Alternative facts and the post-truth world: my February 2017 “Best of Health” article

Best of HealthDealing with “alternative facts” and the post-truth world

It’s pretty clear from what’s happened in recent months that we’re living in what’s being labelled a “post-truth world” where personal opinions and “alternative facts” are used as the basis of policy-making.

I’ve written before about analysis paralysis and the dangers of waiting for the perfect set of data before taking action to address breed health issues (you’ll be waiting a very long time!). Surely, there has to be a middle ground where we can develop plans and implement improvement actions that are evidence-based but where we can be agile enough to change course if new evidence emerges.

Breed Health Coordinators know only too well how hard it can be to have a sensible conversation with a breeder who “has never seen this issue in 25 years of breeding” and therefore believes it cannot possibly be something of concern. BHCs are constantly trying to explain (in plain English) that data from surveys describes what is happening in a population and that may be very different to what’s happening to the health of an individual dog.

One way to build a case to demonstrate action may be needed (or not needed) is to triangulate in on the evidence from several sources. So, for example, the Swedish Agria and VetCompass databases provide a large quantity of data on multiple conditions and thousands of dogs. Individual breed surveys, including the KC’s 2004 and 2014 surveys provide additional data, but typically covering fewer dogs and from different owner samples. A third source is published research papers, many of which focus on very specific health conditions and there will invariably be many of these studies published over the years. A simple search on Google Scholar will find hundreds; for example, I found over 350 papers on IVDD in Dachshunds. You can even create an alert so that you get sent an email every time a new paper is published related to a keyword you choose.

Evidence vs. Data

Of course, one thing everyone needs to understand is the difference between “data” and “evidence”.

I could tell you that as many as 1 in 4 Dachshunds is likely to have some degree of back problem during its life. That’s data, but on its own, it doesn’t have much validity or reliability unless you know something about its context. What was the sample size, how was the data collected and what is it going to be used for? Data can exist on its own but is pretty useless without context.

Evidence, however, can only exist to support a theory, an opinion or an argument. So, if in my opinion too many Dachshunds have back problems, I need to provide some data to support that opinion. That data comes from research, including routine health surveillance.

If you want to improve something, you need to have evidence to support a case for taking action. In the case of Dachshunds, there is evidence to show that the more calcifications you can count in X-rays of a dog’s spine around 24 months of age, the more likely it is to suffer IVDD and its offspring will also be at more risk. There is lots of data to back up that evidence, published in peer-reviewed papers, and that’s why we launched a new X-ray screening programme in November last year.

My 2 Golden Rules for Breed Health Improvement are:

  • There should be no action without evidence
  • There can be no evidence without data

An important point here is that the people expected to implement the action (e.g. owners who we want to screen their dogs) don’t need to understand the data but they do need to believe the evidence. So, those of us who love getting our hands dirty with the data need to become better at storytelling. We need to present the evidence in easy-to-digest formats: infographics are one way, as are success stories from other breeds or other countries.

The wrong solution to the wrong problem

The UK’s National Statistician John Pullinger recently wrote that there is a huge opportunity for statistics in the post-truth world. He said there is great potential to mobilise the power of data to help us make better decisions. But, he points out that with people spending ever more time getting their news from social media channels, we risk connecting only with those with similar views to our own and never encounter those who think differently. This can mean we fall prey to those who choose to support their own opinions with “alternative facts”.

Government is supposed to follow the principles of evidence-based policy-making. The whole point of this approach is that government asks Civil Servants to review and analyse the available data before drafting legislation. They should also be analysing the counterfactuals – what would happen in the absence of the policy or legislation. Both human and veterinary medicine should also be developing evidence-based practice and we need to be doing this with breed health improvement too.

Evidence-based practice is designed to avoid policies being developed either as a knee-jerk reaction to circumstances (exactly what happened with the Dangerous Dogs Act) or on the basis of a politician’s personal agenda or ministerial whim.

