Home Office statistics: numbers, words, and euphemisms

The Home Office has now published its statistical report on the animal research done in Great Britain (i.e. omitting Northern Ireland) during 2015. It shows that 4.14 million ‘procedures’ were completed last year. This is the largest number ever recorded under the 1986 Act, and tends to confirm that the promising drop in the numbers during 2014 (3.87 million) was the result of under-reporting in that year, rather than a sudden change of direction. The new system had just been introduced, whereby the research projects are counted when they finish rather than when they begin, and not everyone seems to have understood it. So the Home Office advises that the new figures should be compared with 2013 rather than 2014 (for VERO’s comment on the 2014 figures, see http://www.vero.org.uk/events.asp.). In that case, there has been a slight increase of 1% or 21 thousand in these ‘procedures’. This in turn means that the real numbers have been rising in every year since 2001, except 2009, which came after a notable jump the year before. During this whole period, the numbers have increased by about 58%.

This new Home Office report makes an exhaustive summary of every countable aspect of the nation’s work as vivisector in 2015. Its own two-page précis can be found at https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/538556/scientific-procedures-living-animals-2015snr.pdf. There are other useful and more critical summaries to be found on the web-sites of the Fund for the Replacement of Animals in Medical Experiments and Cruelty Free International. These notice, for instance, the rise in numbers of primates used in research (from 3,220 to 3,600), and the continuing use of dogs in toxicology studies, one of the most unpleasant areas of research. There’s also a review on the web-site of Understanding Animal Research, which is the promotional arm of the animal research industry. At the end of it the Chief Executive, Dr Wendy Jarrett, is quoted as saying “today’s statistics will help people to find out more about the reality of animal research in the 21st century.”

Yes, on the face of it the statistics ought to help in that way, but I doubt that they will help much. Quite apart from the varying interpretations which statistics notoriously allow, they address a part of the mind (the numerate) which is completely unrelated to the part where ethics or empathy live. What can one feel about this great torrent of numbers? It’s a crowd scene with no foreground. Every now and then, a detail will catch the dazzled attention. For instance, under the category ‘regulatory testing’ (p.49), the astonishing fact emerges that the LD50 and LC50 tests (= the Lethal Dose or Lethal Concentration that kills 50% of the test animals) are still in use. These true products of the mind as computer, giving a specious accuracy to toxicology tests at the cost of human decency, accounted for 8898 animals in 2015 (mice, rats, and fish).

Nearby, now that one’s eye is adjusted to such detail, it seems that something very like the Draize test (listed as “eye irritation/corrosion”) also survives: 173 rabbits went that way. But what: only 173? In most of the categories, that number would simply have disappeared in the ‘rounding down’ of untidy decimals (see User Guide to Annual Statistics, pp.9-10). On the other hand, you’d certainly hate to see the test done to a rabbit you knew, and you’d be quite properly liable to prosecution for cruelty if you did it yourself. And by the way, that’s a useful reminder that the Home Office is wrong to define the Animals (Scientific Procedures) Act in its preamble as “an animal protection measure” (p.5): the Act is also, and much more successfully, an animal-user’s protection measure.

Anyway, such details as the ones mentioned are generally invisible in the glare of the huge numbers. The whole dazzling parade of facts, so competently put together by the Home Office’s statisticians, is therefore a kind of euphemism, tending as much to hide as to show the “reality of animal research in the 21st century”.

A rather more informative source, and a necessary complement to the Annual Statistics, are the ‘non-technical summaries’ of proposed research which the Home Office also publishes (at https://www.gov.uk/government/policies/animal-research-and-testing). There you can see the research in detail, admittedly as presented by its partisans, but in the format required by the Home Office, with answers to questions about purpose, method, the 3Rs, and so on. The animals appear in more comprehensible numbers (150 pigs, 200 chickens), and their kind is more accurately identified (crows, rainbow trout, opossums, voles). What happens to them is more or less picturable, and the scene can be bloody and squalid, even where no suffering is involved: “In parallel to in vivo experiments, we will also carry out in vitro experiments using sheep uteri and ovaries collected from an abattoir” [God, what have we become?]. You get some idea of how scientists may have judged the pain levels which are later to be recorded in the statistics: “The expected adverse effects are the development of skin wounds, inflammation and cancer. In most cases the severity will be mild. However, in some situations, such as tumour development, the severity will be moderate.” [Excellent! Cancer is evidently not as bad as we feared.]

And now, with these and other Home Office publications about animal research to hand, you begin to realize that the word ‘procedure’, the key word in the Animals (Scientific Procedures) Act (ASPA) and the one on which you have to rely if the statistics are to make any sense, is itself a euphemism. Having myself been misled by this word, I shall try to show what’s wrong with it.

For the purposes of the Act, a “regulated procedure” is defined (see the User Guide, p.10) as “any procedure applied to a protected animal for an experimental or other scientific purpose, or for any educational purpose, that may have the effect of causing an animal pain, suffering, distress or lasting harm equivalent to, or higher than, that caused by the introduction of a needle in accordance with good veterinary practice.” The breeding of a genetically altered (GA) animal is quite properly counted as one such procedure under the Act, and we’re told in the 2-page summary that about half of those 4.14 million procedures “related to the creation/breeding of genetically altered animals that were not used in further procedures.” That seems to make good sense. The breeding would be one procedure. Those GA animals for whom that turned out not to be a sufficient contribution to science would become part of other (“further”) procedures, counted as such.

But in fact we know that this isn’t what’s being done. It would mean that there’d be many more procedures than animals in the total count, whereas we’re specifically told that the two numbers are always more or less the same, and that in the rare cases where the number of procedures is higher than the number of animals used “this is due to a re-use of animals” (User Guide, p.9). ‘Re-use’ is a term always meaning ‘used in a different project of research’, which is actually by no means a common practice. And for this purpose, GA breeding apparently doesn’t count as a different project. So the real situation is this: animals which have undergone the GA procedure, and are then used in “further procedures”, still count for only one procedure each.

All right, but even apart from the GA question, ‘procedure’ has a very elastic meaning, which seems to include its own plural. It may just mean an injection, such as the one which is the model for what minimally constitutes a regulated procedure as defined in the Act. On the other hand, it can mean a whole “series of regulated procedures”: that’s the phrase which the Home Office Use, Keeping Alive, and Re-use Advice Note (p.9) uses when reviewing the experience of an animal during one research project, and advising on its suitability for ‘re-use’. The User Guide explains (also p.9): “Each procedure (which may consist of several stages) for a given purpose on an animal is counted as one returnable procedure.” ‘Procedure’, it emerges, is a collective noun, but what exactly it may have collected in any particular instance there’s no way at all of discovering from the statistics.

I don’t know whether I’ve been able to make things clear; probably not, because this key-word in ASPA is not used clearly and consistently even in the official documentation. To summarise, then. A ‘procedure’ is an animal’s whole career of procedures within one research project. If it’s a GA animal, that career will include the procedure which brought it into being, and may or may not include others. In short ‘procedure’ is a term so elastic as to be almost meaningless. The number 4.14 million, therefore, really means 4.14 million multiplied by an unknowable n.

This ambiguity must affect every aspect of the published statistics. For instance, the rule for deciding the painfulness or severity of a ‘procedure’ is that it should be put in the severest of the four classes (sub-threshold, mild, moderate, or severe) which it reaches at any point during the research. But you will see that the meaning of a severity class is itself obscured by the vagueness of the term ‘procedure’. A procedure classed as ‘severe’ may have been a brief torment constituting the whole of an animal’s part in modern science, or it may have entailed that ‘severe’ pain together with a succession of other ‘severe’ or ‘moderate’ or ‘mild’ interventions covering the full period of a research project. It makes a great difference to our understanding and (lest we forget) to the animal concerned, but the difference cannot be indicated in the Home Office statistics.

