OBJECTIVE: In 2004, after an 18-month investigation, the Food and Drug Administration (FDA) directed pharmaceutical manufacturers to add a black box warning to antidepressants regarding increased risk of suicidality in children. Our objective was to evaluate the quality and content of and overall impression conveyed in news coverage of this issue.
METHODS: We collected all news stories on pediatric antidepressant use and suicidality published in a convenience sample of 10 of the highest-circulation print newspapers in the United States, the 3 major television networks, and a major cable news network in 2003 and 2004 (N = 167). Two researchers coded news articles by using a 9-item instrument.
RESULTS: The quality of news reporting on key health messages included in FDA warnings was mixed. The overwhelming majority of news stories correctly described a risk of suicidality associated with pediatric antidepressant use, as opposed to suicide itself. However, other key health messages highlighted in FDA warnings often were absent from news coverage. News stories were more likely to include anecdotes of children harmed versus children helped by antidepressants, whereas expert sources quoted were more likely to emphasize the benefits of antidepressants over their risks. However, the majority of news stories conveyed neither the overall impression that the risks of pediatric antidepressant use outweighed the benefits nor the impression that the benefits outweighed the risks, and coverage became increasingly neutral over time.
CONCLUSIONS: Inclusion of key health messages in FDA safety warnings was not sufficient to ensure their communication to the public through the lay press.
The news media serve a vital function in conveying health care safety information to the public. More than one half of the US public describes national, local, or cable news as the most important source of health information.1 In situations in which there is uncertainty regarding the risks and benefits of medical care, the public looks to the media to flag potential safety problems, to interpret scientific data, and to identify qualified experts to comment on how to weigh risks and benefits. Health care providers also may rely on the news media for timely health safety information.2 Given the prominent role of the news media as an information source for the public and health professionals, assessment of the quality and content of news reporting is essential. The disclosure of new risk information on pediatric antidepressant use and suicidality provides an opportunity to evaluate print and television news reporting on a health safety issue in the context of considerable scientific uncertainty.
Safety concerns regarding pediatric antidepressant use surfaced in 2003, when the manufacturer of the antidepressant Paxil (GlaxoSmithKline, Research Triangle Park, NC) identified evidence of increased risk of suicidality in clinical trial data. The Food and Drug Administration (FDA) issued an advisory in June 2003, warning of possible risks associated with pediatric use of Paxil. In October 2003 and March 2004, the agency issued additional safety warnings regarding a broader set of antidepressants. In September 2004, a FDA advisory panel was presented with results from a FDA-sponsored meta-analysis that found that the rate of suicidality among children assigned to receive selective serotonin reuptake inhibitors (SSRIs) was twice that of the placebo-treated group.3 This evidence led to a 15:8 decision by the panel to recommend a black box warning related to the increased risk of suicidality in pediatric patients for all antidepressant drugs. This recommendation was adopted by the FDA and announced to the public in October 2004.
Supporters of the FDA labeling change argued that evidence of elevated risks of suicidality linked to use of antidepressants by youths was sufficiently serious to warrant warning providers and consumers. Critics countered that no children in the trials actually committed suicide and the FDA actions would reduce the use of an effective treatment for depression, thereby yielding poorer mental health outcomes in an undertreated population. Adding to the uncertainty, several studies using observational data found that pediatric SSRI use was negatively associated with suicide (ie, protective) or that there was no significant association between antidepressant use and suicide.4–9
Concurrently with the FDA investigation and release of safety warnings, dramatic decreases in pediatric antidepressant use occurred (S.H.B., R. G. Frank, PhD, C.L.B., and A. Martin, MD, unpublished data, 2009).10–15 It has been suggested that reporting on the issue by the news media played a critical role in the severity of these decreases,16,17 although to our knowledge no studies have examined the quality or content of coverage in the media. Previous research offered conflicting evidence on how well the news media function as a conduit for communicating health risk information to the public. Moynihan et al18 found that news coverage of medications often included inadequate or incomplete information about the benefits, risks, and costs. Schwartz and Woloshin19 compared news coverage of potential benefits and harms of tamoxifen and found that the negative aspects of tamoxifen received greater emphasis. Another study provided evidence on the positive role of the news media in transmitting warnings about the association between pediatric aspirin use and Reye syndrome.20
We examined the quality and content of and overall impression conveyed in news coverage of pediatric antidepressant use and suicidality risk in 2003 and 2004. We measured the quality of news reporting by determining whether key health messages emphasized in FDA warnings were mentioned in news coverage. We also examined news content related to the inclusion of anecdotes and expert source quotations and the overall impression of risks and benefits conveyed in news reports. In a recent commentary on health care journalism, Dentzer characterized the role of the news media as “delivering to the public accurate, complete, and balanced messages about health,” rather than solely reporting the news, interpreted narrowly as “that which is new.”2 Our goal was to assess whether the news media achieved this standard in reporting on pediatric antidepressant use.
