Book
Bad Pharma: How Drug Companies Mislead Doctors and Harm Patients
Goldacre, a physician and science writer, argues that the evidence doctors use to choose drugs is distorted at every stage and that regulators, universities and professional bodies have let it happen. The book has six chapters: missing and selectively reported trials, with antidepressants, reboxetine, Tamiflu and rosiglitazone as case histories; where new drugs come from; regulators; trial designs that favor the sponsor; a case for larger pragmatic trials; and marketing, covering sales representatives, ghostwriting, journals, medical education and patient groups. Each chapter ends with fixes addressed to named audiences, and the afterword uses the July 2012 GlaxoSmithKline settlement as a recap. Sources are endnoted to journal articles, court documents and regulator papers.
Who made it
- The author's note describes him as 'a doctor and writer' whose first book, Bad Science, reached number one in the UK nonfiction charts (e-text, About the Author).
- Published by Fourth Estate, an imprint of HarperCollinsPublishers, London; first published in Great Britain in 2012; ISBN 978-0-00-735074-2 (copyright page).
- Acknowledgements state that in recent years his day jobs were supported by the National Institute for Health Research, the Scott Trust, the Wellcome Trust, Nuffield College, Oxford, and the NHS (e-text, Acknowledgements).
- The text declares that he sits on a funding committee of the NHS Health Technology Assessment programme, which funds head-to-head drug comparisons that no company would pay for (Chapter 3 endnote 36; Afterword, 'Funders').
- AllTrials, launched in January 2013 after publication, lists Bad Science, the BMJ, the Centre for Evidence-based Medicine, the Cochrane Collaboration, the James Lind Initiative, PLOS and Sense About Science as its initiators and carries a message from Goldacre (archived home page, 2014: https://web.archive.org/web/2014/http://www.alltrials.net/).
- Goldacre is a named author of three studies cited below as post-book follow-up: the COMPare outcome-switching study (Trials 2019), the EU Clinical Trials Register compliance study (BMJ 2018) and the FDAAA compliance study (Lancet 2020, held as exhibit 95).
Its claims checked against the record
27 claims checked:12 supported · 15 partly supported · 0 contradicted · 0 not in the record
A verdict rests on the documents named in its row.
| The work says | Verdict | What the record shows |
|---|---|---|
| Industry-funded trials are more likely to produce a result that favors the sponsor's drug than independently funded trials. The book calls this its core premise. | Supported by the record | The 2017 Cochrane update of 75 studies found industry-sponsored studies more often had favorable efficacy results (RR 1.27, 95% CI 1.17-1.37; 25 papers, moderate-quality evidence) and favorable conclusions (RR 1.34, 1.19-1.51), with no significant difference on harms (RR 1.37, 0.64-2.93; four papers, very low quality). The pooled ratios are smaller than the 'four times more likely' figure the book takes from 2003 reviews. |
| Industry-funded statin trials were twenty times more likely to give results favoring the test drug, after controlling for other factors. | Partly supported | In the multivariable analysis of the industry-funded trials, funding from the test drug's own company was associated with results favoring that drug (odds ratio 20.16, 95% CI 4.37-92.98) and conclusions favoring it (OR 34.55, 7.09-168.4). That is an odds ratio among industry-funded trials, not a rate comparison with independent trials; the same paper reports that 189 of 192 trials used only surrogate outcomes. |
| FDA review records for 12 antidepressants show 38 positive and 36 negative trials; in the journals the picture became 48 positive and 3 negative, because negative trials went unpublished or were written up as positive. | Supported by the record | Exhibit 89: of 74 trials (12,564 participants), FDA judged 38 positive, and 37 of those were published as positive; of 36 negative or questionable trials, 22 were unpublished, 11 were published as positive and 3 as negative, so 48 of 51 published trials read positive (94%) against 51% in FDA review. The book's own text gives the negative count as 36 and as 37; the paper says 36. Exhibit 90 (30 trials of four newer drugs) found transparent publication of negative trials rose from 11% to 47%, with six of 15 negative trials unpublished. |
