Week 1 · Module 1
Congress materials diagnostic
Twenty-four questions across abstracts, posters, presentations and the compliance rules that govern all three
Your scenario
An abstract is accepted. Six weeks before the congress a colleague posts the headline result on social media. The abstract is withdrawn for breaching embargo — a year of work, removed from the programme, over a post that took thirty seconds to write.
What this course covers
Congress work has its own constraints: hard character limits, submission deadlines months ahead of the data, embargo rules with real teeth, and encore restrictions that determine whether the same data can be presented twice. None of it is difficult. All of it is unforgiving.
The craft half — poster architecture, slide design, data display — is where most materials fail quietly rather than dramatically: not withdrawn, just unread.
Take the diagnostic honestly
Do not look anything up. A wrong answer is worth more than a right one, because it is the only way to find the gap.
—
—
Your twelve areas — study the red ones first
Before you move on
Write down your three weakest areas.
Next module
1.2 — The congress ecosystem
Week 1 · Module 2
The congress ecosystem
Deadlines months ahead of the data, embargoes with teeth, and encore rules that decide whether the same work can be shown twice
Your scenario
The abstract deadline is in nine days. The database locks in six weeks. You must submit an abstract on data that do not yet exist — which is the normal condition of congress work, and the reason the late-breaking route exists.
The timeline
Figure 1.2 — The congress timeline. The gap between the abstract deadline and the presentation is where most of the difficulty lives. Internal review has to finish before month 0, not after acceptance — which is why review calendars work backwards from the deadline, not forwards from acceptance.
Submission routes and formats
| Route or format | What it is | Implication |
|---|---|---|
| Regular abstract | Main deadline, most submissions | Data must exist at the deadline |
| Late-breaking | Later deadline, restrictive criteria | For genuinely new high-impact data, not for missed deadlines |
| Encore | Data presented previously elsewhere | Usually must be disclosed at submission; some congresses prohibit it |
| Poster | Displayed, often with a discussion session | Most accepted abstracts |
| Poster discussion | Poster plus a short moderated presentation | Needs slides as well as the poster |
| Oral | Full presentation in a session | Typically 8–15 minutes including questions |
| Plenary or late-breaking oral | Highest-profile slot | Press interest; embargo discipline is critical |
Embargo
Most congresses embargo abstract content until a stated date and time. Until then nothing may be disclosed publicly — no social media post, no press release, no company website update, no slide shown at an internal meeting that anyone outside the company attends.
Breach can mean withdrawal of the abstract from the programme. This happens every year, usually because one person who was not on the distribution list posted something enthusiastic.
Breach can mean withdrawal of the abstract from the programme. This happens every year, usually because one person who was not on the distribution list posted something enthusiastic.
Drills
Drill 1 of 3
Your database locks six weeks after the abstract deadline. What do you do?
ASubmit on available data, or use the late-breaking route if the data qualify
BSubmit projected results
CAsk the organisers for an extension
DSkip this congress entirely
Drill 2 of 3
A colleague posts the abstract's headline result on social media three weeks before the embargo lifts. What is the consequence?
ANone — it was already accepted
BThe abstract may be withdrawn; report the breach to the congress immediately
CDelete the post and say nothing
DPresent anyway and mention it afterwards
Drill 3 of 3
The same data were presented at a national congress last month. You want to submit to an international one. What must you do?
ASubmit without mentioning it
BChange the title so it appears new
CCheck the encore policy and disclose the prior presentation at submission
DNever submit data presented elsewhere
Production drill
Build the working-backwards timeline for this congress.
Congress: 12–16 September. Abstract deadline 15 April. Acceptance notified 20 June. Embargo lifts 10 September, 00:01 local time.
Your organisation: medical review takes 10 working days; legal and regulatory review a further 5; author review 2 rounds at 5 working days each; poster printing needs 10 working days and must ship 5 days before the congress.
Also: the data were presented at a national meeting in February.
Your organisation: medical review takes 10 working days; legal and regulatory review a further 5; author review 2 rounds at 5 working days each; poster printing needs 10 working days and must ship 5 days before the congress.
Also: the data were presented at a national meeting in February.
Abstract, working backwards from 15 April:
· Submission: 15 April
· Final approval: 11 April (buffer for submission-system problems — assume the portal will be slow on deadline day, because it will be)
· Legal and regulatory review complete: 4 April (5 working days)
· Medical review complete: 28 March (10 working days)
· Author review round 2 complete: 21 March
· Author review round 1 complete: 14 March
· First draft circulated: 7 March — roughly six weeks before the deadline
Poster, working backwards from 12 September:
· Ships: 5 September
· Printing complete: 5 September, so files to printer by 22 August (10 working days)
· Final approval: 20 August
· Legal and regulatory: 13 August
· Medical review: 30 July
· Author rounds: 16 and 23 July
· First draft: 9 July — about two and a half weeks after acceptance
Two issues to raise now:
1. This is an encore. The February national meeting must be checked against this congress’s encore policy and disclosed at submission. Do this before drafting, since a prohibition would change everything.
2. Embargo lifts 10 September at 00:01. Circulate the exact date and time to everyone who has seen the data — including authors, their institutional press offices, and anyone internally who might post. The commonest breach source is an author’s university press office issuing a release on its own schedule.
The point of the exercise: the abstract draft has to exist six weeks before a deadline that is itself five months before the congress. Congress work is calendar work, and the calendar runs backwards.
· Submission: 15 April
· Final approval: 11 April (buffer for submission-system problems — assume the portal will be slow on deadline day, because it will be)
· Legal and regulatory review complete: 4 April (5 working days)
· Medical review complete: 28 March (10 working days)
· Author review round 2 complete: 21 March
· Author review round 1 complete: 14 March
· First draft circulated: 7 March — roughly six weeks before the deadline
Poster, working backwards from 12 September:
· Ships: 5 September
· Printing complete: 5 September, so files to printer by 22 August (10 working days)
· Final approval: 20 August
· Legal and regulatory: 13 August
· Medical review: 30 July
· Author rounds: 16 and 23 July
· First draft: 9 July — about two and a half weeks after acceptance
Two issues to raise now:
1. This is an encore. The February national meeting must be checked against this congress’s encore policy and disclosed at submission. Do this before drafting, since a prohibition would change everything.
2. Embargo lifts 10 September at 00:01. Circulate the exact date and time to everyone who has seen the data — including authors, their institutional press offices, and anyone internally who might post. The commonest breach source is an author’s university press office issuing a release on its own schedule.
The point of the exercise: the abstract draft has to exist six weeks before a deadline that is itself five months before the congress. Congress work is calendar work, and the calendar runs backwards.
Reflection
For a congress in your field, when is the abstract deadline relative to the meeting? Look it up — most people guess far too close.
Next module
1.3 — The abstract
Week 1 · Module 3
The abstract
Two thousand characters, four sections, and a permanent citable record. Compression without loss is the entire skill
Your scenario
Your abstract is 2,340 characters against a 2,000 limit. You cut the confidence intervals, the denominators and the safety sentence to fit. It now fits, and it is no longer reportable science — and this version is what will be indexed and cited for years.
The character budget
Under a hard limit, something must go. The order in which you sacrifice determines whether the abstract remains scientifically complete.
| Section | Share | Cut in this order |
|---|---|---|
| Background | 10–15% | Cut first. One or two sentences: the gap and nothing else. No epidemiology. |
| Methods | 20–25% | Cut second. Design, population, intervention, comparator, primary endpoint. Enough to judge the evidence level. |
| Results | 40–50% | Protect. Numbers with denominators and intervals. This is why anyone reads it. |
| Conclusions | 10–15% | Protect. Two sentences, matched to the design. |
What must never be cut to fit
Denominators · confidence intervals · the comparator · the primary endpoint result · the fact that a key endpoint was immature or not met.
If the abstract does not fit with these in place, cut background and methods further, or cut a secondary endpoint entirely. An abstract that reports a percentage with no denominator and an effect with no interval is not a scientific abstract — and unlike a draft manuscript, it is permanent and citable.
If the abstract does not fit with these in place, cut background and methods further, or cut a secondary endpoint entirely. An abstract that reports a percentage with no denominator and an effect with no interval is not a scientific abstract — and unlike a draft manuscript, it is permanent and citable.
Compression techniques
Long
Compressed
In this randomised, double-blind, placebo-controlled Phase III trial, patients were randomly assigned…104 characters
In this randomised, double-blind, placebo-controlled Phase III trial, patients were assigned 1:1…tighter, same content
Median progression-free survival was 8.9 months in the study regimen arm and 5.2 months in the placebo arm
Median PFS was 8.9 vs 5.2 monthsexpand PFS at first use elsewhere
There was a statistically significant improvement, with a hazard ratio of 0.66
HR 0.66 (95% CI 0.53–0.82); p<0.001
Adverse events of grade 3 or higher severity were observed in 37.8% of patients
Grade ≥3 AEs: 147/389 (37.8%)
It is important to note that overall survival data were not yet mature at the time of this analysis
OS was immature
Abbreviations in abstracts
Abbreviating saves a great deal of space, and congresses generally permit it — but the abstract must still stand alone. Expand at first use inside the abstract, even if the term is expanded in the title.
Do not abbreviate something used only once; the expansion plus brackets costs more than the term. And check the congress rules, since some prohibit abbreviations in the title entirely.
Do not abbreviate something used only once; the expansion plus brackets costs more than the term. And check the congress rules, since some prohibit abbreviations in the title entirely.
Drills
Drill 1 of 3
Your abstract is 200 characters over. What do you cut?
AThe confidence intervals
BThe denominators
CThe conclusion
DBackground, then non-essential methods detail
Drill 2 of 3
Which abstract conclusion is acceptable?
A“The regimen significantly prolonged PFS vs placebo; OS data were immature and follow-up continues.”
B“This regimen should become the new standard of care.”
C“The regimen improves survival in this population.”
D“Biomarker-positive patients benefit most from this regimen.”
Drill 3 of 3
You submit the abstract in April. By September you have a later data cut with slightly different numbers. What appears where?
AUpdate the abstract to match
BThe abstract stands as submitted; the presentation may show the updated data with its cut-off stated
CPresent only the April data to stay consistent
DWithdraw the abstract
Congress Abstract Blueprint Multi-Therapeutic 250-Word Abstract Structures
Major international medical congresses impose strict character/word limits (e.g. 250–300 words). Compare how primary efficacy data are compressed across major global congresses:
| Congress & Therapeutic Area | Target Word Limit | Key Primary Finding Phrasing (High-Impact Style) |
|---|---|---|
| AHA / ESC (Cardiology) | 250 words / 2,000 characters | "Primary 3-point MACE occurred in 6.4% of active vs. 9.1% of placebo patients (HR 0.69; 95% CI 0.57–0.83; P < 0.001). CV death was reduced by 32% (HR 0.68; 95% CI 0.52–0.89; P = 0.004)." |
| ADA / EASD (Endocrinology) | 250 words / 2,200 characters | "At Week 40, mean HbA1c change was -2.1% with once-weekly dual agonist vs. -1.4% with active comparator (treatment difference -0.7%; 95% CI -0.9 to -0.5; P < 0.001). Weight change was -9.8 kg vs. -3.6 kg." |
| ERS / ATS (Pulmonology) | 300 words | "The annualized asthma exacerbation rate was 0.41 vs. 0.89 (rate ratio 0.46; 95% CI 0.35–0.60; P < 0.001). Pre-bronchodilator FEV1 improved by 180 mL vs. 50 mL (difference 130 mL; P < 0.001)." |
| ASCO / ESMO (Oncology) | 2,000 characters (late-breaker) | "Median PFS was 14.8 months with targeted doublet vs. 7.2 months with chemotherapy (HR 0.52; 95% CI 0.41–0.66; P < 0.0001). Confirmed objective response rate was 68% vs. 34% (P < 0.001)." |
Production drill
Write the abstract in under 2,000 characters including spaces. Protect the results.
Design: randomised, double-blind, placebo-controlled Phase III; second-line advanced disease; 402 patients (201/201); 389 treated; stratified by region and prior lines. Hierarchy PFS → OS → ORR.
PFS: 8.9 (95% CI 7.4–10.3) vs 5.2 (4.4–6.1) months; HR 0.66 (0.53–0.82); p<0.001.
OS: not reached vs 21.7 months; HR 0.84 (0.62–1.14); p=0.26; immature.
ORR: 44% vs 29% in 371 evaluable; nominal p=0.003.
