Session 8 · Publish: finish the paper, get it out, present

Foundations of Clinical Research: From Clinical Question to Publication

Vương Mỹ Lượng & Bernard Osang’ir

Neudata Consulting Ltd

Session 8 · Publish: finish the paper, get it out, present

Part 1

Week 8 · Saturday, 19:30-22:30 (online)

• 19:30 Recap: Session 7
• 19:35 8.1 Writing the paper (II): Introduction, Discussion, Abstract, Title
• 20:00 Manuscript workshop 2: Q1 together, then your own
• 20:30 8.2 Getting published: from checklist to reviewers
• 21:05 8.3 Be the reviewer: Papers 1 and 2, as a whole class
• 21:30 Proposal presentations (about 5 volunteers), 22:15 closing

Session 8 (19:30). Online, 19:30-22:30, with a 15-minute break at 20:50. The last session. Today we finish writing the FLUID-ICU manuscript (Q1 together, then each participant’s own question), follow a paper from submission to reviewers, practise reviewing as a whole class, and about five volunteers present their own proposals.

Before the session: send the PDFs (Session 8 exercise sheet, Papers 1 and 2 original versions, the class Q1 draft from Session 7). Collect the presenters’ slides and open them on your computer. Load the polls.

Recap: three questions from Session 7

QUICK QUIZ · 5 min

• Who consents for a sedated patient in septic shock?
• Which STROBE item asks for crude and adjusted estimates?
• Where does ‘this may be because sicker patients got more fluid’ belong?

Answers

• The legal representative (or deferred consent, if the committee approved it)
• Item 16 (main results)
• Discussion, not Results

OUTPUT Everyone back in writing mode

Recap (19:30-19:35). Question 1 was last week’s exit question: read two or three of the answers you received and correct any misunderstanding. Questions 2 and 3 are quick polls. Reveal the answers.

Question 3 leads into 8.1: the Discussion is where interpretation lives.

8.1 | 25 minutes

Writing the paper (II)

Introduction, Discussion, Abstract and Title, in that order

Section 8.1 (19:35-20:00). Ten slides with three worked examples from FLUID-ICU Q1 and a 2-minute check.

Order follows the writing order from last Saturday: Introduction, Discussion, then the Abstract and the Title last.

Pace: about 2 minutes per slide. Slow down on the worked Discussion example.

Introduction: a funnel in three paragraphs

KEY MESSAGE Known, unknown, aim. The aim matches the primary outcome.

Teaching point: the introduction is a funnel, not a literature review. Paragraph 1: what is known (burden, mechanism, current practice). Paragraph 2: what is not known: the gap, ideally with conflicting evidence cited on both sides. Paragraph 3: ‘We therefore aimed to …’, one aim sentence that matches the primary outcome in the Methods.

Paper 1 (Boulet 2024, the trial) is the model (annotations #10-#12): known (positive fluid balance is linked to death, hidden fluids are a large share of intake), unknown (earlier trials targeted resuscitation fluids only), aim (primary objective = fluid balance after 5 days), all in under 500 words (pages 2-3).

Paper 2 (Hyun 2023, the cohort) defines the gap in paragraph 2 by citing both sides of a conflict (annotation #9). Copy that. But its introduction runs to about 550 words and the aim repeats the causal word ‘impact’ (#3, #10).

Link to Sessions 1-2: the problem statement and PICO of your proposal already give you this funnel.

Ask (call on two or three people, microphone on): in one sentence, what is not known about the question you chose? That sentence is your paragraph 2.

Worked example: the Introduction funnel for FLUID-ICU (Q1)

1 Known

• Fluids save lives early in sepsis
• A positive balance is linked to death (Paper 2 and other cohorts)

2 Not known

• Sicker patients get more fluid
• Unclear whether the link remains after adjusting for severity

3 Aim

• We examined whether a positive balance by day 3 is associated with 28-day mortality, after adjustment

KEY MESSAGE Three paragraphs, about 350 words. The aim uses ‘associated’.

Worked example. Model opening sentences (adapt, do not copy):

P1: ‘Intravenous fluids are a cornerstone of early sepsis care, but a positive cumulative fluid balance has been associated with higher mortality in several observational studies [refs].’

P2: ‘Because sicker patients receive more fluid, it remains uncertain whether this association persists after adjustment for illness severity [refs].’

P3: ‘We therefore examined whether a positive cumulative fluid balance by ICU day 3 was associated with 28-day mortality in adults with sepsis, before and after adjustment for severity.’

References: participants use the references found in Sessions 1-2 plus Papers 1 and 2.

Point out: the confounding by indication from Session 4 is the gap. That is why the study is worth doing.

Discussion: six moves

1 Answer

• First sentence answers the question
• Paper 1, p.8

2 Compare

• With other studies, and why they differ
• Paper 1: vs CLASSIC, POINCARE-2

3 Explain

• Mechanism, briefly
• Paper 2: four-phase fluid model

4 Limits

• Specific: mechanism, direction, fix
• Paper 1: control-group contamination

5 Implications

• For practice and research, labelled
• Paper 1: ‘would probably be advantageous’ is speculation

6 Conclusion

• Same message as abstract and title

KEY MESSAGE Answer first. Limitations with their direction. Conclusion = abstract = title.

Teaching point: a good discussion makes six moves, in this order: answer the question, compare with other studies and explain why results differ. Explain the mechanism briefly, state the limitations specifically, with their likely direction and a fix. Give implications for practice and research, labelled as such. Conclude with the same message as the abstract and title.

Paper 1 does most of this well: the first sentence answers the primary question (annotation #36). The comparison with the CLASSIC and POINCARE-2 trials explains why results differ (#37). Limitations are specific, with mechanism and a fix (#38, #39). The conclusion matches the abstract (#44). Weak point: ‘would probably be advantageous’ is speculation without data (#46).

Paper 2: one-paragraph mechanism using the four-phase fluid model (#34). Comparison closes the loop with the conflicting studies from the introduction (#35). But its limitations miss survivor bias, residual lactate imbalance and confounding by indication (#36).

Poll: which move do first-time writers usually skip? Expected: the limitations’ direction, saying whether a bias makes the effect look bigger or smaller.

Worked example: the Discussion for FLUID-ICU question 1

Six moves, one sentence each (simulated data).

• 1 Answer: A positive balance by day 3 was associated with higher 28-day mortality (adjusted OR 1.62, 95% CI 1.02 to 2.59).
• 2 Compare: This agrees with Paper 2 (day-2 HR 2.13), although our adjusted estimate is smaller.
• 3 Explain: Most of the crude association (OR 2.49) reflected confounding by indication: sicker patients received more fluid.
• 4 Limitations: Residual confounding and reverse causation may exaggerate the association. The CI of the adjusted Cox HR reached 1.00.
• 5 Implications: These data do not justify restricting fluid. A randomised trial (like Paper 1) is needed.
• 6 Conclusion: ‘associated with’, the same verb as the aim, abstract and title.

