Peeling the onion of financial conflicts of interest - Publishing research

Financial conflicts of interest in research bias towards results, conclusions and recommendations that favour the interests of the sources of the money.  Even though those receiving the money or financial support for their projects 

{Photo Credit - Darwin Bell}
always claim theywere not biased, abundant hard evidence shows just the contrary.

Knowing this, health sciences journals have played a leading role in helping to reveal the conflicts: Initially by providing general guidance for authors about disclosure and then, when that didn’t work, by asking explicit questions about financial relationships with individuals or institutions that might benefit or lose or otherwise be affected by the research, review or opinions expressed.  Editors hoped (without any evidence and indeed with little probability) that published revelations would help readers detect biases or at least increase levels of skepticism.

There have been, however multiple loopholes in the process of revealing and reporting conflict of interest the least of which was addressed by a recent proclamation of the prestigious International Committee of Medical Journal Editors (ICMJE).<1> 

Many authors claimed not to know that their relationship with such and such an institution, company, product or patent etc., was a financial one and therefore did not disclose it. Or they believed they needed to reveal the relationship only if they thought it influenced their research. The most recent ICMJE remedy is a set of definitions of what constitutes a financial conflict of interest along with a prototype for reporting such conflicts.

This timid step is not entirely unhelpful. We should see more honest reporting of the type that Author X owns shares in company Y making product Z that is mentioned in their research, review or commentary.  Medical journals, scarred by multiple humiliations of failure to detect and report authors with financial conflicts of interest to their readers ought to have an easier time of it. Their journals will be less likely to hear, after publication, that a particular author failed to understand that he or she had an unreported conflict of interest.

But will this help the readers, patients or other researchers? Knowing that authors of published research have conflicted financial relationships does not mean that the research is flawed. So how should we, the readers,  interpret such statements as;  author X’s study was funded by company Y or that author X, writing a commentary in a journal, receives speaker fees from company Y?  Given that research, reviews and commentary articles by authors with financial conflicts are more likely to be biased does not explain how these conflicts get translated into results, opinions and conclusions and from there to influencing the reader, often a prescribing physician.

For research articles, which form the basis of knew knowledge, this is especially important. Readers and journal editors need to know more about financial relationships between researchers and their sponsors: Not just that they exist.

We know many of the mechanisms embedded in financial conflicts of interest that result in biased studies and biased study reporting in journals.  The recent revelations made public in the class action law suit pitting Pfizer Inc., the maker of Neurontin (gabapentin), against patients who received the drug show that financial conflicts influence and are directly related to  biases in study design, analysis, interpretation, write-up and publication of research. (an in depth review of some of this is here) A summary article of part of the problem was recently published. (article here)<2> 

The law suit and discovery documents associated with it reveal the many ways that researchers lose control of their studies by relinquishing study design, execution, data collection and analysis, interpretation and even writing up of the study and subsequent publication to company sponsors of the research.

We and others have found that in random samples of researchers these sources of real bias are highly prevalent. For example in our study (in press)  only 11% of over 700 Canadian investigators of randomized trials registered in the clinical trials data base had control over 11 key aspects of their research: study design, data collection, analysis, writing up their papers, deciding when and were to publish and so on. (abstract here - then dig into program abstracts) <3> Most had very little control over multiple aspects of the research.

It is important, that ICMJE and journal editors further peel the onion of financial conflict of interest. In situations where there are declared conflicts of interest authors need to report and journals publish details of the research sponsors' roles in the research and in the write up of the manuscript and decision to publish. Editors need to know who had ultimate responsibility for study design, data collection, analysis, write-up and decision to publish.

Some will argue that supplying such detail is unnecessary, time consuming and bureaucratic. These same individuals will also argue that although they may have competing financial interests, these are not conflicts and their presence did not (and does not) influence their integrity as researchers; contrary, however, to a widening body of research that proves the opposite.

Certainly supplying more detail on the relationships between researchers and sponsors will require additional effort and a bit of time. The process can be simplified. In our work, led by Dr. Paula Rochon at the University of Toronto, we have designed an easily-used electronic reporting template that takes about 10 minutes to complete. The ICMJE should consider adopting this template. For randomized controlled trials, trial registries should insist that these key aspects of human research be revealed when studies are first registered. The financial conflicts could be published along with the other aspects of trial design and recruitment.  Research ethics boards should not approve research on human subjects that is shackled by sponsor control over critical aspects of the research. REBs should have ready access to detailed information on conflict of interest and its effects on research they are evaluating. 

The Canadian Institutes of Heath Research (CIHR) is expected to  adopted a checklist based on the one we developed for use in their grant proposal evaluations.  We urge ICMJE and clinical trials registries to push harder at conflict of interest and adopt a similar template as well.  Fuller disclosure can only be helpful.

1.Drazen JL et al.  Uniform format for disclosure of competing interests in ICMJE journals. N Engl J Med 2009;361:1896-7 (full text here)

2. Vedula SS, Bero L, Scherer RW, Dickersin K. Outcome reporting in industry-sponsored trials of gabapentin for off-label ues. N Engl J Med 2009;361(20):1963=71 (free abstract here)

3. Investigator Experiences With Financial Conflicts of Interest in Clinical Trials. Paula A. Rochon, Melanie Sekeres, John Hoey, Joel Lexchin, Lorraine E. Ferris, David Moher, Wei Wu, Sunila R. Kalkar, Marleen Van Laethem, Andrea Gruneir, Jennifer Gold, M. James Maskalyk, David L. Streiner, An-Wen Chan. Peer Review Congress, Vancouver, 2009 (full abstract here - pdf of conference abstracts)

The sacred and overly screened female breast - new recommendations rightly recommend discontinuing screening mammograms in younger women.

