By outlining the development and use of the Hiroshima and Nagasaki tumor registries, the authors emphasize the critical role these registries play in Life Span Study cancer investigations.

by Kiyohiko Mabuchi, Department of Epidemiology and Epidemiologic Pathology, Hiroshima; and Midori Soda, Department of Epidemiologic Pathology, Nagasaki

This article was originally published in RERF Update 2(2):5-6, 1990.

---

It is well appreciated that in risk assessment, cancer incidence data offer distinct advantages over mortality data. Incidence cases are ascertained at the time of disease onset; thus incidence data can provide a direct measure of cancer risk as it occurs in members of the population–an important feature when studying temporal patterns of cancer. In addition, incidence surveillance is effective in documenting tumors with a favorable prognosis, such as cancers of the breast and thyroid, which may not appear on death certificates. Furthermore, detailed medical data directly obtained from hospital records provide information needed for accurate and specific diagnosis. Such detailed information is not generally available from death certificate-based studies.

 

Active approach to cancer registration

The tumor registries are essential in systematically collecting, managing, and analyzing data relating to individual tumor cases on a population basis. More than 30 years ago, population-based registries were established in Hiroshima and Nagasaki under the auspices of each city’s medical association. From the outset, ABCC-RERF has been responsible for the day-to-day operations of the registries. On a daily basis, RERF personnel visit the major hospitals in the area and extract necessary information from hospital records. This “active” approach to cancer registration, employed by most of the well-known American tumor registries, results in the high quality of registry data from Hiroshima and Nagasaki.

To evaluate the quality of data and, hence, the reliability of registration, two numerical indices are commonly used. These are (i) the proportion of cases registered with histologically verified diagnosis (histological verification, HV) and (ii) the proportion of cases registered for which only information from the death certificate is available (death certificate only, DCO). At the present time, the Hiroshima and Nagasaki registries have an HV rate in excess of 70% and a DCO rate of less than 9%, which make them among the best in Japan and comparable to many internationally established tumor registries (Cancer incidence in five continents, Vol. V, International Agency for Research on Cancer, Lyon, France, 1987).

RERF also supports tissue registries to which tissue-diagnosed tumor cases, both benign and malignant, in the Hiroshima and Nagasaki areas are reported along with tumor tissue slides. The tissue registries serve as a supplemental data source for both case finding and diagnostic confirmation. Recently, a prefecture-wide cancer registry also administered by RERF was started in Nagasaki, adding another source of cancer incidence data.

 

Incidence data for entire LSS population: a product of registry standardization

It is important to realize that maintaining good population-based tumor registries in Hiroshima and Nagasaki is requisite for attaining a high-quality cancer incidence database for RERF cohorts. Whereas the tumor registries have been instrumental for cancer incidence studies of several specific sites, overall analysis of tumor registry incidence data for the LSS and other cohorts has not been undertaken since Life Span Study Report 9 (Part 3) on Nagasaki for 1959-1978 (T. Wakabayashi et al., RERF TR 6-81). This is partly because of delays in data collection in Hiroshima. Major improvements have taken place during the last several years which make it now possible to generate recent incidence data for the entire LSS sample. Data collection problems in Hiroshima have been resolved and cases from certain major hospitals have been updated.

Together with this, a new effort was initiated jointly by the Hiroshima and Nagasaki registry teams to establish a common tumor registry database for the LSS. The objective was to achieve uniformity and consistency in the handling of registry data from the two cities. A painstaking effort was undertaken by the registry staff from the two cities to standardize diagnostic criteria and data handling procedures. The registry staff reviewed all documents for more than 15,000 cancer patients who have been accessed during the last 30 years. To ensure consistency, staff from the two cities met on a monthly basis to review samples of case materials and to discuss cases presenting difficulty in coding. A common set of coding rules was developed and used under the guidance of the registry physicians and pathologist. Data were entered into a new database system, developed with the assistance of the RERF Research Information Center, and various data quality assurance measures including intensive logical checks were implemented.

A brief inspection of the incidence data suggests the potential power of this data set. The Figure shows the number of incidence cases among members of the LSS population for 1958-1985 as compared with cancer deaths for the same period. Noteworthy is the number of incidence cases which total nearly 10,000 (excluding those who were not in the cities at the time of the bombings). This is almost twice the number of mortality cases, which total about 5,800. Proportional excesses in incidence over mortality cases are striking for such sites as the breast, thyroid, skin, uterus (mostly cervix), and urinary tract (mostly bladder). There also is a marked increase in the number of cases (almost 2,000) for cancer of the digestive system, reflecting the high background incidence of cancer of the stomach and liver in Japan. How the increased number of cases together with improved precision in diagnosis affects risk assessment is of special interest to all those who are concerned with radiation carcinogenesis.

Outmigration still a concern

While incidence data have many advantages, it should be noted that cancer incidence data are not free of problems. The major problem, which is already being addressed, results from the considerable number of LSS subjects who have over the years migrated out of the Hiroshima and Nagasaki area. An estimated 20% of the surviving LSS subjects no longer reside in areas covered by the tumor registries, and migration rates are inversely related to age at exposure. Cancers among the migrants are ascertained by the nationwide mortality surveillance but not by the Hiroshima and Nagasaki registries. An address database project was initiated last year and is being vigorously pursued to acquire and maintain information on residence history for individual cohort members.

Meanwhile, our analytical strategy will be to adjust for migratory movement using methods similar to those employed in a previous breast cancer incidence study (M. Tokunaga et al., RERF TR 15-84). Since migration has not been found to differ by radiation dose, relative risk estimates are not expected to be biased by such adjustment. Analysis of this incidence data set for the LSS sample is currently being undertaken jointly with the Department of Statistics and is expected be completed soon (Editor’s note: See Radiation Research 137S:17-67, 1994).