Approach
Method before message.
This is how we decide what to trust, how we keep it current, and where we draw the line between information and advice.
01 — Evidence hierarchy
Sources are tiered, not pooled.
A statistic from a state health department and a statistic from a community blog are not the same kind of fact, even when they state the same number. We tier every source we cite, and every figure in our work inherits the tier of its weakest input.
Government & statutory sources
National and state health departments, statutory agencies, and official statistical collections. Given primary weight because they carry legal reporting obligations, standardised methodology, and regular revision cycles.
Professional & clinical bodies
Medical colleges, professional associations, and clinical guideline bodies. Weighted for consensus positions and peer-reviewed clinical guidance, particularly where government data is silent on practice-level detail.
Universities & registries
Peer-reviewed research, longitudinal registries, and cohort studies. Valued for methodological rigour, but dated to publication and checked against more recent official data where it exists.
Community & sector organisations
Health NGOs and community organisations. Included for lived-experience and access context, and clearly labelled as such — not presented with the same evidentiary weight as clinical or statutory sources.
The order matters because it’s load-bearing: when a lower-tier source is the only one available, we say so, and we flag the claim as provisional rather than quietly borrowing a higher tier’s authority.
02 — Governance
Reviewed, dated, and bounded.
Review and revision
Every piece of content carries a visible last-reviewed date. Content is re-checked against its sources on a set schedule and whenever the underlying statutory data is revised — not left to go stale between rebuilds.
Clinical-safety boundaries
We maintain a hard distinction between information and diagnosis, and between screening and clinical assessment. Nothing we build outputs a diagnosis, a risk score dressed up as one, or advice that substitutes for a clinician.
03 — Health literacy
Written to be understood, not just read.
We write in plain English by default and use progressive disclosure for detail: a claim states its headline first, then lets the reader open it up to see the methodology, the caveats, and the source. Nothing depends on the reader already knowing the jargon — if a term is necessary, we define it on first use. This is tested against how people actually read a page, not against how a data team would prefer to present it.
04 — Accessibility
A stated standard, honestly reported against.
We build to WCAG 2.1 AA as a baseline: colour contrast, keyboard operability, and screen-reader structure are considered from the first layout, not retrofitted. What we won’t do is claim full compliance we haven’t verified. Where a build has had automated checks and manual spot-checks but not a full third-party audit, we say that plainly rather than round up.