Worked example (six moves, one sentence each, numbers from key_results.md, simulated data). OR = odds ratio. HR = hazard ratio. CI = confidence interval.

1 Answer: adjusted OR 1.62 (95% CI 1.02 to 2.59). 2 Compare: Paper 2 found day-2 HR 2.13 (1.48 to 3.06). 3 Explain: the crude OR 2.49 fell to 1.62 after adjustment (confounding by indication). 4 Limitations: residual confounding, reverse causation (patients who are dying accumulate fluid), patients dying before day 3. The adjusted Cox HR 1.46 had a CI reaching 1.00. 5 Implications: no basis to restrict fluid. A trial like Paper 1 is needed. 6 Conclusion: ‘associated with’.

Optional point for those who chose Q1: the association looked stronger in heart failure (adjusted OR 4.39 vs 1.29, p for interaction 0.031, only 90 patients). This is a hypothesis, not a finding.

Why the odds ratio shrank: explain it in the Discussion

FLUID-ICU Q1 (simulated)

• Crude OR 2.49 (1.70 to 3.70)
• Adjusted OR 1.62 (1.02 to 2.59)
• Imputed OR 1.70 (1.09 to 2.64)
• Gap = confounding by indication

KEY MESSAGE Report both, explain the gap. Keep the verb at ‘associated’.

The figure shows the three estimates from results pack Q1 (Table 6): crude (n = 618), adjusted complete-case (n = 527) and adjusted after multiple imputation (n = 620).

Teaching point: the drop from 2.49 to 1.62 is the Session 4 lesson made visible: sicker patients received more fluid and died more. The imputed estimate is close to the complete-case one, so missing lactate did not change the conclusion.

The adjusted CI starts at 1.02, just above 1. Calibrated sentence: ‘a modest association remained’.

Ask: what would you write if the adjusted CI had crossed 1? Expected: ‘no clear evidence of an association after adjustment’, not ‘no effect’.

Calibrated language: match the verb to the design

Stronger claim = needs a stronger design

Verb (strongest at the top)	Needs at least
‘causes’, ‘prevents’, ‘proves’	Several large RCTs agree
‘reduced’, ‘increased’, ‘did not reduce’	A randomised trial (Paper 1)
‘may contribute to’, ‘is consistent with’	Observational + causal reasoning
‘is associated with’, ‘predicts’	Cohort, case-control (Paper 2)
‘was observed in’, ‘was common’	Descriptive, case series

Never: ‘trend’ for p = 0.42. ‘No difference’ (say ‘no evidence of a difference’). ‘First’, ‘ever’, ‘novel’ unchecked. ‘Impact’ in an observational study.

KEY MESSAGE Climb the ladder only as far as your design allows.

Teaching point: the strength of a verb is a claim. Over-interpretation (conclusions that go beyond the data) is one of the commonest reasons for rejection.

Read the ladder from the bottom: descriptive studies ‘observe’. Cohorts and case-control studies find that something ‘is associated with’ an outcome. Strong observational evidence plus causal reasoning may ‘contribute to’. A randomised trial may say ‘reduced’ or ‘did not reduce’. ‘Causes’ or ‘proves’ needs several consistent large trials.

Never: ‘non-significant trend’ for p = 0.42 (Paper 1, page 10, annotation #41). ‘No difference’ when the study was too small (say ‘no evidence of a difference’, #29). ‘First’, ‘ever’, ‘novel’ without checking (‘lowest fluid balance ever published’, #40). ‘Impact’, ‘did not impact’, ‘show’ in an observational study (Paper 2, #1, #27, #28). ‘The most reliable comparison’ (#32). Ranking HRs from different cohorts as ‘the highest risk’ without a test (#37).

Ask (call on two or three people, microphone on): rewrite ‘positive fluid balance increases mortality’ for Paper 2. Expected: ‘a positive fluid balance on ICU day 2 was associated with higher 28-day mortality’.

Check: fix the verb (FLUID-ICU, simulated)

QUICK QUIZ · 2 min

• ‘A positive balance caused higher mortality (OR 1.62).’
• ‘Fluid overload increases AKI (acute kidney injury) (OR 1.27 per 10 ml/kg).’
• ‘There was a trend towards death (Cox HR 1.46, 95% CI 1.00 to 2.13).’
• ‘Lactate predicts death (OR 1.39 per mmol/L).’

Better

• ‘… was associated with …’
• ‘… was associated with higher odds of AKI’
• ‘HR 1.46 (1.00 to 2.13): the CI reaches 1’
• Fine for prediction. For cause, ‘associated with’

Check (2 minutes). Numbers from key_results.md (simulated): Q1 adjusted OR 1.62. Q2 adjusted OR 1.27 per 10 ml/kg. Q1 adjusted Cox HR 1.46 (1.00 to 2.13). Q4 adjusted complete-case OR 1.39 per mmol/L.

One poll per sentence: ‘fine’ or ‘fix’. Call on one person to say the fix with the microphone on.

Sentence 3: no ‘trend’. Give the estimate and say the CI reaches 1.

Sentence 4: ‘predicts’ is acceptable in a prediction sense. If the authors mean cause, ‘associated with’ is the honest verb.

The structured abstract: the shop window

Section	What it must say	Good example
Background	Problem, gap and aim in 1-2 sentences
Methods	Design, setting, participants, exposure, primary outcome, analysis	Paper 2: population, setting and size
Results	Numbers analysed. Main effect with 95% CI. Key secondary results	Paper 2: every result as HR (95% CI)
Conclusion	Answer the aim in words the design allows	Paper 1: negative result stated plainly

250-300 words, plus the registration number for trials. Fix in Paper 1: medians reported, but the estimate called a ‘mean difference’.

Teaching point: most clinical journals want a structured abstract (Background, Methods, Results, Conclusions) of about 250-300 words. Trials add the registration number. Many readers never go further, so it must stand alone.

Good practice: Paper 2’s abstract gives the source population, setting and size (annotation #4) and every result as HR with 95% CI and exact P (annotation #5). Paper 1’s conclusion reports a negative result plainly, with no spin (annotation #6), and the registration number is in the abstract (annotation #7).

Fix (Paper 1, annotation #4): medians are reported but the estimate is called a ‘mean difference’, and p = 0.0506 with a CI touching 0 is given to four decimals. The summary, the estimate and the test must match, and borderline p-values are not ‘nearly significant’.

Write the abstract last, and check that each number appears identically in the abstract, text and tables.

Worked example: the abstract and title for FLUID-ICU question 1

Copy the numbers from your Results (simulated data, about 120 words shown).

• Title: Positive fluid balance by day 3 and 28-day mortality in sepsis: a multicentre cohort study
• Background: A positive fluid balance is linked to death in sepsis, but sicker patients receive more fluid. We examined the association after adjustment for severity.
• Methods: Prospective cohort (simulated teaching data), 620 adults, 4 ICUs, 2023-2024. Exposure: positive balance by day 3. Outcome: 28-day death. Logistic and Cox regression adjusted for severity.
• Results: Of 618 patients, 373 (60.4%) had a positive balance, 35.9% vs 18.4% died. Crude OR 2.49 (1.70 to 3.70), adjusted OR 1.62 (95% CI 1.02 to 2.59).
• Conclusions: A positive balance was modestly associated with 28-day mortality. Most of the crude association reflected severity. A randomised trial is needed.

