Use of different outcome measures in randomised studies of malignant glioma can significantly alter the interpretation of time to progression: Reanalysis of the MRC BR2 study
J. Chataway et al., Use of different outcome measures in randomised studies of malignant glioma can significantly alter the interpretation of time to progression: Reanalysis of the MRC BR2 study, J NEURO-ONC, 43(1), 1999, pp. 87-92
The Medical Research Council (MRC) BR2 study [1] is a randomised trial of t
wo doses of cranial radiation for patients with malignant glioma. We reanal
ysed data to examine the effect of using change in ranked scales of neurolo
gical status (MRC Neurological Status Scale) and performance (World Health
Organisation Scale: WHO) to determine progression rather than clinician's i
mpression.
Four hundred and seventy four patients were studied. Where clinicians recor
ded no progression, ranked scales frequently documented progression (MRC 13
%; WHO 13%). Where clinicians recorded progression, ranked scales frequentl
y did not alter (MRC 33%; WHO 30%) or occasionally improved (MRC 5%; WHO 3%
). When analysing time to progression based on a variety of measures, the e
stimated difference between treatments was most extreme (hazard ratio 0.81,
logrank p = 0.04) when change in WHO status was used, and least extreme wh
en change in MRC neurological status was used (hazard ratio 0.99, p = 0.94)
.
This study highlights how different outcome measures can significantly alte
r the interpretation of randomised studies.