BSE-Fall in Österreich

June 21, 2005

Nun hat auch Österreich seinen ersten echten BSE-Fall.

Und es ist eine Causa, die es in sich hat. Das erkrankte
Tier stammt von einem Bergbauern ohne jegliche Verdachtsmomente. Es
scheint nicht klar warum das Tier erkrankte, die anderen fünf Tiere
wurden zur Untersuchung getötet. Ergebnisse dazu stehen noch aus.

Die Politiker wie Gesundheitsministerin Maria Rauch-Kallat oder Landwirtschaftsminister Josef
Pröll behaupten nun interessanterweise, dass die Erkennung dieses Falls
ein Beweis für das funktionieren der BSE-Testuntersuchungen in
Österreich ist. Das kann ich nicht nachvollziehen. Wenn ich also einen
positiven Fall positiv erkenne heißt das, dass mein Erkennungsystem
hunderprozentig korrekt arbeitet?

Jeder Test — also auch der BSE-Schnelltest — hat eine
Falsch-positiv-Rate (FpR) und eine Falsch-negativ-Rate (FnR), das heißt
eine bestimmte Anzahl von Tests sind positiv auch wenn die Probe negativ
war und vice versa. Nur wenn beide Raten gegen Null gehen, kann davon
ausgegangen werden, das alle Test das korrekte Ergebnis liefern, also
positiv sind für kranke Tiere und negativ für gesunde. Die Auswirkungen
einer hohen FpR sind im Grund rein finanziell, denn es werden gesunde
Tiere vernichtet. Eine hohe FnR jedoch kostet Menschenleben durch
Creutzfeld-Jakob.

Es ist also wichtig aus medizinischen wie finanziellen
Gründen die FnR so niedrig wie nur möglich zu halten ohne gleichzeitig
die FpR zu stark zu erhöhen. Wie es scheint sind diese Test bezüglich FpR “extrem empfindlich“:

But Clifford stated repeatedly that “it’s very likely this animal could be
negative”. He stressed that the BioRad test used was “designed to be
extremely sensitive” to catch any possibly infected animal, some of
which “will end up negative during further testing”.

Asked
the odds of the result being a false positive, Clifford said “we
wouldn’t want to provide that type of information”. But the rate of
false positives for the BioRad procedure was measured by the European
Commission when it evaluated BSE tests for use in the EU in 2003.

Those
data have never been published, but industry sources who have seen them
say the BioRad test had a false positive rate of about one in a
thousand initial tests, a rate borne out subsequently in practical
experience. BioRad is used in Germany and Belgium.

In
Europe and Japan, if a cow’s brain initially tests positive with
BioRad, it is tested twice more. Only if one of those repeated tests is
also positive is the sample sent for IHC confirmation. The false
positive rate after such repeated testing is even lower, around one in
100,000 for BioRad. It is not known whether the USDA is doing this.

However,
Markus Moser of the Swiss firm Prionics, which makes a rival ELISA test
for BSE, points out that if the true incidence of BSE in cattle is much
lower that the test’s false positive rate, most of the positive tests
obtained with the test will indeed be false.

In
Japan, where the BSE incidence is just one tenth of the BioRad false
positive rate, only eight of the first 113 positive BioRad tests were
confirmed.

It
may be that this is what USDA is counting on when it says its positive
test is “very likely” to turn out false. But no one knows if the true
incidence of BSE in the US is so low – that is what the testing
programme is supposed to measure.

Sensitivity is a measure of a test’s ability to detect truly infected
subjects. A test with a high sensitivity will have a high probability
of detecting truly infected subjects. For example, if a screening test
is used to detect BSE in 100 samples known to have BSE and identifies a
positive result in 99 of them, its sensitivity would be 99/100 x 100 =
99%. A test with 99% sensitivity would have a 99% chance of detecting
one infected sample, even if there was only one positive case in one
million negative cases.

Specificity is a measure of as a test’s ability to correctly classify
subjects as being uninfected in the absence of disease. A test with a
high specificity generates few false positive results. For example, if
a screening test is used to detect BSE in 100 samples known to be
uninfected with BSE and produced a negative result in 99 of them, its
specificity would be 99/100 x 100 = 99%. A test with 99% specificity
would be expected to produce false positive results 1% of the time.

Research shows that the probability of Bio-Rad TeSeE ELISA failing to
detect a single case of BSE in a large group of samples is in the range
of 1.1 – 3.6 per 1000 tests (p<
0.05) (sensitivity). Reports from Japan, where every single animal is
tested, show that the false positive rate for the Bio-Rad TeSeE ELISA
test is 1 in 30,000 (specificity). When either of these situations
happens, the test is known to be “inconclusive” and will require
further testing to truly identify whether the sample is positive or
negative.

Interessanter für unsere Gesundheit ist aber die FnR. Dazu fand ich nicht so besonders viele Quellen, ein Zitat von hier (PDF):

We have estimated the false negative rates for normal adult and normal juvenile animals
using a modified version of Harvard’s BSE simulation model. This modified version of the
model reports the characteristics of each BSE-positive animal that dies during the simulation.
Characteristics reported include the animal’s type (dairy, beef, beef reproductive), gender, age
(months), months since the animal was infected with BSE, fraction of the incubation period
elapsed at time of death, and death location (farm or slaughter facility). We assume that animals
with BSE test negative if less than 90% of their incubation period has elapsed. We simulated the
spread of BSE for 20 years following the introduction of contaminated feed (250 ID50s) into the
U.S.1 Our results indicate a false negative rate of 92% for normal adult cattle. For normal
juvenile cattle, the false negative rate is 99.99%. Accounting for these false negative rates and
the potential for BSE among normal juvenile animals suggests that the evaluating the surveillance
data as described here is a relatively insensitive approach for detecting the presence of BSE in the
U.S. cattle population.

Das klingt ja dann nicht mehr besonders gut. Es ist positiv, dass es
fast keine falsch-positiven Fälle von BSE gibt, aber was ist mit den
viel wichtigeren falsch-negativen Fällen?

Also solange wir nicht mehr über die Falsch-negativ-Rate in Österreich wissen, können wir auch nichts über das BSE-Erkennungssystem selbst aussagen, und damit schon gar nicht bei Bekanntwerden eins positiven Falles auf einen Beweis der Korrektheit der Erkennung aller positiven (und damit nur der positiven) Fälle von BSE schließen.

Posted in Austria