The Great TB Debate

Re: The Great TB Debate

Hannah Wynne Richards — Tue, 23 Oct 2012 11:58:53 GMT

Bad news because although I'm far from convinced badgers are the whole problem I think they're definitely part of the problem-and without a cull there'll be no incentive to look at what else is contributing

Wynne

Re: The Great TB Debate

plantagenet — Tue, 23 Oct 2012 11:08:00 GMT

They've just pulled the cull!

Re: The Great TB Debate

Neil Wheadon — Sat, 20 Oct 2012 16:18:50 GMT

Sorry, the formula is cut in half

Specificity = number of true negatives

________________________________________

Number of true negatives + number of false positives

Re: The Great TB Debate

Neil Wheadon — Sat, 20 Oct 2012 16:11:57 GMT

The definition of specificity is

Now I'm no statistician, but I have already stated that the problem with the TB skin test is that it misses the 'anergic' cow. This is the cow that is so full of TB, that it's immune system can't be bothered to react to 0.1ml of tuberculin injected under the skin.

So look at the formula above. The skin test shows almost no false positives, therefore the specificity is very close to 100%, which is the 99.9% figure that other organisations also quote.

There were 21512 cattle culled in the first half of this year see http://www.defra.gov.uk/statistics/foodfarm/landuselivestock/cattletb/national/

Therefore if we used

1) The stat I present and backed by other sources, this means that of these (using 99.9%) 21 were false positives

2) The stat the BVA present being 90% then 2152 of these cattle were false positives - that's a lot of cows.

This 99% statistic is further backed up by the document that DEFRA produce that answers questions relating to TB, see

http://archive.defra.gov.uk/foodfarm/farmanimal/diseases/atoz/tb/documents/tb-facts.pdf

Question 8 relates to the gamma interferon test (blood test that was used in 4 yearly areas in 1999) but is relevant to this discussion

Q8. Does the gamma interferon test give a large percentage of false positives?

Background: Scientific research has shown that the average specificity (accurate identification of uninfected animals) of the gamma interferon test is 97% - which is only slightly lower than the 99% plus for the skin test. Performance evaluation carried out in a number of countries shows that at the laboratory cut-offs used in GB the gamma interferon test has a sensitivity comparable to or marginally better than the skin test – between 73 and 100%, with a median value of about 87%. Scientific research has also shown that the two tests (gamma interferon and skin test) identify different populations of infected cattle. The gamma interferon test can identify infected animals at an earlier stage in the infection as well as infected cattle that simply fail to react to the skin test. An animal that reacts positively to the gamma interferon test and negatively to the skin test will not, in the vast majority of cases, be a false positive.

Answer 8. No. The risk of the gamma interferon test identifying a false positive animal is 3 in 100, this risk is further reduced when the test is applied in a herd known to be TB infected. It is a common misconception that, as 82% of gamma interferon test positive animals do not show post-mortem evidence of TB in the slaughter house or laboratory, they were “false positives”. A failure to find post-mortem evidence of disease does not mean that the animal in question was free of infection.

So yes Luke is right in both respects and there are other stastical factors at work, notably doing the test correctly. (As was discussed a few months ago on this forum)

Neil

Re: The Great TB Debate

Luke Edwards — Sat, 20 Oct 2012 15:25:31 GMT

I think the point I am trying to make - before I get way out of my depth - is that

1. sensitivity and specficity are intrinisic to the test and do not change

2. There are many other statistical factors which dictate how well the test performs in the field - its not just down to senisivity/specficity.

Luke.

Re: The Great TB Debate

Luke Edwards — Sat, 20 Oct 2012 15:09:01 GMT

Dear Mr Wheadon,

You are quite correct - it should read "however I (politely) still think you are wrong about the specificity "

Thus proving, yet again, that specificity is a harder concept to get your head around than sensitivity - at least for me.

Luke.

Re: The Great TB Debate

Malcolm Ness — Sat, 20 Oct 2012 14:03:35 GMT

[quote user="Neil Wheadon"]Can you have another go at explaining this please[/quote]

If you Google "Specificity" then go to the Wikipaedia definition all will be explained.

Re: The Great TB Debate

Neil Wheadon — Sat, 20 Oct 2012 13:36:40 GMT

Hi Luke

[quote user="Luke Edwards"] I think you are correct about sensitivity the test picking up 80/100 disease ridden animals. However I (politely) still think think you are wrong about the sensitivity.[/quote]

Can you have another go at explaining this please.It looks like you're mixing up the sensitivity and specificities (see above) so your point is lost. (And I don't mean your understanding off them ) You're probably making a valid point but I can't quite see it here.

Neil

Re: The Great TB Debate

Luke Edwards — Sat, 20 Oct 2012 13:15:36 GMT

Dear Mr Wheadon,

I think you are correct about sensitivity the test picking up 80/100 disease ridden animals. However I (politely) still think think you are wrong about the sensitivity. A sensitivity 99.9 % means one false postive in 1000. ie one cow in 1000 tests postive but is actually disase free. It has nothing to do with how likely it is that a cow with a lump actually has the disease. For that you need the postive predictive value (PPV) and negative predictve values (NPV).

