Summary This is the first report of a case-control epidemiological study on lymphomas and leukaemias occurring in Yorkshire during 1979-84. This paper deals with the results of the Hodgkin's disease analysis comprising 248 cases and 489 controls. The results indicate support for previous work with respect to small family size and past history of infectious mononucleosis. Positive observations made in a previous pilot study are also confirmed and extended with respect to associations with certain chronic skin lesions, dental anaesthesia and familial factors. Negative associations are described with respect to X-ray exposures and cigarette smoking. It is proposed that these results fit into a general hypothesis that these conditions are the result of interaction between infectious agents and altered immunity in those persons genetically predisposed. the control group but the majority of controls were in hospital either due to an accident or for cold surgery. All interviews were conducted by trained interviewers, usually in hospital, using a standard questionnaire covering all aspects of past life relating to occupation, hobbies, personal habits, drug ingestion, family history and past medical history. Hospital and GP records were checked to confirm the accuracy of drug and medical histories. Cancer in other blood relatives was also cross-checked with the cancer registries or by death certificate perusal. All data were coded, computerised and validated by a trained group not involved with interviewing. The case-control statistics were produced using the programmes of Rothman and Boice (1979), using their stratified techniques.Two levels of analysis were undertaken: firstly, by pooling age groups, disease subtype and sex; and, secondly, by stratifying where possible by sex, age (15-35yr vs. 36+yr of age), and subtypes of disease.
Results
Cases studiedThe total number of HD cases occurring in the Region during the period exceeded that interviewed. The pilot study used the same data base but its results are independent of the analyses presented here. Non-significant or unassessable risk Some topics could not be studied adequately because the number of case or control responses was too small. Table I lists topics with 5 or less eligible cases and controls, which are not considered further. Table II shows non-significant differences at the 5% level of probability, having computed risk ratios less than 2.0. A few factors with higher risk ratios, which are not statistically significant in the pooled data, include the occupation of hand and machine sewers Br.