Wednesday, 16 December 2015

Hip, hip... huh?? (Part One...)

Last month, the British Veterinary Association and Kennel Club celebrated the 50th anniversary of the BVA/KC hip dysplasia scheme.

From the press release marking the event:

"Results from the Hip Dysplasia Scheme showed improvements in the median scores of 20 of the 21 most-scored breeds over the last 15 years, indicating a reduction in the incidence and severity of hip dysplasia in scored dogs."

Said the KC's Caroline Kisko:

“This data goes to show just how much of a positive effect health testing is having on the health and welfare of dogs. 
“The BVA/KC Canine Health Schemes are useful tools to support responsible breeding and, as evidence from the data from the hip and elbow schemes, they are going a long way in protecting the future health of the UK’s dogs.” 
“Breeders who health test their dogs should be tremendously proud that they are having such a sustained positive impact on dog health, and we would encourage any breeder who does not currently use the schemes to do so, to enable the positive results to continue.”

Well that sounds great, doesn't it? But let's have a look at the actual data for the top 21 breeds.

Click to enlarge

As you can see, while there has indeed been a reduction in hip scores, it is a very small one for most of these breeds - down just one or two points in almost 20 years (with perhaps only the Newfie and Gordon Setter showing a truly significant reduction).

"Such small changes may be statistically significant but it is doubtful they are clinically significant," says Gail Smith, Professor of Orthopaedics at the University of Pennsylvania.

And, in fact, if you look at the data provided for the other breeds in this latest report (download link), it is hard to share the KC's enthusiasm.

For the 159 breeds for which comparative data have been provided, 62% per cent have seen no recent improvement in hip scores - and in 25 breeds hip scores have actually increased!

Seriously, it's not much bang for your bucks given that breeders have spent millions hip scoring their dogs over the years.

It's no great surprise to Professor Smith who maintains the UK/OFA/FCI hip schemes are fundamentally flawed. It was in response to this that he developed the alternative PennHIP scheme which measures hip laxity. PennHIP  has been shown to be more effective in identifying which dogs which will go on to suffer degenerative joint disease. (Smith GK, Lawler DF, Biery DN, et al. Chronology of hip dysplasia development in a cohort of 48 Labrador retrievers followed for life, Vet Surg 2012; 41: 20-33)

"The hip-extended radiograph is simply not a good phenotype on which to make breeding decisions. It should be abandoned in favor of using the PennHIP DI. It’s that simple," insists Smith.

That said, not everyone agrees. There has, for instance, been considerable improvement in some breeds in Finland using conventional hip-screening - with the indication being that when there is breed-wide selection against poor hips, it can be effective (Finnish report here.)

Longevity has improved in the St Bernard in Finland, too - it's thought because far fewer dogs are being euthanised because of severe hip dysplasia.  In the 1990s, Finnish St Bernard's died on average at 5 years old. In the 2000s, it had increased to 7yrs 1mth - and today, it is 7yrs 6mths - quite an achievement.

Finnish geneticist Katariina Maki says: "All you have to do is get reliable results and choose breeding dogs from the better half of the population."  In fact, Maki says you don't even have to breed only from animals from the very best hips; just those that are better than the breed average. Progress will be slower, but doing it this way helps maintain genetic diversity.

I had a discussion about PennHIP a few years back with Tom Lewis - then at the Animal Health Trust, now a full-time geneticist at the Kennel Club. Lewis acknowledged that PennHIP might be better but was adamant that the BVA/KC scheme was still useful and would become more useful with the introduction of estimated-breeding values (EBVs). These launched at Crufts earlier this year and are available for 28 breeds. Lewis also pointed out that it was very hard to justify throwing away 50 years of data collected under the existing scheme.

I have some sympathy with that... but it does make you wonder how much better those EBVs would be if they were built on PennHIP data rather than on the current scheme.

Incidentally, I have gone back to the BVA to ask for more comprehensive historical hip-score data from which to better assess the success of the scheme, and will report further when I get it. Perhaps if it made more grading information available we could see more of an improvement?

