iatroX is developed by a practising UK GP and combines UK guideline retrieval with relevant international medical evidence. This article sets out, directly and without hedging, how that combination is actually prioritised when a clinical question is asked.
The proposed hierarchy for treatment and prevention questions
For most treatment and prevention questions, the working order runs: current authoritative UK clinical guidelines first, followed by high-quality systematic reviews and meta-analyses, followed by high-quality randomised controlled trials, followed by appropriate cohort and case-control studies, and finally case series, case reports and mechanistic evidence where nothing stronger is available for the specific question being asked.
Why guidelines can reasonably sit above individual meta-analyses at the point of care
Placing authoritative guidelines ahead of individual meta-analyses in a UK point-of-care context is a deliberate choice, not an oversight of the evidence hierarchy's usual ordering. Guidelines integrate evidence quality with safety considerations, cost, feasibility and real-world implementation within the specific healthcare system a UK clinician is working in. They translate raw literature into a healthcare-system-level recommendation. And they are, by construction, more directly and immediately relevant to NHS practice than a standalone meta-analysis that may not have considered UK-specific formulary, referral, or resourcing constraints at all.
The exceptions worth stating explicitly
This guideline-first default is not absolute, and three situations reasonably override it. A guideline may simply be outdated, not yet reflecting evidence published since its last review. New, high-quality evidence may have emerged after a guideline's search date closed, evidence the guideline committee never had the chance to consider. And a guideline may not cover the specific patient group or clinical question being asked, leaving a genuine gap that the underlying evidence base, rather than a non-existent recommendation, has to fill directly.
What iatroX deliberately does not do
iatroX does not naively count citations. Ten weak studies pointing the same direction should not automatically outweigh one large, well-conducted, definitive trial. Several papers drawing on the same underlying dataset or cohort should not be treated as independent confirmation simply because they were published separately. And a meta-analysis should not be preferred by default where its own component studies are individually poor or too heterogeneous to have been meaningfully combined.
A question-specific hierarchy, not one hierarchy for everything
The specific ordering above is calibrated for therapy questions. Diagnosis questions are weighted towards prospective diagnostic-accuracy studies. Prognosis questions favour inception cohort studies. Harm questions weight observational and pharmacovigilance data more heavily than the treatment-question hierarchy would suggest. And screening questions draw on population-level evidence and modelling in a way none of the other categories require.
iatroX's UK and EU positioning, stated plainly
The combination this reflects is deliberate: international evidence as the underlying evidence base, UK guideline interpretation as the primary lens for point-of-care relevance, UK medicines and care-pathway relevance built into how answers are framed, and clinician-led product design throughout, reflecting the fact that the platform was built by a practising GP working directly with the specific retrieval problems NHS clinicians actually encounter.
Why this ordering is a defensible, not an arbitrary, choice
It would be reasonable to ask why guidelines outrank a fresh meta-analysis at all, given that the classical evidence-based medicine hierarchy usually places meta-analysis at the very top. The answer is that the classical hierarchy was built primarily to answer "which study design gives the most reliable estimate of an effect," a slightly different question from "what should a UK clinician actually do right now, given everything relevant including cost, safety, feasibility and system constraints." Guidelines answer the second, broader question directly; a standalone meta-analysis, however rigorous, only ever answers the first.