Contents
On this page
- Distinction between an audit and a checklist
- Measured levels of reporting adherence
- Limitations of aggregate adherence scores
- Direction of change in individual reporting items
- The item that declined and its implications
- Design of the audit
- Separate audit of the abstract
- Declarations not scored by reporting checklists
- Limitations of audit scores
- Key points
- Application to your study
- Frequently asked questions
Academy · Research Publication · auditing GPP reporting
Auditing a Publication Against the Reporting Standards Required by Good Publication Practice
GPP expects authors to follow the reporting guideline for their study design but does not specify how adherence should be checked. A reporting audit provides that check, and it must be conducted item by item rather than as a score, because the measured evidence shows that aggregate adherence conceals item-level changes in both directions. In 463 randomised controlled trial articles from five high-impact journals, mean adherence to CONSORT for Abstracts was 64.3% where professional writing support was declared and 66.5% where it was not, a difference that was not significant at p=0.30, while individual items differed by factors of two or more in either direction.
Most organisations that have adopted GPP check reporting by asking whether a checklist was completed. This provides very little information, and there is evidence for that claim. When adherence to CONSORT for Abstracts was measured across 463 randomised controlled trial articles from five high-impact journals, the mean was 64.3% with declared writing support and 66.5% without, essentially identical (p=0.30); beneath that flat aggregate, however, reporting of harms and funding source was approximately twice as likely in supported papers, and reporting of the study setting was substantially less likely. The aggregate score concealed the informative differences, and an audit that produces a single number per paper would do the same.
Learning objectives
- Why an item-level audit and an adherence score answer different questions
- What the measured adherence data show, including where they show nothing
- Which reporting items move most, and in which direction, when writing support is involved
- Design of the audit, including unit of analysis, scoring, double-scoring and handling of disagreement
- Why the abstract must be audited as a separate object
- How to audit the GPP declarations that no reporting checklist scores at all
Common misconceptions
- “A completed checklist means the reporting is adequate.” A checklist records that a review took place; an audit records, item by item, what was found and where in the manuscript. These produce different documents, and only the second is useful.
- “An overall adherence percentage tells you how good the reporting is.” An overall percentage discards the information that is required. Across 463 articles, mean adherence to CONSORT for Abstracts was 64.3% with writing support and 66.5% without (p=0.30, no detectable difference), while specific items ranged from approximately twice as likely to less than half as likely to be reported in the same comparison.
- “Professional writing support improves reporting across the board.” The evidence does not support this, and an internal claim to that effect would be contradicted by a single audit. Harms reporting was more likely with support (RR 2.04, 95% CI 1.37 to 3.03), as was reporting of funding source (RR 1.75, 95% CI 1.18 to 2.60), but study setting was reported considerably less often (RR 0.40).
- “These are observational findings, so there is nothing to act on.” The confounding is real, but the item-level pattern remains actionable because it identifies where to look. An item that moves in the unfavourable direction when a writer is involved should be checked explicitly in the publication process.
- “The abstract is covered by auditing the paper.” It is not. The abstract has its own item set, a large proportion of readers never read beyond it, and it is the part of the publication that is carried into congress materials and summaries where the full paper does not follow.
Distinction between an audit and a checklist
GPP expects publications to follow the reporting guideline appropriate to the study design, and leaves the mechanism to the author 5,9. In most organisations the mechanism is a completed checklist filed before submission. That is a record that the exercise happened; it is not a finding.
The difference is concrete. A completed checklist is a column of ticks against item numbers. An audit is a judgement per item, with three things attached to each: whether the item is addressed, where in the manuscript it is addressed, and if it is not, what is missing. The first can be produced without reading the manuscript carefully. The second cannot.
Two consequences follow for how the work is organised. An audit must be conducted by someone who did not write the paper, because the author already believes the items are covered, and that belief is precisely what the audit tests. The audit output must also be the gap list rather than the score, because the gap list can be acted on before submission and the score cannot.
This page describes how to conduct that audit and what the published measurements of adherence indicate about where to direct it. Our companion page on what GPP requires you to report and declare sets out the two layers of content; this page concerns checking them.
Measured levels of reporting adherence
A published measurement anchors the exercise. Across 463 randomised controlled trial articles from five high-impact journals, mean adherence to CONSORT for Abstracts was 64.3% where professional medical writing support was declared and 66.5% where it was not, with p=0.30 for the difference 7,18.
Three points follow from this. The first is the absolute level. Roughly two-thirds of the abstract items were addressed, in articles published in high-impact journals, which are the articles most likely to have been through the most editorial scrutiny 7,11. If that is the ceiling in those journals, an organisation assuming that its own abstracts are near-complete is probably mistaken, and the audit is worth conducting.
The second is that the two groups were not distinguishable. A difference of 2.2 percentage points at p=0.30 is a null result, and the appropriate reading is that declared writing support was not associated with better overall abstract adherence in this sample 7. Any internal case for writing support on the grounds of improved reporting compliance should acknowledge that the aggregate evidence does not support it.
