Early detection through mammography screening has contributed substantially to the reduction of breast carcinoma morbidity and mortality.1–5 Several studies have been conducted to assess the use of mammography by different patient groups6–8 and in different health care settings9, 10 using various study methods. Those studies produced conflicting results. For example, using self-report through mailed questionnaire, Ostbye et al.8 found that mammography use among women ages 50–64 years was 70–80%, which declined to 40% among women ages 85–90 years. Using data from the 2000 National Health Survey (Cancer Control Module),11 Rakowski et al.7 found that, among women ages 55–79 years, 49% had returned for a repeat screening examination at the 12-month interval, and 64.1% had returned within 24 months. In another study, Blanchard et al.9 tracked actual mammography events in a single urban hospital between January 1985 and February 2002 and found that the mean number of mammograms received by women in that period was only 5.06, or 51% of the number recommended by the American Cancer Society.12
The studies described above either assessed self-reported use of mammography several days, weeks, months, or years after its occurrence or assessed mammography use within a single health care setting. Population-based mammography registries, like the New Hampshire Mammography Network (NHMN), provide a more comprehensive picture of the process of screening as it actually occurs in community-based settings, which is the predominant model for breast cancer detection in the U.S. To understand the utilization of mammography in New Hampshire, we undertook a study comparing population statistics for New Hampshire women from the 2000 Census compared with women who underwent mammography screenings and were enrolled in the NHMN between May 1, 1996 and December 31, 2000. This allowed us to assess the utilization of mammography in New Hampshire using actual population statistics as our denominator.
MATERIALS AND METHODS
The NHMN is a voluntary, population-based registry that records breast-imaging events with linkages to pathology diagnoses for consenting women in New Hampshire. Approximately 90% of all women who receive mammography at participating facilities consent to provide data to the NHMN. The design and development of the registry and the performance characteristics of mammography in New Hampshire have been described in detail elsewhere.13, 14 Briefly, the project was funded in late 1994, and data collection began in May 1996 and was implemented fully by December 1996. Women's enrollment in the NHMN includes completing a baseline demographic and epidemiologic survey (e.g., family history of breast and ovarian carcinoma, current height and weight, parity, age at first birth); participating in an in-person interview with a mammography technologist to gather additional risk information (e.g., hormone use, menopausal status, previous breast procedures); consenting to release medical records, including pathology information for linkage to radiology data; and agreeing to future contact for research purposes. All data collection forms and procedures have been approved by our Committee for the Protection of Human Subjects.
Benign and malignant pathology data are collected from participating pathology laboratories, and data regarding patients with breast carcinoma are supplemented by the New Hampshire State Cancer Registry. The ascertainment, linking, and coding of pathology data are described in detail elsewhere.13 Participation in NHMN is voluntary and, during the study period, 40 of 44 mammography facilities (91%) were enrolled in the registry. The composition of these sites includes hospitals (67.5%), clinic-based or outpatient facilities and breast imaging centers (22.5%), and physicians' private offices (10.0%). In addition, during the study period, facility zip codes revealed that 51.5% of these facilities were located in state or federally designated underserved areas. Of these, 47.1% of facilities were located in health professional shortage areas, 29.4% were located in medically underserved areas, 17.6% were located among medically underserved populations, and 5.9% were located in state (governor-designated) underserved areas.
Utilization Assessment Procedures
In the spring of 2004, we obtained New Hampshire population figures from 2000 Census data for age-eligible women (ages 40 yrs and older) who should have received a recommendation for screening mammography. Ninety percent of the counties in New Hampshire are well represented in the NHMN. Some facilities in one county have participated inconsistently, by pausing for periods of time due to staffing shortages, which is a reality of delivering mammography services. Therefore, we excluded the 35,000 women who received mammography in that county and dropped the Census data from the county in our analyses. We then stratified women into the following age groups: ages 40–49 years, 50–59 years, 60–69 years, 70–79 years, and 80 years and older. We identified women who enrolled in the NHMN between April 1, 1996 through December 31, 2000 and stratified those women into corresponding age groups for comparison with the New Hampshire 2000 Census data. To account for one county with a non-NHMN facility, we estimated that facility's screening volume by using data collected from a comparable NHMN facility within the same county. This allowed us to estimate adjustments needed to account for registry capture. One additional facility in another county was not participating, but its volume was small that we kept that county in our analysis. To assess NHMN women's screening adherence, we stratified women by age and screening interval. These intervals included women who adhered within a 14-month interval; women who adhered within a 26-month interval (categories mutually exclusive), allowing 2 months to make mammography appointments; and women who did not return for any repeat mammography. We allowed a follow-up of 27 months (into March 31, 2003) after screening mammography to ascertain adherence time intervals. Women who did not return for screening within 27 months of their prior screening examination were considered irregular screenees. Data analysis involved the use of descriptive statistics.
