SEARCH

SEARCH BY CITATION

Keywords:

  • breast cancer;
  • access;
  • mammography;
  • stage

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

BACKGROUND

There is a lack of data on the access to mammography and its relation to the incidence of early breast cancer. In this study, the authors evaluated access by correlating geographically the number of U.S. Food and Drug Administration (FDA)-certified mammography facilities and the AJCC stage of breast cancer at diagnosis by county.

METHODS

Breast cancer incidence rates and stage at diagnosis were compared with the number of FDA-certified mammography facilities by county in the Surveillance, Epidemiology, and End Results reporting areas. The objective was to determine whether the number of certified facilities was associated with the percent of breast cancers diagnosed at the in situ stage. This was a multiple-group ecologic study, and counties were used as the units of analysis.

RESULTS

There was a strong correlation between the number of mammography facilities and the population of a county, whereas there was no correlation between the number of mammography facilities and the land area of a county. A correlation existed between the percent of incident breast cancers that were diagnosed as in situ disease and the number of mammography facilities per 10,000 women among both whites and African Americans.

CONCLUSIONS

There was an association between the number of mammography facilities and population. In counties with ≥ 30,000 black and white females, 1) the percent of in situ breast cancers in black women and white women was correlated with the number of facilities per 10,000 women, indicating that population density is a factor in access for both racial groups; 2) except for 2 counties with ≥ 30,000 black women, the percent of incident in situ cases was similar in both black women and white women, indicating equal access in both groups; and 3) there was a correlation between the percent in situ incidence and number of facilities per 1000 square miles in white women, but not in black women. There was a direct correlation of statewide mammography rates with the number of facilities per 1000 square miles, indicating that the rate of screening depends on availability. Maximum rates of statewide screening were achieved when there were > 15 mammography facilities per 1000 square miles. Cancer 2005;103:1571–80. © 2005 American Cancer Society.

Breast cancer is the most common malignancy in American women and is the second leading cause of death among women of all ages.1 Diagnosis in an early stage, before metastasis, is important for survival.2 Mammography is one of the most effective procedures for detecting breast cancer early. Mammography rates, by state, have been correlated positively with breast cancer incidence and inversely with case fatality.3 At the county level, higher mammography rates also have been associated with higher incidence rates.4

However, access to mammography may be a challenge for some women. Traveling a long distance for a mammogram may produce an additional burden, making mammography less accessible in certain geographic areas. In older women in rural Kansas, increasing distance from a mammography facility was associated with a lower mammography rate.5 In addition, mammography rates were higher in counties that had permanent facilities compared with counties that had only mobile facilities or neither type of facility.5 However, in a study of women in rural northern Michigan, measures of distance and travel time to a mammography site were not associated with whether women had received a mammogram in the previous 2 years.6 In an Iowa study, it was found that there were no significant differences in mammography rates between rural women in farms and women in nonfarm households.7 In a New Jersey study, a geographic information system was used to characterize geographic areas with high proportions of women who were diagnosed with distant stage breast cancer: The investigators found that those areas had high proportions of African-American and Hispanic women and high proportions of “linguistically isolated” households.8

Access to mammography may be associated with breast cancer stage at diagnosis. A study in Georgia showed that rural patients who had breast cancer were more likely to have advanced disease at diagnosis compared with urban patients, which may be related to different rates of screening.9

It is difficult to measure access to screening. Whereas previous studies of access focused on measuring rates of mammography in women, in the current study, we examined the association between the number of mammography facilities available in a county and the AJCC stage of breast cancer at diagnosis.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Design

Using the counties participating in the Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute as the unit of measure, an ecologic study was conducted. Data from SEER, the U.S. Food and Drug Administration (FDA), and the Behavioral Risk Factor Surveillance System (BRFSS) were used in the analyses.

Sampling

The population represented in the study included African-American and white women living in the 200 counties participating in the SEER Program. The geographic coverage included the entire states of Connecticut, Iowa, New Mexico, and Utah and several counties within the states of California, Georgia, Michigan, and Washington. The number of counties represented in each state is indicated in Table 1.

Table 1. The Number of Counties in Each State Participating in The Surveillance, Epidemiology, and End Results Program
StateNo. of counties included in SEER
  • SEER: Surveillance, Epidemiology, and End Results.

  • a

    Indicates states in which all counties are included in the SEER Program.

California10
Connecticuta8
Georgia5
Iowaa99
Michigan3
New Mexicoa33
Utaha29
Washington13

Data Collection

The data for the independent variable, FDA-certified mammography facilities, are available online from URL: http://www.fda.gov/cdrh/mammography/certified.html [accessed January 31, 2004]. This website lists all FDA-certified mammography units by Zip code. All mammography facilities that meet baseline quality standards under the Mammography Quality Standards Act (MQSA) of 1992 and the Mammography Quality Standards Reauthorization Act amendments are listed. For a facility to be certified, it must be accredited by a federally approved accreditation body. All mammography facilities that produce, process, or interpret mammograms must meet the MQSA requirements, with the exception of facilities under the Department of Veterans Affairs, which has its own regulatory program.10 Only current data on FDA-approved mammography facilities for 2003 were available for the current study.

Because data on FDA-certified mammography facilities are given by Zip code and SEER Program data are given by county, the numbers of mammography facilities were translated into the appropriate counties using the Melissa Data website (available from URL http://www.melissadata.com/Lookups/countyzip.asp [accessed January 31, 2004]). When a Zip code was included in more than one county, the addresses of the mammography facilities were typed into the URL http://www.semaphorecorp.com/cgi/form.html [accessed January 31, 2004], which named the counties in which the facilities are located.

The SEER Program provided data for age-adjusted breast cancer incidence rates and stage at presentation for African-American and white women in the participating counties. In addition, the white female and African-American female population numbers were provided for each of the counties in the program. Data for the year 2000 were used, which is the latest year for which incidence rates were available. The SEER data are collected from 11 population-based cancer registries and 3 supplemental registries. SEER Program case ascertainment is 98%.11

Data for state mammography rates were obtained from the BRFSS. The percentage of women age ≥ 40 years who had a mammogram within the last year for the year 2000 survey, by state, was considered in the data analysis. Data on total population, on the female population age ≥ 18 years, and for the land area of each county participating in SEER were obtained from the U.S. Census Bureau website (http://quickfacts.census.gov/qfd/ [accessed July 1, 2004]).

Plan of Analyses

All data analyses were performed using SAS software (version 8; SAS Inc., Cary, NC). For all analyses, a correlation and regression analysis was performed. First, it was determined whether here was an association between the number of FDA-certified mammography facilities and the county population and whether there was an association between the number of mammography facilities and the land area in square miles of a county. The results from this analysis determined how much of an affect population size had on the number of mammography units.

The objective of the next analysis was to determine whether there was an association between the number of FDA-certified mammography facilities and the age-adjusted incidence rate by county. Separate analyses were performed for African-American women and white women. A second subanalysis was performed that examined the correlation between incidence and the number of mammography facilities per 10,000 women in each county. Another subanalysis was performed that examined the correlation between incidence and the number of mammography units per 1000 square miles. Each analysis was performed 1) with all counties included and 2) including only counties with a female population of each race ≥ 30,000 to insure a minimum of 50 incident cases.

Another analysis determined whether there was a correlation between the number of FDA-certified mammography facilities and the percent of in situ diagnoses. The percent of in situ diagnoses was calculated by dividing the age-adjusted in situ incidence rate by the age-adjusted total incidence rate. Separate analyses were performed for African-American women and white women. A subanalysis was performed that examined the correlation between the in situ incidence and the number of mammography facilities per 10,000 women in each county. Another subanalysis examined the correlation between the in situ incidence and the number of mammography units per 1000 square miles. Each analysis was performed both with all counties included and including only counties with a female population of each race ≥ 30,000.

Another analysis was used to determine whether there was an association between the number of FDA-certified mammography facilities and the percent of diagnoses made at Stage III or IV (advanced). Separate analyses of the percent in situ were performed for white women and African-American women, and subanalyses were performed that examined mammography facilities per 10,000 women and per 1000 square miles. Each analysis was performed twice, for all counties, and for only counties with ≥ 30,000 females of a given race.

To incorporate the important factor of mammography rates into the correlation, two correlation analyses were performed using state mammography rate data. First, a correlation analysis was performed for the percent of women age ≥ 40 years who received mammograms in the last year and the number of mammography facilities per 10,000 women. Second, a correlation analysis was performed for the percent of women age ≥ 40 years who had a mammogram in the last year and the number of mammography facilities per 1000 square miles.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

The results of all correlation analyses of the county data are summarized in Tables 2 and 3.

Table 2. Correlations Using All Counties
AnalysisRace
WhiteAfrican American
rP valuerP value
  • r: correlation coefficient (a measure of the strength of the correlation).

  • a

    Statistically significant.

Incidence and no. of mammography facilities0.130.0690.210.0039a
Incidence and no. of mammography facilities per 10,000 women−0.140.057−0.210.0040a
Incidence and no. of mammography facilities per 1000 square miles0.170.02a0.150.04a
Percent in situ and no. of mammography facilities0.050.500.040.80
Percent in situ and no. of mammography facilities per 10,000 women0.040.62−0.130.42
Percent in situ and no. of mammography facilities per 1000 square miles0.080.270.120.43
Percent Stages III and IV and no. of mammography facilities−0.00080.910.080.62
Percent Stages III and IV and no. of mammography facilities per 10,000 women−0.0370.610.280.077
Percent Stages III and IV and no. of mammography facilities per 1000 square miles−0.0020.970.090.59
Table 3. Correlations using Counties with > 30,000 White or African-American Women
AnalysisRace
WhiteAfrican-American
rP valuerP value
  • r: correlation coefficient (a measure of the strength of the correlation).

  • a

    Statistically significant.

Incidence and no. of mammography facilities0.160.240.180.52
Incidence and no. of mammography facilities per 10,000 women0.170.200.120.67
Incidence and no. of mammography facilities per 1000 square miles0.300.03a−0.830.0001a
Percent in situ and no. of mammography facilities0.170.240.180.54
Percent in situ and no. of mammography facilities per 10,000 women0.340.012a0.680.0074a
Percent in situ and no. of mammography facilities per 1000 square miles0.330.02a0.520.06
Percent Stages III and IV and no. of mammography facilities0.080.550.120.68
Percent Stages III and IV and no. of mammography facilities per 10,000 women−0.260.052−0.330.25
Percent Stages III and IV and no. of mammography facilities per 1000 square miles0.100.46−0.090.76

Correlation of Facilities with the Total Population and Land Area

There was a strong correlation between the number of mammography units in a county and the total population of a county (correlation coefficient [r] = 0.98; P < 0.001) (Fig. 1). No correlation was found between the number of mammography facilities in a county and the land area in square miles.

thumbnail image

Figure 1. This graph illustrates the correlation between county population and the number of mammography facilities (correlation coefficient [r] = 0.98). Note that Los Angeles County is not shown here because it has a population of > 9 million and, if it had been included, all other counties plotted in this figure would appear close together. Los Angeles County falls along the regression line.

Download figure to PowerPoint

Association between Incidence and the Number of Mammography Facilities

For white women, no significant correlation was found between breast cancer incidence and the number of mammography facilities. For African-American women, when all counties were included in analysis, a weak, statistically significant correlation was found (r = 0.21; P = 0.0039). However, when only the counties with ≥ 30,000 African-American women were included, there was no correlation.

Association between Incidence and the Number of Mammography Facilities per 10,000 Women and Mammography Facilities per 1000 Square Miles

In white women, no correlation was found between the incidence of breast cancer and the number of mammography facilities per 10,000 women, However, weak, positive correlations were found between the incidence and the number of mammography facilities per 1000 square miles, both with all counties included (r = 0.17; P = 0.02) and when only counties with 30,000 white females were included (r = 0.30; P = 0.03).

Among African-American women, when all counties were included, there was a weak, negative, statistically significant correlation (r = − 0.21; P = 0.0040) with the number of mammography facilities per 1000 square miles. A small, positive correlation also was found in African-American women between the incidence and the number of mammography facilities per 1000 square miles when all counties were included in the analysis (r = 0.15; P = 0.04). However, a very weak, negative correlation was found when only counties with ≥ 30,000 African-American women were included (r = − 0.83; P = 0.0001).

Association between the Percent of In Situ Incidence and the Number of Mammography Facilities

In white and African–American women, no association was found between the percent of in situ incidence and the number of mammography facilities.

Association between the Percent In Situ Incidence, the Number of Mammography Facilities per 10,000 Women, and the Number of Mammography Facilities per 1000 Square Miles

Among white women, there was no correlation between the percent in situ incidence and the number of mammography facilities per 10,000 women when all counties were included. However, when only counties with ≥ 30,000 white females were included, there was a positive, significant correlation between the percent in situ incidence and the number of mammography facilities per 10,000 women (r = 0.34; P = 0.012). This is shown in Figure 2. Similar results were found when mammography facilities per 1000 square miles were considered. There was no correlation when all counties were included, whereas there was a correlation when only counties with ≥ 30,000 white women were included (r = 0.33; P = 0.02).

thumbnail image

Figure 2. Association between the percent of in situ diagnoses and the number of mammography facilities per 10,000 white women (correlation coefficient [r] = 0.34). This analysis included only counties with at least 30,000 white females (n = 55).

Download figure to PowerPoint

Likewise, among African-American women, no correlation was found when all counties were included. However, when only counties with ≥ 30,000 African-American women were included, there was a significant correlation between the percent in situ incidence and the number of mammography facilities per 10,000 women (r = 0.68; P = 0.0074). The results are displayed in Figure 3. When considering facilities per 1000 square miles, there was no correlation with the percent in situ incidence among African-American women.

thumbnail image

Figure 3. Association between the percent of in situ diagnoses and the number of mammography facilities per 10,000 African-American women (correlation coefficient [r] = 0.68). This analysis included only counties with at least 30,000 African-American females (n = 14).

Download figure to PowerPoint

Association between the Percent of Stage III and IV (Advanced) Disease and the Number of Mammography Facilities

For both white women and African-American women, no correlations were found between the percent of advanced breast cancers (Stage III and IV) detected and the number of mammography facilities, the number of mammography facilities per 10,000 women, or the number of mammography facilities per 1000 square miles.

State Mammography Rate Analyses

Just like at the county level, there was a strong, significant, positive correlation at the state level between the number of mammography units and the total population (r = 0.97; P < 0.0001), whereas there was no significant correlation between the number of mammography units and the land area (r = 0.067; P = 0.64). There was no correlation between the percent of women age ≥ 40 years who had a mammogram in the past year and the number of mammography facilities per 10,000 women (r = 0.185; P = 0.19). However, there was a correlation between the percent of women who had a mammogram in the last year and the number of mammography facilities per 1000 square miles (r = 0.556; P < 0.0001) (Fig. 4). This relation approached a limit of nearly 73% of women age ≥ 40 years having a mammogram in the last year where there were > 15 mammography facilities per 1000 square miles. Table 4 shows the top 10 states and the bottom 10 states according to the percent of women who had a mammogram in the last year along with the number of mammography facilities per 1000 square miles.

thumbnail image

Figure 4. This graph illustrates the percent of women age ≥ 40 years who reported having a mammogram in the last year and the number of mammography facilities per 1000 square miles by state. Mammography rate data are from the Centers for Disease Control and Prevention Behavioral Risk Factor Surveillance System.

Download figure to PowerPoint

Table 4. The Percent of Women Age ≥ 40 Years who Had a Mammogram in the Last Year, with the Number of Mammography Facilities per 1000 Square Miles and the Actual Number of Mammography Facilities
Mammography rate rankaStatePercent of women age ≥ 40 years who had a mammogram in the last yearbNo. of mammography facilities per 1000 square milesActual no. of mammography facilities
  • a

    The mammography rate rank is the rank among the 50 states of the percent of women age ≥ 40 years who reported having a mammogram in the last year based on data obtained from the Centers for Disease Control and Prevention Behavioral Risk Factor Surveillance System.