Safety (n=389): grade ≥3 in 147 (37.8%) vs 121 (31.2%); discontinuation for AE 31 (8.0%) vs 22 (5.7%).
Follow-up: median 19.8 months.
PFS: 8.9 (95% CI 7.4–10.3) vs 5.2 (4.4–6.1) months; HR 0.66 (0.53–0.82); p<0.001.
OS: not reached vs 21.7 months; HR 0.84 (0.62–1.14); p=0.26; immature.
ORR: 44% vs 29% in 371 evaluable; nominal p=0.003.
Safety (n=389): grade ≥3 in 147 (37.8%) vs 121 (31.2%); discontinuation for AE 31 (8.0%) vs 22 (5.7%).
Follow-up: median 19.8 months.
Model abstract — 1,412 characters including spaces
Background: Outcomes after failure of first-line therapy for advanced disease remain poor, and no randomised trial has evaluated this agent in the second-line setting.
Methods: In this randomised, double-blind, placebo-controlled Phase III trial, 402 patients with previously treated advanced disease were assigned 1:1 to the study regimen or placebo, stratified by region and prior lines. The primary endpoint was progression-free survival (PFS); secondary endpoints were tested in a prespecified hierarchy.
Results: At a median follow-up of 19.8 months, median PFS was 8.9 (95% CI 7.4–10.3) vs 5.2 (4.4–6.1) months; HR 0.66 (0.53–0.82), p<0.001. Median overall survival (OS) was not reached vs 21.7 months; HR 0.84 (0.62–1.14), p=0.26, and formal testing stopped at this endpoint. Objective response rate was 44% vs 29% (n=371 evaluable; nominal p=0.003). Among 389 treated patients, grade ≥3 adverse events occurred in 147 (37.8%) vs 121 (31.2%), and 31 (8.0%) vs 22 (5.7%) discontinued due to adverse events.
Conclusions: The study regimen significantly prolonged PFS vs placebo in previously treated advanced disease, with a manageable safety profile. OS data were immature and follow-up continues.
Six things this does under the limit:
1. Background is two clauses — the problem and the gap. No epidemiology.
2. All denominators and intervals survive. This is what the 588 characters of headroom bought.
3. The hierarchy break is stated. Eight words, and without them ORR reads as a confirmed finding.
4. ORR is marked nominal and its denominator given.
5. Safety uses n=389, the treated population, not 402.
6. The conclusion matches the primary endpoint — PFS, not survival — and discloses immaturity.
What was cut and never missed: the stratification factor levels, the number treated per arm, the exact p value for ORR beyond nominal status, and any mention of subgroups. None of those is needed for a reader to judge the result.
Background: Outcomes after failure of first-line therapy for advanced disease remain poor, and no randomised trial has evaluated this agent in the second-line setting.
Methods: In this randomised, double-blind, placebo-controlled Phase III trial, 402 patients with previously treated advanced disease were assigned 1:1 to the study regimen or placebo, stratified by region and prior lines. The primary endpoint was progression-free survival (PFS); secondary endpoints were tested in a prespecified hierarchy.
Results: At a median follow-up of 19.8 months, median PFS was 8.9 (95% CI 7.4–10.3) vs 5.2 (4.4–6.1) months; HR 0.66 (0.53–0.82), p<0.001. Median overall survival (OS) was not reached vs 21.7 months; HR 0.84 (0.62–1.14), p=0.26, and formal testing stopped at this endpoint. Objective response rate was 44% vs 29% (n=371 evaluable; nominal p=0.003). Among 389 treated patients, grade ≥3 adverse events occurred in 147 (37.8%) vs 121 (31.2%), and 31 (8.0%) vs 22 (5.7%) discontinued due to adverse events.
Conclusions: The study regimen significantly prolonged PFS vs placebo in previously treated advanced disease, with a manageable safety profile. OS data were immature and follow-up continues.
Six things this does under the limit:
1. Background is two clauses — the problem and the gap. No epidemiology.
2. All denominators and intervals survive. This is what the 588 characters of headroom bought.
3. The hierarchy break is stated. Eight words, and without them ORR reads as a confirmed finding.
4. ORR is marked nominal and its denominator given.
5. Safety uses n=389, the treated population, not 402.
6. The conclusion matches the primary endpoint — PFS, not survival — and discloses immaturity.
What was cut and never missed: the stratification factor levels, the number treated per arm, the exact p value for ORR beyond nominal status, and any mention of subgroups. None of those is needed for a reader to judge the result.
Reflection
Take an abstract you have written. Count its characters. Which section is over-weight relative to the 15/25/45/15 budget?
In most drafts it is background, and the space it takes comes out of results.
Next module
1.4 — The abstract clinic
Week 1 · Module 4
The abstract clinic
A degraded abstract with fourteen planted faults, against a hard character limit and a title requirement
How to do this
Timer at 40 minutes. Rewrite the abstract to under 2,000 characters and write a title. Then reveal the rubric. Reading it first measures nothing.
The draft
Rewrite the title and abstract. Fourteen faults are planted.
Title: A breakthrough treatment for advanced disease
Background: Advanced disease affects over two million people worldwide each year and is a leading cause of mortality. Treatment options after first-line failure are limited, and there is significant unmet need. Many agents have been studied over the past decade with mixed results, and the optimal approach remains unclear.
Methods: Patients were randomised to receive the study regimen or placebo. The primary endpoint was progression-free survival.
Results: The primary endpoint was met. Median PFS was 8.9 vs 5 months (HR .66, p=0.000). Response rate was significantly higher (44% vs 29%, p=0.003). Overall survival showed no benefit. Grade 3 or higher events occurred in 36.6% of the 402 randomised patients. Quality of life improved significantly.
Conclusions: The regimen is highly effective and well tolerated, and should be considered the new standard of care in this setting. It compares favourably with existing options.
Background: Advanced disease affects over two million people worldwide each year and is a leading cause of mortality. Treatment options after first-line failure are limited, and there is significant unmet need. Many agents have been studied over the past decade with mixed results, and the optimal approach remains unclear.
Methods: Patients were randomised to receive the study regimen or placebo. The primary endpoint was progression-free survival.
Results: The primary endpoint was met. Median PFS was 8.9 vs 5 months (HR .66, p=0.000). Response rate was significantly higher (44% vs 29%, p=0.003). Overall survival showed no benefit. Grade 3 or higher events occurred in 36.6% of the 402 randomised patients. Quality of life improved significantly.
Conclusions: The regimen is highly effective and well tolerated, and should be considered the new standard of care in this setting. It compares favourably with existing options.
The fourteen faults
Title (2)
1. “Breakthrough” is a promotional characterisation, and a declarative title stating a result is prohibited by many congresses.
2. The title names no intervention, comparator, population or design. It is unsearchable and uninformative.
Background (2)
3. Far too long — roughly 320 characters of epidemiology and hand-waving, in a section that should be 150–250.
4. No gap stated. “The optimal approach remains unclear” is not a gap; it does not say what was unknown that this study addressed.
Methods (2)
5. No design — randomised is stated, but not double-blind, not placebo-controlled, not Phase III, no n, no ratio. A reader cannot judge the evidence level.
6. No mention of the testing hierarchy, which is what makes the Results section interpretable.
Results (6)
7. 5 months should be 5.2 — decimal precision must match across arms.
8. .66 needs a leading zero, and no confidence interval is given.
9. p=0.000 is not a possible value; use p<0.001.
10. Response rate called significant — it sits below the hierarchy break at OS and is nominal only.
11. “OS showed no benefit” — OS was immature and non-significant, which is not the same as showing no benefit. Give the HR and interval.
12. Safety denominator wrong — 36.6% is computed against 402 randomised; safety uses 389 treated, giving 37.8%. No denominators anywhere in the section.
Conclusions (2)
13. “Highly effective and well tolerated… new standard of care” — three unsupported claims. A single trial does not establish standard of care, and the quality-of-life claim has no supporting data given.
14. “Compares favourably with existing options” — a comparative claim with no head-to-head data.
Model title: Study regimen versus placebo in previously treated advanced disease: a randomised, double-blind, placebo-controlled Phase III trial
Model abstract: as in Module 1.3 — 1,412 characters, with all denominators and intervals intact, the hierarchy break stated, and the conclusion matched to the primary endpoint.
Scoring
11–14 — pass. Your abstracts would go through review.
7–10 — close. Note which section your misses cluster in.
Under 7 — redo Module 1.3 before retrying.
The three that matter most: fault 10, because a nominal endpoint presented as significant becomes a citable claim; fault 11, because immature is not the same as negative and the difference is clinically important; and fault 12, because a wrong safety denominator understates harm in a permanent record.
Title (2)
1. “Breakthrough” is a promotional characterisation, and a declarative title stating a result is prohibited by many congresses.
2. The title names no intervention, comparator, population or design. It is unsearchable and uninformative.
Background (2)
3. Far too long — roughly 320 characters of epidemiology and hand-waving, in a section that should be 150–250.
4. No gap stated. “The optimal approach remains unclear” is not a gap; it does not say what was unknown that this study addressed.
Methods (2)
5. No design — randomised is stated, but not double-blind, not placebo-controlled, not Phase III, no n, no ratio. A reader cannot judge the evidence level.
6. No mention of the testing hierarchy, which is what makes the Results section interpretable.
Results (6)
7. 5 months should be 5.2 — decimal precision must match across arms.
8. .66 needs a leading zero, and no confidence interval is given.
9. p=0.000 is not a possible value; use p<0.001.
10. Response rate called significant — it sits below the hierarchy break at OS and is nominal only.
11. “OS showed no benefit” — OS was immature and non-significant, which is not the same as showing no benefit. Give the HR and interval.
12. Safety denominator wrong — 36.6% is computed against 402 randomised; safety uses 389 treated, giving 37.8%. No denominators anywhere in the section.
Conclusions (2)
13. “Highly effective and well tolerated… new standard of care” — three unsupported claims. A single trial does not establish standard of care, and the quality-of-life claim has no supporting data given.
14. “Compares favourably with existing options” — a comparative claim with no head-to-head data.
Model title: Study regimen versus placebo in previously treated advanced disease: a randomised, double-blind, placebo-controlled Phase III trial
Model abstract: as in Module 1.3 — 1,412 characters, with all denominators and intervals intact, the hierarchy break stated, and the conclusion matched to the primary endpoint.
Scoring
11–14 — pass. Your abstracts would go through review.
7–10 — close. Note which section your misses cluster in.
Under 7 — redo Module 1.3 before retrying.
The three that matter most: fault 10, because a nominal endpoint presented as significant becomes a citable claim; fault 11, because immature is not the same as negative and the difference is clinically important; and fault 12, because a wrong safety denominator understates harm in a permanent record.
Reflection
Which of the four sections did you find hardest to compress without losing content?
Next week
2.1 — Poster architecture
Week 2 · Module 1
Poster architecture
You have three seconds to earn a reader and ninety seconds to keep one. Layout decides both, before a word is read
Your scenario
Your poster is scientifically excellent and nobody stops at it. The title is a 24-word sentence in small type, the primary result sits bottom-right, and the top-left panel is a 300-word background section. Everything a passer-by sees first is the least interesting material on the board.
The reading path
Figure 2.1 — Poster layout. The centre panel is the poster. Everything else supports it. Readers arrive at the title, drop to whatever is visually largest, and only then read left to right — so the primary result must be the largest thing on the board, not the thing at the end of the argument.
Three tests
| Test | How to run it | Pass condition |
|---|---|---|
| Three seconds | Glance at the whole board for three seconds | You can state what the study found |
| Three metres | Stand back three metres | Title and primary result are readable; axis labels legible |
| Ninety seconds | Read only the largest elements | You have design, result, and conclusion — enough to decide whether to ask a question |
Minimum type sizes
On an A0 board: title around 72–100 pt, section headings 36–48 pt, body text 24–28 pt, figure axis labels no smaller than 20 pt, references and disclosures 16–18 pt.
If body text falls below 24 pt to make the content fit, the content is too long. Shrinking type is not a solution, it is the symptom.
If body text falls below 24 pt to make the content fit, the content is too long. Shrinking type is not a solution, it is the symptom.
Drills
Drill 1 of 3
Where does the primary result figure belong?
ABottom right, as the conclusion of the argument
BTop left
CCentre, as the largest element on the poster
DBehind the QR code, to save space
Drill 2 of 3
Your content does not fit at 24 pt body text. What do you do?
