Margaret A.WinkerMD, Deputy EditorIndividualAuthorPhil B.FontanarosaMD, Interim CoeditorIndividualAuthor
To the Editor: Dr Fleischmann and colleagues1 provided a detailed summary of 2 commonly used
noninvasive cardiac test modalities: exercise ECHO and exercise SPECT. We
are concerned about the heterogeneity of the studies included in their meta-analysis.
Midgette et al2 state that when data
from a number of reports are poorly correlated and heterogeneous, neither
summary point estimates or summary receiver operating characteristic (SROC)
curves are appropriate. Petitti3 has also
reported that use of the fixed-effects model to create summary point estimates
requires homogenous data. A test for homogeneity in the reports cited by Fleischmann
et al reveals significant heterogeneity in both the exercise SPECT and exercise
ECHO reports (χ2= 51.02 for SPECT, and 61.61 for ECHO, both P<.001). This leads us to question whether there might
be significant underlying differences in the component reports, creating bias
in the results.
Kymes S, Shaw L, Fletcher J. Exercise Echocardiography vs Exercise SPECT Testing. JAMA. 1999;282(17):1621-1623. doi:10-1001/pubs.JAMA-ISSN-0098-7484-282-17-jbk1103