The third point is the most important operationally: the null aggregate coexisted with substantial item-level differences in both directions. That combination is possible only if the item-level movements partly cancel, which is what the next two sections describe. It is also a strong argument against reporting an audit as a single percentage.
The aggregate result, which is null. Mean adherence to CONSORT for Abstracts measured across 463 randomised controlled trial articles from five high-impact journals, as reported in the analysis of declared medical writing support cited for this article. The item-level differences underneath this flat aggregate run in both directions.
Limitations of aggregate adherence scores
Scoring a paper as a percentage of items addressed has an obvious appeal: it is one number, it can be tracked over time, and it resembles a metric. It has three defects that make it counterproductive for improving reporting.
Items are not equivalent and the score treats them as such. Failing to report the primary outcome and failing to report the trial registration number both cost one item. One of those omissions makes the trial uninterpretable and the other makes it unfindable, and averaging them produces a number that corresponds to nothing.
Movements in opposite directions cancel. This concern is not hypothetical. The 463-article comparison produced a flat aggregate — 64.3% against 66.5%, p=0.30 — while harms reporting was about twice as likely with writing support and study setting reporting was substantially less likely 7. An organisation tracking only the mean would have concluded that nothing was happening, while two items were moving sharply.
A score cannot be acted on. A score of 68% gives an author no indication of what to change. Informed that the primary outcome is specified in the methods but not in the abstract, that the trial registration appears only in the funding footnote, and that the setting is described as “multiple centres” with no country or care level, an author can correct all three in twenty minutes.
The audit should therefore be reported as a gap list with locations. If a tracked metric is required for governance purposes, the frequency of specific gaps across papers should be tracked rather than the mean score of each paper, which at least indicates which part of the process is failing.
Direction of change in individual reporting items
The item-level data indicate where the audit should be directed. The items most improved where professional medical writing support was declared were specification of the primary outcome (RR 1.77), the participant flow diagram (RR 1.96), recruitment and follow-up dates (RR 2.04) and trial registration (RR 1.64) 7,8. On the separate abstract analysis, harms reporting was better with support (RR 2.04, 95% CI 1.37 to 3.03) and so was funding source (RR 1.75, 95% CI 1.18 to 2.60) 7,18.
These items have features in common. Primary outcome specification, flow diagram, recruitment dates, registration and funding source are all items with a definite, locatable answer. Either the primary outcome is named or it is not; either the dates are given or they are not. They are the items that a trained person working through a checklist reliably identifies, because identifying them requires attention rather than judgement.
That is the useful generalisation, and it directs the audit to the complement. The items a checklist pass does not reliably catch are the ones requiring judgement about adequacy rather than presence: whether the stated limitations are the real ones, whether the conclusion follows from the primary outcome rather than from a secondary one, and whether the setting description is sufficient for a reader to judge generalisability. An audit designed to add value beyond a checklist should concentrate there.
One caution applies. These are associations in observational data, and the comparison is between papers that declared writing support and papers that did not — groups that differ in sponsorship, trial size and much else 7,8. The pattern identifies where to look; it does not establish that writing support caused the differences.
Item-level audit targets, and how each item behaved in the measured data
| Item | What a complete answer contains | Measured association with declared writing support | Audit priority |
|---|---|---|---|
| Primary outcome specified | The outcome named, with its measurement and timepoint, and the same one as in the protocol | RR 1.77 — more likely to be reported | Medium — reliably caught by a checklist pass; check it matches the protocol |
| Participant flow diagram | Numbers reconciling at every stage, with reasons for loss | RR 1.96 — more likely to be reported | Medium — check the arithmetic, not just the presence |
| Recruitment and follow-up dates | Start and end of recruitment, and end of follow-up | RR 2.04 — more likely to be reported | Low — definite answer, easily caught |
| Trial registration | Registry name and number, in the paper and the abstract | RR 1.64 — more likely to be reported | Low — but check it appears in the abstract, not only the footnote |
| Harms | Harms reported in the abstract as well as the paper, including when none were observed | RR 2.04 (95% CI 1.37 to 3.03) — more likely to be reported | High — frequently absent from abstracts altogether |
| Funding source | The funder named, distinct from the statement of the sponsor’s role | RR 1.75 (95% CI 1.18 to 2.60) — more likely to be reported | Low for presence, high for sufficiency — funding named is not role described |
| Study setting | Type and number of centres, countries, level of care, and the period | RR 0.40 — less likely to be reported | Highest — the item that moved backwards, and the one least amenable to a template |
| Limitations | The limitations specific to this trial, not the ones common to all trials | Not reported in this analysis | High — requires judgement of adequacy, which a checklist pass does not supply |
| Conclusion | A statement of what the primary outcome showed, not a recommendation about the intervention | Not reported in this analysis | High — the commonest place a secondary result is promoted |
Relative risks are associations from observational analyses of randomised controlled trial articles comparing those that declared professional medical writing support with those that did not; the groups differ in sponsorship, size and therapeutic area, so these identify where reporting differs rather than establishing cause. Confidence intervals are given where the source reports them. The final two rows carry no measured association and are included because an audit that only checks the templated items will miss them. Audit priority is a practical judgement for authors, not a published ranking.
