In message <[log in to unmask]>, Declan Fox
<[log in to unmask]> writes
> I think that doing exercise
>ECGs on asymptomatic 40 somethings (esp with normal resting ECG) is gross
>misconduct, there is no evidence that they are of any value and since they
>are not perfect tests they have the potential for harm.
Too right! I had fun with these figures when I worked them out:
(assuming asymptomatic patient):
If prevalence of coronary artery stenosis is 1% (Sackett et al cite this
for women aged 30-39) and sensitivity=69% and specificity=88% for
exercise ECG 1mm S-T depression (Martin TW et al, Clin Card
1992;15:641-6) then you get these figures:
True positive 0.7%, true negative 87.1%, false negative 0.3% but false
positive is a wopping 11.9%.
If prevalence is 5% (Sackett et al cite this for men aged 30-39 but
seems a bit high to me) then the results are not much better:
TP 3.4%, TN 83.6%, FN 1.6%, FP 11.4%
and at 10% prevalence:
TP 6.9%, TN 79.2%, FN 3.1%, FP 10.8% - so you're still almost twice as
likely to get a false positive as a true positive!
It's only when the prevalence reaches 40% to 50% that exercise ECG
starts to look sensible. At 50% prevalence:
TP 34.5%, TN 44.0%, FN 15.5% and FP 6% (but note the increase in false
negatives as prevalence increases - at higher prevalence it makes the
test as dodgy as at low prevalence, but the other way round)
Toby
--
Toby Lipman 7, Collingwood Terrace, Jesmond, Newcastle upon Tyne. Tel
0191-2811060 (home), 0191-2437000 (surgery)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|