Worked example (simulated data, about 120 words shown, and a full abstract is up to 250).

Title: names the exposure, outcome, population and design. No ‘impact’.

Methods: must say the data are simulated (course rule).

Results: the crude and adjusted ORs, as in the Results. The participant numbers (618 analysed, 373 positive).

Conclusions: ‘modestly associated’, which matches an adjusted CI of 1.02 to 2.59.

Ask: which abstract sentence would change if the adjusted CI had been 0.95 to 2.40? Expected: Results and Conclusions, which would read ‘no clear evidence of an association after adjustment’.

Title and keywords: your PICO in one line

Formula from your PICO: design, I vs C, in P, on O. 10-16 words. No ‘impact’, ‘novel’, ‘first’.

Paper 1 (trial)

• Published: ‘Impact on fluid balance of an optimized restrictive strategy targeting non-resuscitative fluids in … septic shock: a single-blind, … pilot study’
• Problem: over 25 words, value words
• Better: Restricting non-resuscitative fluids in septic shock: a multicentre pilot randomised trial

Paper 2 (cohort)

• Published: ‘Impact of a cumulative positive fluid balance during the first three ICU days in patients with sepsis: a propensity score-matched cohort study’
• Problem: ‘impact’ is a causal word for a cohort
• Better: Positive fluid balance in the first three ICU days and 28-day mortality in sepsis: a propensity-matched cohort

KEY MESSAGE Keywords: 3-6 MeSH terms (PubMed subject headings) not already in the title.

Teaching point: the title is the most-read sentence of the paper. Build it from the PICO: design, intervention or exposure vs comparator, population, outcome. Aim for about 10-16 words. State the design. Avoid value-laden words (‘impact’, ‘optimized’, ‘novel’, ‘first’).

Paper 1 (annotation #1): the design is in the title (good), but it is over 25 words and uses value words. A shorter descriptive version is shown in green.

Paper 2 (annotation #1): ‘Impact’ is causal wording for an observational cohort. Reviewers routinely ask for ‘Association between …’.

Keywords (Paper 1, annotation #9): its keywords repeat words already in the title. Choose 3-6 MeSH (Medical Subject Headings) terms that are NOT in the title, so the paper is found by more searches.

You will write your own title in the workshop that follows.

Before you submit: references, checklist, final read

References

• Use a reference manager
• Check every citation
• Paper 2 cites the wrong reference for its covariates

Reporting checklist

• Fill it in while writing
• Add page numbers
• Neither paper names CONSORT or STROBE

Final read

• Text = table, for every number
• Call-outs point to the right figure
• A colleague reads the tables

KEY MESSAGE Most errors in these two papers needed no new data, only a careful co-author.

References: use a reference manager (Zotero, Mendeley, EndNote) and check every citation. In Paper 2 the covariate list cites reference 26 (the Sepsis-3 definitions paper), almost certainly the wrong number (annotation #22). Never cite a reference an AI tool suggested without reading it.

Reporting checklist: fill in CONSORT (trials) or STROBE (observational studies) while you write, with page numbers. Many journals require it at submission. Neither paper names its guideline (both annotated summaries flag this).

Final read: check every number in the text against the tables (Paper 1’s creatinine, #25), every figure call-out (Paper 1 refers to Fig. 1 where it means Fig. 2A, #28), and typography of numbers (an IQR whose upper limit is below the lower limit, #35).

Key message from the annotated summaries: almost all of Paper 1’s errors would have been caught by a co-author reading the tables against the text.

Manuscript workshop 2 | 30 minutes

Complete the draft: together, then on your own

22 minutes drafting Q1 live with the whole class, then 8 minutes on your own aim and title

Manuscript workshop 2 (20:00-20:30). Course dataset.

Part 1 (20:00-20:22): you add the Introduction, Discussion, Abstract and Title to last Saturday’s class draft for Q1, live on the shared screen.

Part 2 (20:22-20:30): each participant writes the aim sentence and title for their own question, then finishes the draft and submits it by 23:59 tonight (Course/Assignment/manuscript_brief.md).

Instructions: Course/Practicals/s8_exercises.md, Part B. Model paragraphs for Q1: Course/Solutions/s8_solutions.md.

Manuscript workshop 2, part 1: we complete Q1 together

WHOLE-CLASS TASK · 22 min

• Introduction (6 min): three paragraphs. You suggest sentences aloud when called on
• Discussion (9 min): six moves, one or two sentences each
• Abstract (5 min): numbers copied from the class Results
• Title and verb check (2 min): every ‘cause’ or ‘impact’ becomes ‘associated with’

Use

• Class Q1 draft from Session 7 (PDF)
• Results pack Q1
• The worked examples (slides)

OUTPUT A complete class draft for Q1, shared as a PDF after the session

Workshop Part 1 (20:00-20:22). Whole class. Open last week’s class draft and type into it. Call each phase.

Prompts: ‘Does our first Discussion sentence answer the question with the number?’ ‘Is there a limitation with its direction?’ ‘Does our conclusion use the same verb as our aim?’ ‘Is every number in the abstract identical to the Results?’

Share the completed class draft as a PDF after the session. It is the model for everyone’s own manuscript.

Manuscript workshop 2, part 2: your own aim and title

INDIVIDUAL TASK · 8 min

• Write the aim sentence for your question (Q1 to Q5)
• Write your title: design stated, 10-16 words, ‘associated’
• Two volunteers read their title aloud. The instructor comments
• Tonight: add the Introduction, Discussion and Abstract. Submit by 23:59

Submit

• Complete draft, about 2,000 words
• By 23:59 tonight
• Course folder or email to the course lead
• Then the peer review (closing slides)

OUTPUT Your aim sentence and working title. The complete draft submitted by 23:59 tonight

Workshop Part 2 (20:22-20:30). Individual.

Question-specific reminders (post them in the chat): Q2, AKI may also cause fluid accumulation (reverse causation). Q3, ventilator-free days count death as zero. Explain the measure. Q4, compare complete-case and imputed estimates and say what MAR (missing at random) means for the conclusion. Q5, survivors only: say who the result applies to.

Deadline: the complete draft by 23:59 on the day of Session 8 (Course/Assignment/manuscript_brief.md). It then goes to the individual peer review (closing slides).

8.2 | 20 minutes

Getting published

From checklist to reviewers, and the traps on the way

Section 8.2 (20:30-20:50). Six slides, the 3-minute live demo at 20:36 and a 2-minute check at 20:48.

Pace: about 2 minutes per slide. Point to the slide, give one example, move on.

Link to the manuscripts: a FLUID-ICU paper would go through exactly this route: STROBE checklist, journal choice, submission package, peer review. Preprints, licences, what happens after acceptance and rejection are in the Extra slides.

Reporting guidelines: choose by design

Study design	Reporting guideline	Our example
Randomised trial	CONSORT	Paper 1
Cohort, case-control, cross-sectional	STROBE	Paper 2, FLUID-ICU
Diagnostic accuracy study	STARD
Systematic review or meta-analysis	PRISMA
Trial protocol	SPIRIT
Case report	CARE

All checklists are free at the EQUATOR Network (equator-network.org). Name your guideline in Proposal Part 4 and use it as a writing checklist.

Teaching point: for each study design there is an internationally agreed reporting checklist. all of them are collected free at the EQUATOR Network (Enhancing the QUAlity and Transparency Of health Research).

CONSORT for randomised trials (updated as CONSORT 2025), STROBE for observational studies (cohort, case-control, cross-sectional), STARD for diagnostic accuracy, PRISMA for systematic reviews, SPIRIT for trial protocols (SPIRIT 2025), CARE for case reports.

Running example: Paper 1 is an RCT (CONSORT). Paper 2 is a cohort (STROBE). Interesting finding from the annotated papers: neither paper names its guideline. Paper 1 does include the CONSORT flow diagram in the main text. Paper 2 keeps its flow chart in the supplement.

Many journals require the completed checklist at submission, with page numbers.

Misconception: ‘reporting guidelines tell you how to do the study’. They tell you what to report, but reading them before you start is the best design checklist you will find.

Poll: which guideline fits your own proposal? Expected: most ICU proposals are cohort or cross-sectional (STROBE). A few are trials (CONSORT, and SPIRIT for the protocol).

Choosing a journal: four checks

Decide the target journal before you write: it sets the rules.

Fit

• Who needs your result?
• Read 3 recent issues
• Similar designs published?

Trust

• Indexed in MEDLINE, DOAJ, Scopus or Web of Science
• Clear peer-review policy

Rules

• Guide for Authors
• Word limit, checklist
• Registration required?

Cost and speed

• Article fee (APC) and waivers
• Licence: who may re-use the paper
• Paper 1: 7 weeks. Paper 2: 7 months

KEY MESSAGE A good journal is one your readers read, and one that is honestly run.

Teaching point: choose the journal at the start, because its Guide for Authors sets word limits, structure, checklist and registration rules.

Fit: the people who need your answer (ICU nurses? intensivists?) decide the journal. Trust: indexed in MEDLINE/PubMed, the Directory of Open Access Journals (DOAJ), Scopus or Web of Science, with a clear peer-review and retraction policy.

Cost: many open-access journals charge an article-processing charge (APC). Ask about waivers and institutional agreements. Licence: Paper 1 is CC BY-NC-ND (share, but no adaptation or commercial use), Paper 2 is CC BY (re-use with attribution). CC = Creative Commons.

Speed: Paper 1 went from submission to acceptance in about 7 weeks (10 Sept to 31 Oct 2024). Paper 2 took about 7 months (15 Jan to 24 Aug 2023). Speed varies. It is not a sign of quality by itself.

Ask (call on two or three people, microphone on): name one journal your unit reads regularly. That is often a sensible first target.

Spot the predatory journal

KEY MESSAGE If a journal finds you by flattering e-mail, check it twice.

Teaching point: predatory journals take publication fees without providing real peer review, editing or indexing. Your work disappears into a journal nobody trusts, and you cannot submit it elsewhere.

Walk the five red flags on this (invented) e-mail: flattering mass e-mail from a free address. A vague, huge scope and a name that imitates a real journal. Impossibly fast review, fake or unknown ‘impact’ metrics. Fees hidden until after acceptance. Other flags: editorial board members who cannot be verified, no physical address, many spelling errors.

Checks: Think.Check.Submit (a free checklist), DOAJ for open-access journals, the NLM Catalog (US National Library of Medicine) to see if a journal is really indexed in MEDLINE. Also ask the hospital librarian.

Poll: did you receive an e-mail like this in the last month? Usually most say yes: every clinician who has published once receives them.

The Session 8 exercise sheet (Part C) has a short ‘spot the predatory journal’ exercise for homework.

Live demo: find a checklist and check a journal

LIVE DEMO · 3 min

• Open the EQUATOR Network website and search ‘CONSORT’
• Show the checklist and the flow diagram (Paper 1 has one)
• Find STROBE the same way (Paper 2 and FLUID-ICU)
• Check a journal in DOAJ and in the NLM Catalog
• Open the Think.Check.Submit checklist

Watch for

• Where the checklist PDF is
• How to see if a journal is in MEDLINE
• Three questions to ask any journal

OUTPUT Everyone checks the journal and guideline in their own publication plan at home

Demo (20:36-20:39), on the shared screen. Full script with offline fallback: Course/Demos/s8_demo.md.

Keep it to 3 minutes: EQUATOR home page, search box, CONSORT 2025 page, checklist and flow diagram, then STROBE. Then a DOAJ search for a journal title, and the NLM Catalog with the filter ‘currently indexed in MEDLINE’. End on the Think.Check.Submit checklist. Paste the four web addresses in the chat.

If your internet fails, use the screenshots described in the demo script.

The submission package

Item	What it contains
Cover letter	Why this journal, what is new. Not under review elsewhere
Title page	Authors, affiliations, ORCID iDs (free researcher identifiers), corresponding author, word count
Manuscript	IMRaD, formatted as the Guide for Authors says (blinded if asked)
Tables and figures	Separate files, right format and resolution
Reporting checklist	CONSORT, STROBE …, with page numbers
Declarations	Ethics number, consent, conflicts of interest, funding, data sharing, author roles, AI use
Registration number	Trials and systematic reviews
Suggested reviewers	Experts with no conflict, not recent co-authors
Supplementary files	Protocol, extra tables, long flow charts

Teaching point: journals check the package before any scientific review. Missing items send the manuscript straight back.

Cover letter: one page covering why this journal, what is new, that the work is not under consideration elsewhere, and any preprint. Title page: all authors, affiliations, ORCID iDs (a free, permanent researcher identifier that every participant can register at orcid.org), corresponding author, word count. Manuscript in the journal’s format (blinded if the journal uses double-blind review). Declarations: ethics approval number, consent, competing interests, funding, data availability, author contributions (CRediT) and AI use. Suggested reviewers: experts without conflicts. The editor decides whether to use them.

Running example: Paper 1 has a complete declarations block, exactly what the submission system asks for, item by item (annotation #48). Templates: Course/Templates/Cover_Letter_Template.

Ask: who already has an ORCID iD? Encourage everyone to register this week.

The peer-review journey

Rejected? Improve the paper, then submit to another journal. Never to two at once.

Submit

• Manuscript, cover letter, checklist

Editor

• Screens the paper
• Desk rejection is common

Reviewers

• 2 or 3 peer reviewers comment

Decision

• Accept, minor or major revision, or reject

Revise

• Change the paper
• Reply point by point

Accepted

• Proofs, then publication

KEY MESSAGE Most papers are revised before acceptance. Reply to every point politely.

