The Lancet done screwed up

Chelsea Polis and Kathryn Curtis wrote a paper that asked whether hormonal contraceptives affected your likelihood of being infected with HIV, Use of hormonal contraceptives and HIV acquisition in women: a systematic review of the epidemiological evidence. Here's the abstract:

Whether or not the use of hormonal contraception affects risk of HIV acquisition is an important question for public health. We did a systematic review, searching PubMed and Embase, aiming to explore the possibility of an association between various forms of hormonal contraception and risk of HIV acquisition. We identified 20 relevant prospective studies, eight of which met our minimum quality criteria. Of these eight, all reported findings for progestin-only injectables, and seven also reported findings for oral contraceptive pills. Most of the studies that assessed the use of oral contraceptive pills showed no significant association with HIV acquisition. None of the three studies that assessed the use of injectable norethisterone enanthate showed a significant association with HIV acquisition. Studies that assessed the use of depot-medroxyprogesterone acetate (DMPA) or non-specified injectable contraceptives had heterogeneous methods and mixed results, with some investigators noting a 1·5—2·2 times increased risk of HIV acquisition, and others reporting no association. Thus, some, but not all, observational data raise concern about a potential association between use of DMPA and risk of HIV acquisition. More definitive evidence for the existence and size of any potential effect could inform appropriate counselling and policy responses in countries with varied profiles of HIV risk, maternal mortality, and access to contraceptive services.

In short, hormonal contraceptives don't affect your chances of getting AIDS, with the possible exception of DMPA (better known as Depo-Provera), which a few studies with different methods found to elevate the risk. So, basically, it's saying that there's not a problem with contraception endangering women in this regard, but that inconsistent results with Depo-Provera warrant further investigation (later, they would publish an update (pdf) that suggests women ought to be warned about the uncertainty of this side effect of Depo-Provera). It seems reasonable. So they sent it off to The Lancet for review and publication, and that's where the mess began.

There are actually two problems here: The Lancet's sloppy review policies and their failure to adequately respond to a screw-up, and the abuse of the data in the paper by a group of religious ideologues.

1. The paper was sent out to three reviewers, only one of whom bounced it back with an unprofessional and strongly negative review. It happens. The authors responded with a thorough rewrite of the paper, which is a cautious over-reaction, but I think they were trying their best to be meticulous. Then it turns out that the negative reviewer is a peculiar kind of AIDS-denialist crank with absolutely no professional affiliation -- he's a guy with a degree crusading for his pet theory, that HIV is not sexually transmitted at all, but is spread by medical injections.

   So one question is why this guy is even on The Lancet's list of reviewers. They've removed him now, which is one positive response, but it's unfortunate that he was given the opportunity to play the poison pill on papers that didn't share his weird hobby horse. At the very least, his bad reviews ought to have flagged him to the editors as someone who ought to be checked out.

   Overall, though, I think The Lancet's response was muted but appropriate. It should have acknowledged the loss of time and effort by researchers that was lost because they let a wackaloon have his hand on the brake of publication.

2. The second problem is one The Lancet has little control over: it's that anti-science kooks tend to be unscrupulous and unethical. That third reviewer not only tried to squelch the paper, but then published the reviewer's copy on an anti-contraception web site, and somehow distorted it's message to be that Depo-Provera was inarguably deadly, and that there was a government cover-up to conceal it.

   DEPO-PROVERA -- DEADLY REPRODUCTIVE VIOLENCE AGAINST WOMEN, they announced. It has been cited on LifeSite News, the mouthpiece for a conservative Christian "pro-life" group that opposes contraception and abortion.

   The second classified document obtained by Fosu was authored by Chelsea Polis of USAID and Kathryn Curtis with CDC. It heavily influenced the WHO review staff to determine there was “insufficient reason” to withdraw Depo Provera and enforce mandatory warnings. Fosu claims the Polis and Curtis analysis “water[ed] down available evidence” by ignoring key research.

   Oh, really? It was published in The Lancet. That doesn't sound very "classified" to me. It was stolen in violation of copyright and reviewer ethics before publication -- I don't quite see how they can claim an unpublished paper that was just working its way through the review process could have "heavily influenced" the World Health Organization, and further, its conclusion is a moderate call for caution and further research.

Polis is rightly annoyed at all this. The Lancet is not responsible for the theft and ignorant misuse of a scientific paper by a gang of fanatics, but they do have a responsibility to do a better job screening out said fanatics, and they ought to do a better job of recognizing the concerns of their contributors.

In sum, Lancet Infectious Diseases failed to perform due diligence in selecting an unbiased and professional peer reviewer, allowed unscientific and offensive comments to move forward in the review process, failed to protect my work from being leaked by an unethical reviewer who the journal was unable to follow up on, dragged their feet in taking my case formally to COPE, failed to fully describe the situation to COPE, and shrugged their shoulders at the violation of copyright given "reputational concerns". They've never apologized for this situation, or shown any interest in implementing more meaningful solutions to prevent this from happening in the future (e.g., telling reviewers that if they breach confidentiality that their names will be publicly shared so that other scientists can avoid being reviewed by them, or putting additional measures in place when asking unaffiliated individuals to serve as peer reviewers). Why not?

I agree that naming and shaming ought to be the very least action taken, so it's a little odd that she didn't come right out and name David Gisselquist as the unethical reviewer (although the links in her article do make it very easy to figure out who it was).

One more annoying little twist, that I've heard a thousand times before…

Recently, the journal did a Tweetchat called #AskLancet - and I had some follow up questions on my situation. How this conversation unfolded is best encapsulated in a Storify; in short, my questions were not responded to, until they were retweeted by a man who shared my concern. Click here to read the Storify.