A citation is only as trustworthy as the source behind it. Before this Review treats any figure, claim, or recommendation as citable evidence, it runs that source through a set of practical checks — not a ranking of study types (that’s covered elsewhere), but a filter for whether the source itself deserves to be quoted at all. Evaluating health information sources this way is a discipline borrowed from UK statistics practice, health journalism guidance, and clinical evidence review — applied here to a civic archive project with no measured backlink authority of its own, where sourcing discipline is one of the few forms of credibility actually within its control.
This piece sets out those checks: what counts as a primary source, what trust signals a secondary source needs before it’s used, how conflicts of interest are treated, and where evidence-quality frameworks come in when a claim rests on thinner ground.
Primary Sources First
The starting point is always the original data, not someone else’s summary of it. For anything statistical — life expectancy figures, air quality readings, screening uptake — that means going to the body that actually collected and published the numbers: the Office for National Statistics, NHS Digital, or the relevant public body, rather than a news article reporting on them secondhand. The Office for National Statistics publishes its data under a formal Code of Practice for Statistics, which requires that statistical releases be presented clearly, link back to the underlying dataset, and include a plain account of how the data was collected and how reliable it is. NHS Digital similarly maintains a Data Model and Dictionary that sets out approved definitions and standards for health data used across England. When a claim traces back to one of these sources directly, it is treated as primary. When it only traces back to a press article, a summary post, or an advocacy report describing what the data showed, it is treated as secondary — useful for context, but not sufficient on its own to support a specific figure.
Trust Signals for Secondary Sources
Not every claim can be sourced to a raw dataset, and secondary sources have a real role — expert commentary, sector reporting, patient-facing guidance. Before using one, this Review looks for the same basic signals that health-information quality checklists such as DISCERN were built around: is the publisher clearly identified, is there a dated publication (and, ideally, a review date), and does the source itself show its evidence rather than asserting conclusions without support? A domain ending in .gov.uk or .nhs.uk carries a different weight than an unaffiliated commercial site, and an “About us” page that names who is actually responsible for the content is a minimum bar, not a bonus. Sources that fail these basic checks — undated, unattributed, or unwilling to show their working — are treated as unreliable regardless of how confidently they’re written.
Declaring Conflicts of Interest
Medical and health journalism has a well-established approach to this problem. The International Committee of Medical Journal Editors’ disclosure framework asks authors to declare financial and other relationships — payments, consultancies, sponsorships — that could plausibly bias a piece of work, generally covering the preceding three years, while ordinary public funding from government bodies or charities is not treated the same way as funding from a commercial party with a stake in the outcome. Health journalism bodies apply a parallel principle: a source’s link to a company or product that stands to benefit from the finding being reported should be disclosed and weighed, not ignored. This Review applies the same logic before citing a study, white paper, or industry report: who funded it, and does that funding relationship match what the source is claiming? A study or report that is silent about its funding, or that comes from a body with an obvious commercial stake in its own conclusion, is either flagged as such or left out.
Leaning on Established Evidence-Quality Frameworks
Where a claim rests on a single study rather than an official dataset, this Review looks at how that evidence has itself been assessed, rather than taking the headline finding at face value. The National Institute for Health and Care Excellence produces evidence-based guidance that is treated as a reliable reference point precisely because it has already been through independent scrutiny. Cochrane’s risk-of-bias tools — RoB 2 for randomised trials and ROBINS-I for non-randomised studies — are the standard mechanism clinical researchers use to judge whether a study’s design lets its conclusions be trusted; when a cited study has been through that kind of scrutiny, or has been assessed against reporting guidelines from bodies such as the EQUATOR Network, that is a meaningfully stronger basis for a claim than an unreviewed press release.
Primary vs Secondary Sources at a Glance
| Source type | Example | How this Review treats it |
|---|---|---|
| Primary | ONS statistical bulletin, NHS Digital dataset, official GLA report | Cited directly for figures; preferred wherever available |
| Reviewed secondary | NICE guidance, a Cochrane-assessed study | Cited with attribution; treated as strong support |
| Unreviewed secondary | News coverage, undated blog summary | Used for context only, never as the sole basis for a figure |
| Undisclosed-interest source | Vendor white paper with no stated funding | Flagged or excluded until funding is clear |
Why This Matters for a Site Like This One
This Review has no measured backlink profile and makes no claim to be the active London Health Commission, the Greater London Authority, or Lord Ara Darzi’s office — it is an independent archive-and-commentary project working from the public record of the 2014 Better Health for London report and what has followed it. Given that, sourcing discipline isn’t a formality; it’s close to the whole basis of whatever credibility this project has. Evaluating health information sources properly — primary before secondary, disclosed funding before hidden funding, reviewed evidence before a single unchecked study — is the mechanism that keeps that credibility earned rather than assumed.