The United States Preventive Task Force recently released its latest guidance on screening healthy women for breast cancer. To general astonishment the Task Force recommended that mass screening for breast cancer in healthy women between the ages of 40 ad 40 be stopped, along with efforts to teach women to regularly palpate their own breasts, searching for lumps and bumps.

Image Ductal Carcinoma in Situ From - Breast Cancer.org

The Task Force, an unbiased group of experts in cancer screening, epidemiology and disease modeling of effectiveness determined that the potential benefits were slim and greatly outweighed by the very real harms.Public and political reaction to the announcement was immediate, appropriately embedded in current socio-political narrative of women’s health and widely dismissive of science and evidence: Responses were irrational and often personal - claiming that the members of the Task Force were ‘Bush appointees’.  

I have sympathy for a cult that celebrates an organ which is both a procreative and nourishing. I too would wish it no harm: The point of the Task Force recommendation, however, is that screening for small cancers in the breasts of women between the ages of 40 and 49 years is harmful. How can this be so?

Harms and benefits need to be considered together. It is a rare human environments that sees benefits without risks or harms. For breast cancer, the benefits are often tritely and misleading summed as ‘curing’ cancer. Breast cancer advocates regularly proclaim that breast cancer can be cured, beaten, vanquished, extinguished, destroyed, survived. There are testimonials. But what exactly is a cure?

Common sense has it that ‘cure’ means the particular disease has been eliminated and will never, ever, ‘come back’. We don’t know that for sure about breast cancer, or indeed about almost all cancers. We do know that some women with breast cancer live a long time with the disease and it does not seem to bother them very much. Others suffer greatly and live only a short time. These women can have what appear to be similar types of breast cancer when they are first diagnosed.

Scientists and rationalists, therefore, don’t use the word ‘cure’: They use ‘survival’ and survival rates. What is measured is survival after diagnosis for say 5 or 10 years. Thus we can measure the effects of different treatments including screening with mammography for breast cancer in metrics of 5 or 10 year survival rates. It appears true that the 10 year survival rates for women aged 50 to 69 who are screened by mammography for breast cancer are slightly more likely to still be alive and less likely to die of breast cancer in the 10 years after screening than women who are not screened. This is not cure necessarily (and usually isn’t), but it is an improved chance of survival.

These ‘odds’ are often expressed as the number of women who would need to have a mammogram and the associated treatments if the mammogram was positive in order that one woman’s life would be extended. For women aged 40 to 49 this number is 1,904. This means that 1,904 women would have to be screened in order for 1 woman to benefit from the screening. Looked at this way we can say that 1,903 women would have had an unnecessary mammogram.

How can a mammogram be harmful? There are several harms that are relatively trivial and can be ignored - the hassle of getting the mammogram, the discomfort of the procedure, possible anxiety while waiting for the result. But there are others that are more serious.

First, there are the false-positive mammograms. The mammogram shows a potential lesion and there is then a biopsy or lumpectomy that turns out to be negative for cancer. Both procedures are more than trivial and the latter may deform the breast.

Secondly, and most importantly, by far the most common type of cancer detected is called a ductal carcinoma in situ (DCIS). The cancer is classified by looking at it under a microscope. Almost all cancers detected by mammograms are of the DCIS type. We don’t in fact know much about this type of ‘cancer’. But it is clear that many of these cancers spontaneously revert to normal with no treatment.

Unfortunately we don’t know how to separate those that require treatment fro those that don’t. So the standard practice is to treat them all as malignant (even though we know that most are not). The harms associated with treating a benign tumour are unnecessary surgery (which is often deforming) and the associated chemotherapy and radiotherapy. There is also real anxiety that persists for years and decades of follow-up visits to check for spread of the tumour. Cancer is not a trivial diagnosis.

Difficult decisions are always difficult.

The Task Force could have sidestepped this decision by simply presenting the information - the data - and leaving it to the woman to assess risk. But it didn’t. The Task Force, recognizing that such a decision is all but impossible on individual grounds, made a clear recommendation - for the average woman aged 40 to 49, the odds of improved survival from breast cancer that come from mammography are so small and the odds of having to go through unnecessary surgery and radio/chemotherapy so large that most women in this age group should not have mammograms.

The implications of this recommendation are multiple. First the recommendation ought to mean that as a society we cease to promote mammography screening for women aged 40 to 49. Promotion means organizing screening programs and advertising and promoting them. The evidence indicates that such promotion puts 1903 women at risk of harm in order to possibly prolong the 5 or 10 year survival of 1 woman.

To my mind the Task Force has it right. For the average woman aged 40 to 49, mammography is harmful and should be avoided. Individual women and their physicians might decide otherwise, based on other factors such as family history: But as a public mammogram screening program promising that benefits outweigh harms is a misleading and itself harmful.


Reference:

The Natural History of Invasive Breast Cancers Detected by Screening Mammography
Arch Intern Med. 2008;168(21):2311-2316. get article