But they’re too complicated!


That dirty open secret in biomedical research: bias gets built into the study design.

For decades, scientists have embarked on the long journey toward a medical breakthrough by first experimenting on laboratory animals. Mice or rats, pigs or dogs, they were usually male: Researchers avoided using female animals for fear that their reproductive cycles and hormone fluctuations would confound the results of delicately calibrated experiments.

“Delicately calibrated” seems to be used as a synonym for “not robust”. If your results are so finicky that they don’t hold true in translation from male to female rats, why would you expect them to hold up in translation from rats to people? There are detectable differences in male and female physiology that turn out to matter, so this business of ‘simplifying’ by focusing entirely on one sex means women’s medicine suffers.

There is good news. Now the NIH is cracking down and telling researchers that they must test mixed sexes, or no money for you. That’s an effective incentive.

It also turns out that the decision to ‘simplify’ by studying only male experimental animals is a bad one, borne of a bias that women are hypervariable because of their menstrual cycle — other studies find that male animals tend to be more variable.

Bias in mammalian test subjects was evident in eight of 10 scientific disciplines in an analysis of published research conducted by Irving Zucker, a professor of psychology and integrative biology at the University of California, Berkeley. The most lopsided was neuroscience, where single-sex studies of male animals outnumbered those of females by 5.5 to 1.

Contrary to the conventional wisdom in laboratories, there is far more variability among males than among females on a number of traits and behaviors, Dr. Zucker has found. Yet even when researchers study diseases that are more prevalent in women — anxiety, depression, thyroid disease and multiple sclerosis among them — they often rely on male animals, according to another analysis led by Dr. Zucker, who has written extensively on gender bias in scientific research.

At least I can say I’m safely innocent of this bias: all my work is on zebrafish embryos between fertilization and about 48 hours, and they don’t have sexes yet. I couldn’t sort out male and female embryos if I wanted to.

Comments

  1. David Marjanović says

    So I wondered if zebrafish have sex chromosomes.

    Quoth Wikipedia: “The sex of juveniles cannot be distinguished except by dissection, and sex determinants are not clearly understood.”

    I’ll take that as a “no”…?

  2. davidnangle says

    See, scientists don’t want to worry those pretty, little, very small, tiny heads.

  3. chris61 says

    It may get even worse with mice. According to a recent paper, pain responses in mice are influenced by the sex of the researcher.

  4. mikeyb says

    That is interesting that no studies have been done apparently to see which sex is more variable. One reason perhaps one sex is used in research vs both is that using 2 sexes is more complicated and costly. Say you need n=7 animals to do a specific experiment with males. Including females you now need n=14 and have to determine if there are sex difference correlative effects involved along with the effects you are looking for to begin with. Of course this makes sense, but it may double the cost of doing research, in already limited time and budgets involved. I’m not saying it is a good excuse, but it is a very real practical reason for doing research on one sex or another.

  5. says

    I can’t find the item to cite, but I read once how such attitudes among doctors and scientists occurred during the rise in HIV/AIDS rates in the 1980s. Doctors were only looking for symptoms that appeared in men, and arrogantly assumed women would have the same symptoms, or that women who had HIV/AIDS were engaging in the same sexual behaviours as men.

    It took years before they recognized there are differences. The consequence was that women with HIV had a much shorter life expectancy than men because the disease wasn’t being diagnosed early enough. This is not the item I recall reading, but it highlights the same issue:

    http://www.womenshealthresearch.org/site/PageServer?pagename=hs_healthfacts_hiv

  6. kalkin says

    Male mice tend to fight compared to females. Obviously, that can affect results, particularly if you are doing behavioral studies. That’s one reason you may see a female bias when using those animals.

  7. ibbica says

    One reason perhaps one sex is used in research vs both is that using 2 sexes is more complicated and costly.

    Sure, but of course that’s no excuse for nearly everyone to choose males.

    Male mice tend to fight compared to females.

    Which is a not-bad reason to expand our study subject pool out not only to include females, but other model species as well. If you want to study ssocial interactions between males in a situation where they won’t try to kill each other, you either need to work on giving them a much larger and complicated home setup, or use a different species. Some researchers seem to get hung up on their one specific model as the be-all and end-all, and refuse to think that any other might be better suited to answering the question at hand. And some seem to think that a question that can’t be answered using their model isn’t worth asking :/

  8. robertfoster says

    Apparently they’ve never seen a 50 or 60-something guy with low-T. Now that’s variable.

  9. Kevin Kehres says

    The biggest problem isn’t the sex of the rodents — it’s that we’re using rodent models that do not translate to people.

    A while back, I had a gig editing a series of articles on female sexual dysfunction on behalf of a pharma company developing a drug. The rat studies looked pretty good — female rats given the drug were much more receptive to mating.

    Problem was, when human females took it, they had symptoms of morning sickness. Oopsie.

    I’m currently working in another field entirely, where the animal models seriously and truly suck balls. The inside joke is that “everything works in rats”. There is not one teeny bit of usable data that comes from the rat models — except to find out if a compound is a (rat) teratogen or turns the rats into porpoises or something.

  10. kalkin says


    I think this was the subject of a recent article in Nature (?), regarding mouse models. Particularly and issue in using mouse models of ALS. The good thing about mice is the relative ease of creating mutant strains, the bad thing about mice is the relative ease of creating mutant strains…..

  11. ambassadorfromverdammt says

    I would have thought that the thalidomide debacle would have cured the powers that be (both private and regulatory) of not adequately testing drugs on females if the drug is to be administered to females. It looks like they took the wrongest possible lesson from it.

  12. Blattafrax says

    Females cost more. As far as laboratory mice are concerned – I make no comment on the truth of that in other species. Although I agree in principle with PZ, I’d suggest that restricting an experiment to either males or females will give less variation in experimental outcomes and mean that fewer animals would be needed for the experiment (since reaching statistical significance will be easier). Using fewer animals is a good thing.

    #12 All new drugs are tested in males and females for safety. The issue here is testing hypotheses experimentally.

    #10 It’s only a problem in that an imperfect system will inevitably produce failures. Since we can’t use people to test ideas, then we have to use the next best available. Rodents. Cell culture systems. Stem cells.
    Most drugs developed cannot be shown to work as expected in people – this is true. We wouldn’t be in even that sorry state if it were not for our animal models.

  13. busterggi says

    “Mice or rats, pigs or dogs, they were usually male: Researchers avoided using female animals for fear that their reproductive cycles and hormone fluctuations would confound the results of delicately calibrated experiments.”

    And that’s why I’ll never vote for a mouse, rat, pig or dog for POTUS.

  14. says

    Does anyone know of a place on the net, a forum or something, where someone that is reading this sort of literature can ask questions?

    I know enough to be dangerous and have been studying my own cognitive issues in some pretty deep detail and those issues have been getting intimately involved in systems that have to do with male-female differences and how they are established. I’m at the level where I can actually search the literature to come up with my own hypotheses (partially for fun, partially for my own satisfaction, partially because who knows I might find something) and I’m the sort of person that actually likes to eliminate my biases with extreme prejudice. Lying to myself is one of the few things I would call a “sin”.

    Any ideas?

  15. says

    But this is, to some extent, baked into our medical system from its founding.

    Else, why is the study of male reproduction given to urologists, and of female reproduction to gynaecologists? Because our system of medicine has been built on the concept that the male form is the default, and the female is a male with some extra bits. There are a number of serious diseases and disorders that present very differently in women, and we’re not quite sure how many women have died because they were carefully watching for symptoms that are only common in men, and missing the actual symptoms of their killers. We’re slowly beginning to see the change of this, with new differentials on stroke, heart failure, and several other things being developed and disseminated* in the last couple of decades.

    The surprising/sad part to me is that people are surprised to discover this bias in effect, when it’s pretty much what people like Greer and de Beauvoir were saying forty and fifty years ago. They should be put into the curriculum in high school. Maybe we can drop Hemingway to a specialist class.

    * LOL at the word we have for “passing along knowledge”, and its relation to the topic.

  16. The Very Reverend Battleaxe of Knowledge says

    ambassadorfromverdammt @ 12:

    I’m not sure that was the problem with thalidomide. Wasn’t it supposed to be a remedy for morning sickness? I think the problem was that in a lab setting, the thalidomide used was freshly prepared, and hadn’t had time to racemize: i.e. had some of it flip to the opposite stereoisomer, which is what caused the problem.

  17. The Very Reverend Battleaxe of Knowledge says

    And I used the word “problem” three times in three lines…I’m making another pot of coffee. (Sometimes I wish I spoke a language that didn’t have 12 words for each thing, and if you mean something you can say it as many times as you need to.)

  18. Anton Mates says

    Well, of course males are more variable than females. That’s why all the smartest, best and most awesome people are male, and should be in charge of society. Also, all the dumbest, most violent and useless people are male, and should be in jail. (They’re conveniently identifiable by the fact that they’re brown.)

    Thus the status quo is preserved, as it will always be if you interpret new scientific findings Correctly.