In the meantime, I've discovered something else interesting about the BVA/KC hip scores - something every breeder should know.

Stand by for Part Two...


  1. I'm interested in where this fits with the ICB recent thoughts that all pups are born with good hips and deteriorate (or not) with age - the implication being that environmental impact is high in early life.

  2. Comparison of three radiographic methods for diagnosis of hip dysplasia in eight-month-old dogs

    Objective—To compare the accuracy of the extended- hip radiographic (EHR) score, the distraction index (DI), and the dorsolateral subluxation (DLS) score for identifying hip dysplasia in dogs at 8 months of age.

    Design—Cohort study

    Animals—129 Labrador Retrievers, Greyhounds, and Labrador Retriever-Greyhound crossbreds.

    Procedure—Radiography was performed when dogs were 8 months of age. Dogs were euthanatized at 8 to 36 months of age; hip dysplasia was diagnosed at the time of necropsy on the basis of results of a gross examination of the articular cartilage of the hip joints for signs of osteoarthritis.

    Results—The EHR score, DI, and DLS score at 8 months of age were all significantly correlated with degree of cartilage degeneration at necropsy. Sensitivity and specificity of using EHR score at 8 months of age to diagnose hip dysplasia (scores > 3 were considered abnormal) were 38 and 96%, respectively; sensitivity and specificity of using DI (values > 0.7 were considered abnormal) were 50 and 89%; and sensitivity and specificity of using DLS score (scores ≤ 55% were considered abnormal) were 83 and 84%.

    Conclusions and Clinical Relevance—Results suggested that specificities of the 3 methods for diagnosing hip dysplasia in dogs at 8 months of age were similar. However, the DLS score had higher sensitivity, indicating that there were fewer false-negative results. (J Am Vet Med Assoc 2001;219:1242–1246)

    Evaluation of multiple radiographic predictors of cartilage lesions in the hip joints of eight-month-old dogs

    Objective—To determine the radiographic methods that best predict the development of osteoarthritis in the hip joints of a cohort of dogs with hip dysplasia and unaffected dogs.

    Animals—205 Labrador Retrievers, Greyhounds, and Labrador Retriever-Greyhound crossbred dogs.

    Procedure—Pelvic radiography was performed when the dogs were 8 months old. Ventrodorsal extendedhip, distraction, and dorsolateral subluxation (DLS) radiographs were obtained. An Orthopedic Foundation for Animals-like hip score, distraction index, dorsolateral subluxation score, and Norberg angle were derived from examination of radiographs. Osteoarthritis was diagnosed at the time of necropsy in dogs ≥ 8 months of age on the basis of detection of articular cartilage lesions. Multiple logistic regression was used to determine the radiographic technique or techniques that best predicted development of osteoarthritis.

    Results—A combination of 2 radiographic methods was better than any single method in predicting a cartilage lesion or a normal joint, but adding a third radiographic method did not improve that prediction. A combination of the DLS score and Norberg angle best predicted osteoarthritis of the hip joint or an unaffected hip joint. All models that excluded the DLS score were inferior to those that included it.

    Conclusions and Clinical Relevance—A combination of the DLS score and Norberg angle was the best predictor of radiographic measures in 8-month-old dogs to determine whether a dog would have normal or osteoarthritic hip joints. (Am J Vet Res 2003;64:1472–1478)

  3. The 'average' score for a breed is meaningless. Look at the first Finnish graph and compare the green and blue bars. The earlier results (blue) are fairly evenly spread across the categories, while the later ones (green) are higher in the better categories, but both may produce a similar average value. The difference between the two sets of figures is something called 'skewness', which is a well-defined statistical measure. The green values are positively skewed i.e. there are more entries in the better categories. We need to see the skewness values for each year as well as the 'average' values. It would also be useful to know whether they are using mean or median as their average. Median just means the middle value when they are ordered smallest to largest, while mean is the classic average of total divided by number of objects. If you have a few very high values the mean can be affected, but the median will be unaffected. A good result would be to see the mean and median values falling, but also the skewness becoming more positive.

