Because most risk factors for breast carcinoma are not readily amenable to primary prevention, and early diagnosis is a powerful prognostic determinant, screening for the disease is crucial. Consequently, assessment of the progress and comprehensiveness of screening and other breast carcinoma early detection activities is important. The relative frequency of early diagnosis may provide a useful indicator of such activities. Nationwide, time trends in the early diagnosis of breast carcinoma have been improving for decades, but not all population subgroups may have benefited equally.
Using 1994–1997 data from the California Cancer Registry (CCR), a review of diagnostic patterns of in situ and local stage breast carcinoma was undertaken. For analytic purposes, the CCR includes 10 regional registries and 36 county reporting groups. Three early diagnostic measures were designated, including in situ breast carcinoma with tumor size < 10 mm in greatest dimension, in situ breast carcinoma, and localized breast carcinoma with tumor size < 21 mm in greatest dimension. These are referred to hereinafter as early diagnosis breast carcinomas.
The percentage of early diagnosis breast carcinomas differed markedly by age, ethnicity, diagnosis year, and county of residence. Lower percentages of early diagnosis breast carcinomas were diagnosed in older women age ≥ 85 years. Hispanic women were diagnosed with lower levels of in situ breast carcinoma. Hispanic and black women were diagnosed with less localized breast tumors of small size. There was an increase in the percentage of early diagnosis breast carcinomas over the 4-year observation period. Lower percentages of early diagnosis breast carcinomas were reported for the nonurban county/county groups, which were characterized by greater distances, lower population density, and lower household incomes.
In the year 2000, approximately 183,000 women in the United States will be diagnosed with breast carcinoma, and more than 40,000 women will die from this disease.1–3 Many risk factors for breast carcinoma have been identified, including a family history, several menstrual and reproductive characteristics, and levels of endogenous and exogenous hormones.4 However, even with knowledge of such factors, breast carcinoma is not readily amenable to primary prevention; hence screening for the disease is crucial to detect it at an early, curable stage.
Improvements in the treatment patients with of breast carcinoma, patient quality of life, and the outcome of patients from an early diagnosis of breast carcinoma have been well documented5, 6 and underscore the importance of early detection wherever possible. Although the 5-year survival rate for women with breast carcinoma in the United States was 85% for women with all disease stages combined (1989–1995), this figure varied from a high of 96% for women with breast carcinoma diagnosed at an early stage (in situ or local disease) to only 21% for women with distant disease at the time of diagnosis.2 Therefore, early diagnosis is a powerful prognostic indicator, and public health screening efforts must be directed toward ensuring early diagnosis.
For planning purposes, the relative frequency of early diagnosis may provide a useful surrogate for the efficacy of local mammographic screening activities in different U.S. subpopulations.7, 8 Nationwide, time trends in the early diagnosis of breast carcinoma have been improving for decades, but all population subgroups may not have benefited equally.9–11
Using data from the California Cancer Registry (CCR), a review of diagnostic patterns of in situ and local stage breast carcinoma was undertaken to better understand the current efficacy of screening activities in California and to determine potential opportunities for improvement in early detection. This effort builds on earlier analyses of disease stage at the time of diagnosis,12 including the influence of MediCal insurance status on breast carcinoma stage at the time of diagnosis.7
MATERIALS AND METHODS
Sources of Data
The CCR is a population-based cancer registry that monitors the occurrence of all newly diagnosed cancers and cancer-related deaths among the population of California.13 A statewide cancer reporting law in 1985 created the CCR. The CCR includes 10 regional registries that are responsible for collecting the data on which this work is based. These data were based largely on cases abstracted and computerized by hospital cancer registrars in the regions. These regions include Region 1, Monterey, San Benito, Santa Clara, and Santa Cruz Counties; Region 2, Cancer Registry of Central California, including Fresno, Kern, Tulare, Kings, Madera, Merced, Stanislaus, Mariposa, and Tuolumne Counties; Region 3, Cancer Surveillance Program, including the 13 counties surrounding Sacramento County; Region 4 (Central Coast), including Ventura, Santa Barbara, and San Luis Obisbo Counties; Region 5 (Inland Empire), including San Bernadino, Riverside, Inyo, and Mono Counties; Region 6, the northernmost 16 counties of the state; Region 7, Orange County; Region 8, the 5 counties of the San Francisco Bay Area; Region 9, Los Angeles County; and Region 10, San Diego County.
The methodology of the CCR has been described previously.14 The statewide tumor registry commenced operation in 1988, and current reporting is complete and edited through 1997. The registry accessions all cancers except for nonmelanoma skin cancers and in situ cancers of the uterine cervix. Information on several demographic variables (e.g., age, race, gender, and residence), diagnostic variables (including stage at diagnosis, tumor size, histology, and grade of tumor), and first course of treatment are collected for all cases. Cases routinely are coded with regard to anatomic stage of disease using the general summary stage schema for 1988–199315 and SEER extent of disease for 1994–1997.16 For the purposes of this report, attention is focused on the period from 1994 to 1997, because diagnostic information on tumor size was available for those years only.