The item that declined and its implications
One finding in this data is more instructive than the rest: writing support was associated with worse reporting of the study setting, at RR 0.40 — less than half as likely 7.
This finding merits careful consideration rather than dismissal, because the plausible explanation is unflattering and useful. Setting is the item least amenable to a template. Primary outcome, registration number, flow diagram and funding source all have a standard slot and standard wording; a writer working to a house structure fills them because the structure has a place for them. An adequate setting description has no standard form: it requires knowing which features of the care context a reader would need in order to judge whether the result transfers, and that is specific to the trial and usually known only to the investigators. A drafting process optimised for completeness on templated items can compress exactly this kind of content, because it reads as background.
Whether or not that is the mechanism, the operational lesson holds. An audit should treat the items that resist templating as high-risk items, and setting is the clearest example. A setting statement that says “conducted at multiple centres in Europe” is syntactically a setting statement and conveys nothing a reader can use. The before-and-after example below presents that sentence and its revision.
The same logic applies to two further items that merit attention in any audit: the limitations paragraph, which templates into a generic list of the limitations all trials have, and the conclusion, which templates into a claim about the intervention rather than a statement of what the primary outcome showed. Our page on the GPP disclosure topics most often missed covers the declaration side of the same problem.
The study setting statement — the item that went backwards
Syntactically a setting statement, conveys nothing usable
The study was conducted at multiple centres in Europe between 2019 and 2021. Patients were recruited from participating sites according to the inclusion and exclusion criteria.
“Multiple centres in Europe” names neither a number nor a country. Between two and two hundred sites, in any of some forty health systems: the statement excludes nothing and therefore informs the reader of nothing.
No level of care. University hospital, regional centre and primary care are different populations with different baseline management, and the result may transfer to one and not the others.
“Between 2019 and 2021” is not a recruitment period. It conflates recruitment with follow-up and gives neither an opening nor a closing date, which matters wherever standard care changed during the window.
“According to the inclusion and exclusion criteria” is a tautology. It describes no setting feature at all and occupies the space where the setting description should be.
Nothing about capability or volume. If the intervention needs facilities or expertise that participating sites happened to have, that is the central generalisability fact and it is absent.
Lets a reader judge whether the result transfers
The trial was conducted at 14 sites in four countries — Germany, Spain, Poland and the Netherlands — comprising 11 university hospital departments and 3 regional secondary-care centres, with no primary-care recruitment. Recruitment ran from 3 February 2019 to 18 November 2020 and follow-up ended on 30 November 2021. All participating centres had on-site access to the index investigation and to the specialist team required to deliver the intervention, which is not universally available at secondary-care level in these health systems; readers applying these results in settings without that access should note that the control pathway in this trial reflected specialist rather than general management. Median annual case volume at participating sites was 112 patients (range 41 to 390).
Number of sites, the countries and the mix of care levels are all stated, so a reader can locate the trial in a health system they recognise.
Recruitment and follow-up dates are separated and given precisely, which also satisfies the recruitment-dates item that a checklist pass catches anyway.
The capability requirement is stated explicitly, together with what it implies for settings that lack it — which is the generalisability judgement the item exists to enable.
The control pathway is characterised as specialist rather than general management, so a reader comparing against their own usual care knows what the comparator was.
Case volume with a range gives a concrete measure of site experience, which is the kind of detail a reader uses to decide whether the result applies to a lower-volume centre.
Study setting was the one item reported substantially less often where writing support was declared, at RR 0.40, and the plausible reason is that it is the item with no template — so it is the item an audit has to check by reading rather than by looking for a slot.
Design of the audit
Six decisions, each of which changes what the audit can detect.
The unit of analysis is the item, not the paper. One judgement per item per document, recorded separately. If the output has one row per paper, the result is a scoring system rather than an audit.
The scoring categories need a middle. Addressed, partially addressed, not addressed, because “partially” is where most real problems lie, and a binary scheme forces them into one of two incorrect categories. A partial judgement requires a note stating what is missing.
Every item gets a location. Section and paragraph, or page and line. An item marked addressed with no location has not been checked, and the location requirement is what makes the audit auditable in turn.
The auditor is not the author. This requirement is not negotiable. The author believes the items are covered, and that belief is what is being tested.
Double-score a sample. A proportion of items scored independently by a second auditor, with disagreements recorded rather than quietly reconciled, indicates whether the audit is reproducible. Disagreement concentrated on particular items indicates that the item definitions are loose, which can be corrected.
The output is a gap list with required actions. Each not-addressed or partially-addressed item becomes a line stating what is missing, who will correct it and by when. That document is returned to the author; the score, if one is calculated at all, goes to governance.
On timing, the audit should be conducted while the paper can still change. An audit at submission is a record; an audit at the point where the author can still obtain the missing information from the study team is an intervention 12. Our reporting checklist review service and publication-stage checklists cover where this sits in the workflow.
Separate audit of the abstract
The abstract needs a separate pass against its own item set, and the measured data is the argument: the 64.3% and 66.5% figures are abstract adherence, in high-impact journals, which means roughly a third of the required abstract items were missing in the articles least likely to have that problem 7.