AReduce to 18 pt
BReduce the margins
CRequest a larger board
DCut the content and move detail to QR-linked material
Drill 3 of 3
Which title works best on a poster?
AA short title stating the population and the question or finding, readable at 3 m
BA 24-word title naming every element of the design
CA one-word title for impact
DThe full abstract title, unchanged
Production drill
Plan the poster. Allocate every element to a position and a size.
Board: A0 portrait.
Material available: title and authors · background (3 sentences) · study design schematic · baseline characteristics table (18 rows) · PFS Kaplan–Meier figure · OS Kaplan–Meier (immature) · forest plot, 6 subgroups · grade ≥3 adverse events table (22 preferred terms) · discontinuation figures · conclusions · references (14) · disclosures and funding · acknowledgement of writing support.
Material available: title and authors · background (3 sentences) · study design schematic · baseline characteristics table (18 rows) · PFS Kaplan–Meier figure · OS Kaplan–Meier (immature) · forest plot, 6 subgroups · grade ≥3 adverse events table (22 preferred terms) · discontinuation figures · conclusions · references (14) · disclosures and funding · acknowledgement of writing support.
On the board:
· Title and authors — full width, top. Title 80 pt, authors 32 pt.
· Background — top left, small panel. Three sentences reduced to two bullets ending on the gap. 24 pt.
· Design schematic — below background, left column. Replaces a methods paragraph entirely.
· PFS Kaplan–Meier — centre, largest element on the board. Redrawn at poster scale: thicker lines, 24 pt axis labels, numbers at risk legible. One sentence beneath stating what it shows.
· Forest plot — right column, upper. Six subgroups fits; interaction p values must remain legible.
· Grade ≥3 adverse events — right column, cut from 22 preferred terms to the 6–8 most frequent plus an “any grade ≥3” row.
· Conclusions — bottom right, gold panel, one or two lines, 28 pt.
· QR code — beside conclusions.
· Disclosures, funding, writing support — foot of board, 16–18 pt. On the poster, not behind the QR.
· References — cut to the 3–4 actually cited on the board; the rest go to QR.
Moved behind the QR code: the 18-row baseline table (a 3-row summary of key characteristics stays on the board), the immature OS curve, full adverse event listing, full reference list, and the plain-language summary.
The two judgement calls:
1. OS comes off the board. It is a key secondary endpoint, but immature and non-significant, so the curve shows little and would compete with the primary figure for attention. The result is stated in one line in the conclusions panel, so nothing is hidden.
2. The baseline table is cut to three rows. Eighteen rows at 24 pt occupies a quarter of the board to answer a question nobody stopped to ask. Age, biomarker status and prior lines are enough for a passer-by to know who was studied.
· Title and authors — full width, top. Title 80 pt, authors 32 pt.
· Background — top left, small panel. Three sentences reduced to two bullets ending on the gap. 24 pt.
· Design schematic — below background, left column. Replaces a methods paragraph entirely.
· PFS Kaplan–Meier — centre, largest element on the board. Redrawn at poster scale: thicker lines, 24 pt axis labels, numbers at risk legible. One sentence beneath stating what it shows.
· Forest plot — right column, upper. Six subgroups fits; interaction p values must remain legible.
· Grade ≥3 adverse events — right column, cut from 22 preferred terms to the 6–8 most frequent plus an “any grade ≥3” row.
· Conclusions — bottom right, gold panel, one or two lines, 28 pt.
· QR code — beside conclusions.
· Disclosures, funding, writing support — foot of board, 16–18 pt. On the poster, not behind the QR.
· References — cut to the 3–4 actually cited on the board; the rest go to QR.
Moved behind the QR code: the 18-row baseline table (a 3-row summary of key characteristics stays on the board), the immature OS curve, full adverse event listing, full reference list, and the plain-language summary.
The two judgement calls:
1. OS comes off the board. It is a key secondary endpoint, but immature and non-significant, so the curve shows little and would compete with the primary figure for attention. The result is stated in one line in the conclusions panel, so nothing is hidden.
2. The baseline table is cut to three rows. Eighteen rows at 24 pt occupies a quarter of the board to answer a question nobody stopped to ask. Age, biomarker status and prior lines are enough for a passer-by to know who was studied.
Reflection
Find a poster photo from a congress in your field. Run the three-second test. Could you state the finding?
Next module
2.2 — Data display
Week 2 · Module 2
Data display
A manuscript figure dropped onto a poster is unreadable. Redrawing for distance is the difference between a figure that works and one that is decoration
Your scenario
The Kaplan–Meier curve from the manuscript goes onto the poster unchanged. At the board, the axis labels are illegible, the numbers-at-risk row is a grey smear, and the two curves are the same weight in similar colours. The most important figure in the study communicates nothing at viewing distance.
Redrawing for distance
| Manuscript figure | Poster figure |
|---|---|
| Read at 30 cm, in a column | Read at 1–3 m, at scale |
| 8–10 pt axis labels | 20–24 pt minimum |
| Thin lines, subtle distinctions | Thick lines, high contrast, distinguishable at distance |
| All detail included | Only what carries the message; detail to QR |
| Legend below, dense | Labels on the curves where possible — no legend to cross-reference |
| Grey scale acceptable | Colour, with contrast that survives colour-blindness and poor lighting |
Label the curves directly
A legend forces the reader to look away from the data, decode a key, and look back. At a poster, standing up, in a crowd, most people do not bother.
Put the arm name at the end of each curve, in the curve's own colour. It costs nothing and removes a step from every reader's path.
Put the arm name at the end of each curve, in the curve's own colour. It costs nothing and removes a step from every reader's path.
Reducing tables
A 22-row adverse event table on a poster is read by nobody. Cut it to the six to eight most frequent events plus a summary row, and move the full table behind the QR code.
The same applies to baseline characteristics: three to five rows on the board, the full table linked. Completeness is achieved by the QR link; readability can only be achieved on the board.
The same applies to baseline characteristics: three to five rows on the board, the full table linked. Completeness is achieved by the QR link; readability can only be achieved on the board.
Drills
Drill 1 of 3
The manuscript KM curve has 9 pt axis labels. What is required for the poster?
AEnlarge the image file
BRedraw with 20–24 pt labels, thicker lines and direct curve labelling
CRemove the numbers-at-risk row to reduce clutter
DNothing — readers can step closer
Drill 2 of 3
A 22-row adverse event table must go on the poster. What do you do?
AInclude all 22 rows at whatever size fits
BOmit safety data from the poster
CShow the most frequent 6–8 plus a summary row; link the full table
DShow only the events that favour the product
Drill 3 of 3
Why label curves directly rather than using a legend?
AIt saves space
BLegends are prohibited at congresses
CIt makes the figure cheaper to print
DIt removes a decoding step the standing reader will often skip
Production drill
Specify the redraw. List every change from the manuscript version.
Manuscript Kaplan–Meier figure as supplied:
· 9 pt axis labels, 8 pt numbers-at-risk row
· Two curves, both 0.75 pt, one solid black one dashed black
· Legend box in the upper right identifying the arms
· HR, 95% CI and p value in small text inside the plot area
· Censoring ticks at 3 pt
· Y-axis 0 to 1.0 in decimals
· X-axis "Time (months)" 0 to 30, ticks every 6
· 9 pt axis labels, 8 pt numbers-at-risk row
· Two curves, both 0.75 pt, one solid black one dashed black
· Legend box in the upper right identifying the arms
· HR, 95% CI and p value in small text inside the plot area
· Censoring ticks at 3 pt
· Y-axis 0 to 1.0 in decimals
· X-axis "Time (months)" 0 to 30, ticks every 6
Nine changes:
1. Axis labels 9 pt → 24 pt. Non-negotiable; below 20 pt the figure is decorative.
2. Numbers at risk 8 pt → 20 pt, and keep the row. It is the most informative part of the figure and the most often deleted.
3. Curve weight 0.75 pt → 3–4 pt. Thin lines vanish at distance.
4. Black solid vs black dashed → two contrasting colours, checked for colour-blind distinguishability and for poor congress-hall lighting. Keep the dash pattern as a secondary cue so the distinction survives if colour fails.
5. Legend box removed; arms labelled directly at the end of each curve, in the curve colour, at 24 pt.
6. HR, CI and p moved out of the plot area to a clear space at 24 pt, or into the one-line statement beneath the figure.
7. Censoring ticks 3 pt → 6–8 pt so they are visible.
8. Y-axis decimals → percentages, 0 to 100%. Faster to read standing up.
9. Add a one-sentence statement beneath the figure saying what it shows — not a legend, a finding. “Median PFS was 8.9 vs 5.2 months (HR 0.66, 95% CI 0.53–0.82).”
What you have not changed: the data, the censoring, the numbers at risk, or the axis ranges. Redrawing for distance is a legibility operation, not an editorial one. If a redraw makes a result look stronger than the original figure did, something has gone wrong.
1. Axis labels 9 pt → 24 pt. Non-negotiable; below 20 pt the figure is decorative.
2. Numbers at risk 8 pt → 20 pt, and keep the row. It is the most informative part of the figure and the most often deleted.
3. Curve weight 0.75 pt → 3–4 pt. Thin lines vanish at distance.
4. Black solid vs black dashed → two contrasting colours, checked for colour-blind distinguishability and for poor congress-hall lighting. Keep the dash pattern as a secondary cue so the distinction survives if colour fails.
5. Legend box removed; arms labelled directly at the end of each curve, in the curve colour, at 24 pt.
6. HR, CI and p moved out of the plot area to a clear space at 24 pt, or into the one-line statement beneath the figure.
7. Censoring ticks 3 pt → 6–8 pt so they are visible.
8. Y-axis decimals → percentages, 0 to 100%. Faster to read standing up.
9. Add a one-sentence statement beneath the figure saying what it shows — not a legend, a finding. “Median PFS was 8.9 vs 5.2 months (HR 0.66, 95% CI 0.53–0.82).”
What you have not changed: the data, the censoring, the numbers at risk, or the axis ranges. Redrawing for distance is a legibility operation, not an editorial one. If a redraw makes a result look stronger than the original figure did, something has gone wrong.
Reflection
Print a manuscript figure at A4 and stand two metres away. Can you read the axis labels?
Next module
2.3 — Poster copy and the QR ecosystem
Week 2 · Module 3
Poster copy and the QR ecosystem
How much text a board can carry, what belongs behind a code, and why linked material is still congress material
Your scenario
Your poster is approved. The QR code links to a landing page that includes the full adverse event listing and a plain-language summary. Neither went through review. Everything a QR code on a congress poster points to is congress material, and it is subject to the same approval as the board itself.
The text budget
| Panel | Words | Form |
|---|---|---|
| Background | 30–50 | Two bullets, the second ending on the gap |
| Methods | 40–70 | Bullets, or replaced entirely by a design schematic |
| Results | 60–100 | One sentence per display item, stating what it shows |
| Conclusions | 25–40 | One or two lines, matched to the design |
| Total body text | under 300 | If you are past 400, the board will not be read |
Sentences, not paragraphs
A poster is read standing, often at an angle, sometimes while someone else is talking. Paragraphs do not survive those conditions.
Bullets, one idea each, under about 15 words. The exception is the one-line statement under each figure, which should be a full sentence because it carries the finding.
Bullets, one idea each, under about 15 words. The exception is the one-line statement under each figure, which should be a full sentence because it carries the finding.
What goes behind the QR code
| Good QR destinations | Not acceptable |
|---|---|
| Full methods and eligibility criteria | Anything not reviewed and approved |
| Complete adverse event listing | Off-label content, unless the whole poster is compliant for it |
| Full baseline table and reference list | Promotional material of any kind |
| Plain-language summary | Data more recent than the approved poster |
| A downloadable copy of the poster | Anything that collects personal data without a compliant privacy notice |
Two rules for QR codes
1. Whatever it links to is congress material. Same review, same approval, same compliance rules as the board. A landing page assembled the week before by someone outside the process is an unapproved material.
2. Disclosures do not go behind it. Funding, conflicts of interest and medical writing support must be visible on the poster itself. Placing them behind a code does not satisfy transparency requirements, and it is a finding.
2. Disclosures do not go behind it. Funding, conflicts of interest and medical writing support must be visible on the poster itself. Placing them behind a code does not satisfy transparency requirements, and it is a finding.
Drills
Drill 1 of 3
Your poster body text is 620 words. What does that mean?
AIt is roughly double the workable budget; cut to under 300 and link the rest
BIt is thorough and complete
CIt is fine if the type is small enough
DRequest a larger board
Drill 2 of 3
Where do funding and conflict-of-interest statements belong?
