An examination of the sensitivity of reported trends in childhood leukemia incidence rates to geographic location and diagnostic coding (United States).

BACKGROUND: There have been conflicting reports of increased incidence of childhood leukemia in the United States with some, but not other, registries reporting increasing rates over the past two decades. Because of the reported discrepancy in childhood leukemia incidence rates an analysis of the SEER database was undertaken.

METHODS: The latest SEER data (1973-1995) were analyzed for trends in childhood (age 0-14) leukemia incidence rates by histologic group (all leukemia combined, acute lymphocytic leukemia (ALL), acute mylogenous leukemia (AML), and other acute leukemia) for each SEER reporting region.

RESULTS: A significant increase in ALL during 1973-1995 was observed in the combined SEER data, but the increase was a function of whether the first 3 years, data from the SEER Detroit reporting region are included. For the years 1973-1975 the Detroit region reported to much lower rates for ALL and much higher rates for "other acute leukemias" relative to the other SEER regions, resulting in an exaggerated temporal increase in ALL.

CONCLUSION: Excluding both the temporal variability, and coding differences for Detroit in the 1973-1975 time frame, there has been no significant increase in childhood leukemia of any histologic group or age category in the United States from the 1970s to 1990s.