There are three reasons the abstract warrants a separate pass. A large share of readers never read beyond it, so an omission there is an omission for most of the audience. Its item set is genuinely different: the abstract must carry the primary outcome, the effect estimate with its precision, the trial registration and harms, in a word count that invites all four to be cut 18. And the abstract is the component that travels: into congress submissions, into plain-language summaries, into internal materials, in each case detached from the paper that would otherwise supply the missing context.
The abstract should be audited for four things specifically. The first is whether the primary outcome named in the abstract is the primary outcome in the protocol, which is where selective emphasis enters. The second is whether the effect estimate appears with a measure of precision rather than as a bare point estimate or a p-value. The third is whether harms appear at all, given that harms reporting was one of the items that moved most in the measured data 7. The fourth is whether the conclusion is a statement about the primary outcome result rather than a recommendation about the intervention.
Our abstract writing support and congress content work cover the downstream objects, and plain language summaries cover the lay version, which inherits whatever the abstract got wrong.
Declarations not scored by reporting checklists
A reporting guideline covers the study. GPP also expects a set of declarations about how the publication came about, and no reporting checklist scores most of them — so an audit limited to the reporting guideline will pass a paper with every one of them missing or wrong 5,9.
These declarations should be audited separately, as a second pass with its own short item list: authorship criteria met and documented for every named author; contributions stated; everyone who contributed but did not qualify for authorship acknowledged, including professional writers, with the funder of that support named; the role of the sponsor stated in substance rather than as a funding acknowledgement; author access to the data described accurately; prior disclosure of the same data at congress or as a preprint declared; and a data availability statement that says something operable 1,5,21.
Two of these merit particular attention because they fail without obvious signs. The sponsor role statement tends to default to a formula that names the funding and omits the involvement: who designed the study, who held the data, who analysed it and who had the final decision on submission. The data availability statement is frequently a sentence that cannot be acted on; a reader who cannot tell from it whom to ask and what would be provided has not been given a statement. Our pages on what GPP3 requires, ICMJE guidance on authorship and publication governance cover this layer, and data management planning is where an operable availability statement comes from.
Limitations of audit scores
Four honest limits.
The adherence evidence is observational and confounded. The comparison across 463 articles is between papers that declared professional writing support and papers that did not, and those groups differ in sponsorship, funding, trial size and therapeutic area 7,8. The relative risks identify where reporting differs; they do not establish that writing support caused the differences, in either direction.
The aggregate result is null and should be reported as such. Mean adherence of 64.3% against 66.5% at p=0.30 means no detectable difference in overall abstract adherence. An internal business case for writing support that rests on improved reporting compliance is not supported by this evidence, and the item-level findings are not a substitute for the aggregate one.
Complete reporting is not good conduct. An audit establishes that the manuscript says what happened; it says nothing about whether what happened was adequate. A fully reported trial with an inappropriate comparator is fully reported. These require different remedies, and an audit that blurs them produces false reassurance — the same distinction our page on PRISMA as a reporting rather than conduct instrument sets out for reviews.
Moreover, the audit does not generalise beyond the design it was run against. CONSORT items apply to randomised trials; an observational study audited against them will score badly for reasons that are not defects 14,20. The guideline should be matched to the design — CONSORT for trials, STROBE for observational studies, PRISMA 2020 for reviews, CARE for case reports, SPIRIT for protocols — and the guideline used should be recorded on the audit form, because an audit that does not name its instrument cannot be interpreted later.
Key points
Key points
GPP expects authors to follow the reporting guideline for their design and leaves the checking to them, and that checking must be item-level rather than a score. The evidence for this is direct: across 463 randomised controlled trial articles from five high-impact journals, mean adherence to CONSORT for Abstracts was 64.3% where professional writing support was declared and 66.5% where it was not, a null difference at p=0.30, while beneath that figure specification of the primary outcome (RR 1.77), the participant flow diagram (RR 1.96), recruitment and follow-up dates (RR 2.04), trial registration (RR 1.64), harms (RR 2.04, 95% CI 1.37 to 3.03) and funding source (RR 1.75, 95% CI 1.18 to 2.60) were all more likely to be reported, and study setting was substantially less likely at RR 0.40. The items that improve are those with a defined place in the manuscript; the item that declined is the one without a template. The audit should therefore focus on the items requiring judgement (setting, limitations and conclusion), score each item separately with a location and an intermediate category, be conducted by someone other than the author while the paper can still be changed, treat the abstract as a separate object, and include a second pass over the GPP declarations that no reporting checklist scores.
Application to your study
Application to your study
- Items, rather than papers, should be scored. Each item should receive one judgement with a location in the manuscript, and the output should be a gap list rather than a percentage.
- Three categories should be used rather than two. The categories should be addressed, partially addressed and not addressed, with a note on every partial rating specifying exactly what is missing.
- The audit should be conducted by someone other than the author. The author believes that the items are covered, and it is this belief that the audit tests.
- The audit should focus on items without a template. Setting, limitations and the conclusion are the items for which a checklist pass adds least, and setting is the item that the measured data show declining.
