x million rats and mice are used in US labs every year —- so . . . . ?

published my estimate of over 110 million mice annually in US labs and got some criticism of my methods and the resulting estimate. Fair enough. So let’s suppose there are “only” some 30 or 50 million per year. What should we do with that info? 

Two opposite conclusions from seeing these large numbers:

This is too many animals to leave out of the federal protections of the US Animal Welfare Act (AWA)

This is too many animals for the USDA ever to be able to extend AWA protections to

As I said in my article, the truly large number of mice in labs should make us ask whether they should become “animals” in the US Animal Welfare Act. I say Yes; others say No.

Knowing my concern that Congress has said that mice are not “animals” in the eyes of the Animal Welfare Act, told David Grimm two more-or-less opposite reasons for keeping things as they are:

  1. Speaking of Research “disagrees with the idea that the animals are at risk of being mistreated because they are not covered by the AWA… that all labs that receive federal funding must answer to institutional animal care and use committees, which follow the three Rs and oversee the welfare of all lab animals. Those that don’t receive federal funds typically submit to AAALAC inspections,… It is unlikely that there are large number of facilities that are uncovered by any regulation.”
  • Except, according to the National Association for Biomedical Research, it would do something. Maybe that something would not improve these animals’ welfare, but the government would not “have the time or money to track so many animals, and that doing so would drain vital resources. Now is not the time to be seeking additional restrictions on biomedical research or endeavoring to make it more difficult and more expensive.” 
  1. To the first point, I agree (anecdotal experience and informal communication) that accreditation site visits are really quite rigorous and helpful. I don’t see that accreditation alone, without AWA inspections, is sufficient (see below).  As for their claim that it is unlikely that there are large number of facilities that are uncovered by any regulation — first, nobody knows how many [ SoR estimates 10 – 25 million, which is highly implausible, so I’m not sure I buy their other estimates]. Second, perhaps we should worry that there are labs with no regulatory oversight and that are not accredited [I certainly know of several] and why should we be confident in the quality of their animal care?
  1. On the second point, I agree that the USDA’s resources would be stretched if they upped their game to covering the other 99% of mammals in labs. From the point of view of the inspected, again, I’m not convinced. Places that have NIH regulations to follow and/or are already accredited are already doing committee reviews and annual reports, which is what AWA coverage would require. Places that aren’t, should be, and AWA coverage would make that happen.   

So, are current welfare protections good enough for mice?

I say “no,” for four main reasons:

  • There is close to zero public transparency about mouse use
  • There are many labs (No One Knows how many, because only the AWA that excludes them requires publicly reporting animal numbers) that are not accredited and are not under jurisdiction of NIH animal welfare laws and 
  • It is impossible to monitor trends in animal use, including in painful experiments, without sound publicly-available statistics
  • There’s nothing like an unannounced USDA inspection to keep people on their toes.

What to do? 

Well, it will take an act of Congress, quite literally, for mice to become Animal Welfare Act animals. That will not happen. American politics is a bit of a mess now, eh? Still there should be a system in which these animals at least occasionally receive the kind of spot-checks from government veterinary animal welfare inspectors that hamsters get.

The NIH could certainly commit to greater transparency, for that fraction (half? fewer? more?) of lab mice under NIH jurisdiction. They could add up the numbers of animals reported to them annually and post that, as USDA does for Animal Welfare Act-covered animals. They could (with resources they don’t presently have) do more site visits and make their findings public (you can get them, but only via FOIA requests).

Meanwhile, many good people are working on ways to make laboratory life better for our most important and numerous animals, if only their efforts can trickle up to administrators who know their institution can escape scrutiny of much of their animal research.  

Over 100 Million Mice per Year in US labs? No One Knows

No One Knows.

January 2021, I published my article estimating how many mice and rats per year we use in United States labs every year. The reality: No One Knows. For two reasons:

  1. The patchwork of US regulations and accreditations means that no one is actually reporting mouse and rat use in a publicly transparent way, given that these animals are excluded from the Animal Welfare Act definition of “animals.”
  2. No one has a consistent definition of “used.”  Every mouse that becomes a data point in a paper or a regulatory test has parents, siblings, and colony-mates who do not get that far. If you include breeders and culled animals, you find a larger number “used” than if you only count those who become data points in published papers. 

My approach relied on getting annual reports to the accreditation agency, AAALAC. They get annual reports from labs they accredit, but never release such confidential info. Some labs were willing to give me their info as reported to AAALAC. Some gave me the info via their state’s Public Records act.  For the 16 labs whose data I received, mostly well-funded universities, I then compared annual rat/mouse use to annual Animal Welfare Act “animal” use on annual reports on the USDA website.  When I put those together, I found that at those 16 large campuses whose data I got, Rats/Mice were more about 99.3% of the mammals. Extrapolating from an annual report in 2018 of over 780,000 AWA “animals,” I estimated over 110 million mice and rats, which is a good three times what others – research advocacy groups and animal advocacy groups – have been saying. 

I did this project because I’d pretty much guessed pretty much this number in my 2004 book, What Animals Want, and always figured someone would do some actual data-gathering to check on this. Almost no one did, and the one group that did used European stats to get at US numbers, and I do believe they ended up under-estimating US mice and rats in the process. So, I did it myself. 

As David Grimm found writing about my estimate in Science, there was some pushback. Some pretty vehement pushback, as these things go. 

Here’s my review of the criticisms of my paper, and some response

First, my own criticisms of my project (which I do discuss in my paper): 

  1. This is hampered by having no standard definition of “used,” even for labs that do report to NIH or to AAALAC. My guess — but remember: no one knows — if you include breeders and culled animals (including the ones at commercial breeding labs) I imagine there’s a 10-fold increase compared to just counting those animals enrolled in an experiment or euthanized for their tissues. But no one knows.
  2. Even if we had an agreed-upon definition of “used,” NIH and AAALAC have not given a uniform definition of “how many” or “average daily census.” At some places, especially if they count cagesto bill to grants, there’s no standard industry conversion factors for converting numbers of cages on a given day to numbers of mice on that day, and even less, for going from average daily number of cages to total annual numbers of animals used. I found a wide range of conversion factors in my project, but the truth is: no one knows.
  3. When there is such an imbalance (my data found a 99.3 to 0.7 difference, though I actually think my 99.3% is on the low side), any small difference in the estimated balance of mice to larger animals is magnified. So, while 99.2% seems not far off from 99.6%, there’s a large difference between 0.8% and 0.4%   So if all you can really get are counts of the small numbers of animals reported to the USDA, that 0.4, 0.8 or other really could make a big difference in your estimated totals. 
  4. The big challenge with my project: if all I can get are mouse counts from a handful of institutions, mostly from those subject to states’ sunshine laws, is that sample representative enough of all US labs in academia and industry?
  5. I also only looked at one year’s data. That may be more faith than warranted in our counting and reporting systems. 

I tried to lay out my methods in enough detail that even if I’m over (or under-estimating) the project might be reproducible enough to show trends over the years. 

Others’ critiques:

Nadia Jackson at Jackson Labs thinks my university sample would not be representative, and I’m wrong to work with that assumption. She’d rather make assumptions based on commercial info from Charles River labs, which seems to be the world’s largest supplier of lab rodents. If you make certain assumptions about what percent of their revenue is mouse and rat sales, and what the average price of the mice they sell is, and an assumption that people across the country buy about three-fourths of their mice and only breed one fourth in-house AND you assume that the numbers of unsold (breeders; culled animals) at vendors selling 15 million mice per year are negligible, the total number is only about 20 million.  I do think that’s a lot of assumptions to come up with that number, and I see no reason to think her assumptions are any more valid than mine. Because as we know, with no clear windows into mouse labs, No One Knows.

Speaking of Research does not like my look over to Europe (or up to Canada or over to Australia) for the idea that the US should transparently report on numbers of animals used. Guilty as charged, but they do also critique my estimate. They point out (correctly) that my data are skewed toward large universities, or rather the subset of those that I was able to get data from (and yes: thanks to those of you who told me your mouse numbers. I promised you anonymity, and I’m honoring that, as well as to those who refused my request). True. They point out that 7 large private companies use more USDA-covered animals (e.g., hamsters, dogs, monkeys, some pigs) than the 50 largest NIH-funded universities. They don’t say how I could get at those places’ mouse and rat numbers to let me make a nation-wide extrapolation to total mouse numbers. They may be write that my sampling is not representative enough. Maybe if someone uses my methods, they can go further down the level of funding hierarchy and see if lesser-funded universities do indeed have different mouse rat ratios. That would be great. Until then, I still think I’ve got the most reproducible method.

At the National Association for Biomedical Research and Foundation for Biomedical Research, I don’t see any dispute about my estimate, just their concerns about what people would do with my estimate, which I’ll write about separately. Their estimate is that mice and rats are 95% of lab animals (notice we’re all avoiding mentioning zebra fish, where lack of transparency or standardized counting are even greater, so No One Knows if they outnumber mice. They may). That would mean that for the year I covered, there were about 15 – 16 million mice and rats. 

Analgesia for Animals: My Lottery Theory of Multimodal Pain Treatment


My Lottery Theory of Multimodal Animal Pain Management:

Good lab animal welfare requires effective treatment of any pain scientists cause. It is way too easy to pick a drug from the list, pick a dose, and never really know if the drugs is actually helping the animal feel better. One variation on this is to use three drugs, or multimodal analgesia treatment — going Beyond Buprenorphine (the most common single-agent treatment) to maximize pain relief

It’s been years now that I’ve been advocating combining different classes of pain medicines to treat research animal pain. I haven’t always prevailed; most notably, I consistently failed to get the vets or the animal committee on my recent job to enforce this standard on one particularly resistant monkey-user scientist.

I think a lot about evidence-based ethical treatment of animals when the evidence just isn’t there. When it comes to fine-tuning pain treatments for animals, our evidence is so very sparse. Whether it’s monkeys, mice, fish or others, how do we evaluate the level of pain they’re experiencing? How do we evaluate if pain medicines truly make them feel better? How then can we know if combining analgesics is even better (or worse) than using single drugs? What about side-effects of pain drugs? What about how either drugs or untreated pain might affect the experiments the animals are on?

We totally lack the most important info for even our most common drugs, and that is, what if our animal patients could self-medicate or somehow tell us they need another dose, or a stronger dose, of their pain meds?

Frankly, we don’t have the info we need to pronounce on the best pain management — other than the obvious, which is to first, cause no pain. And so IACUCs end up making an ethical decision, in what I call the ethics-of-uncertainty. Do they err on the side of caution, and require more aggressive pain treatment than some scientists want to provide? Or do they privilege concerns about how drugs will affect research data?

With my recommended 3-drug multimodal regimen, we combine an opioid (usually buprenorphine; an MD would rely more on morphine or codeine) with a Non-steroidal anti-inflammatory NSAID (for animals, we use carprofen or meloxicam; an MD might rely on ibuprofen) and also, a local “block” at the surgical incision (bupivacaine or lidocaine; your dentist uses a lot of lidocaine). Our doses for these drugs are not wild guesses, but neither are they very solid science either. We don’t know exactly how much to use, we don’t know when to re-dose, we don’t give mice or monkeys their own medicine chests, and we don’t have the best skills at round-the-clock pain evaluations.

Hence, the Lottery Theory: use all three and hope you’re getting at least one of them mostly right. While I’m watching the science develop and hoping we’ll get good answers on just what pain management, my short term approach is that we should buy three tickets in the analgesics lottery, and hope at least one of them is a winner. And yes, let’s scale back on doing painful things to animals in the first place

A new mouse painkiller in your medicine chest

This is what vets in animal laboratories need: a painkiller for mice (and other creatures) that is

— Fully effective for all strains, ages and sexes

— Completely safe with no unpleasant side effects for the mouse

Will not interfere with the experiments – for neurobiologists, no drowsiness, disruption of Learning or other effects on brain and nerve function; for cancer biologists, no effects on immune function; for immunologists, no effects on inflammation or immunity.

Long-acting, since you might get a monkey user to come in at night to re-dose the painkillers, but you will not get many mouse and rat users to do that

We do not have that, and so the search goes on.

Years ago, I did this wrong (I’ve written about that here): Meloxicam was a hot new dog-cat non-steroidal anti-inflammatory drug (NSAID), that might have advantages over Carprofen that we were current using, if only because the oral version is really tasty. Mouse doses of carprofen are about the same as for dogs and cats, so I started prescribing Meloxicam for mice at the same dose (around 0.2 to 0.3 mg/kg) that I’d use for dogs or cats. Two years and many under-treated mice later, the good vets in Newcastle did actual studies, and my mouse dose was about 3% of what would have actually helped those deserving animals.

And as we look for the magic long-acting safe effective non-interfering NSAID, we’re also hoping for an opioid that fits that bill.  Opioids are drugs like morphine, fentanyl, codeine and hydrocodone that work on different pathways than NSAIDs, and have less effect on the immune system. The best we’ve had to date is Sustained-Release Buprenorphine (SR-Bup). It’s not perfect and still in need of many more studies to best establish the right doses and dosing frequency, but it is now well-established in many laboratories. Scientists should resist quick changes in the pain medicines they use, as there can be subtle differences in experimental outcomes when they change things around.

SR-Bup has a big problem limiting its use, a regulatory rather than a biological problem. It’s classified as a compounded drug, not as an FDA-labeled medicine, which requires a vet’s prescription. By contrast, FDA-labeled drugs can be purchased by research facilities without relying on an individual vet’s narcotics license. For some vets, that’s a personal legal risk they don’t want to take, and I can see their point.

Enter Simbadol, a newer long-acting version of buprenorphine FDA-labeled for cats. It doesn’t require a vet’s narcotics license (though it is still an off-label use to use it for mice; not many drug companies are investing in the FDA labeling process to make mouse drugs).  NOTE: I have no financial stake in either version of buprenorphine, though I did publish a paper on SR-Bup’s use in mice.    

So, let’s all jump to Simbadol for our mice, shall we? I say, not so fast. Responsible veterinary care means evidence-based pain management. What we know so far is that it seems to work for cats, at a dose and dosing frequency determined in lab and in feline clinical trials. Will it work in mice, help them feel better and not worse (i.e., effective with minimal side effects)? How long can it last before it needs re-dosing? Will it change experimental outcomes compared to whatever painkillers the lab is currently using?

Well, we don’t know. I find two articles on-line (here and here), both very small studies, not independently replicated, and in rats, not mice. Neither found compelling evidence of effective pain management, and then, for only about an hour, not the 48 – 72 hour efficacy we are hoping to see. It’s possible that the assay they used (time until the rats move their foot from an obnoxiously hot light shone on the sole) is not the best approximation of surgical pain, so it’s possible the drug performs better in clinical use than in the lab. Possible, but we do not know. We know that very few drugs use the same amount and frequency in mice and cats, given their vastly different metabolic rates and possible differences in opioid processing.  

BOTTOM LINE #1: it would be poor practice to jump on Simbadol for treating mouse post-surgical pain at this point, without a lot more studies showing that it works, and for a long enough period of time

And BOTTOM LINE #2: we still need to know more about the other Buprenorphine formulation, especially as a sole agent without concurrent NSAID analgesics, and choice of painkillers is therefore much less important than finding non-surgical ways of conducting an experiment.

Pain management details – full reporting for better science and better welfare

I’ve blogged on this before a few years back, here and here. I have published a literature review with Jamie Austin on the topic, and I’m concerned that scientists do not describe how they manage pain in their laboratory animals. In a world of on-line publishing without page limits, many journals allow thorough Methods descriptions in which scientists can hit more of the guidance in the ARRIVE guidelines for publishing animal experiments. For projects that include surgery on animals, that means describing all use (doses, frequency, duration of treatments) of anesthetics and analgesics, methods of pain evaluation, and an important poiint that none of some 800 papers I reviewed included: a clear statement that pain medicines were withheld (and why).

When scientists DO include pain management information they accomplish several good things:

They allow other scientists more information for a critical reading of the reported findings

They allow all readers the evidence that they used animals as humanely as possible

They model better animal treatment for others building on their work

When scientists do NOT include this information, they leave other scientists to think that pain medicines are optional, or worse yet, inappropriate for a particular set of experiments. Even in 2019, too many scientists in my opinion resist full use of analgesic pain medicines for their animals, and too many do that because they think others in their field will actually fault their work for using painkillers — patience: I will write about that, but it’s in some of my published writing here and here.

Mouse & Monkey Vet My Animal Welfare blog

In this blog, I will take you behind the scenes in animal research. My goal is to show you the animals behind news stories about medical advances. There is a harm-benefit balance in how scientists use animals, and only with detail and transparency can you evaluate for yourself the harms animals experience and the benefits scientists seek. Some of this is wonky and technical, and some is graphic (I will warn you). Some of it only tangentially touches on laboratory animals, as our impact on animal welfare reaches everywhere there are animals.