Monday, November 30, 2009

How Industry Views the Research It Sponsors

We have posted frequently about threats to the integrity of the clinical evidence-based, and to the practice of evidence-based medicine.  In particular, we have discussed how research may be manipulated in favor of vested interests, or suppressed when the results do not favor such interests.

Last week, the British Medical Journal electronically published a set of guidelines for how industry sponsored clinical research ought to be published, sponsored by the International Society for Medical Publication Professionals.  The authors came from pharmaceutical companies (Johnson & Johnson, AstraZeneca, Pfizer and Cephalon), medical device companies (LifeScan), and medical publishing and medical education and communication companies (John Wiley & Sons, Excerpta Medica, Field Advantage Medical Communications, PharmaWrite, and Knowledgepoint 360 Group).  [Graf C, Battisti WP, Bruce-Winkler V et al. Good publication practice for communicating company sponsored medical research: the GPP2 guidelines.  Brit Med J 2009; 339:b4430.  Link here.]

These guidelines are remarkable for the questions they raise about how people from industry view clinical research and how it should be reported in medical journals. 

Who's in Charge?

Nowhere does the article acknowledge that any one person has overall responsibility for a research project.  Clinical research projects funded by the US National Institutes of Health, Agency for Healthcare Quality and Research, and other federal agencies must have "principal investigators," who are the people who take overall scientific responsibility for the project.  The Graf et al article does not use this or an equivalent term.   There is no sense that they expect anyone to be in completely in charge of industry sponsored research. 

Particularly confusing is the following passage:
Before writing begins one author (a lead author, who may also be guarantor) should take the lead for writing and managing each publication or presentation. One author (identified as guarantor) should take overall responsibility for the integrity of a study and its report.

Here are some questions it raises.  If the guarantor could be chosen only when writing a paper is contemplated, which presumably could be years after the study that the paper would report was designed and implemented, who then would have been responsible for study "integrity" before the guarantor was chosen? Who would chose the guarantor for a particular paper? If a study generates more than one paper, could they each have a guarantor, and then how could they share responsibility for study "integrity?"  If the guarantor and lead author of a paper are different people, how would they share responsibilities, what would happen if they were not to agree, and who would be finally accountable?

So the guidelines seem to completely diffuse accountability for research projects, and the reports written about them.

Who is an Author?

In my experience with US federally and foundation funded research, research papers are written by the investigators, the people who actually did the research project.  However, in the guidelines by Graf et al, the concept of authorship is also ambiguous.  They suggest that "recognised criteria should be used to determine which of the contributors to an article should be identified as authors."  This is already confusing, since how could one be a "contributor" to an article without authoring it?  A later discussion of "contributors" as "investigators, sponsor employees, and individuals contracted by the sponsor" was not very helpful.  Why should a "sponsor employee" who was not an "investigator" be a "contributor" or an "author," whatever the distinction is between them? 

So the guidelines blur distinctions among people who do research, and people employed by companies that may have vested interests in the research favoring their products or services.

What do Professional Medical Writers Do?

There has been considerable recent controversy, not directly acknowledged by Graf et al, about the role of professional medical writers in the reporting of research and the writing of ostensibly scholarly medical publications, particularly in cases where the writers were paid by and reported to corporate sponsors, but were not recognized as such in the publications they wrote (a type of ghost-writing).  The guidelines by Graf et al do not clearly explain what the roles of professional medical writers ought to be:
Professional medical writers must be directed by the lead author from the earliest possible stage (for example, when the outline is written), and all authors must be aware of the medical writer’s involvement. The medical writer should remain in frequent contact with the authors throughout development of the article or presentation. The authors must critically review and comment on the outline and drafts, approve the final version of the article or presentation before it is submitted to the journal or congress, approve changes made during the peer review process, and approve the final version before it is published or accepted for presentation.

Note that this would not prevent a professional medical writer from writing an initial outline, the first draft, and all subsequent drafts. including the final draft of a paper. (The role of an "author" above might be restricted to simply commenting on and accepting the outline and all drafts.) Thus, the "authors" could function as distant editors, and the professional writer would assume authorship, as most people would define it. (Merriam-Webster: "1. one that originates or creates 2. the writer of a literary work.")

Furthermore, while the professional writer could take one the role of authorship, the guidelines do not require him or her to publicly acknowledge this role:
Professional medical writers, depending on the contributions they make, may qualify for authorship. For example, if a medical writer contributed extensive literature searches and summarised the literature discovered, and by doing so helped define the scope of a review article, and if he or she is willing to 'take public responsibility for relevant portions of the content' then he or she may be in a position to meet the remaining ICMJE criteria for authorship.

Presumably, a professional writer could dodge authorship by simply being "unwilling" to take such public responsibility.

So the guidelines apparently condone nearly all functions commonly assumed to be those of an author to be performed by a professional writer paid directly by the sponsor, without the writer being listed as an author.  The guidelines thus appear to condone ghost-writing in its most pernicious form.

Who Owns and Analyzes the Data?

Cases in which various the implementation and analysis of clinical research seems to have been manipulated to favor vested interests have raised concerns about the integrity of the data collected in the course of a research project, and how it is analyzed.  This is what Graf et al say about the ownership and use of the data:
Sponsors have a responsibility to share the data and the analyses with the investigators who participated in the study. Sponsors must provide authors and other contributors (for example, members of a publication steering committee or professional medical writers) with full access to study data and should do so before the manuscript writing process begins or before the first external presentation of the data. Information provided to the authors should include study protocols, statistical analysis plans, statistical reports, data tables, clinical study reports, and results intended for posting on clinical trial results websites. Sufficient time should be allowed for authors and contributors to review and interpret the data provided and to seek further information if they wish (for example, access to raw data tables or the study database).