- The abstract should be audited separately against its own item set. The audit should confirm a primary outcome matching the protocol, an effect estimate with a measure of precision, the presence of harms, and a conclusion about the result rather than the intervention.
- A second pass should be conducted over the GPP declarations. This pass should cover authorship, contributions, acknowledgement of non-author contributors and of who funded them, the substantive role of the sponsor, data access, prior disclosure and an operable availability statement.
- A sample should be double-scored and the disagreements retained. Disagreement clustered on particular items indicates loosely defined items, which can be corrected.
- The guideline should be named on the audit form. CONSORT, STROBE, PRISMA, CARE or SPIRIT should be recorded according to design; an audit that does not record its instrument cannot be interpreted later.
Frequently asked questions
A completed checklist records that somebody looked; an audit records what was found. Concretely, an audit produces three things per item (whether it is addressed, where in the manuscript, and if not, what is missing), whereas a checklist produces a column of ticks. The location requirement makes the difference, because an item marked addressed with no location has not been checked. Two organisational consequences follow. The audit must be conducted by someone other than the author, since the author already believes the items are covered and that belief is what is being tested. The output must also be a gap list with named corrections, because a gap list can be acted on before submission whereas a score cannot.
Adherence is lower than most authors assume, even at the top of the literature. Across 463 randomised controlled trial articles from five high-impact journals, mean adherence to CONSORT for Abstracts was 64.3% where professional medical writing support was declared and 66.5% where it was not. Roughly a third of required abstract items were therefore missing in articles published in journals with the most editorial scrutiny. If that is the level in those journals, an organisation assuming that its own abstracts are near-complete is probably mistaken, which is the practical argument for conducting an audit at all. The figure refers specifically to abstract adherence: the abstract has its own item set and is the component most often under-reported.
The evidence is mixed, and any internal case should acknowledge both aspects. On overall abstract adherence there was no detectable difference: 64.3% with declared support against 66.5% without, p=0.30. At item level the picture is more informative. Specification of the primary outcome (RR 1.77), the participant flow diagram (RR 1.96), recruitment and follow-up dates (RR 2.04) and trial registration (RR 1.64) were all more likely to be reported with support, as were harms (RR 2.04, 95% CI 1.37 to 3.03) and funding source (RR 1.75, 95% CI 1.18 to 2.60). Study setting, however, was reported substantially less often, at RR 0.40. These are observational comparisons between confounded groups, so they identify where reporting differs rather than establishing cause.
There are three reasons. Items are not equivalent, and a score treats them as such: omitting the primary outcome and omitting the registration number each cost one item, although one makes the trial uninterpretable and the other merely unfindable. Movements in opposite directions cancel, and this is not hypothetical: the 463-article comparison produced a flat aggregate at p=0.30 while harms reporting roughly doubled and setting reporting more than halved, so an organisation tracking the mean would have concluded that nothing was happening. Finally, a score cannot be acted on: “68%” gives an author no direction, whereas “the primary outcome is in the methods but not the abstract and the setting says multiple centres with no country” represents twenty minutes of work.
The items that benefit most are those requiring judgement rather than attention. The items that move most reliably when a trained person works through a checklist are the ones with a definite, locatable answer: primary outcome named, flow diagram present, recruitment dates given, registration number stated, funder named. A second pass adds little value there. It adds considerable value on study setting, the limitations paragraph and the conclusion, which are items where the question is adequacy, not presence. A setting statement reading “multiple centres in Europe” passes a presence check and conveys nothing; a limitations paragraph listing the limitations common to all trials also passes. Study setting is also the item that the measured data show declining, at RR 0.40.
The abstract has its own item set, its own audience and its own downstream use. The measured 64.3% and 66.5% figures are abstract adherence, so about a third of abstract items were missing in high-impact journal articles. A large share of readers never read past the abstract, which makes an omission there an omission for most of the audience. The abstract must carry the primary outcome, the effect estimate with its precision, the registration and harms in a word count that invites all four to be cut. The abstract is also the component that travels into congress submissions, plain-language summaries and internal materials, detached from the paper that would have supplied the missing context. It should be audited for four things: outcome matching the protocol, estimate with precision, harms present, and conclusion about the result.
Most of it. A reporting guideline covers the study; GPP also expects declarations about how the publication came about, and a reporting checklist scores few of them, so an audit limited to CONSORT or STROBE will pass a paper with every declaration missing. A second short pass should cover: authorship criteria met and documented for each named author; contributions stated; all non-author contributors acknowledged, including professional writers, with the funder of that support named; the sponsor’s role described in substance, meaning who designed the study, held the data, analysed it and decided to submit; author access to the data described accurately; prior disclosure of the same data at congress or as a preprint; and a data availability statement on which a reader could act.
No, and keeping these separate is the main discipline the audit requires. An audit establishes that the manuscript reports what happened. It says nothing about whether what happened was adequate: a trial with an inappropriate comparator, an underpowered sample or a questionable analysis can be reported completely and will pass. Complete reporting exposes problems rather than removing them, which is its purpose. The two failures require different remedies (a reporting gap is corrected by writing; a conduct problem is not), and an audit that blurs the line produces false reassurance. A returned gap list should indicate which items are reporting failures and which are conduct issues that the reporting has revealed.