Walk the six steps. Submit (with cover letter, reporting checklist and declarations). The editor screens, and many papers are rejected here without review (‘desk rejection’) because they do not fit or are poorly reported. Two or three peer reviewers comment. The editor decides: accept, minor revision, major revision or reject. You revise and answer every comment point by point. Accepted, then proofs (read them carefully!) and publication.

Running example: our annotators judged Paper 1 ‘minor revision’ and Paper 2 ‘major revision’. Both were published, and both still contain errors a careful reviewer would have caught. Paper 1 took about 7 weeks from submission to acceptance. Paper 2 about 7 months.

Rule: submit to one journal at a time. After rejection, improve the paper using the reviews, then submit elsewhere.

Ask: a reviewer is wrong about something. What do you do? Expected: thank them, explain politely with evidence, and clarify the text so the next reader is not confused.

Responding to reviewers: the point-by-point table

Comment, response, and exactly where the paper changed. Illustrative rows built on the annotated Paper 1.

R1.1

• Comment: please do not call p = 0.42 a ‘non-significant trend’.
• Accept: thank you, we agree. We report the estimate with its 95% CI and removed ‘trend’.
• Change: Discussion, p.10, lines 4-6

R1.2

• Comment: how was allocation concealed with blocks of 2?
• Clarify: allocation was central and computer-generated. Block size added as a limitation.
• Change: Methods p.3, limitations p.10

R2.1

• Comment: please add 90-day mortality.
• Rebut, politely: 90-day status was not a pre-specified outcome and was not collected.
• Change: noted as a limitation, p.10

KEY MESSAGE Thank, answer every point, show what changed, and disagree politely with evidence.

Teaching point: the response letter is a table: every reviewer comment, your response, and exactly where the manuscript changed (page and line). Number the comments (R1.1, R1.2, R2.1 …).

Tone: thank the reviewers, agree where you can, never ignore a point, never argue emotionally. When you disagree (a rebuttal), explain politely with evidence and, where possible, still clarify the text so the next reader is not confused.

The three rows are what the annotators say a reviewer would have asked Paper 1: remove the ‘non-significant trend’ (accept). Justify blocks of 2 (clarify and add a limitation). A request for an outcome that was never collected (polite rebuttal: you cannot add data that do not exist, but you can name it as a limitation).

Common mistakes: changing the manuscript without saying where. Answering ‘done’ with no detail. accepting a request that makes the paper wrong just to please the reviewer.

Template: Course/Templates/Response_to_Reviewers_Template. Participants use this format when they answer their peer reviewer after the course.

Check: true or false?

QUICK QUIZ · 2 min

• ‘A journal that e-mails me first is probably predatory’
• ‘I can send my paper to two journals at once to save time’
• ‘A preprint has been peer reviewed’
• ‘I must answer every reviewer comment, even if I disagree’
• ‘The FLUID-ICU paper should follow CONSORT’

Answers

• Often: check it twice
• False: duplicate submission
• False: posted before peer review
• True: politely, with evidence
• False: STROBE (cohort)

Check (20:48-20:50). One true/false poll per statement. Reveal.

Statement 1 is ‘often true’ rather than always. Real journals also invite papers, but a flattering mass e-mail from an unknown journal deserves the five-flag check.

Statement 3: a preprint is a manuscript posted publicly (e.g. on medRxiv) before peer review (Extra slides).

Break

15 min

The slide is only a picture on purpose. Before the break, say: we are back at 21:05 for 8.3, be the reviewer (25 min), then the proposal presentations (45 min) and the closing (15 min). During the break, open Paper 1 and Paper 2 (original PDFs) and the Session 8 exercise sheet, Part D (the answer grid). Presenters: check that your slides reached the instructor.

Break (20:50-21:05).

Open the presenters’ slides on your computer and check you can share each one. Confirm the running order by chat with the volunteers.

Prepare your screen for 8.3: Paper 1 and Paper 2 (original, unannotated PDFs) in two tabs.

8.3 | 25 minutes

Be the reviewer

The whole class appraises Paper 1 and Paper 2 together

Section 8.3 (21:05-21:30). Full instructor guide: Course/Resources/appraisal_instructor_guide.md (model answers, prompts, key messages).

Structure: 3 minutes set-up (two papers, the 10 questions), 15 minutes whole-class appraisal with polls, 7 minutes on what each design may claim.

Why now: after the course each participant reviews one peer’s FLUID-ICU draft with the same 10 questions. This is the rehearsal.

Reassure: ‘Can’t tell’ is a perfectly good answer if the paper does not say.

Two papers, one question

Same clinical question (fluids in septic shock), two designs. Both PDFs: Course/Papers.

Paper 1: randomised trial

• Boulet 2024, Critical Care
• Pilot trial, 48 patients analysed
• Restrictive vs usual fluids in septic shock
• Primary outcome: fluid balance on day 5

Paper 2: cohort study

• Hyun 2023, Annals of Intensive Care
• Propensity-matched cohort, 2516 patients
• Positive fluid balance in sepsis
• Outcome: 28-day mortality

KEY MESSAGE Same question, two designs: compare what each may claim.

Explain the set-up in one minute. Both papers address our running question (fluids in septic shock) with different designs. The whole class appraises both, with the same 10 questions (next two slides).

Each participant keeps their own copy of the answer grid (Session 8 exercise sheet, Part D) and fills it in as we go: Yes / No / Can’t tell, plus the page.

Link to Session 4: the 10 questions are a simplified version of the formal risk-of-bias tools: RoB 2 for Paper 1, Newcastle-Ottawa (or the newer ROBINS-E) for Paper 2. We do not run the formal tools today.

The 10-question checklist (1-5)

1 Question

• Can you write the study’s PICO in one line?

2 Design

• What is the design, and is it the right one for the question?

3 Participants

• Who was included, and who was left out?
• Would your patients have been included?

4 Comparison groups

• Were the groups similar at the start (Table 1)?
• How was this achieved: randomisation, matching or adjustment?

5 Bias

• Who knew which group patients were in (blinding)?
• Could that change the outcome?

The checklist, part 1. Read each question aloud once. Do not explain the answers.

Q1-Q2 are about the question and the design (Sessions 1-4). Q3 is about who was studied, external validity: would your patients be in this study? Q4 is about whether the groups were comparable at baseline (look at Table 1). Q5 is about blinding.

Answers are Yes / No / Can’t tell, plus the page.

The 10-question checklist (6-10)

6 Outcome

• Was there one pre-specified primary outcome, measured the same way in both groups?

7 Numbers

• What is the main result as an effect size with 95% CI?
• Is the CI narrow or wide?

8 Words vs design

• Do the conclusions match the design (‘associated’ vs ‘caused’, no ‘trend’ for p = 0.42)?

9 Limitations

• Are the main limitations stated honestly?
• Name one the authors missed.

