Tuesday, August 15, 2006

Going Dutch. "A 2mm difference..."

A 2 mm difference in swelling, will mean 24 percent of incidents are missed or detected later"

We remarked on this 2mm in our post below 'It Ain't what you do ...", but have now had time to read the whole file. A "small difference in performance", the lady said. Define 'small' in the context of the above passage from the full statement, and redefine 'small' after reading the description of the reactions described in more precise epidemiological terms by her staff at VLA as a "statistically significant difference". And remember - just 2mm.

The Chief Veterinary Officer has issued a report (317 KB) on the recent reduction in the number of new TB incidents in Great Britain. See in particular p. 12 and 14 of the 23 page pdf.

Note: The slaughterhouse cases referred to in the paper are 'Confirmed', and have been added to CNI found at herd tests. The samples submitted by MHS for culture from slaughterhouse examinations of carcasses are up by a not insignificant 35 - 75 percent on last year. (Defra figures Jan - March +75%, Jan -May +50% and Jan - June +35%)

Also, we make the 'herds under restriction from a TB incident to total 4.55 percent, (up slightly on last year) and not 3.5 per cent quoted. Defra are probably using the lower Confirmed Incident figure.

That 2mm figure is also of significance in the new TB 64 Interpretation Chart revised by Defra in February (TB64 & TB64a Revised 02/06) and issued to LVI's conducting bTb intradermal testing, which we also mentioned in the posting below.

The interpretation sheet instructs:
"All skin swellings with an increase of more than 2mm in the thickness of the skin fold must be recorded as positive reactions, as should any swelling irrespective of size, showing oedema .

+ Positive - an increase of more than 2mm in skin thickness or any reaction with oedema.

Veterinary practitioners who have had the oportunity to study the new test interpretation chart, have remarked on a 'marked tightening up' of reactions which may be described as 'oedemous', with the result that even a 0 (zero measurable) rise on the bovine lump, but if showing oedema will classed as an Inconclusive on a standard interpretation, and a Reactor on severe interpretation of the test.

So what have we got here? An annex and graph towards the end of a long pdf file - which bears little relation to the CVO's press statement at the beginning. A new, improved 'Revised Interpretation Chart' issued at the peak of the 'drop' in cases, and coincidentally at the peak of the use of ID Lelystad tuberculin - which is not mentioned at all. And of course the continued mention of that 'small difference in performance' by the lady, which is a rather glib translation of the report's actual description of a "statistically significant difference" between ID Lelystad bovine antigen and the UK product. Of 2mm perhaps? Which when number crunched through Defra's computers arrives at the staggering conclusion, that "a 2mm difference in swelling will mean (not might, or maybe - WILL) 24 per cent of incidents are missed or detected at a later stage". And if the test is a routine one, that 'later stage' will be a year or more.

And as we said in our post below, it may be small to the CVO but that 2mm in a scale of 0-5mm for a Reactor on standard and 0-3mm on severe interpretation, with the extra proviso of 'any oedemous reaction' to be classed as positive even if not measurable, is life or death to some poor cow, and a pass or a fail for the herd - and yet another inconsistant in the convoluted life of the RBCT, the data for which, prior to its ritual torture in the hands of the statisticians, is gleaned from the numbers of herd breakdowns, as determined by the intradermal skin test...

And if we have joined the dots correctly, and we have no reason after reading this report, to think that we have not, the Lelystad bovine tuberculin antigen has missed almost a quarter of bTb cases. This up until February 2006, when the new 'TB64 / revised 02/06' interpretation chart made a belated attempt to tighten up the readings to cope with a 'statistically significant difference" in reaction. In which case, the drop in cases is not a drop in incidence of bTb at all, it is a drop in detection, and that is quite different.

No comments: