Dissemination of 2014 dual antiplatelet therapy (DAPT) trial results: a systematic review of scholarly and media attention over 7 months

Objective To explore how the results from the 2014 dual antiplatelet therapy (DAPT) trial were disseminated to the scientific community and online media. Design A a systematic review of scholarly and public attention surrounding the DAPT study. Settings Data were collected from the ISI Web of Knowledge, Google Scholar, PubMed Commons, EurekAlert, the DAPT study website (www.daptstudy.org) and the New England Journal of Medicine website (for scholarly attention) and Altmetric Explorer, Snap Bird, YouTube (for public attention) citing DAPT study results appearing from 16 November 2014 to 10 June 2015. Participants No participants were involved in this study. Main outcome measure Proportion of contents highlighting the increased risk of mortality and critical to the author’s interpretation of the results. Results We identified 425 items reported by seven sources; 164 (39%) disseminated the authors’ interpretation via an electronic link or a reference, with no additional text. Among 81 items (19 %), the message favoured prolonged treatment and consequently overstated the article conclusions. Among 119 items (28 %), the text was uncertain about the benefit of prolonged treatment but was reported with no or inappropriate mention of increased risk of mortality. Only 34 items (8 %) were uncertain about the benefit of prolonged treatment and mentioned increased risk of mortality. In all, 27 items (6 %) did not favour prolonged treatment, and only 12 of these (3 %) clearly raised some concerns about the reporting of increased risk of death. Conclusion Dissemination of the DAPT study results to the scientific community and on different media sources rarely criticised the interpretation of the study results.


Strengths and limitation of this study
• Our method involved a broad search strategy, which ensured to capture an extensive and representative sample of contents citing DAPT trial for both scholarly and public attention.
• Our systematic approach to analyze the text of contents provides a comprehensive overview of dissemination of this study results.
• This study only focussed on a specific trial publication and results are not generalizable to other studies.

INTRODUCTION
Development of optimal coronary stent replacement has progressed rapidly over recent years 1 . In the United States, almost 700,000 stents are placed every year and there is an increasing trend in Europe in its use 2 . Dual antiplatelet therapy (DAPT) (i.e., P2Y12-receptor inhibitor combined with aspirin) is recommended after placement of coronary stents to prevent thrombotic complications 3 . The optimal duration of DAPT has been debated [4][5][6][7][8] .
In December, 2014, the Harvard Clinical Research Institute (HCRI) released the results of the DAPT study, the largest international randomized controlled trial to date 9 . The trial aimed to determine the benefits and risks of continuing DAPT beyond 1 year after placement of a coronary stent 9 . A total of 9,961 adult patients were randomly assigned to continue thienopyridine treatment or to receive a placebo for 30 months. Continued therapy reduced the rates of stent thrombosis (0.4% vs.1.4%; p<0.001) and major adverse cardiovascular and cerebrovascular events (MACCEs) (2.1% vs. 4.1%; p<0.001) with an expected increase in the rate of moderate or severe bleeding (2.5% vs. 1.6%; p=0.001) 9 . However, continued therapy was also associated with an increase of 36% in all-cause mortality (2.0% vs. 1.5%; hazard ratio 1.36 [95% CI, 1.00 to 1.85]; P=0.05).
The results of the DAPT study were published in the New England Journal of Medicine (NEJM) 9 after the presentation of results at the American Health Association Conference, in November 2014. However, the reporting of the results raised some concerns 10,11 .
Particularly, the abstract conclusions did not mention the increased risk of mortality. Further, the discussion included questionable explanations based on post-hoc analyses to clear the role of prolonged thienopyridine treatment on this increased risk of mortality. For this purpose, the authors had split the analysis by cause of death, which reduced the power to show a statistically significant difference. Then, they focused on the increase in cancer-related death (0.62% vs 0.28%, p=0.02). However, instead of raising the hypothesis that prolonged treatment could increase the risk of cancer or the risk of dying from cancer, they interpreted this finding as being related to an imbalance at baseline in patients with a history of cancer before enrollment (9.8% vs 9.5%). To confirm this hypothesis, the authors performed a posthoc analysis excluding all deaths that could be related to cancer diagnosed before enrolment.
This post-hoc exclusion of patients with an event is a concern. As expected, the results became statistically non-significant (0.50% vs 0.28%, p=0.11). The authors did not mention other studies showing that prasugrel, one thienopyridine used in this trial, has been associated with a significantly increased risk of incident cancer 12 and has been specifically investigated by the US Food and Drug Administration 13 .
Here we aimed to explore how these results from the DAPT trial were disseminated to the scientific community and the public. Particularly, we aimed to determine whether the scholarly and public attention raised by this study highlighted the increased risk of mortality and criticized the authors' questionable interpretation of the findings.

Content of scholarly and public attention surrounding DAPT study
Two researchers (MS, RH) read the items from each source independently and evaluated them by using a preliminarily tested extraction form. Disagreements were resolved by discussion until consensus was reached. If needed, a third researcher (IB) appraised the content.
We determined whether the source consisted of a reference or a link to the NEJM article reporting the DAPT study only or was a text commenting on the DAPT study. For a text commenting on the DAPT study, we checked whether the original study authors were involved in writing the text or not. Our main outcome of interest was the proportion of contents highlighting the increased risk of mortality and critical to the author's questionable interpretation of the results. We determined whether; • the primary efficacy outcomes (i.e., stent thrombosis and MACCE) were reported • the safety outcomes related to moderate or severe bleeding were reported • the increased risk of mortality with prolonged treatment was reported • the authors' questionable explanation clearing the responsibility of prolonged treatment in the increased risk of mortality was reported or criticized.
• the content of the text was 1) favourable about the prolonged treatment and consequently overstating the article conclusion, 2) uncertain about the benefit of the prolonged treatment (i.e., statement of both the beneficial effect, and increased risk of bleeding, text ending with a question mark, use of "may or might" or reporting that the study needs further research), or 3) not favourable about the prolonged treatment 18 .

Identification of scholarly and public attention surrounding DAPT study
From all sources, we selected and appraised 425 items: 118 communications, 12 news items, 3 blogs, 189 Facebook posts or comments, 75 tweets or replies, 8 videos on YouTube, 14 DAPT media pages, 5 DAPT website pages and 1 video on the DAPT website ( Figure 1). The original study authors were directly involved in 35 items.

Reporting of the content
The items are described in Figures 2 and 3. Overall, 164 items (39%) involved disseminating the authors' questionable reporting and interpretation via an electronic link (n=151, 36%) or reference (n=13; 3%), with no additional text or message. Among 81 items (19%), the message favoured the prolonged treatment and therefore overstated the article conclusions.
For example, the DAPT study website dedicated to patients reported that "It is important that patients who currently take a thienopyridine anti-clotting medication (clopidogrel or prasugrel) do not stop taking their medication. […] The benefits of continuing dual antiplatelet therapy for one year, according to current guidelines, far outweigh the risks." Among 119 items (28%), the text was uncertain about the benefit of prolonged treatment but was reported with no mention of the increased risk of mortality (100, 24%) or the questionable explanation clearing the responsibility of prolonged treatment (n=19; 4%).
Overall, 34 items (8%) were uncertain about the benefit of prolonged treatment but mentioned the increased risk of mortality. Only 27 (6%) did not favour prolonged treatment and only 12 of these (3%) clearly raised some concerns about the reporting of the increased risk of death.
Further information on items by source is in appendix 1.

Figure 3: Content of scholarly and public attention surrounding DAPT study by source
Participants: No participants were involved in this study.
Main outcome measure: Proportion of contents highlighting the increased risk of mortality and critical to the author's questionable interpretation of the results.

Results:
We identified 425 items reported by 7 sources; 164 (39%) disseminated the authors' questionable interpretation via an electronic link or a reference, with no additional text.
Among 81 items (19%), the message favoured prolonged treatment and consequently overstated the article conclusions. Among 119 items (28%), the text was uncertain about the benefit of prolonged treatment but was reported with no or inappropriate mention of increased risk of mortality. Only 34 items (8%) were uncertain about the benefit of prolonged treatment and appropriately mentioned increased risk of mortality. In all, 27 items (6%) did not favour prolonged treatment, and only 12 of these (3%) clearly raised some concerns about the reporting of increased risk of death.

Conclusion:
The amount of contents criticizing the interpretation of the DAPT study results was limited.

Strengths and limitation of this study
• Our method involved a broad search strategy, ensured to capture an extensive and representative sample of contents citing the 2014 DAPT trial for both scholarly and public attention.
• Our systematic approach to analyze the text of contents provides a comprehensive overview of dissemination of the study results.
• This study focused on only a specific trial publication and results are not generalizable to other studies. The results of the DAPT study were published in the New England Journal of Medicine (NEJM) 9 after their presentation at the American Health Association Conference, in November 2014. However, the reporting of the results raised some concerns 10,11 .
Particularly, the abstract conclusions did not mention the increased risk of mortality.
Furthermore, the discussion included questionable explanations based on post-hoc analyses to clear the role of prolonged thienopyridine treatment in this increased risk of mortality. For this purpose, the authors had split the analysis by cause of death, which was not powered to show a statistically significant difference. They focused on the increase in cancer-related imbalance at baseline in patients with a history of cancer before enrolment (9.8% vs 9.5%).
To confirm, the authors performed a post-hoc analysis excluding all deaths that could be related to cancer diagnosed before enrolment. Consequently, the results became statistically non-significant (0.50% vs 0.28%, p=0.11). This post-hoc exclusion of patients with an event is a concern.
We aimed to explore how the distorted interpretation of results from the DAPT trial was disseminated to the scientific community and online media and to assess whether this interpretation was criticized or not.

METHODS
We performed a cross-sectional study of scholarly and public attention surrounding the DAPT study.

Identification of scholarly and public attention surrounding the DAPT study Scholarly attention
On June 2015, we searched the following electronic databases to identify responses to the DAPT study: ISI Web of Knowledge, Google Scholar, and PubMed Commons. We also searched the comments and citing articles on the NEJM website for the original article 9 .

Public attention
We and the search terms "DAPT" and "dual antiplatelet therapy". We also searched EurekAlert! (a free online database for science press releases, www.eurekalert.org) for press releases dedicated to the DAPT study; YouTube (search terms "DAPT" and "dual antiplatelet therapy"); and pages dedicated to patients, clinicians and media at the DAPT study website (http://www.daptstudy.org).

Content of scholarly and public attention surrounding the DAPT study
Two researchers (MS, RH) read the items from each source independently and evaluated them by using a preliminarily tested extraction form. Disagreements were resolved by discussion to reach consensus. If needed, a third researcher (IB) appraised the content.
We determined whether the source consisted of a reference or a link to the NEJM article reporting the DAPT study only or was a text commenting on the DAPT study. For a text commenting on the DAPT study, we checked whether the original study authors were involved in writing the text or not. Our main outcome of interest was the proportion of contents highlighting the increased risk of mortality and critical to the author's questionable interpretation of the results. We determined whether • the primary efficacy outcomes (i.e., stent thrombosis and MACCE) were reported • the safety outcomes related to moderate or severe bleeding were reported • the increased risk of mortality with prolonged treatment was reported • the authors' questionable explanation clearing the responsibility of prolonged treatment in the increased risk of mortality was reported or criticized • the content of the text was 1) favouring the prolonged treatment and consequently overstating the article conclusion, 2) uncertain about the benefit of the prolonged treatment (i.e., statement of both the beneficial effect, and increased risk of bleeding, text ending with a question mark, use of "may or might" or reporting that the study needs further research), or 3) not favouring the prolonged treatment 16 .

Statistical analysis
We calculated frequencies and percentages (%) for qualitative variables and median (interquartile range) for quantitative variables.
A total of 100 items (24%) did not mention mortality, but when mortality was mentioned, in 19 items (5%), it was reported with the authors' questionable justification for prolonged treatment. However, this is the first study to our knowledge to focus on both scholarly and public dissemination of study results. Our study highlighted an unmet need of scientific communication in the media, whose importance in dissemination of scientific data is becoming increasingly relevant. These findings could be helpful for the entire community for better understanding how scientific knowledge is disseminated.
Our approach involved a broad search strategy and multiple search engines, which ensured the capture of an extensive and representative sample of contents discussing the DAPT study results. Each social media item from Altmetric was systematically reviewed for additional content that may have been missed, and several different search engines were used. We captured items that were published over the course of many months, which highlighted the perpetuation and continuation of the dissemination of the questionable interpretations. The inclusion period for sources seemed to be more than sufficient because tweets linked to scientific articles have been shown to taper off well before our cut-off point (7 months) 25 . In addition, 2 independent researchers assessed each source by using a standardized data extraction form and disagreements were resolved by consensus. Altmetric. Second, the data extraction involved some subjectivity; however, we tried to address this by using a standardized data extraction form and independent assessment as well as consensus among 2 researchers. Third, despite our best efforts, we cannot ensure that our search strategy was all-encompassing because of the breadth of social media. Finally, we did not explore the balance between efficacy and safety outcomes with DAPT treatment.
Our aim was not to resolve the controversy about DAPT duration and this debate is still ongoing. The OPITUDAL trial did not find an increased risk of death with the prolonged treatment; on the contrary, the risk of death was lower with the prolonged treatment 26 .
Several meta-analyses found conflicting results 4,5,8,27,28 . The researchers involved in the DAPT trial concluded in a meta-analysis published in The Lancet that prolonged DAPT duration was not associated with a difference in risk of all-cause mortality 29 . Three metaanalyses, published later by different teams, showed prolonged DAPT associated with increased risk of all-cause mortality 4,5,8 . More recently, other meta-analyses did not find a statistically significant increase in all-cause mortality 27,28 . Most of these meta-analyses warranted further research with extended DAPT.
However, these results are difficult to interpret because of different definitions of short (1, 3, 6, or 12 months) and extended (6, 12, 24 or > 24 months) durations, which varied across studies. Furthermore, different durations of follow-up and types of stents could also influence the results.

Acknowledgements
We thank Elise Diard for help in creating Figure 2. We acknowledge support from Altmetric for free access to "Altmetric Explorer". We acknowledge the assistance in English language proofreading by Laura Smales (BioMedEditing, Toronto, Canada).
Isabelle Boutron and Philippe Ravaud submitted a letter to the NEJM following the publication of the DAPT study to highlight the inadequate reporting in the abstract conclusions, but the letter was rejected.

Funding Source
This study did not receive any funding.

Competing interests
None declared.

Ethical approval
Not needed

Introduction 4
Background/rationale 2 Explain the scientific background and rationale for the investigation being reported 4, 5 Objectives 3 State specific objectives, including any prespecified hypotheses 5

Strengths and limitation of this study
• Our method involved a broad search strategy, ensured to capture an extensive and representative sample of contents citing the 2014 DAPT trial for both scholarly and public attention.
• Our systematic approach to analyze the text of contents provides a comprehensive overview of dissemination of the study results.
Particularly, the abstract conclusions did not mention the increased risk of mortality. The results were interpreted as being related to an imbalance at baseline in patients with a history of cancer before enrolment (9.8% vs 9.5%). To confirm, the authors performed a post-hoc analysis excluding all deaths that could be related to cancer diagnosed before enrolment. Consequently, the results became statistically non-significant (0.50% vs 0.28%, p=0.11). This post-hoc exclusion of patients with an event is questionable.

METHODS
We performed a cross-sectional study of scholarly and public attention surrounding the DAPT study.

Identification of scholarly and public attention surrounding the DAPT study Scholarly attention
On June 2015, we searched the following electronic databases to identify responses to the DAPT study: ISI Web of Knowledge, Google Scholar, and PubMed Commons. We also searched the comments and citing articles on the NEJM website for the original article 9 .

Public attention
We searched Altmetric Explorer 12-15 to identify all online attention (news, blogs, Twitter, Facebook, Google+, Mendeley, CiteULike) given to the DAPT study. Each identified social media source was then systematically evaluated to determine whether other posts were not captured by Altmetric Explorer. In addition, each original tweet was reviewed to find retweets, replies and favourites. Since Altmetric.com captures only tweets attached to the DOI (Digital Object Identifier) of the original DAPT article, we also used snapbird.org, a search engine that can search an individual Twitter account by using the NEJM's Twitter account and the search terms "DAPT" and "dual antiplatelet therapy". We also searched EurekAlert! (a free online database for science press releases, www.eurekalert.org) for press releases dedicated to the DAPT study; YouTube (search terms "DAPT" and "dual antiplatelet therapy"); and pages dedicated to patients, clinicians and media at the DAPT study website

