Anatomy of a meta-analysis: a critical review of "exercise echocardiography or exercise SPECT imaging? A meta-analysis of diagnostic test performance".

BACKGROUND Accurate diagnosis of coronary heart disease has the potential to contribute substantially to cost-effective delivery of health services. Recent work by Fleischmann et al (JAMA 1998;280:913-20) represents an effort to summarize the accuracy of exercise echocardiography and exercise single photon emission computed tomography (SPECT). METHODS AND RESULTS A critique of the previous work was constructed, obtaining the 44 articles used. These articles were reviewed and summarized with established techniques for meta-analysis. The studies summarized by Fleischmann et al were found to be significantly heterogeneous (echocardiography and SPECT, both P<.001). In the SPECT cohort, combination of different radioisotopes and reading techniques, and inclusion of reports using experimental techniques, were sources of heterogeneity. In the echocardiography cohort, experimental techniques and an individual series were identified. When the sample was stratified for sources of heterogeneity, it was found that there was no significant difference in diagnostic accuracy between the echocardiography and SPECT techniques used in current clinical practice. Meta-regression with summary receiver operating characteristic curve techniques, after adjustment of the model for multicolinearity and outliers, revealed that there were no significant differences between SPECT as used in current clinical practice and echocardiography. CONCLUSION The report by Fleischmann et al contains serious flaws that limit its validity and generalizability.