Before starting medical school, the author spent a summer doing chemistry research at his state university.
A published photo in a journal showed a tiny, gray comma-shaped particle. It was discovered by taking a photo of any object and then copying it again and again with a copy machine, set to highest magnification. Whether the original photo was that of a person, car, or dinosaur, this process eventually revealed a sheet of these particles, proving them to be the building blocks of all matter. For a while, I was thinking Mike was serious, but then he pointed to the front page of the scientific publication, The Journal of Irreproducible Results,1 and started laughing. Who says scientists aren't funny?
Of course, Franklin was ahead of his time in understanding the beauty of the scientific method, in which carefully performed and described experiments can be repeated and the results confirmed by others. In instances where results are not confirmed, the search for the explanation of conflicting outcomes can also lead to scientific insights.
I have to admit that, these days, the quality of science and the rigor with which scientists perform and publish their work had led me to imagine that lack of reproducibility of peer-reviewed, published experiments was becoming pretty rare. When such work cannot be reproduced, or is actually the result of fraudulent activities, this deficiency is quickly revealed by the scientific community.
But a series of articles suggests that scientific research, published in the most prestigious journals, is often not reproducible, and this can be undiscovered for decades. A recent issue of the journal Science discusses a professor who fabricated data for 20 years and 100 publications, before the data fabrication was discovered.2
Reportedly, it is also common that drug manufacturers interested in finding potential new products for diseases cannot reproduce the tantalizing results published by university laboratories, causing the pharmaceutical companies to waste time and many millions of dollars in unproductive research.3
Scientists at the company Bayer were apparently only able to replicate fully 21% of 67 studies published in academic journals. They partially replicated 12%, and could not replicate 64%. A professor at Stanford University reported that he could not reproduce16 out of 18 papers published in one of the most prestigious journals in the genetics field.
This lack of repeatability of lab work by other scientists is discouraging. I wish I could say that we in ophthalmology never have this issue. But as a young ophthalmologist I remember being perplexed by reading the results and listening to the talks of a prominent surgeon whose results with a refractive surgical procedure were far beyond those achieved by other surgeons.
"Why do his patients see so much better than everyone else's?" I asked a senior colleague. He responded with a look that said "how naïve can a person be?" and told me that this surgeon with the great results reported a very low rate of a certain complication.
"Every single one of his patients with that complication must come to see me, because I am seeing them all the time in my clinic," he said.
By Peter J. McDonnell, MD director of the Wilmer Eye Institute, Johns Hopkins University School of Medicine, Baltimore, and chief medical editor of Ophthalmology Times.
He can be reached at 727 Maumenee Building 600 N. Wolfe St. Baltimore, MD 21287-9278 Phone: 443/287-1511 Fax: 443/287-1514
2. Crocker J, Cooper ML. Addressing scientific fraud. Science. 2011;333:1182.
3. Naik G. Scientists' elusive goal: Reproducing study results. Wall Street Journal. Dec. 2, 2011.