Published on

Conspiricy Theory For the critical minds

Published in: Health & Medicine, Business
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide


  1. 1. B R I T I S H J O U R N A L O F P S YC H I AT RY ( 2 0 0 3 ) , 1 8 3 , 2 2 ^ 2 7 Interface between authorship, industry METHOD This article distinguishes between traditional and science in the domain of therapeutics{ and non-traditional authorship on the basis of a judgement as to whether the authors DAVID HEALY and DINAH CAT TELL are free in a traditional manner to share with others the raw data from studies they author. We have assumed that authors working on company-sponsored articles are, in general, not at liberty to share proprietary raw data and are even less likely to do so if they have not seen the Background Changes in the character Traditionally scientific authors generate, raw data in the first instance. By raw data of medical authorship. analyse and have access to raw data and here is meant untabulated data; tabulation prepare an article that disinterested is arguably a primary and key act of Aims Tocomparetheimpactof industry- observers would accept reflects an appro- authorship. In pharmaceutical-company- linked andnon-industrylinkedarticles. priate interpretation of those data. Author- sponsored clinical trials, this initial tabula- ship has been changing, however, and tion is invariably performed either within Method We comparedarticleson journals now accept that articles may be the company or within a contract research sertraline beingcoordinatedbya medical authored by individuals who have made a organisation that passes on tabulated data substantial contribution to the conception and trial reports to medical writing writingagencywitharticlesnotcoordinated and design or the acquisition of data or agencies. This practice, almost by defini- inthisway. We calculatednumbersof inthisway . calculatednumbers of analysis and interpretation of data in a tion, gives rise to a non-traditional form Medline-listedarticlesper author, journal study, or who have drafted or critically of authorship. In contrast, we have as- impactfactors, literature profiles and revised the intellectual content of an article sumed that individuals who conduct studies and who have approved the final version of of their own design, regardless of funding citationratesof bothsetsof articles. both setsof the published article (International Com- source, can share raw data, if necessary. Results Non-agency-linked articles on mittee of Medical Journal Editors, 2000; We have used two data sources: Rennie et al, 2002). This new authorship al, Medline and EMBASE literature retrieval EMBASE sertraline had an average of 2.95 authors matrix is consistent with many articles services searching for the word sertraline per article, a mean length of 3.4 pages, a being ghostwritten (Davidoff et al, 2001). al, in the titles of articles from 1998, which mean Medline listing of 37 articles per Unacknowledged editorial or writing were scrutinised for articles referring to author (95% CI 27^47) and a mean assistants to academic authors – so-called the therapeutic uses of sertraline; and a literature profile of 283 per article (95% ghostwriters – are often employed by document prepared for Pfizer Pharma- medical communication agencies working ceuticals by Current Medical Directions CI130^435). Agency-linked articles on for pharmaceutical companies. Efforts have (CMD) on 29 January 1999, which gives sertraline had an average of 6.6 authors been made to quantify the extent to which a worldwide status update for 85 articles per article, a mean length of10.7 pages, a ghostwriting is happening, with Flanagin on Pfizer’s antidepressant sertraline, some mean Medline listing of 70 articles per et al (1998) reporting that up to 11% of of which had been published in 1998 and articles published in six peer-reviewed others subsequently in 1999, 2000 or early author (95% CI 62^79) and a mean journals in 1996 involved the use of ghost- 2001. The CMD document was made literature profile of1839 per article (95% writers. available to us on a non-confidential basis CI1076^2602).The citation rate for There are a number of delicate issues in the course of legal proceedings. agency articles was 20.2 (95% CI13.4^ that need to be teased out in this area, Current Medical Directions is a medical 27.0) and for non-agency articles it was ranging from the practicalities of regulating information company, based in New York authorship to the more profound questions and set up in 1990 to deliver scientifically 3.7 (95% CI 3.3^8.1). of whether ghostwriting is an unfortunate accurate information strategically devel- Conclusions The literature profiles accidental development in the scientific oped for specific target audiences (http:// enterprise or whether it reflects some This