Audience determines the document
A journal manuscript addresses specialists who will interrogate the method. It leads with the question, carries full methodological detail and hedges claims to the evidence.
A regulatory document addresses assessors against a defined framework. Structure is prescribed, completeness matters more than elegance, and every claim must be traceable to source data.
Clinical communication addresses practitioners deciding what to do. It leads with the bottom line, gives absolute effects and numbers needed to treat rather than relative risks alone, and states applicability.
Patient communication addresses people affected by the result. Plain language, absolute risks in natural frequencies, honest uncertainty, and no implied recommendation the evidence does not support.
What must not change between them
The finding, the uncertainty around it and the limitations. Simplifying for a lay audience means removing jargon, not removing caveats. A relative risk reduction of 50% becomes "2 in 100 instead of 4 in 100" — which is simpler and also more accurate about the magnitude.
The commonest failure in science communication is that uncertainty is dropped first, because it is the hardest thing to say briefly.
Structure that works in all four
Put the conclusion where the reader starts, not where the writer finished. One idea per paragraph. Name the thing the same way throughout. Define each abbreviation once and only where it earns its place.
Figures carry more than text for quantitative results, and a figure needs a caption that stands alone — a reader should understand it without the body text.
Numbers
Absolute risks alongside relative ones, always. Confidence intervals rather than bare p-values. Natural frequencies for lay audiences, since percentages of percentages defeat most readers. Significant figures matched to the precision actually achieved — a mean reported to four decimal places claims a precision the measurement did not have.
In short
What to take from this.
- Never drop: the uncertainty, however short the format.
- Give: absolute risks alongside relative ones, in every format.
- Write captions that stand alone without the body text.
Sources
- ICMJE Recommendations, on reporting and presentation
- Plain language and health literacy guidance
- SAMPL guidelines, on numerical reporting