How long will my dog live? – cautionary comments by Dr Brenda Bonnett

I am grateful to Dr Brenda Bonnett for the following comments on my blog post “How long will my dog live?“. Brenda is an epidemiologist and CEO of the International Partnership for Dogs.

Thanks for an informative overview of this paper.  As an epidemiologist, however, I think a few cautions should be added.  Owner-reported mortality frequently lists ‘old-age’ as a cause of death, whereas this is essentially never listed by veterinarians or clinical data.  To me, what that designation means is that the dog achieved an age that approximated the owner’s expectation, and of course that would vary across breeds and owners.  Where breeds have a lower median age at death, owners might be more likely to pursue veterinary care, thereby resulting in a diagnosis.  This could confuse a comparison across breeds.  Not to discount the information, but just to suggest a little caution in its interpretation. 

A more important caution is in the use of proportional mortality.  As this value does not account for the actual rate of death or the base population, using it to compare across breeds is very risky. Let me give you one example from Swedish insurance statistics on deaths before 10 years of age in a population of millions of dog-years-at-risk, and reflecting almost 40% of all dogs in Sweden.  Bernese Mountain Dogs, Flat-coated Retrievers and Golden retrievers all had as their #1 cause of death Lymphosarcoma (a type of cancer).  So, if one had looked at proportional mortality, that was the highest cause.  On that basis, on might be tempted to say that these breeds were similar for this cause of death.  However, when one looks at the rates of death, and compares them as relative risk, the picture is very different as seen in the graphic below.  Essentially, very few Goldens died before 10 years of age, but among those who did, lymposarcoma was the most likely cause. It is fairly obvious that the importance of cancer as a cause of death before 10 years of age is very different for these breeds.  So – proportional mortality can be useful within a breed, but it is very dangerous to use it to compare across breeds.

Again – not to detract from the useful information in the paper, just to caution us all on how we apply it.

Advertisements

Longevity study – how long will my dog live?

How long will my dog live?

It seems a long time ago, but in 2014 the KC ran its pedigree dogs breed health survey with an online survey that attracted just under 50,000 responses. Among these were 5663 reports of dogs that had died. Now, that set of mortality data has been analysed and published in an Open Access paper: “Longevity and mortality in Kennel Club registered dog breeds in the UK in 2014”. The co-authors are Tom Lewis, Bonnie Wiles, Aimee Llewellyn-Zaidi, Katy Evans and Dan O’Neill; names that will be familiar to many readers.

There are some interesting findings in the paper and I’d like to share a few of those, this month.

The most commonly reported causes of death were old age (13.8%), unspecified cancer (8.7%) and heart failure (4.9%); with 5.1% of deaths reported as unknown cause. Overall median age at death was 10.33 years. Breeds varied widely in median longevity overall from the West Highland Terrier (12.71 years) to the Dobermann Pinscher (7.67 years). There was also wide variation in the prevalence of some common causes of death among breeds, and in median longevity across the causes of death.

What do dogs die of?

All dogs are going to die of something (!) so it’s perhaps good news to find that owners reported “old age” as the most common cause of death. Interestingly, “old age” as reported by the owners ranged from just under 6 years old to just over 22 years old. The median age of death under the “old age”category was 13.7 years.

At the recent Breed Health Coordinator Symposium, Dr Mike Starkey told us that 1 in 4 dogs will be affected by cancer so it’s probably not surprising to see Cancer (of unspecified types) as the second highest cause of death. The median age of cancer deaths was just over 10 years, again suggesting it is as most people would expect, a condition of older age. The range for age of death due to cancers was very wide: 2 months to 21 years.

What do different breeds die of?

It’s well-known that canine longevity varies considerably depending on the size of the breed; giant breeds have shorter lifespans while smaller breeds tend to live longer. Of particular interest to me was a previous VetCompass study that showed Miniature Dachshunds to be among the longest-lived breeds. This breed was subsequently chosen to be a long-lived representative in a genome-wide association study that Cathryn Mellersh (AHT) and other colleagues conducted to compare the genomes of long and short-lived breeds.

The latest paper shows data for “Within Breed Proportional Mortality” (WBPM). This is a way to look at the relative differences between the various causes of death for each breed where there were sufficient reports. (Unfortunately, from my personal point of view there were too few Dachshund reports to be included in this analysis).