Cancer data for California are compiled and computerized for diagnosis years 1988–1997 and are available for scientific analysis.17 For the period 1988–1977, 1,343,094 cancers in total were reported to the CCR, including 208,914 breast cancers in both genders and all ethnic groups.
There are 58 counties in the state of California. In the available computer file,17 data for several counties with low populations had been combined for confidentiality purposes, resulting in 31 individual counties and 5 combined county groups. These 36 analysis units are referred to hereinafter as county/county groups.
For one part of the analysis, the county/county groups were dichotomized into categories hereinafter referred to as large area/low population (nonurban), and urban. Counties that include at least 10,000 acres and a population density < 30 inhabitants per acre were classified as nonurban.
The numbers of patients with breast carcinoma by reporting region and county/county group are presented in the tables along with the number and percent of patients who were diagnosed with breast carcinoma diagnosed when the disease was still at the in situ or local stage with a tumor size < 21 mm in greatest dimension. In situ breast tumors are not necessarily obligate precursors of invasive breast carcinoma, and, indeed, only 20–30% of women with these lesions but who have received no further treatment subsequently develop breast carcinoma.18
Whatever their biologic activity, the detection of in situ tumors may reflect the intensity of screening efforts for breast cancer in the population.16 The sum of in situ and small tumor local disease is also included in the tables. Although this sum does not necessarily have biologic significance, it is included as a mathematic convenience to simplify tabular assessment. Reviews and reports on the completeness and accuracy of these data have been published.19
In this report, variations in the percentage of patients diagnosed with breast carcinoma when the tumor was still in situ or at the local stage and still measured < 21 mm in greatest dimension are examined by age at the time of diagnosis, race/ethnicity, calendar year of diagnosis, and county of residence for the years 1994–1997. Selected census data by county and/or region are presented as putative variables of interest that may influence disease stage at the time of diagnosis.20
To guide the reader, limited use of significance tests has been made. Significance tests measure not only the likelihood of chance findings but also sample size. Caution should be exercised in interpreting significance because of the large numbers of cases included in these data and the fact that these data are not a sample but represent the universe of breast cancer cases. Analysis of the data was performed primarily with SPSS software.21 Time trend significance was tested using the chi-square for trend test.22
For the period 1994–1997, 89,210 patients were diagnosed with breast carcinoma in California. After excluding males (589 patients) and patients with unstaged breast tumors (2813 patients or 3.2%), there were 85,808 female patients with breast carcinoma with staging information that was available for analysis. Overall, 14.5% of patients were diagnosed when their tumors were still at the in situ stage, and 49% of patients were diagnosed with tumors at the local stage with a size < 21 mm in greatest dimension (Table 1).
Table 1. Percentage and Number of Patients Diagnosed with Early Stage Breast Carcinoma by Age Group, Ethnicity, and Year of Diagnosis by Anatomic Stage of Disease, 1994–1997 (85,808 Patients)
The percentage of patients diagnosed with in situ breast carcinoma differed markedly by age. Lower percentages of in situ tumors were diagnosed in older women, with 11.4% among those diagnosed at ages 75–84 years and 7.1% among those diagnosed at age ≥ 85 years. Higher percentages of in situ breast carcinoma were detected in the younger age groups: For patients ages 45–54 years and 55–64 years at the time of diagnosis, the percentages diagnosed with in situ tumors were 17.6% and 15.4%, respectively (Table 1). These differences were statistically significant (P < 0.0001). Similarly, the percentage of women who were diagnosed with localized breast carcinoma who had small tumors was also lowest among women age ≥ 85 years.
The distribution of early breast carcinomas also differed by ethnic group. Non-Hispanic white women were diagnosed with 14.4% in situ breast carcinoma and 51.4% localized breast carcinoma with small tumor size. Hispanic women were diagnosed with lower levels of in situ breast carcinoma: 12.0%. Hispanic and black women were diagnosed with less localized breast carcinoma with small tumor size: 41.4% and 39.1%, respectively. Asian/Pacific Islander women were diagnosed with 16.6% in situ breast carcinoma, and 45.1% localized breast carcinoma with small tumor size. These findings were statistically significant (P < 0.0001) (Table 1).
For all ethnic groups combined, there was an increase in the percentage of in situ breast carcinomas diagnosed over time, increasing steadily from 13.7% in 1994 to 15.3% in 1997. This represents an 11.7% increase in the number of women who were diagnosed at an early stage during the 4-year observation period or an average annual increase of approximately 3%. During this same time period, the percent of women diagnosed with localized breast carcinoma with small size increased from 48.4% to 49.3% or approximately 2% during the 4 years (Table 1).