The guidelines by Graf et al suggest that the company that sponsors the research should own the data. The investigators who collected the data and implemented the research project should not. At best, the company should "share" summaries, analyses, or pieces of the data, but at best investigators could have only "sufficient time" to "seek ... access to raw data tables or the study database."

So the guidelines would allow corporate research sponsors to analyze the data from studies evaluating their own products and services as they see fit, and the scientists who implemented the study and collected the data could only ask for access to it. 


The guidelines by Graf et al seem based on a very strange conceptualization of clinical research. In their view, no individual may be responsible for a clinical research project. Research data is controlled by the company that paid for the project, not scientists who implemented the research and collected the data. Research papers may be written by anonymous professional writers while the scientists who did the research only need to review and approve what they have written.

So why should anyone give credence to industry sponsored research?

We have discussed numerous instances in which clinical research was manipulated in favor of vested interests, and when clinical research whose results did not favor vested interests was suppressed. In most cases, the vested interests were held by for-profit pharmaceutical, biotechnology or device anufacturers acting as research sponsors. The guidelines by Graf et al seem to have been cleverly written to to employ comforting platitudes while licensing manipulation and suppression.  They should inspire no confidence in the integrity of industry sponsored research.


Anonymous said...

On the very first day, of my first undergraduate stat class, over 35 years ago, the instructor looked around, closed the door and said: The only question you ask is: How do you want the numbers to turn out.

This should explain the business attitude towards these research projects and the willingness to skew the date to achieve the desired results.

Steve Lucas

Anonymous said...

Um, the founder and first president of ISMPP is Laurence Hirsch. He was also in charge of publications at Merck, during the Vioxx ghostwriting scandal.

Nuff said.

InformaticsMD said...

Roy, since pharmas now treat employees as expendable assets, there really cannot be a PI of research. They may not be around for long.

In that sense, the pharma model of treating employees as expendable assets is incompatible with good biomedical research.

-- SS

Anonymous said...

Two points: 1)Over 60% of all research is funded by medical companies - not the NIH; and 2)the writer is confusing the person responsible to complete a publication with the role of an investigator - of course the investigator is involved in the collection of trial data - but they are not always interested, or so inclined to participate in the development of publications.

Anonymous said...

Um, Dr. Hirsch was one of many founders of the group and was the 3rd President. One should take care and responsibility for ensuring information is accurate. Interesting that folks still post with anonymity in this day and age of transparency. Are we sure it wasn't a "ghostblogger?"

Michael S. Altus, PhD, ELS said...

Dr. Poses writes in part: "The guidelines thus appear to condone ghost-[authoring] in its most pernicious form.”

In a Dec. 3, 2009, ISMPP teleconference about the GPP2, I posted a question seeking a reaction to Dr. Poses’ comments from presenter Chris Graf, first author of the GPP2. The teleconference did not get to my question.

However, when Mr. Graf discussed the role of professional medical writers, he said, “if they [medical writers] can be [authors] maybe they should be.”

Sadly, “The guidelines thus appear to condone ghost-[authoring] in its most pernicious form.”

Competing interests: I am a medical writer-editor who provides services to individuals and organizations.

Roy M. Poses MD said...

To the anonymous poster of 2 December at 4:32 PM -

True, some investigators may not want or deserve to be authors of a particular paper, but it makes little sense for someone who was not an investigator to be an author.

To the anonymous poster of 2 December at 7:08 PM -

The pot is calling the kettle black.

Anonymous said...

I'm sorry but so much of this article seems to be a willfull misinterpretation of the intentions of GPP2 to promote ethical, transparent publication practice. The whole section querying the difference between 'authorship' and 'contribution' could have been dispensed with by simply following the guidance of GPP2 and referring to the ICMJE authorship criteria. Of course someone can contribute to a publication without being an author - should a junior statstician who makes a significant contribution to the overall data analysis, but who did not have any input on the a-priori methodology be an author taking responsibility for study design and conduct? The answer is no - they should be disclosed and acknowlegded for their contribution however.

We have to start being realsitic about the fact that Pharmaceutical sponsored trials and the timely publication of their data are heavily regulated these days, and that without industry sponsored articles there would be very little evidence base for clinicians to work from. Academic trials are far more likely in this envirnment to flaunt the guidelines and make unsound clinical conclusions, and I say this as an academic...

Roy M. Poses MD said...


Sorry, but the ICMJE criteria still allow various abuses. As I note above, they allow a writer paid by the sponsor to draft the manuscript, and make most of the revisions, including the final revision, but still dodge identification. This, of course, could allow a sponsor to pay such a writer to spin a manuscript in favor of its vested interests without publicly leaving a trace of such influence.

There may be regulation of pharmaceutical sponsored trials, but it has not prevented the numerous instances of research manipulated to favor the sponsors' vested interests, and research suppressed when it fails to to favor such interests discussed on this blog, and elsewhere.

Where is it written in stone that sponsors with vested interests in having research results favor their products should be the predominant source of research funding? Why is it "realistic," rather than cynical to accept this state of affairs as inevitable?

What is your evidence that "academic trials" are likely to be less reliable? And by the way, do you mean trials sponsored by organizations that have no vested interests in the results turning out any particular way, and implemented by academics with no conflicts of interest? Or do you mean industry sponsored trials implemented by academics with numerous personal financial ties to the same companies?

Finally, since you are anonymous, how can we tell whether you are an academic, and if so, whether you are free from relevant conflicts of interest?