It’s no wonder, now I come to think of it, that Understanding Animal Research has been content to present the Home Office statistics on its web-site as the “reality” of animal research. In truth, they’re a mixture of understatement, euphemism, and unintelligibility. Despite all the varieties of show and tell that the animal research industry now agrees to, the essential secrecy remains. And I should say that outsiders will never really know what’s going on until we get the number of ‘procedures’ down to nought.



For Oxford University’s part in the 2015 numbers, see ‘Multitudes, multitudes’ in this blog (posted 24 April).

The Home Office’s Annual Statistics of Scientific Procedures on Living Animals, Great Britain 2015 can be seen at https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/537708/scientific-procedures-living-animals-2015.pdf 

Its User Guide to Annual Statistics of Scientific Procedures on Living Animals  is at https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/538549/guide-animal-procedures.pdf

Use, Keeping Alive and Re-use (dated October 2015) is at https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/470008/Use__Keeping_Alive_and_Re-use_Advice_Note.pdf .

Other references are to be found on the relevant web-sites.




9 thoughts on “Home Office statistics: numbers, words, and euphemisms

  1. Hi Matthew,

    Firstly, our official coverage of the UK stats can be found here: https://speakingofresearch.com/2016/07/20/animal-experiments-in-the-uk-statistics-show-4142631-procedures-in-2015/

    A few reflections. Your point about the large rise compared with 2001 is correct, but not necessarily fair. You have picked the lowest year since the early 1950s. A comparison with the mid-70s would find the numbers over a million lower for instance. Overall, though, we can agree there has been a rise in recent years – though this probably reflects increases in science funding (which will often have a delayed effect on the animal studies – as money can take time to filter through).

    If I’m understanding you correctly on genetically altered animals, I think you may be slightly mistaken. If an animal is genetically altered and then not used in other procedures – then they will be categorised under “related to the creation/breeding of genetically altered animals that were not used in further procedures”, if they the GA animal is bred and then used (the two will be related – they will have been bred for the purpose of the subsequent experiment), then they will go down as a single procedure somewhere else (probably basic research). So your sentence:”Those GA animals for whom that turned out not to be a sufficient contribution to science would become part of other (“further”) procedures, counted as such.” is incorrect. It should read “Those GA animals intended for further study would become part of subsequent interventions as part of the same procedure, counted as such.”

    Procedure is certainly an odd word in this context. What it means, is any number of (ASPA regulated) interventions that make up a single experiment on a single animal (in reality an experiment will nearly always involve multiple animals – each of whom will under a procedure [which can be multiple interventions]).

    Procedure is accurate – it basically means the intervention on a single animal during an experiment. To count individual interventions would be meaningless – when doing surgery would you count the anaesthetic as one intervention, the surgery as a second, the post-surgery anaesthetic as a third etc? Does opening up an animal AND suturing the wound count as one or two interventions?

    To argue the severity bands are being understated by UAR is also ridiculous. The severity banding – as you say – is the maximum suffering the animal may undergo. Sure, some Severe procedures are more severe than others, but that is still the maximum severity the animal feels. Some mild procedures are worse than others, but mild is still the “worst” it gets for that specific animal. Such severity reporting is a huge step forward in the animal stats.


  2. Thank you very much for reading the post and for your comments, Tom.

    It’s certainly true that the U.K. numbers were at their highest ever in the 1970s. But as you say, the direction in recent years has unfortunately been pretty steadily upward, which is all I meant to show. The HO itself makes the same point in its two-page summary, though it takes the starting point at 2006.

    The point about GA animals: yes, as you quote me, what I say is wrong, but at that stage I’m only offering what might be a reasonable deduction from the Home Office documents as to what’s going on, hence the conditional tenses. I then say, “in fact we know that isn’t what’s being done”.

