Public-health writing lives or dies on what it is willing to cite and how carefully it weighs what it finds. This piece sets out the hierarchy of evidence this Review applies when it reports on London's health infrastructure, prevention agenda, and the legacy of the 2014 "Better Health for London" report — and what happens when the evidence behind a claim turns out to be weaker than first presented.
Why an Archive-and-Commentary Project Needs a Sourcing Standard
This site is an independent archive-and-commentary project built around the London Health Commission's work and Professor Lord Ara Darzi's 2014 review of the capital's health system. It is not the active Commission, the Greater London Authority, or the NHS, and it does not speak for any of them. That distinction only means something if the writing itself is held to a visible standard: readers should be able to tell the difference between a settled finding, a plausible inference, and an open question. A methodology page exists so that standard is written down rather than assumed.
The Hierarchy of Evidence We Follow
Public-health research is not all equally reliable, and treating a single small study the same way as a systematic review misleads readers even when every individual sentence is technically accurate. The Review follows the standard evidence hierarchy used across public-health research and guideline-writing bodies, from strongest to weakest:
| Tier | Evidence type | How it is used here |
|---|---|---|
| 1 | Systematic reviews and meta-analyses | Treated as the most reliable synthesis of a question; cited as the primary basis for a claim wherever one exists |
| 2 | Randomised controlled trials | Cited as strong evidence for intervention effectiveness, with the study's size and population noted |
| 3 | Cohort and case-control studies | Used for association-level claims, always flagged as observational rather than causal |
| 4 | Case series and case reports | Used sparingly, mainly to illustrate a mechanism, never as the sole basis for a policy claim |
| 5 | Expert opinion and anecdote | Attributed explicitly as opinion, never presented as settled fact |
This is the same ordering used by evidence-based public-health practice generally: reviews and meta-analyses sit above trials, which sit above observational studies, which sit above case reports and opinion. It is a ranking of reliability, not of interest — a well-reported case series can still be worth writing about, provided it is labelled as what it is.
Where We Look First
When a claim touches on clinical or population-health guidance, the Review looks first to the bodies that already do critical appraisal as their core function rather than attempting to re-adjudicate primary research from scratch. The National Institute for Health and Care Excellence (NICE) is the reference point for UK clinical and public-health guidance, since its recommendations are themselves built on structured evidence review. Cochrane serves the same role for intervention effectiveness questions, producing systematic reviews that are widely treated as a high-water mark for rigour in evidence-based healthcare. Where a NICE guideline or a Cochrane review exists on a topic, it is the anchor citation; where neither exists, the Review says so rather than substituting a single study or a press release and presenting it as equivalent.
The same discipline applies to how a finding is described, not just what is cited. A study of a few hundred participants is described as such, not inflated into "research shows." An association between two variables is described as an association, not smuggled into causal language. A preliminary or unreplicated finding is labelled preliminary. These habits track the same practices UK health-journalism bodies expect of specialist reporting: naming the source of a claim precisely enough that a reader could look it up, seeking independent scrutiny of extraordinary claims, and placing new findings in the context of the wider evidence rather than treating each new study as if it stood alone.
Corrections
Errors happen in any project that cites primary and secondary sources at volume, and the test of a sourcing standard is what happens after one is found, not whether one was ever made. The Review's approach mirrors the principle that underpins UK press self-regulation: a material inaccuracy is corrected promptly and visibly, not quietly edited out of a page with no trace. Where a figure, a study finding, or an attribution turns out to be wrong or outdated, the correction is made to the article itself and the nature of the change is not concealed. Minor wording tightening is treated differently from a correction that changes what a reader would have understood a claim to mean — the latter always gets flagged, not folded silently into a routine edit.
What We Don't Claim
The same conservatism applies to how the Review describes its own standing, not just the research it cites. This project makes no claim to be affiliated with the Mayor of London, the Commission, or Lord Darzi's office, and it does not assert a backlink profile, a ranking position, or a reach it has not verified. Where a data point about the Review's own authority is unverified, it is left out rather than implied. That is the same rule applied to every clinical and policy claim on the site: state what is supported, mark what is uncertain, and never let an absence of evidence get rewritten as evidence of absence.
Conclusion
A methodology page is only useful if it changes how the rest of a site reads. Applied consistently, the hierarchy of evidence above means a systematic review gets weighted differently from a single small trial, an association is never quietly upgraded to causation, and a correction is a visible event rather than an invisible edit. That is the standard this Review is holding itself to across its coverage of London's health infrastructure — and the standard by which its own claims, including about its own reach and authority, should be judged.