10 Use

• Would this change practice in your unit?
• Why, or why not?

The checklist, part 2. Q6 and Q7 use what we learnt in Session 6: one primary outcome, an effect size with its 95% confidence interval.

Q8 is the heart of today’s exercise: are the words in the title, abstract and discussion as strong as the design allows? ‘Caused’ or ‘impact’ needs randomisation. Observational studies say ‘associated’. And p = 0.42 is never a ‘trend’.

Q9: the authors’ own limitations section, plus one they missed. Each paper has a clear one (see the instructor guide).

Q10: clinical judgement. There is no single right answer, but a reason is needed.

Be the reviewer: the whole class appraises both papers

WHOLE-CLASS TASK · 15 min

• The instructor shows the right page of each paper on the shared screen
• Paper 1: questions 4, 5, 7, 8 and 9
• Paper 2: questions 4, 5, 8 and 9
• For each question, vote in the poll: Yes, No or Can’t tell
• A volunteer gives the page and the sentence, microphone on

Where to look

• Abstract: question and result
• Methods: design, blinding, outcome
• Table 1: groups similar?
• Discussion: words and limitations

OUTPUT A class answer grid for both papers, shared as a PDF. The other questions as homework

Whole-class appraisal (21:08-21:23). About 1.5 minutes per question. Read the question, show the page, run the poll, then call on one person to give the evidence with the microphone on. Type the class answer into the grid on the shared screen.

The most useful prompts (more in the instructor guide): Paper 1: ‘Who was not blinded, and who decided how much fluid the patients got?’ (clinicians were unblinded and they prescribe the fluids that make up the outcome). ‘Find p = 0.42 in the Discussion. What word did the authors use?’ (page 10, ‘non-significant trend’). Paper 2: ‘Who decided which patients got more fluid?’ (the doctors, and sicker patients receive more fluid: confounding by indication). ‘Patients had to stay at least 3 days in the ICU. Who is missing?’ (early deaths and early discharges: survivor bias).

The remaining questions (1-3, 6 and 10) are homework on the answer grid: good practice for the peer review after the course.

What may each design claim?

Paper 1: randomised trial (n = 48)

• Chance decided the groups: known and unknown factors balanced
• Day-5 balance: difference 13.2 ml/kg (95% CI -15.2 to 41.6), p = 0.42
• May say: ‘did not reduce’
• But too small to be sure: no ‘trend’

Paper 2: matched cohort (n = 2516)

• Doctors decided the fluid: only measured factors balanced
• Day-2 positive balance and 28-day death: HR 2.13 (95% CI 1.48 to 3.06)
• May say: ‘is associated with’
• Not ‘impact’ or ‘show’: sicker patients get more fluid

KEY MESSAGE Randomisation earns the word ‘caused’. Observation earns ‘associated’.

Plenary (21:23-21:30), with the next table.

Paper 1: chance decided who got the restrictive strategy, so known and unknown factors are balanced on average, so a randomised trial may make a causal claim. Its claim is negative: the strategy did not reduce day-5 fluid balance (difference 13.2 ml/kg, 95% CI -15.2 to 41.6, p = 0.42). But with 48 patients the CI is wide: this is ‘no evidence of an effect’, not ‘evidence of no effect’, and certainly not a ‘trend’.

Paper 2: doctors decided how much fluid each patient got. Propensity matching balances only the factors that were measured. Sicker patients receive more fluid (confounding by indication, Session 4) and some severity is never fully measured. So the paper may say a positive balance on day 2 ‘is associated with’ higher 28-day mortality (HR, hazard ratio, 2.13, 95% CI 1.48 to 3.06), not that it ‘impacts’ or ‘shows’ anything causal, although its title says ‘Impact’.

Poll: if you had to change ICU practice tomorrow, which paper is stronger evidence for causation, and which tells you more about real-world risk? Expected: the RCT for causation (but it is a small pilot). The cohort for a large, real-world association that needs a trial to confirm.

The two papers side by side

Paper 1: RCT (Boulet 2024)	Paper 2: cohort (Hyun 2023)
Groups made similar by	Randomisation 1:1 (blocks of 2)
Who knew the group	Clinicians knew, patients, statistician blinded
Main result	13.2 ml/kg (95% CI -15.2 to 41.6), p = 0.42
Missed limitation	Blocks of 2 with unblinded clinicians
Wording problem	‘Non-significant trend’ (p = 0.42)
May claim	Did not reduce day-5 balance (small pilot)

RCT = randomised controlled trial. Pages and details: Course/Resources/appraisal_instructor_guide.md

Reveal this table at the end of the plenary as the summary, after the class has given its own answers, not before.

Row by row: how the groups were made comparable. Who knew the allocation. The main effect with its CI. The limitation the authors missed (Paper 1: permuted blocks of 2 make the next allocation predictable when clinicians are unblinded, page 3. Paper 2: requiring at least 3 ICU days excludes patients who died or left early (survivor bias), page 2). The wording problem (Paper 1 page 10. Paper 2 title and page 4). And the claim each design supports.

Both papers are good, published papers in Q1 journals, and both have fixable flaws. That is the normal state of the literature, which is why appraisal is a clinical skill.

Land the four messages: design sets the verb. A wide CI means ‘we don’t know yet’. Published is not the same as correct. Name the reporting guideline.

Transition: ‘Now it is time to hear some of your own proposals.’

Proposal presentations | 45 minutes

Your proposals and publication plans

About 5 volunteers, 6 minutes each, then 2 minutes of feedback. Everyone submits slides.

Presentations (21:30-22:15). About five volunteers present their own proposal (volunteers told the course lead by the Wednesday before Session 8): 5 x (6 minutes + 2 minutes feedback) = 40 minutes, then a 5-minute closing round in which you sum up the common strengths and problems seen in the submitted proposals and remind everyone of the deadlines. Show this slide and the scoring slide in the first minute.

Everyone, presenting or not, submits slides and the written proposal before 19:30 today (Course/Assignment/proposal_brief.md).

The co-instructor keeps time and posts ‘2 min’ at 4:00, ‘1 min’ at 5:00 and ‘Time’ at 6:00 in the chat. The chair stops the talk at 6:30.

Remind the class: the reviewing skills we just used now apply to our own proposals, but feedback is given the way you would like to receive it.

How proposals and presentations are scored

Criterion	What we look for	Points
12 Structure and content	All 6 slides. Question, design, outcome, sample size, ethics, journal and guideline visible	0-6
13 Clarity	Every doctor and nurse in the class understands. One message per slide	0-6
14 Visual design	Readable on a shared screen. Pictures, not text	0-6
15 Timekeeping	Finished between 5:00 and 6:00	0-6
16 Questions and discussion	Answers directly and honestly, names the limitations	0-6

Written proposal (criteria 1-11, /70): everyone. Criteria 12-16 (/30): presenters only. Rubric: Course/Assignment/proposal_rubric.md

Scoring: the written proposal (criteria 1-11 of Course/Assignment/proposal_rubric.md, out of 70) is scored for every participant, after the course. Criteria 12-16 (6/4/2/0 points each, out of 30) score the presentation, and only for those who present. Non-presenters: the written subtotal is scaled to 100 (rule at the top of the rubric). Presenting never lowers anyone’s score.