  19. says

    Interesting discussion. I would say that most of my research has used female outbred mice, but then I study infectious disease not drugs or behavior. In our system, males tend to fight more, and also seemed to have more severe disease. If we are asking questions related to sex differences, then we use mixed female and male. And for specific immunological questions, where we are breeding mice deficient in a certain immune cell receptor or something, we always use a mixed model. I would be slammed during review of my paper if I didn’t! I always wonder when I read these reports- how the hell did they get away with publishing such crap!?!

  20. Ichthyic says

    Give me cell-lines any day, even if they’re so genetically unstable that you get different numbers of chromosomes in cells in the same flask….

    sarcasm, right? it’s so hard to tell any more.

  21. Ichthyic says

    Does anyone know of a place on the net, a forum or something, where someone that is reading this sort of literature can ask questions?

    you mean, other than here?

    overall, my recommendation to people has always been to ask questions of the people doing the research themselves. Most researchers are in fact approachable, and if you ask them a question that really shows you actually have read their work, they tend to even be eager to answer your questions.

    use email.

    general science forums tend to get bogged down with personal affairs inevitably, and i find that they actually often aren’t the best places to seek specific answers to specific questions.

  22. Ichthyic says

    I’d suggest that restricting an experiment to either males or females will give less variation in experimental outcomes and mean that fewer animals would be needed for the experiment

    uh, it’s that very LACK of variability that is causing problems with clinical drug trials.

    you WANT variablity, and large sample sizes.

  23. Ichthyic says

    There is not one teeny bit of usable data that comes from the rat models

    if you are speaking specifically to your particular experimental regime, I’ll accept that.

    as a general statement, it “truly sucks balls”.

    In fact, people are overgeneralizing what the NIH is saying here.

    do you REALLY think nobody has ever done studies on variability between sexes in rodents in medical testing regimes?

    I challenge you to spend a few minutes on PubMed and decide for yourself.

  24. Tigger_the_Wing, Back home =^_^= says

    Well, I can’t say I’m surprised.

    Ask any woman, or female-coded person, what happens when they come into contact with the medical profession with something regarded as primarily a ‘male’ problem, or with ‘atypical’* symptoms, i.e. not ones usually seen in middle-aged white men.

    Diagnosis can take decades. Women die of heart disorders in greater numbers than men because the doctors

    (a) do not recognise the symptoms; e.g. when in hospital for an overdue diagnosis of my own disorders, I met a woman recovering from quadruple-bypass surgery. She had driven herself 200km to the hospital, in terrible pain, because the GP on duty that night had refused to come out ‘for a case of indigestion’. Her stubbornness had saved her life. My own unstable angina was labelled indigestion for ten years.

    (b) have a picture in their minds of a ‘normal’ sufferer of a particular disorder, and ‘woman-shaped’, especially young, doesn’t fit the template. I don’t know what the statistics are, but every woman I’ve known or interacted with who has atrial fibrillation was diagnosed with ‘panic attacks’, delaying the actual diagnosis by years (in my case, 32 years).

    The Heart Foundation in Australia has been mounting a campaign in recent years, with fridge magnets, leaflets and wallet cards, with a list of the actual symptoms of angina/heart attack, as opposed to the ‘Hollywood Model’ that most people are familiar with (overweight, middle-aged white man, collapsing dramatically whilst clutching his chest and screaming “my pills!!!”).

    (*that’s right, the majority of the human race is labelled ‘atypical’)

  25. Blattafrax says

    #27
    If your mice and money supply are unlimited and you want to show your treatment is generally applicable to all mice, then I’d agree with you.

    But the first two don’t happen in real life and in the pharma industry (since you mention clinical trials) it’s the huge difference between humans and rodents that drives the failure of new drugs**, not the relatively tiny variation in mice strains. There’s certainly no reason to believe that a drug, universally successful across every strain, age, weight and sex of mouse will be any more likely to succeed in people than one tested only in a well designed in-bred model of human disease. Well, maybe there is a marginal benefit, but by the time you’ve worked that out, 8 other companies will be in phase 3 trials.

    **Where the failure is due to the non-predictive nature of animal models rather than e.g. safety or economics.

  26. Blattafrax says

    #27 #30
    Obviously, researchers tend not to see one positive result in a single model and then plonk the drug into phase 1 trials. It’s encouraged to get broad evidence for efficacy – including in more than one sex – but not all in the same experiment. That way lies madness.

  27. Kevin Kehres says

    @28 — In the disease I’m specifically talking about, yes. I did not make the hasty generalization. You lept to that conclusion.

    But it annoys me no end, because researchers keep using the same model over and over and over again. They basically are poisoning rats to give them an analog of the disease in question. And when they intervene — VOILA! — that intervention works. In rats. And then on to people and … nope … nada … zip … oops, a case of porpoise … makes it worse … and on and on. The model sucks. It’s worse than just throwing darts.

    In that particular disease entity with a very very flawed model.

    And while I’m sure there are dozens if not hundreds of other similarly flawed animal models, I did not denigrate animal testing in general.

  28. palefury says

    Given that I work on breast, ovarian and uterine cancers, I am kind of stuck with using females.

  29. chrislawson says

    CatieCat@17:

    While there are many sexist elements in medicine, that particular example isn’t really one of them. Urologists specialise in the urological system, which both men and women have. So while urologists deal with prostate disease, which is obviously male-only, they also manage problems of the bladder, ureter, and kidneys in people of all sexes.

  30. says

    @Ichthyic 26

    What about possible bias from the researcher that I might be trying to contact? I know there is at least one paper that discusses “masculinization of behavior” among women with Tourette’s and that would definitely be one to get some other opinions on. There are some interesting possible links between andrenarche, neuroactive steroids, and a bunch of other little things with hypothalamic connections that I’m kind of excited about, which is a pretty good warning sign about making sure I don’t see things that are not there.

    I suppose I could always ask what people think on the Thunderdome or something.

  31. says

    . There are a number of serious diseases and disorders that present very differently in women, and we’re not quite sure how many women have died because they were carefully watching for symptoms that are only common in men, and missing the actual symptoms of their killers.

    Well, in German they even started talking about an “Eva Infarkt” for heart attacks in women. Unbefuckinglievable. They build “women are a weird exception to the disease of heart attacks” into the very word. Because men get heart attacks, and heart attacks have symptoms X, Y and Z. Women don’t get heart attacks. They get a cute “Eva Infarkt”. It’s not like “People get heart attacks and depending on whether you’re male or female your symptoms are more likely to be this or that”

  32. cubist says

    Given the highly nontrivial hormonal cycles associated with menstruation, I can sympathize with any researchers who entertain the tentative hypothesis that women really do have a lot worse variance in relevant matters than men do. Can’t really condemn it as sexism until the researchers don’t bother to test said tentative hypothesis, perhaps because “well, heck, isn’t it obvious?”

  33. Maureen Brian says

    cubist @ 37,

    Have there actually been studies which indicate that the degree of variability within women is greater than that in men? And has this been confirmed several times in different studies over decades?

    Has anyone compared the effects of these “woman” hormones with the great variability and the possibly cyclical nature of testosterone production in both men and women? What about circadian rhythms and their impact on the efficacy of particular medicines?

    Until you can show us the links to real studies then this remains one of the myths which underpins a whole load of medical assumptions and failures to grasp things, sometimes with fatal effect as illustrated.

    It is thus sexist and discriminatory by definition – until you prove it has a scientific base.

  34. says

    cubist
    It is sexism because the male gets seen as the default and the female as an icky variance. Results from such male-only studies then get published not as studies about male mice, rats, whatever, but as studies about mice and rats, rendering half the species invisible and making them an unusual variety. As mentioned by others, such things have serious consequences in medicine where women get underdiagnosed, misdiagnosed and wrongly medicated because they don’t fit the “universal” i.e. male model.

  35. =8)-DX says

    @cubist #37

    Um, excepting a tentative hypothesis that hasn’t been tested, but exists due tosexist prejudice is.. well sexist. Unless you have specific evidence for a particular biological process being influenced by the female hormonal cycle (if your study includes basal temperature measurements I guess), in comparison to male hormonal levels (and their changes), it’s best to assume none is there.

    If peopel were really sincere about controlling for menstrual cycle, surely questionnaires would include “how long since your last onset of menstration” for females. Oh, and “how long since your last erection” for males. Hmmm.

  36. David Marjanović says

    It reminds me of this:
    http://people.cs.uchicago.edu/~dinoj/scilies.html
    Phrases in research papers and what they really mean.

    One of them is wrong:

    “Of great theoretical and practical importance.”
    Interesting to me.

    Hah. Nobody gives a shit what I find interesting. It means “I hope this is interesting to journal editors and funding agencies”!

    * LOL at the word we have for “passing along knowledge”, and its relation to the topic.

    That’s a coincidence, though: the image here is that you strew your knowledge around like sowing a field. Semen meant literal seeds long before it was metaphorically extended to man-slime, which itself happened long before the metaphor was translated into modern languages.

    Given that I work on breast, ovarian and uterine cancers, I am kind of stuck with using females.

    Breast cancer does occur in men. And when it happens, it’s usually overlooked for a long time!

    What about possible bias from the researcher that I might be trying to contact?

    :-| Contact several. Contact the whole field if you like.