Very few BMJ original research articles are cited by 1600 other publications, so it’s dispiriting to discover that the message of one of our citation classics may have been wrong.

A meta-analysis by Neils Skakkebaek and colleagues reported in 1992 that sperm counts had halved over the preceding 50 years. It became big news, no doubt chiming with contemporary anxieties about “the crisis of masculinity” and the ill effects of endocrine disrupters (eg the Everglades overrun by hermaphrodite ‘gators’). Who knows:  the resulting concern about falling sperm counts might have contributed to the shift from tighty-whities to boxer shorts.

Now the same research team have come up with a much stronger study design. Semen samples from 5000 18 year old Danish military recruits were analysed over a 15 year period.  Their reassuring findings:  whichever way you tilt your head at the graph of the average sperm counts against year it all seems remarkably stable. So maybe falling sperm counts is one less thing we should be frightened of.

To allow you to make up your own mind, I would normally be directing you to the research paper in a peer reviewed journal, (and logically the authors might have considered the BMJ) but the findings appear in a commentary in Epidemiology. 

According to Time magazine the process went something like this: “ The data wasn’t released by Dr. Niels Erik Skakkebaek of the University of Copenhagen, who initiated the study; he hasn’t commented yet on the results, pending publication in a scientific journal. Instead, the research group’s current leader, Niels Jorgensen, sent the data to the Danish Ministry of Health, which helped fund the study. The journal Epidemiology then got the data off the health ministry’s website and published it in a commentary.”

Whatever the editor’s justification for his actions it looks like sharp practice to me. In his accompanying editorial, Allen Wilcox asks who owns data generated with public funds, while admitting  that the researchers didn’t want to release their findings before they had at least 20 years of results. “The presentation of a few raw data on a Web site— or in a commentary—is hardly the preferred way to advance science,” he writes.  “But neither is it acceptable for valuable data to be held in storage.”

You could argue, with some justification,  that the bald graph reproduced in Epidemiology advances science not one jot. It’s hard to imagine what would have been lost if the researchers who’d been collecting the data published them in their own good time and in sufficient detail to allow proper interpretation.

Meanwhile, spare a thought for the subjects who volunteered “a specimen.” If I understand reference 6 to the Epidemiology commentary correctly, the 18 year olds also had their testicular volumes measured and completed a questionnaire on self rated health.

I can’t quite see military recruits in the UK or US readily agreeing to this level of intrusion in the interests of science, but then Danes are the happiest people on earth and its young men were presumably “cool” about it.

Tony Delamothe is deputy editor, BMJ.