Who scores the presentations: the instructors. Two or three people give feedback with the microphone on (two stars and a wish).

Feedback in 2 minutes (Course/Assignment/presentation_guide.md): call on two people to give two stars and a wish with the microphone on (45 seconds). instructor 1 minute (one strength, the most important improvement, one question). Presenter 15 seconds (‘thank you’, no defending). Fill in the feedback form at the end of the rubric and email it with the saved chat after the session.

The score is feedback to help each participant polish the proposal for the ethics committee.

Proposal presentations

PRESENTATIONS · 45 min

• About 5 volunteers present their own proposal: 6 minutes, 6 slides
• Slides: title, problem and PICO, design, data and analysis, ethics and feasibility, publication plan
• Time signals in the chat: ‘2 min’ at 4:00, ‘1 min’ at 5:00, ‘Time’ at 6:00
• Feedback, 2 minutes: two people give two stars and a wish (microphone on), then the instructor
• Everyone submits slides and the written proposal

Running order

• 21:30 Presenter 1
• 21:38 Presenter 2
• 21:46 Presenter 3
• 21:54 Presenter 4
• 22:02 Presenter 5
• 22:10 Closing round (5 min)

OUTPUT About five proposals presented. Every proposal and slide set submitted

Presentations (21:30-22:15). Share each presenter’s slides from your computer, or let the presenter share their screen. Keep this slide up between presenters.

Instructor feedback: start with a genuine strength. Give the most important improvement for the written proposal or publication plan (usually design, outcome definition, feasibility or the choice of journal and guideline). One question if time allows.

Check slide 6 of each presenter: target journal and correct guideline (STROBE for a cohort, CONSORT for a trial), working title in calibrated words, author roles, one line of the abstract skeleton.

Typical issues to watch for: outcome not measurable in routine data. A causal claim planned from an observational design (link back to the appraisal!). No plan for consent in sedated patients. unrealistic recruitment.

If the session runs late, shorten the closing round, never the last presentations.

Extra slides | Session 8

Extra slides: Session 8

Use only if time allows, or for self-study

Extra: the slides after this divider are optional. Use them if time allows or point participants to them for self-study.

Key words of Session 8

Term	In plain words
Introduction funnel	Known, not known, aim: three paragraphs
Six moves	Answer, compare, explain, limitations, implications, conclusion
Calibrated language	Verbs as strong as the design allows (‘associated’, not ‘caused’)
Predatory journal	Takes fees without real peer review or indexing
Point-by-point response	A table answering every reviewer comment with the change made
Preprint	Public manuscript posted before peer review

Full glossary (English-Vietnamese): Course/References/glossary_EN_VN (PDF in the course folder)

Extra: the key words of Session 8, for self-study. They also appear in the post-course quiz.

Write it right: fix the spin, then draft your title

INDIVIDUAL TASK · 6 min

• Rewrite Paper 1’s ‘non-significant trend’ sentence honestly (2 min)
• Rewrite Paper 2’s title with calibrated wording
• Draft the working title of your own proposal from its PICO (4 min)
• One line each for Background, Methods, expected Results, Conclusion

Publication plan (Part 4)

• Working title
• Abstract skeleton
• Authors and their roles
• Target journal and guideline (8.2)

OUTPUT Working title and abstract skeleton for your own presentation

Extra: optional 6-minute writing drill on the two papers (use it as homework). Worksheet: Course/Practicals/s8_exercises.md, Part A.

Spin sentence (Paper 1, page 10): ‘considering the non-significant trend in fluid balance reduction in the optimized group, our approach … remains worth exploring in a larger randomized controlled trial’. Model rewrite: ‘The point estimate favoured the restrictive strategy, but the 95% CI was wide (-15.2 to 41.6 ml/kg) and included no effect. A larger trial would be needed to detect a smaller difference.’

Title (Paper 2): ‘Positive cumulative fluid balance during the first three ICU days and 28-day mortality in sepsis: a propensity score-matched cohort study’. Use ‘association’, not ‘impact’.

Own title: check 10-16 words, design stated, no value words. Abstract skeleton: Results are Expected results, so write the format, e.g. ‘28-day mortality, RR (95% CI)’.

Preprints, open access and licences

Preprint

• Posted before peer review (e.g. medRxiv)
• Fast, timestamped, open feedback
• Not peer-reviewed: say so
• Check the journal’s policy

Open access

• Free for everyone to read
• Gold: the journal charges a fee (APC)
• Green: accepted version in a repository
• Ask about fee waivers

Licence

• CC BY: re-use with credit (Paper 2)
• CC BY-NC-ND: share, no changes (Paper 1)
• Read it before you sign

KEY MESSAGE Know who can read, re-use and pay before you submit.

Extra. Preprints: a manuscript posted on a public server (for health research, medRxiv is the best known) before or during peer review. Benefits: speed, a public timestamp, early feedback. Risks: it has not been peer-reviewed, and the media or colleagues may treat it as established. Most journals accept submissions that were preprinted. Check the journal’s policy first.

Open access: gold OA means the journal makes the article free and usually charges an article-processing charge (APC). Both our papers are gold OA. Green OA means you deposit the accepted manuscript in a repository, often after an embargo.

Licences (CC = Creative Commons): Paper 2 is CC BY (anyone may re-use and adapt with credit). Paper 1 is CC BY-NC-ND (share, but no adaptation or commercial use). This also matters when you want to re-use a figure in teaching.

After acceptance: proofs, sharing, corrections

Accepted

• Congratulations

Proofs

• Check every number, name and table

Published

• Online, with a DOI

Share

• Repository, plain-language summary, your unit

Correct

• Openly, if errors appear

KEY MESSAGE Read the proofs as carefully as the data. Errors become permanent.

Extra. Teaching point: after acceptance the publisher sends proofs, often with a short deadline (days). Read them slowly: every number, every name and affiliation, every table and figure. This is the last chance.

Running example: Paper 1 prints the whole ‘Study protocol’ paragraph twice: in the Methods (page 4) and again after the Conclusions (page 11). It is a duplication that should have been caught at proof stage (annotation #45). Paper 2’s affiliations have inconsistent capitalisation (#42).

Sharing: post the article as its licence and the journal’s policy allow (Paper 2 is CC BY and allows re-use. Paper 1 is CC BY-NC-ND and allows no adaptation): institutional repository, a plain-language summary for patients and colleagues, a short talk in your department.

Corrections: if you find an error after publication, contact the journal. A correction is honest, a retraction is for serious errors or misconduct.

Rejected? Resubmit smarter

Rejection is normal: most papers are rejected at least once.