References
- International Committee of Medical Journal Editors. Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals. ICMJE. 2026. read 9 October 2026. icmje.org
- International Committee of Medical Journal Editors. The New ICMJE Recommendations (August 2013). ICMJE. 2013. read 9 October 2026. icmje.org
- Committee on Publication Ethics. Core Practices. COPE. read 9 October 2026. equator-network.org
- International Society for Medical Publication Professionals. GPP 2022: Good Publication Practice guidelines for company-sponsored biomedical research published in Annals of Internal Medicine (press release). ISMPP. 2022. read 9 October 2026. ismpp.org
- DeTora LM, Toroser D, Sykes A, Vanderlinden C, Plunkett FJ, Lane T, Hanekamp E, Dormer L, DiBiasi F, Bridges D, Baltzer L, Citrome L. Good Publication Practice (GPP) Guidelines for Company-Sponsored Biomedical Research: 2022 Update. Annals of Internal Medicine. 2022;175(9):1298-1304. doi:10.7326/M22-1460 read 9 October 2026. doi.org/10.7326/M22-1460
- Riley DS, Barber MS, Kienle GS, Aronson JK, von Schoen-Angerer T, Tugwell P, et al. CARE guidelines for case reports: explanation and elaboration document. Journal of Clinical Epidemiology. 2017;89:218-235. read 9 October 2026. community.the-hospitalist.org
- Mills I, Sheard C, Hays M, Douglas K, Winchester CC, Gattrell WT. Professional medical writing support and the reporting quality of randomized controlled trial abstracts among high-impact general medical journals. F1000Research. 2017;6:1489. doi:10.12688/f1000research.12268.2 read 9 October 2026. doi.org/10.12688/f1000research.12268.2
- Gattrell WT, Hopewell S, Young K, Farrow P, White R, Wager E, Winchester CC. Professional medical writing support and the quality of randomised controlled trial reporting: a cross-sectional study. BMJ Open. 2016;6(2):e010329. doi:10.1136/bmjopen-2015-010329 read 9 October 2026. doi.org/10.1136/bmjopen-2015-010329
- Battisti WP, Wager E, Baltzer L, Bridges D, Cairns A, Carswell CI, Citrome L, Gurr JA, Mooney LA, Moore BJ, Pena T, Sanes-Miller CH, Veitch K, Woolley KL, Yarker YE. Good Publication Practice for Communicating Company-Sponsored Medical Research: GPP3. Annals of Internal Medicine. 2015;163(6):461-464. doi:10.7326/M15-0288 read 9 October 2026. doi.org/10.7326/M15-0288
- Gagnier JJ, Kienle G, Altman DG, Moher D, Sox H, Riley D. The CARE guidelines: consensus-based clinical case reporting guideline development. Journal of Medical Case Reports. 2013;7(1):223. doi:10.1186/1752-1947-7-223 read 9 October 2026. doi.org/10.1186/1752-1947-7-223
- Turner L, Shamseer L, Altman DG, Schulz KF, Moher D. Does use of the CONSORT statement impact the completeness of reporting of randomised controlled trials published in medical journals? A Cochrane review. Systematic Reviews. 2012;1:60. read 9 October 2026. care-statement.org
- Hopewell S, Ravaud P, Baron G, Boutron I. Effect of editors' implementation of CONSORT guidelines on the reporting of abstracts in high impact medical journals: interrupted time series analysis. BMJ. 2012;344:e4178. read 9 October 2026. care-statement.org
- Fang FC, Steen RG, Casadevall A. Misconduct accounts for the majority of retracted scientific publications. Proceedings of the National Academy of Sciences of the USA. 2012;109(42):17028-17033. doi:10.1073/pnas.1212247109 read 9 October 2026. doi.org/10.1073/pnas.1212247109
- Schulz KF, Altman DG, Moher D. CONSORT 2010 statement: updated guidelines for reporting parallel group randomized trials. Annals of Internal Medicine. 2010;152:726-732. read 9 October 2026. care-statement.org
- Graf C, Battisti WP, Bridges D, Bruce-Winkler V, Conaty JM, Ellison JM, Field EA, Gurr JA, Marx M-E, Patel M, Sanes-Miller C, Yarker YE. Good publication practice for communicating company sponsored medical research: the GPP2 guidelines. BMJ. 2009;339:b4330. doi:10.1136/bmj.b4330 read 9 October 2026. doi.org/10.1136/bmj.b4330
- Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Medicine. 2009;6:e1000097. read 9 October 2026. care-statement.org
- Gotzsche PC, Kassirer JP, Woolley KL, Wager E, Jacobs A, Gertel A, et al. What should be done to tackle ghostwriting in the medical literature? PLoS Medicine. 2009;6(2):e1000023. read 9 October 2026. archive.emwa.org
- Hopewell S, Clarke M, Moher D, Wager E, Middleton P, Altman DG, Schulz KF; CONSORT Group. CONSORT for reporting randomized controlled trials in journal and conference abstracts: explanation and elaboration. PLoS Medicine. 2008;5:e20. read 9 October 2026. care-statement.org
- Norris R, Bowman A, Fagan JM, Gallagher ER, Geraci AB, Gertel A, et al. International Society for Medical Publication Professionals (ISMPP) position statement. Current Medical Research and Opinion. 2007;23:1837-1840. read 9 October 2026. archive.emwa.org