    It may all seem a bit pedantic, this question of what ‘procedure’ means or can mean, as your example of the anaesthetic, etc., suggests. That’s why I’m glad to have comment from the animal research side: to find out whether it really is pedantic or not. But I don’t feel that you have dealt with the essential difficulty: namely that if ‘procedure’ can also mean (as I find from HO material) a “series of regulated procedures”, it is a term too elastic to be really informative. You say that a procedure “basically means the intervention on a single animal during an experiment”, but you’ve earlier said it means “any number of (ASPA regulated) interventions”. So ‘intervention’ seems similarly ambiguous. Is there a limit to what it can imply? I make the suggestion in the post that what ‘procedure’ actually means is one animal’s “whole career of procedures”, whether brief or extended, within one research project (or experiment, to use your term). Is that incorrect?

    About severity: please note that I don’t say that UAR is misleading anyone here, only that it endorses the HO’s euphemistic presentation (and I should add that I haven’t seen this point made by any other organisation on either side, which is why I feel uncertain about it and would like to see it discussed). I wasn’t saying that experiences within any one stated category can vary in unpleasantness, though of course that is necessarily true if the number of categories is to be kept manageable. What I was saying is that if ‘procedure’ can mean ‘series of procedures’, then any stated pain-category for a given procedure may imply a whole series of experiences of that severity or less. It makes a great but hidden difference to animal suffering, and that’s after all the one thing which all these numbers are supposed to give information about.

    I agree with you that the new HO rules for providing information about animal research are a great improvement. What I’m suggesting is that the numbers are still much less informative than they seem.


  3. I’d like to comment on the linguistic issue of procedures – I think this may also elucidate the issue of severity.

    “I make the suggestion in the post that what ‘procedure’ actually means is one animal’s “whole career of procedures”,”
    – I’m uncomfortable with this definition as it is self referential (in the sense that you’re defining the word procedure using the word procedure. To use your choice of words, I would say that that a “Procedure is the whole career of *interventions* on a single animal”. So we’re using “intervention” to a mean a single act on a single animal, and procedure to mean the collective set of interventions that make up a procedure on an animal. To extend the language discussion further, an experiment would be a collection of procedures on a set of animals (one procedure [multiple interventions] per animal).

    Now this may help understand the issue of severity. When classifying the severity of a procedure , the researcher must take into account the impact of all interventions, as well as post-surgical recuperation. While severity is partially determined by the highest severity caused by an individual intervention, it is also determined by long lasting effects, so long lasting mild pain would likely class the study as moderate (not mild).
    So, to quote from the Severity Reporting Guidelines released by the Home Office:
    “Animals that undergo procedures that produce chronic low-level pain or discomfort or dysfunction such as altered gait will usually be classified as moderate. A higher level of pain that persists, such as non-weight bearing lameness without improvement, even in the absence of other signs of severe pain, would be considered severe unless a diagnosis can be made that indicates the condition is associated with pain of a lower intensity.”

    Click to access NotesActualSeverityReporting.pdf


  4. Thank you very much for these helpful observations, Editor.

    About the severity: yes, I do see your point that duration is taken into account in the Home Office Guidelines. However, that only applies in cases where the pain to the animal is sustained or cumulative. In other cases – i.e. “When there is no increasing impact with multiple steps or if the suffering resolves completely between each step in a procedure” – the highest band actually reached in any of the steps is the one used for reporting purposes. Of course it wouldn’t anyway affect the particular example I give in the post, which is of a ‘procedure’ already in the ‘severe’ band, and therefore not amenable to pushing into a higher band in the way you speak of.

    About defining ‘procedure’: your suggested terminology seems excellent to me. I wasn’t myself meaning to define ‘procedure’ in any particular way, only to work out how the Home Office was using the term in its statistics and other documents. I agree that ‘procedure = career of procedures’ is a perverse definition, but it is the one on which the Home Office usage seems to be based.

    In fact I wish you could persuade the Home Office to adopt your terminology, provided that the statistics would then include a figure for the total of interventions. Since these must be recorded anyway as part of the severity calculus, it surely wouldn’t add significantly to the office work in the laboratories, though no doubt it would make the annual recording a bit more laborious. On the other hand, it would dramatically improve the information-value of the statistics.


    • It is frustrating that the word “procedure” appears to have a standard meaning separate from the Home Office definition.

      The problem with including statistics on total number of interventions is that defining an individual intervention is very difficult. Consider the following series of interventions and ask yourself how many interventions are each of these.
      – Food restricted diet for 3 nights, then animal anaesthetised, MRI scanned, placed in isolation for 1 day of post-anaesthesia observation before being returned to cage mates.
      – Food restricted diet for 2 nights, animal anaesthetised, animal weighed, animal is opened up, a catheter is implanted, wound is sutured, animal placed in isolation for 2 days with post-op anaesthetics provided on both days
      – 3 hour water restriction, animal placed in transport cage and moved to another room, animal weighed (trained to jump on scales), animal given visual health check, blood sample taken, animal trained to sit in MRI scanner, animal taken out after an hour for a biscuit and leg stretch, animal returns to MRI scanner, animal returned to cage mates.
      – 2 day water restriction (limited water, not none), animal placed in transport cage to weighing room, weighed (picked up and put on scales) then put in transport cage to surgery room. Animal anaesthetised and given analgesic. Blood sample is taken. surgery starts by opening the animal, animal injected with pathogen, animal scanned over the course of 9 hours, blood samples taken every hour, tissue sample from tail taken every 2 hours, animal euthanised without being woken up.


  5. Well, yes, I see the difficulty, but in each of the examples you give it would surely be quite reasonable to see the separate parts as forming one ‘intervention’, the end being signalled by a return to the cage, or death in the last case. In fact, this is the sort of thing that the term ‘procedure’ would naturally be taken to imply. It’s when the business is repeated, or a different sort of test or intervention is used, that a new count ought to begin for that animal: intervention two, etc.

    I can see that there would be many doubtful cases. But when the question is whether a series adds up to two, three, five, nine, or what, it’s quite a strange (though convenient) answer to say that the total shall never exceed one.

    I wonder how much this has ever been discussed in revisions of the legislation. In some ways I think that it was less misleading simply to count the animals, though I’m not advocating a return to that.


    • Procedures are routed in project licences. So one procedure = one animal in one project licence. Add up all the animals in a single project licence and you have the number of procedures. I can’t see a non-arbitrary way of adding interventions. If you were to count returns to the enclosure then what about injections given while the animal remains in its pen (common for animals like monkeys which are trained to come to the cage and offer a limb for an injection).


  6. My point about counting animals is not that the total would be much different (see the original post on that subject), but that the term ‘animal’ is unambiguous, whereas ‘procedure’ is not.

    I suggested ‘return to cage’ as a reasonable closure only because you put those particular examples to me. It certainly wouldn’t work for the sort of in-cage injection which you now mention. But since such an injection is cited (in the User Guide) as the paradigm of a ‘procedure’, and must certainly be individually recorded, I don’t see why it shouldn’t constitute a single and complete ‘intervention’. But please understand that I’m not proposing to wrap up the procedure/intervention problem with these amateur suggestions, only to say (what your helpful comments themselves have shown me) that there really exists, or could exist, a much more informative statistical unit than the ‘procedure’.

    I take it that the point of the Concordat on Openness is not just to defend animal research, but also genuinely to make it more open. Here is an opportunity to make the crucial annual statement on the subject, the Home Office statistics, much less misleading and very much more informative. Wouldn’t that be a good thing to attempt?

    I ask that question without meaning that you should give any more of your time to this post. You’ve certainly done your duty by it – for which, thank you.


  7. Pingback: For We Are Many | Voice for Ethical Research at Oxford

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s