News Sources Studied
We collected all news stories on pediatric antidepressant use and suicidality reported by a convenience sample of 10 of the highest-circulation print newspapers in the United States, the 3 major television networks (ABC, NBC, and CBS), and a major cable news network (CNN) in 2003 and 2004 (N = 167). The news outlets included in our study were USA Today, Wall Street Journal, New York Times, Los Angeles Times, Washington Post, Chicago Tribune, New York Daily News, Philadelphia Inquirer, Denver Post/Rocky Mountain News, and Houston Chronicle. We used data from the Audit Bureau of Circulation to identify daily US newspaper circulation rates21 and data from Nielsen Media Research to identify television viewership.22 We chose a convenience sample of news outlets with large circulation or viewership to analyze news coverage transmitted to a large subset of the US news-viewing public. In 2004, the daily circulation for the sampled newspapers exceeded 10 million, the evening and morning network news programs were watched by 17 million and 15 million viewers, respectively, and the average daytime and primetime viewership for CNN was 490500 and 799750 viewers, respectively.
News Coverage Selection
We used Lexis-Nexis, Factiva, and newspaper online archives to collect newspaper articles and Lexis-Nexis to collect television transcripts, by using the search terms “suicide,” “antidepressant,” and “children.” For a story to be included, all 3 terms or a close variant (eg, suicidality, SSRI, or pediatric) needed to appear in the story. We identified 327 newspaper articles and 191 television transcripts that met these search criteria. Our interest was in analyzing only hard-news stories, feature stories, or health column stories. We excluded stories coded as editorials, corrections, book reviews, letters to the editor, business/stock information, question-and-answer format, duplicate wire stories, indices, or obituaries. We also excluded stories in which antidepressants and suicidality were not the primary focus and stories in which children or adolescents were not the primary focus. All exclusions were determined by Drs Barry and Busch, yielding a total of 167 stories analyzed.
To analyze the quality and content of and overall impression conveyed in each news story, we developed a 9-item coding instrument. The instrument was pilot-tested by the authors, and minor adjustments were made to clarify question wording and item response coding. After the instrument was finalized, each of the 167 articles was coded independently by each author. Interrater reliability for all items, measured with κ statistics, was high. Raw agreement for a single, more-subjective item assessing overall impression was 92% (κ = .84). Raw agreement was higher for the other items, ranging from 93% to 99% (κ = .85–.99). To resolve coding disputes regarding specific items, the authors discussed and reached agreement on the final coding. There were no items for which agreement could not be reached.