| Trials with positive results are about twice as likely to be published as negative ones, and about half of all trials go unpublished. | Supported by the record | Schmucker's systematic review pooled 46.2% (95% CI 40.2-52.4%) of ethics-committee-approved studies and 54.2% (42.0-65.9%) of registry-listed studies as published, and found significant results more likely to be published (OR 2.8, 2.2-3.5). Song's report confirms the direction; the book's 1.62 and 2.4 ratios are in the report body and were not checked. |
| Reboxetine looked effective in the published trials; when unpublished trials were added it was no better than placebo, did worse than other antidepressants and did more harm than good. | Partly supported | Eyding analyzed 13 trials with 4,098 patients, of whom 74% (3,033) had unpublished data, and reported no significant remission difference against placebo (OR 1.17, 0.91-1.51), inferiority to SSRIs (OR 0.80) and more harm outcomes than placebo, concluding reboxetine is 'an ineffective and potentially harmful antidepressant'. The MHRA and the European Pharmacovigilance Working Party reported in September 2011, a year before the book, that Eyding's response analysis used 7 of 11 placebo-controlled studies; with all 11 the response odds ratio was 1.47 (1.10-1.97), no new safety signal was found, and the benefit-risk balance 'remains positive', with benefit shown in severe depression and none shown in mild or moderate depression. The book states Eyding's conclusion without the regulators' review. |
| Researchers switch the pre-specified primary outcome between protocol and paper, and do so mostly in favor of positive results; gabapentin trials are the example. | Supported by the record | Chan: in 102 Danish trials, 62% had at least one primary outcome changed, introduced or omitted, and significant outcomes were more likely to be fully reported (OR 2.4 for efficacy, 4.7 for harm). Vedula: of 21 protocol primary outcomes in published gabapentin trials, 6 were not reported and 4 were reported as secondary, and 12 of 28 reported primary outcomes were new, matching the book's figures. COMPare, run by the author's group after the book, found 58 of 67 trials in five high-impact journals (87%) had outcome discrepancies; journals published 23 of the correction letters (40%). |
| Study 329 of paroxetine in adolescents was reported as positive although it showed no benefit on its pre-specified outcomes; positive findings came from outcomes added later. | Supported by the record | Exhibit 92, the 2015 reanalysis using the protocol and trial records, states efficacy of paroxetine and imipramine 'was not statistically or clinically significantly different from placebo for any prespecified primary or secondary efficacy outcome'. The Justice Department alleged that GSK took part in preparing a journal article that misreported the trial as showing efficacy in patients under 18, and GSK pleaded guilty to misbranding Paxil. The original investigators dispute the reanalysis and say the secondary measures were added before the blind was broken (Exhibit 93); the journal attached an expression of concern to the 2001 article in September 2025 (Exhibit 94). |
| GSK's own trials showed paroxetine did not work in children; GSK did not tell doctors or the regulator, held back suicide-risk data under a reporting loophole, and the UK regulator opened a four-year criminal investigation. | Partly supported | The Justice Department release alleges GSK promoted Paxil for patients under 18 from April 1998 to August 2003, 'did not make available data from two other studies in which Paxil also failed to demonstrate efficacy', and pleaded guilty to misbranding Paxil as to use under 18. The internal memo the book quotes ('commercially unacceptable to include a statement that efficacy had not been demonstrated') and the off-label reporting loophole are not in the documents cited. Press reports say the MHRA closed its criminal investigation in March 2008 without charges, saying the law in force then had 'important weaknesses'. |
| The European Medicines Agency refused the Nordic Cochrane Centre access to clinical study reports on orlistat and rimonabant for over three years, citing commercial interests, until the European Ombudsman intervened. | Supported by the record | The Ombudsman found 'EMA did not establish that the requested documents fall within the scope of the commercial interests exception' (para. 83), found the burden argument insufficiently explained (para. 37) and noted patients appear only by study number (para. 86); EMA accepted the draft recommendation and agreed to grant access. The record has since moved: EMA's Policy 0070 was published on 2 October 2014, covers data submitted from 1 January 2015, and publishing began in October 2016; EMA later suspended it for non-COVID-19 medicines and resumed for new active substances with a CHMP opinion from September 2023. |
