Previous studies of cancer incidence among persons living in municipalities within one mile of two nuclear materials processing and fabrication plants in Pennsylvania were extended for the years 1998–2004. It had been shown that mailing addresses for residents of rural areas often did not reflect the actual municipality of residence and, if not corrected, would bias study results. The previous studies had corrected for this bias. Accordingly for the extended study, we obtained mailing addresses from the Pennsylvania Department of Health (PDH) for 866 persons with cancer who presumably lived in one of eight minor civil divisions (MCDs) near or encompassing the former nuclear facilities, designated as Area 1 in previous studies conducted by the PDH. Street addresses were geocoded and local postmasters were asked to place rural delivery addresses, post office boxes and street addresses that could not be geocoded into the correct MCD of actual residence. Over 15% of the mailing addresses were found not to be within the boundaries of the Area 1 municipalities. After the mailing addresses of individuals with cancer were placed in their proper MCD of residence, the number of persons diagnosed with cancer (n = 708) and confirmed to have lived in Area 1 was as expected (728.4) based on cancer incidence rates in the general population of Pennsylvania (SIR 0.97; 95% CI 0.90–1.05). To further evaluate the patterns of cancer rates near these nuclear facilities and the influence of improved reporting and geocoding of addresses over time, analyses were conducted of publicly available cancer incidence data from 1990 through 2004. Based on mailing addresses, a steady decrease in the number of cancers reported in the Area 1 proximal MCDs was seen, in contrast to a steady rise in the number of cancers reported in seven adjacent but more distant MCDs from the nuclear facilities, designated as Area 2. These patterns were attributed to improvements over time in the geocoding of residential mailing addresses coupled with the gradual elimination and replacement of rural delivery addresses with street addresses. The incorrect placement of mailing addresses in residential Area 1 municipalities prior to about 2002 overestimated the number of cancers occurring among residents living in close proximity to the nuclear facilities and, correspondingly, underestimated the number among Area 2 residents. Summing Area 1 and Area 2 data showed that there was no change in cancer rates over time. These results are consistent with previous studies indicating that living in municipalities near the former Apollo-Parks nuclear facilities was not associated with an increase in cancer occurrence.
* International Epidemiology Institute, 1455 Research Boulevard, Suite 550, Rockville, MD 20850; † Vanderbilt University, Department of Medicine and Vanderbilt—Ingram Cancer Center, Nashville, TN 37232; ‡ University of Pittsburgh, Department of Pathology, School of Medicine, Department of Epidemiology, Graduate School of Public Health and University of Pittsburgh Cancer Institute/Hillman Cancer Center, Pittsburgh, PA 15213; § Epidemiology and Surveillance Research, American Cancer Society (retired), 674 Plantation Point Road, Woodbine, GA 31569.
For correspondence contact: J. D. Boice, Jr., International Epidemiology Institute, 1455 Research Blvd., Suite 550, Rockville, MD 20850, or email at firstname.lastname@example.org.
(Manuscript accepted 10 July 2008)