Strata Academy

PROSPERO Amendments & Protocol Deviations Explained

When to amend your registration, how to document protocol changes, and what examiners and journals expect in the final manuscript

• Material changes need a PROSPERO amendment before you implement them.
• Document every deviation in a protocol comparison table for examiners.
• Outcome switching after seeing results without amendment is misconduct territory.
• Editors compare your CRD record to the submitted manuscript line by line.

1. What counts as a protocol deviation?

A protocol deviation is any difference between what you registered on PROSPERO and what you actually did. Some deviations are justified and transparent; others — especially post-hoc outcome changes — undermine credibility.

Material deviations typically include: changing the primary outcome, widening or narrowing inclusion criteria after screening, adding databases mid-search, switching from fixed to random effects after seeing heterogeneity, or dropping risk-of-bias assessment you pre-specified.

Minor clarifications — spelling out operational definitions, adding secondary outcomes that do not replace the primary, or correcting typos — may not need a formal PROSPERO amendment but should still appear in your methods or appendix.

UK dissertation examiners increasingly download your PROSPERO PDF. Treat the registration as a contract you can explain, not a box you ticked once.

• Material — amend PROSPERO and report in manuscript
• Minor — document in methods; amendment optional
• Undocumented — red flag for selective reporting
• Post-hoc — changes driven by results without protocol update

2. When to submit a PROSPERO amendment

Submit an amendment before implementing the change when possible. If you have already screened 200 records and realise your age criterion was wrong, stop, amend, and document how many records were re-screened.

PROSPERO's amendment workflow creates a version history. Readers can see what changed and when — this is a feature, not a confession. Transparent updates are preferable to silent drift.

Common student scenarios requiring amendment: supervisor feedback narrows the population; librarian adds Embase after a pilot search; you pre-specify ROBINS-I for non-randomised studies you initially excluded; subgroup analysis becomes primary after feasibility discussion (with justification, not after peeking at results).

Do not amend retrospectively to match what you already published in a draft chapter. Align the record to honest methods, then align the manuscript.

Note: Changing your primary outcome because the pooled result 'was not significant' — without a pre-specified amendment and rationale — is outcome switching. Examiners and journals treat this seriously.

3. How to amend a PROSPERO record (practical steps)

Log in to PROSPERO, open your registration (CRD420xxxxxx), and select the amendment option. You will be prompted to identify which fields changed and provide a brief reason.

Write amendment notes in plain language: 'Primary outcome changed from HbA1c at 12 weeks to HbA1c at 24 weeks because no included trial reported 12-week data; amendment submitted before full-text screening completed.'

Save the amended PDF and note the amendment date in your project folder. Reference both the original registration date and latest amendment date in your dissertation methods.

Allow time for editorial review — same as initial registration. Do not leave amendments to the week before submission.

1. Identify which PROSPERO fields differ from current practice
2. Draft a one-paragraph rationale (supervisor sign-off helps)
3. Submit amendment via PROSPERO portal
4. Download updated PDF; update dissertation methods and PRISMA checklist
5. Add a 'differences from protocol' subsection in results or discussion

4. Reporting deviations in your manuscript

PRISMA 2020 expects transparency about protocol changes. Include a table or subsection listing: registered plan, final method, reason for change, and whether the change was amended on PROSPERO before implementation.

For each deviation, state whether it could introduce bias (e.g. narrowing inclusion after seeing study characteristics) or is unlikely to (e.g. adding a secondary outcome).

If you ran sensitivity analyses to test robustness after a deviation — such as repeating meta-analysis with the originally registered outcome — report those results explicitly.

Dissertation examiners reward honesty. A well-explained deviation with amendment date is stronger than hoping nobody checks PROSPERO.

5. Research integrity and selective reporting

Selective reporting means presenting outcomes or analyses that favour a desired conclusion while omitting others. Protocol registries exist partly to detect this. Comparing PROSPERO to ClinicalTrials.gov for included RCTs is standard in Cochrane reviews — examiners may expect similar rigour in student work.

Outcome switching is the most common integrity concern in student dissertations: registering outcome A, finding non-significance, then writing up outcome B as primary. Amendment after the fact without a blinded rationale does not fix this.

If you discover your question is infeasible mid-project, discuss with your supervisor whether to amend scope, convert to scoping review, or stop. Transparent re-scope beats silent shortcutting.

University research integrity policies apply to systematic reviews. When unsure, ask your supervisor before changing methods.

6. Common student scenarios (and what to do)

Scenario A — Too many hits after search: narrow inclusion criteria with supervisor agreement, amend PROSPERO before further screening, document how many records were re-assessed.

Scenario B — No RCTs found: if you registered RCT-only, amend to include quasi-experimental designs with ROBINS-I, or re-scope the question with a new registration.

Scenario C — Meta-analysis not possible: amend analysis plan to narrative synthesis only; do not claim pooling was 'always intended' if PROSPERO stated meta-analysis.

Scenario D — Supervisor suggests extra subgroup: if suggested before any analysis, amend with rationale; if suggested after you have seen forest plots, document as post-hoc exploratory analysis, not pre-specified primary subgroup.

Tip: Keep a project changelog from Week 1 — date, decision, PROSPERO action. It saves hours at write-up.

7. Pre-submission checklist

Before dissertation or journal submission, download the latest PROSPERO PDF and compare field-by-field to your methods section. Mismatches must be explained or corrected.

Verify PRISMA flow numbers match your final eligibility criteria — not the original draft criteria.

Confirm primary and secondary outcomes in the abstract match the amended registration.

Include PROSPERO ID and amendment history in the abstract and methods. Link to the public record.

• PROSPERO PDF matches methods section
• Differences-from-protocol table completed
• All amendments dated before implementation where feasible
• Supervisor has reviewed deviation narrative
• PRISMA checklist updated for final methods

Frequently asked questions

Do I need to amend PROSPERO for a typo fix?

Pure typos that do not change meaning can be corrected without a formal amendment, but note the fix in your dissertation appendix. If the typo affected eligibility (e.g. wrong age range), treat it as a material deviation and amend.

Can I amend PROSPERO after my dissertation is written?

You can submit a retrospective amendment, but it will be visible on the public record. Examiners prefer amendments made before or during the work, with honest reporting of when changes occurred. Do not amend to retroactively justify decisions you already made without documentation.

What if my supervisor disagrees with the registered PICO?

Discuss before screening begins. Amend PROSPERO to the agreed PICO and document the change. If screening already started, note how many records were screened under the old criteria and whether re-screening was needed.

How do journals handle protocol deviations?

Many journals require PROSPERO ID and compare registration to manuscript. Unexplained deviations trigger editorial queries or rejection. A clear 'differences from protocol' section with amendment dates usually satisfies reviewers.

Is adding a new secondary outcome a deviation?

Adding secondary outcomes is common and usually minor if the primary outcome is unchanged. Amend PROSPERO if the addition is substantial (e.g. new patient-reported measure domain). Always report all registered outcomes in results — do not silently drop ones that were non-significant.

Interactive walkthroughs and quizzes load when JavaScript is enabled — the checklist and tables above are fully readable without it.