Content of scholarly and public attention surrounding the DAPT study
Two researchers (MS, RH) read the items from each source independently and evaluated them by using a preliminarily tested extraction form. Disagreements were resolved by discussion to reach consensus. If needed, a third researcher (IB) appraised the content.
We determined whether the source consisted of a reference or a link to the NEJM article reporting the DAPT study only or was a text commenting on the DAPT study. For a text commenting on the DAPT study, we checked whether the original study authors were involved in writing the text or not. Our main outcome of interest was the proportion of contents highlighting the increased risk of mortality and critical to the author's interpretation of the results. We determined whether • the primary efficacy outcomes (i.e., stent thrombosis and MACCE) were reported • the safety outcomes related to moderate or severe bleeding were reported • the increased risk of mortality with prolonged treatment was reported • the authors' explanation clearing the responsibility of prolonged treatment in the increased risk of mortality was reported or criticized • the content of the text was 1) favouring the prolonged treatment and consequently overstating the article conclusion, 2) uncertain about the benefit of the prolonged treatment (i.e., statement of both the beneficial effect, and increased risk of bleeding, text ending with a question mark, use of "may or might" or reporting that the study needs further research), or 3) not favouring the prolonged treatment 16 .

Statistical analysis
We calculated frequencies and percentages (%) for qualitative variables and median (interquartile range) for quantitative variables.
A total of 100 items (24%) did not mention mortality, but when mortality was mentioned, in 19 items (5%), it was reported with the authors' justification for prolonged treatment. However, this is the first study to our knowledge to focus on both scholarly and public dissemination of study results. Our study highlighted an unmet need of scientific communication in the media, whose importance in dissemination of scientific data is becoming increasingly relevant. These findings could be helpful for the entire community for better understanding how scientific knowledge is disseminated.
Our approach involved a broad search strategy and multiple search engines, which ensured the capture of an extensive and representative sample of contents discussing the DAPT study results. Each social media item from Altmetric was systematically reviewed for additional content that may have been missed, and several different search engines were used. We captured items that were published over the course of many months, which highlighted the perpetuation and continuation of the dissemination of the authors' interpretations. The inclusion period for sources seemed to be more than sufficient because tweets linked to scientific articles have been shown to taper off well before our cut-off point (7 months) 25 . In addition, 2 independent researchers assessed each source by using a standardized data extraction form and disagreements were resolved by consensus. Altmetric. Second, the data extraction involved some subjectivity; however, we tried to address this by using a standardized data extraction form and independent assessment as well as consensus among 2 researchers. Third, despite our best efforts, we cannot ensure that our search strategy was all-encompassing because of the breadth of social media. Finally, we did not explore the balance between efficacy and safety outcomes with DAPT treatment.
Our aim was not to resolve the controversy about DAPT duration and this debate is still ongoing. The OPITUDAL trial did not find an increased risk of death with the prolonged treatment; on the contrary, the risk of death was lower with the prolonged treatment 26 .
Several meta-analyses found conflicting results 4,5,8,27,28 . The researchers involved in the DAPT trial concluded in a meta-analysis published in The Lancet that prolonged DAPT duration was not associated with a difference in risk of all-cause mortality 29 . Three metaanalyses, published later by different teams, showed prolonged DAPT associated with increased risk of all-cause mortality 4,5,8 . More recently, other meta-analyses did not find a statistically significant increase in all-cause mortality 27,28 . Most of these meta-analyses warranted further research with extended DAPT.

Acknowledgements
We thank Elise Diard for help in creating Figure 2. We acknowledge support from Altmetric for free access to "Altmetric Explorer". We acknowledge the assistance in English language proofreading by Laura Smales (BioMedEditing, Toronto, Canada).
Isabelle Boutron and Philippe Ravaud submitted a letter to the NEJM following the publication of the DAPT study to highlight the inadequate reporting in the abstract conclusions, but the letter was rejected.

Funding Source
This study did not receive any funding.

Competing interests
None declared.

Introduction 4
Background/rationale 2 Explain the scientific background and rationale for the investigation being reported 4, 5 Objectives 3 State specific objectives, including any prespecified hypotheses 5