Sensitivity and Specificty are inherent features of the test -they do not change with changing circumstances. However PPV ands NPV do change especially with changes in disease prevalence rate.

Imagine you have a test with a very high sensitivity (ie picks up all the affected animals) and a rate of false positives= 1 in 100 i(ie 1 in 100 animals testing positive is actually disease free) . Now imagine the population also has a true disease prevalence of 1 in 100. You can see that in this test the rate of true postives = the rate of false postives. In other words when you find a lump in this population it is 50:50 whether the animal actually has the disease or not. Might as well flip a coin.

This might explain why different groups get different results - different rate of disease.

If it is any consolation, then you will note I had to go back in and edit my original post after posting when I realised I had written it wrong - thus elegantly proving my own point that specificity is a harder concept to get your head around than sensivity.

Luke

Re: The Great TB Debate

Neil Wheadon — Sat, 20 Oct 2012 10:59:13 GMT

[quote user="Hannah Wynne Richards"]subsequently have no sign of disease at PM? Are they newly infected, or is there another reason?[/quote]

Yes

Think about the process here. When a cow goes to slaughter, you are looking for an abscess in the lymph nodes and these need to be visible. Also think about the looking at the lymph node process. It's big and to find a single abscess, it's like slicing a Christmas pudding to find the sixpence. In certain cases the lymph node is meticulously sliced much like a supermarket ham slicer, then you can definitively say that it is NVL (Sorry No Visible Lesion)

After this process you grind up a lymph node and try to grow a slow growing bacteria on a plate, is it any surprise that in some cases you don't find it?

It takes time to develop Visible Lesions, it's the nature of the infection. Testing as often as we do, is it any surprise that they come back negative?

I hope that helps

Neil

Re: The Great TB Debate

Neil Wheadon — Sat, 20 Oct 2012 10:33:54 GMT

Dear Luke

The way I understand it is. Imagine every cow on the farm has TB You inject 100 cows and you get a positive reaction in 80. As the specificity of the test is 99.9 per cent, you can justify to a farmer that these cattle have had exposure to M Bovis. However as the sensitivity is 80 per cent you know that statistically there are 20 cows in the herd that are positive but weren't picked up. This herd will now be under restriction and you have 2 more chances of finding this hidden 20 per cent, so using these statistics you test the 20 cows and you'll pick up 16 out of the 20. At the next test of the remaining 4 you should get the lot.

This analogy though seemingly ridiculous is an illustrative example but I have seen this on some farms in specific groups. I can vividly recall farms with young stock (it's always cattle aged under 3) where a group of cattle were affected in this way and from a group of about 25 cattle you we left with 1 or none at the end of testing. This was in parallel to the rest of the herd testing clear. Why? Because they were geographically in a separate place.

So returning to the majority case. The question of this missing 20 per cent that have had exposure what happens at the next test? You will find most of them, but there will remain the 'anergic' cow. The one so riddled with TB and potentially an open case that won't react to the TB test. Fortunately these are usually older cattle, and these won't be sold and moved about the country, they will appear as a slaughter house case when finally sent for slaughter. However and this is why I strongly believe that TB testing is a veterinary job, as vets we need to be aware that she could still be in he herd infecting others. A few years ago after a chat with a farmer in a dreadful outbreak we worked out that there may be an open case. 5 cows were presented and 2 sent for slaughter and the clinical case was there. So to conclude this sensitivity is an issue and one that justifiably Is seized on. However in my experience young stock have a higher sensitivity than older cattle and these are the ones likely to be moved about the country. If anyone has statistics to back this up, please say.

Neil

Re: The Great TB Debate

Hannah Wynne Richards — Sat, 20 Oct 2012 10:06:16 GMT

Neil Why are my farming friends forever complaining that cows have been removed as reactors, and subsequently have no sign of disease at PM? Are they newly infected, or is there another reason?

Wynne

Re: The Great TB Debate

Luke Edwards — Sat, 20 Oct 2012 00:59:57 GMT

Mr Wheadon,

I might be wrong but doesn't 99.9% specificity refer to negative results? ie the abscence of a lump in a correctly tested cow excludes the cow from having TB 99.9% of the time. 1 in 1000 will be a false positive (test positive, disease negative)

If you get a lump (postive test) and want to know whether to belive the result you need the positive predictive value ... or something like that.

Luke

Re: The Great TB Debate

Neil Wheadon — Fri, 19 Oct 2012 23:03:12 GMT

It's late but I've watched the first 10 minutes

First proper statistic

Specificity and sensitivity is just well how can I put it ......Wrong

The presentation quotes 90% sensitivity and 90% specificity. These are important numbers

In fact the TB skin test is 80% sensitive and 99.9% specific. This means that when you test a cow, if you get a skin lump, there is a 99.9% chance that this test is detecting M Bovis. The trouble is that out of 5 reactors you're only finding 4 of them at each test, that's the sensitivity

See below

ref http://www.bovinetb.info/testing.php

The risk of false negatives and positives when testing cattle

In the extract below quoted specificities imply that in Great Britain the blood test is on average about 40 times more likely than the skin test to indicate that a healthy animal is infected. Average specificity of the skin test is quoted to be 99.9% (1 in 1000 wrongly classified as a reactor) whereas that of the blood test is quoted to be 96% (4 in 100 wrongly classified as a reactor).

