You are here

Article Text

Article menu

Download PDFPDF

Research report
Academics and competing interests in H1N1 influenza media reporting
  1. Kate L Mandeville1,
  2. Sam O'Neill2,
  3. Andrew Brighouse3,
  4. Alice Walker4,
  5. Kielan Yarrow5,
  6. Kenneth Chan6
  1. 1Department of Global Health and Development, London School of Hygiene and Tropical Medicine, London, UK
  2. 2School of Medicine, Imperial College London, London, UK
  3. 3Accident and Emergency Department, Princess Alexandra Hospital NHS Trust, Harlow, UK
  4. 4Accident and Emergency Department, Chase Farm Hospital, Enfield, UK
  5. 5Department of Psychology, City University London, London, UK
  6. 6Barts and The London School of Medicine & Dentistry, London, UK
  1. Correspondence to Dr Kate L Mandeville, Department of Global Health and Development, London School of Hygiene and Tropical Medicine, 15-17 Tavistock Place, London WC1H 9SH, UK; kate.mandeville{at}lshtm.ac.uk

Abstract

Background Concerns have been raised over competing interests (CoI) among academics during the 2009 to 2010 A/H1N1 pandemic. Media reporting can influence public anxiety and demand for pharmaceutical products. We assessed CoI of academics providing media commentary during the early stages of the pandemic.

Methods We performed a retrospective content analysis of UK newspaper articles on A/H1N1 influenza, examining quoted sources. We noted when academics made a risk assessment of the pandemic and compared this with official estimations. We also looked for promotion or rejection of the use of neuraminidase inhibitors or H1N1-specific vaccine. We independently searched for CoI for each academic.

Results Academics were the second most frequently quoted source after Ministers of Health. Where both academics and official agencies estimated the risk of H1N1, one in two academics assessed the risk as higher than official predictions. For academics with CoI, the odds of a higher risk assessment were 5.8 times greater than those made by academics without CoI (Wald p value=0.009). One in two academics commenting on the use of neuraminidase inhibitors or vaccine had CoI. The odds of CoI in academics promoting the use of neuraminidase inhibitors were 8.4 times greater than for academics not commenting on their use (Fisher's exact p=0.005).

Conclusions There is evidence of CoI among academics providing media commentary during the early H1N1 pandemic. Heightened risk assessments, combined with advocacy for pharmaceutical products to counter this risk, may lead to increased public anxiety and demand. Academics should declare, and journalists report, relevant CoI for media interviews.

  • PUBLIC HEALTH POLICY
  • INFLUENZA
  • ETHICS
  • EPIDEMICS

This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/

https://doi.org/10.1136/jech-2013-203128

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    The UK spent an estimated one billion pounds on pharmaceutical products during the 2009 to 2010 A/H1N1 influenza pandemic, including neuraminidase inhibitors (NI) and H1N1-specific vaccine.1 Pharmaceutical companies made profits of 4.5–6.5 billion pounds from H1N1 vaccines alone.2 This was despite the evaluation of the pandemic as less severe than previous pandemics3 ,4 and uncertainty over the effectiveness of neuraminidase inhibitors (a type of antiviral medication) in reducing transmission and complications of influenza.5

    In the postpandemic period, there were significant concerns about competing interests (CoI) among experts on influential advisory committees, including the WHO Emergency Committee.2 ,6 ,7 Members of these committees have been linked to manufacturers of both neuraminidase inhibitors and influenza vaccines.7 ,8 There have been repeated calls for greater transparency around the potential influence of the pharmaceutical industry on the decisions made by these committees.2 ,6 ,7 ,9

    Public health academics are often asked to provide commentary and analysis on emerging health risks by the media. Media coverage of health issues has been shown to influence the public's perception of risk, demand for new drugs and policy decisions.10–13 In the UK, extensive media advocacy of the breast cancer drug trastuzumab (Herceptin) resulted in a ‘fast-track’ approval from the National Institute for Health and Clinical Excellence,14 but there was subsequent debate over the cost-effectiveness of the drug.15 It has been suggested that optimistic media portrayals may be more successful for pharmaceutical companies than explicit promotional campaigns as “the message is separated from an obvious marketing agenda and often includes a trusted voice, such as a university-based researcher. Paradoxically, this trust is based in part on a belief in the perceived independence of university researchers”.16 Like those on advisory committees, academics quoted in the media may also have possible CoI. Media commentaries, therefore, represent an alternative route to exert pressure on public demand and one in which CoI are not routinely declared.