Politicians and those in positions of power, such as ministers, are notoriously bad at asking for data and evidence, let alone using them to inform decisions. Steve Dean also noted this in one of his Our Dogs articles on the outcomes of the EFRACom review of canine welfare issues. His article “Poor research and little science” discussed the lack of critical information to support the committee’s views and recommendations. He concluded by saying “attempting to impose sanctions on the majority, to deal with a disreputable minority, is a repetitive misdemeanor of governing bodies“.

Politicians and animal welfare campaigners too often look for simple solutions to complex problems. The last thing they want to do is to look at the data or evidence because, often, these would undermine the rationale for their current “pet policy”. As a consequence, they end up implementing the wrong solution to the wrong problem which is what has happened with the Dangerous Dogs Act. They also end up with unintended consequences and even more bad publicity!

The New York Times’ Andrew Revkin blames pervasive misinformation in part on “single-study syndrome,” in which agenda-driven fringe groups promote studies supporting a predetermined position — no matter how questionable the research behind them may be.

We mustn’t fall into that trap with breed health improvement. We need just enough data and evidence-based policy-making.

I’ll end with a quote from Jill Abramson writing in the Guardian: “Alternative facts are just lies, whatever Kellyanne Conway (advisor to Donald Trump) says”.

.

.

.

.

.

.

.

.

 

 

 

Genetic Diseases: Not so simple after all? January 2017 “Best of Health” article

Best of HealthA recent paper, “The continuum of causality in human genetic disorders” by Nicholas Katsanis caught my eye because it describes some emerging thinking that has relevance in the world of canine genetics.

Katsanis writes about recent advances in our understanding of the overlap between so-called simple and complex diseases. These two extremes are characterised as either diseases caused by mutations in a single gene (“simple”) or those caused by mutations in many genes (“complex”).

In the case of the simple diseases, genetic technology has enabled organisations such as the AHT, Laboklin and VetGen (to name but a few) to commercialise DNA tests for autosomal dominant and recessive mutations. Most readers will be aware of these DNA tests for Progressive Retinal Atrophies, Primary Lens Luxation, Hereditary Cataract and other eye conditions. Tests also exist for conditions such as Urate Stones and neurological conditions such as Cerebellar Ataxia and Lafora’s Disease.

The University of Sydney’s database (Online Mendelian Inheritance in Animals) currently lists 691 traits or disorders known in dogs, of which 289 are Mendelian (Dominant or Recessive) and 215 of them have an identified mutation. That doesn’t mean there are DNA tests for all of these, though!

Complex diseases in dogs include Hip and Elbow Dysplasia, Epilepsy, BOAS and Intervertebral Disc Disease. These are often described as also being associated with multiple environmental factors.

The first challenge Katsanis discusses is “one gene, one phenotype” where we now have examples of single genes actually driving multiple disorders. The best-known is a mutation (CFTR) which is primarily associated with Cystic Fibrosis but is also known to cause male infertility and lung disease. The interaction of CFTR with other genes or environmental factors also now needs to be studied.

Not all “affected” dogs go blind

The second challenge relates to “penetrance”, a term that has basically been created to explain why classically defined dominant or recessive traits do or do not actually manifest as particular phenotypes. Penetrance is defined as the percentage of individuals with a given genotype that exhibit the phenotype associated with that genotype. So, you might find that not all dogs who have two copies (“affected”) of a particular mutation associated with PRA actually go blind. We have an example of this in Dachshunds where people have been asking why so few Miniature Smooths have been reported to have gone blind despite being DNA tested as “affected” for Cord1 PRA.

I am grateful to Dr Cathryn Mellersh for her explanation of this situation and the emerging evidence. The original research that established this association was undertaken with a very inbred colony of Miniature Longhaired Dachshunds that all descended from a small number of animals. In the original colony of dogs, there was absolute correlation between a mutation in the gene RPGRIP1 and the CORD1 phenotype.

Following the original publication, it became apparent that there was considerable variation in the age of onset of disease in dogs that carried two copies of the mutation, indicating (an) additional modifying gene(s) might be involved in determining the age of onset. Additional studies provided firm evidence of a second, modifying gene that influences the effects of the RPGRIP1 mutation.

