W e read with interest the report of Hjalmars et al., documenting an increase in childhood brain cancer incidence in Sweden for the period 1973-1992. 1 The authors note average annual increases of 2.6% for this period, with the increase restricted to the astrocytoma category and not noted for either the primitive neuroectodermal tumor (PNET)/medulloblastoma subcategory or the ependymoma subcategory.
The incidence of brain cancers among children younger than 15 years in the United States also increased during this time period, with an average annual increase of 1.8% from 1973 to 1994. 2 We have presented a detailed analysis of the temporal trend for this increase in brain cancer incidence among children in the U.S. 3 We demonstrated that this increase was best explained by a model with a step increase in incidence occurring in the mid-1980s (termed the "jump model"), compared with the alternative model of a continuous increase in incidence from 1973 to 1994. Similar to the data for Swedish children, the increase in brain cancer incidence for children in the U.S. from the 1970s to the 1990s was restricted to the astrocytoma subcategory, with no increase observed for the PNET/medulloblastoma subcategory. 4 The significantly better fit of the incidence data by the jump model (in the absence of a jump in mortality rates) supports the hypothesis that the observed increase in incidence somehow resulted from changes in detection (e.g., the availability of magnetic resonance imaging for brain imaging in the mid 1980s) and/or the reporting of childhood primary malignant brain tumors during the mid 1980s. 3,5 Subsequent to our initial analysis demonstrating the superior fit of the jump model to the continuous increase model, data for childhood brain cancer incidence for 1995 and 1996 have become available, which provide further support for the jump model. 6 These most recent incidence data from the SEER program document essentially stable brain cancer incidence rates among children in the U.S. since 1986 (with an estimated annual percentage change of 0.2%). 6 In addition, the California Cancer Registry recently reported that brain cancer rates for children in California were essentially constant for the years 1988 -1995 (with an estimated annual percentage change of Ϫ1.7%, P Ͼ 0.05). 7 Visual inspection of the temporal trends in childhood brain cancer incidence presented by Hjalmars et al. suggest that the "jump model" may be applicable to their data for children in Sweden. 1 For example, visual inspection of the incidence pattern for all malignant brain tumors (Fig. 2b in their article), for the astrocytoma subcategory (Fig. 2c), and for the astrocytoma subcategory among girls (Fig. 2e)