Initially, 749 newspaper articles were retrieved, of which 161 articles, 74 from UK papers and 87 from US newspapers remained following removal of duplicates and excluded articles (Figure 1). Inter-rater kappa values ranged from 0.5 to 1.0, indicating moderate to perfect agreement between coders . The mean inter-rater kappa value across all of the coded variables was 0.65; this is similar to the level of inter-rater agreement reported in previous quantitative content analyses involving the newspaper media [19, 25].
Frequency of newspaper reporting
The number of articles retrieved from UK and US newspapers ranged from 7 in 2012 to 24 in 2009, indicating a low publication frequency relating to this topic. The highest numbers of articles were published in 2003 (21 articles), 2008 (21 articles) and 2009 (24 articles), with a marked decline in reporting on online health information after 2009 (Figure 2). In the UK, the overall trend in reporting remained relatively constant throughout the 10-year period, while an overall downward trend was observed in the US. During the same period there was an increase in scientific articles on this topic archived in PubMed® (Figure 3), indicating that, during this time, online health information is a topic that has been researched actively.
Newspaper type and positioning of articles
Strictly speaking, the terms ‘broadsheet’ and ‘tabloid’ refer to newspaper dimensions, however, broadsheet newspapers are perceived to be more intellectual in content in comparison to tabloids, which tend to report more sensationalist and celebrity material. Articles relating to online health information were published more frequently in ‘broadsheet’ newspapers than in ‘tabloid’ newspapers. Indeed, only one relevant article was found in the US tabloid press over the entire 10-year period of interest. In the UK, on average, 4.9 articles (SD 2.8) were published in broadsheet newspapers per year, which was significantly higher than the average of 2.5 (SD 2.0) articles published in tabloid newspapers annually (p = 0.04). In approximately two thirds (68.3%) of articles, it was obvious from the headline that the article related to health information on the Internet. Approximately a quarter of the selected articles (24.8%) were published in health sections and approximately one fifth appeared in feature (18.6%) and business (18.0%) sections. Interestingly, on only one occasion did the topic feature in the editorial/leader section, indicating the low priority given to the topic by newspaper editors.
Authorship and information sources
Journalists wrote a substantial proportion of the articles (83.2%) and health professionals wrote relatively few (9.9%), although, in approximately a quarter (22.4%) of the articles authored by journalists, a health professional was cited as the main source of the information. Other sources included published reports or articles, or their authors (18.6%), spokespersons from the IT industry (12.4%) or from a Government/National Health Service (NHS) department (9.3%). Thirty articles were informed by a scientific report or journal article. The most frequently cited reports were those published by the Pew Internet and American Life Project. Almost two thirds of the articles (62.7%) included quotations from patients, medical or industry experts.
Content of newspaper articles
Online health information was the main theme in the majority (65.2%) of articles. Other themes included the Internet as a medium for health-related communication between the public and/or health professionals (11.8%), access to online personal health records (8.7%), developments in Internet technology (5%) and online disease management tools (4.3%). The majority of articles (67.1%) mentioned or recommended specific web sites. In approximately one fifth (19.3%) of articles, the main focus was on the Internet as a channel for conveying health information in a public health context, for example, during the 2009 swine flu pandemic USA Today reported: “Internet users have ramped up their searching, chatting and blogging of up-to-the-minute news on the symptoms and spread of swine flu since its sudden appearance this month. It's a trend health experts say is effective in rapidly pushing out public health information, using technology not available during the deadly, worldwide flu outbreaks of decades past” (Gillum J. “People mine Net for everything flu; technology provides wealth of information – not all scientific”. USA Today. 29 April 2009; News, p7a).
Approximately half (49.1%) of the selected articles linked online health information to specific diseases, disease groups (e.g. cancer) or general health topics (e.g. women’s health). Using the BNF classification, the most frequently mentioned diseases related to the central nervous system (Table 1). Diabetes was the most frequently mentioned single disease, cancer the commonest group of diseases and sexual health was the most common general health topic. There was no significant difference between UK and US newspaper reporting in relation to the frequencies of mentioning diseases in each of the BNF classifications (p > 0.05). In addition, lifestyle issues, such as weight loss, alcohol consumption and exercise featured in approximately one fifth (19.2%) of the articles and ten articles focused on the Internet as an information source during pregnancy.Overall, 80% of articles mentioned benefits and 55% mentioned risks associated with health information on the Internet. Public access to health information was the most frequently reported benefit (64%) and access to misleading information was the most frequently cited risk (39.8%) (Figure 4). Most articles (41%) were written with a mixed slant, portraying benefits and risks equally. A slightly smaller proportion (38.5%) was positively slanted, i.e. mainly expressing benefits, and relatively few articles had a negative (11.2%) or neutral (9.3%) slant (i.e. no benefits or risks expressed). Interestingly, articles in US newspapers mentioned benefits more often than UK articles (81.6% vs. 77.0%) and risks less often (50.6% vs. 59.5%), although these differences were not significant (p > 0.05).
There was no significant difference between UK and US newspapers in the frequency of reporting of facilitators and barriers to using online health information in routine clinical practice (p > 0.05). Facilitators were mentioned in 55.3% of articles (Figure 5); ease of Internet access and the expression of positive views by health professionals were the most frequently reported facilitators, for example “We need to help them sort through it, not discourage the use of information. We have to acknowledge that patients do this research. It's important that instead of fighting against it, that we join them and become their coaches in the process” (Parker-Pope, T. You’re sick. Now what? Knowledge is power. The New York Times. 30 September 2008; Science Desk, p1). Barriers were stated in 37.3% of articles (kappa = 0.5); the most frequently cited barrier was the negative viewpoint of health professionals “Some doctors are less enthusiastic. People think all they need is some basic medical information and off they go. They even suggest that doctors could soon be out of a job" (Bird J. ‘More like a conversation between equals’. The Financial Times. 27 June 2011; FT Health, p3).
Balance and quality of newspaper reporting
The majority of articles (83.9%) were rated as having balanced judgement, i.e. the authors neither made exaggerated nor understated claims in comparison with the generally accepted status of online health information. The quality of information presented in each article was rated with the aid of descriptors on a scale of 1-10. Higher scores indicated higher quality reporting. A typical high quality article had balanced judgement, was based on evidence, and included quotations from subject experts, whereas, an article was rated as poor if it was anecdotal, lacked balanced judgement and did not include any evidence in support of its claims. Overall, 47.2% of the articles were rated as having excellent quality reporting (scored 8-10), 32.9% presented average/good quality information (scored 4-7) and 19.9% reported poor quality information (scored 1-3). We found no difference in the quality of reporting in UK articles compared to US articles (p > 0.05).