| Registries were never enforced: the 2007 FDA Amendments Act requires results within a year, but only one in five trials complied, the $10,000-a-day fine has never been levied, and the EU register was kept secret. | Supported by the record | Prayle found 163 of 738 trials (22%) subject to mandatory reporting had reported within a year. Exhibit 95 found 1,722 of 4,209 trials (40.9%) reported on time by September 2019, with industry sponsors more likely to comply than others (OR 3.08). FDA's table of noncompliance notices (Exhibit 96) begins on 27 April 2021, the first notice to Acceleron Pharma, and states no civil money penalty outcome for any entry. For the EU register, 49.5% of 7,274 trials due had reported results by the December 2016 deadline, 68.1% for commercial sponsors against 11.0% for non-commercial sponsors. The book's account that the EudraCT contents were closed to the public was not checked against EMA documents. |
| Governments spent hundreds of millions of pounds stockpiling Tamiflu to reduce complications (pneumonia and death), and enough was bought to treat 80 per cent of the UK population. | Supported by the record | The NAO states the Department of Health bought nearly 40 million units for 424 million pounds (2011-12 prices) between 2006-07 and 2012-13 (para. 2), that the Cabinet Office review found 'no published evidence on a reduction in mortality due to antiviral use and very limited evidence on reduced complications', and that the advisory committee told modelers to assume a 40 to 50 per cent reduction in hospitalizations and deaths, based partly on unpublished evidence (para. 15). Stockpile coverage briefly reached 80% during the 2009 pandemic and the Department held it at 50% (para. 17). |
| Roche withheld the full clinical study reports on Tamiflu from Cochrane for years, so the evidence on whether the drug reduces pneumonia and death could not be assessed. | Partly supported | The NAO (para. 13) records that Roche told Cochrane in April 2013 it would release all its Tamiflu study reports, edited for confidentiality and commercial interest; the 2014 Cochrane review rests on 107 study reports obtained from EMA, GSK and Roche, all from manufacturer-sponsored trials. With the reports, Cochrane found oseltamivir cut time to first symptom relief in adults by 16.8 hours (7 to 6.3 days), no significant effect on hospitalization (RD 0.15%, -0.78 to 0.91) or serious complications, and a 1.00% reduction in unverified self-reported pneumonia (NNTB 100) that was not significant in the five trials with a detailed diagnostic form; nausea (RD 3.66%) and vomiting (4.56%) rose. A 2015 pooled analysis of Roche-sponsored adult trials (Dobson) reported complications needing antibiotics at RR 0.56 and hospital admission at RR 0.37 and drew critical responses; WHO's 2017 committee moved oseltamivir to the complementary list on the basis of a smaller estimated effect. Record partly withheld. Roche's full clinical study reports and trial protocols for its Tamiflu trials were supplied to Cochrane only in part from 2010 (about 3,200 pages, which Roche called sufficient) and Roche undertook to release all of them in April 2013; Doshi et al., PLoS Medicine 9(4):e1001201 (2012) prints Roche's dated letters of 8 October 2009 to 26 April 2011, and the NAO memorandum (21 May 2013, paras 3.14-3.15) records the dispute and the April 2013 undertaking. |
| The published Tamiflu papers differed from the clinical study reports: adverse events were omitted and one 'placebo' was not inert. | Partly supported | The 2014 review found 'evidence of selective reporting' in oseltamivir and zanamivir studies, found inadequate protection against performance bias because placebo was not identically presented in 11 oseltamivir studies, and states the placebo in both sets of trials 'may have contained active substances'. The book's specific examples (ten serious adverse events in 'module 1' documents, three possibly drug-related; the placebo ingredient dehydrocholic acid) are from a 2011 BMJ paper that is not among the documents cited. |
| Nissen's 2007 meta-analysis showed a 43 per cent increase in heart problems on rosiglitazone, later work confirmed it, and in 2010 the drug was withdrawn or restricted around the world. | Partly supported | Nissen's 42-trial meta-analysis found an odds ratio of 1.43 for myocardial infarction (95% CI 1.03-1.98) and 1.64 for cardiovascular death (0.98-2.74, borderline), and its authors noted they lacked source data. FDA's December 2015 communication says that in 2013 FDA required removal of prescribing and dispensing restrictions 'after determining that data did not demonstrate an increased risk of heart attack with rosiglitazone medicines compared to the standard type 2 diabetes medicines metformin and sulfonylurea', and eliminated the REMS in 2015 after finding 'no new pertinent safety information'. The book (2012) predates both decisions. |