Characteristics of the women enrolled in NHMN (Table 1) revealed that participation in screening decreases with age and that the New Hampshire screened population is well insured (97%) and well educated (61% with some college or college degree). New Hampshire 2000 Census data revealed that 205,084 women were eligible for mammography screening (ages 40 yrs and older), and NHMN captured 116,480 women in the period between 1996 and 2000 (Table 2). Approximately 36% of New Hampshire women either never had a mammogram or had not had 1 within the past 27 months. We classified these women as irregular screenees. Older women (age 80 yrs and older) were less likely to be screened (79% irregular screenees) compared with younger women (ages 40–69 yrs; 28–32% irregular screenees). Combining both NHMN encounter data and estimates of women screened at a nonparticipating site in New Hampshire, the proportion of women who received routine screening during the study period was 65%.
Table 1. Characteristics of Women in the New Hampshire Mammography Network: May 1, 1996– December 31, 2000
|Demographic characteristics|| |
| 40–49 yrs||46,919 (40)|
| 50–59 yrs||32,242 (28)|
| 60–69 yrs||19,868 (17)|
| 70–79 yrs||13,718 (12)|
| ≥ 80 yrs||3632 (3)|
| Marital status||109,624|
| Married/living as married||75,073 (68)|
| Not married||34,551 (32)|
| High school or less||42,587 (39)|
| Some college or technical school||31,486 (29)|
| College or postcollege graduate||35,058 (32)|
| Health insurance||109,023|
| None||2908 (3)|
| Any||106,115 (97)|
|Risk characteristics|| |
| Menopausal status||115,995|
| Premenopausal/perimenopausal||43,703 (38)|
| Postmenopausal||72,292 (62)|
| Body mass index||102,527|
| < 25 kg/m2 (under/normal)||46,626 (45)|
| 25–29 kg/m2 (overweight)||31,510 (31)|
| 30–34 kg/m2 (obesity I)||14,920 (15)|
| ≥ 35 kg/m2 (obesity II–III)||9471 (9)|
| Family history of breast carcinoma||116,480|
| None||79,936 (69)|
| Any||34,584 (30)|
| Not sure||1658 (1)|
| Age at menarche||108,765|
| ≤ 12 yrs||50,897 (47)|
| 13 yrs||32,979 (30)|
| 14 yrs||14,057 (13)|
| ≥ 15 yrs||10,832 (10)|
| Age at first birth||94,110|
| < 20 yrs||19,266 (20)|
| 20–24 yrs||39,838 (42)|
| 25–29 yrs||22,547 (24)|
| 30–34 yrs||8955 (10)|
| ≥ 35 yrs||3504 (4)|
| No. of live births||107,298|
| 0||10,306 (10)|
| 1||12,546 (12)|
| 2||31,887 (30)|
| 3||24,170 (23)|
| 4||14,029 (13)|
| ≥5||14,360 (13)|
Table 2. Age-Appropriate (≥ 40 yrs) Women Eligible for Mammography Screening in New Hampshire using 2000 Census Data Compared with New Hampshire Mammography Network Screenees and Non-New Hampshire Mammography Network Screenees: May 1, 1996–December 31, 2000
|40–49 yrs||73,374||46,919 (64)||5874 (9)||20,581 (28)|
|50–59 yrs||51,999||32,343 (62)||4044 (8)||15,612 (30)|
|60–69 yrs||32,972||19,868 (60)||2480 (8)||10,624 (32)|
|70–79 yrs||27,516||13,718 (50)||1702 (6)||12,096 (44)|
|≥ 80 yrs||19,223||3632 (19)||441 (2)||15,150 (79)|
|Total/average||205,084||116,480 (57)||14,541 (7)||74,063 (36)|
We also examined the time interval between screening examinations (Table 3) and found that 65% of New Hampshire women were adhering to a recommended screening interval between approximately 1–2 years. Forty-four percent of women were adhering to screening within 14 months, and 21% were adhering to screening within 15 months and 26 months. Thirty-five percent of women had only 1 or 2 mammograms and then did not return within at least 27 months, an interval that is longer than the current recommendation for routine screening in the U.S.1 Screening adherence rates by decade of age at the 14-month interval were from 36% for women ages 40–49 years to 49% for women ages 50–59 years. Adherence to screening at the 15–26 month interval ranged from 16% among women age 80 years and older to 23% among women ages 40–49 years.
Table 3. Adherence Status among Women in the New Hampshire Mammography Network: May 1, 1996–December 31, 2000
|40–49 yrs||46,919||16,899 (36)||10,952 (23)||19,068 (41)|
|50–59 yrs||32,343||16,003 (49)||6751 (21)||9589 (3)|
|60–69 yrs||19,868||10,340 (52)||3977 (20)||5551 (28|
|70–79 yrs||13,718||6430 (47)||2728 (20)||4560 (33)|
|≥ 80 yrs||3632||1396 (38)||575 (16)||1661 (46)|
|Total||116,480||51,068 (44)||24,983 (21)||40,429 (35)|
The current results are is important, because we used 2000 Census data for population estimates stratified by age and compared those data with actual mammographic screening events ascertained from a representative, statewide screening mammography registry. This allowed us to assess time intervals between screening examinations. Several studies have assessed adherence to mammography screening and have reported a broad range in the use of mammography using different research methods,6–10 which may be influenced by the accuracy of self-report. We know of very few studies that used actual population-based Census data compared with actual mammographic events in diverse, community-based screening facilities. We estimated that overall utilization of mammography was 64%. Results of the 2000 New Hampshire Behavioral Risk Factor Surveillance System data (NH-BRFS) indicate that 82% of New Hampshire women age 50 years and older reported that they had undergone a mammogram within the past 2 years.
Cronin et al.15 conducted a modeling study to assess the dissemination of mammography in the U.S. using data from the National Cancer Institute's Breast Cancer Surveillance Consortium,16, of which NHMN is a member, for 2001. The current study data concurred with estimates generated using the model proposed by Cronin et al., who estimated that 56.7% of women ages 40–49 years would undertake screening at either an annual or biennial interval. In the current study, we found that this proportion was 59%. For women ages 50–59 years, according to Cronin et al., the annual/biennial screening rate was 70.8 whereas in the current study, it was 70%. Among women ages 60–69 years, the estimate by Cronin et al. was 70.7% whereas that of the current study was 72%; among women ages 70–79 years, their estimate was 67.3% and the estimate from the current study was 67%. Finally, among the oldest age group of women (80 yrs and older), Cronin et al. estimated an annual/biennial screening rate of 57.2% compared with 54% in the current study. This correlation supports the model of Cronin et al. and provides confidence in our own assessment of mammography utilization in New Hampshire.
This assessment of mammography in New Hampshire identified two issues of concern. First, we believe that the NH-BRFS data overestimate the utilization of mammography in New Hampshire. Using the NHMN, we found that adherence to screening was much lower than the 80% reported in NH-BRFS. Second, we found that adherence to recommended screening intervals also was lower than what currently is endorsed. These findings were lower compared with what has been reported in many other studies of mammography use that have reported use rates of between 65% and 85%, although this rate is much lower in several underserved populations.17, 18 However, reports of higher rates tended to be based on self-reported screening events, like the Behavioral Risk Factor Survey.19 Other studies of defined populations receiving care at health maintenance organizations also have reported mammography utilization at higher rates. Those studies also use actual screening events, which may be more reliable than self-reports. At least 1 study10 compared self-reported mammography use with actual events among 480 women in a managed-care setting. Those investigators found a high correlation (88.4%) between self-reported use and actual events. Their findings may indicate that women who know their actual use can be validated easily and are more truthful in their responses to a survey on this topic. In addition, these kinds of reports in diverse community settings may not reveal such findings.
An important group of women identified in this study was women who had one or two mammograms and who then did not return for any subsequent mammograms (irregular screenees). This group represented women who may be most reticent regarding routine mammography screening. Several studies have examined factors associated with adherence to screening20–24 and found that previous negative mammography experiences; body mass index20, 21; and fear, worry, or anxiety22–24 may affect adherence to screening.
In 1 retrospective, cross-sectional study,25 we used the NHMN to identify and recruit > 500 women age 50 years and older. Approximately 50% of those women adhered to mammography screening (within 24 mos), and approximately 50% and had not. The characteristics (objective and perceived risk factors) and psychologic profiles of the women in each group (274 adherents and 265 nonadherents) were compared. Briefly, body mass index was found to be associated with less adherence to screening, and the use of hormone-replacement therapy was associated with more adherence to screening.25 Although we had hypothesized that anxiety, dread, or misperceptions concerning risk would influence adherence, our findings did not support this hypothesis. It is noteworthy, however, that nonadherent women were more likely to report a negative experience with mammography that involved pain or discomfort during the procedure. In addition, 30–40% of participants reported that communication by technologists was not valuable or was neutral to them.
A recent meta-analysis of interventions that were tested to motivate those most resistant to screening found that enhancing access and counseling individual patients in health care settings had the highest impact on adherence to mammography screening.26 Other studies have found conflicting results regarding counseling interventions.27–29 Clearly, this is an area for continued research. One important tool for a future analysis may be the use of geocoding to assess where women who live in underserved areas go for their mammogram and whether adherence to screening at the level of the individual woman is affected by distance to facilities in underserved areas. In our data, facility zip codes revealed that 51.5% of mammography facilities in New Hampshire were state or federally designated underserved areas. Unfortunately, undertaking the geocoding and assessment was beyond the scope of the current study.
The findings of the current study regarding the use and periodicity of mammography by patient age both reveal new insights and confirm prior reports. The decrease in screening utilization with increasing age is similar to other, previously studies,8 in which it was reported that older women were more likely to decrease their use of screening over time. We found that 54% of elderly women (80 yrs and older) had been screened within the past 2 years, which is comparable to the 40% reported by Ostbye et al.8 The upper age limit of screening mammography remains undefined and is somewhat controversial, most likely because the incidence of breast carcinoma continues to rise, but more severe competing comorbidities call into question the need for early detection of breast carcinoma. Perhaps more importantly, we found that screening compliance was fairly consistent in patients ages 40–69 years and did not drop off significantly until age 70 years.
We also observed some interesting trends in the adherence to different screening intervals by age decade. The frequency of yearly screening was consistent between ages 50–79 years but dropped off at the extremes of age. The lower rate in young women in their 40s may be explained by conflicting medical societal recommendations for either yearly or biennial screening, whereas the similar low rate in elderly women likely reflects the issue of competing health demands.
Some relevant caveats should be mentioned here. Our registry's data collection process is voluntary and is based on the good will of mammography technologists, radiologists, and staff in pathology laboratories in New Hampshire. With the volume and complexity of care involved in mammography screening and follow-up care, it is impossible to avoid missing data completely. However, we specifically explored sources of missing data that were relevant for registry management and the effects these data may have on studies of mammographic accuracy and adherence to screening, which were minimal in both cases. In addition, our overall estimate of 64% utilization depended on the size of the window we used. We identified all age eligible women who had mammography screening and followed them for at least 27 months to assess their adherence. The primary window of this assessment was 4.5 years, with an additional 27 months of follow-up (total window, 6.7 yrs), a stable estimate for the period assessed.
Routine mammography screening may be occurring less often than believed when survey data alone are used. An important, compelling concern is the reason why women did not return for additional screening after only one or two mammograms. This issue deserves additional research.