Writing Conventions for Public Health
The unit of analysis is a population, so denominators matter
A count means nothing without the population it came from. Rates, prevalence, and incidence are reported with the denominator, the time window, and the standardization method, because a reader has to be able to compare across settings.
Two audiences, two documents, from the same evidence
The journal article and the policy brief say the same thing to different readers. A brief leads with the recommendation and puts the method in a box; a paper builds to the recommendation and puts the method in the middle. Writing one and hoping it works as the other fails both.
Equity language is scrutinized as closely as the statistics
Describe people by circumstance rather than by category — “people experiencing homelessness,” not a label that fuses the person to the condition. Avoid deficit framing that locates a problem in a community rather than in the conditions and policies acting on it.
Grey literature is legitimate evidence and awkward to cite
Agency reports, surveillance bulletins, and government statistics often carry the best data available, and they rarely have a clean author-date-journal shape. Record the issuing body, report number, and access date at the moment you use them, because reconstructing them later is painful.
Mixed-methods papers report each strand on its own terms
The quantitative strand gets estimates and intervals; the qualitative strand gets themes with illustrative quotations and a note on how they were coded. The integration paragraph — where the two strands meet — is the part reviewers most often say is missing.
Implications are written for people who act, not just for readers
A discussion that ends at “further research is needed” underdelivers in this field. State who would do what, at what level — clinic, district, national policy — and be explicit about the cost and feasibility assumptions behind the recommendation.
