Dear Allstat,
I am looking at the diagnostic value of certain physiotherapy tests for
detecting a particular condition. I would like to calculate the
sensitivity, specificity and positive and negative likelihood ratios
(along with their 95% confidence intervals) for each test but I am
dealing with clustered data. Each person contributes two observations -
one from their symptomatic hip and the other from their asymptomatic
hip.
(1) Does anyone know of a straightforward way of incorporating the
clustering into the calculation of the variances?
(2) Is it correct to say that if I ignore the clustering and just
calculate a standard 95% confidence interval this would be conservative
since the within-cluster correlation coefficient should be negative? Is
this recommended?
I'll summarise to the list, if appropriate.
Thank you,
Vicki
Dr Vicki Livingstone,
Lecturer,
Dept. of Preventive and Social Medicine,
University of Otago Medical School,
P O Box 913,
Dunedin,
New Zealand.
Phone: + 64 3 479-5138
Fax: + 64 3 479-7298
email: [log in to unmask]
|