- von Elm E, Altman DG, Egger M, Pocock SJ, Gotzsche PC, Vandenbroucke JP; STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ. 2007;335:806-808. read 9 October 2026. care-statement.org
- Jacobs A, Wager E. EMWA guidelines on the role of medical writers in developing peer-reviewed publications. Current Medical Research and Opinion. 2005;21:317-321. read 9 October 2026. archive.emwa.org
- von Elm E, Poglia G, Walder B, Tramer MR. Different patterns of duplicate publication: an analysis of articles used in systematic reviews. JAMA. 2004;291(8):974-980. doi:10.1001/jama.291.8.974 read 9 October 2026. doi.org/10.1001/jama.291.8.974
- Steinbrook R. Public registration of clinical trials. New England Journal of Medicine. 2004;351:315-317. doi:10.1056/NEJMp048191 read 9 October 2026. doi.org/10.1056/NEJMp048191
- Hamilton CW, Royer MG; AMWA 2002 Task Force on the Contributions of Medical Writers to Scientific Publications. AMWA position statement on the contributions of medical writers to scientific publications. AMWA Journal. 2003;18:13-16. read 9 October 2026. archive.emwa.org
- Wager E, Field EA, Grossman L. Good publication practice for pharmaceutical companies. Current Medical Research and Opinion. 2003;19(3):149-154. doi:10.1185/030079903125001767 read 9 October 2026. doi.org/10.1185/030079903125001767
- Wager E, Field EA, Grossman L. Good publication practice for pharmaceutical companies: why we need another set of guidelines. Current Medical Research and Opinion. 2003;19(3):147-148. read 9 October 2026. ismpp.org
- Antes G, Chalmers I. Under-reporting of clinical trials is unethical. Lancet. 2003;361:978-979. read 9 October 2026. ismpp.org
- Stern JM, Simes RJ. Publication bias: evidence of delayed publication in a cohort study of clinical research papers. BMJ. 1997;315:640-645. read 9 October 2026. ismpp.org
- Easterbrook PJ, Berlin JA, Gopalan R, Matthews DR. Publication bias in clinical research. Lancet. 1991;337:867-872. read 9 October 2026. ismpp.org
Every source above is a peer-reviewed article, a publisher’s record of one, or a standard issued by the body that owns it. Each carries the date it was last read. Where a standard moves, this article is revised rather than left up with a note.
How to cite this article
Pick your referencing style and copy it. Your university or target journal may vary the punctuation — check its guide if it has one.
Pubrica Academy. (2026, October 9). Auditing a Publication Against the Reporting Standards Required by Good Publication Practice. Pubrica. https://pubrica.com/academy/concepts-definitions/reporting-standards-of-gpp-2/
Pubrica Academy. “Auditing a Publication Against the Reporting Standards Required by Good Publication Practice.” Pubrica, 9 Oct. 2026, pubrica.com/academy/concepts-definitions/reporting-standards-of-gpp-2/.
Pubrica Academy (2026) Auditing a Publication Against the Reporting Standards Required by Good Publication Practice. Pubrica. Available at: https://pubrica.com/academy/concepts-definitions/reporting-standards-of-gpp-2/ (Accessed: 15 October 2026).
Pubrica Academy. “Auditing a Publication Against the Reporting Standards Required by Good Publication Practice.” Pubrica. October 9, 2026. https://pubrica.com/academy/concepts-definitions/reporting-standards-of-gpp-2/
Pubrica Academy, ‘Auditing a Publication Against the Reporting Standards Required by Good Publication Practice’ (Pubrica, 9 October 2026) <https://pubrica.com/academy/concepts-definitions/reporting-standards-of-gpp-2/> accessed 15 October 2026.
Pubrica Academy. Auditing a Publication Against the Reporting Standards Required by Good Publication Practice. [Internet]. Pubrica; 2026 Oct 9 [cited 2026 Oct 15]. Available from: https://pubrica.com/academy/concepts-definitions/reporting-standards-of-gpp-2/
Pubrica Academy, “Auditing a Publication Against the Reporting Standards Required by Good Publication Practice,” Pubrica, Oct. 9, 2026. [Online]. Available: https://pubrica.com/academy/concepts-definitions/reporting-standards-of-gpp-2/
@misc{reporting_standards_of_gpp_2_2026,
author = {{Pubrica Academy}},
title = {Auditing a Publication Against the Reporting Standards Required by Good Publication Practice},
howpublished = {Pubrica},
year = {2026},
month = {oct},
url = {https://pubrica.com/academy/concepts-definitions/reporting-standards-of-gpp-2/},
note = {Accessed: <span data-cite-acc="iso">2026-10-15</span>}
}This article is written to be learned from. Cite it if you are referring to what it says; if you are using the evidence behind it, cite the sources listed above instead.