    We set out to examine media commentary on A/H1N1 influenza provided by academics during the period in which the UK government decided its policy on public provision of NI and H1N1-specific vaccine (NI/vaccine). We then independently searched for CoI for each academic to determine whether commentary from academics with and without CoI was significantly different.

    Methods

    Design and setting

    This study was a retrospective content analysis of UK newspaper reporting. We excluded television and radio coverage, as audiovisual reporting is often limited by time constraints and presents less divergent viewpoints and in-depth analysis compared with print media.17 ,18

    Selection of newspaper articles

    Figure 1 shows the flow of articles through the study. We used the Nexis-UK database, which provides full-text access to all UK national newspapers. Twelve UK national newspapers were included in the sample (January 2009 circulation figures are given in parentheses19): Daily Mirror (1 366 891), Sunday Mirror (1 244 007), The Sun (3 146 006), News of the World (3 031 025), Daily Mail (2 200 398), The Mail on Sunday (2 134 809), The Guardian (358 844), The Independent (215 504), The Observer (427 867), Daily Telegraph (783 210), The Times (617 483) and The Sunday Times (1 198 984). These were selected in order to ensure coverage from tabloid, middle-market and broadsheet publications, daily and Sunday newspapers, and left and right political orientations so that a range of perspectives and reporting styles were represented. This typology has been used in previous content analyses.20 ,21

    Figure 1
    Figure 1

    Flow of articles through study.

    The database was searched using the following terms (an exclamation mark is used as a truncator in this database): H1N1, Influenza A, Swine !flu!, Pandemic !flu!, Pig !flu!. Only articles that contained at least three mentions of the search terms were eligible for inclusion in order to select articles where H1N1 influenza was the main theme. Articles with a different focus entirely, such as business, sports and non-news articles like obituaries, were excluded. Search dates were between 20 April and 5 July 2009, the period in which the major decisions on pharmaceuticals as part of the pandemic response were taken by the UK government. Key events and policy decisions within this period are summarised in table 1.1 ,22 News coverage dropped off considerably after this period.20

    View this table:
    Table 1

    Key events, official risk assessments and UK policy decisions during study period

    Using these criteria, 712 articles were eligible for inclusion. These were extracted into Microsoft Word and screened by one of the authors. Duplicate articles from later editions of the newspapers and any remaining articles as per exclusion criteria above were excluded, leaving 425 articles in the final sample. These provided a good coverage of formats, frequencies and political orientation, taking into account the circulation figures above (figure 1).

    Coding framework

    Each article was assessed independently by two authors using a standardised coding framework consisting of two sections.

    The first section categorised the sources quoted in each article. The main categories consisted of Health Secretary/Minister (of England and the Devolved Administrations—Wales, Scotland and Northern Ireland); Department of Health (of England and the Devolved Administrations); Chief Medical Officer (of England and the Devolved Administrations); World Health Organization (WHO); the UK's Health Protection Agency (HPA), the Centers for Disease Prevention and Control (CDC); pharmaceutical company representative; and named academic (defined here as a researcher or academic clinician affiliated with a higher educational body or research institute in the article).

