Patient Safety and Quality

Studies examine the source of diagnostic errors in thyroid and lung cancers

The diagnosis of thyroid cancer is usually based on the pathologist's evaluation of cell and tissue samples obtained through fine needle aspiration (FNA) and surgical excision of the thyroid gland. When the diagnoses from both samples do not agree, root cause analysis determines whether the source of error is sampling (diagnostic material was obtained in one but not the other sample) or interpretation (diagnostic material is present in both samples but misinterpreted in one sample). One-fourth of patients with thyroid cancer are misdiagnosed as not having cancer due to errors in specimen quality and misinterpretation of thyroid gland fine-needle aspiration samples, according to a new study. A second study found that pathologists at one hospital are more likely to agree with one another than those from different hospitals on the cause of diagnostic error in pulmonary specimens sent for lung cancer diagnosis. Both studies were supported by the Agency for Healthcare Research and Quality (HS13321) and led by Stephen S. Raab, M.D., of the University of Pittsburgh. They are discussed here.

Raab, S.S., Vrbin, C.M., Grzybicki, D.M., and others (June, 2006). "Errors in thyroid gland fine-needle aspiration." American Journal of Clinical Pathology 125, pp. 873-882.

The use of FNA to biopsy the thyroid gland has improved the care of patients with thyroid gland nodules. Yet, FNA has higher false-positive and false-negative error rates than other specimen types. This study revealed that over one-third of patients were incorrectly diagnosed. Twenty-five percent of patients who underwent FNA were diagnosed as cancer-free, when they had cancer (false-negative diagnosis). Nearly 10 percent of patients who did not have cancer were diagnosed with cancer (false-positive diagnosis).

Researchers performed more detailed analyses of thyroid gland aspirates obtained over a 2-year period. Surgical pathology followup was obtained in 364 patients. Based on examination of the original thyroid gland FNA reports, the diagnoses were reclassified independently by two study investigators as benign, atypical, adenoma, or malignant. The goal of the study was to determine whether FNA could accurately classify lesions into 1 of 2 categories: lesions that could be excised (including carcinomas and all neoplasms) and those that did not need to be excised (benign, nonneoplastic lesions).

About 5 patients per month had a thyroid gland lobectomy for a benign condition or a delay in diagnosis for a malignant condition due to failure to correctly diagnose their condition. Most diagnostic errors involved poor sampling or interpretation. To improve diagnostic accuracy at their hospital, the researchers increased the number of pathologist-performed (instead of clinician—performed) FNAs and required immediate interpretation of clinician- and radiologist-performed FNAs. They also adopted strict use of specimen adequacy criteria that are based on optimally adequate specimens rather than minimally adequate specimens. Finally, they cautioned clinicians not to interpret inadequate specimens.

Raab, S.S., Meier, F.A., Zarbo, R.J., and others (June, 2006). "The 'big dog' effect: Variability assessing the causes of error in diagnoses of patients with lung cancer." Journal of Clinical Oncology 24(18), pp. 2808-2814.

As many as 15 percent of patients with a lung mass are misdiagnosed due to pathology errors. However, pathologists often disagree on the cause of errors in interpreting lung specimens sent for cancer diagnosis. Pathologists at one hospital are more likely to agree with one another about the cause of a lung pathology error than pathologists from different hospitals, according to this study. This is partly due to the status of the pathologists in a given hospital, suggest the researchers. They attribute intra-hospital agreement and lack of inter-hospital agreement to the 'Big Dog' effect.

Senior experienced pathologists at each hospital serve as the final arbitrator for error cause, and use different methods and approaches to decide whether discrepancies exist and their causes. Most hospitals have only one 'Big Dog' to whom other pathologists ('Little Dogs') defer diagnostic judgment, explaining the greater agreement among pathologists at the same hospital. Yet, when 'Big Dogs' at one hospital are confronted with differing assessments from 'Big Dogs' at other hospitals, they remain reluctant to change their opinions. In this study, the locally dominant pathologists in every case did not significantly change their assessments.

The investigators asked pathologists from 6 institutions to review the slides of 40 patients who had a false-negative diagnosis of a lung specimen (the specimen was diagnosed as noncancerous when it was cancerous). They were asked to attribute the diagnostic error to clinical sampling (diagnostic material was obtained in one but not the other sample) or interpretation (the pathologist failed to identify the salient diagnostic features of the sample). Agreement about the source of diagnostic error among pathologists at the same hospital was better than agreement between pathologists from different hospitals.

Return to Contents
Proceed to Next Article