CHICAGO – Many of you are aware of the phenomenon of “publication bias” in the scientific literature. This most commonly refers to the fact that “negative” results find their way into publication much less commonly than “positive” results.

Given human nature, rarely does one want to publicly show their own hypothesis being disproved. While scientists are certainly a competitive sort, the profession finds it more dignified to go about constructively proving one’s own hypotheses than necessarily trashing others.

The lack of public disclosure about negative results is a shame since one of the most important scientific experiments of all time – the Michelson-Morley experiment of 1887 – utterly failed to prove the hypothesis of the stationary ether. This would have experimentally implied that the speed of light would vary according to its orientation with respect to the earth’s movement through space.

The hypothesis of the stationary ether, which buttressed the careers of nearly all professional physicists at the time, was trashed. Forward thinkers needed to entertain the concept that the speed of light was a fixed constant. The negative Michelson-Morley result made possible the more radical, positive steps that Einstein took with his theory of relativity.

Human nature was not enough to suppress publication of the Michelson-Morley experiment. In fact, this result was so important (albeit negative) that Michelson and Morley gained worldwide renown (and Albert Michelson a Nobel Prize in 1907) for their work. That’s not typically the case.

In our less-innocent modern age where money often rules science more than gentility, there are additional forces influencing publication bias. Who knows what would have happened if International Ether, Inc. had sponsored the experiment.

Would the publication have been suppressed? Would the conclusions have been made sufficiently obtuse that proper interpretation was not possible? For the price to maintain International Ether’s stock, the world – with no nuclear power, no atom bomb and no advanced wireless technology, which are all dependent on relativity – would have been strikingly different.

Especially under the influence of financial pressures, publication bias can be very subtle. I came across this on Dec. 26, 2006 as I read a Wall Street Journal article headlined “Medical Journal Spikes Article on Industry Ties of Kidney Group”.

The article reported on the publication turmoil surrounding recent studies linking the use of the drug erythropoietin (EPO) in kidney disease patients to an increased risk of heart failure and an accelerated progression to dialysis.

The article looked into a complex set of circumstances in which the New England Journal of Medicine refused to publish (the more colorful term is “spiked”) an editorial by former deputy editor Robert Steinbrook. Instead, Steinbrook submitted the editorial for publication by a rival publication: the British journal Lancet.

Steinbrook’s editorial in question pointed out that another article in the New England Journal of Medicine did not sufficiently reveal underlying financial conflicts of interest. Except for the fact that very illustrious and accomplished physician scientists were involved, it certainly sounded like something out of Jerry Springer.

The reason the article caught my eye was not necessarily because of the journalistic politics but because I had also written about this very topic as a medical student back in 1989. At that time, I sent a letter to the Annals of Internal Medicine commenting on what was then one of the very first studies evaluating the use of EPO in pre-dialysis kidney disease patients.

In that letter – which is published for your perusal here – I acknowledged that EPO “clearly ameliorated the anemia”. I also pointed out that “it is important to consider the possibility of two significant side effects: systemic hypertension as well as an acceleration in the progression of chronic renal failure”.

I based my conclusion on the fact that while the original paper stated in both the abstract and its summary that there were no changes in mean blood pressure, careful reading of the methods section revealed that three of the 11 treated patients required an increase in their anti-hypertensive medications.

I interpreted this to mean that this dangerous, hypertensive side effect was in fact masked by the change in anti-hypertensive regimen and that the conclusions of the paper could – to put it graciously – be misinterpreted.

To put it less graciously, the authors were covering up a “negative result” (e.g. a potential adverse effect) by medicating it away and not forthrightly presenting that in the conclusions to their paper. Though these 1989 studies were sponsored by Ortho Pharmaceuticals, I decided not to point that out thinking this would be too inflammatory.

As a mere medical student, I certainly had no industry ties, no fancy consulting contracts and no stock options (nor do I have at present any such industry ties related to EPO). Nevertheless, my 1989 letter was never published. As the phrase goes, it was spiked.

In retrospect, this is a shame since greater awareness as early as 1989 of this problem (e.g. getting past the veil of publication bias) could have saved untold millions of dollars and perhaps many patient lives. Ironically, 1989 was a very different time than 2007. Back then, not having significant industry and financial support put into question the legitimacy of what one had written.

Now, the sensitivities and tables have turned. It is those who have significant financial conflicts of interest whose legitimacy now begins to be questioned. So what is to be done?

A problem as complex as publication bias – and all its attendant consequences – cannot be addressed in a short article such as this column. The onus recently has been placed on authors to proactively document all their financial conflicts of interest. That trend should continue to be encouraged. However, bias goes beyond just financial incentives.

All of this is human endeavor with human frailties and human powers. As circumstances dictate, it is important for readers to understand some of the other, more subtle (yet still powerful) forces that determine bias. The presence of bias – which is always there – is not automatically bad nor does it necessarily invalidate information.

It just needs to be recognized and taken into consideration.

Even Einstein was biased. He was certainly biased against Nazi Germany when he brought the possibilities of the atomic bomb to President Roosevelt. Though relatively agnostic, Einstein did invoke his belief that “God does not play dice” in his disbelief of quantum mechanics. Biased as he was, Einstein certainly did make enormous contributions to the world.

Perhaps you might be asking: “What is the bias of the author of this column?” Financial incentives? No. As a columnist and in my own line of work, I try to anticipate future directions in biotech and medical technology. Certainly my bias would be to exercise and demonstrate such a predictive capability.

Feel free to accuse me of bias in the context of showing that I predicted 15 years ago the developments in EPO that are hitting the press now. I’ll leave it up to you whether that invalidates the opinions expressed here.