agency writes and citation rates of industry-linked and fundamental aspect of the way modern up studies, review articles, abstracts, jour- non-industry-linked articles differ.The science is conducted. There can be few if nal supplements, product monographs, ex- emerging style of authorship in industry- any of these issues or questions, however, pert commentaries and textbook chapters. linked articles can deliver good-quality that would not benefit from some quantifi- It conducts meta-analyses and organises cation of what is happening. Against this journal supplements, satellite symposia articles, but it raises concerns for the background we have sought to quantify and consensus conferences, as well as scientific base of therapeutics. the literature profile of articles on one drug, advisory boards for its clients. sertraline, that were in production in 1998. The CMD document indicates that Declaration of interest D.H. has had CMD was coordinating articles on sertra- consultancies or other links with all major line. These articles appear to involve pro- { pharmaceutical companies. See editorial, pp. 3^4, this issue. prietary data in almost all instances. There 22
  2. 2. I N T E R F A C E B E T W E E N AU T HO R S H I P, IN D U S T RY A N D S C I E N C E IN AC I N DU ND were a number of publications that the The impact factor of a journal and the This leaves 37 papers dealing with the document suggests originated within com- Medline listings linked to academic authors therapeutic effects of sertraline. In addition munication agencies, with the first draft of give an estimate of the potential impact of to those articles retrieved by Medline, we articles already written and the authors’ an article, and as such can be expected to have included a further four EMBASE- EMBASE- names listed as ‘to be determined’. In the guide the considerations of a company such listed papers not found in the CMD case of subsequently published articles in as CMD, which ‘strives to exceed the ex- document. EMBASE also retrieved a EMBASE this series, the authors’ names are available. pectations’ of their clients and ‘assist them further seven papers on toxicology and A further series of articles had very similar in achieving their strategic objectives’ three on biochemical studies, which are academic and company authors, already ( The actual not considered further. published or with authors’ names desig- impact of an article may, however, differ Thus, in total, there are 41 non-CMD nated. Finally, there were articles that do significantly from its apparent potential. articles on therapeutics with sertraline: 19 not appear to have been written within a We have therefore also established the ac- are categorised by the Medline retrieval communication agency and do not have a tual citation rates for the articles in both process as journal articles or clinical trials; Pfizer name on them, but they acknowledge series using the Institute for Science 16 are categorised as letters; and 6 are cate- Pfizer funding or support. Some of these Information Web of Science database. For gorised as randomised controlled trials. Of articles involve economic models, con- citation rates we have restricted the com- the 41 articles, 3 report ambiguous findings structed on the basis of tabulated rather parison to articles from both series that for sertraline, 20 report negative findings than raw data. Others are review articles. were published in 1998. and 18 report positive findings, including Three involved clinical trials. In this study, the appearance of an indi- positive results for depression (3), for The Medline and EMBASE articles on EMBASE vidual’s name on an article is designated premature ejaculation (2) and for dialysis sertraline include articles listed in the as an authorship. An individual author, hypotension (1). Of the 41 non-CMD CMD series and some not listed in that therefore, may have several authorships. studies, three received support from Pfizer series, henceforth called the non-CMD but the authors appeared likely to be in series. In the non-CMD series, the majority possession of the raw data. are reports of studies not supported by RESULTS Of the 20 papers offering negative find- Pfizer and only one appears to have in- ings for sertraline, 16 detail adverse effects, volved the generation of proprietary data; Using Medline with sertraline as a keyword including serotonin syndrome, hypomania, accordingly we have assumed that these and searching article titles we found 59 hyponatraemia, suicide attempts, extra- authors are in a position to share raw data distinct articles in 1998 with sertraline in pyramidal problems, urinary retention and if requested. the title. Altogether for 1998, 1999 and priapism. There were, in addition, one re- We have attempted to estimate the 2000 Medline listed 81% of the CMD