ABehind the QR code, to save board space
BVisible on the poster, at the foot of the board
COnly in the abstract
DThey are optional for investigator-initiated studies
Drill 3 of 3
A colleague builds a landing page for the QR code the week before the congress. What must happen?
ANothing — it is a web page, not congress material
BAdd a disclaimer that it is not approved
CIt goes through the same review and approval as the poster
DRestrict it to healthcare professionals only
Production drill
Write the poster copy for all four panels. Under 300 words total.
Same study. Second-line advanced disease; 402 randomised; 389 treated; median follow-up 19.8 months.
PFS 8.9 vs 5.2 months; HR 0.66 (0.53–0.82); p<0.001. OS immature: NR vs 21.7; HR 0.84 (0.62–1.14). ORR 44% vs 29% (nominal). Grade ≥3 in 147/389 (37.8%) vs 121/388 (31.2%). Biomarker interaction p=0.03.
The gap: no randomised evaluation of this class in second line.
PFS 8.9 vs 5.2 months; HR 0.66 (0.53–0.82); p<0.001. OS immature: NR vs 21.7; HR 0.84 (0.62–1.14). ORR 44% vs 29% (nominal). Grade ≥3 in 147/389 (37.8%) vs 121/388 (31.2%). Biomarker interaction p=0.03.
The gap: no randomised evaluation of this class in second line.
Model poster copy — 218 words
BACKGROUND
· Outcomes after first-line failure in advanced disease remain poor
· No randomised trial has evaluated this class in the second-line setting
METHODS
· Randomised, double-blind, placebo-controlled Phase III
· 402 patients, second-line advanced disease, assigned 1:1
· Stratified by region and prior lines of therapy
· Primary endpoint: progression-free survival (PFS)
· Secondary endpoints tested in a prespecified hierarchy
RESULTS
· At median follow-up of 19.8 months, PFS was significantly longer with the study regimen: median 8.9 vs 5.2 months (HR 0.66, 95% CI 0.53–0.82; p<0.001) [Figure 1]
· Overall survival was immature: not reached vs 21.7 months (HR 0.84, 95% CI 0.62–1.14). Formal testing stopped at this endpoint
· Objective response rate was 44% vs 29% (nominal p=0.003)
· A significant interaction was observed by biomarker status (interaction p=0.03) [Figure 2]
· Grade ≥3 adverse events occurred in 147/389 (37.8%) vs 121/388 (31.2%) [Table 1]
CONCLUSIONS
· The study regimen significantly prolonged PFS vs placebo in previously treated advanced disease
· OS data were immature; follow-up continues
· The biomarker interaction was one of six prespecified subgroup analyses and requires prospective confirmation
Five things this copy does:
1. Under 300 words, with room for the figures to dominate.
2. Bullets throughout, none over about 25 words.
3. Every figure and table cross-referenced from the text.
4. The hierarchy break is stated in eight words, and ORR marked nominal.
5. The biomarker finding is hedged on the board itself — not left for the conversation. A poster is read when you are not standing beside it.
Behind the QR: full methods and eligibility, complete AE listing, full baseline table, all references, plain-language summary, downloadable poster. Disclosures stay on the board.
BACKGROUND
· Outcomes after first-line failure in advanced disease remain poor
· No randomised trial has evaluated this class in the second-line setting
METHODS
· Randomised, double-blind, placebo-controlled Phase III
· 402 patients, second-line advanced disease, assigned 1:1
· Stratified by region and prior lines of therapy
· Primary endpoint: progression-free survival (PFS)
· Secondary endpoints tested in a prespecified hierarchy
RESULTS
· At median follow-up of 19.8 months, PFS was significantly longer with the study regimen: median 8.9 vs 5.2 months (HR 0.66, 95% CI 0.53–0.82; p<0.001) [Figure 1]
· Overall survival was immature: not reached vs 21.7 months (HR 0.84, 95% CI 0.62–1.14). Formal testing stopped at this endpoint
· Objective response rate was 44% vs 29% (nominal p=0.003)
· A significant interaction was observed by biomarker status (interaction p=0.03) [Figure 2]
· Grade ≥3 adverse events occurred in 147/389 (37.8%) vs 121/388 (31.2%) [Table 1]
CONCLUSIONS
· The study regimen significantly prolonged PFS vs placebo in previously treated advanced disease
· OS data were immature; follow-up continues
· The biomarker interaction was one of six prespecified subgroup analyses and requires prospective confirmation
Five things this copy does:
1. Under 300 words, with room for the figures to dominate.
2. Bullets throughout, none over about 25 words.
3. Every figure and table cross-referenced from the text.
4. The hierarchy break is stated in eight words, and ORR marked nominal.
5. The biomarker finding is hedged on the board itself — not left for the conversation. A poster is read when you are not standing beside it.
Behind the QR: full methods and eligibility, complete AE listing, full baseline table, all references, plain-language summary, downloadable poster. Disclosures stay on the board.
Reflection
Count the words on the last poster you produced or reviewed. Over or under 300?
Next module
2.4 — The poster clinic
Week 2 · Module 4
The poster clinic
A poster brief with sixteen planted problems, spanning layout, data display, copy and compliance
How to do this
Timer at 40 minutes. Produce a revised layout plan and a list of findings with corrections. Do not read the rubric first.
The poster as drafted
Find every problem.
Board: A0 portrait.
Title (28 pt): "A randomised, double-blind, placebo-controlled, multicentre Phase III study evaluating the efficacy and safety of the study regimen compared with placebo in adult patients with previously treated advanced disease who have received one or two prior lines of systemic therapy"
Top left (40% of board): Background, four paragraphs, 380 words, 20 pt.
Top right: Baseline characteristics, 18 rows, 16 pt.
Centre: Study design schematic.
Bottom left: Adverse event table, 22 preferred terms, 14 pt.
Bottom right (small): PFS Kaplan–Meier, copied from the manuscript at original size, 9 pt axis labels, legend box, black solid and black dashed curves.
Bottom edge: "Conclusions: this regimen is highly effective and well tolerated and should be considered a new standard of care." 18 pt.
QR code: links to a landing page containing the full AE listing, a plain-language summary, and an updated data cut from last week. Page built by the brand team, not reviewed.
Not on the board: funding statement, conflict-of-interest disclosures, medical writing acknowledgement.
Title (28 pt): "A randomised, double-blind, placebo-controlled, multicentre Phase III study evaluating the efficacy and safety of the study regimen compared with placebo in adult patients with previously treated advanced disease who have received one or two prior lines of systemic therapy"
Top left (40% of board): Background, four paragraphs, 380 words, 20 pt.
Top right: Baseline characteristics, 18 rows, 16 pt.
Centre: Study design schematic.
Bottom left: Adverse event table, 22 preferred terms, 14 pt.
Bottom right (small): PFS Kaplan–Meier, copied from the manuscript at original size, 9 pt axis labels, legend box, black solid and black dashed curves.
Bottom edge: "Conclusions: this regimen is highly effective and well tolerated and should be considered a new standard of care." 18 pt.
QR code: links to a landing page containing the full AE listing, a plain-language summary, and an updated data cut from last week. Page built by the brand team, not reviewed.
Not on the board: funding statement, conflict-of-interest disclosures, medical writing acknowledgement.
The sixteen problems
Layout and hierarchy (5)
1. Title is 42 words at 28 pt. Should be short and 72–100 pt, readable at 3 m.
2. Background occupies 40% of the board in prime top-left space. Should be a small panel, two bullets.
3. The primary result is bottom-right and small. It must be the centre and the largest element.
4. The design schematic holds the centre, which is prime space spent on methods.
5. Conclusions at 18 pt on the bottom edge, below readable size and in the least-seen position.
Data display (4)
6. KM figure copied at manuscript size with 9 pt labels — must be redrawn at 20–24 pt with thicker curves.
7. Black solid vs black dashed — indistinguishable at distance; use contrasting colours with direct labelling.
8. Legend box instead of labels on the curves.
9. 22-row AE table at 14 pt — unreadable; cut to 6–8 plus a summary row, full table to QR.
Copy (3)
10. 380-word background, four paragraphs, against a 30–50 word budget in bullets.
11. 18-row baseline table on the board — cut to 3–5 key rows.
12. No one-line statement under any figure saying what it shows.
Compliance (4)
13. Conclusion overclaims: highly effective, well tolerated, and a standard-of-care recommendation from a single trial. Also no mention that OS was immature.
14. Disclosures, funding and writing acknowledgement absent from the board. These must be visible on the poster.
15. QR landing page not reviewed. It is congress material and requires the same approval.
16. The QR page carries an updated data cut not in the approved poster — unapproved data presented at a congress, and inconsistent with the board and the abstract.
Scoring
13–16 — pass. You would catch this before it printed.
8–12 — close. Note whether your misses cluster in design or compliance; most people are strong in one and weak in the other.
Under 8 — redo Modules 2.1 to 2.3.
The two most serious: problems 15 and 16. Everything else produces a poster nobody reads, which is a waste. An unreviewed landing page carrying unapproved data, reachable from a congress board, is a compliance breach that outlives the meeting — the code stays scannable long after the poster comes down.
Layout and hierarchy (5)
1. Title is 42 words at 28 pt. Should be short and 72–100 pt, readable at 3 m.
2. Background occupies 40% of the board in prime top-left space. Should be a small panel, two bullets.
3. The primary result is bottom-right and small. It must be the centre and the largest element.
4. The design schematic holds the centre, which is prime space spent on methods.
5. Conclusions at 18 pt on the bottom edge, below readable size and in the least-seen position.
Data display (4)
6. KM figure copied at manuscript size with 9 pt labels — must be redrawn at 20–24 pt with thicker curves.
7. Black solid vs black dashed — indistinguishable at distance; use contrasting colours with direct labelling.
8. Legend box instead of labels on the curves.
9. 22-row AE table at 14 pt — unreadable; cut to 6–8 plus a summary row, full table to QR.
Copy (3)
10. 380-word background, four paragraphs, against a 30–50 word budget in bullets.
11. 18-row baseline table on the board — cut to 3–5 key rows.
12. No one-line statement under any figure saying what it shows.
Compliance (4)
13. Conclusion overclaims: highly effective, well tolerated, and a standard-of-care recommendation from a single trial. Also no mention that OS was immature.
14. Disclosures, funding and writing acknowledgement absent from the board. These must be visible on the poster.
15. QR landing page not reviewed. It is congress material and requires the same approval.
16. The QR page carries an updated data cut not in the approved poster — unapproved data presented at a congress, and inconsistent with the board and the abstract.
Scoring
13–16 — pass. You would catch this before it printed.
8–12 — close. Note whether your misses cluster in design or compliance; most people are strong in one and weak in the other.
Under 8 — redo Modules 2.1 to 2.3.
The two most serious: problems 15 and 16. Everything else produces a poster nobody reads, which is a waste. An unreviewed landing page carrying unapproved data, reachable from a congress board, is a compliance breach that outlives the meeting — the code stays scannable long after the poster comes down.
Reflection
Did your findings cluster in design or in compliance? Whichever you missed is your development area.
Next week
3.1 — Slide architecture
Week 3 · Module 1
Slide architecture
Ten minutes, twelve slides, and an audience that cannot re-read anything. The narrative arc has to carry them
Your scenario
Thirty-two slides for a ten-minute slot. You reach the primary endpoint at minute eight, rush it, and the chair cuts you off before the conclusion. The one thing the audience came for is the one thing they did not get.
The arc
Figure 3.1 — The slide budget. Roughly one slide per minute, weighted toward the primary result. The commonest failure is front-loading: five background slides, then a rushed result and no conclusion. Backup slides sit after the end and are where the full tables live.
Rehearse with a timer, out loud
Reading slides silently takes about half as long as presenting them. A deck that feels like eight minutes on screen runs to thirteen at the podium.
Rehearse once, aloud, timed. It is the single highest-return thing you can do, and almost nobody does it.
Rehearse once, aloud, timed. It is the single highest-return thing you can do, and almost nobody does it.
Drills
Drill 1 of 3
How many slides for a 10-minute slot including questions?
AAround 30
BAround 4–5
CAs many as the data need
DAround 11–13, plus backup slides
Drill 2 of 3
You are at minute eight of ten and have four slides left including the conclusion. What do you do?