  • b

    Data obtained from the Centers for Disease Control and Prevention Behavioral Risk Factor Surveillance System.

1Delaware73.215.3530
2Connecticut72.229.52143
3Massachusetts70.222.84179
4Rhode Island69.239.2341
5 (tie)Michigan68.95.60318
5 (tie)Maryland68.914.22139
7Maine66.81.9861
8New York66.713.66645
9New Jersey66.234.51256
10New Hampshire65.95.1346
41Nevada57.20.5965
42Missouri57.12.54175
43Montana57.10.3450
44 (tie)Texas56.12.08544
44 (tie)Colorado56.11.05109
46Wyoming54.80.2827
47Oklahoma54.21.50103
48 (tie)Utah52.00.5747
48 (tie)Mississippi52.02.30108
50Idaho50.40.5344

Tables 5 and 6 show county percent in situ incidence data for whites and African Americans, respectively. These tables rank counties that participate in SEER by the percent of in situ diagnoses and also include the number of mammography facilities per 10,000 women. In Tables 6 and 7, it can be seen that, generally, the counties with low percentages of incident in situ breast cancers have fewer facilities. The rank of in situ incidence is compared with the statewide mammography rate rank among the SEER-participating states. In general, the counties with higher percentages of in situ incidence are located in states with higher mammography rates.

Table 5. In Whites, the Percent of In Situ Breast Cancer Diagnoses by Countya
Percent in situ rankbCountyStatePercent in situNo. of mammography facilities per 10,000 womenState's mammography rate rankc
  • a

    Includes only the counties with at least 30,000 white females.

  • b

    Counties are ranked based on the percent of breast cancers diagnosed at the in situ stage. The rank is among the 55 counties participating in the Surveillance, Epidemiology, and End Results (SEER) Program that have at least 30,000 white females.

  • c

    Mammography rate rank among the eight states participating in SEER. The mammography rate is the percent of women age ≥ 40 years who reported having a mammogram in the last year according to the Centers for Disease Control and Prevention Behavioral Risk Factor Surveillance System.

1DubuqueIowa28.31.734
2FultonGeorgia25.31.076
3New HavenConnecticut25.01.211
4New LondonConnecticut24.91.001
5HartfordConnecticut24.91.281
6FairfieldConnecticut23.70.841
7SandovalNew Mexico23.30.305
8ScottIowa23.10.664
9WayneMichigan22.30.602
10OaklandMichigan22.00.842
46GwinnettGeorgia13.70.386
47WhatcomWashington13.30.467
48WashingtonUtah13.10.318
49LinnIowa12.50.674
50WoodburyIowa11.91.284
51CacheUtah11.90.628
52StoryIowa11.30.954
53Santa FeNew Mexico9.80.205
54PottawattamieIowa8.20.594
55San JuanNew Mexico7.90.765
Table 6. In African Americans, the Percent of In Situ Breast Cancer Diagnoses by Countya
Percent in situ rankbCountyStatePercent in situNo. of mammography facilities per 10,000 womenState's mammography rate rankc
  • a

    Includes only the counties with at least 30,000 African-American females.

  • b

    Counties are ranked based on the percent of breast cancers diagnosed at the in situ stage. The rank is among the 55 counties participating in the Surveillance, Epidemiology, and End Results (SEER) Program that have at least 30,000 African-American females.

  • c

    Mammography rate rank among the eight states participating in SEER. The mammography rate is the percent of women age ≥ 40 years who reported having a mammogram in the last year according to the Centers for Disease Control and Prevention Behavioral Risk Factor Surveillance System.

1HartfordConnecticut30.71.261
2KingWashington28.80.707
3New HavenConnecticut25.91.211
4WayneMichigan23.70.602
5OaklandMichigan20.20.842
6DeKalbGeorgia16.90.426
7AlamedaCalifornia16.70.573
8FairfieldConnecticut16.40.841
9Los AngelesCalifornia15.40.563
10FultonGeorgia14.81.076
11GwinnettGeorgia14.70.386
12Contra CostaCalifornia13.60.683
13ClaytonGeorgia7.320.126
14CobbGeorgia5.30.396
Table 7. Comparison of the Percent of In Situ Diagnoses in Whites and in African Americansa
CountyStatePercent in situ (95% CI)No. of mammography facilities per 10,000 women
WhitesAfrican Americans
  • 95% CI: 95% confidence interval.