    The second section looked in greater detail at those articles that quoted academic sources. First, we examined whether academics made a risk assessment of the emerging pandemic. For example, quotes such as “this is going to affect millions of people in England” or “thousands of people could die from this virus” would be a risk assessment. We then checked whether the academic cited official figures or whether there was a risk assessment made by an official body relevant to the UK population quoted within the same article (defined as WHO, Health Secretary/Minister(s), Chief Medical Officer(s), Department(s) of Health or HPA). Table 1 presents examples of risk assessments from these agencies during the study period. We used the official risk assessments as a benchmark to measure each academic risk assessment: judging whether it concurred with the official estimate, or was higher or lower (ie, implying more or less risk to the public).

    All quotes by academics were then examined for reference to the use of NI or influenza vaccine. Those that made reference to NI/vaccine were further analysed as to whether they promoted or rejected the use of these products. The analysis was performed according to a pre-agreed consensus on terms. ‘Promotion’ was defined as advocacy of the effectiveness, need for or supply shortages of NI/vaccine. Conversely, ‘rejection’ referred to statements highlighting the adverse effects, ineffectiveness of or lack of need for NI/vaccine.

    The coding framework was piloted on 20 articles by both coders, with subsequent minor modifications made to definitions before coding of the complete data set. Cohen's kappa was calculated to determine inter-rater agreement between the qualitative measures of risk assessment and promotion/rejection of pharmaceutical products.23 ,24 Disagreements between coders were assessed by a third researcher for final arbitration. Microsoft Excel was used for all coding and calculations.

    Evidence of CoI

    For each named academic, we performed a comprehensive search for CoI based on the protocol from a recent study examining CoI in authors of clinical practice guidelines.25 This was undertaken by two researchers who did not take part in the coding in order to minimise bias. We used the International Committee of Medical Journal Editors’ definition that “Conflicts of interests exists when an author (…) has financial or personal relationships that inappropriately influence (bias) his or her actions)”.26 For each academic, we looked for associations with pharmaceutical or biotechnology companies, in the form of grants (including research), honorariums, speakers’ fees, consultant/adviser/employee relationships and stock ownership.25 These could be personal, indicating benefit to that individual (eg, honorariums), or non-personal, indicating benefit to a department or organisation for which an academic has managerial responsibility (eg, research grants).16 We searched for CoI from the 4 years before the start of the pandemic, that is, March 2005 to March 2009. This is consistent with the WHO's standard that CoI should be declared if incurred in the 4 years before acting in an expert advisory role.25 ,27

    For each academic, we made the following searches in a sequential manner, stopping after each stage if a CoI was identified:

    • The CoI statements (where available) for four major scientific advisory committees relevant to this issue: Joint Committee on Vaccination and Immunisation (UK), Scientific Advisory Group on Emergencies (UK), WHO Emergency Committee and WHO Strategic Advisory Group of Experts.

    • Funding sources detailed on the individual's profile page on the website of affiliated institution.

    • A general internet search using Google linking “(name of academic)” with respectively “vaccine”, “neuraminidase inhibitor”, “antiviral”, “Oseltamivir”, “Zanamivir” and the name of the main pharmaceutical companies producing neuraminidase inhibitors (Roche, GlaxoSmithKline) and influenza vaccine (Novartis, GlaxoSmithKline, Baxter International, Sanofi-Pasteur). The list of manufacturers was obtained through the electronic Medicines Compendium (http://www.emc.medicines.org.uk).

    • CoI and funding declarations on all publications in the past 4 years. These were identified through the PubMed/Medline database.

    Two authors identified CoI, and a separate author verified the presence of CoI.

    We calculated the likelihood of a risk assessment being higher than official estimates if it was made by an academic with CoI compared with those without CoI. As some academics made multiple risk assessments, we used a variant of the generalised linear model (generalised estimating equations, using a binary logistic link function, with an exchange correlation matrix) to take account of clustering.28 We also calculated the likelihood of an academic who promoted or rejected the use of NI/vaccine having CoI compared with academics who provided general commentary, using Fisher's exact test. All statistics were calculated in SPSS V.19.