view paper on extrapyramidal problems as- impact of these two different groups of articles published in 1998, 1999 and 2000 sociated with sertaline use and one negative articles as follows. The impact factor for (excluding supplement and health econom- trial on the use of sertraline in chronic pain. each journal was established using Journal ics articles). In 1998, 12 of the 20 articles Five of the 41 papers appeared in the Citation Report for 1999 from the Institute appearing in the CMD document appeared Journal of Clinical Psychopharmacology, Psychopharmacology, for Scientific Information Inc. For one in the Medline search. Of the eight not three in the Journal of Clinical Psychiatry, Psychiatry, journal in the CMD series and for four appearing in Medline, five came from the two in Psychosomatics, one in the Ameri- Psychosomatics, journals in the non-CMD series it was not only supplement in the CMD series and can Journal of Psychiatry and the rest in a possible to obtain impact factors and so three from health economics journals; none number of lower-impact journals, including these have not been included. Using of these eight articles had sertraline in their some non-English-language journals. There Medline, we systematically searched for title. were 121 authorships from 120 individual the number of Medline listings for each authors and an average of 2.95 authors author in both the CMD and non-CMD per article. These articles were 3.4 pages author series. This permits us to offer three Non-CMD articles in length on average. estimates of literature impact. First, we Excluding from the Medline series those have estimated the mean number of Med- articles listed by CMD leaves 47 papers line listings for a CMD author versus a that included sertraline in the title. We have The CMD articles non-CMD author, giving an estimate of excluded a further ten papers as follows. The CMD’s document outlined 85 papers author impact. Second, we have assumed One was a Pfizer-funded large multi-centre in the production process during 1998: the total number of Medline listings for study that outlined therapeutic advantages two appeared in 1998 but have a 1997 date all authors of an article and multiplied by for sertraline in depression. The nature of (these are excluded from the analysis); 20 the journal impact factor; taking a mean this paper and its funding suggest an over- appeared in 1998; 18 appeared in 1999; of these values for all articles in a respective lap with the CMD series of articles laid and 17 appeared in 2000 or early 2001. series gives a literature profile per article out below, in that the data are proprietary. We have used the results for all these from each series. Third, we have multiplied This article has not been counted in either articles published from 1998 to 2000 to the mean literature profile per article, as series. Eight papers that deal with animal, generate mean impact factors for authors, calculated above, by the number of articles healthy volunteer or non-therapeutic meta- journals and articles, but have generated in each series to give an annual literature bolic research were excluded, as was one the overall annual literature impact for the impact factor for both the CMD and non- letter that offered a comment on the CMD series from the articles published in CMD series. methodology of a sertraline trial. 1998 alone. 23
  3. 3. H E A LY & C AT T E L L The 85 articles cover depression (14), Table 1 Journals taking articles on sertraline linked with Current Medical Directions Table seasonal affective disorders (1), dysthymia (7), panic disorder (8), post-traumatic stress Name of journal No. of articles Journal impact factor disorder (2), general anxiety (2), obsessive– (1999 ISI data) compulsive disorder (1), differentiation between selective serotonin reuptake Journal of Clinical Psychiatry 7 4.2 inhibitors (17), what is termed ‘outcomes Journal of Psychopharmacology 7 2.8 research’ (largely pharmaco-economic American Journal of Psychiatry 6 6.3 articles) (10), the use of sertraline in Journal of the American Medical Association 3 11.4 the elderly (10), the use of sertraline in Archives of General Psychiatry 3 11.0 children (6), the use of sertraline in women Journal of Affective Disorders 3 2.1 (4), sertraline pharmacokinetics (2) and sertraline in paedophilia (1). Journal of the American Academy of Child and Adolescent 3 3.6 The 55 published articles that form the Psychiatry basis of this analysis have a mean length of Journal of Clinical Psychopharmacology 3 5.7 10.7 pages, with 365 authorships drawn International Clinical Psychopharmacology 2 1.1 from a total of 207 individual authors, Archives of Family Medicine 2 1.4 giving a mean of 6.6 authors per article. ISI, Institute for Science Information. Of these, there are 182 academic and 25 company authors. Two of these articles follow current journal guidelines and acknowledge writing support from indivi- duals not listed as authors. These 55 articles Table 2 Table The literature profile of Current Medical Directions (CMD) and non-CMD articles offer the results of 25 clinical trials from a number of different therapeutic areas, Literature status Results including areas in which Pfizer were seek- ing licenses at that time for sertraline, in Non-CMD articles (n¼41) (n 41) addition to eight review articles and six Authorships/authors (n) (n 121/120 articles offering economic models based Authors per article (mean) 2.95 on Pfizer trial data. All of the clinical trial Pages per article (mean) 3.4 results were favourable to Pfizer, as were Medline listing per academic author (mean (95% CI)) 37 (27^47) the economic analyses. One of the review Literature profile per article (mean (95% CI)) 283 (130^435) articles, from a Pfizer author, offered a frank acknowledgement of the capacity of Annual literature profile of series1 10 471 sertraline to induce agitation/akathisia and Series citation rate (n¼37) (mean (95% CI)) (n 37) 3.7 (3.3^8.1) the links between this and treatment- Citation rate of series, excluding letters (n¼23) (mean (95% CI)) (n 23) 7.7 (4.1^11.3) induced suicidality (Lane, 1998). (This CMD articles (n¼55: 18 per year) (n 55: may be because the intention of Lane’s Authorships/authors (n) (n 365/207 (122/69 per year) paper was to place the clinical problem in Authors per article (mean) 6.6 context rather than to identify specific Pages per article (mean) 10.7 problems.) The 55 papers appeared in the Medline listing per academic author (mean (95% CI)) 70 (62^79) journals listed in Table 1. The British Medline listing per company author (mean (95% CI)) 17 (14^20) Medical Journal, European Psychiatry, the Journal, Psychiatry, Literature profile per article (mean (95% CI)) 1839 (1076^2602) Psychiatry, British Journal of Psychiatry, the American Annual literature profile of series1 34 941 Heart Journal, Pharmacoeconomics and 13 Journal, other journals published a single article Citation rate of series (n¼19) (mean (95% CI)) (n 19) 20.2 (13.4^27.0) each. 1. The annual literature profile for each series of articles was calculated by multiplying the mean literature profile per 1.The Of the 85 articles, 23 are listed as poss- article by the number of articles from the series for that year. ibly originating within communications agencies. Of the 55 published articles, the names of several academic authors appear on more than one article, with one indivi- dual being named as a co-author on 12 of 13 are review articles appearing in a Comparison of CMD these articles. Of the published articles, 13 company-sponsored symposium supple- and non ^ CMD articles of the 55 do not appear to have a company ment. The remaining four articles acknowl- author or to have been through an agency. edge support funding, of which three In Table 2 we list the mean number of Four of these 13 articles involve economic involve clinical trials. Of these three trials, authors per article, the mean number models based on data provided by Pfizer, Pfizer personnel are listed as having re- of pages per article and the mean number and it is assumed that these authors do viewed draft articles in two, but the authors of Medline listings per author for each not have access to raw data. Five of the appear to hold the data. series. There are statistically significant 24
  4. 4. I N T E R F A C E B E T W E E N AU T HO R S H I P, IN D U S T RY A N D S C I E N C E IN AC I N DU ND differences between the two series of DISCUSSION ghostwriting. Authorship lines from per- articles on each of these features. In addi- ceived opinion-leaders with minimal com- tion, as outlined we list the mean literature These data address two issues in the scienti- pany representation and non-declaration profile per article in each series. Finally, we fic literature. First, they offer a further of other non-academic authorship inputs constructed an annual literature profile for quantification of the number and impact increase the likelihood that these articles each series by multiplying the mean litera- of articles based on proprietary data, and will be influential with prescribers and ture profile per article by the number of the possible extent of ghostwriting based purchasers. articles from the series that year. Using a on a single drug. Second, they offer a first paired sample t-test, the two series of arti- set of figures on the likely impact of a series cles differed significantly in terms of Med- of articles prior to publication and the Effect of ghostwriting on academia line listing per authorship (P40.001; 95% (P subsequent citation rates of those articles. One of the expressed concerns about ghost- CI for the difference between the series writing has been the way in which this pro- was 11.2–42.4) and in terms of the litera- cess leads to a lack of recognition for the ture profile per article (P40.002; 95% CI (P Literature impact people who actually write the articles. The for the difference between the series was In the debate on how to evaluate the scien- converse of this point is that academics 623–2570). tific literature, Seglen (1992) has argued become opinion leaders in a therapeutics In addition to the above, we have deter- that citation rates rather than journal field because they appear to have their mined the citation rates for the CMD and impact factors should be used. We have names on a larger proportion of the litera- non-CMD series of articles. We have com- used both measures and, in addition, a ture appearing in the most prestigious pared the CMD articles both with the mean composite literature impact measure. The journals than do others and because they citation rates for the whole non-CMD debate on how to evaluate articles has get asked to national and international series (n¼37) and for the non-CMD series (n 37) hitherto focused on the extent to which meetings to present data with which they with the letters excluded (n¼23). In each (n 23). scientific articles may or may not have may not have first-hand acquaintance. case, there is a significant difference be- moved a scientific field forward. Citation Whether or not the academic authors in this tween the data-sets. Using a paired t-test, rates arguably reflect the true scientific series saw the raw data from the studies the the difference between the two series was worth of an article better than the impact CMD articles are based on, these authors statistically significant at P40.001 (95% factor of a journal. However, in the field cannot share proprietary raw data with CI for the difference between the series of therapeutics, pharmaceutical companies colleagues in the way that has been was 9.6–22.7). Comparing the CMD series may be more interested in short-term gains traditional in the scientific domain. This, with the non-CMD series excluding letters with major purchasers than in developing allied to the volume of industry-linked gives a result of P40.001 (95% CI the science base of the field. To the extent authorship, indicates a process of changing for the difference between the series was that this is happening, the prestige of scientific authorship that could conceivably 7.1–20.8). journals and their apparent authors will culminate in a situation in which the domi- In addition, in the non-CMD series be of greater importance to them than the nant figures in therapeutics actually have there was a mean journal impact factor of actual citation rates of articles. Indeed, in comparatively little first-hand research 3.0 (95% CI 2–4) for articles reporting a mirror image of Seglen’s arguments for experience and few raw data that they can beneficial effects of sertraline, versus 1.78 other scientific domains, citation rates seem share with others. (95% CI 1–2.5) for those reporting nega- at risk of being artificially boosted by It should be acknowledged that there tive effects. The mean literature profile for ghostwriters for companies in a way that are a number of good aspects to the ghost- favourable articles was 351 (95% CI 59– is less likely to happen for journal impact writing process. First, authorship by a 643), versus 172 (95% CI 1.7–337) for factors. The findings reported here, how- communications agency or within a negative articles. ever, appear unaffected by the method used company makes it more likely that at least Finally, when the CMD articles were to evaluate the respective literatures. some of the results of research will enter considered on their own, there was a statis- The profile of the articles reported here the public domain than if the production tically significant correlation between suggests that the background of certain of articles were left to the senior clinicians journal impact factors and citation rates authors may have increased the possibility involved in clinical trials. Second, the (r¼0.67; P40.01). When both CMD and 0.67; of the company’s publications appearing quality of the writing is probably consis- non-CMD articles for 1998 were consid- in the most prestigious journals. Specific tently superior as a consequence. Third, ered, there was a significant correlation journals seem to have been targeted. The there is every reason to believe that at least between journal impact factor and citation combination of distinguished journal, some communications agencies will take rates (r¼0.71; P40.01), which increased (r 0.71; distinguished author, an efficient distri- the efforts by journal editors to encourage further if the letters in the non-CMD series bution system and sponsored platforms disclosure of interests more seriously than were excluded (r¼0.74; P40.01). There (r 0.74; appears to have led to an impact on the many academic investigators will. Fourth, were comparable statistically significant therapeutics domain greatly in excess of there are data to indicate that the reporting correlations between citation rates and the 50% of the impact of the rest of the litera- of adverse events in company-sponsored composite potential literature profile ture on sertraline. The impact of this literature and monitored clinical trials is more measures that we constructed, as well as on third-party payers and other interested comprehensive than the reporting of between journal impact factors and the parties is at present unquantifiable. The adverse events in government-sponsored composite potential literature profile question of literature impact would or other independent studies (Shamoo, measures. seem to be tied closely to the nature of 2001). 25
  5. 5. H E A LY & C AT T E L L However, analyses of published results on antidepressant studies in recent years CLINICAL IMPLICATIONS have made it clear that a considerable pro- portion of negative results are not pub- & This paper makes it clear that an increasing proportion of the clinical trial lished, to the extent that the sponsorship literature in pharmacotherapeutics is managed through medical writing agencies. of a published study is now a demonstrable predictor of the findings of that study (Free- & The articles reported here contain significant discrepancies between published mantle et al, 2000; Gilbody & Song, 2000). al, data and the raw data from the actual clinical trials. There is, however, little reason to believe that this bias does not affect the entire & In addition to already recognised implications for ghostwriters, the new style of domain of therapeutics, including psy- authorship has implications for academic authors. chotherapy, whether supported by pharma- LIMITATIONS ceutical companies or not. The tensions involved show in the figures reported here. & From the available material it is not possible to know what data the apparent On the one hand, the CMD-linked articles authors of articles were privy to. report universally positive results. On the other hand, the CMD-linked articles & In the absence of clear agreement on how to assess the impact of academic contain a much higher proportion of articles, it is not possible to know for certain the relative longer-term impact of randomised controlled trials, convention- company-supported articles versus independent articles. ally seen as offering a superior calibre of data, than do the non-CMD articles. & It is unclear what proportion of studies or data on sertraline for the relevant period have been published. Problems with ghostwriting If the methods employed in industry-linked authorship make the publication process both more efficient and more effective, then DAVID HEAL FRCPsych, DINAH CATTELL, North Wales Department of Psychological Medicine,University of Y, what, if any, are the problems linked to Wales College of Medicine, Bangor, UK new styles of authorship? In addition to having the potential to produce a set of Correspondence: David Healy, North Wales Department of Psychological Medicine,University of Wales College of Medicine, Bangor,Wales LL57 2PW,UK.Tel: 01248 384 452; fax: 01248 371397; authorities on therapeutics with little Hergest @ e-mail: Healy___ clinical experience, we list two other issues. First, most studies are now sponsored, (First received 25 April 2002, final revision 18 December 2002, accepted 6 January 2003) designed and analysed, in addition to being efficiently written, by pharmaceutical companies. This is a process that in psycho- pharmacology picked up pace from 1980 (Healy, 2002a). As long as the greatest 2002a proportion of studies are both undertaken sertraline who committed suicide, and three Attention has previously been drawn to by and published by pharmaceutical others on sertraline who reported increas- possible incongruities in the reporting of companies, the primary questions being ing suicidal ideation necessitating treatment suicidal acts on recent antidepressants in a asked in the therapeutics domain may well discontinuation, in contrast to just one case meta-analysis by Khan et al (2000) (see relate to the marketing interests of those of emergent suicidality on a comparable Healy, 2002b). Importantly, the categorisa- 2002b companies rather than to unanswered drug and no problems on placebo. There tion of suicidal acts on placebo from these scientific questions, as the CMD series of is no reference to these data in the final trials reported by Khan et al is also reported articles outlined here demonstrates. Recent published article. Second, of the six in a number of other articles, suggesting efforts to encourage pharmaceutical published paediatric psychopharmacology that these academic authors may all be companies to publish the results of all of CMD articles, only one article mentions using data previously tabulated by the their studies imply that therapeutics will one suicidal act. There were in fact six respective companies. This has clear impli- become scientific if all studies are pub- suicidal acts on sertraline and three further cations for any assessment of the hazards lished. Complete publication of studies cases of suicidality in the subject group of these drugs, and for the confidence that would, in fact, only bring the field of from which these articles come, including can be placed in the process by which these therapeutics up to an acceptable business four suicidal acts in 44 patients with articles were written. ethics standard. A field is only scientific if depression given sertraline, which is a scientific questions are addressed. rate of 9% (Pfizer Expert Report, 1997). The second issue relates to the corre- The effects of sertraline in paediatric de- A possible solution spondence between published articles and pression were outlined by Alderman et al raw data. The current CMD series throws (1998), who reported only the adverse Problems of this sort could be overcome if, up issues of concern in this area. First, events that occurred in more than 10% of in addition to making available the gross de- one study in this series had one patient on patients. tails of negative studies, as some companies 26
  6. 6. I N T E R F A C E B E T W E E N AU T HO R S H I P, IN D U S T RY A N D S C I E N C E IN AC I N DU ND do, companies also made the raw data or REFERENCES International Committee of Medical Journal Editors (2000) Uniform requirements for manuscripts primary data tables from therapeutic trials submitted to biomedical journals. May. http:// available. This may be seen as a counsel of Alderman, J.,Wolkow, R., Chung, M., et al (1998) J., Wolkow, Sertraline treatment of children and adolescents with perfection, but if pharmacotherapy is to be OCD or depression: pharmacokinetics tolerability and a scientific business rather than just a busi- efficacy. Journal of the American Academy of Child and Khan, A.,Warner, H. A. & Brown,W. A. (2000) Adolescent Psychiatry, 37, 386^394. Psychiatry 37, , Symptom reduction and suicide risk in patients treated ness adopting the appearances of science, with placebo in antidepressant clinical trials: an analysis no less than this is needed. It should be Davidoff, F., DeAngelis, C. D., Drazen, J. M., et al of the Food and Drug Administration database. Archives remembered that the capitalisation of the (2001) Sponsorship, authorship, and accountability. New accountability. of General Psychiatry, 57, 311^317. Psychiatry 57, , England Journal of Medicine, 345, 825^827. Medicine, 345, industry depends entirely on the voluntary Flanagin, A., Carey, L. A., Fontanarosa, P. B., et al Lane, R. M. (1998) SSRI-induced extrapyramidal side participation of health care consumers in (1998) Prevalence of articles with honorary authors and effects and akathisia: implications for treatment. Journal studies of the kind reviewed here. If ghost- ghost authors in peer-reviewed medical journals. JAMA, JAMA, of Psychopharmacology, 12, 192^214. Psychopharmacology 12, , writing is an inevitable feature of modern 280, 222^224. 280, scientific writing, the potential availability Freemantle, N., Mason, I. M. & Y Young, P. (2000) oung, Pfizer Expert Report (1997) Sertraline Hydrochloride of the raw data would do more to ensure a Predictive value of pharmacological activity for the for Obsessive Compulsive Disorder in Paediatric Patients relative efficacy of antidepressant drugs. Meta- (approved 20 October 1997). New York: Pfizer. York: correspondence between those data and a regression analysis. British Journal of Psychiatry, 177, Psychiatry 177, , published end result than could be achieved 292^302. Rennie, D., Flanagin, A. & Yank,V. (2002) The by any other mechanism. contributions of authors. JAMA, 284, 89^91. JAMA, 284, Gilbody, S. M. & Song, F. (2000) Publication bias and the integrity of psychiatry research. Psychological Medicine, 30, 253^258. Medicine, 30, Seglen, P. O. (1992) Why the impact factor of journals ACKNOWLEDGEMENT should not be used for evaluation research. BMJ, 314, BMJ, 314, Healy, D. (2002a) The Creation of Psychopharmacology. (2002a Psychopharmacology. 498^502. The authors acknowledge input from Professor E. Cambridge, MA: Harvard University Press. Young of Ann Arbor, Michigan, USA, to the drafting _ (2002b) SSRIs and deliberate self-harm (letter). (2002b Shamoo, A. E. (2001) Adverse event reporting ^ the of this article. British Journal of Psychiatry, 180, 547. Psychiatry 180, , tip of an iceberg. Accountability in Research, 8, 197^218. Research, 27