We included 4 measures of quality in news reporting. First, we examined whether news reports described correctly the association between pediatric antidepressant use and suicidality (versus suicide). Reporting an association between antidepressant use and suicide itself would constitute poor-quality news reporting. We also examined inclusion of other key health messages emphasized in FDA safety warnings, including the importance of monitoring a child using antidepressants, the importance of tapering medication doses during antidepressant discontinuation, and the fact that fluoxetine is the only FDA-approved pharmaceutical treatment for pediatric depression. We chose to be expansive in our measure of monitoring and included references to medication monitoring as well as more-general references to parent or physician monitoring. We noted that the warning stressed not only that fluoxetine was the only drug approved for pediatric depression but also that it was the only drug with proven efficacy in this population. Before FDA action, all 3 pieces of health information were available to the medical profession. However, because the FDA emphasized these health messages in each of its advisories beginning in October 2003, we viewed their inclusion in coverage in the media as an indicator of high-quality reporting.
Next, we identified whether the news story referenced an individual child/adolescent who it was claimed was harmed or helped by antidepressant use. To be included, the child/adolescent needed to be referenced specifically by name or as the child of a named parent. For stories referencing an individual child, we examined whether children harmed or helped were referenced more often. We also identified whether the news story included ≥1 quote from an expert source suggesting that the risks of pediatric antidepressant use may outweigh the benefits or that the benefits may outweigh the risks. For news stories including an expert quote on risks and benefits, we examined whether risk quotations or benefit quotations were more likely to be included.
Finally, we included 1 measure to assess whether the news story left the overall impression that the risks of pediatric antidepressant use outweigh the benefits, the benefits of pediatric antidepressant use outweigh the risks, or neither. We chose a conservative approach and, when only a weak overall impression of risk or benefit was conveyed, the story was coded as neither.
We analyzed the quality and content of and overall impression conveyed in news coverage in 2003 and 2004 and during the 2-week periods after 3 FDA actions, namely, the March 2004 advisory, the September 2004 panel vote, and the October 2004 decision. The news story was the unit of analysis. We tested differences between proportions of print and television news stories by using logistic regression. We adjusted SEs for lack of independence within news source. All analyses were conducted with Stata 9 (Stata, College Station, TX).
As indicated in Table 1, the study included 126 newspaper and 41 television stories reporting on pediatric antidepressant use and suicidality in 2003 and 2004. Figure 1 shows that the volume of news stories in our sample increased after FDA actions and that peak coverage occurred with the vote of the FDA advisory panel to recommend issuance of a black box warning.
Key Health Messages Emphasized in FDA Warnings
Table 2 indicates that the quality of news reporting on key health messages emphasized in FDA warnings was mixed. The overwhelming majority (98%) of news stories correctly described a risk of suicidality associated with pediatric antidepressant use, as opposed to suicide itself. However, other key health messages highlighted in FDA warnings often were absent from news coverage. Only 43% noted the importance of monitoring children using antidepressants, 13% mentioned that medication doses should be tapered when administration is discontinued, and 40% mentioned fluoxetine as the only FDA-approved pharmaceutical treatment for pediatric depression. Generally, news stories in the 2 weeks after FDA actions were no more likely to mention health information than were those in other time periods. The notable exception was the high level (88%) of coverage noting the importance of monitoring after the March 2004 advisory.
Use of Anecdotes and Expert Quotations
Table 3 indicates that television news was significantly more likely than print news to include anecdotes involving individual children's experiences with antidepressants, and both print and television new stories were more likely to include references to specific children harmed (35%) than helped (14%). Among news stories including an anecdote about a child (N = 66), 89% mentioned a child who was harmed, whereas only 36% mentioned a child who was helped.
In contrast, fewer news stories included a quote from an expert source suggesting that the risks of pediatric antidepressant use outweighed the benefits (20%) than suggested that the benefits outweighed the risks (32%). Among just the subset of news stories including an expert quote mentioning risks and benefits (N = 73), a greater proportion mentioned that the benefits of antidepressant use outweighed the risks (74%) than that the risks of antidepressant use outweighed the benefits (47%). We did not detect statistically significant differences in expert source quotations related to risks and benefits for print versus television news.
Overall Impression of Risks and Benefits
The majority (62%) of stories conveyed neither the overall impression that the risks of pediatric antidepressant use outweighed the benefits nor that the benefits outweighed the risks (Table 4). The 38% of news stories that were not neutral were significantly more likely to convey the impression that the risks of pediatric antidepressant use outweighed the benefits (35%) than that the benefits outweighed the risks (4%). News stories became more neutral over time in their portrayal of risks and benefits. After the March 2004 advisory, 53% of news stories emphasized the risks of antidepressant use, compared with only 8% of news stories after the October 2004 decision.
The purpose of this study was to assess the quality and content of and overall impression conveyed in news reporting on FDA warnings pertaining to pediatric antidepressant use, among national and regional news sources with large circulation or viewership. We found mixed evidence on the quality of news reporting. The overwhelming majority of news reports accurately described the association between pediatric antidepressant use and suicidality rather than suicide itself, an indicator of good-quality reporting. However, other key health messages highlighted in FDA warnings, pertaining to the importance of monitoring, the need to taper doses during medication discontinuation, and FDA approval of fluoxetine to treat depression in children, often were absent from news reports.
These findings suggest that including key health messages in FDA safety warnings was not sufficient to ensure communication to the public through the lay press, although this information might have mitigated the risks of pediatric antidepressant use. One research study found that monitoring did not increase after warnings, despite the emphasis of the FDA on monitoring,23and use of fluoxetine by incident child antidepressant users was still only 27% in September 2004 (up from 15% in October 2003) (S.H.B., R. G. Frank, PhD, R. Rosenheck, MD, D. Leslie, PhD, C.L.B., A. Martin, MD, E. G. Martin, PhD, unpublished data, 2009). Beyond issuing safety warnings and contraindicating medications, the FDA relies on a number of other methods to inform consumers and providers about safety risks, including the use of medication guides that include practical, reader-friendly, product safety information. However, the agency depends in part on the press to fulfill its mandate to communicate risk and benefit information to the public. Simply including key health messages in FDA press releases was not sufficient to ensure that journalists mentioned this information in news stories.
News stories, in particular television news, were more likely to include anecdotes of children harmed by antidepressants than children helped, whereas expert source quotations were more likely to emphasis the benefits of antidepressants over their risks. The finding that television news reporting relied heavily on anecdotes of children harmed by antidepressants is consistent with previous research indicating that television news tends to frame coverage in more-personal terms by emphasizing individual experiences.24 Given evidence that patients may weigh anecdotal versus statistical evidence disproportionately,25 the news coverage use of anecdotes of children harmed might have led to larger decreases in pediatric antidepressant use than would have occurred otherwise.
Although the greater emphasis on the benefits of antidepressants in expert source quotations may serve as a balance to the use of anecdotes, increasing attention to conflicts of interest in medicine raises concerns about journalists' use of physicians or researchers as sources, in the absence of disclosure policies. In this study, none of the news stories coded included specific information about whether the experts interviewed had specific conflicts of interest (eg, received funding from the pharmaceutical industry). This is consistent with a recent study that reported that news articles reporting on medication studies often fail to report pharmaceutical company funding.26 Finally, the majority of news stories conveyed neither the overall impression that the risks of pediatric antidepressant use outweighed the benefits nor the impression that the benefits outweighed the risks, and coverage became increasingly neutral over time.
A number of limitations are worth mentioning. Our results are limited to the extent that the public and medical professionals learned about safety concerns related to pediatric antidepressant use through nontraditional news sources, including blogs and other Internet news sites. Given the current turmoil in the newspaper industry, the public is likely to rely increasingly heavily on these alternative sources of health information in the future. Similarly, it was not feasible to include local television news or smaller newspaper outlets, because these news sources are not readily available to researchers for study. Finally, although families' exposure to risk and benefit information through the news media may explain subsequent treatment patterns, assessing this causal relationship is beyond the scope of our study.
Because depression is an undertreated disease27 with the potential for long-term negative consequences, the sharp decreases in pediatric antidepressant use that occurred in the aftermath of the FDA risk disclosure are troubling, to the extent that they represent an increase in unmet need. Clinical trial data indicated that one antidepressant, fluoxetine, was clinically effective in treating children with major depressive disorder.28 The efficacy of other antidepressants in treating children has not been proved. In the 1990s, increased antidepressant use after the introduction of SSRIs was hailed by many as an important advancement in public health. Steady decreases in youth suicide rates suggested to some that antidepressants were saving lives. This trend reversed unexpectedly with the upswing in national youth suicide rates in 2004 and 2005, leading to the speculation that less antidepressant use after FDA safety warnings might have increased rather than decreased youth suicide rates, although other explanations also are possible (eg, changes in youth access to firearms, the prevalence of alcohol use, and Internet-based social networks).29
In future years, the public may be expected to rely more on both traditional and nontraditional media to obtain health information, given decreasing consumer trust in the medical profession30 and the movement toward a more–consumer-oriented health system. As medical decision-making grows more complex, health care journalists are increasingly challenged to convey complex health information to the public. Further empirical research is needed for better understanding of how news reporting on medical risks and benefits may affect who does and does not receive treatment.
This study was supported by a grant from the National Institute of Mental Health (grant R01 MH 080883).
- Accepted July 18, 2009.
- Address correspondence to Colleen L. Barry, PhD, Yale University, School of Public Health, 60 College St, New Haven, CT 06520. E-mail:
Financial Disclosure: The authors have indicated they have no financial relationships relevant to this article to disclose.
What's Known on This Subject:
Dramatic decreases in pediatric antidepressant use occurred after FDA safety warnings. It has been suggested that news reporting played a critical role in the decreases, although no studies have examined the quality or content of coverage in the media.
What This Study Adds:
We examined the quality and content of and overall impression conveyed in news coverage of pediatric antidepressant use and suicidality risk in 2003 and 2004.
- ↵Kaiser Family Foundation; Harvard School of Public Health. September/October 2001 Health News Index. Menlo Park, CA: Kaiser Family Foundation; 2001
- Hall WD, Mant A, Mitchell PB, et al. Association between antidepressant prescribing and suicide in Australia, 1991–2000: trend analysis. BMJ.2003;326 (7397):1008– 1013
- ↵Valuck RJ, Libby AM, Sills MR, Giese AA, Allen RR. Antidepressant treatment and risk of suicide attempt by adolescents with major depressive disorder. CNS Drugs.2004;18 (15):119– 132
- ↵Rosack J. New data show declines in antidepressant prescribing. Psychiatr News.2005;40 (17):1
- Harris G. Study finds less youth antidepressant use. New York Times. September 21, 2004. Available at: http://query.nytimes.com/gst/fullpage.html?res=990DEFDC1239F932A1575AC0A9629C8B63&sec=health. Accessed November 2, 2009
- ↵Arehart-Treichel J. Suicide attempts decline with psychotherapy or antidepressants. Psychiatr News.2007;45 (15):1
- ↵Satel S. Bad medicine? National Review. September 12, 2004. Available at: www.nationalreview.com/comment/satel200409121424.asp. Accessed November 2, 2009
- ↵Audit Bureau of Circulation. Newspaper circulation figures for the six months ending on March 31, 2004. Available at: www.burrellesluce.com/top100/2004_Top-100List.pdf. Accessed November 2, 2009
- ↵Pew Research Center's Project for Excellence in Journalism. 2005 annual report: network TV audience. Available at: www.journalism.org/node/883. Accessed November 2, 2009
- ↵Iyengar S. Is Anyone Responsible? How Television Frames Political Issues. Chicago, IL: University of Chicago Press; 1991
- ↵Fagerlin A, Wang C, Ubel PA. Reducing the influence of anecdotal reasoning on people's health care decisions: is a picture worth a thousand statistics? Med Decis Making.2005;25 (4):398– 405
- Copyright © 2010 by the American Academy of Pediatrics