ASummarise the middle two in a sentence each and give the conclusion its full time
BSpeed up and cover all four
CSkip the conclusion and take questions instead
DAsk the chair for two more minutes
Drill 3 of 3
Where does the conflict-of-interest disclosure slide go?
AAt the end, after the conclusion
BSecond, immediately after the title
CIn the backup slides
DIt is optional if the study was investigator-initiated
Production drill
Build the slide plan for a 10-minute oral. Give each slide a title that states its message.
Same study. Second-line advanced disease, 402 randomised. PFS 8.9 vs 5.2 months, HR 0.66 (0.53–0.82), p<0.001. OS immature: NR vs 21.7, HR 0.84 (0.62–1.14). ORR 44% vs 29% (nominal). Grade ≥3 37.8% vs 31.2%. Biomarker interaction p=0.03, one of six prespecified. Gap: no randomised evaluation in second line.
Twelve slides, ten minutes
1. Title, authors, congress ID — 20 sec
2. Disclosures — 20 sec
3. “No randomised trial has evaluated this class in second-line disease” — the gap, one slide — 1 min
4. “Randomised, double-blind, placebo-controlled Phase III in 402 patients” — design schematic — 1 min
5. “Endpoints were tested in a prespecified hierarchy” — hierarchy diagram — 30 sec
6. “Baseline characteristics were balanced between arms” — 5 rows only — 30 sec
7. “PFS was significantly prolonged: 8.9 vs 5.2 months, HR 0.66” — KM curve — 1.5 min
8. “OS data were immature; formal testing stopped at this endpoint” — 1 min
9. “Response rate was 44% vs 29% (nominal)” — 30 sec
10. “Grade ≥3 events: 37.8% vs 31.2%” — top 6 preferred terms — 1 min
11. “A biomarker interaction was observed, but requires prospective confirmation” — forest plot — 1 min
12. Conclusions — three lines — 1.5 min
Backup: full baseline table, all six subgroups in detail, sensitivity analyses, full AE listing by system organ class, PH diagnostics, prior trials in the class.
Three things about these titles:
1. Every title states a message, not a topic. “PFS was significantly prolonged…” not “Progression-free survival”. The audience gets the point from the screen while you are still speaking.
2. Slide 5 exists because without the hierarchy on screen, slides 8 and 9 are not interpretable — and it costs 30 seconds.
3. Slide 11 hedges in its own title. If you run out of time and only the title is read, the hedge still lands.
Where the cuts come from if you run over: slides 6 and 9 first, then slide 11 reduced to a sentence. Slides 7, 8 and 12 are never cut.
1. Title, authors, congress ID — 20 sec
2. Disclosures — 20 sec
3. “No randomised trial has evaluated this class in second-line disease” — the gap, one slide — 1 min
4. “Randomised, double-blind, placebo-controlled Phase III in 402 patients” — design schematic — 1 min
5. “Endpoints were tested in a prespecified hierarchy” — hierarchy diagram — 30 sec
6. “Baseline characteristics were balanced between arms” — 5 rows only — 30 sec
7. “PFS was significantly prolonged: 8.9 vs 5.2 months, HR 0.66” — KM curve — 1.5 min
8. “OS data were immature; formal testing stopped at this endpoint” — 1 min
9. “Response rate was 44% vs 29% (nominal)” — 30 sec
10. “Grade ≥3 events: 37.8% vs 31.2%” — top 6 preferred terms — 1 min
11. “A biomarker interaction was observed, but requires prospective confirmation” — forest plot — 1 min
12. Conclusions — three lines — 1.5 min
Backup: full baseline table, all six subgroups in detail, sensitivity analyses, full AE listing by system organ class, PH diagnostics, prior trials in the class.
Three things about these titles:
1. Every title states a message, not a topic. “PFS was significantly prolonged…” not “Progression-free survival”. The audience gets the point from the screen while you are still speaking.
2. Slide 5 exists because without the hierarchy on screen, slides 8 and 9 are not interpretable — and it costs 30 seconds.
3. Slide 11 hedges in its own title. If you run out of time and only the title is read, the hedge still lands.
Where the cuts come from if you run over: slides 6 and 9 first, then slide 11 reduced to a sentence. Slides 7, 8 and 12 are never cut.
Reflection
Take a deck you have made. Do the slide titles state messages or label topics?
Next module
3.2 — Data slides
Week 3 · Module 2
Data slides
One message per slide, stated in the title. Everything else on the slide exists to support that one message
Your scenario
The slide carries a fourteen-row table, and you say “I know this is busy, but if you look at the third row from the bottom…” Everyone is now reading rows one to thirteen instead of listening to you. Apologising for a slide is proof it should have been cut.
The rules
| Rule | Why |
|---|---|
| The title states the message | The audience reads the title before you finish your first sentence. Make it do work. |
| One message per slide | Two messages means one is lost; the audience cannot pause and re-read |
| Cut tables to what the message needs | Full table goes to backup, where it belongs for questions |
| Highlight the row or point you are discussing | Colour, weight or a box — do not ask the audience to find it |
| Type no smaller than 18–20 pt | Back of a congress hall, on a screen that may be dim |
| No apology | “I know this is busy” means the slide needed cutting |
Topic titles
Message titles
Progression-free survival
PFS was significantly prolonged: 8.9 vs 5.2 months
Overall survival
OS data were immature; testing stopped at this endpoint
Safety
Grade ≥3 events were modestly higher with the study regimen
Subgroup analyses
A biomarker interaction was observed but requires confirmation
Baseline characteristics
Baseline characteristics were balanced between arms
Drills
Drill 1 of 3
A slide needs to show both efficacy and safety. What do you do?
APut both on the slide with a split layout
BShow efficacy only and mention safety verbally
CUse two slides, each with its own message title
DPut safety in the backup slides
Drill 2 of 3
Which slide title is best?
A“Overall survival”
B“No OS benefit observed”
C“Encouraging OS trend”
D“OS was immature at this analysis; formal testing stopped here”
Drill 3 of 3
You find yourself about to say "I know this slide is busy". What does that mean?
AThe slide should have been cut down before the talk
BNothing — acknowledging it is polite
CThe data are inherently complex and nothing can be done
DYou should spend longer on it
Production drill
Turn this manuscript table into slide content. Write the title and specify what stays.
Table: adverse events, 22 preferred terms × 6 columns (any grade n and %, grade ≥3 n and %, for each of two arms).
Key facts: any-grade 95.4% vs 90.7%; grade ≥3 37.8% vs 31.2%. Largest differences in grade ≥3: neutropenia 12.1% vs 4.6%; fatigue 8.5% vs 6.4%; nausea 5.1% vs 4.9%. Discontinuation for AE 8.0% vs 5.7%. No treatment-related deaths.
Key facts: any-grade 95.4% vs 90.7%; grade ≥3 37.8% vs 31.2%. Largest differences in grade ≥3: neutropenia 12.1% vs 4.6%; fatigue 8.5% vs 6.4%; nausea 5.1% vs 4.9%. Discontinuation for AE 8.0% vs 5.7%. No treatment-related deaths.
Slide title: “Grade ≥3 events were modestly higher with the study regimen, driven by neutropenia”
On the slide — a 4-row summary table:
· Any-grade AE — 95.4% vs 90.7%
· Grade ≥3 AE — 37.8% vs 31.2% (highlighted)
· Discontinuation for AE — 8.0% vs 5.7%
· Treatment-related deaths — none
Plus a small bar pair for the three most divergent grade ≥3 events: neutropenia 12.1% vs 4.6%, fatigue 8.5% vs 6.4%, nausea 5.1% vs 4.9%.
To backup: the full 22-term table, unchanged, for questions.
Four decisions:
1. 22 terms → 4 summary rows plus 3 highlighted events. The message is the overall picture and its driver, not a listing.
2. The title names the driver. Neutropenia is the actual finding; “modestly higher” alone would be vague and slightly evasive.
3. Grade ≥3 row is highlighted because that is the row you will be speaking about. Do not make the audience find it.
4. Nausea is included even though the difference is trivial (5.1% vs 4.9%). Showing the three largest differences and having one of them be negligible is honest — showing only the two divergent ones implies a pattern that is not there.
The check: if someone photographs this slide and nothing else, do they have an accurate picture of the safety profile? Here, yes.
On the slide — a 4-row summary table:
· Any-grade AE — 95.4% vs 90.7%
· Grade ≥3 AE — 37.8% vs 31.2% (highlighted)
· Discontinuation for AE — 8.0% vs 5.7%
· Treatment-related deaths — none
Plus a small bar pair for the three most divergent grade ≥3 events: neutropenia 12.1% vs 4.6%, fatigue 8.5% vs 6.4%, nausea 5.1% vs 4.9%.
To backup: the full 22-term table, unchanged, for questions.
Four decisions:
1. 22 terms → 4 summary rows plus 3 highlighted events. The message is the overall picture and its driver, not a listing.
2. The title names the driver. Neutropenia is the actual finding; “modestly higher” alone would be vague and slightly evasive.
3. Grade ≥3 row is highlighted because that is the row you will be speaking about. Do not make the audience find it.
4. Nausea is included even though the difference is trivial (5.1% vs 4.9%). Showing the three largest differences and having one of them be negligible is honest — showing only the two divergent ones implies a pattern that is not there.
The check: if someone photographs this slide and nothing else, do they have an accurate picture of the safety profile? Here, yes.
Reflection
Find a busy data slide you have used. What was its one message? Could the slide be cut to just that?
Next module
3.3 — Delivery and questions
Week 3 · Module 3
Delivery and questions
Timing, the room, and the two minutes of questions where a good presentation is confirmed or undone
Your scenario
The talk went well. The first question asks about an analysis you did not perform, and you speculate about what it would probably have shown. The speculation is now the thing the room remembers — and it is the only part of the session that was not reviewed by anyone.
Delivery
| Practice | Why |
|---|---|
| Rehearse aloud, timed, once | Silent reading is about half real speed; almost nobody does this and it shows |
| Know your first and last sentences exactly | The two moments where nerves do most damage |
| Face the audience, not the screen | You know what is on the slide; they need your voice pointed at them |
| Point at the specific number you are discussing | Pointer or highlight — do not let them search |
| Pause after the primary result | Two seconds of silence is how the room registers the key finding |
| Have a cut plan | Decide before you start which two slides go if you are behind |
Questions
| Question type | Response |
|---|---|
| About an analysis you did not do | Say it was not performed and, if relevant, that it was not prespecified. Do not speculate. |
| About a limitation you already stated | Agree, restate it briefly, and say what would resolve it |
| Hostile or dismissive in tone | Answer the substance, ignore the tone entirely |
| Asking you to compare with another product | State that no head-to-head data exist, if that is the case |
| About off-label use | State the approved indication; do not extend into unapproved territory from a podium |
| You do not know the answer | Say so, and offer to follow up. This costs nothing and guessing costs a lot. |
The podium is not exempt
Everything that constrains written congress material constrains what you say at the microphone. Off-label claims, comparative claims without head-to-head data, and unapproved-product discussion are no more permissible spoken than printed — and the Q&A is the only unreviewed part of the session.
“That was not prespecified, so I would not want to speculate” is a complete, professional answer. Senior people use it constantly.
“That was not prespecified, so I would not want to speculate” is a complete, professional answer. Senior people use it constantly.
Drills
Drill 1 of 3
A questioner asks what the result would have been with a different censoring rule. You did not run it. What do you say?
AGive your best estimate of what it would have shown
BSay it was not performed or prespecified, and point to the sensitivity analyses that were
CSay the question is not relevant
DRefer it to a co-author in the audience
Drill 2 of 3
From the floor: "Would you use this instead of [competitor] in practice?"
A“Yes, based on these data I would.”
B“I cannot answer that.”
C“No head-to-head data exist; treatment selection depends on factors this trial did not address.”
DCompare the two trials’ hazard ratios
Drill 3 of 3
You do not know the answer to a factual question about the study. What is the best response?
AGive your best guess
BAnswer a different question
CSay the analysis team did not provide it
DSay you do not have it to hand and offer to follow up
Production drill
Write your answer to each question, as you would say it.
1. "Your OS curves look like they're separating late. Doesn't that violate proportional hazards?"
2. "The biomarker result is the interesting finding here. Shouldn't patients be tested before treatment?"
3. "This trial excluded ECOG 3 patients. Isn't that the population we actually see?"
4. "Frankly, a three-and-a-half month PFS gain isn't clinically meaningful."
5. "What was the median treatment duration in the placebo arm?" (You do not know.)