    Chris R.

    1. These breeds have estimated breeding values (EBVs) in Finland, where average is set to 100.

  4. Two things I noticed in the first table that make me more optimistic than Jemima:

    1) At first glance, the numbers don't seem like they're changing that much: 10 to 9, or 14 to 12. However, this is actually a very fast change in 14 years compared to the normal speed of evolutionary changes in skeletal structures. I've lived in several developing countries, and--believe me--poor conformation does not go away in a few generations when dogs take over their own breeding choices.

    2) I'm assuming the dogs being tested are a somewhat random group from each breed; that is, dogs whose owners took them in for testing. The average hobby breeder breeds for <5 years, so that means that the dogs tested in 2010 are not necessarily the offspring of the dogs tested and "approved for breeding" in 2007 (for example). This makes the results even more meaningful, because it shows that changes are happening across the overall population, and not only in a tiny subgroup of the dogs whose owners get them tested one generation after another.

  5. AVA, NZVA are the same as BVA and they are meaningless! It all depends on the scorer, whereas pennHIP cannot be faked. The problem is that pennHIP will never be accepted by the FCI and therefore the qualified vets in Europe are few and far between. I have yet to add a single pennHIP score for a Leonberger located anywhere in Europe on the International Leonberger Database, so therefore the dogs x-rayed by this method only come from USA, Australia and New Zealand and the numbers most certainly are not representative. All the same, they give a far more true evaluation.

  6. It's hard to know what these results mean for many reasons:
    1. Are all results tabulated, or is there a chance that breeders who end out with very bad scores opt not to report them? Ie., are the samples biased. Bias is especially likely in breeds where HD is not a recognized problem. E.g., if very few fox terriers (or any other small breed not known for hip problems) gets a radiographic exam, a litter that for some reason had extremely bad scores could greatly skew the statistics.
    2. Are those bilateral scores, or average of both sides?
    3. What is the association between radiographic score and prognosis in the real world? In my experience (with Labradors), it's only the dogs with really bad scores who suffer. Moderate dysplasia . . . likely to get some arthritis at 8 to 10 years of age, but nothing crippling. Plus, some cases of HD have clear environmental origin. I had a 'friend' of sorts who ran pups behind a bicycle on hard surfaces. By golly if all her mature dogs didn't end out with bad hips.
    4. Medians are confusing. Hopefully, breeders are increasingly taking dogs out of the breeding pool when they have seriously bad hips. But 10 is not seriously bad for a medium/large breed dog. The more important question is whether the clinically important numbers, eg., 30's and above, are being culled. My guess is they are in breeds that worry about HD.
    5. I don't know of any studies of hip scores for wolves, or canids selected by natural selection. It wouldn't surprise me at all if imperfect hip joints were pretty common in nature. Severe imperfections that left youngsters unable to run would be culled by natural selection. But I'd guess a hypothetical wolf with hip scores of 6:6 would be at no evolutionary disadvantage. It may be that the 0:0 hip score is actually a bit of a freak, and overemphasis on perfect scores is causing some breeders to make sacrifices elsewhere to be able to advertize excellent hip scores.
    6. There is reason to suspect that bad hips are more serious in large, heavy dogs whose hips bare a lot of weight. It may be that there is little point in looking for perfect hips in dogs weighing 10 kg or less, and hip conformation in the 10-20 kg class is only mildly important.

    In my experience with Labradors, I find most breeders are strict in avoiding sires and dams with bad hip scores. It's clear that the inheritance pattern is complex: I have seen pups with awful hips come out of a sire and dam with excellent scores.

    Btw. JH, 2000-2014 is not 'over twenty years'. I'd say a drop of a point or two in 14 years is pretty impressive if it's not some statistical artifact.

    1. Should have been "almost 20 years" (have corrected)- these are rolling means and as it explains the 2000 data covers from 1996.