| GSK held internal analyses showing cardiac risk with rosiglitazone years before 2007, intimidated Dr John Buse when he raised concerns in 1999, and obtained Nissen's unpublished manuscript; GSK later admitted failing to report safety data to FDA. | Supported by the record | The 2010 staff report, from over 250,000 pages of documents, found GSK analyses completed in fall 2005 (hazard ratio 1.29 for myocardial ischemia, 11,586 subjects) and summer 2006 (1.31, 14,237 patients), concluded GSK was aware of the risk in late 2004 or early 2005, and states that on 3 May 2007 an NEJM peer reviewer, Dr Steve Haffner, faxed Nissen's unpublished paper to a GSK executive. It describes the 1999 plan to stifle Dr Buse. The Justice Department release says GSK pleaded guilty to failing to report Avandia safety data to FDA between 2001 and 2007. that GSK and FDA held the 2005 and 2006 analyses without comment until 2008 was not checked. |
Show the other 12 claims checked
| The work says | Verdict | What the record shows |
|---|---|---|
| In July 2012 GSK pleaded guilty to a vast range of charges, bribed doctors with gifts and hospitality, withheld paroxetine data and made false statements about rosiglitazone. | Partly supported | GSK pleaded guilty to three counts: misbranding Paxil, misbranding Wellbutrin, and failing to report Avandia safety data to FDA (criminal fine and forfeiture $1 billion). The Justice Department describes the other conduct, including kickbacks to physicians and false statements about Avandia safety, as allegations resolved in a $2 billion civil settlement with the federal government and the states. |
| Regulators in the United States and European Union are paid for almost entirely by drug companies through fees. | Partly supported | Exhibit 127: user fees were $1,377,172,919 of $1,772,198,497 in obligations for FDA's prescription drug review program in FY 2024, printed as 77 percent; the share was 73 percent in FY 2020. Fees therefore pay a majority of the review program, short of 'almost entirely'. A search excerpt of the CRS report gives 51% of the human drugs program in FY 2012. The EU share was not checked. |
| A third of the 197 drugs approved by FDA from 2000 to 2010 had no evidence comparing them with the best available treatment; the rest of the time regulators accept 'better than nothing'. | Partly supported | Of 197 new molecular entities, 100 (51%) had comparative efficacy data in the approval package; the figure rose to 70% after excluding 37 orphan products and 17 drugs for conditions with no alternative treatment. The book gives the 70% figure and the exclusion for untreated conditions but omits the orphan-product exclusion. |
| Seeding trials, such as Merck's ADVANTAGE on Vioxx and Pfizer's STEPS on Neurontin, were marketing exercises presented to doctors and patients as research. | Supported by the record | Hill's review of Merck litigation documents found ADVANTAGE was designed by Merck's marketing division, which handled the data, and that Merck hid the marketing nature of the trial from participants, investigators and ethics committee members. Krumholz's review found STEPS 'was a seeding trial posing as a legitimate scientific study' and that none of the company analyses of investigators' prescribing were reported in the two published articles. The memo wording the book quotes was not checked against the papers' full text. |
| Ghostwriting is common: about 8 per cent of articles in six leading journals, and 75 per cent of Danish industry-initiated trials, had ghost authorship. | Supported by the record | Wislar: 49 of 622 respondents' articles (7.9%) had ghost authors, 11.9% of research articles; honorary or ghost authorship together was 21.0%, down from 29.2% in 1996, and ghost authorship fell from 11.5% to 7.9%. Gøtzsche: ghost authorship in 33 of 44 industry-initiated trials (75%), 91% counting acknowledged writers. Ross, from about 250 Merck litigation documents, describes Merck employees and medical publishing companies preparing manuscripts and recruiting academic authors. The Senate staff report describes the practice as drafting by company-paid writers for physicians who sign as authors. |
