A study of space-time clustering in Hodgkin's disease in the Manchester region.

MacMahon (1966), in a comprehensive review of the epidemiology of Hodgkin's disease, provided evidence for the hypothesis that Hodgkin's disease may consist of several separate entities. Cole, MacMahon, and Aisenberg (1968) supported this hypothesis by analysis of regional and age variation in mortality within the United States. Smithers (1970) has reviewed the epidemiological and clinical evidence and he concludes that it is unlikely that Hodgkin's disease is two distinct conditions. Families with more than one affected member have been reported, in which the disease occurred at the same time among relatives of different ages (DeVore and Doan, 1957; Razis, Diamond, and Craver, 1959). On the basis of this, together with the clinical and pathological features of Hodgkin's disease, MacMahon (1966) suggested that one variety may be triggered off by infections. Fraumeni and Li (1969) and Cridland (1961) have shown that an excess of cases presents in the winter months. Clemmesen, Busk, and Nielsen (1952), who used a crude mapping technique, reported microclusters of the disease. Fraumeni and Li (1969) looked for time and space clustering in 314 affected children; they were unable to identify any evidence of clustering of cases using the procedure developed by Ederer, Myers, and Mantel (1964). Alderson and Nayak (1971) carried out a preliminary analysis of incidence data for 370 patients presenting with Hodgkin's disease over a period of four years in the Manchester area; they showed a trimodal age distribution and variation in the preponderance of males at different ages, as reported by other authors (MacMahon, 1966; Clemmesen, 1965). There was no appreciable variation in the social class or marital status of the patients compared with the population from which they were drawn, but analysis of the date of first attendance at hospital revealed a significant peak of patients first attending in the second quarter of the year. There was an excess of patients close together in space and time, the clustering being signi-