OBJECTIVES Exposure assessment and the ascertainment of cases are topics of main concern in epidemiologic research. To investigate validity aspects of this kind practically, a comparative evaluation was performed of information sources for case-referent data.

METHODS Cases of Hodgkin's disease and non-Hodgkin's lymphoma diagnosed in 1975-1984 were collected from the Regional Cancer Register in Linkoping. The study population was restricted to men 20-80 years of age at diagnosis. Living cases included in a previous case-referent study and the complementary cases from the register were compared with regard to the exposures indicated by job title and industrial branch, as given in the National Population Census in 1975. Furthermore, questionnaire-based exposure information from the case-referent study was compared with exposures derived from census information.

RESULTS Only 21% of all of the cases in the study base had been included in the questionnaire-based case-referent study. No major differences in the distribution of exposure were found for the two case series, except that fewer subjects were economically inactive in the previous case-referent study. No clear relations appeared between survival and particular exposures. Risk estimates derived from the present cancer register study were compared with corresponding estimates from the previous case-referent study. With restriction to only economically active cases, the risk estimates approached those of the case-referent study.

CONCLUSIONS This comparison showed that incomplete ascertainment of cases in a case-referent study was not an important source of bias, at least not in relation to what could be achieved by more complete case ascertainment through register linkage.