Science Communication for UK Journalists: A Practical Guide

Understanding Peer Review

Peer review is the process by which a submitted scientific paper is independently assessed by other experts in the same field before publication. It is the primary quality-control mechanism in academic science, but it is not infallible, and journalists who treat a peer-reviewed paper as automatically reliable are at risk of amplifying significant errors.

Understanding the mechanics of peer review helps reporters ask the right questions. When a study appears in a high-impact journal such as The Lancet, BMJ, or Nature, the peer-review process is rigorous and multi-stage. A study posted to a preprint server — such as medRxiv or bioRxiv — has not been peer-reviewed at all. During the COVID-19 pandemic, many preprints received widespread media coverage before their methodology had been scrutinised by independent experts; several were subsequently found to be deeply flawed or retracted after formal peer review. Reporters should always check whether a study is peer-reviewed and published, or merely a preprint.

• Open-access and paywalled journals: The BMJ and some Lancet articles are freely accessible. Others require institutional or personal subscription. The University of Nottingham and many other UK universities operate open-access repositories where authors deposit accepted manuscripts. PubMed Central (PMC) hosts a large archive of freely available biomedical research funded by public grants.
• Impact factor: Journals are ranked by impact factor — a measure of how frequently their papers are cited. Higher-impact journals (such as Nature, The Lancet, NEJM, and BMJ) apply more stringent peer review. A study in a low-impact or predatory journal deserves less weight than one in a top-tier publication.
• Retractions: Published papers can be retracted after errors or misconduct are discovered. The Retraction Watch database (retractionwatch.com) tracks retracted papers. Before reporting on a study, check whether it has been the subject of post-publication corrections or expressions of concern.
• Conflicts of interest: Peer-reviewed papers are required to declare funding sources and author conflicts of interest. Research funded by an industry with a financial interest in the outcome should be reported with that context included. Independent replication by unfunded researchers is the gold standard.

The Science Media Centre Embargo System

The Science Media Centre (SMC) at sciencemediacentre.org is one of the most valuable resources available to UK science journalists. It operates a unique service that provides accredited journalists with early access to scientific papers under embargo, alongside expert reaction written by independent scientists who have reviewed the study before it is publicly released.

The SMC's embargo system works as follows: when a major journal prepares to publish a significant study, it shares the paper with the SMC under embargo. The SMC recruits independent UK scientists to read and comment on the study. These expert comments — typically a mix of methodological assessment, contextualisation, and caveats — are then made available to registered journalists alongside the embargoed paper, usually 24-48 hours before the paper is published. This gives reporters time to read the study, understand the expert reaction, seek additional comment, and file a more considered story than they could produce under a same-day embargo.

• How to register: Journalists with a regular byline in an accredited UK news outlet can register for SMC access at sciencemediacentre.org. The SMC covers all scientific disciplines but specialises in health, medicine, climate, and technology.
• Expert reaction alerts: Once registered, you receive email alerts when new expert reactions are published. These cover both newly published studies and breaking science stories — such as a government health announcement or an IPCC report — where independent expert assessment is valuable.
• Embargo discipline: Embargoes must be observed strictly. Publishing before the agreed embargo lift time risks losing SMC access and breaching the trust of the journal. Most embargoes lift at 23:01 UK time on the date of online publication.
• Beyond the SMC: Major UK universities operate their own press offices and distribute embargoed releases via services including EurekAlert! and AlphaGalileo. These provide additional access to embargoed research, though without the independent expert reaction that makes the SMC particularly valuable.

Types of Scientific Studies

Different types of study provide different levels of evidence. Understanding the hierarchy of evidence is essential for accurately communicating what a study can and cannot establish:

• Randomised Controlled Trial (RCT): Participants are randomly assigned to a treatment or control group. Because randomisation distributes confounding factors between groups, an RCT can establish causation more reliably than any other study design. RCTs are the gold standard in medical research. Their limitations include cost, ethical constraints, and the fact that they may not reflect real-world conditions (external validity).
• Systematic review and meta-analysis: A systematic review synthesises all available evidence on a question using defined search and selection criteria. A meta-analysis goes further, statistically combining data from multiple studies to produce a pooled estimate. Both sit at the top of the evidence hierarchy and are the basis for most NICE guidance. A good meta-analysis is more reliable than any single RCT; a poorly conducted one can mislead by pooling incompatible studies.
• Cohort study: A group of people (a cohort) is followed over time and their exposures and outcomes are recorded. Cohort studies can identify associations between exposure and disease over long periods. They cannot establish causation as definitively as an RCT because the groups being compared may differ in other ways that affect the outcome (confounding). The UK Biobank, which has followed over 500,000 participants, is a major source of cohort data.
• Case-control study: People with a disease (cases) are compared with similar people without the disease (controls), and researchers look back to identify differences in past exposures. Case-control studies are useful for rare diseases where following a large cohort would be impractical. They are susceptible to recall bias (cases may remember past exposures differently from controls) and selection bias.
• Cross-sectional study: Data is collected at a single point in time, providing a snapshot. Cross-sectional studies can describe associations but cannot establish which came first — the exposure or the outcome. They generate hypotheses for more rigorous investigation rather than definitive conclusions.
• Animal and laboratory studies: Results from animal models and in vitro (test tube) experiments are frequently reported in the press as if directly applicable to humans. Many findings do not replicate in humans. These studies are early-stage and should be reported as such, with explicit caveats about the translation gap.

Correlation vs Causation

The most common error in science reporting is conflating correlation with causation. Two variables can be strongly correlated — they vary together — without one causing the other. The classic example is the correlation between ice cream sales and drowning deaths: both increase in summer, but ice cream does not cause drowning. The explanation is a confounding variable (warm weather) that drives both.

Observational studies (cohorts, case-controls, cross-sectionals) identify associations. They cannot, on their own, establish that A causes B. Additional criteria — most famously Bradford Hill's criteria for causation in epidemiology, which include factors such as strength of association, consistency, biological plausibility, and dose-response relationship — are used to assess whether a causal interpretation is warranted. Reporters should reflect the appropriate level of causal confidence in their language:

• Observational finding: “The study found an association between X and Y” or “people who did X were more likely to develop Y.”
• Experimental finding (RCT): “The trial found that X reduced the risk of Y” is permissible, because randomisation allows a causal inference.
• Avoid: “X causes Y” or “X leads to Y” for observational studies unless there is a strong, replicated body of evidence and scientific consensus supporting the causal claim.
• Reverse causation: A further pitfall is reverse causation. A study might find that people who take vitamin supplements are healthier, but the explanation may be that healthier people are more likely to take supplements, rather than that supplements cause better health.

NICE Guidance and Health Claims

The National Institute for Health and Care Excellence (NICE) at nice.org.uk is the body responsible for producing evidence-based guidance on health treatments, public health interventions, and social care in England. NICE guidance represents a rigorous synthesis of the best available evidence and is the authoritative benchmark for clinical practice in the NHS. For journalists reporting on health topics, NICE guidance is both a primary source and a calibration point for assessing other claims.

• Guidelines: NICE publishes clinical guidelines on specific conditions and treatments, incorporating systematic reviews and evidence ratings. If a treatment is recommended by NICE, it has passed a threshold of evidence and cost-effectiveness assessment. If it is not recommended, there is usually a documented reason.
• Technology appraisals: NICE's technology appraisals assess whether specific drugs or medical devices should be made available on the NHS. A positive appraisal is a major health news story; a negative one can generate public controversy, particularly around expensive cancer drugs.
• Evidence search: NICE Evidence Search (evidence.nhs.uk) provides access to health evidence including NICE guidance, systematic reviews, and NHS quality standards. It is a useful starting point for any health story.
• Scotland, Wales, and Northern Ireland: NICE guidance applies in England. Scotland has the Scottish Medicines Consortium (SMC) and Healthcare Improvement Scotland. Wales has the All Wales Medicines Strategy Group (AWMSG). Northern Ireland generally follows NICE but has its own implementation processes. Always check which guidance applies in the jurisdiction you are reporting on.

The Royal Society and Expert Sources

The Royal Society (royalsociety.org) is the UK's national academy of science and one of the world's most prestigious scientific institutions. It publishes policy reports, responds to government consultations, and issues statements on matters of scientific consensus. For journalists, it is an authoritative source for the scientific mainstream position on contested questions, from climate change to AI safety to vaccine policy.

• Royal Society reports: When the Royal Society publishes a report on a topic — such as its reports on COVID-19, machine learning, or ocean acidification — it represents the considered view of leading scientists across the field. These reports undergo their own rigorous review process and are reliable secondary sources for complex scientific topics.
• Finding experts: Fellows of the Royal Society (FRS) are among the most distinguished scientists in the UK. The Royal Society website allows you to search Fellows by research area. Similarly, the Royal Society of Medicine, the Academy of Medical Sciences, and the British Medical Association all operate expert-finding services for journalists.
• Expert databases: The SMC maintains a database of scientists willing to comment to the media. The Association of British Science Writers (ABSW) and the British Science Association provide networking and training resources for science journalists. Many universities also operate expert-finding services through their press offices.
• Independence matters: When seeking expert comment, always ask about potential conflicts of interest. A scientist who has received research funding from a pharmaceutical company, or who holds a patent related to the topic, may have financial interests that should be disclosed. The best expert sources are those with no financial stake in the outcome and whose institution is independent of the funder of the research being discussed.

IPSO Clause 1 and Accuracy Obligations

IPSO Clause 1 (Accuracy) of the Editors' Code of Practice requires that the press does not publish inaccurate, misleading, or distorted information. In science and health reporting, this creates specific obligations around the accuracy of claims, the representation of evidence, and the inclusion of appropriate caveats. IPSO has upheld complaints against newspapers that misrepresented health study findings, overstated evidence, or failed to correct inaccurate science claims promptly.

• Headlines vs body text: The majority of IPSO accuracy complaints about science stories involve headlines that overstate the findings in the body copy. “X cures cancer” is a different claim from “early-stage research finds X may inhibit cancer cell growth in laboratory conditions.” Subeditors writing science headlines should accurately reflect the body text and the underlying study's design and findings.
• Quantifying risk: Relative risk statements are regularly misunderstood by readers and misrepresented by reporters. A headline stating that X “doubles the risk” of Y sounds alarming; if the absolute risk increases from 0.01% to 0.02%, the real-world impact is very small. Best practice is to report both relative and absolute risk where possible.
• Obligation to correct: Clause 1 requires that significant inaccuracies are corrected promptly and with due prominence. If you have published a science story that overstates a finding, the correction must appear in a position that gives it reasonable visibility — not buried in a corrections column. Online articles should carry corrections as an update notice attached to the original piece.
• Distinguishing comment from fact: Scientists and commentators may express speculative or contested views. If you report these views, make clear they are the opinion of the source, not an established scientific finding. Attributing a speculative claim to the source rather than presenting it as fact is a basic accuracy safeguard.

Reading a Scientific Paper

Journalists who learn to read the key sections of a scientific paper rather than relying solely on the abstract or press release will consistently produce more accurate and more sophisticated science coverage. Most peer-reviewed papers follow a standard structure: Abstract, Introduction, Methods, Results, Discussion, and Conclusion. Each section is relevant to the journalist for different reasons:

• Abstract: The summary of the entire paper. Useful for a quick overview, but the abstract is sometimes written to emphasise the most favourable findings. Always read the Methods and Discussion sections before writing your story.
• Methods: This section explains how the study was conducted: the sample size, the selection criteria, the measurement tools, and the statistical analysis used. Sample size matters: a study of 40 people can rarely support universal claims. Check whether the study population is representative of the group to whom the findings are being applied.
• Results: The raw findings, usually presented with statistical measures including confidence intervals and p-values. A result is described as “statistically significant” when the p-value is below a conventional threshold (usually 0.05), meaning there is less than a 5% probability the result occurred by chance. Statistical significance does not imply clinical or practical significance — a drug that produces a statistically significant but tiny improvement in a symptom score may not provide meaningful patient benefit.
• Discussion and Limitations: Authors are required to discuss the limitations of their study — what the findings cannot establish, what alternative explanations exist, and what further research is needed. This section is essential reading: if the authors themselves acknowledge significant limitations, your story should reflect them.
• Funding and conflicts of interest: Declared at the end of most papers. Industry-funded research is not automatically unreliable, but funding source is relevant context for your readers. Research funded by a tobacco company on the health effects of smoking, or by a food manufacturer on the benefits of ultra-processed food, warrants particular scrutiny.

Health Reporting Pitfalls

Health journalism carries particular responsibilities because inaccurate coverage can influence readers' health decisions. The following pitfalls are among the most frequently identified by the NHS Behind the Headlines service (which assessed the accuracy of health stories in UK media before it was discontinued) and by IPSO complaints panels:

• The “breakthrough” problem: The word “breakthrough” is systematically overused in health reporting. Genuine therapeutic breakthroughs — new treatments that substantially change clinical practice — are rare. Most positive health study findings are incremental steps in an ongoing research programme, not clinical solutions ready for patient use.
• Animal study translation: Results from mouse or rat models frequently do not translate to humans. The history of medicine contains many failed drug candidates that showed promise in animals but were ineffective or harmful in humans. Animal studies should always be reported with explicit language that describes them as early-stage and pre-clinical.
• The BMJ Christmas edition: The BMJ publishes a well-known Christmas edition featuring deliberately humorous or unusual studies. These are peer-reviewed but are intended as light-hearted commentary rather than serious clinical evidence. They are regularly misreported as genuine health guidance.
• Screening and prevention stories: Stories about cancer screening or preventive interventions are frequently oversimplified. Benefits are typically reported without harms — false positive rates, overdiagnosis, the psychological burden of recall appointments, and the risks of unnecessary treatment. Balanced reporting on screening should include both the potential benefit (lives saved) and the potential harm (unnecessary procedures arising from false positives).
• Drug approval vs NHS availability: A drug approved by the Medicines and Healthcare products Regulatory Agency (MHRA) as safe and effective is not automatically available on the NHS. It must first receive a positive NICE appraisal and be commissioned by NHS England. Stories about drug approvals should make this distinction clear.

Practical Checklist

Before publishing any science or health story:

• Confirm the study is peer-reviewed and published, not a preprint or press release summary
• Identify the study type (RCT, cohort, case-control, systematic review) and reflect its evidential weight in the copy
• Read the Methods and Limitations sections of the paper, not just the Abstract
• Check for conflicts of interest and funding sources, and include them as context if relevant
• Obtain independent expert comment from someone not involved in the study
• Use language that reflects the level of evidence: “associated with” for observational studies, not “causes”
• Report absolute risk alongside relative risk where the distinction is material
• Ensure the headline accurately reflects the body copy and does not overstate the finding
• For health claims, cross-reference against current NICE guidance at nice.org.uk
• Check Retraction Watch if the paper is more than a few months old

Common Mistakes

• Reporting preprints as if peer-reviewed: Preprint servers allow scientists to share work before formal peer review. A preprint may not survive peer review. Always check publication status before reporting.
• Using relative risk without absolute risk: A 50% increased risk sounds alarming. If the baseline risk is 2 in 10,000, the elevated risk is 3 in 10,000 — a real but very small difference. Reporting only the relative figure is misleading.
• Treating association as causation: Observational studies identify associations. Only well-designed experiments, typically RCTs, can reliably establish causation. Reporting an observational association as a causal finding is one of the most common and consequential errors in science journalism.
• Over-generalising from small samples: A study of 30 people cannot support population-level conclusions. Check sample size and whether the study population is representative of the group to whom findings are applied.
• Ignoring study limitations: Authors flag limitations in their Discussion section for a reason. A reporter who omits material limitations is presenting an incomplete and potentially misleading account of the evidence.
• Confusing statistical and clinical significance: A finding can be statistically significant (unlikely to have occurred by chance) without being clinically meaningful. Always ask: does this difference matter in practice for patients?
• Taking press releases at face value: University and journal press releases are written to maximise coverage. They routinely inflate the significance of findings and omit caveats that appear in the paper itself. Go to the paper, not the press release.

Red Flags

• A study published in a journal you have not heard of, or one that charges article processing fees without clear peer-review standards — potential indicator of a predatory journal
• No conflicts of interest or funding declaration — a significant omission that should prompt further investigation before reporting
• A study whose conclusions substantially exceed what the data can support — the Discussion section will often reveal this
• Expert commentators who are all from the same institution as the study authors or funded by the same source
• A press release that uses language such as “proves,” “cures,” or “eliminates” for a single study finding — science rarely proves anything definitively from one study
• A health supplement, food product, or treatment being promoted on the basis of a single industry-funded study with no independent replication
• Statistics presented without denominators — “X cases were diagnosed” is meaningless without knowing the size of the population studied
• A spokesperson or expert who refuses to name the study they are citing, or whose claimed credentials cannot be verified

Primary Sources

• Science Media Centre — Expert reaction, embargoed papers, and journalist registration
• BMJ (British Medical Journal) — Key peer-reviewed medical journal, much content freely accessible
• NICE (National Institute for Health and Care Excellence) — Clinical guidelines and technology appraisals
• Royal Society — Scientific consensus statements, policy reports, and Fellow finder
• IPSO Editors' Code of Practice — Clause 1 (Accuracy) and its application to science and health reporting
• Retraction Watch — Database of retracted scientific papers
• PubMed / PubMed Central — Free access to biomedical research abstracts and many full-text papers
• MHRA (Medicines and Healthcare products Regulatory Agency) — UK medicines and medical device regulation
• IPSO Editors' Code 2025: A Full Guide — Detailed walkthrough of all 16 Clauses including Clause 1
• Investigative Journalism Techniques for UK Reporters — Source verification and document analysis skills