    Results

    Quoted sources

    Ministers of Health were the most frequently quoted sources (144/425, 33.9% of articles), while academics were the second most commonly quoted (29.7%, 126/425). Other common sources included WHO (27.8%, 118/425), Departments of Health (21.6%, 92/425), HPA (19.1%, 81/425), Chief Medical Officers (16.2%, 69/425) and CDC (5.6%, 24/425). Pharmaceutical companies were quoted in just eight articles (1.9%). A total of 61 named academics were quoted within the sample.

    Risk assessments

    Academics made 74 risk assessments, over half of which were higher than with those made by official agencies in the same article (59.5%, 44/74). In nearly a quarter, 23.0%, 17/74), academics concurred with official risk assessments and in 17.6% (13/74), academics estimated the risk as lower. Table 2 gives some examples of these different categories of risk assessments.

    View this table:
    Table 2

    Examples of risk assessments made by academics and official agencies, by category assigned to academic risk assessment

    Use of NI/vaccine

    Twenty academics commented specifically on the use of NI/vaccine in 36 articles (8.5% of total articles). Ten academics (50%) promoted the use of NI whereas four specifically rejected their use (20%). Nine academics (45%) promoted the use of a vaccine, while none rejected its use. Three academics (15%) promoted the use of both NI and vaccine. Examples of quotes for these categories are illustrated in table 3. Cohen's kappa for inter-rater agreement was 0.66 (values between 0.61 and 0.80 indicate substantial inter-rater agreement).24

    View this table:
    Table 3

    Comments promoting or rejecting the use of neuraminidase inhibitors or vaccine

    Competing interests

    A total of 61 named academics were quoted within the sample. We identified CoI in a third of these academics (29.5%, 18/61), through CoI declarations for scientific advisory committees (5), profile pages (2), internet searches (9) and publications (2). Most CoI were personal in nature (13/18, 72.2%), consisting of paid consultancies or advisory roles, directorships or stock in companies specialising in antiviral products. Seven CoI were non-personal (38.9%), relating to research grants or commercial work funded by pharmaceutical companies. Two academics held both personal and non-personal CoI.

    Out of the 44 risk assessments that were higher than official sources, 35 were made by academics with CoI. In contrast, 10 of the 30 risk assessments that concurred with or were lower than official sources were made by academics with CoI. As several academics made more than one risk assessment, data were fitted using generalised equalising equations. In this analysis, risk assessments were categorised as either being higher than official estimates or concurring with/lower than the official position, forming a binary dependent variable. The best-fitting model revealed that for risk assessments made by academics with CoI the odds of a higher risk assessment were 5.8 times greater compared with assessments made by academics without CoI (Wald p value=0.009).

    Out of the 20 academics who commented on the use of NI/vaccine in the pandemic, one in two had CoI (10, 50%). This is a higher proportion than the one in three academics on the WHO's Emergency Committee advisory group who declared CoI.8

    When we correlated CoI by type of comment, 7 out of 10 academics (70%) promoting the use of NI had CoI compared with 10 out of 47 (21.3%) of academics not commenting on their use (table 4). The odds of COI in academics promoting the use of NI were 8.4 times greater than for academics not commenting on the use of NI (Fisher's exact p=0.005). The odds of CoI in academics rejecting the use of NI were not significantly different to the odds in those not commenting their use (Fisher's exact p=1.0). Five out of nine academics (55.6%) promoting the use of a vaccine had a CoI compared with 13 out of 52 (25.0%) not commenting on its use, a non-significant trend (Fisher's exact p=0.11).

    View this table:
    Table 4

    Number of academics with competing interests by type of comment

    Only three articles in the entire sample noted that the quoted academics had a potential conflict of interest, with one columnist commenting that, “it would be helpful if newspapers informed us of these things”.

    Discussion

    During the period in which the UK government took its major decisions on pharmaceutical policy, one in two academics commenting on NI/vaccine use in UK national newspapers had CoI. The odds of CoI in academics promoting the use of NI were 8.4 times greater than for academics not commenting on the use of NI. If academics with CoI made an assessment of the risk of the pandemic, the odds of this risk assessment being higher than official sources were 5.8 times greater compared with assessments made by academics without CoI.

    CoI among academics on influential advisory committees have led to intense debate worldwide.2 ,6 ,7 This study estimates, for the first time, the prevalence of CoI among academics providing media commentary during the early H1N1 pandemic. We combined a rigorous search for CoI with a comprehensive sample of nationally prominent media during a critical policymaking period. Our findings are based on a small sample, however, and should be viewed as a scoping study. They are corroborated by a study by Moynihan et al29 examining news coverage of three medications for non-communicable diseases, which found that out of 170 stories citing an expert or a scientific study, 50% (85) cited those with a financial tie to the drug manufacturer. Indeed, a study looking at UK newspapers’ representations of the H1N1 pandemic found little discussion of the profits that pharmaceutical companies would make from the development of a H1N1-specific vaccine and few articles describing the potential side effects of vaccines.20

    It is clear from our results that academics constitute an accessible and trusted source for journalists. Academics were the second most commonly quoted source after Ministers of Health, and therefore hold a unique and powerful position for communication on emerging public health issues. However, in a third of cases, academics estimated the risk of the emerging pandemic as higher than official sources. We recognise that academics may be involved in modelling outcomes based on early estimates and may therefore predict higher risks than is borne out by more comprehensive data. In addition, journalists may seek out divergent viewpoints in order to provide balance within a story or to increase its newsworthiness. However, consensus among risk assessors during public health emergencies is important to decrease public anxiety, increase the effectiveness of risk communication and promote adherence to personal protective measures.30–32 We would suggest that this responsibility extends to the media as well, who may need to balance their investigative role with the need to provide a clear and consistent message during the early stages of a public health emergency. Indeed, content analyses of UK20 and European media reporting on H1N1 influenza33 found predominantly factual reporting with little evidence of sensationalism.

    Our results provide some evidence that the provision of higher risk assessments and the promotion of NI are associated with CoI among academics. These add to the growing body of literature highlighting the potential influence of the pharmaceutical industry on policy decisions through multiple avenues, including advisory committees6, drafting of guidelines25 and media commentary.16 This type of influence may be stronger for more familiar health issues, such as cancer, as the public response to emerging health risks is usually one of scepticism.30 Indeed, uptake of H1N1-specific vaccine during the pandemic among those in clinical risk groups was only 37.6%,34 which suggests that both the official vaccination campaign and any media support for vaccination had limited impact.

    There were several limitations to our study. Although this sample was drawn from a large number of articles, the number of academics actually commenting on the use of NI/vaccine was small. More quotes may have been obtained if the study period was extended to the end of the H1N1 pandemic in the UK, but any CoI would be less relevant after the main decisions on pharmaceutical products were taken. While newspaper articles provide a limited set of quotes, the actual interviews with academic sources were undoubtedly longer and may have contained more nuanced views than those represented by the quotes. The definitions and coding of promotion/rejection could be criticised as subjective, although similar definitions have been used in other content analyses.35 Finally, we performed a comprehensive search for CoI, but there may be further conflicts (disclosed or undisclosed) that were not identified here.

    Rather than trying to decrease commentary on public health issues from academics with CoI, a pragmatic approach would be to focus on the complete transparency of these interests36 and allow readers in any capacity to judge comments from academics with these in mind. Indeed, there have been repeated calls for journalists to investigate CoI in their quoted sources in science articles.16 ,37 ,38 In the study by Moynihan et al,29 financial ties to drug manufacturers that were disclosed in the scientific literature were only reported in 39% of the news stories. In our analysis, disclosure was present in only 3% of articles, which may reflect the more fast-moving nature of the pandemic news coverage. In spite of potential logistical difficulties, we echo Caulfield16 in his demand that all “reporters should always ask for and researchers should always offer information about [financial associations]”.

    There are, admittedly, limitations to disclosure. Kassirer points out that disclosure currently tells us nothing about the magnitude of CoI.39 In addition, the interpretation of declared CoI can be subtle, as the emphasis is on complete disclosure of any CoI that may potentially influence an author outside of any judgement of their actual influence.40 It is not known whether this distinction would be appreciated by those unversed in the particularities of scientific CoI declarations. Researchers may be understandably reluctant to put this to the test as news stories about scientific CoI are often high profile. In a 10-year analysis of news media coverage of scientific CoI, McCormas and Simone found that nearly 1 in 10 stories appeared on the front page, suggesting a high degree of newsworthiness.40 Finally, journalists themselves may have undisclosed CoI that would impede truly impartial reporting.16

    Despite these obstacles, we would argue that undisclosed CoI degrades public confidence in medical research, to the detriment of the whole scientific community. We would recommend that these principles are extended to more settings. We call on all academics to declare any potential CoI when providing commentary to the mass media. We encourage journalists to ask for and report any CoI in their interviewees, so that readers can judge their comments in full light of the facts. As Caulfield puts it,16 complete transparency should now be the understood standard practice. Through these measures, the academic voice will retain its credibility in public health issues.

    What is already known on this subject

    • Considerable public funding was spent on vaccines and antiviral medication during the 2009 to 2010 A/H1N1 pandemic.

    • Subsequently, there were concerns over competing interests of academics serving on scientific advisory committees during the pandemic.

    • Many academics also provide media commentary on emerging health risks, and the media has been shown to influence public risk perception and demand for new drugs.

    What this study adds

    • Academics with competing interests were more likely to predict a higher risk to the public from the pandemic than official agencies compared with those without any competing interests.

    • Academics promoting the use of antiviral medication were more likely to have a competing interest than those not commenting on its use.

    • Given the evidence of competing interests among academics providing media commentary, these should be declared before media interviews in order for public health to retain its independent voice.

    References

    1. ↵
      1. Hine D
      . Independent review into the response to the 2009 swine flu pandemic. London: The Cabinet Office, 2010.
    2. ↵
      1. Flynn P
      . The handling of the H1N1 pandemic: more transparency needed. Brussels: Council of Europe, 2010.
    3. ↵
      1. Donaldson L,
      2. Rutter P,
      3. Ellis B,
      4. et al
      . Mortality from pandemic A/H1N1 2009 influenza in England: public health surveillance study. BMJ 2009;339:b5213.
      OpenUrlAbstract/FREE Full Text
    4. ↵
      1. Pebody R,
      2. McLean E,
      3. Zhao H,
      4. et al
      . Pandemic influenza A (H1N1) 2009 and mortality in the United Kingdom: risk factors for death, April 2009 to March 2010. Euro Surveill 2010;15:pii=19571.
    5. ↵
      1. Jefferson T,
      2. Jones M,
      3. Doshi P,
      4. et al
      . Neuraminidase inhibitors for preventing and treating influenza in healthy adults and children. Cochrane Database Syst Rev 2012;(1):CD008965.
    6. ↵
      1. Cohen D,
      2. Carter P
      . WHO and the pandemic flu “conspiracies”. BMJ 2010;340:c2912.
      OpenUrlFREE Full Text
    7. ↵
      1. Epstein H
      . Flu warning: beware the drug companies! The New York Review of Books. New York: The New York Review of Books, 2011.
    8. ↵
      World Health Organisation. List of Members of, and Advisor to, the International Health Regulations (2005) Emergency Committee concerning Influenza Pandemic (H1N1) 2009. Secondary List of Members of, and Advisor to, the International Health Regulations (2005) Emergency Committee concerning Influenza Pandemic (H1N1) 2009 2010. http://www.who.int/ihr/emerg_comm_members_2009/en/index.html (accessed Jul 2013).
    9. ↵
      1. Godlee F
      . Conflicts of interest and pandemic flu. BMJ 2010;340:c2947.
      OpenUrlFREE Full Text
    10. ↵
      1. Coleman C-L
      . The influence of mass media and interpersonal communication on societal and personal risk judgements. Communic Res 1993;20:611–28.
      OpenUrlAbstract
    11. ↵
      1. Booth C,
      2. Dranitsaris G,
      3. Gainford M,
      4. et al
      . External influences and priority-setting for anti-cancer agents: a case study of media coverage in adjuvant trastuzumab for breast cancer. BMC Cancer 2007;28:110.
      OpenUrl
    12. ↵
      1. Otten A
      . The influence of the mass media on health policy. Health Affairs 1992;11:111–8.
      OpenUrlFREE Full Text
    13. ↵
      1. Benelli E
      . The role of the media in steering public opinion on healthcare issues. Health Policy 2003;63:179–86.
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
      1. Ferner R,
      2. McDowell S
      . How NICE may be outflanked. BMJ 2006;332:1268.
      OpenUrlFREE Full Text
    15. ↵
      1. Barratt A,
      2. Roques T,
      3. Small M,
      4. et al
      . How much will Herceptin really cost. BMJ 2006;333:1118–20.
      OpenUrlFREE Full Text
    16. ↵
      1. Caulfield T
      . The Commercialisation of Medical and Scientific Reporting. PLoS Med 2004;1:e38.
      OpenUrlCrossRefPubMed
    17. ↵
      1. Adelman R,
      2. Verbrugge L
      . Deaths makes news: the social impact of disease on newspaper coverage. J Health Soc Behav 2000;41:347–67.
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      1. Stephens M
      . The history of news. New York: Oxford University Press, 2007.
    19. ↵
      Audit Bureau of Circulations (UK). National newspapers & bulk distribution: January 2009. Berkhamsted: Audit Bureau of Circulations, 2009.
    20. ↵
      1. Hilton S,
      2. Hunt K
      . UK newspapers’ representations of the 2009–10 outbreak of swine flu: one health scare not over-hyped by the media? J Epidemiol Community Health 2011;65:941–6.
      OpenUrlAbstract/FREE Full Text
    21. ↵
      1. Hilton S,
      2. Hunt K,
      3. Langan M,
      4. et al
      . Newsprint media representations of the introduction of the HPV vaccination programme for cervical cancer prevention in the UK (2005–2008). Soc Sci Med 2010;70:942–50.
      OpenUrlCrossRefPubMed
    22. ↵
      Health Protection Agency. The role of the Health Protection Agency in the ‘containment’ phase during the first wave of pandemic influenza in England in 2009. London: Health Protection Agency, 2010.
    23. ↵
      1. Lowry R
      . Kappa as a Measure of Concordance in Categorical Sorting. VassStats: Vassar College Statistical Computation Website. Secondary Kappa as a Measure of Concordance in Categorical Sorting. VassStats: Vassar College Statistical Computation Website 2001–2010. http://faculty.vassar.edu/lowry/kappa.html (accessed Jul 2013).
    24. ↵
      1. Landis J,
      2. Koch G
      . The measurement of observer agreement for categorical data. Biometrics 1977;33:159–74.
      OpenUrlCrossRefPubMedWeb of Science
    25. ↵
      1. Neumann J,
      2. Korenstein D,
      3. Ross J,
      4. et al
      . Prevalence of financial conflicts of interest among panel members producing clinical practice guidelines in Canada and United States: cross sectional study. BMJ 2011;343:d5621.
      OpenUrlAbstract/FREE Full Text
    26. ↵
      International Committee of Medical Journal Editors. Ethical Considerations in the Conduct and Reporting of Research: Conflicts of Interest. Uniform Requirements for Manuscripts Submitted to Biomedical Journals. Secondary Ethical Considerations in the Conduct and Reporting of Research: Conflicts of Interest. Uniform Requirements for Manuscripts Submitted to Biomedical Journals 2009. http://www.icmje.org/ethical_4conflicts.html (accessed Jul 2013).
    27. ↵
      World Health Organization. Declaration of Interests for WHO experts. http://keionline.org/node/1062 (accessed Jul 2013).
    28. ↵
      1. Hanley JA,
      2. Negassa A,
      3. deB. Edwardes MD,
      4. et al
      . Statistical analysis of correlated data using generalized estimating equations: an orientation. Am J Epidemiol 2003;157:364–75.
      OpenUrlAbstract/FREE Full Text
    29. ↵
      1. Moynihan R,
      2. Bero L,
      3. Ross-Degnan D,
      4. et al
      . Coverage by the news media of the benefits and risks of medications. N Engl J Med 2000;342:1645–50.
      OpenUrlCrossRefPubMedWeb of Science
    30. ↵
      1. Rubin G,
      2. Amlot R,
      3. Page L,
      4. et al
      . Public perceptions, anxiety, and behaviour change in relation to the swine flu outbreak: cross-sectional telephone survey. BMJ 2009;339:b2651.
      OpenUrlAbstract/FREE Full Text
    31. ↵
      1. Freudenburg W
      . Perceived risk, real risk: social science and the art of probabilistic risk assessment. Science 1988;242:44–9.
      OpenUrlAbstract/FREE Full Text
    32. ↵
      Department of Health. Pandemic flu: a national framework for responding to an influenza pandemic. London: Department of Health, 2007.
    33. ↵
      1. Duncan B
      . How the media reported the first days of the pandemic (H1N1) 2009: results of EU-wide media analysis. Euro Surveill 2009;14:pii=19286.
    34. ↵
      1. Sethi M,
      2. Pebody R
      . Pandemic H1N1 (Swine) influenza vaccine uptake amongst patient groups in primary care in England 2009/10. London: Department of Health, 2010.
    35. ↵
      1. Blair N,
      2. Yue S,
      3. Singh R,
      4. et al
      . Depictions of substance use in reality television: a content analysis of The Osbournes. BMJ 2005;331:1517–9.
      OpenUrlAbstract/FREE Full Text
    36. ↵
      1. Leung G,
      2. Nicoll A
      . Reflections on pandemic (H1N1) 2009 and the international response. PLoS Med 2010;7:e1000346.
      OpenUrlCrossRefPubMed
    37. ↵
      1. Schwitzer G,
      2. Mudur G,
      3. Henry D,
      4. et al
      . What are the roles and responsibilities of the media in disseminating health information? PLoS Med 2005;2:e215.
      OpenUrlCrossRefPubMed
    38. ↵
      1. Shuchmann M,
      2. Wilkes M
      . Medical scientists and health news reporting: a case of miscommunication. Ann Intern Med 1997;126:976–82.
      OpenUrlCrossRefPubMedWeb of Science
    39. ↵
      1. Kassirer J
      . On the take: how medicine's complicity with big business can endanger your health. New York: Oxford University Press, 2004.
    40. ↵
      1. McComas K,
      2. Simone L
      . Media coverage of conflicts of interest in science. Sci Commun 2003;24:395–419.
      OpenUrlAbstract

    Supplementary materials

    • Press release notice

      Files in this Data Supplement:

    Footnotes

    • Correction notice The license of this article has also changed since publication to CC BY 4.0.

    • Contributors KM conceived and designed the study, and collected initial data. SON and KC performed the content analysis. AB and AW performed the search for competing interests. KM, KC, SON and KY analysed the data. KY performed the statistical analysis. KM wrote the first draft of the manuscript, and all authors contributed to and approved the final manuscript.

    • Funding KM is funded by the Wellcome Trust (grant number 09401). The funder had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.

    • Competing interests None.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data sharing statement Unpublished data from the content analysis are available from the authors.