  7. p.s. The Finnish data are probably more significant as reporting is mandatory, hence not likely to be biased by people withholding data on dogs that had bad scores. At least in Finland, it looks like problem dogs/bitches are being culled from the breeding pool and the outcome is gradual improvement. Given that hips do not breed true, you cannot ask for more than gradual improvement.

    1. Moreover, for many breeds (whose breed clubs have approved this measure) it is a requirement that both parents have been tested and the radiographs evaluated by Finnish KC in order to register the puppies resulting from the pairing.

  8. I think you'll find that environment plays a roll in dogs (breeds) that are genetically predisposed (carry the genes) for HD but in dogs that don't carry the genes the environment has little effect on if they will develop HD. Compare racing bred Greyhounds to Labs; this comparison has been recognized by researchers which is why Cornell bred a colony of crosses of these two breeds for the HD studies.

  9. One of the vets I worked with in Australia was a GSD breeder before she went to vet school. She did her honors thesis on HD . . . I'd guess it was in the mid 80s. She said that really awful hips, eg. scores above 30, used to be quite common in GSD's and Rottis. They still pop up . . . but not that often.

    I wish we had longer term stats.

    re. PipedreamFarm. Why does everyone pick on Labs as the basket case for HD. All the stats I've seen show Labs as 'middle-of-the-pack' or below for HD?

  10. Don't think about it as being picked on, think about it as having more studies on your breed. You don't need to extrapolate results from another breed to yours.

  11. PennHip doesn't provide a publicly-available database, whereas OFA does--so I am not sure how we know that PennHip is better?

    1. The published research I posted above indicates that PennHip is not "better" (similar accuracy) than the other radiological examination methods.

  12. Let me first say, I hate the deliberate breeding of animals, for many reasons. But I also want to mention several other causes of deformities, besides breeding. Author Weston Price wrote a book called 'Nutrition and Physical Degeneration' (you can read the book on Project Gutenberg Australia), which talks about deformities in humans caused by malnutrition during and before their time in the womb, and early in life. It also shows photos of similar deformities in dogs, pigs, and other animals caused by malnutrition of the mother during her pregnancy. Another possible cause of deformities in dogs is spaying and neutering, especially at an early age before their bones form. They need their sex hormones to properly develop their bones, and hip dysplasia in particular is strongly associated with spaying and neutering animals early in life. Universal spaying and neutering doesn't happen in all countries, and they have fewer problems with dysplasia in the countries where they don't routinely spay and neuter animals. Hip dysplasia even happens to non-breed mutts who shouldn't have any inbreeding, and it's caused by either the prenatal malnutrition, or the sterilization at an early age. Again, I'm opposed to breeding in general, but it helps to know that some other factors can also cause deformities.

    1. If you "hate deliberate breeding of animals, for many reasons", does this mean you do not use any animal products because all farm raised animals were produced by deliberate breeding.

    2. Animal products can come from natural (non farmed) sources and there are a lot of people who are vegan and don't use any animal products at all.

      I'm in partial agreement with eagledove - I dislike purebred animals of any species except those which are bred for a very specific purpose which cannot be achieved without very selective breeding, even still I think all studbooks should be open to allow crossbreeding and new genetic information into each breed.

      I also very much disagree with the practice of intensive farming - I think we should ban battery farming entirely and force all farms to go free range. This would drastically reduce the amount of meat (and milk and eggs etc) available to us so every human would have to eat far less meat than they do (I. E. Not every day of the week). Which would not only be good for the farm animals, but would improve human nutrition and health too.

      We also need to work on reducing world population of humans too - and rewilding a lot of developed and industrial land to provide habitat for all the endangered species of the world (which cannot be helped until more habitat becomes available).

      But I have very low hopes of humanity doing any of this in my lifetime because of attitudes (that persist in USA especially) of being able to do anything one pleases at the cost of anyone and everyone else, and the idea that any compromise is an infringement of some "right" or privilege of being human (no such thing exists in any meaningful sense).

      So I'm expecting some kind of human induced apocalypse to occur which will obliterate most of the world's population of humans, and all because people are mostly too short sighted to see it coming.

      At Least I'll be long dead by that point.