Pre-submission reporting audit worksheet
One sheet per publication, completed by someone who did not write it, at a point where the paper can still change. The scoring includes a middle category because that is where real problems lie, and every item requires a location because an item marked addressed without one has not been checked.
PRE-SUBMISSION REPORTING AUDIT
PUBLICATION: _______________________________________
DESIGN: ____________________ TARGET JOURNAL: ____________
AUDITOR: ____________________ DATE: ____________
[ ] The auditor is NOT an author of this publication
STAGE: [ ] first draft [ ] pre-circulation [ ] pre-submission
(an audit at submission is a record, not an intervention)
REPORTING GUIDELINE USED (name it — the audit cannot be
interpreted later without this)
[ ] CONSORT (randomised trial)
[ ] STROBE (observational) — variant: ______________
[ ] PRISMA 2020 (systematic review)
[ ] CARE (case report)
[ ] SPIRIT (protocol)
[ ] other: ______________________
Extension applied, if any: ______________________
SCORING KEY
A = addressed
P = partially addressed — a note on what is missing is REQUIRED
N = not addressed
NA = not applicable to this design, with a reason
PART 1 — THE PAPER, ITEM BY ITEM
Copy each guideline item into its own row. Do not summarise.
Item | Score | Location (section / para / line) | What is missing
----------------------------------------------------------------
____ | ____ | _______________________________ | ______________
____ | ____ | _______________________________ | ______________
____ | ____ | _______________________________ | ______________
____ | ____ | _______________________________ | ______________
(continue for all items)
PART 2 — HIGH-PRIORITY ITEMS (judgement, not presence)
These pass a presence check while conveying nothing. Read them.
STUDY SETTING score ____ location ______________
[ ] number of centres given (not "multiple")
[ ] countries named
[ ] level of care stated (university / secondary / primary)
[ ] recruitment dates separate from follow-up end date
[ ] capability or volume stated where it affects transferability
[ ] what the control pathway actually was
Missing: ____________________________________________
LIMITATIONS score ____ location ______________
[ ] limitations are specific to THIS study
[ ] generic limitations common to all studies removed
[ ] direction of likely bias stated, not just its existence
Missing: ____________________________________________
CONCLUSION score ____ location ______________
[ ] states what the PRIMARY outcome showed
[ ] no secondary outcome promoted to the conclusion
[ ] no recommendation the design cannot support
Missing: ____________________________________________
PART 3 — THE ABSTRACT (separate object, separate pass)
Abstract item set used: ______________________
[ ] Primary outcome named, and it MATCHES the protocol
[ ] Effect estimate given WITH a measure of precision
[ ] Harms reported — including an explicit statement where none
[ ] Trial registration present in the abstract itself
[ ] Recruitment / follow-up dates present
[ ] Conclusion is about the result, not the intervention
[ ] Funding named
Gaps: ______________________________________________
PART 4 — GPP DECLARATIONS (no reporting checklist scores these)
[ ] Authorship criteria met and DOCUMENTED for each author
[ ] Contributions statement present and specific
[ ] All non-author contributors acknowledged
including the professional writer, if any
and WHO FUNDED that support: ______________________
[ ] Sponsor's role stated in substance:
designed the study? ______ held the data? ______
analysed the data? ______ decided to submit? ______
[ ] Author access to the data described accurately
[ ] Prior disclosure declared (congress / preprint / companion
paper): ______________________
[ ] Data availability statement is OPERABLE — a reader can tell
who to ask and what they would receive
[ ] Conflicts of interest complete for every author
PART 5 — DOUBLE SCORING
Proportion of items independently second-scored: ______%
Second auditor: ______________________
Disagreements recorded (do not reconcile silently): ______
Items with clustered disagreement: ______________________
-> item definition needs tightening
PART 6 — OUTPUT: GAP LIST (this is what goes to the author)
# Item | What is missing | Reporting or conduct? | Owner | By
----------------------------------------------------------------
1 ____ | _______________ | [ ]R [ ]C | _____ | ____
2 ____ | _______________ | [ ]R [ ]C | _____ | ____
3 ____ | _______________ | [ ]R [ ]C | _____ | ____
(R = fixable by writing. C = a conduct issue the reporting
has revealed; it is not fixable by writing and must be
escalated, not edited away.)
[ ] Completed checklist with locations attached to submission
[ ] NO single adherence percentage reported to the author
(if governance requires a metric, track gap FREQUENCY
across publications instead) Pubrica AcademyWritten and maintained by Pubrica’s subject matter experts. Articles are revised when the underlying standards change, and each article states the date on which it was last reviewed.
How Pubrica can help
Pubrica conducts this audit as an independent pass, providing one judgement per guideline item with its location in the manuscript, a three-category score that makes partial reporting visible, and a gap list that separates issues correctable through writing from conduct issues exposed by the reporting. Pubrica audits the abstract against its own item set rather than inferring its status from the paper’s result, and adds a second pass over the GPP declarations (authorship documentation, non-author contributors and who funded them, the substantive role of the sponsor, and the operability of the data availability statement), which no reporting checklist scores.