Lung cancer histologic type in the surveillance, epidemiology, and end results registry versus independent review

Abstract

Because few studies have assessed the accuracy of lung cancer histologic diagnoses reported by state cancer registries, we examined whether the Iowa Surveillance, Epidemiology, and End Results Cancer Registry (i.e., the Iowa Cancer Registry)-reported lung cancer histologic diagnoses were reliable. We investigated agreement between lung cancer histologic types reported for 413 patients with lung cancer by the Iowa Cancer Registry and those obtained through an independent review of diagnostic slides. Among lung cancer histologic types, small-cell carcinoma had the highest sensitivity (94.1%, 95% confidence interval [CI] = 85.6% to 98.4%), positive predictive value (94.1%, 95% CI = 85.6% to 98.4%), negative predictive value (98.8%, 95% CI = 96.9% to 99.7%), and highest percent exact agreement (98.0%, 95% CI = 96.6% to 99.4%). The lowest sensitivity (21.9%, 95% CI = 9.3% to 40.0%) and positive predictive value (23.3%, 95% CI = 9.9% to 42.3%) were noted for large-cell carcinoma, probably because other more specific features of adenocarcinoma or squamous carcinoma were absent. Adenocarcinoma had the lowest specificity (84.4%, 95% CI = 79.0% to 88.9%), negative predictive value (85.2%, 95% CI = 79.9% to 89.6%), and percent exact agreement (82.9%, 95% CI = 79.2% to 86.6%). Samples collected by cytologic examination (odds ratio [OR] = 2.4, 95% CI = 1.1 to 5.2) or biopsy examination (OR = 2.2, 95% CI = 1.1 to 4.2) were more likely to be misclassified than samples obtained via resection. Thus, the histologic type obtained by the Iowa Cancer Registry is reasonably reliable, but independent slide review is needed for precise histologic typing of lung cancer.