SEARCH

SEARCH BY CITATION

Keywords:

  • mammography;
  • breast cancer;
  • cancer survivors;
  • retrospective cohort study;
  • administrative databases

Abstract

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

BACKGROUND

Improvements in cancer detection and treatment have resulted in increasing numbers of breast cancer survivors. Information regarding the use of mammography by breast cancer survivors is limited.

METHODS

The use of surveillance mammography was examined over a 5-year period in a retrospective cohort of women age ≥ 55 years who were diagnosed with incident primary breast cancer (1996–1997) while enrolled in 1 of 4 geographically diverse integrated health systems.

RESULTS

Of the 797 women included in the study, 80% (n = 636) underwent mammograms during the first year after treatment for breast cancer. The percentage of women having mammograms during each yearly period decreased significantly over time. In multivariable analyses, older women with comorbid illnesses or those with late-stage disease were less likely to undergo mammograms, whereas those who underwent breast-conserving therapy (adjusted odds ratio [OR] of 1.38 [95% confidence interval (95% CI), 1.09–1.75]) were more likely to have mammograms. Women who had outpatient visits with a gynecologist (adjusted OR of 3.49 [95% CI, 2.55–4.79]), or a primary care physician (adjusted OR of 2.21 [95% CI, 1.73–2.82]) during the year were more likely to undergo mammograms in that year.

CONCLUSIONS

The use of mammography among breast cancer survivors declines over time. Efforts are needed to increase awareness among healthcare providers and breast cancer survivors of the value of follow-up mammography. The current findings highlight the importance of maintaining ongoing contact with primary care physicians and gynecologists. Cancer 2006. © 2006 American Cancer Society.

Improvements in detection and treatment outcomes have resulted in increasing numbers of breast cancer survivors over the last 2 decades.1, 2 There are reported to be approximately 2.3 million women in the U.S. who were previously treated for breast cancer.2 Compared with those with no prior history of breast cancer, women with such a history have an approximately 3-fold increased risk of cancer in the contralateral breast.3, 4 This elevated risk of a second primary cancer combined with increasing survival rates after diagnosis of breast cancer underscores the importance of regular mammography in women with a history of breast cancer. In addition, the increasing use of breast-conserving therapy5–9 along with the risk of local recurrences after this therapy10 emphasizes the need for regular surveillance mammography.

Currently, to our knowledge, annual mammography is the only routine surveillance examination recommended in the follow-up of women with a history of breast cancer to detect local recurrences and/or contralateral breast cancer.11, 12 Population-based screening for cancers of the breast have been shown to reduce the risk of death from breast cancer13, 14 and routine mammograms after breast cancer diagnosis have been found to detect subsequent contralateral disease at an earlier stage than the initial breast cancer.15–17

Although underutilization of mammography services is well documented for the general population,18, 19 to our knowledge only a few studies to date have examined the utilization of mammography by survivors of breast cancer.20–23 In addition, little is known regarding factors associated with the use of mammography in this population, especially in settings in which health insurance is not a barrier. Previous studies examining the use of mammography among breast cancer survivors have focused mainly on Medicare beneficiaries,20 or in single regions in the U.S.21, 23 Enrollees of integrated health systems have healthcare insurance that includes mammography as a covered benefit and include women younger than those typically enrolled in Medicare. Therefore, integrated health systems provide a unique setting to examine factors associated with the use of mammography that is independent of the effects of health insurance. The objective of the current study was to examine the patterns and factors associated with the use of surveillance mammography among female breast cancer survivors enrolled in four integrated health systems.

MATERIALS AND METHODS

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

Population and Setting

Data for this study were derived from the Cancer Surveillance in HMO Administrative Data project, an affiliated project of the Cancer Research Network. The Cancer Research Network is a consortium of research organizations affiliated with not-for-profit integrated health systems and the National Cancer Institute.24 The Cancer Surveillance in HMO Administrative Data project identified a retrospective cohort of 897 women age ≥55 years enrolled in 4 geographically diverse integrated health systems who were diagnosed with incident primary breast cancer between January 1, 1996, and December 31, 1997. The diagnosis of breast cancer was confirmed through standardized medical record reviews by trained abstractors using predefined criteria at 2 sites and from institution cancer registries at the other 2 sites. Interobserver reliability assessment found high levels of agreement (κ = 0.70–1.00) between abstractors on all data elements assessed. Patients known to have undergone bilateral mastectomy (n = 12) were excluded from this report. We restricted this report to those who remained alive and enrolled with their respective integrated health systems for at least 2 years after the date of diagnosis. For each year of follow-up, women were excluded from the cohort if they died or disenrolled. Therefore, we had declining numbers of patients during each of the years of follow-up after cancer treatment, as shown in Figure 1. Data for the fifth year of follow-up was only available for eligible women diagnosed in 1996 (n = 262). The Institutional Review Boards of the participating healthcare organizations approved the study.

