Hello,
I am currently in the process of analysing data from a
comparative assessment of a medical device in patients.
Population:
20 patients (no sample size calculations)
Study:
Patients were monitored for 2-3 hours using two medical device.
Parameters monitored were heart rate (HR), respiration rate(RR),
body temperature (BT) & BP (systolic & diastolic).
(A nurse assessment was also made for HR & RR)
Four assessments on each of the parameters were made on each
subject
within the 2-3 hour period.
In analyzing this data I have been referring to the article by
Bland & Altman - 'Statistical Methods For Assessing Agreement
Between Two Methods of Clinical Measurement' (The Lancet). The
study under consideration has more than two repeated measurements
on each subject and three methods for measuring HR & RR and
therefore need to use analysis of variance. However the paper
doesn't go into the analysis.
Am I correct in assuming that a one-factor repeated measures
analysis is appropriate? Where the blocks are Subjects,
treatments are Devices
and repeated measurement is assessments. For each of the
variables I have looked
at the distributional properties and decided upon a parametric or
non-parametric approach.
I would be grateful if anyone knows of other articles dealing
with assessing agreement between methods of clinical measurements
within the field of medical devices or any suggestions on how to
analyse the study outlined above.
Regards
Prafull
*****************************************
Prafull Mistry
Medical Statistician
Nexan Telemed Ltd
Tel: (01223) 713500
Fax: (01223) 713501
e-mail: [log in to unmask]
*****************************************
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|