| Elsevier produced six journals for industry, among them an Australasian journal made entirely of Merck advertising, and its chief executive admitted they were made to look like journals. | Partly supported | As reported, Elsevier's Health Sciences chief executive said in May 2009 that from 2000 to 2005 its Australia office published sponsored compilations that 'were made to look like journals and lacked the proper disclosures'. In June 2009 Elsevier said the Australasian Journal of Bone and Joint Medicine was 'the only title identified as single sponsored without proper disclosure', with Merck its sole sponsor, and that the other titles were funded by advertising from several companies. The book's wording that all were 'entirely advertising projects for that one company' does not match the second statement. |
| Drug-rep visits, sponsored dinners and sponsored education change prescribing, and the industry spends twice as much on marketing as on research. | Partly supported | Spurling's review of 58 studies found 38 associated exposure to company information with higher prescribing frequency and 13 found no association; the book's count of 29 rep-visit studies (17 positive) was not checked. Gagnon and Lexchin estimated $57.5 billion on US promotion in 2004 against $31.5 billion on industrial R&D (National Science Foundation), 'almost twice as much', or 24.4% of sales against 13.4%, but say the true promotion figure 'could range from a low of US$27.7 billion to a high of US$57.5 billion' and that IMS data put R&D ($29.6 billion) above promotion. The Senate staff report says industry funded over a billion dollars a year of continuing medical education. The book's report that doctors deny being influenced was not checked. The Sunshine Act disclosure the book endorses began with reporting from March 2013 per the Senate staff report; CMS posted the first Open Payments data on 30 September 2014 (search excerpt). |
| Many patient groups are funded by drug companies, often without declaring it, and funded groups take positions that favor the industry. | Supported by the record | Fabbri's review of 26 studies found industry funding of patient groups ranged from 20% to 83%, 27% of funded groups disclosed it on their websites, and in four studies industry-funded groups generally supported the sponsors' interests. The book's figures (two thirds of groups working with EMA funded; average donation rising from €185,500 to €321,230) are from the 2010 HAI report, which is not among the documents cited. |
| Statins roughly halve the risk of a heart attack whatever the person's starting risk, so the benefit is large in absolute terms only for high-risk patients. | Partly supported | The 2012 analysis of individual data from 27 trials found a 21% reduction in major vascular events per 1.0 mmol/L LDL reduction (RR 0.79, 0.77-0.81), with proportional reductions at least as large in the lowest-risk categories (RR 0.62 and 0.69 per mmol/L), so the constant-proportion argument holds. The 'roughly halve' figure is larger than the per-mmol/L estimate, and the book states no LDL reduction. |
| MHRA decided in February 2008 to add depression, sleep disturbance, memory loss and sexual dysfunction to the label of all statins, but a drug company's disagreement over wording delayed the change until November 2009. | Partly supported | The November 2009 report states the review decided that sleep disturbances, memory loss, sexual disturbance, depression and interstitial pneumopathy were a class effect of all statins and recommended updating product information; MHRA's Drug Safety Update says the review found 'the balance of risks and benefits of statins as a class remains positive'. |
| Professor Richard Eastell was censured by the General Medical Council over a statement of full data access, after a delay of several years, after Aubrey Blumsohn raised the issue and was forced out of Sheffield. | Partly supported | As reported, the GMC found the statement 'was untrue, since you did not have full access to the raw data', and found that failing to give it due consideration 'may have been negligent' but 'was not deliberately misleading or dishonest'; the report says there was 'no finding of misconduct, and no warning has been issued'. Blumsohn's suspension and settlement are not in the report that was opened. |
| Sir John Bell, Oxford professor of medicine and president of the Academy of Medical Sciences, is on the board of Roche. | Partly supported | Roche's release says that after 18 years of service on the Board, Professor Sir John Bell would not stand for re-election at the 2020 annual meeting, so he sat on the Board in 2012. His Academy of Medical Sciences role was not checked. |
What the record shows that it leaves out
- MHRA and the European Pharmacovigilance Working Party reanalyzed all 11 placebo-controlled reboxetine studies in September 2011 and found response OR 1.47 (1.10-1.97) and a positive benefit-risk balance; the book, finished in August 2012, does not mention it (MHRA UK Public Assessment Report, September 2011).
- FDA required removal of rosiglitazone prescribing and dispensing restrictions in 2013 and eliminated the REMS in December 2015, finding data did not demonstrate an increased heart attack risk against metformin and sulfonylurea (FDA Drug Safety Communication, archived at https://web.archive.org/web/2016/http://www.fda.gov/Drugs/DrugSafety/ucm476466.htm).
- The Cholesterol Treatment Trialists' individual-patient meta-analyses of 27 trials, which depend on trial data shared with the collaboration, give the statin effect as RR 0.79 per 1.0 mmol/L LDL reduction (Lancet 2012;380:581), the quantity the book's risk-reduction illustration omits.
- Wislar et al. report ghost authorship in six leading journals fell from 11.5% in 1996 to 7.9% in 2008, a change the book's 8 per cent figure does not show (BMJ 2011;343:d6128).
- The Cochrane 2017 review of sponsorship found no significant difference between industry and non-industry studies on harms results (RR 1.37, 0.64-2.93), a result absent from the book's account (MR000033).
Documented efforts and withheld records
Each item is a documented act, with its source: an act against the work, its author or its sources, or the withholding of records on its subject. The list states what was done and does not state why.
- 2009-10-08 and 2009-12-08. Roche declined Cochrane reviewers' request for its Tamiflu trial data on 8 October 2009, saying 'a similar meta analysis is currently commencing'; on 8 December 2009 it proposed a confidentiality agreement that the reviewers declined because it barred disclosing 'the existence and terms of this Agreement'. Doshi, Jefferson and Del Mar, 'The Imperative to Share Clinical Study Reports: Recommendations from the Tamiflu Experience', PLoS Medicine 9(4):e1001201
- 2010-2012. Roche supplied partial reports for ten trials in early 2010 and wrote on 20 August 2010 that 'around 3,200 pages of information have already been provided', asking for the reviewers' 'full analysis plan for review by Roche' before releasing more; it wrote on 26 April 2011 'You have all the detail you need to undertake a review' and on 20 January 2012 that it 'has made full clinical study data available to health authorities around the world'. The authors report these were subsets of the full reports. Doshi, Jefferson and Del Mar, PLoS Medicine 9(4):e1001201 (Roche letters quoted in the paper)
- 2010-02 to 2013-04. Cochrane withdrew its 2009 Tamiflu review in February 2010 after a query about its reliance on a Roche-sponsored meta-analysis of ten trials, eight unpublished; the NAO describes a 'prolonged dispute between the Cochrane team and Roche concerning access to the original data, conducted increasingly in the public domain'. In April 2013 Roche told Cochrane it would release the reports of all Tamiflu trials it sponsored, edited for patient confidentiality and commercial interests. National Audit Office, 'Access to clinical trial information and the stockpiling of Tamiflu', HC 125, paras 3.14-3.15
- 2013-02-27. Roche announced on 27 February 2013 that it would set up an independent expert body to assess requests for patient-level data and said health authorities 'need to remain the gatekeeper for drug assessment and approval'; Sense About Science director Tracey Brown (an AllTrials initiator) said 'Does Roche expect applause for announcing that it will continue to keep clinical trial findings hidden?', and Carl Heneghan said confidentiality clauses would likely apply. PharmaTimes, 'Critics unconvinced as Roche expands access to data'
- 2013-01 to 2014. AbbVie asked the EU General Court to annul EMA's January 2013 decision to release three Humira clinical study reports; the Court's President granted interim measures blocking release on 25 April 2013, and the Vice-President of the Court of Justice overturned that order on 28 November 2013. EMA and AbbVie then agreed a redacted release in 2014. (The Ombudsman's decision also covers the parallel InterMune case only in outline; InterMune is named in press reports of the same April 2013 orders.) European Ombudsman, decision closing own-initiative inquiry OI/3/2014 (Humira clinical study reports), 8 June 2016
- 2015 to 2018-02-06. Pari Pharma, PTC Therapeutics and MSD Animal Health each sued EMA to stop it releasing documents they had submitted (a clinical study report in PTC's case); the General Court dismissed all three challenges on 6 February 2018 because the companies gave no concrete evidence of commercial harm. EMA, 'General Court confirms EMA approach to transparency' (cases T-235/15, T-718/15, T-729/15)
- 2013-02-27. ABPI chief executive Stephen Whitehead, at a PharmaTimes debate with Goldacre, accused him of using 'The Big Lie' and conceded the ABPI had never set up a system to check that members followed existing data-disclosure rules, while agreeing summary results should be published for all trials: 'It's not about what, it's about how'. Pharmafile, report of the PharmaTimes debate 'Ben Goldacre versus Stephen Whitehead'
- 2012-12-14. Goldacre published a post (14 December 2012) saying the ABPI chief executive and president had falsely told MPs and journalists over many years that he refused to engage with them, described an 'energetic personal smear campaign' by the ABPI, and published his letter to Whitehead asking for a list of the people misled. This is an act by the author's side. Goldacre, Bad Science blog post, reposted on Pharmagossip 8 February 2013
- 2009-02. Lawyers for LBC sent Goldacre a letter in early February 2009 demanding he remove from his Bad Science blog a 44-minute clip of a Jeni Barnett broadcast, claiming it infringed the station's copyright; Goldacre removed it. This concerns the author's earlier blog, not the book. Technollama, 'Bad Science meets bad copyright'
Published responses
- Association of the British Pharmaceutical Industry (Stephen Whitehead, chief executive): At a PharmaTimes debate with Goldacre on 27 February 2013, Whitehead said: 'We agree that summary trial information should be published for all trials. It's not about what, it's about how', said the ABPI had never set up a system to ensure members followed existing data-disclosure rules, and criticized the book. The ABPI's earlier October 2012 statement that the examples were historical was reported by Research Professional News.
- Roche: Told Cochrane in April 2013 it would release all its Tamiflu clinical study reports over the following months, with edits for patient confidentiality and commercial interest (NAO memorandum, para. 13).
- MHRA and European Pharmacovigilance Working Party: Reviewed reboxetine in response to Eyding et al. and published its conclusion in September 2011 that benefit-risk 'remains positive' (before the book's publication).
- Elsevier: In May and June 2009 gave the statements on sponsored 'journals' summarized in the Elsevier claim above, as reported by The Scientist.
- Original Study 329 investigators (Keller et al.): Disputed the 2015 reanalysis, saying secondary measures were added before the blind was broken (held exhibit 93).
- No suit or correction recorded: No lawsuit over the book and no published list of corrections were found in the sources searched (Europe PMC, web searches, ABPI and PharmaTimes coverage). The book invites corrections and states no libel case has been brought against the author (e-text, 'A Note on Errors').
Where it connects
- The Filtered Record: Reboxetine, Study 329 and Tamiflu are the chapter's case histories of trial data unpublished or released only in part, and the Roche, EMA and AbbVie disputes document who held the reports and on what terms.
- What Was Known and When: Paroxetine in adolescents and rosiglitazone: what GSK held internally, the Senate Finance staff findings and the 2012 Justice Department plea.
- Money, Fees and the Revolving Door: User fees at FDA (MED-6.1), EMA staff moves to industry (Lonngren, not yet checked) and sponsored medical education; the 'almost entirely' fee claim needs the FY 2024 figure.