Stephen J.LurieMD, PhD, Senior EditorIndividualAuthorJody W.ZylkeMD, Contributing EditorIndividualAuthor
Copyright 2001 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2001
In Reply: Dr Smith-Bindman notes that the studies in our meta-analysis had a high prevalence of malignancy, which raises the possibility that PET imaging may not be as accurate in populations with a lower prevalence of malignancy. As she notes, this is a concern only if the higher prevalence of disease in the studies is correlated with a more severe spectrum of disease. Studies will overestimate sensitivity if the disease is more severe in the study population than in the clinically relevant population.1 To assess whether this problem existed in the literature on PET imaging, we compared studies with a very high prevalence (mean, 80%) with those with a lower prevalence (mean, 64%) and found that PET imaging was more accurate in studies with a lower prevalence of malignancy (log odds ratio [OR], 5.19; 95% confidence interval [CI], 4.50-5.88) than those with the higher prevalence (log OR, 3.91; 95% CI, 3.34-4.58; P<.001 for difference). Thus, the available data do not support a correlation between prevalence and disease spectrum. Nevertheless, we agree that it is important to study a population with a full spectrum of disease severity.
Gould MK, Owens DK. Positron Emission Tomography to Evaluate Lung Lesions—Reply. JAMA. 2001;285(21):2711-2712. doi:10.1001/jama.285.21.2711