The overwhelming conclusion from these studies was that the previously reported RPGRIP1 mutation, upon which the widely used DNA test is based, causes cone photoreceptors to malfunction and, together with an additional mutation in a modifying gene, causes early onset cone-rod dystrophy.

Identification of the additional mutation has been the subject of intense research over the last few years and last year reached a successful conclusion with the identification of a mutation in a second gene, called MAP9. This acts as an age of onset modifier of the RPGRIP1 mutation.

Dogs with two copies of both the RPGRIP1 and MAP9 mutations will have early onset Cord1 PRA. Dogs with two copies of the RPGRIP1 mutation but one or no copies of the MAP9 mutation will have impaired eyesight but may not display clinical signs until middle or old age and some may retain vision throughout their life. The fact that the MAP9 mutation is not as common as the RPGRIP1 mutation explains why so many Dachshunds do not have early onset blindness (PRA).

More unexplained variation

Another Dachshund example is the Lafora mutation in Miniature Wirehaired Dachshunds. The age of onset of the disease and the severity and nature of the symptoms are quite variable in dogs that have been DNA tested and found to have two copies of the mutation. The mutation is a dodecamer repeat in the EPM2B gene and it has been speculated that variation in the number of repeats may be associated with the diverse clinical manifestation of the disease. A few years ago, we had conversations with a European genetics research group who felt the Lafora mutation had some similarities with the behaviour of the mutation associated with Huntington’s Chorea, a neurological disease affecting people. In that disease, which is inherited as an autosomal dominant mutation, they said variation in the triplet repeat affected symptoms and age of onset. The more repeats of the mutation there are, the earlier the age of onset. It would be interesting to investigate this for Lafora.

With nearly 50% of Lafora DNA-tested Mini Wires carrying the mutation, it is easy to understand why UK breeders of the other Dachshund varieties are so concerned not to introduce this to their gene pools.

This brings us to Katsanis’ third challenge point; the traditional view that alleles associated with rare disease are themselves rare in a population. Recent research has found that the adverse effects of an allele can be modulated by neutral alleles in the same gene or elsewhere. Another recent study of canine mast cell tumours demonstrated a stepwise accumulation of numerical DNA copy number aberrations (CNAs) as tumor grade increases. So, the concept of simple Mendelian relationships between genes and diseases is becoming less clear-cut.

At the recent KC Breedwatch Education Day, Dr Eleanor Raffan from Cambridge University spoke on the subject of canine obesity and described her research into the genetics of obesity in Labradors. This work identified the POMC mutation in overweight dogs: dogs with 1 copy of the mutation weighed more than wildtype and those with 2 copies weighed yet more and this explained most of the variation in body weight. The presence of the mutation also correlated with dogs’ food motivation. The POMC mutation interrupts the leptin pathway, which regulates appetite, so dogs with the mutation don’t know when to stop eating!

75% of Labradors are wildtype (2 copies of the normal gene) but this was as low as 20% in Guide Dogs. Temperament and “trainability” are the main drivers for selection of assistance dogs, and “positive reinforcement” with food reward is a mainstay of puppy training. Eleanor therefore hypothesised that dogs carrying the POMC deletion may be more likely to be selected as assistance dogs. There are also, of course, genetic factors associated with temperament.

The POMC mutation also exists in Flatcoated Retrievers, a breed not normally associated with obesity. Interestingly, Golden Retrievers don’t have the POMC mutation. The observations of these two breeds were particularly interesting for the Breedwatch audience who felt FCRs were generally not considered to be overweight in the show ring, in contrast to some GRs, which can be overweight.

No “simple” choices to improve health

The Katsanis paper concludes by saying that, “for clinical diagnosis, the emphasis on rare penetrant alleles must persist in order to understand causality and develop intervention strategies; however, this must be coupled with improved statistical models for assessing the contribution of multiple factors, both genetic and non-genetic, to complex traits”.

This is equally true of canine genetics and diseases; simple DNA tests will continue to be helpful to reduce the risk of breeding clinically affected puppies. However, the more tests that become available, the more difficult it will be to make simple choices about breeding away from particular diseases. When you then consider environmental factors, the potential relationships between genes and the effects of genes on the biochemical processes that cause disease, it is clear that “testing” our way out of canine health problems is doomed to fail.