2. "The biomarker result is the interesting finding here. Shouldn't patients be tested before treatment?"
3. "This trial excluded ECOG 3 patients. Isn't that the population we actually see?"
4. "Frankly, a three-and-a-half month PFS gain isn't clinically meaningful."
5. "What was the median treatment duration in the placebo arm?" (You do not know.)
1. “That is a fair observation. We tested the proportional hazards assumption using Schoenfeld residuals and it was supported for the primary endpoint. OS data are immature, so the shape of those curves should be interpreted cautiously; we will reassess at the next analysis.” — Concedes the observation, gives the diagnostic, does not over-defend.
2. “The interaction was statistically significant, but it was one of six prespecified subgroup analyses in a trial powered for the overall population. I would call it hypothesis-generating. It needs prospective evaluation before it could support biomarker-directed treatment selection.” — The question invites you to recommend a practice change from a subgroup. Declining is the answer.
3. “Yes, and that is a genuine limitation. Eligibility was restricted to ECOG 0 to 2, so these results may not generalise to patients with poorer performance status. That population would need separate evaluation.” — Agree, do not defend. Conceding a real limitation is the strongest available position.
4. “Clinical meaningfulness is a reasonable question and it is not one this trial was designed to answer. What I can say is that the hazard ratio was 0.66 with a confidence interval from 0.53 to 0.82, and that this was the prespecified primary endpoint. Whether that magnitude is meaningful will depend on the setting and on the OS data as they mature.” — Answers the substance, ignores the tone, does not argue.
5. “I do not have that figure to hand. If you find me afterwards or send an email, I will get it to you.” — Complete answer. Ten words. Do not extend it.
What all five have in common: none speculates, none over-claims, and none argues with the questioner. Answers 3 and 4 in particular concede ground — which is what makes the rest of the presentation credible.
2. “The interaction was statistically significant, but it was one of six prespecified subgroup analyses in a trial powered for the overall population. I would call it hypothesis-generating. It needs prospective evaluation before it could support biomarker-directed treatment selection.” — The question invites you to recommend a practice change from a subgroup. Declining is the answer.
3. “Yes, and that is a genuine limitation. Eligibility was restricted to ECOG 0 to 2, so these results may not generalise to patients with poorer performance status. That population would need separate evaluation.” — Agree, do not defend. Conceding a real limitation is the strongest available position.
4. “Clinical meaningfulness is a reasonable question and it is not one this trial was designed to answer. What I can say is that the hazard ratio was 0.66 with a confidence interval from 0.53 to 0.82, and that this was the prespecified primary endpoint. Whether that magnitude is meaningful will depend on the setting and on the OS data as they mature.” — Answers the substance, ignores the tone, does not argue.
5. “I do not have that figure to hand. If you find me afterwards or send an email, I will get it to you.” — Complete answer. Ten words. Do not extend it.
What all five have in common: none speculates, none over-claims, and none argues with the questioner. Answers 3 and 4 in particular concede ground — which is what makes the rest of the presentation credible.
Reflection
Which of the five questions would you find hardest to answer well? Rehearse that one.
Next module
3.4 — The presentation clinic
Week 3 · Module 4
The presentation clinic
A deck plan with fourteen planted problems across structure, slide design and what may be said aloud
How to do this
Timer at 40 minutes. Produce a revised slide plan and a findings list. Do not read the rubric first.
The deck as drafted
Find every problem. Slot is 10 minutes including 2 minutes of questions.
Slide 1: Title
Slides 2–7: Background — disease epidemiology, treatment history, six prior trials in the class, mechanism of action (2 slides)
Slide 8: "Methods" — full eligibility criteria, 24 bullet points, 14 pt
Slide 9: "Baseline characteristics" — full 18-row table
Slide 10: "Results" — PFS, OS, ORR and safety all on one slide
Slide 11: "Subgroup analyses" — forest plot with all 11 subgroups shown
Slide 12: "Conclusions: this regimen is highly effective and should be adopted as standard of care in biomarker-positive patients"
Slide 13: Disclosures
No backup slides.
Presenter's plan for Q&A: "If anyone asks about the competitor I'll point out our HR is better than theirs."
Slides 2–7: Background — disease epidemiology, treatment history, six prior trials in the class, mechanism of action (2 slides)
Slide 8: "Methods" — full eligibility criteria, 24 bullet points, 14 pt
Slide 9: "Baseline characteristics" — full 18-row table
Slide 10: "Results" — PFS, OS, ORR and safety all on one slide
Slide 11: "Subgroup analyses" — forest plot with all 11 subgroups shown
Slide 12: "Conclusions: this regimen is highly effective and should be adopted as standard of care in biomarker-positive patients"
Slide 13: Disclosures
No backup slides.
Presenter's plan for Q&A: "If anyone asks about the competitor I'll point out our HR is better than theirs."
The fourteen problems
Structure and timing (5)
1. Six background slides of thirteen — nearly half the deck before any data. Should be one, maybe two.
2. The primary result gets a fraction of one slide. It should have its own slide and the most time.
3. Thirteen slides for an 8-minute talk (10 minus 2 for questions) is already tight, and the front-loading makes it unworkable.
4. No backup slides. Every question will be answered from memory or not at all.
5. No cut plan. When the talk runs long, the conclusion is what gets lost.
Slide design (5)
6. Slide 8: 24 bullets at 14 pt. Unreadable from the room. Key criteria only; full list to backup.
7. Slide 9: full 18-row baseline table. Cut to 5 rows.
8. Slide 10 carries four messages. Should be four slides.
9. All slide titles are topic labels — “Methods”, “Results”, “Subgroup analyses” — not messages.
10. Slide 11 shows 11 subgroups where six were prespecified. Show the six; the rest are exploratory.
Compliance and content (4)
11. Disclosures are slide 13, after all the data. Must be slide 2, before any data.
12. The conclusion overclaims twice: “highly effective” is a promotional characterisation, and “should be adopted as standard of care in biomarker-positive patients” is a practice recommendation built on one interaction test in a trial not powered for it.
13. The conclusion omits OS immaturity entirely.
14. The Q&A plan is a cross-trial comparison. Comparing your HR to a competitor's from a separate trial is scientifically invalid and, from a podium, a comparative claim with no head-to-head basis. This must not happen.
Revised plan: the twelve-slide structure from Module 3.1 — title, disclosures, gap, design, hierarchy, baseline (5 rows), PFS, OS, ORR, safety, subgroups (six, hedged), conclusions — with backup slides for the full tables, and a cut plan naming slides 6 and 9 as the first to go.
Scoring
11–14 — pass.
7–10 — close; note whether you missed design or compliance.
Under 7 — redo Modules 3.1 to 3.3.
The one that would do most damage: problem 14. Everything else produces a poor presentation. Comparing hazard ratios across trials from a podium, in a room that may include press and competitors, is a claim the organisation cannot retract and did not approve.
Structure and timing (5)
1. Six background slides of thirteen — nearly half the deck before any data. Should be one, maybe two.
2. The primary result gets a fraction of one slide. It should have its own slide and the most time.
3. Thirteen slides for an 8-minute talk (10 minus 2 for questions) is already tight, and the front-loading makes it unworkable.
4. No backup slides. Every question will be answered from memory or not at all.
5. No cut plan. When the talk runs long, the conclusion is what gets lost.
Slide design (5)
6. Slide 8: 24 bullets at 14 pt. Unreadable from the room. Key criteria only; full list to backup.
7. Slide 9: full 18-row baseline table. Cut to 5 rows.
8. Slide 10 carries four messages. Should be four slides.
9. All slide titles are topic labels — “Methods”, “Results”, “Subgroup analyses” — not messages.
10. Slide 11 shows 11 subgroups where six were prespecified. Show the six; the rest are exploratory.
Compliance and content (4)
11. Disclosures are slide 13, after all the data. Must be slide 2, before any data.
12. The conclusion overclaims twice: “highly effective” is a promotional characterisation, and “should be adopted as standard of care in biomarker-positive patients” is a practice recommendation built on one interaction test in a trial not powered for it.
13. The conclusion omits OS immaturity entirely.
14. The Q&A plan is a cross-trial comparison. Comparing your HR to a competitor's from a separate trial is scientifically invalid and, from a podium, a comparative claim with no head-to-head basis. This must not happen.
Revised plan: the twelve-slide structure from Module 3.1 — title, disclosures, gap, design, hierarchy, baseline (5 rows), PFS, OS, ORR, safety, subgroups (six, hedged), conclusions — with backup slides for the full tables, and a cut plan naming slides 6 and 9 as the first to go.
Scoring
11–14 — pass.
7–10 — close; note whether you missed design or compliance.
Under 7 — redo Modules 3.1 to 3.3.
The one that would do most damage: problem 14. Everything else produces a poor presentation. Comparing hazard ratios across trials from a podium, in a room that may include press and competitors, is a claim the organisation cannot retract and did not approve.
Reflection
Did you catch the Q&A plan? It is the only problem not visible in the deck itself.
Next week
4.1 — Embargo, encore and disclosure
Week 4 · Module 1
Embargo, encore and disclosure
Three rules with real consequences: withdrawal from the programme, rejection at submission, and a compliance finding
Your scenario
An author's university press office issues a release about the findings four days before the embargo lifts. They were never told there was an embargo. Nobody put them on the distribution list, because nobody thought of them as part of the project.
Embargo
| Rule | Detail |
|---|---|
| What is embargoed | Abstract content, and usually presentation content, until a stated date and time |
| Who must know | Everyone — all authors, their institutional press offices, internal teams, agency partners |
| What counts as breach | Social media, press release, company website, public presentation, an open meeting |
| Consequence | Withdrawal from the programme; occasionally sanctions on future submissions |
| If a breach occurs | Report it to the congress immediately. Self-reporting is treated far more favourably than discovery. |
The press office problem
The commonest embargo breach is not a company employee. It is an author’s institutional press office, operating on its own calendar, publicising its researcher’s work.
They are almost never on the project distribution list, because they are not part of the project. Write to them by name, with the exact embargo date and time, as a specific task on the timeline.
They are almost never on the project distribution list, because they are not part of the project. Write to them by name, with the exact embargo date and time, as a specific task on the timeline.
Encore
| Situation | What is required |
|---|---|
| Data presented at a prior meeting | Check the congress encore policy before drafting |
| Encore permitted | Disclose the prior presentation at submission, usually in a designated field |
| Encore prohibited | Do not submit; consider a different analysis or a different congress |
| Substantially updated data | May not be an encore at all — but check rather than assume |
| Undisclosed encore discovered | Rejection, or withdrawal after acceptance |
Disclosure
| Element | Where |
|---|---|
| Conflicts of interest for all presenters | Slide 2 of an oral; a section of the poster |
| Funding source | On the poster and in the abstract |
| Medical writing support | Named, with the funder, on the material itself |
| Trial registration number | Abstract and material |
| Prior presentation, if an encore | Submission form, and often on the material |
Drills
Drill 1 of 3
Who is the commonest source of an embargo breach?
AAn author’s institutional press office
BThe company communications team
CThe congress organisers
DJournalists
Drill 2 of 3
An embargo breach has occurred. What is the first action?
ADelete the content and hope it was not noticed
BReport it to the congress immediately
CWait and see whether it is raised
DWithdraw the abstract yourself
Drill 3 of 3
Where must medical writing support be acknowledged?
ABehind the QR code
BOnly in the manuscript, not at congress
COn the material itself, naming the writer and the funder
DIt is optional if the writer is not an author
Policy Compliance Global Congress Encore & Embargo Rules
Encore presentations (re-presenting data at a regional or national meeting after a major international congress) must strictly adhere to GPP 2022 and specific congress bylaws:
| Congress Entity | Encore Policy Stance | Mandatory Writer / Presenter Action |
|---|---|---|
| Primary Congress (e.g. ASCO, AHA) | Requires 100% original, unpublished, non-presented data at time of submission. | Strictly honor official embargo lift time (usually 00:01 on first day of meeting or time of oral presentation). |
| Regional / National Congress | Many permit encore presentations if explicitly declared and translated into local language. | Include explicit footnote on poster/slides: "Originally presented at [Congress Name], [City, Year]. Re-presented with permission." |
| Target Peer-Reviewed Journal | Congress abstracts/posters do NOT constitute prior publication under ICMJE guidelines. | Disclose previous congress presentation in journal cover letter and manuscript title page footnote with DOI. |
Production drill
Write the embargo briefing note. Who receives it, and what does it say?