thumbnail image

Figure 1. Identification of the retrospective cohort breast cancer survivors among women diagnosed with breast cancer 1996–1997. *Women diagnosed in 1997 were censored because utilization information was not available for the fifth year of follow-up.

Download figure to PowerPoint

Study Parameters

Information abstracted from the medical records was linked to the computerized information systems of the healthcare organizations that included utilization and enrollment data. The use of mammography was determined using codes from the Common Procedural Terminology (76090, 76091, and 76092), the International Classification of Diseases, Ninth Edition, Clinical Modification (87.36, 87.37), and the Health Care Common Procedure Coding System (G0202) in electronic administrative data from outpatient services. Because patients were likely to be under active treatment and/or surveillance for breast cancer during the first year after incident diagnosis, we defined the end of the first year after diagnosis (treatment period) as time zero. We therefore determined the use of yearly mammography for each of the 5 years of follow-up beginning in the 13th month after diagnosis. A patient was considered to have had a mammogram if the administrative databases contained a code for mammography during a 12-month (1-year) period. To account for minor deviations from this strict definition, we also defined the use of mammography according to a more dynamic schedule at 9-month to 15-month intervals. Electronic utilization information was available from the date of diagnosis through December 31, 2002, providing information on the use of mammography for up to 5 completed years of follow-up after the treatment period for those diagnosed in 1996 and for up to 4 years for those diagnosed in 1997. Among all eligible women, the average follow-up time after the initial treatment period was 3.84 years.

Information regarding age at the time of diagnosis (categorized as 55–64 years, 65–74 years, or ≥75 years), date of diagnosis, race/ethnicity (categorized as non-Hispanic white, non-Hispanic black, or other), comorbid medical conditions, tumor stage at the time of diagnosis, and breast cancer-related therapies received, were obtained through chart review or from tumor registry data. Surgical therapies received were categorized as mastectomy, breast-conserving therapy, or other (for those whose surgical treatment modality was unknown or missing). Tumor stage was categorized as in situ, AJCC Stage I, Stage II, Stage III/IV, or unknown. We derived the Charlson comorbidity index25 (categorized as 0, 1, or 2 and higher) for each patient using information on comorbid medical conditions abstracted from the medical records in the 2-year period before cancer diagnosis. To examine the impact of physician specialty on the use of mammography, we identified the specialty of physicians (categorized as gynecologist, primary care physician, or other) associated with outpatient visits for women in the study using physician specialty codes contained in electronic administrative records. These outpatient office visit categories were defined for each year of follow-up in a mutually exclusive, hierarchical fashion to account for visits to multiple physicians of varying specialties during each of the study years. We categorized an outpatient visit as “gynecologist” if at least 1 visit with a gynecologist was identified in the electronic administrative data during a specified study year, irrespective of visits to physicians of other specialties. Of the remaining women, those with at least 1 outpatient visit with a family physician, an internist, and/or a geriatrician during a study year were then categorized as “primary care physician.” The remaining women, including those for whom the specialty of the physician associated with an outpatient visit was unknown, were categorized as “other.”

Statistical Analyses

We determined trends in the use of yearly mammography during the follow-up period under study (years 1–5 after cancer treatment) using graphical plots and the chi-square test for trend. We also determined the percentage of women having at least 1 mammogram per year for each of the years of follow-up after cancer treatment. Differences between patients who had mammograms and those who did not were assessed using the chi-square test or Student t-test, as appropriate.

We examined the association between selected characteristics and the use of yearly mammography over the 5-year period using generalized estimating equations (GEE) with a logit link based on a binomial distribution. We used an unstructured correlation covariance approach based on examination of the correlation matrix. The GEE method accounts for the correlation of repeated measurement and clustering within healthcare organizations, uses all available data points, and is sensitive to the pattern of change over time.26, 27 This model also provides an estimate of the effect of time elapsed since breast cancer treatment on the use of mammography. Covariates in the model included age, race/ethnicity, comorbidity, tumor stage at diagnosis, breast cancer-related therapies received, years of follow-up after cancer treatment, and physician specialty categories for outpatient visits. The age and physician specialty categories were updated for each year of follow-up after cancer treatment. Because we observed a linear dose-response relation between years of follow-up after treatment and the use of yearly mammography, we entered years of follow-up as a continuous variable in the final model. We also examined 2-way interaction effects between age and race, comorbidity, physician specialty, and years of follow-up, and between years of follow-up and comorbidity, race, and physician specialty, respectively. We found a significant interaction effect between age and comorbidity. Therefore, we used indicator variables for the interaction between age and comorbidity in our final model. Consideration of the other interactions did not change our results. Women with early-stage breast cancer have better long-term prognosis than those with late-stage disease. Therefore, we repeated the multivariable analyses in the subgroup of women diagnosed with in situ, Stage I, or Stage II tumors. To examine the impact of loss-to-follow-up in the fifth year under study, we repeated these analyses by restricting the sample to the first 4 years of follow-up. Consideration of year of cancer diagnosis and other breast cancer-related therapies (hormone, radiation, and chemotherapy) in our regression models did not change our findings.

The analyses for this study were generated using SAS software (version 9.1) of the SAS System for Windows (SAS Institute, Inc., Cary, NC), and STATA (version 8.2; StataCorp, College Station, TX).

RESULTS

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

Of the 797 women included in the study, 36% were between 55 and 64 years of age and the mean age was 69.7 (±8.7) years. The majority of the women were non-Hispanic white (82%), diagnosed with in situ or Stage I/II tumors (88%), and had no documented comorbid medical conditions (65%) (Table 1). Mastectomy was performed in approximately 40% of the women and 55% had breast-conserving therapy. Each year, between 22% and 26% of the women had outpatient visits with gynecologists (Table 2). The majority of the women had outpatient visits with a family physician or internist each year. Outpatient visits with a primary care physician increased, whereas visits to gynecologists decreased slightly over time.

Table 1. Characteristics of Breast Cancer Survivors at Time of Diagnosis, 1996–1997
CharacteristicNo.%*
  • *

    Due to rounding error, the sum of the column percentages may not equal 100.

Age, y  
 55–6428736.0
 65–7427434.4
 ≥7523629.6
Race  
 White, non-Hispanic65482.1
 Black8110.2
 Other627.8
Charlson comorbidity index  
 052065.2
 115920.0
 2+11814.8
Tumor stage at diagnosis  
 In situ13116.4
 Stage I42152.8
 Stage II15219.1
 Stage III/IV718.9
 Unknown222.8
Surgical therapy within 1 year of diagnosis  
 Breast-conserving43855.0
 Mastectomy31839.9
 Other415.1
Years of follow-up  
 1797100.0
 273291.8
 366883.8
 460475.8
 526232.9
Table 2. Distribution of Age and Physician Outpatient Visits by Years of Follow-Up
CharacteristicsYears after Breast Cancer Treatment
1 (n=797)2 (n=732)3 (n=668)4 (n=604)5 (n=262)
No.%*No.%*No.%No.%No.%
  • *

    Due to rounding error, the sum of the column percentages may not equal 100.

Age group, y          
 55–6423529.519126.115322.911118.44115.7
 65–7428035.126035.523034.420533.98632.8
 ≥7528235.428138.428542.728847.713551.5
Physician office visit          
 Gynecologist17622.118925.816224.314223.55320.2
 Family physician/internist43454.543859.842363.340166.418570.6
 Others18723.510514.38312.46110.1249.2

A total of 80% of women had mammograms during the first year after breast cancer treatment (Fig. 2). The percentage of women having mammograms decreased over time. During the fifth year after breast cancer treatment, only 63% of the women had mammograms. A similar pattern was found when use of mammography was determined at a less conservative 9-month to 15-month interval with the percentage undergoing mammography declining from 87% during the first year to 63% during the fifth year. Of those with at least 4 years of follow-up after breast cancer treatment (n = 604), only 45% (n = 269) consistently received mammography during each of the 4 years. Among women with 5 years of follow-up (n = 262), only 33% (n = 87) consistently underwent a yearly mammography. The findings were similar when use of mammography was determined using the less conservative 9-month to 15-month interval or when the analyses were restricted to women with in situ or early stage (Stage I/II) tumors.

thumbnail image

Figure 2. Percentage use of yearly mammography by years since the completion of breast cancer treatment. (Test for trend, P<.001.)