The badger trust too, also quote this fact see http://www.badger.org.uk/_Attachments/Resources/15_S4.pdf

“The figures that the hon. Gentleman gives the House are not quite correct. The skin test’s sensitivity, which is its identification of infected animals, is between 77 and 95 per cent. Its specificity, which is its identification of uninfected animals, is 99 per cent. The test is much better than he suggests,

These figures were in the AHVLA figures that I frequently used with farmers

I might watch the rest, but if the first stat is so far out, what about the rest?.

Neil

Re: The Great TB Debate

J G Wray — Fri, 19 Oct 2012 22:24:49 GMT

[quote user="Neil Wheadon"]I spoke to a practioner that had seen the webinar and we are pretty much saying the same thing, so perhaps you'd like to justify this broad brush statement or do you have anything factual to add?[/quote]

Then it must have been a different presentation to the one I took part in the other Monday evening. I note you're relying on hearsay.

Personally, I believe the science as presented by the Sainted Padgett on behalf of BVA, a development of the BCVA Roadshow stuff of yesteryear, Andy Biggs et al - we all remember that don't we?

I don't like my subscription being used this way, but it was a free hour of CPD and a rational overview. It'll be available again to members at some point. Correction -ah it already is http://www.bva.co.uk/vetstv/2952.aspx

JGW

Re: The Great TB Debate

Neil Wheadon — Fri, 19 Oct 2012 15:09:23 GMT

[quote user="J G Wray"]Who's right d'ya suppose?[/quote]

All I can do is present the eveidence as I saw it, and this was backed up by actual references so that if you are interested, maybe you can read them

I spoke to a practioner that had seen the webinar and we are pretty much saying the same thing, so perhaps you'd like to justify this broad brush statement or do you have anything factual to add?

Neil

Re: The Great TB Debate

Tim Cheyne — Fri, 19 Oct 2012 13:50:27 GMT

It seems that opposition and doubt is resurfacing. These very reently from the BBC website.

http://www.bbc.co.uk/news/uk-19939393

http://www.bbc.co.uk/news/science-environment-19903430

http://www.bbc.co.uk/news/science-environment-19981171

Tim.

Re: The Great TB Debate

plantagenet — Fri, 19 Oct 2012 08:27:12 GMT

Is it in the RSPCA's charitable remit to threaten farmers if they take part in the trial cull?

Re: The Great TB Debate

J G Wray — Sat, 13 Oct 2012 14:15:39 GMT

Hmmm, I've been following this and thinking how far from what Mr Padgett presented in his webinar the discussion has been.

Who's right d'ya suppose?

JGW

Re: The Great TB Debate

Hannah Wynne Richards — Sat, 13 Oct 2012 09:43:37 GMT

Lets wait and see As I've said before I'll be delighted to be proved wrong on this one-and badgers are certainly part of the problem

Wynne

PS Did the yoghurt business use sewage sludge?

Re: The Great TB Debate

Neil Wheadon — Fri, 12 Oct 2012 17:44:00 GMT

One large farm with an associated yoghurt business had TB for years. They spent £60.000 on a proper wildlife proof fence all the way round.

No TB since

Re: The Great TB Debate

Neil Wheadon — Fri, 12 Oct 2012 17:41:31 GMT

Because

1) Not all the badgers were removed - see statistics in the Wilsmore article

2) The effect was measured after the cull had taken place. As these rings had a 3km radius and Mr Badger move at 3km a night then the setts that had been cleared within the ring will have been gradually refilled.

In fact considering the statements above only lends weight to a 28% reduction being quite remarkable.

It also shows why the proposals for these areas to be used now for the cull were 1) Large 2) Had well defined geographical barriers all the way round. I'm not sure of the boundaries in the 2 areas approved but large waterways, motorways (these are fenced) etc. The point that once areas were cleared re population will be difficult, so adding weight to any finding (it will also prevent pertubation)

In one area within an endemic area there was no TB for years, reason? Bristol Channel to the north, Large Rhymes (Somerset waterways) to the south and rivers east and west.

Hope that explains that one

Neil

Re: The Great TB Debate

Hannah Wynne Richards — Fri, 12 Oct 2012 17:22:25 GMT

Neil 28% reduction is a vast improvement-but nowhere near the almost total eradication we once achieved Why?

Wynne

Re: The Great TB Debate

Tim Cheyne — Fri, 12 Oct 2012 13:01:48 GMT

[quote user="Neil Wheadon"]I hope that helps and 5 stars please Arlo!![/quote]

Well, I cannot speak for Arlo but I certainly think that you deserve the stars for some masterly explanations.

Tim.

Re: The Great TB Debate

Hannah Wynne Richards — Fri, 12 Oct 2012 11:58:20 GMT

This is one where I'll be absolutely delighted to be proven wrong if badger culling (together with cattle testing and removal ) gets levels down to that we once took for granted

Wynne