Desk rejection

• Out of scope or poorly reported
• Often within days
• Re-target the journal

After review

• Read the reviews, wait a day
• Fix what they found
• Never resubmit unchanged

Next journal

• Reformat to its Guide for Authors
• Use transfer offers if given
• Appeal only for a clear error

KEY MESSAGE The next reviewer may be the same person. Fix the problems first.

Extra: Teaching point: rejection is part of publishing, not a verdict on you. Many good papers are rejected once or twice before acceptance.

Desk rejection: the editor declines without review, usually because the topic does not fit the journal or the reporting is poor. Re-target: choose a journal whose readers need your result.

After peer review: read the reviews, wait a day before reacting, then fix what they found. The same reviewers are often invited by the next journal in the same field. An unchanged manuscript annoys them.

Next journal: reformat to the new Guide for Authors. Some publishers offer to transfer the manuscript and reviews to a sister journal, which can save months. Appeal only if there is a clear factual error in the decision.

Never submit to two journals at once (duplicate submission is misconduct).

Conference abstracts and posters

KEY MESSAGE A poster is a billboard: one message, one figure, readable from 2 metres.

Extra: Teaching point: a conference abstract is often the first output of a study. It follows the same IMRaD logic as the paper’s abstract, usually 250-300 words, with real numbers and confidence intervals. An abstract that says ‘results will be presented’ is usually rejected. Follow the conference template and deadline.

Poster: a billboard, not a paper on a wall. A short title that states the finding, a few words of background and methods (a flow chart), one big figure, a take-home message in one sentence, and a QR code linking to more detail. It should be readable from 2 metres.

Presenting an abstract does not normally prevent publication of the full paper, but check the journal’s policy. Do write the full paper: many abstracts never become publications.

A conference abstract template is in Course/Templates/Conference_Abstract_Template.

From question to publication: what you can now do

Protect

• Ethics, consent, registration
• Honest data and authorship

Write

• IMRaD in the right order
• Numbers with CIs
• Verbs that match the design

Publish

• Reporting checklist
• Trusted journal
• Polite, complete responses

Review

• 10 questions
• What each design may claim

KEY MESSAGE Next: submit your proposal and your manuscript, then review one peer’s draft.

Extra: a one-slide summary of Sessions 7-8 for self-study. The closing slides cover what the whole course built.

Closing: what you built and what happens next

Part 2

Session 8 · 22:15-22:30

• Post-course quiz (the same 24 questions)
• What you built in eight sessions
• Course evaluation
• After the course: submit, then review one peer’s manuscript
• Next steps

Closing (22:15-22:30), the last 15 minutes of Session 8. Keep it brisk and positive.

Thank the volunteers who presented, and everyone who submitted slides, before starting the quiz.

Paste the links to the post-course quiz and the evaluation form in the chat.

Post-course quiz

QUICK QUIZ · 8 min

• The same 24 questions as in Session 1, in an online form (link in the chat)
• Individual, no discussion
• Use the same participant code as in Session 1
• The answer key is shared after the course

OUTPUT Quiz submitted (participant code only)

Post-course quiz (22:15-22:23). The same online form as the pre-course quiz in Session 1.

Later: compute the mean pre/post change with the table in the pre/post quiz answer key (Course/Assessment) and report it to the department head in the course report.

What you built in eight sessions

Ask & find

• A PICO question
• 5 key references
• A working title

Design

• A justified design
• Outcomes
• A bias plan

Analyse

• A CRF and analysis plan
• The course dataset read and interpreted

Write

• Methods, Results
• Introduction, Discussion, Abstract

Publish

• Your own proposal
• Your own draft manuscript

KEY MESSAGE Each of you leaves with a research proposal and a draft paper.

What you built (1 minute). Celebrate: each participant has a proposal they could take to the ethics committee and a draft paper written from the course dataset. CRF = case report form.

Ask (call on two or three people, microphone on): what is your next step with your proposal? (Who to discuss it with, a target date for the ethics committee.)

Course evaluation

INDIVIDUAL TASK · 2 min

• Rate each session: content, pace, usefulness
• One thing to keep, one thing to change
• Interested in Level 2 (statistical modelling in R)?

OUTPUT Evaluation form submitted (anonymous)

Evaluation (2 minutes, finished at home if needed). Short anonymous online form (link in the chat). The pace question matters: we planned the course to be slow and online in the evening, so tell us whether it was right.

After the course: peer review like real authors

Tonight

• Submit your complete draft by 23:59
• Course folder or email

2 days

• The course lead sends you one peer’s draft to review

Week 1

• Send your review report
• Use the 10-question checklist

Week 2

• Revised draft and response letter
• Use the response-to-reviewers template

KEY MESSAGE Individual homework: everyone reviews one draft and receives one review (Course/Assignment/peer_review_instructions).

Peer review (2 minutes). This is individual homework. Drafts are due by 23:59 tonight. Within 2 days the course lead sends each participant one peer’s draft to review (a simple list so that every draft is reviewed exactly once, where possible on a different question), by email or through the course folder. Reviews are due within 1 week. The revised draft and response letter within 2 weeks.

Reviews follow Course/Assignment/peer_review_instructions.md and the 10-question checklist from 8.3. Authors reply with Course/Templates/Response_to_Reviewers_Template.

This is the last step of the whole research process, exactly what happens after a real submission.

What comes next

Your toolkit

• Slides, handouts, glossary
• Templates: proposal, protocol, consent form, case report form (CRF), analysis plan
• Templates: manuscript, response to reviewers

Keep going

• Take your proposal to the ethics committee
• Level 2: statistical modelling in R
• Level 3: advanced methods

Presenting is not required.

Level 2 (hands-on, for those who want to run analyses themselves): statistical modelling in R, one model family per session: linear, logistic, count (Poisson, negative binomial) and Cox models, with their assumptions and how to report them. Data cleaning, Table 1 and the basics of missing data are taught as far as the analysis needs them, and each participant does an individual analysis project. Prerequisite: this course plus the free Neudata course Clinical Data Analysis in R.

Level 3 (on request, workshops and consultancy): reproducible projects with version control and Quarto, figures that meet journal requirements, missing-data and sensitivity analyses in depth, systematic review and meta-analysis, and other advanced methods (prediction, diagnostic accuracy, advanced survival, longitudinal and clustered data). See Course_Levels/.

Give the trainers’ contact details for follow-up questions in the chat.

Thank you!

Good questions. Sound designs. Honest analysis. Clear writing.

Thank the department leadership, the organisers and the participants. If everyone agrees, take a screenshot of the class with cameras on.

Copyright © Neudata, 2026

All property rights and copyright are reserved.

This presentation contains proprietary data, information, and materials developed by Neudata for dedicated use only and may not be communicated, copied, reproduced, distributed, published, or cited, in whole or in part, without the prior written consent of Neudata. Where explicit written permission has been granted, appropriate attribution must be included, followed by “by courtesy of Neudata”.

Any unauthorized use or infringement of these materials may give rise to legal action and claims for damages, without prejudice to any other rights of Neudata, including rights relating to patents, trademarks, or other forms of intellectual property protection.