in technical language | Have a non-clinician read it; simplify every sentence they stumble on |
| Aims and methods don't match | Every aim needs a matching method and a matching test in the analysis plan |
| No statistical expertise visible in the team | Name who will run the analysis in the role table |
| Copy-pasted data-protection boilerplate | Be concrete: which files, on which encrypted system, for how many years, accessed by whom |
| Retrospective review without date range / record count | State "Jan 2020 – Dec 2024, ~850 records" precisely |
| Missing conflict-of-interest / funding declaration | Declare "none" explicitly rather than leaving the section out |
| No permission for a copyrighted scale/questionnaire | Attach the licence or the author's permission as an appendix |
6. Animal studies: what the board additionally expects
- The 3Rs — Replacement, Reduction, Refinement: why animals are unavoidable, how numbers were minimized, how suffering is reduced
- Anaesthesia and analgesia protocol: agent, dose, monitoring
- Humane endpoints: which findings end the experiment for that animal
- Euthanasia method and the post-experiment tissue plan
- Certification of everyone performing procedures
FAQ
Does a case report need board approval?
In most institutions a single case report needs the patient's written consent for publication rather than board approval — and virtually every journal requires that consent. A case series (often 3+, institution-dependent) does trigger board review.
Can I start collecting data while approval is pending?
No — and it cannot be repaired afterwards. Boards do not grant retroactive approval, and journals compare the approval date with the data-collection dates.
Sources and standards
- WMA Declaration of Helsinki — ethical principles for medical research involving humans
- ICH-GCP — Good Clinical Practice guideline
- Institutional IRB/REC guidance documents; national regulator guidance for trials
- The 3Rs framework and institutional animal-care (IACUC/AWERB) guidelines
- GDPR and national data-protection law provisions on health data
You don't have to write this file from scratch
After stress-testing your research idea against the literature, KONSİL drafts all 17 sections of this file for your study: it builds the power analysis with sourced parameters, explicitly marks every field it cannot honestly fill as "to be completed", routes animal studies to the animal-ethics track, and exports the file to Word.
See the ethics file in the sample report → Apply for the closed beta →