To minimize confounding by ethnicity, the analysis of geographic differences in disease stage has been restricted to the non-Hispanic white segment of the population. For 1994–1997, the percentage of breast carcinomas in non-Hispanic white women diagnosed with disease still in situ averaged 14.4%, and varied from a low of 11.2% in the Inland Empire Region to a high of 16.4% in the San Francisco/Oakland Region (Table 2). The pattern of localized breast carcinoma of small tumor size was similar. The Inland Empire counties had the lowest proportion of women who were diagnosed with breast carcinoma that was still at the local stage (49.3%).
Table 2. Rank Order of Cancer Reporting Regions by Percentage of Patients with In Situ and Local (Tumor Size 0–20 mm) Breast Carcinoma: Non-Hispanic Whites, 1994–1997 (65,056 Patients)
The county/county groups within California were dichotomized into urban and nonurban analysis units. The nonurban county/county groups were selected based on large geographic size and a low population density. They form a geographically contiguous area comprising most of the eastern and northern portions of California (Fig. 1).
The eight nonurban county/county groups, compared with the urban county/county groups, were characterized by somewhat lower percentages of elderly patients (25.3% vs. 27.0%, respectively), low population density (14 inhabitants per square mile vs. 174 inhabitants per square mile, respectively), greater geographic area (31,593 square miles vs. 5327 square miles, respectively), and lower median family income ($29,054 vs. $37,093, respectively) (Table 3).
Table 3. Comparison of Percentages of Patients with In Situ and Local (Tumor Size 0–20 mm) Breast Carcinoma by Type of County/County Group (Large Area/Low Population Density or Nonurban vs. Urban): Non-Hispanic White Females, 1994–1997
Ranking the 36 county/county groups on the basis of median household income, the eight nonurban groups included the 19th, 26th, 27th, 28th, 30th, 32nd, 34th, and 36th lowest income groups. Among the nonurban county/county groups, the percentage of patients with breast carcinoma diagnosed in situ was 12%, and the percentage of patients with localized breast carcinoma of small tumor size was 49.9% (Table 3). For the urban county/county groups, the percentage of patients with tumors diagnosed in situ was 14.7%, and the percentage of patients with localized, small tumors was 51.6%. The difference was statistically significant (P < 0.001). The county with the highest percentage of patients diagnosed with in situ breast carcinoma, San Mateo County, had the fourth greatest population density and was the third highest in average family income.
An additional analysis was performed using as a third measure of early breast carcinoma detection (in addition to in situ and localized, small tumors): the percentage of in situ breast tumors measuring < 1 cm in greatest dimension. It is believed that these tumors are those most likely to be detected by mammography; 5.5% of patients with known stage and tumor size were diagnosed with in situ breast tumors measuring < 1 cm in greatest dimension.
The patterns of in situ breast tumor percentages in patients who were diagnosed with tumors of this small size were remarkably similar to the percentages of total in situ and localized breast tumors measuring < 21 mm in greatest dimension. Specifically, lower percentages of in situ breast tumors measuring < 1 cm in greatest dimension were reported for elderly patients: 4.1% for patients age 75–84 years and 2.5% for patients age ≥ 85 years. By ethnicity, 4.4% of Hispanics and 4.7% of blacks were diagnosed with 0–9 mm in situ breast tumors compared with 5.5% in non-Hispanic whites and 6.0% in Asian-Pacific islanders. The county/county group patterns of percentages of 0–9 mm in situ breast tumors were similar to the patterns of total in situ and 0–20 mm localized breast tumors.
The early diagnosis of breast carcinoma in women in California differed significantly by age, ethnicity, year of diagnosis, population density, and household income. Breast carcinoma was diagnosed at an early stage most frequently in women of younger age, in non-Hispanic white women or women of Asian and Pacific Island ancestry, in women who lived in urban counties characterized by higher household incomes, and in women who were diagnosed in more recent years. Important improvements were reported in the frequency of early breast carcinoma diagnosis over the 4-year reporting period. Between 1994 and 1997, there was a 12% increase in the diagnosis of in situ tumors and a 2% improvement in the diagnosis of local stage tumors of small size. This is a notable success.
The data in this report identify those women in California who are at high risk not for receiving a diagnosis of breast carcinoma but receiving such a breast carcinoma diagnosis when their disease is at a later stage and, hence, less treatable. Specifically, these high-risk groups include the elderly, Hispanic and Black female populations of the state and those who reside in nonurban areas.
Several underlying circumstances may cause this excess risk. It may be speculated that elderly women see their doctors less often compared with women of childbearing age, that rural residents have farther to drive to doctor's offices, and that these rural residents and certain ethnic and lower income groups may have different attitudes toward health screening and may receive less frequent medical examinations.
Taken together these data suggest that lower access to medical care increases risk for later stage of disease at the time of initial diagnosis. Unfortunately, this results in more aggressive treatment and poorer survival. The results of this analysis are consistent with earlier reports.7, 9 From these findings, we conclude that the identified groups should be targeted as high-priority subpopulations for mammographic screening.
The authors thank Drs. Gordon Honda and Rebecca Stickler for their helpful comments.