This is where the paper gets really interesting. The data shows, for example that the WBPM for “old age” ranged from 3.85% in Bernese Mountain Dogs to 25.0% in Bearded Collies. In other words, significantly fewer BMDs die of old age than Bearded Collies. The WBPM for ‘cancer – unspecified’ ranged from 0.00% in Gordon Setters to 19.56% in Flat Coated Retrievers. The WBPM for ‘heart failure’ ranged from 0.00% in Whippets to 19.82% in Cavalier King Charles Spaniels. Again, these reflect what most people know about cancer risk in FCRs and heart disease in Cavaliers. The analysis also shows that Border Terriers had the highest WBPM for dying as a result of road traffic accidents.

This WBPM data enabled the authors to identify how individual breeds’ causes of death compared with the Overall Proportional Mortality (OPM) based on reports for all the dogs in the survey. Boxers and FCRs were the 2 breeds less likely to die of old age compared with the OPM. Cavaliers were less likely to die of cancers, compared with the OPM, but, as you would expect, were more likely to die of heart conditions. If your breed is among the 25 analysed in this way, it’s well worth looking at the data to see how it compares with your own experience.

Healthspan vs. Lifespan

Healthspan is an interesting concept that has become quite topical. A dog’s healthspan is the length of time it is healthy, not just alive. The paper says “Although death may be postponed by improved healthcare, extended longevity by itself does not necessarily imply an improved or even a good quality of life, so a delicate balancing act exists between longevity and acceptable quality of life.” This leads to challenging ethical debates about whether a shorter but healthy lifespan with a short, rapid decline to death, might be preferable to a longer life with long periods of illness and a slow decline to death. There is, inevitably, a difficult decision to be made by owners, with vets, about treatment options to prolong life, quality of life and when might be the right time to consider euthanasia.

The concept of healthspan means that longevity almost certainly means different things in different breeds. A giant breed would, typically, be expected to die younger than a toy breed but as long as the dog was healthy during that lifespan, most people would not consider there to be welfare issues. Conversely, long-lived breeds should not necessarily be considered as being “healthier”, particularly if much of their lifespan is subject to a debilitating illness.

4 categories of breed

The authors merged the results of longevity by breed with Within Breed Proportional Mortality (WBPM) and came up with 4 categories of breed:

  1. Long-lived with no specific cause of death at a raised proportional mortality (e.g. WHW Terrier, Bearded Collie, Gordon Setter)
  2. Long-lived with at least one cause of death at a raised proportional mortality (e.g. Labrador, Golden Retriever, Border Collie)
  3. Short-lived with no specific cause of death at a raised proportional mortality (e.g. GSD, Whippet)
  4. Short-lived with at least one cause of death at a raised proportional mortality (e.g. Flat Coated Retriever, Dobermann)Category 4 breeds are short-lived with serious, breed-specific, life-limiting conditions. Categories 1 and 3 are breeds where there is a wide variation in longevity associated with factors that apply across all dogs (such as size) and there is no obvious disease that accounts for death.

The paper concludes: “This study has identified individual breeds that have both a low median lifespan and also a high proportional mortality for one or more specific causes of death. Breeds with this combination are highlighted with potential welfare concerns that may need to be addressed.”

If your breed is one of the 25 breeds with causes of death with more than 50 reports, the paper is well worth reading and reflecting on what actions your breed clubs might need to be taking.


Please also read the comments by Dr Brenda Bonnett on this paper.

.

.

.

.

.

.

.

.

.

A review of breeding policies at 15 national Kennel Clubs

This month saw the publication of a paper in the Veterinary Journal titled “Breeding policies and management of pedigree dogs in 15 national kennel clubs”. The authors include Dr Tom Lewis from our Kennel Club.

The authors investigated approaches being adopted by Kennel Clubs internationally and what they see as high priority issues. They issued a questionnaire to 40 KCs and received responses from 15, 11 in Europe and 4 elsewhere (Australia, Mexico, Uruguay, and the USA). The European responses were from Austria, Belgium, Denmark, France, Germany, Ireland, Latvia, The Netherlands, Norway, Sweden and the UK. Also among the authors were Sofia Malm and Gregoire Leroy who I met at the IPFD’s 3rd International Dog Health Workshop last year. They were facilitating the workstream on Breeding Strategies and Gregoire blogs regularly on the IPFD website (dogwellnet.com).