Download figure to PowerPoint

We modeled the factors associated with yearly mammography using the generalized estimating equations approach. In our initial models, we found that women age ≥ 75 years (adjusted odds ratio [OR] of 0.63; 95% confidence interval [95% CI], 0.48–0.84) and those with a Charlson comorbidity index of ≥2 (adjusted OR of 0.58; 95% CI, 0.43–0.80) were less likely to have yearly mammography. Because we found significant interaction effects between age and comorbidity, we used indicator variables for the interaction between these variables in our final models. We found no difference in the use of mammography between women ages 55 years to 64 years and those ages 65 years to 74 years, so we used a dichotomous age variable (ages 55–74 years vs. age ≥75 years) in the final model. In this model, we found an inverse correlation between yearly mammography and the combined effects of increasing comorbidity and older age (Table 3). Among women age ≥ 75 years, those with higher levels of comorbidity were significantly less likely to have mammograms, whereas an increasing comorbid disease burden did not significantly impact the use of mammography among women ages 55 years to 74 years. This model corroborated the decline in use of yearly mammography over time after cancer treatment (adjusted OR per year of follow-up of 0.78; 95% CI, 0.73–0.83). Women undergoing breast-conserving therapy were more likely to have mammograms than those undergoing mastectomy. The strongest predictors for the use of yearly mammography were an outpatient visit to a gynecologist (adjusted OR of 3.50; 95% CI, 2.55–4.80) or a primary care physician (adjusted OR of 2.22; 95% CI, 1.73–2.83) during the year. We found similar results when the regression analyses were repeated with use of mammography determined at the less conservative 9-month to 15-month interval. The results were also similar when the regression analyses were restricted to women with in situ or Stage I/II tumors (data not shown). In subgroup analyses among subjects with 4 years of follow-up the results were unchanged (data not shown).

Table 3. Adjusted Relative Odds of Receipt of Mammography Among Breast Cancer Survivors
 Adjusted ORs* (95% CI)
  • ORs: odds ratios; 95% CI: 95% confidence interval.

  • *

    The model was based on the conservative definition of mammography use at every 12-month interval and included all the variables listed. The model included all women in the study and also was adjusted for study site.

Age (y) and Charlson comorbidity index 
 55–74 and 
  Comorbidity index of 01.00 (referent)
  Comorbidity index of 11.05 (0.72–1.51)
  Comorbidity index of 2+0.97 (0.59–1.59)
 ≥75 and 
  Comorbidity index of 00.79 (0.60–1.04)
  Comorbidity index of 10.53 (0.37–0.75)
  Comorbidity index of 2+0.35 (0.23–0.52)
Race 
 White, non-Hispanic1.00 (referent)
 Black0.98 (0.65–1.49)
 Other1.32 (0.89–1.95)
Tumor stage at diagnosis 
 In situ1.00 (referent)
 Stage I0.93 (0.67–1.28)
 Stage II0.80 (0.54–1.18)
 Stages III/IV0.59 (0.37–0.94)
 Unknown0.52 (0.31–0.86)
Surgical therapy within a year of diagnosis 
 Mastectomy1.00 (referent)
 Breast-conserving surgery1.38 (1.09–1.75)
 Other0.72 (0.42–1.23)
Physician office visit 
 Gynecologist3.49 (2.55–4.79)
 Family physician/internist2.21 (1.73–2.82)
 Other1.00 (referent)
 Years of follow-up (per year)0.78 (0.73–0.83)

DISCUSSION

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

This population-based study of insured women for whom mammography is a covered benefit found that the majority of breast cancer survivors had mammograms performed during the first year after treatment. However, the percentage of women undergoing yearly mammograms decreased steadily over time. Older women, particularly those with comorbid illnesses, were less likely to have mammograms after treatment for breast cancer. Women who had outpatient visits to a gynecologist or a primary care physician were much more likely to have yearly mammography.