  • a

    This table includes only the 14 counties participating in the Surveillance, Epidemiology, and End Results Program with at least 30,000 African-American females.

  • b

    The 95% confidence intervals do not overlap for the white and African-American percent in situ in these counties.

FultonbGeorgia25.3 (21.2–29.4)14.8 (10.3–19.4)1.07
New HavenConnecticut25.0 (21.7–28.3)25.9 (15.8–36.0)1.21
HartfordConnecticut24.9 (21.7–28.1)30.7 (21.1–40.3)1.26
FairfieldConnecticut23.7 (20.7–26.7)16.4 (7.9–25.0)0.84
WayneMichigan22.3 (19.6–25.0)23.7 (20.7–26.8)0.60
OaklandMichigan22.0 (19.5–24.5)20.2 (12.9–27.5)0.84
DeKalbGeorgia21.1 (16.2–26.0)16.9 (12.6–21.2)0.42
KingWashington20.7 (18.6–22.8)28.8 (19.3–38.3)0.70
CobbbGeorgia18.9 (15.1–22.7)5.3 (−0.4–11.0)0.39
AlamedaCalifornia16.6 (14.0–19.3)16.7 (11.1, 22.3)0.57
Los AngelesCalifornia16.5 (15.5–17.5)15.4 (12.9–17.9)0.56
ClaytonGeorgia14.9 (7.0–22.8)7.3 (2.0–12.6)0.12
Contra CostaCalifornia14.5 (11.9–17.1)13.6 (5.7–21.5)0.69
GwinnettGeorgia13.7 (10.3–17.1)14.7 (7.2–22.3)0.38

Table 7 compares the percentages of in situ diagnoses among white women and African-American women in the 14 SEER-participating counties that had ≥ 30,000 African-American women. The table shows that, whereas the percent of in situ diagnoses were similar in most counties between whites and in African-Americans, there were two counties with disparities: Fulton County and Cobb County in Georgia.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

In the current study, we used data listed for 2003 for the number and location of approved mammography facilities. The SEER breast cancer incidence data and the BRFSS mammography rate data were collected in 2000. Although this may affect the results, it seems unlikely that the number of mammography facilities has changed significantly since 2000.

Each county-level analysis was performed twice: once for all counties that participated in the SEER Program and again only for those counties with ≥ 30,000 white women or African-American women. A cut-off level of 30,000 women was used, because this is approximately the minimum number of women of a given race needed for there to be at least 50 incident cases each year. We decided to include both analyses, because both were important in the study. The higher population counties, with ≥ 30,000 white women or African-American women, yielded more reliable estimates, because they are not as susceptible to the mathematical fluctuations in the incidence rate as the counties with smaller populations. The correlation analyses using all counties may not be as reliable, because the plots include points that may not be accurate. Conversely, it is important to include counties with smaller populations in the analysis, because it is more likely that these counties will not have enough facilities to serve their populations.

It was important to determine whether county population and land area were correlated with the number of mammography facilities in a county. This was useful to take into account the population of a county in the associations. Because we found a strong correlation between population and the number of mammography facilities, subsequent analyses included a measure of mammography facilities per 10,000 women. Although no correlation was found between land area and the number of mammography facilities, we still accounted for the size of a county still in subsequent analyses at the county level.

For white women, there was no association between the number of mammography facilities and the total incidence of breast cancer at all stages, either when all 200 SEER counties were included in the analysis or when only those counties with ≥ 30,000 white women were included. In African-American women, an association between the number of mammography facilities and total incidence was shown when all 200 SEER counties were included in the analysis, but not when only counties with ≥ 30,000 African-American women were included.

Total breast cancer incidence is a function of both screening mammography and clinical examination or breast self-examination. Because in situ breast cancers usually are not detected by physical breast examination,12 determining the percent of all breast cancers detected at the in situ stage is a method to determine the effects of screening mammography on disease stage at diagnosis. In the current study, we determined that the percent of in situ cancers was correlated with the number of mammography facilities per 10,000 women only when the counties with ≥ 30,000 white women or black women were included in the analyses.

Table 7 compares the percent of in situ diagnoses in white women and in African-American women. Although most of the counties represented had similar percentages of in situ diagnoses, several counties have large disparities. These counties were Cobb County and Fulton County in Georgia, with a higher percent in situ incidence in whites than in African Americans. It has been reported that black women continue to be diagnosed with breast cancer at a later stage compared with white women, even though the mammography rates for the two groups are converging.13 However, the performance of the mammograms does not seem to be a factor in the disparity, because sensitivity, specificity, and positive predictive values are similar when comparing screening mammograms of black women and white women.14

In addition to the in situ incidence, advanced stage incidence also was analyzed. It was expected that, if there was a correlation between the number of mammography facilities and the percent of advanced stage incidence, then the correlation would be negative, because fewer mammography facilities would be associated with the detection of cancer at later stages. In the current study, no significant associations were found between the number of mammography facilities in a county and the percentage of the incidence of Stage III or IV disease. This result is not unexpected, because Stage I–IV breast cancer can be detected through either self-examination or clinical breast examination, and not necessarily through mammography.

With the exception of the correlation between the total county population and the number of mammography facilities, most of the statistically significant correlations were weak, with correlation coefficients of ≈ 0.3. For example, the correlation coefficient for the relation between the percent in situ incidence in white women and the number of mammography facilities per 10,000 women in counties with ≥ 30,000 white females was 0.34. The r-square value for this coefficient, 0.1156, means that only 11.56% of the variation in the percent in situ incidence can be explained by variation in the number of mammography facilities per 10,000 women. This suggests that there are other factors involved in the correlation.

There are several possible confounders, including differences related to urban versus rural populations, the number of mammographic units in each facility, availability of public or private transportation, local publicity, underlying breast cancer patterns in the counties, and the presence of mobile mammography units. Another confounder was the mammography rate in each county. Because there are minimal standardized data on mammography rates at the county level, for the current study, we looked at the state level to obtain a general concept of the differing rates.

A correlation was found between the state mammography rate and the number of mammography facilities per 1000 square miles in a state. Figure 4 shows that this correlation approached a limit at a number of mammography facilities per 1000 square miles. The highest reported mammography rate (73%) was observed in Delaware, with 15 facilities per 1000 square miles. Other states (see Fig. 4) had more facilities per 1000 square miles but had a mammography rate of < 73%. The high Delaware rate most likely reflects a very active publicity program for screening in the state. States with lower rates but with more facilities than 15 per 1000 square miles probably have less intense publicity campaigns for screening. In Rhode Island, for instance, with 39.23 mammography facilities per 1000 square miles, 69% of women reported having a mammogram,. Nonetheless, it is clear from Table 4 that, the fewer the number of facilities per 1000 square miles, the lower the percent of women who report having a mammogram.

Tables 5 and 6 show a correlation between the percent of in situ diagnoses in a county and the mammography rate of the county's state. In general, it can be seen that the counties with a higher percent of in situ diagnoses are located in states with higher mammography rates. Obviously, conclusions cannot be reached from these tables, because they compare county data with state data, which can vary widely within a state.

Previous studies did not consider the relation between the number of mammography facilities and the stage of breast cancer at diagnosis. Studies have shown associations between access to mammography facilities and mammography rates in women5 and between mammography rates and breast cancer incidence.3 However, to our knowledge, there has been a lack of studies attempting to associate access to mammography facilities with disease stage at diagnosis. One study that investigated this association showed that, in Georgia, women who lived in urban counties were more likely to be diagnosed with breast cancer at an early stage compared with women who lived in rural counties.9 Those authors suggested that the more advanced incident cancers were due to poor access to health care, fewer cancer prevention activities, and decreased receptivity to health resources.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES