Reporting research in medical journals and newspapers

Journalists' exposure to journal articles

The journalists interviewed said that at least one specialist journalist on each British broadsheet newspaper routinely scans every issue of the BMJ and Lancet for potential news stories and that they expect to find at least one story from these journals each week. Other medical journals (including general practice magazines) are scanned and used less regularly. Journalists tend to look down contents pages for “key” diseases and eminent authors because they often find the titles of articles incomprehensible. Key diseases included big killers such as cancer and currently topical diseases such as AIDS. The BMJ and Lancet usually reach the news rooms by Thursday lunchtime, and articles for Friday editions must be submitted by the evening so quick decisions are needed. Major news stories from the BMJ and Lancet appear on Fridays because the journals give journalists advance information but put an embargo on its publication until Friday. The embargo is intended to allow journalists time to prepare stories without having to race to beat their rivals into print. Despite time pressures, most journalists claim to go through the BMJ and Lancet from cover to cover, reading the abstract and conclusions of each article for a first impression of its potential for a story. Close attention is paid to letters pages, which are thought likely sources of news articles because they often cover topical subjects about which people feel strongly and are less likely to be reported by all the newspapers.

Both the BMJ and the Lancet issue press releases each week alerting journalists to articles they think newsworthy. At the time of the interviews, press releases from the Lancet summarised several articles in lay English and provided details for contacting the author. Journalists usually received these on Thursdays. The BMJ circulated briefer highlights of a few articles together with authors' details earlier in the week and encouraged journalists to request the full text of those which interested them. The journalists considered the press releases useful as either simple introductions to articles (Lancet), appetite whetters (BMJ), or indicators of what the journals considered important. However, they would not rely on press releases alone, regarding access to the full text of the journal as essential both to ensure they did not miss potential stories and to provide them with adequate information for a news article.

Research institutions and funding organisations also issue press releases about particular journal articles. The journalists generally appreciated those which concisely and clearly explained what the news was and why it was important. They were reluctant to wade through waffle to find a buried story and suggested that the best press releases were written by press officers with journalistic experience and a good feel for what journalists need. Journalists were wary of being manipulated, particularly by commercial interests, and preferred to come across a story for themselves—for example, while scanning journals. However, media relations efforts can influence selection of stories by bringing information to journalists attention, presenting it as newsworthy, and making a news article easier to write.

Content of medical news stories

I identified 90 articles in Friday newspapers based on 57 BMJ or Lancet articles over 18 weeks (an average of 1.25 news articles in each newspaper each Friday, with a range of 0-4). Journalists sometimes reported more than one journal article in one news piece. Five BMJ or Lancet articles were reported on a Friday by all four newspapers, four by three newspapers, and 16 by two.

The table shows the proportions of BMJ and Lancet articles from different sections that were reported as news articles. In total, five letters, five editorials, and 47 research or review papers were used in six, seven, and 87 news articles respectively. Only a small proportion of letters were reported, and they were less likely than research papers to be reported by more than one newspaper. Relatively few professional update type articles were reported.

Most (81%) articles reported had appeared in a journal press release. These accounted for 77 (86%) of the newspaper articles. At least three of the reported journal articles which had not appeared in journal press releases had been the subject of press releases issued by research funding bodies.

The journalists stressed that medically worthy information is not necessarily newsworthy. They said they were more likely to cover currently topical subjects; common and fatal diseases; rare but interesting or quirky diseases; those with a sexual connection; new or improved treatments; and controversial subject matter or results. They expressed a preference for research papers with British authors. The results of the content analysis echoed these comments. Box 1 lists the five most widely reported subjects (human insulin was topical and controversial at the time, and legal wrangles were being reported in some newspapers).

Box 2 gives the journal articles reported in newspapers but not put in the press release issued by the source journal. Journals seemed more reluctant to include case reports and material from editorials and letters in their press releases than newspapers were to report them. The journals' press releases were dominated by summaries of research papers. One notable subset of papers not picked up by the newspapers included nine summaries of BMJ general practice articles looking at topics such as the underrecognition of visual problems in elderly people; the use of questionnaires to help general practitioners target asthma care; the possibility that the British medical system discriminated against Asian doctors; and the effects of an announcement that a housing estate would be demolished on the number of consultations residents made with their general practitioners. These topics did not involve high technology medicine and related more to social problems than biomedical problems, which are more commonly covered by the media. The content analysis confirmed journalists' preference for British authors. The only reported articles without a British author covered topics of interest to British readers. Cholera transmission in Africa, pneumonia in Nepalese children, and Venezuelan haemorrhagic fever were all ignored.

Development of news stories

The journalists would often write news reports from the full text of a journal article, but shortage of time discouraged them from seeking extra information. Sometimes, however, they contacted authors to check they had understood the article, to “humanise” the research by including quotes, and to obtain stronger statements; they hoped authors would be less cautious over the telephone than they were in print. They attended news conferences for similar reasons. Many of the journalists were critical of the impersonal and inaccessible style of research papers.

It is a basic rule of journalism to give two sides to every story, but the journalists did not always feel obliged to do so for peer reviewed articles in respected journals. Although they would usually report defence statements if a journal article claimed that a product was harmful or a service ineffective and reaction comments on controversial or sensitive research, they would not routinely seek an opposing view from another researcher or practitioner. They were concerned that potential sources of such comment would often not have seen the article in question, might be prejudiced by rivalry, and might weaken a story by questioning the claims made. Summarising a research paper in a few hundred words is hard enough, but having to summarise comments on it as well, with no extra word allowance, is even harder.

Journalists preferred to quote recognised leaders in the field and trusted contacts who had previously supplied lively comments. Many approached medical research charity press offices to identify suitable experts for them, giving these organisations a chance to shape media reporting.

The journalists were keen that their articles should not give rise to undue optimism (for example, for a new cure) or pessimism (for example, about an environmental health hazard) among their readers. They would therefore consider their choice of language (several claimed never to use the word breakthrough) and were to some extent less trusting of research on small samples than of major studies. They were aware, however, that if they were too cautious their stories would not get printed at all, and if they could find a recognised expert to speak with enthusiasm about the latest results his or her comments were considered fair game.

Constraints of daily news reporting

Newspapers have limited space, and, since there are no reserved slots on news pages, articles about medicine compete with other stories. The competition is judged by editors, whose decisions reflect established news values such as the size and impact of an event, its relevance to readers, and strength of human interest.1 2 Stories that are not considered newsworthy are not printed. News journalists working to tight deadlines on daily newspapers said they had little time to identify and develop news articles from the wide range of potential stories available to them, so they value conveniently packaged information.

They were also constrained by their inability to evaluate the quality of evidence and argument presented in medical journals. None of the correspondents interviewed was medically trained and none had sophisticated knowledge about research methods. Although they routinely translated medical jargon into readable news, they were less able to judge and report the credibility or importance of research. They relied heavily on journal peer review processes and the opinions of medical experts to guide them in the selection and development of stories. The journalists surveyed were keen to get stories right and had a sense of responsibility about reporting medical research. Their values tended to converge with those of the medical journals, and they acknowledged the tension between writing responsibly about medical research and producing articles considered newsworthy by news editors. Faced with a strict word limit, journalists said that they found it impossible to include all the caveats and qualifying statements that are often found in research reports without killing their story.

Importance of peer review

General medical journals are a routine part of a medical journalists' “beat.” They are recognised as authoritative, the information in them is regarded as new, and their publication dates make the stories current. Peer review gives research papers an independent stamp of approval, and the journalists saw no need to check the information published in peer reviewed journals. They regarded the process as a quality filter and safeguard and reported news from journal articles safe in the knowledge that if their story subsequently turned out to be flawed the blame would fall on the experts, the editors, and the referees and not on them.

Medical and scientific communities encourage journalists' reliance on peer reviewed articles. Peer opinion and journal policies discourage authors from discussing their work with journalists until it has appeared in academically and professionally acceptable print (the journalists were acutely aware of this). They depended on medical journals as regular sources and cooperated with embargo arrangements because not doing so would risk no longer being given advance copies of material.

The main arguments for ensuring that research is peer reviewed before it is published are that the public should be protected from the dissemination of flawed research and exaggerated claims3 and that doctors should have access to medical research findings before their patients.4 Journal peer review is widely believed to improve the quality of manuscripts before publication5 but is not infallible. It is currently the subject of much research.6 Peer review is not intended to function as a censor for the public interest.

Conclusions

The way information flows from medical journals to newspapers influences the balance of medical topics reported, the quality of the research reported (and its appropriateness for public attention), and the quality of news reporting. Medical correspondents on quality newspapers rely quite heavily on a few journals as sources of medical research news, so the publication policies of these journals largely determine the pool of information from which stories are selected. (Media relations efforts also contribute to this.) Research published in specialist journals is less likely to come to journalists' attention and is less often used as a basis for news reports.7 Journal publication bias in favour of positive results is less likely to affect the balance of stories that the public receives because lay journalists apply similar biases themselves.8 Whatever topics the journalists are presented with, they will select stories in accordance with news values which are shaped primarily by media and political agendas. Medicine in the media tends to focus on hospital based medicine and pay scant attention to social causes of ill health.9 10

Concern about the quality of news coverage is at least partly rooted in concern about the effect it is likely to have. Media reports may increase the attention paid to particular projects by research communities,11 but the effect they have on the opinions and behaviour of practitioners, decision makers, or the public is uncertain. It is not known to what extent lay audiences appreciate that single journal articles are often not definitive and rarely supply enough evidence to justify a change of behaviour. Until research provides information about the ways in which audiences understand and are affected by news reports of medical research we cannot say what kind of news coverage is in the best public interest, and recommendations for changes in the behaviour of journal decision makers, media relations staff, and journalists as these affect news reports of medical research will be based only on opinion and personal judgment.

The study was supported by a research studentship from the Department for Education and Science. I thank Trevor Sheldon for his helpful comments.