.

.

.

.

.

.

.

.

 

20 videos on how to strip a Wirehaired Dachshund’s coat

I’ve updated my Wirehaired Dachshund grooming page and there are links to 20 YouTube videos on how to strip a Wire’s coat.

.

.

.

.

.

.

.

.

.

.

 

.

..

.

.

 

Let’s celebrate our Breed Clubs and Breed Health Coordinators – December 2016 “Best of Health”

Best of HealthI’ve had many interesting conversations with Breed Health Coordinators (BHCs) about what they are trying to achieve in their breed. One of the great things about most BHCs is that they are so passionate about their breed, its health and its development. There’s no lack of enthusiasm and motivation to make a difference, but there’s never enough funding or resource to match these.

Our conversations often lead us to discuss the available data which could give an insight into where and how the BHC could make the most difference and how this could be achieved. So, for example, what’s the value of a big, expensive and time-consuming Breed Health Survey that engages lots of owners, if nothing happens with the results? At the other end of the spectrum, Dave Brailsford’s “1% Principle” might suggest that finding a low-cost (online) way of educating lots of owners about good husbandry, breeding practices and the need for health testing could aggregate into a sizeable pool of benefits. This latter point is particularly relevant when you look at the most common health conditions identified in the VetCompass Project: Otitis, Periodontal disease and Impacted Anal Glands.

Passion without data will take you a long way, but in an increasingly complex and uncertain world, the danger is that you simply won’t be able to demonstrate the impact you are making and won’t be able to persuade owners and breeders to participate in future initiatives.

In our conversations, we usually decide that data without passion is simply “boring”. Nobody really cares about numbers. When was the last time you got excited to hear someone saying “I’m going to do a t-test” or “Let’s draw the regression line”?

I’m convinced that Passion plus Data is a winning combination for any breed that is serious about safeguarding its future.

BHC of the Year Award

The recent Breed Health Coordinator of the Year Award was a great way to recognise all the hard work of these often unsung heroes. This year’s winner was Liz Branscombe (Flatcoated Retrievers) and the other finalists were Sam Goldberg (Beagles), Penny Rankine-Parsons (French Bulldogs), Margaret Woods (Golden Retrievers) and Kathryne Wrigley (Gordon Setters). Among the characteristics being looked for by the judging panel were their ability to motivate breeders and breed clubs in relation to health matters as well as their ability to encourage participation in health surveys and research projects. Passion plus data once again!

There’s probably also a good argument for having a diverse Health Committee in addition to a passionate and numerate BHC. There’s usually someone in each breed whose day-job involves doing things with data as well as all those enthusiasts who bring breed-specific knowledge, an understanding of genetics and veterinary science, or healthcare skills. If that committee also includes teachers, trainers, fundraisers and marketing people, you’ve probably got most bases covered. However, that probably also means there needs to be a critical mass of Breed Clubs from which to draw this pool of talent.

Breed Clubs make a difference

It’s three years since Philippa Robinson published her KarltonIndex review of the work being done by Breed Clubs to safeguard and improve the health of their breeds. I thought it might be interesting to see if there was any link between the number of clubs in a breed and their score on the KarltonIndex.

The KI assesses breeds against four parameters:

  • Leadership (having a strong health team and plan)
  • Communication and engagement (of breeders and owners)
  • Participation (of breeders/owners in health improvement activities)
  • Impact (on breed health)

I’ve done the analysis for the Hound Group and I found that, on average, Hound breeds with fewer than 5 breed clubs scored 12 (out of 100) on the KI, while breeds with more than 5 clubs scored 26 on the KI. (For the number jockeys reading, this was a statistically significant difference).

Breeds with more clubs tended to have a better KI score. I haven’t had the time to do the analysis for every breed and it would be unwise to say there is a direct cause and effect relationship between the two variables.