We know in the UK that our KC believes it registers around 35% of pedigree dogs which leaves a large number of breeders and dogs that fall outside its direct influence. I suspect that, historically, the KC and most breed clubs have taken the view that they can only influence dog owners among the registered population. Given the high percentage of unregistered dogs, the question therefore arises: who is looking after their interests? Certainly, in the Dachshunds, with our Pet Advisors among our Health Committee and Breed Clubs who are proactive on Dachshund Facebook Groups, we have taken the view that we need to help ALL Dachshund owners and potential owners. The dogs don’t know or care whether they are KC registered and if we can provide advice to all owners, that has to be a good thing.

The survey results from the 15 countries showed a range from less than 1% to 78% of dogs registered by their KC. The Nordic countries, in general, had a higher proportion of their pedigree dogs registered by their KCs but it’s worth remembering that the total dog population sizes in these countries are relatively small compared with say the UK and USA. The lowest proportions registered were in India, Nicaragua, the Dominican Republic and Hong Kong. The highest were in Finland, Sweden, Iceland and New Zealand. Our KC reported a figure of 35%.

One size does not fit all

When I wrote about the discussions at last year’s IDHW, I specifically commented on the international and cultural aspects that can significantly influence the choice of approach that will work for a Breed Health Strategy and the likely compliance from breeders and owners. This latest paper reinforces those comments. It is clear that what might work in the Nordic countries with smaller pedigree dog populations and a high compliance among breeders, is almost certainly not going to work in the UK, USA or Australia. That’s not to say we can’t learn from each other but a simple “cut and paste” solution that assumes “one size fits all” is doomed to fail. Each Kennel Club and each Breed Club needs to understand not only their specific challenges and priorities but also the context within which they are operating.

The paper goes on to discuss the different issues each of the KCs prioritised. It should be no surprise that exaggerated morphological features and inherited disorders ranked as the most important issues. It has been obvious for at least the past decade that these issues are significant and are not going to go away. The evidence that some breeds need serious action is overwhelming and anyone still calling for more data is, in my view, simply in denial. In the UK, we have seen the formation of the Brachycephalic Working Group whose report and action plan was published last year. To me, this seems like a model for collaboration and practical steps that the diverse range of interested parties (stakeholders!) can sign up to.Our KC ranked issues in the following order (most important, first): exaggerated morphological features, inbreeding and genetic variability, inherited disorders, puppy farming, legislative constraints to breeding, dog behaviour and economic constraints to breeding.

Health and breeding recommendations

Individual Kennel Clubs’ responses to these issues are also discussed and we can see how widely adopted different approaches are and the proportion of breeds these cover. “Health recommendations prior to breeding were provided for more than half of the breeds in 11 countries, health status for breeding was required in 10 countries, and the maximum numbers of litters or/and puppies produced by a single dog were restricted in seven countries. Three countries indicated they do not have any specific restrictions on choosing mating partners, while another three countries reported that specific restrictions on choosing mating partners were implemented for all breeds.” Only 1 of the responding KCs said they have no health recommendations in place prior to breeding. It’s not possible to tell from the paper or its supplementary data which countries place restrictions on choosing mating partners or the limits on puppies produced by a single dog (so-called Popular Sires). Similarly, we don’t know how compliant breeders are where these rules exist or their impact on dog health or genetic diversity.

Austria, Sweden and The Netherlands have breeding strategies covering all of their breeds. Five countries reported that they provide Coefficient of Inbreeding information online for 100% of their breeds (presumably that includes our KC via MateSelect). Three countries provide online advice mating tools for all of their breeds. The paper says that our KC provides EBV data on Hip and Elbow Dysplasia for 28 breeds (Sweden does this for 42 breeds). What’s interesting here is that there is a wealth of expertise available around the world and there should be many lessons learnt that can be applied to help KCs catch up, where they need to. I’m sure some of those lessons learnt would relate to the design and implementation of software solutions, as we often read about how easy or difficult it can be to navigate and find health or pedigree information in different countries. Applying those lessons learnt won’t necessarily be easy, particularly when KCs have legacy IT systems that really weren’t designed to meet the needs of today’s breeders or to cope with the newly emerging data and breeding tools.

Learning from each other

One of the other analyses was the pairing of countries with similar question response profiles. Our KC was most similar to the Danish KC and, perhaps surprisingly, France and the USA were paired. Uruguay/Mexico were also paired, as were Austria/Germany. There is potential for cooperation between these pairs of countries because of their similar responses. However, they might actually find equally useful insights by looking at countries with whom they have little similarity. Apparently, the French KC has already benefited from learning about our Mate Select system to develop their online database.

My main takeaways from this paper are (a) that the issues facing Kennel Clubs and breeders of pedigree dogs around the world have a lot in common and (b) that, by taking an international perspective, there is huge potential for more joined-up solutions to be developed. Solutions will necessarily cover access to and sharing of information on pedigrees, health conditions and test/screening programmes. In terms of creating real change and breed health improvement, I think the key will be the development of Breed-specific Improvement Strategies (Breed Health & Conservation Plans in the UK). Sharing these documents internationally could prove to be a critical success factor in accelerating the rate of improvement in dog health, particularly if we are able to learn what works and what doesn’t in different countries and cultures. Readers will not be surprised, therefore, to see me conclude that I believe the International Partnership for Dogs has a major role to play over the next decade.

What is it they say about “statistics”?

There seem to be endless discussions about the evidence for or against the prevalence of health conditions in specific breeds of pedigree dogs. The “front line” of these battles over data has been among the Brachycephalic breeds in the past 12 months. Numerous other breeds crop up for debate with a predictable regularity (GSDs, Cavaliers, Dachshunds, BMD, Flatcoats – the list goes on).

To use that dreadful cliche, “at the end of the day”, there is no single RIGHT answer for the prevalence of any given condition, or conditions, in each breed. The published answers are very dependent on who is doing the research, what their objectives were, how they designed the study, what dogs were used as sources of data and, finally, how the data was analysed and presented.

This month, I want to focus on that latter aspect; how the data was “manipulated” and presented. However, with an eye on unanticipated consequences, please don’t use this article as a checklist of ways to spin your data. It would be better viewed as a starting point for being curious (sceptical?) about studies that are being published and data that is being presented.

What answer would you like?

Cherry-picking is probably the easiest way to spin data; simply select the results that support your case and ignore the rest. It is not unusual for research studies to come up with different answers to previously published material. For example, Packer et al (2012) studied the relationship between body length and back disease (IVDD) and concluded that the longer and lower the dog, the higher the odds of it having IVDD. That clearly plays to an agenda that links exaggerated conformation to health issues. A subsequent analysis of a much larger dataset collected by the Dachshund Breed Council also published by Packer et al (2015) did not reproduce those findings. It would be wrong to cherry-pick the latter study as a way of justifying exaggerated conformation (particularly when our Breed Standard calls for moderation in body length and asks for sufficient ground clearance).

The Cobra Effect occurs when an incentive produces the opposite result to the one intended (also known as “perverse incentive”). A classic example here would be the decision to publish the results of a screening programme to showcase dogs with, for example, good hips and to show an overall improvement in scores over time. If owners choose only to submit “good” scores for publication, the published results will give a false impression of the state of the breed.

False causality occurs when you assume that if 2 events occur together, one has caused the other. There is, for example, data that suggests Pugs with a higher Body Condition Score tend to have a higher risk of BOAS. It might be unwise to conclude that “being overweight causes BOAS”. It may be more appropriate to suggest that there is an association between being overweight and BOAS, and therefore good husbandry advice to owners would be to keep their dogs at an ideal body condition score. Having said that, we know that being overweight is generally unhealthy and leads to all sorts of adverse health outcomes!

Don’t be surprised by contradictory results

Sampling bias is a great argument for anyone who wants to challenge a set of results. In its purest sense, it means that the sample chosen is unrepresentative of the general population. For most canine studies, the reality is that particular sampling frames were chosen either deliberately or by default and the results will inevitably reflect that decision. The sample frame might be “pet dogs”, “show dogs”, “dogs seen at first opinion vets”, “dogs seen at referral practices” and so on. That’s one reason why it is perfectly possible for apparently contradictory results to be obtained.

There are other aspects of sampling bias which can affect the results obtained in a survey or research exercise. There may be Area Bias which means the geographic origin of the sample is not representative of the whole population. For example, our 2015 Dachshund Health survey includes data on about 90 Australian Dachshunds. This group has a high prevalence of skin conditions compared with UK dogs and this is likely to be an area bias related to climate and environment.

Self-selection bias is perhaps one of the most used “excuses” for results being challenged. The argument is usually along the lines of “people whose dogs have been ill are more likely to respond” or “you can’t rely on show people to report honestly, if at all”. Both of these might be true and would lead to biased samples and results.

Social desirability bias occurs when people don’t want to admit to doing something that is perceived to be socially undesirable or, in the case of their dogs, is undesirable for the dog. Typically, owner-reported estimates of a dog’s body condition underestimates the degree to which dogs are overweight and the amount they are fed. Similarly, owners may report an overestimate of the amount of exercise their dog gets; e.g. 40 minutes is rounded-up and reported as “an hour”.

Of course, adding in a sampling bias to your data collection is an important consideration if you want to lie or mislead with your study results!

Averages can hide a multitude of sins

Finally, the use of Summary Statistics can be misleading. Calculating an Arithmetic Mean (average) may hide a large amount of variation and/or multiple causes of that variation. Dachshunds are generally considered to be a long-lived breed and were used as one of the breeds in a recent GWAS project comparing the genomes of long and short-lived breeds. A look at the age of death (AoD) histogram for the breed shows a Mean AoD of 9 years but this is skewed by the number of deaths due to IVDD. On average, these IVDD dogs die at 6, whereas all other causes of death occur at an average age of 10.

Age_of_death_2010-12.png

The most worrying misuse of summary statistics I have come across is the choice of the denominator in the calculation of the mean. Say, for example, a large multi-breed population survey of 1,000,000 dogs explores a health condition which is known to be prevalent in particular breeds. The prevalence in the total population might be just 1% (10,000 dogs). If there are 20,000 examples of one breed and, of those, 1000 have the condition, it would be misleading to say the prevalence was 0.1% “among dogs”. The most meaningful calculation is to report that the prevalence is 5% “in that breed” or that it is 50 times more common in that breed than in dogs on average in the sample population of 1 million. We need to understand whether health conditions should be addressed at the level of dogs in general, or if they are breed-specific. Both types of issue exist and masking breed-specific issues by reporting population prevalence is simply avoidance and denial.

So, next time somebody shares some statistical analysis with you, approach it with curiosity and try to figure out if they have some ulterior motive to manipulate your opinion. It might just be their lies, damned lies and statistics (to quote Disraeli).

This article was inspired by “Data fallacies to avoid” published at http://www.datasciencecentral.com

Happy, healthy dogs: breeders and owners want the same thing!

Best of Health

Sometimes, it’s easy to lose sight of what we’re all striving for. With vocal and polarised “camps”, the noise we hear often boils down to “ban x breed” at one end of the spectrum and “breed x is perfectly healthy” at the other. These positions are sometimes exacerbated by cherry-picking of data and evidence to prove a point or reinforce a position. What we all have in common is a desire to breed and own happy, healthy dogs, whatever our chosen breed might be.

I’ve written before that improving canine health and welfare requires a whole-systems approach. This is difficult because people with particular interests (or prejudices) are often only interested in proposing “simple” solutions to what is actually a complex problem. Another reason it is difficult is that there are too few tools and techniques that are widely understood for people to use in order to get a better understanding of the problem and to evaluate potential solutions.

I was therefore particularly interested to read a paper published in the Canine Genetics and Epidemiology Journal last month. The paper is open access and, although it includes some complicated analytical techniques, is well worth a read because of the novel approach it discusses. “Assessing the relative importance of health and conformation traits in the Cavalier King Charles Spaniel” asked a sample of breeders, judges and owners to assess the relative importance of health and conformation traits when selecting a Cavalier. The study was carried out by a team in Belgium (Wijnrocx, François, Goos, Buys and Janssens) with support from the Belgian Cavalier Club and the Cavaliers for Life Foundation.

By way of background, the plain English summary of the paper says: When selecting a future breeding dog, different disease characteristics and other traits have to be balanced against one another, which makes it a complicated task. In the case of selecting for a large number of traits, the exclusion of all affected animals might be very inefficient, since this may reduce the genetic diversity in a population or breed. A solution could be the use of a selection index, in which all traits of interest are combined into one single value according to their relative weight.

Attributes and choices

Wijnrocx et al chose Mitral Valve Disease, Syringomyelia and Eye Disease as important health conditions in Cavaliers that breeders/owners might be concerned about. They also chose a set of conformational attributes that might influence people’s choice of a Cavalier: Coat Colour, Eye Shape and Muzzle Length. Finally, they included 2 attributes that buyers and breeders would, typically, also consider: Price and Level of Inbreeding. These factors were identified and agreed through consultation with experts and key stakeholders. Clearly, many more factors could have been identified for inclusion in the research but there is a balance to be struck in order not to over-complicate the choices or to make them too burdensome for respondents. Even still, with the number of attributes in this study, the questionnaire was limited to 17 choice sets of 2 alternatives. Respondents were asked to choose between 2 different dogs that they would prefer to use as a breeding animal, given the various attributes.

The initial analysis showed that all the attributes were statistically significant, i.e. they mattered to the respondents, EXCEPT for price. Price was then excluded from the final modelling/analysis. Interestingly, the choices showed there to be no differences in preferences between breeders or owners. Both groups chose Syringomyelia (SM) and Mitral Valve Disease (MVD) as their top 2 traits to consider when buying a Cavalier. Incidentally, the choices they were presented with for SM and MVD were: (a) Tested and present, (b) Not tested or (c) Tested and free. Unsurprisingly, the respondents attributed a higher value to “tested and free” than the other 2 options. SM was prioritised above MVD, followed by Eye Disease. SM was more than twice as important as MVD and 5 times more important than Eye Disease, which possibly reflects the publicity given to the condition as well as respondents’ awareness of its welfare impact and the costs of veterinary diagnosis and treatment.

Respondents’ views of the level of inbreeding in their choices showed that they were concerned with levels over 6%. I checked the KC’s health website and discovered that the average Inbreeding Coefficient for Cavaliers is currently reported as 6.3%. So, breeding litters below this average would have been considered a positive factor by this survey’s respondents.

The “baby-face” factor

I was surprised to see that Eye Shape was scored as the third highest attribute (above Inbreeding) with the preference being for a “prominent” eye (the other choices being “small” and “wall-eyed”). This would seem to tie in with other research into buyer behaviour that suggests people are attracted to baby-like features when looking for a pet (hence the popularity of many of the brachycephalic breeds). The worry, of course, from a health and welfare perspective is that a preference for a prominent eye might result in dogs at risk of some eye diseases or of damaging their eyes. Muzzle length and coat colour were the lowest rated attributes.

While it would be easy to say that the findings of this study and the researchers’ conclusions are unsurprising and/or predictable, I think it’s an interesting approach to try to quantify a range of people’s opinions and place an order of priority on traits. Breeders have always had to juggle multiple traits when making their breeding decisions and, these days, the issues of genetic diversity and complex diseases make those decisions even more difficult and important. The authors describe this as “a first investigation in the rational thoughts of breeders and owners towards some non-economic traits such as desired conformation or beauty traits in the selection of a CKCS“. The work could be extended to include other traits but I’m not convinced that this would shed more light on what’s really important for the health of the dogs. It might be possible, eventually, to create a Selection Index for each dog that would be a measure of its value in the gene pool but the authors acknowledge that Estimated Breeding Values (EBVs) would also need to be known for each trait as well as some estimate of genetic correlation.

Priorities and practical steps

The reason this sort of analysis is important is that there are no simple answers to complex problems and, therefore, there will always be a need to identify and prioritise potential trade-offs when looking for solutions. The biggest risk of cherry-picking “simple” solutions is that they actually make things worse as a result of either unintended or unanticipated consequences. Techniques such as the choice experiment described in the Wijnrocx paper could be really useful to help develop a consensus on priorities and practical steps that can be taken to improve the health of dogs.

My closing thought for this month, with apologies to Eliyahu Goldratt: “The world of pedigree dogs is awash with ill-considered solutions to ill-defined problems”.

.

.

.

.

.

.

.

.

Trust or trash? Just what can you believe?

The annual KC Breed Health Coordinator (BHC) Symposium was, for the first time, opened up to people who are not BHCs. As a result, around 200 people attended the event which featured a packed agenda of topics. There has already been an overview of the day published in Our Dogs (13/10/17) but, this month, I want to focus on one of the sessions I found particularly interesting.

Sniffing out the science – how to interpret information on dog health”, was presented by Dr Zoe Belshaw from the Centre for Evidence Based Veterinary Medicine at Nottingham University. I am an avid searcher for, and reader of, scientific papers and other published information on canine (and human) health and genetics. There are a couple of basic challenges; knowing where to look and knowing what to believe. Remember, we are increasingly living in a world where so-called Fake News pops up, especially via social media. Zoe’s presentation addressed both of the challenges. Let’s start with knowing what to believe.

Almost every day, new scientific papers are published or there’s a blog post (often cross-posted onto Facebook or Twitter) with a mix of data, evidence, insights and opinions. There is ALWAYS a risk of bias in these; every author brings their own agenda. As readers, we also bring our own biases and I’ve written before about Cognitive Dissonance. For example, many people simply look for information that supports their existing opinions and tend to reject anything that goes against that opinion.

Another reality of published information is that there may well be different conclusions drawn about the same issue. That might be a result of bias or it might be related to the methodology used by the author in their study. An obvious example would be the different results published by Dan O’Neill’s VetCompass project compared with results from a Breed Club’s Health Survey. The two sample populations are completely different. VetCompass data comes from first opinion vets using standard VeNom classification codes for different health conditions. Many breed surveys will not use those codes; they may use categories that reflect terms in common use by “ordinary” owners. The population being sampled is also likely to be different; mostly breeders and exhibitors. It might therefore not be surprising for VetCompass to find the most prevalent diagnoses are Otitis, Dental disease, Anal sac impaction and overgrown nails, whereas a breed survey may be more likely to find issues such as PRA, epilepsy or allergies. They are both “right”, based on the methodology and the sample. In my breed, our Health Committee has always taken the view that we need to triangulate in on priority health issues by using a variety of evidence sources. As we find new information from different sources, we can tweak our approach and our priorities. For example, we amended some of the categories and conditions we list in our ongoing health survey as a result of the 2014 KC survey.

The Trust Triangle

Zoe talked about the Trust Triangle which describes the different types of information you might come across and the levels of trust that can be associated with each.

Trust TriangleAt the bottom of the Trust Triangle are non-experts with opinions. Facebook and social media are awash with these! Journalists and experts with a commercial interest also fall into this category. Next comes expert opinion; these are people who are widely acknowledged to be experts in their field. Many of them will know an awful lot about a very narrow field of science. They too come with their biases and personal agendas but, mostly, they will have years of experience and scientific data to back up their opinions. Moving up the Trust Triangle, we find primary scientific research. This is made public via “papers”, the best of which will be peer-reviewed, rigorous, well-reported and independent. At the pinnacle of trustworthy published scientific research are papers that present systematic reviews of multiple other studies. These publications dissect and critique a set of primary research papers in order to arrive at “the best evidence” to support a particular case (or to disprove it). This is the sort of work that Zoe’s colleagues do at the Nottingham Centre for EBVM and they then publish what can be considered to be best practice for vets and clinicians to adopt. As with all science, “best practice” today could well change if new research evidence emerges. A topical example is vaccination protocols where advice from WSAVA should have moved vets away from “annual boosters” of core vaccines to a less frequent regime (no more frequent than every 3 years, generally) or the use of titre testing to assess levels of immunity.

One potential issue with the Trust Triangle is that, the higher up you go, the less accessible the information becomes and, often, it also becomes more difficult to interpret and understand. At the base of the triangle, keyboard warriors and instant experts often use language pitched at a level that you’d expect to find in the Daily Mail (or Beano). At the top of the triangle, the language is (rightly) riddled with scientific terminology and often supported by statistical analysis that is impenetrable to the ordinary dog owner or breeder. Those of us who are interested in this material may need help from subject matter experts to interpret it and clarify the meaning so we can share it with other owners. The KC’s Health team and other Breed Health Coordinators with science, veterinary or research backgrounds are invaluable in this respect. Most BHCs will also have built a network of trusted specialist advisors to whom they can turn for advice when a new paper is published. We should be hugely grateful that people like Cathryn Mellersh, Dan O’Neill, Clare Rusbridge and Sheila Crispin, to name but a few, are so generous with their time and support to our breeds.

There’s a useful resource which Zoe pointed us at to help decide if we should “Trust it or trash it” (trustortrash.org). This takes you through a series of questions to identify: who said it, when they said it and how they know it. She also gave examples of good ways and places to find the information we need to support the development of breed health improvement strategies. These include Google Scholar, RCVS Knowledge, PubMed, BestBETS for Vets and VetSRev.

As champions of breed health improvement, BHCs can make good use of social media to communicate with owners. It enables them to reach a wide audience, for example through breed-specific Facebook Groups. What they communicate, however, needs to be distilled from evidence sources higher up the Trust Triangle.

.

.

.

.

.

.

.

.

.

Show me the data and I will show you insights and solutions

Some readers will remember me writing about the Dachshund Breed Council’s 2015 Breed Survey: DachsLife 2015. The aim of the survey was very specific – to identify any lifestyle factors that contributed to the risk of back disease (IVDD). The original survey design was done in conjunction with Rowena Packer and her colleagues at the Royal Veterinary College (RVC). We deliberately involved them in the design because we had ambitions to gather lots of data that could be used to build on Rowena’s earlier work on the conformational risk factors associated with IVDD. In 2013, Rowena published her paper “How long and low can you go?” which showed that the longer a Dachshund’s body, the higher was its risk of IVDD. This clearly had implications for breeders and judges, but we also wanted to explore what owners could do.

After we had reported our results, we handed the data over to Rowena to work her magic with the more sophisticated statistical analysis tools she has available. Now, we have just published that further level of analysis in the peer-reviewed journal of Canine Genetics and Epidemiology.

One of the great things about this journal, which is sponsored by the Kennel Club, is that all the articles are Open Access. I strongly believe that sharing data and making the results of scientific studies widely available is key to making progress with any sort of improvement. So, not only can readers read the full text of our paper, but they can also access the anonymised source data. Anyone can repeat our analysis and interrogate the data further.

“Without data, you’re just another person with an opinion” (Dr. Deming)

As soon as data, research and analysis are put into the public domain, people can begin to make informed decisions.

Of course, the data and analysis may well challenge existing preconceptions and urban myths. Our Dachshund analysis was no different. Conventional wisdom would have you believe that allowing your Dachshund to jump on/off furniture, or to run up and down stairs is potentially bad for their backs. The analysis actually showed the reverse; those dogs who were allowed to do those things had a lower odds of having IVDD.

We do have to be careful to remember that correlation does not imply causation and we point out in the paper that there may be a danger of “reverse causation” with some of the results. For example, it may be that dogs whose backs have herniated were subsequently not allowed to jump on/off furniture or go up/down stairs. The missing data is “when” did those activities start/stop in relation to the herniation. We can, however, find other studies which also showed that this sort of activity reduced the risk of IVDD. The practical advice for owners arising from the activity/exercise findings is that keeping your Dachshund fit and active is one way to reduce the risks of IVDD. It’s common sense and any reasonable person would expect a fit, well-muscled dog to be able to lead a normal lifestyle coping with furniture and stairs.

Another surprise from our analysis was that neutered/spayed Dachshunds had an increased odds of IVDD compared with entire dogs. In this case, we can be more confident that it is not a case of reverse causation (e.g. dogs with IVDD were subsequently neutered) because we know the age of neutering and the age of IVDD diagnosis. Indeed, we found that the younger a dog was neutered, the higher its odds of IVDD. This finding should not surprise us in light of numerous other studies in dogs (and people) that challenge the claimed health benefits of neutering.

This finding is going to be a hard one to sell to many vets who still seem keen to encourage new puppy owners to book their dog in to be spayed/neutered as soon as possible after its vaccinations or before 6 months of age. However, in this emerging age of evidence-based veterinary medicine, vets will, I suspect, find some of their prejudices and biases challenged.

Big data gives you big insights

One of the interesting challenges when the paper was peer-reviewed was around biases in the data. The survey participants were self-selecting, they were willing to “own up” to health issues in their dogs, they may have been keen to “make a point”. They all responded via the internet in an online survey, so there may have been demographic biases in the respondent population.

Our argument was that, if you have large volumes of data, these sorts of bias are less significant. We had over 2000 responses and looked at more than 50 variables in the lifestyle factors. The VetCompass project is similar; with literally millions of case reports from first opinion veterinary practices, the volume of data permits deep insights to be obtained with high statistical confidence.

Of course, we have to remember that epidemiological studies draw conclusions about populations, not individual dogs. Inevitably, there will be owners whose personal experience differs from the findings at population level. In our case, there will be comments like “my dog jumped off the chair and ended up paralysed” or “my dog was fit and well-muscled, but still had IVDD”.

Those individual cases are entirely to be expected and do not diminish the advice we are able to give owners based on population-level insights. We know that, fundamentally, IVDD has a strong genetic component and is highly heritable. Nevertheless, if every owner adopted the good lifestyle choices we have been able to identify, I am convinced that it would make some inroads into the risks of IVDD. I’ve written previously about Dave Brailsford’s 1% Principle: lots of small changes can aggregate to make a significant difference.

While I have focused on Dachshunds and IVDD in this article, I am sure the principles apply more widely:

  • Breed Health Coordinators and their committees should think about the design of their health surveys with a view to carrying out deeper analyses
  • There are many situations where breeds already have lots of data; collecting more or different data is not the solution; we often just need more insight from analysis of existing data
  • The answers lie in the data, but don’t be surprised if some people (owners, vets, breeders) feel challenged or exposed by this, particularly when the results run counter to their perceived wisdom (see my article on Cognitive Dissonance)

Finally, remember “Prejudice is a great time-saver; it enables you to form opinions without having to gather the facts”.