Monday, April 14, 2014

A Study of Common Veterinary Billing Line Items

A top health concern for all dogs is obesity.
A recent study published in PlosOne and entitled “Prevalence of Disorders Recorded in Dogs Attending Primary-Care Veterinary Practices in England,” tells us a great deal and not much.

If this was a survey of cars, it would be a survey of ALL cars.

And apparently, when you look at maintenance issues of all cars, you find that they are all very safe, and only rarely burst into flame on their own volition.

Among the most common problems in cars are tears in seat fabric, bent radio antennae, and broken latches on glove boxes.

The biggest serious problems occur when diesel fuel is put in lead-free engines, and when engine belts are not maintained, resulting in overheating and/or overworked alternators.

Right. But what about all those recalled Toyotas that suddenly accelerated?

Well, you have to put that in context. Toyota has made a number of really excellent cars, and we really must look at that one problem within the context of all Toyotas.

Yes. But what about the recall?

The recall? We have nothing to say about the recall. It is not a significant data point when we look at all cars, and it is an outlier even with Toyotas.

Right. But Toyota was manufacturing defective cars, wasn’t it? And they knew they were manufacturing defective cars. And instead of redesigning and recalling those cars on their own, they carried on manufacturing them until they get caught, had to do a massive recall, and had to pay a $1.2 billion settlement to end the government probe. Right?

Well yes, but statistically, when we look at ALL cars, Toyota is really not worse than the others, and most cars are really quite safe.

You mean, from 30,000 feet, all the cars look just like ants crawling in a line on the freeway, and from that height, you cannot really see what particular model of car is too dangerous to be on the road?

Yes. Exactly. But here’s what out research does show: Most cars have small scratches, many have very small dings on their bumpers or fenders, and a surprising number have sticky vestigial soda spills in their drink cup holders.

----- * * * * ----- 

Now, if this sounds like I am making fun of this study, be advised that I am.

It is a massive look at massive numbers of dogs by someone who simply failed to know what to look for or ask.

As a study of common line-item billing concerns for a general veterinary practice in the U.K., it is a smashing success. But was that the goal of this study? Who knows! The authors themselves seem to have simply decided to collect a lot of data and load it into a statistical analysis software in order to run standard deviations based on whatever was commonly coded.

As a result, we learn that a lot of dogs have otitis externa (a minor ear problem treatable with over-the-counter drops for mites), periodontal disease, impacted anal sacs, and overgrown toe nails. A lot of dogs are obese, some get traumatic injuries, others get lacerations, dog bites, or split toe nails.

Seriously. This is the kind of “deep think” that went into the operationalization of the collected data.

Despite the tremendous numbers of dogs looked at, the data set for any one breed and the health issues for that breed, are really quite small.

For example, everyone who actually owns and is involved  with Jack Russell Terriers can tell you that this is a breed that has problems with deafness and eye disease, but from 30,000 feet a Jack Russell’s genetic problems are a never-mind; washed out by the white noise created by dogs fed too much, exercised too little, or which are very old.

And is a Jack Russell or a Border Collie even a "pure breed"? Really?

Well, I guess it depends on who you ask.

The man in the hedge or on the hill is not looking at a scrap of paper when he decides the value of  a dog. Oh sure, he may know the dog's breeding back five or six generations, but a pure breed? Well, NO. Not if you are defining a pure breed as a Kennel Club dog.

The border collie man or the terrier man is not going to to the Kennel Club for his dogs, is he? He is looking for a TYPE. That type has a name, and it almost certainly has a pedigree, but the paper is simply a window into the working history and perhaps the size and the mind of the dog that might grow out of a jumble of puppies in a basket. It is understood that when it comes to health, the hedge, the hill, Mother Nature and Father Time will judge them. But the Kennel Club? No, probably not. Real Jack Russells and real Border Collies rarely come with Kennel Club papers.


Peter Apps said...

From the abstract "Among the twenty most-frequently recorded disorders, purebred dogs had a significantly higher prevalence compared with crossbreds for three: otitis externa (P = 0.001), obesity (P = 0.006) and skin mass lesion (P = 0.033)"

This is hardly new news, but it supports the campaign against excessive inbreeding in "purebred" dogs, and due to the size of the study it cannot be dismissed by the line breeders as an artifact of small sample size.

PBurns said...

Yes, there is no question they found a few things and then buried them in the typing.

If you read slowly you find:

** "Purebred dogs showed a significantly higher prevalence than crossbreds for four of the twenty most-prevalent mid-level disorders: dermatological (P = 0.004), aural (P = 0.001), ophthalmological (P = 0.032) and obesity (P = 0.009) (Table 4). Statistically significant differences in prevalence values were shown between the most popular breeds in eight of the twenty most-frequent mid-level disorders: musculoskeletal (P = 0.002), claw/nail (P = 0.008), dental (P = 0.007), neoplastic (P = 0.001), anal sac (P = 0.006), obesity (P = 0.004), cardiac (P = 0.005) and brain (P = 0.003) (Table 5)."

** "Purebreds had significantly higher prevalence values than crossbreds for two of fifteen organ systems, namely integument (P = 0.001) and auditory (P = 0.002) (Table 6). The most prevalent pathophysiologic processes recorded were inflammation (1,246, 32.1%, 95% CI: 29.8–34.3), mass/swelling (625, 16.1%, 95% CI: 14.6–17.6) and traumatic (557, 14.3%, 95% CI: 12.8–15.9). Purebreds had significantly higher prevalence values than crossbreds for two of twenty-one pathophysiological processes: inflammatory (P = 0.006) and nutritional (P = 0.0014) disorders (Table 7). Statistically significant differences in prevalence values between the most popular breeds were shown for 5/8 body location terms, 5/15 organ system terms and 5/21 pathophysiologic processes (Tables 8, 9 &10)."

** "Purebreds showed significantly higher prevalence values for 13 of the 84 (15.5%) disorders and syndromes evaluated. No instances were identified in which prevalence values were significantly higher in crossbred than in purebred dogs."

But is the study mostly typing?


If you compare apples and eggs (cracked toe nails and cancer), do not operationalize your definitions (what is a breed?), and sort nothing by cohort, then you do not have much in the end. The KC breed health surveys, with their obvious breed bias, are actually more illuminating.

Peter Apps said...

Hi Patrick

Your concerns about important findings being buried in and obscured by sophisticated (in the pejorative sense) statistical processing are completely valid. Sadly this problem is not restricted to studies of dog illness, it crops up all over the biological sciences and can be crudely characterized as the substitution of indiscriminate number crunching for thinking.