I have, however, looked at the Top 10 KI scoring breeds and compared them with the 14 breeds that scored zero points in 2013. The top breeds have an average of 7 breed clubs per breed, whereas the zero-points clubs average 2 clubs per breed. You’d probably also expect there to be a relationship between the number of breed clubs and the number of dogs registered. The top-scoring breeds accounted for around 56,000 registrations in 2015 (about a quarter of the KC’s registrations). In comparison, the zero-scoring breeds only accounted for 5,400 registrations (approx. 2% of the total).

There are, of course, good examples of breeds with few clubs who are also doing a great job (as measured by the KI). However, you just need to look at what Philippa found in her top-performing breeds to draw some conclusions about why breeds with more clubs might be making more effort and progress on health matters:

  • There are likely to be more people on committees who are passionate about making a difference for their breed
  • It is easier to find people willing and able to join their Health Committee
  • There are more people to call on to help run health seminars, screening sessions and to promote health initiatives
  • There are more clubs who can fund-raise and make donations to research programmes
  • There are more events at which health matters can be communicated and owners can be engaged in learning how to breed healthier dogs
  • There are more Facebook Groups, Club websites and social media channels being used to reach and involve owners and potential owners in health improvement

Ultimately, the KI score that a breed achieves is not that important; what matters is whether there are enough people who feel passionate enough about their breed to get on and do something. Breed Clubs are the catalyst for most of the good work being done across numerous pedigree breeds.

Let’s celebrate the work of our Breed Clubs and our Breed Health Coordinators! Best wishes for Christmas and the New Year.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

 

The dangers of evidence-free policy-making

rspca_-_bsl_dogs_dinnerI was interested to read David Cavill’s latest blog where he sings the praises of the RSPCA for their report on the Dangerous Dogs Act (DDA). ‘Breed Specific Legislation – a Dog’s Dinner‘ sets out the history of Breed Specific Legislation (BSL), its lack of effectiveness and proposes a number of solutions and recommendations.

The report concludes: ‘The RSPCA strongly believes that the evidence presented in this report clearly shows that BSL has been ineffective in achieving its goals of protecting public safety and reducing the number of prohibiting types of dogs.  Since its introduction in 1991 a significant proportion of dogs involved in fatal incidents are not those prohibited by law and hospital admissions due to dog bites have increased substantially in the past decade despite the provisions.’

It is truly shocking that, despite all the evidence, the Government still seems to believe that breed specific legislation is the answer. It was evident after just 5 years that it was not working as there had been no significant reduction in dog bites between 1991 and 1996.

Quite how this situation has been allowed to continue is puzzling since government is supposed to follow the principles of evidence-based policy-making. The whole point of this approach is that government asks Civil Servants to review and analyse the available data before drafting legislation. They should also be analysing the counterfactuals – what would happen in the absence of the policy or legislation.

Of course, all this is designed to avoid policies being developed either as a knee-jerk reaction to circumstances (exactly what happened with the DDA) or on the basis of a politician’s personal agenda or ministerial whim.

David describes the Dangerous Dogs Act as a “car crash piece of legislation” with a whole load of unintended consequences. He also points out that many of the parties who could make a difference seldom work together in a coordinated way because they are more interested in protecting their own “brand”. With true collaboration, pooled resources and a coordinated approach there might just be a chance of reducing dog bites and fatalities. After all, that’s what everyone wants to achieve.

Politicians and those in positions of power, such as ministers, are notoriously bad at asking for data and evidence, let alone using it to inform decisions. Steve Dean also noted this recently in his Our Dogs article on the outcomes of the EFRACom review of canine welfare issues. His article “Poor research and little science” discussed the lack of critical information to support the committee’s views and recommendations. He concluded by saying “attempting to impose sanctions on the majority, to deal with a disreputable minority, is a repetitive misdemeanor of governing bodies“.

Politicians too often look for simple solutions to complex problems. The last thing they want to do is to look at the data or evidence because these would undermine the rationale for their current “pet policy”. As a consequence, they end up implementing the wrong solution to the wrong problem which is what has happened with the Dangerous Dogs Act. They also end up with unintended consequences and even more bad publicity!

You can read the RSPCA Report here.

.

.

.

.

.

.

.

.

 

%d bloggers like this: