Background: Valid and reliable diagnoses of disease are key both to meaningful epidemiologic and clinical investigations and to decision-making about appropriate treatment. One previous study highlighted the lack of precision in diagnosing primary brain tumors in a neuropathology referral practice. The current study explores diagnostic discrepancies in a population-based adult glioma series by hospital of origin, specialty training of the original diagnosing pathologist, and clinical significance.

Methods: To confirm patients' eligibility for the San Francisco Adult Glioma Study, the authors obtained participants' pathology specimens and conducted a uniform secondary neuropathology review. Eligible patients were all adults age 20 years or older newly diagnosed with glioma between August 1, 1991, and March 31, 1994, who resided in 1 of 6 San Francisco Bay Area counties.

Results: Overall, the original and secondary diagnoses were the same (concordant) for 352 (77%) of the 457 cases available for study. Twenty-six percent of the cases from community hospitals were discordant, compared with 12% of the cases from academic hospitals P= 0.004. Of the 105 discordant diagnoses, 17 (16%) were determined to be clinically significant, defined as a difference that could significantly alter patient management and/or prognosis. Sixteen of these 17 cases originated at community hospitals, and only 1 originated at a hospital with a neuropathologist. Based on the distribution of review diagnoses, subjects presenting at nonacademic hospitals were more likely than those presenting at academic hospitals to have glioblastoma (61% vs. 52%; P = 0.07).

Conclusions: The percentage of cases with discrepant original and review diagnoses was higher among those originally diagnosed at community hospitals without a neuropathologist than among those originally diagnosed at an academic hospital with a neuropathologist. Clinically significant discrepancies were much more likely to have originated at a community hospital without a neuropathologist. These data highlight the importance of review of brain tumors by a neuropathologist prior to decision-making regarding treatment. A separate implication of this study is that glioma cases selected exclusively from academic or nonacademic institutions in a particular geographic area are unlikely to be representative of all cases occurring in that area.