Congress: embargo lifts 10 September, 00:01 CEST. Oral presentation 12 September, 14:20.
People involved: 9 authors across 6 institutions; sponsor medical, communications and commercial teams; a medical communications agency; two authors' universities have active press offices; one author is a prolific social media user.
People involved: 9 authors across 6 institutions; sponsor medical, communications and commercial teams; a medical communications agency; two authors' universities have active press offices; one author is a prolific social media user.
Distribution list: all 9 authors individually (not a group alias) · named press office contacts at all 6 institutions, obtained deliberately · sponsor medical, communications and commercial leads · the agency team · anyone internally who has seen the data.
The note:
“Subject: EMBARGO — [study] data are confidential until 10 September, 00:01 CEST
The abstract has been accepted for oral presentation at [congress] on 12 September at 14:20. The congress embargo lifts on 10 September at 00:01 CEST.
Until that moment, no part of the abstract or presentation may be disclosed publicly. This includes social media posts, institutional or company press releases, website updates, newsletters, and presentations at any meeting open to people outside the study team.
Breaching the embargo can result in the abstract being withdrawn from the programme.
If you work with a press office or communications team, please forward this note to them by name today and confirm to me that you have done so. Institutional press offices operate on their own schedules and are the most common source of accidental breach.
If a breach occurs, or you think one may have occurred, tell me immediately — do not delete anything and do not wait. Self-reported breaches are treated far more leniently than discovered ones.
Materials for use from 10 September will be circulated on 8 September.”
Four things the note does:
1. Exact date, time and time zone in the subject line and the first paragraph. “Early September” is how breaches happen.
2. Names the specific channels, because people do not think of a newsletter or an internal town hall as public.
3. Asks for confirmation of onward forwarding to press offices — converting a hope into a tracked action.
4. Makes self-reporting safe. If the person who breaches is afraid to say so, you find out from the congress instead.
The prolific social media user: do not single them out in writing. Have a short direct conversation instead.
The note:
“Subject: EMBARGO — [study] data are confidential until 10 September, 00:01 CEST
The abstract has been accepted for oral presentation at [congress] on 12 September at 14:20. The congress embargo lifts on 10 September at 00:01 CEST.
Until that moment, no part of the abstract or presentation may be disclosed publicly. This includes social media posts, institutional or company press releases, website updates, newsletters, and presentations at any meeting open to people outside the study team.
Breaching the embargo can result in the abstract being withdrawn from the programme.
If you work with a press office or communications team, please forward this note to them by name today and confirm to me that you have done so. Institutional press offices operate on their own schedules and are the most common source of accidental breach.
If a breach occurs, or you think one may have occurred, tell me immediately — do not delete anything and do not wait. Self-reported breaches are treated far more leniently than discovered ones.
Materials for use from 10 September will be circulated on 8 September.”
Four things the note does:
1. Exact date, time and time zone in the subject line and the first paragraph. “Early September” is how breaches happen.
2. Names the specific channels, because people do not think of a newsletter or an internal town hall as public.
3. Asks for confirmation of onward forwarding to press offices — converting a hope into a tracked action.
4. Makes self-reporting safe. If the person who breaches is afraid to say so, you find out from the congress instead.
The prolific social media user: do not single them out in writing. Have a short direct conversation instead.
Reflection
On your last congress project, was anyone's institutional press office on the distribution list?
Next module
4.2 — Authorship and review
Week 4 · Module 2
Authorship and review
The same contribution criteria as a manuscript, on a calendar that runs backwards from a deadline that cannot move
Your scenario
Two days before the abstract deadline, a senior investigator asks to be added as an author and a second asks for the author order to change. Neither conversation happened in the previous three months, and the deadline does not move for either of them.
Congress authorship
| Principle | Detail |
|---|---|
| Same criteria as publication | Substantial contribution, drafting or critical revision, approval, accountability |
| Agreed before drafting | The list and the order, in writing, at project start |
| The presenting author | Must be able to attend and to answer questions on the data |
| Writing support | Acknowledged and funded-disclosed, never an author unless criteria are met |
| Consistency with the manuscript | Congress and publication author lists should be consistent, or the difference explained |
The presenting author is a real commitment
Presenting requires attendance, familiarity with the analysis, and the ability to handle questions on methodology. Confirm availability at submission, not at acceptance — and confirm again once travel is booked.
A presenting author who withdraws two weeks out, with the poster already printed carrying their name, is a common and entirely avoidable emergency.
A presenting author who withdraws two weeks out, with the poster already printed carrying their name, is a common and entirely avoidable emergency.
The review calendar
| Stage | Typical duration | Note |
|---|---|---|
| Draft preparation | 1–2 weeks | Longer if the data are not final |
| Author review, round 1 | 1 week | Nine authors will not all respond in three days |
| Author review, round 2 | 1 week | Assume two rounds; one is optimistic |
| Medical review | 2 weeks | The longest single stage in most organisations |
| Legal and regulatory review | 1 week | Runs after medical, not in parallel, in most systems |
| Final approval and submission | 2–3 days | Buffer for portal problems on deadline day |
| Total | 6–8 weeks | Before the deadline, not after acceptance |
Drills
Drill 1 of 3
Two days before the deadline, a senior investigator asks to be added as an author. What is the correct response?
AAdd them — seniority warrants it
BRefuse — it is too late
CAdd them to the manuscript instead
DAssess against the contribution criteria and act accordingly
Drill 2 of 3
When should the first draft of a congress abstract exist?
ASix to eight weeks before the deadline
BOne week before the deadline
CAfter acceptance is confirmed
DThe day before submission
Drill 3 of 3
The presenting author withdraws two weeks before the congress. The poster is printed. What matters most?
AFind anyone available to stand at the board
BA replacement from the author list who can answer questions on the data, notified to the congress
CWithdraw the poster
DDisplay it unattended
Production drill
Resolve the authorship situation and write the note you would send.
Three days before the deadline.
· Dr A — designed the study, will present. First author, agreed at project start.
· Dr B — enrolled the most patients, reviewed drafts. On the list.
· Mr C — performed all analyses, wrote the statistical text. On the list.
· Prof D — head of department, secured the grant, has not seen a draft. Now asks to be added as last author.
· Dr E — on the list, contributed to interpretation. Now asks to move from fifth to second.
· Dr A — designed the study, will present. First author, agreed at project start.
· Dr B — enrolled the most patients, reviewed drafts. On the list.
· Mr C — performed all analyses, wrote the statistical text. On the list.
· Prof D — head of department, secured the grant, has not seen a draft. Now asks to be added as last author.
· Dr E — on the list, contributed to interpretation. Now asks to move from fifth to second.
Resolution
Prof D — acknowledgement, not authorship. Securing funding and heading a department are explicitly acknowledgement-level contributions, and not having seen a draft fails the drafting, approval and accountability criteria. Offer a named acknowledgement.
Dr E — not a decision for you. Author order is agreed among the authors, not adjudicated by the writer. Refer it to Dr A as first author and corresponding author, flag the deadline, and hold the current order unless they agree otherwise.
Note to Dr A:
“Two authorship items ahead of Friday’s deadline.
Prof D has asked to be added as last author. On the contribution criteria we have been applying — the same as for the manuscript — securing funding and departmental leadership are acknowledgement-level rather than authorship contributions, and Prof D has not reviewed a draft. My recommendation is a named acknowledgement for funding and departmental support. That is your call rather than mine, but it needs settling before Friday, and it will need settling again for the manuscript.
Dr E has asked to move from fifth to second author. Author order is for the author group to agree, so I have not changed anything. Could you decide with Dr E and let me know by Thursday morning? If there is no agreement, I will submit in the order agreed at project start.
The abstract is otherwise through review and ready.”
Three things the note does:
1. Applies a stated external standard rather than the writer’s judgement. This is what makes the Prof D conversation survivable.
2. Routes the decision to the first author, which is where it belongs. A writer who adjudicates author order acquires a problem that is not theirs.
3. Sets a hard internal deadline a day before the real one, with a stated default if nobody decides. Without the default, this drifts to Friday afternoon.
The underlying lesson: both requests are late because the author list was agreed at project start but never re-confirmed. A one-line email to all authors six weeks out — “the author list and order for this submission is as follows; please confirm” — surfaces both of these while there is still time.
Prof D — acknowledgement, not authorship. Securing funding and heading a department are explicitly acknowledgement-level contributions, and not having seen a draft fails the drafting, approval and accountability criteria. Offer a named acknowledgement.
Dr E — not a decision for you. Author order is agreed among the authors, not adjudicated by the writer. Refer it to Dr A as first author and corresponding author, flag the deadline, and hold the current order unless they agree otherwise.
Note to Dr A:
“Two authorship items ahead of Friday’s deadline.
Prof D has asked to be added as last author. On the contribution criteria we have been applying — the same as for the manuscript — securing funding and departmental leadership are acknowledgement-level rather than authorship contributions, and Prof D has not reviewed a draft. My recommendation is a named acknowledgement for funding and departmental support. That is your call rather than mine, but it needs settling before Friday, and it will need settling again for the manuscript.
Dr E has asked to move from fifth to second author. Author order is for the author group to agree, so I have not changed anything. Could you decide with Dr E and let me know by Thursday morning? If there is no agreement, I will submit in the order agreed at project start.
The abstract is otherwise through review and ready.”
Three things the note does:
1. Applies a stated external standard rather than the writer’s judgement. This is what makes the Prof D conversation survivable.
2. Routes the decision to the first author, which is where it belongs. A writer who adjudicates author order acquires a problem that is not theirs.
3. Sets a hard internal deadline a day before the real one, with a stated default if nobody decides. Without the default, this drifts to Friday afternoon.
The underlying lesson: both requests are late because the author list was agreed at project start but never re-confirmed. A one-line email to all authors six weeks out — “the author list and order for this submission is as follows; please confirm” — surfaces both of these while there is still time.
Reflection
On your last congress submission, was the author list confirmed in writing before drafting?
Next module
4.3 — After the congress
Week 4 · Module 3
After the congress
Reuse, plain-language summaries and the relationship between the abstract and the manuscript that follows it
Your scenario
The poster is emailed to a mailing list of clinicians the week after the congress. Presenting it publicly did not make it a promotional material — but sending it unsolicited to a list does exactly that, and it now needs an approval it never had.
Reuse
| Use | What governs it |
|---|---|
| Sent on request to an attendee | Usually acceptable; follow the medical information process |
| Emailed unsolicited to a list | Proactive dissemination — requires promotional approval, and off-label content is prohibited |
| Posted on a company website | Separate approval; often region-specific restrictions |
| Reused in a field slide deck | Requires its own review; on-label content only |
| Posted on social media by an author | Permitted after embargo, but institutional and sponsor policies apply |
| Included in a manuscript | Congress presentation is not prior publication; disclose it in the cover letter |
Abstract and manuscript
Presenting at a congress does not normally count as prior publication and does not prevent journal submission — but you must disclose the presentation in the cover letter, and some journals have specific rules about how much detail may already be public.
Where the manuscript reports a later data cut than the abstract, say so explicitly. Reviewers do check, and an unexplained difference between a published abstract and a submitted manuscript reads badly.
Where the manuscript reports a later data cut than the abstract, say so explicitly. Reviewers do check, and an unexplained difference between a published abstract and a submitted manuscript reads badly.
Plain-language summaries
| Feature | Requirement |
|---|---|
| Audience | Patients, carers and the public — not a simplified version for clinicians |
| Reading level | Typically aimed at around age 12–14 reading age |
| Content | Why the study was done, what was done, what was found, what it means, limitations |
| Tone | Non-promotional; no favourable characterisation; balanced on harms |
| Numbers | Natural frequencies (“about 4 in every 10 people”) rather than percentages and hazard ratios |
| Approval | Same review as any other congress material |
Drills
Drill 1 of 3
After the congress, the poster is emailed unsolicited to 400 clinicians. What is the problem?
ANone — it was presented publicly
BOnly a problem if more than 500 recipients
CUnsolicited distribution is proactive and requires separate promotional approval
DOnly a problem if a disclaimer is missing
Drill 2 of 3
Does presenting an abstract at a congress prevent journal publication?
AYes — it constitutes prior publication
BOnly if published within 12 months
COnly for oral presentations
DNo, but it must be disclosed in the cover letter
Drill 3 of 3
Which is the right way to express a hazard ratio of 0.66 in a plain-language summary?
A“On average, the time before the disease got worse was about 3.7 months longer for people taking the medicine. This was an average — some people did better and some did worse.”
B“The hazard ratio was 0.66 (95% CI 0.53–0.82).”
C“The medicine worked 34% better.”
DOmit the number entirely
Production drill
Write the plain-language summary. Around 200 words.
Same study. 402 people with advanced disease whose first treatment had stopped working. Half received the study medicine, half a placebo; neither they nor their doctors knew which.
Time before the disease got worse: 8.9 months vs 5.2 months.
Overall survival: too early to know.
About 4 in 10 people on the medicine had a serious side effect, vs about 3 in 10 on placebo. The commonest was low white blood cell counts.
About 8 in 100 stopped the medicine because of side effects, vs about 6 in 100 on placebo.
Time before the disease got worse: 8.9 months vs 5.2 months.
Overall survival: too early to know.
About 4 in 10 people on the medicine had a serious side effect, vs about 3 in 10 on placebo. The commonest was low white blood cell counts.
About 8 in 100 stopped the medicine because of side effects, vs about 6 in 100 on placebo.
Model plain-language summary — 214 words
Why was this study done?
When the first treatment for advanced disease stops working, there are few options left. Doctors wanted to know whether a new medicine could help people in this situation.
What did the study do?
402 people took part. Half were given the new medicine and half were given a placebo — a dummy treatment with no active ingredient. Who got which was decided at random, and neither the participants nor their doctors knew which one anybody was taking.
What was found?
People taking the medicine went longer before their disease got worse: about 8.9 months on average, compared with about 5.2 months for people taking the placebo. That is around 3.7 months longer. This is an average — some people did better than this and some did worse.
It is still too early to know whether the medicine helps people live longer. The study is continuing so that this can be answered.
What about side effects?
About 4 in every 10 people taking the medicine had a serious side effect, compared with about 3 in every 10 taking the placebo. The most common was a drop in white blood cells, which help fight infection. About 8 in every 100 people stopped the medicine because of side effects, compared with about 6 in every 100 on placebo.
Six things this does:
1. Questions as headings. Readers scan for the question they have.
2. Natural frequencies — 4 in every 10 — not percentages or hazard ratios.
3. Placebo and randomisation explained in the sentence where they appear.
4. “This is an average” — without it, a reader takes 3.7 months as a promise.
5. Survival honesty: too early to know, and the study continues. Not omitted, not spun.
6. Harms are given equal weight and equal clarity, with the commonest one explained rather than named.
What is absent: any adjective describing the medicine. No effective, no well tolerated, no promising. The reader is given the numbers and left to judge.
Why was this study done?
When the first treatment for advanced disease stops working, there are few options left. Doctors wanted to know whether a new medicine could help people in this situation.
What did the study do?
402 people took part. Half were given the new medicine and half were given a placebo — a dummy treatment with no active ingredient. Who got which was decided at random, and neither the participants nor their doctors knew which one anybody was taking.
What was found?
People taking the medicine went longer before their disease got worse: about 8.9 months on average, compared with about 5.2 months for people taking the placebo. That is around 3.7 months longer. This is an average — some people did better than this and some did worse.
It is still too early to know whether the medicine helps people live longer. The study is continuing so that this can be answered.
What about side effects?
About 4 in every 10 people taking the medicine had a serious side effect, compared with about 3 in every 10 taking the placebo. The most common was a drop in white blood cells, which help fight infection. About 8 in every 100 people stopped the medicine because of side effects, compared with about 6 in every 100 on placebo.
Six things this does:
1. Questions as headings. Readers scan for the question they have.
2. Natural frequencies — 4 in every 10 — not percentages or hazard ratios.
3. Placebo and randomisation explained in the sentence where they appear.
4. “This is an average” — without it, a reader takes 3.7 months as a promise.
5. Survival honesty: too early to know, and the study continues. Not omitted, not spun.
6. Harms are given equal weight and equal clarity, with the commonest one explained rather than named.
What is absent: any adjective describing the medicine. No effective, no well tolerated, no promising. The reader is given the numbers and left to judge.
Reflection
Try writing your own study's main finding as a natural frequency. Did it change how strong it sounded?
Next module
4.4 — Capstone and exit assessment
Week 4 · Module 4
Capstone and exit assessment
A full congress project with eighteen planted problems across timeline, materials and compliance
How to do this
Timer at 50 minutes. Produce a findings list with corrections and a revised project timeline. Pass mark 14 of 18. Do not read the rubric first.
The project
Review the whole congress project and find every problem.
Congress: abstract deadline 15 April; acceptance 20 June; embargo lifts 10 September 00:01; meeting 12–16 September.
Current date: 25 March.
Status: abstract not yet drafted. Data lock was 10 March.
Abstract plan: "We'll draft it the week before and get it signed off quickly — it's only 2,000 characters."
Data: the same study was presented at a national meeting in February. No one has checked the encore policy.
Authors: 9 across 6 institutions. List agreed 8 months ago, not re-confirmed. Presenting author not yet asked whether they can attend.
Draft conclusion: "This regimen is highly effective and should be considered standard of care, particularly in biomarker-positive patients."
Poster plan: "We'll adapt the manuscript figures directly to save time." QR code will link to a page the brand team is building, containing the full AE listing and a plain-language summary.
Communications plan: sponsor press release scheduled for 8 September. Two authors' universities have press offices; neither has been contacted.
Disclosures: "We'll add those at the end if there's room."
Current date: 25 March.
Status: abstract not yet drafted. Data lock was 10 March.
Abstract plan: "We'll draft it the week before and get it signed off quickly — it's only 2,000 characters."
Data: the same study was presented at a national meeting in February. No one has checked the encore policy.
Authors: 9 across 6 institutions. List agreed 8 months ago, not re-confirmed. Presenting author not yet asked whether they can attend.
Draft conclusion: "This regimen is highly effective and should be considered standard of care, particularly in biomarker-positive patients."
Poster plan: "We'll adapt the manuscript figures directly to save time." QR code will link to a page the brand team is building, containing the full AE listing and a plain-language summary.
Communications plan: sponsor press release scheduled for 8 September. Two authors' universities have press offices; neither has been contacted.
Disclosures: "We'll add those at the end if there's room."
The eighteen problems
Timeline (4)
1. Drafting the week before the deadline leaves no time for two author rounds plus medical, legal and regulatory review. The draft is needed now — 25 March is already three weeks late for a 15 April deadline.
2. “Signed off quickly—it’s only 2,000 characters” misunderstands review: length does not drive duration, and nine authors across six institutions will not respond in three days.
3. Poster production not planned at all. Printing and shipping need roughly 15 working days after final approval, so the poster draft is due mid-July.
4. No buffer for the submission portal on deadline day.
Encore and authorship (4)
5. February national meeting is an encore and the policy has not been checked. Do this before drafting — a prohibition changes everything.
6. If permitted, the prior presentation must be disclosed at submission. Undisclosed encores are rejected or withdrawn.
7. Author list not re-confirmed in 8 months. Circulate now for written confirmation; late authorship disputes are otherwise guaranteed.
8. Presenting author has not been asked whether they can attend. Confirm before submission, not at acceptance.
Content (4)
9. “Highly effective” — promotional characterisation.
10. “Should be considered standard of care” — a practice recommendation from a single trial.
11. “Particularly in biomarker-positive patients” — a claim built on one interaction test in a trial not powered for it.
12. No mention of OS immaturity in the conclusion.
Poster and QR (3)
13. “Adapt the manuscript figures directly” — manuscript figures must be redrawn for poster scale, not rescaled.
14. QR landing page built by the brand team, not in the review process. It is congress material and needs the same approval.
15. Disclosures “at the end if there’s room” — funding, conflicts and writing support must be on the poster, planned in, not fitted around.
Embargo (3)
16. Press release scheduled 8 September, embargo lifts 10 September. This is a planned embargo breach and would get the abstract withdrawn. Move it to 10 September at the earliest.
17. Two institutional press offices not contacted. The commonest breach source. Write to them by name with the exact date and time.
18. No embargo briefing to the author group at all.
Revised timeline, working backwards
· Today: check encore policy; circulate author list for confirmation; ask presenting author about attendance; start the draft.
· 1 April: draft to authors, round 1.
· 8 April: round 2. · 11 April: review complete. · 14 April: submit, one day early.
· Late June: poster drafting begins on acceptance.
· Mid-July: poster draft; figures redrawn; QR page enters review with the poster.
· 20 August: poster approved. · 22 August: to printer. · 5 September: ships.
· Immediately: embargo briefing to all authors and both press offices; press release moved to 10 September.
Scoring
14–18 — pass. You can run a congress project unsupervised.
9–13 — close; note which block you missed.
Under 9 — work through Weeks 1 and 4 again.
The one that would end the project: problem 16. A press release scheduled two days before the embargo lifts is not a risk, it is a breach already in the calendar — and it would be executed by a team that has no idea an embargo exists.
Timeline (4)
1. Drafting the week before the deadline leaves no time for two author rounds plus medical, legal and regulatory review. The draft is needed now — 25 March is already three weeks late for a 15 April deadline.
2. “Signed off quickly—it’s only 2,000 characters” misunderstands review: length does not drive duration, and nine authors across six institutions will not respond in three days.
3. Poster production not planned at all. Printing and shipping need roughly 15 working days after final approval, so the poster draft is due mid-July.
4. No buffer for the submission portal on deadline day.
Encore and authorship (4)
5. February national meeting is an encore and the policy has not been checked. Do this before drafting — a prohibition changes everything.
6. If permitted, the prior presentation must be disclosed at submission. Undisclosed encores are rejected or withdrawn.
7. Author list not re-confirmed in 8 months. Circulate now for written confirmation; late authorship disputes are otherwise guaranteed.
8. Presenting author has not been asked whether they can attend. Confirm before submission, not at acceptance.
Content (4)
9. “Highly effective” — promotional characterisation.
10. “Should be considered standard of care” — a practice recommendation from a single trial.
11. “Particularly in biomarker-positive patients” — a claim built on one interaction test in a trial not powered for it.
12. No mention of OS immaturity in the conclusion.
Poster and QR (3)
13. “Adapt the manuscript figures directly” — manuscript figures must be redrawn for poster scale, not rescaled.
14. QR landing page built by the brand team, not in the review process. It is congress material and needs the same approval.
15. Disclosures “at the end if there’s room” — funding, conflicts and writing support must be on the poster, planned in, not fitted around.
Embargo (3)
16. Press release scheduled 8 September, embargo lifts 10 September. This is a planned embargo breach and would get the abstract withdrawn. Move it to 10 September at the earliest.
17. Two institutional press offices not contacted. The commonest breach source. Write to them by name with the exact date and time.
18. No embargo briefing to the author group at all.
Revised timeline, working backwards
· Today: check encore policy; circulate author list for confirmation; ask presenting author about attendance; start the draft.
· 1 April: draft to authors, round 1.
· 8 April: round 2. · 11 April: review complete. · 14 April: submit, one day early.
· Late June: poster drafting begins on acceptance.
· Mid-July: poster draft; figures redrawn; QR page enters review with the poster.
· 20 August: poster approved. · 22 August: to printer. · 5 September: ships.
· Immediately: embargo briefing to all authors and both press offices; press release moved to 10 September.
Scoring
14–18 — pass. You can run a congress project unsupervised.
9–13 — close; note which block you missed.
Under 9 — work through Weeks 1 and 4 again.
The one that would end the project: problem 16. A press release scheduled two days before the embargo lifts is not a risk, it is a breach already in the calendar — and it would be executed by a team that has no idea an embargo exists.
Retake the diagnostic
Return to Module 1.1 and take the 24-item diagnostic again. The difference from your baseline is the only honest measure of what changed.
What to do next
Four habits. Build the calendar backwards from the deadline before anything else. Check the encore policy before drafting. Redraw every figure for the medium rather than rescaling it. And send the embargo briefing to institutional press offices by name.
Then pair it with the rest. Writing the Manuscript covers what each section may claim; Reading Source Documents covers where the numbers come from; Medical Information covers what may be said reactively. Congress work draws on all three under a harder time constraint.
Then pair it with the rest. Writing the Manuscript covers what each section may claim; Reading Source Documents covers where the numbers come from; Medical Information covers what may be said reactively. Congress work draws on all three under a harder time constraint.
Final reflection
Your diagnostic score at the start, and now. Which module changed how you would run a congress project most?
You have completed the course
Congress Materials
IMRAD Academy — Read like a scientist. Write like a professional.
Congress Materials · v1.0