Previous studies have reported underuse of surveillance mammography by breast cancer survivors.20–22 Schapira et al.20 found that only 62% of Medicare beneficiaries had annual mammography claims during the first 2 years after initial treatment of early breast cancer. Our study found higher rates but the inclusion of multiple years of follow-up demonstrated a steady decline over time. The higher rates of surveillance mammography use found in the current study may reflect better adherence in settings of integrated health systems in which this service is a covered benefit and in which systems are in place to promote follow-up testing after cancer treatment. Decreasing concern about the risk of breast cancer over time and problems in the transition of care from oncologists to primary care physicians may be possible explanations for declining mammography use after breast cancer treatment.

Several sociodemographic (e.g., age, race/ethnicity, and insurance status), clinical (e.g., comorbid conditions), and health system (e.g., type of clinician) determinants of cancer screening have been identified for the general population,19, 28 but to our knowledge little is known regarding predictors of cancer screening among cancer survivors. We found that among breast cancer survivors, women age ≥ 75 years, particularly those with comorbid medical conditions, were significantly less likely to have yearly mammograms. This is similar to findings in the general population.19, 28 Competing risks for death increase with advancing age and increased disease burden, leading to uncertainties regarding the benefits of mammography in older women.29 However, population-based studies have found that older women undergoing screening mammography testing had earlier stage cancers and were less likely to die of their disease.16, 17, 30 We also found that women with late-stage tumors were less likely to receive follow-up mammography. This finding is surprising, because those with late-stage breast cancer are at a greater risk of recurrent disease. We did not find any significant differences in the receipt of yearly mammography between non-Hispanic black and non-Hispanic white women, in contrast to the results of previous studies set in a general insured population.31 However, the current study included a relatively small number of non-Hispanic black women, resulting in wide 95% confidence intervals around our estimate.

The strongest association with surveillance mammography in the current study was found with outpatient visits to gynecologists or primary care physicians. Although this finding is similar to previous reports concerning the effect of physician specialty on mammography screening,19 to our knowledge this is the first study to report this association in a population of breast cancer survivors. The primary explanation may be that women who visit gynecologists or primary care physicians have greater awareness of their preventive care needs. This finding suggests that involving primary care physicians and gynecologists in the follow-up of breast cancer survivors and improvements in the transition of care may be beneficial.

There are several specific interventions that may increase use of surveillance mammography among breast cancer survivors. In general, cancer care is often fragmented.32 Therefore, mechanisms to improve coordination of care between cancer specialists and primary care physicians are needed that includes defined roles and responsibilities for surveillance testing. In addition, breast cancer survivors and their primary care physicians should receive a comprehensive care summary and follow-up plan at the completion of primary cancer treatment, as recommended by the Institute of Medicine report on the care of cancer survivors.32 Previous studies have shown that physician recommendation is a strong predictor of cancer screening.19 Thus, increasing awareness of the need for regular surveillance mammography among primary care physicians may be beneficial.

Strengths and Limitations

This population-based study was conducted using data derived from administrative databases associated with integrated health systems in which all subjects were insured. Therefore, the findings are independent of the effects of differential healthcare insurance. We had complete utilization data regarding all women included in the study during the period examined, reducing the impact of information bias. We were also able to examine 5-year follow-up trends in the use of yearly mammography. Our data analytic approach allowed us to include all available data points while accounting for change over time. The current study included women age ≥ 55 years, in contrast to previous studies limited to Medicare enrollees. Therefore, our findings are more generalizable than those from previous reports.

A limitation of electronic administrative data is the impossibility of determining whether mammograms were intended for surveillance or were undertaken in response to symptoms. An unknown portion of the mammograms we identified may not indicate ongoing breast cancer surveillance. In addition, we did not have information regarding the socioeconomic status of women included in the study; even within this insured population, socioeconomic status may impact the receipt of mammography.19

Conclusion

There was a progressive decline over time in the use of mammography among this cohort of breast cancer survivors. Efforts are needed to increase awareness among healthcare providers and breast cancer survivors on the value of follow-up mammography. The current study findings highlight the importance of maintaining ongoing contact with primary care physicians and gynecologists.

Acknowledgements

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

The authors thank Ann M. Hanson, programmer/analyst at HealthPartners Research Foundation, and Maribet C. McCarty, Ph.D., R.N., whose hard work on the IMPACT study made the research project possible, and Wenjun Li, Ph.D., biostatistician at the University of Massachusetts, for assistance with data analysis.

REFERENCES

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