Nonconcurrence in abstracted diagnoses of non-Hodgkin's lymphoma

The availability of two independent sets of abstracted diagnoses on 289 cases of non-Hodgkin's lymphoma (NHL), one from the Iowa Surveillance, Epidemiology, and End Results (SEER) Program and the other from an epidemiologic study in Iowa of factors affecting rural males (FARM), allowed us to determine the disagreement between abstracted diagnoses. For both sets the reported diagnosis was translated to International Classification of Disease for Oncology terminology and then to the Working Formulation (WF). Comparison of the WF diagnoses between the FARM study and the SEER Program showed disagreement in 68 of 290 cases (23.4%). Apparent causes of disagreement were as follows: coding errors-9 cases, unconventional or ambiguous terminology on reports resulting in different interpretation of the NHL subtype--30 cases, differences in coding rules for the FARM study and SEER Program--9 cases, acquisition of different reports with different diagnoses on the same case by the FARM study and SEER Program--20 cases. Several corrective measures that might successfully decrease the incidence of disagreement include: education of pathologists in the use of conventional terminology, and use of the category "unclassifiable" when abstracting reports with unconventional or ambiguous terminology. This 23.4% disagreement rate in the abstracting method adds to the known problem of nonconcurrence in the pathologic diagnosis of subtypes of NHL.