Advertisement

Screening Mammography in a Public Hospital Serving Predominantly African-American Women: A Stage–Survival–Cost Model

Published:April 21, 2015DOI:https://doi.org/10.1016/j.whi.2015.02.006

      Abstract

      Background

      Ethnic and socioeconomic disparities pervade breast cancer patterns and outcomes. Mammography guidelines reflect the difficulty in optimizing mortality reduction and cost-effectiveness, with controversy still surrounding the 2009 U.S. Preventive Services Task Force (USPSTF) recommendations. This study simulates USPSTF and American Cancer Society (ACS) guidelines' effects on stage, survival, and cost of treatment in an urban public hospital.

      Methods

      Charts of 274 women diagnosed with stage I, II, or III breast cancer (2008–2010) were reviewed. Published tumor doubling times were used to predict size at diagnosis under simulated screening guidelines. Stage distributions under ACS and USPSTF guidelines were compared with those observed. Cohort survival for observed and hypothetical scenarios was estimated using national statistics. Treatment costs by stage, calculated from Georgia Medicaid claims data, were similarly applied.

      Results

      Mean age at diagnosis was 56 years. African Americans predominated (82.5%), with 96% publically insured or uninsured. Simulated stages at diagnosis significantly favored ACS guidelines (43.1% stage 1/38.3% stage 2/9.9% stage 3 vs. USPSTF 23.0%/53.3 %/15.0%), as did 5-year survival and cost of treatment relative to both observed and USPSTF-predicted schema (p < .0001). Following USPSTF guidelines predicted lower survival and additional costs.

      Conclusions

      Following ACS guidelines seems to lead to earlier diagnosis for low-income African-American women and increase 5-year survival with lower overall and breast-specific costs. The data suggest that adjusting screening practices for lower socioeconomic status, ethnic minority women may prove essential in addressing cancer disparities.
      The advent of breast cancer screening in the 1970s established that mammography decreases breast cancer mortality (
      • Bjurstam N.
      • Björneld L.
      • Warwick J.
      • Sala E.
      • Duffy S.W.
      • Nyström L.
      • Wahlin T.
      • et al.
      The Gothenburg Breast Screening Trial.
      ,
      • Hellquist B.N.
      • Duffy S.W.
      • Abdsaleh S.
      • Björneld L.
      • Bordás P.
      • Tabár L.
      • Jonsson H.
      • et al.
      Effectiveness of population-based service screening with mammography for women ages 40 to 49 years: Evaluation of the Swedish Mammography Screening in Young Women (SCRY) cohort.
      ,
      • Miller A.B.
      • To T.
      • Baines C.J.
      • Wall C.
      The Canadian National Breast Screening Study-1: Breast cancer mortality after 11 to 16 years of follow-up. A randomized screening trial of mammography in women age 40 to 49 years.
      ,
      • Moss S.M.
      • Cuckle H.
      • Evans A.
      • Johns L.
      • Waller M.
      • Bobrow L.
      Trial Management Group. Effect of mammographic screening from age 40 years on breast cancer mortality at 10 years' follow-up: A randomised controlled trial.
      ,
      • Norman S.A.
      • Russell Localio A.
      • Weber A.L.
      • Coates R.J.
      • Zhou L.
      • Bernstein L.
      • Nadel M.R.
      • et al.
      Protection of mammography screening against death from breast cancer in women aged 40-64 years.
      ,
      • Shapiro S.
      • Strax P.
      • Venet L.
      Periodic breast cancer screening in reducing mortality from breast cancer.
      ,
      • Tabár L.
      • Vitak B.
      • Chen T.H.
      • Yen A.M.
      • Cohen A.
      • Tot T.
      • Duffy S.W.
      • et al.
      Swedish two-county trial: Impact of mammographic screening on breast cancer mortality during 3 decades.
      ), leading to widespread implementation aimed at earlier detection and treatment. However, in today's era of increasing health care costs (
      • Mariotto A.B.
      • Yabroff K.R.
      • Shao Y.
      • Feuer E.J.
      • Brown M.L.
      Projections of the cost of cancer care in the United States: 2010-2020.
      ) and enhanced consumer autonomy, programs are being reevaluated to weigh financial impact and potential harms against concrete health benefits. The American Cancer Society (ACS) and the U.S. Preventive Services Task Force (USPSTF) both publish screening recommendations based on such investigation, but differences in these statements in 2009 generated considerable controversy. The ACS guidelines endorse mammography starting at age 40 and continuing annually as long as health status and life expectancy allow (
      • Smith R.A.
      • Cokkinides V.
      • Brooks D.
      • Saslow D.
      • Shah M.
      • Brawley O.W.
      Cancer screening in the United States, 2011: A review of current American Cancer Society guidelines and issues in cancer screening.
      ). In contrast, USPSTF recommendations shifted significantly from 2002, when they roughly matched ACS policies (
      U.S. Preventive Services Task Force (USPSTF)
      Screening for breast cancer: U.S. Preventive Services Task Force recommendation statement.
      ). The update eliminated routine mammography in women aged 40 to 49 years (instead basing the decision on discussion between the patient and her primary care provider), increased the screening interval from 1 to 2 years, and withheld recommendations for women over 75 years. This policy update was informed by two major studies after years of debate about screening women in their 40s (
      • Green B.B.
      • Taplin S.H.
      Breast cancer screening controversies.
      ,
      • Salzmann P.
      • Kerlikowske K.
      • Phillips K.
      Cost-effectiveness of extending screening mammography guidelines to include women 40 to 49 years of age.
      ).
      The first group,
      • Nelson H.D.
      • Tyne K.
      • Naik A.
      • Bougatsos C.
      • Chan B.K.
      • Humphrey L.
      U.S. Preventive Services Task Force. Screening for breast cancer: An update for the U.S. Preventive Services Task Force.
      , used existing data from randomized, controlled trials to compare how mammography impacts the risk of breast cancer death in women of varying ages. The trials that were meta-analyzed by Nelson et al. emerged from the United Kingdom, Sweden, Canada, and New York State. The group found the same reduction in relative risk for breast cancer death (about 15%) for women in the 4th and 5th decades of life, but used mortality rates from the National Cancer Institute's Surveillance Epidemiology and End Results (SEER) database to predict that 1,904 women from the 39 to 49 group would need to be invited to screening to prevent one additional death versus 1,334 from the 50 to 59 group. By comparison, for women 60 to 69 years old (with consensus for screening), mammography offers a 30% risk reduction, and only 377 need to be invited to screening to save one life. The authors stated, however, that data applicability was only fair owing to a lack of U.S.-based studies and that the “number needed to invite to screening could be misleading if…risk for mortality…varied between studies,” a mortality variation that occurs in ethnic minority, socioeconomically disadvantaged women. Basing outcomes modeling solely on age-specific variables neglects evidence demonstrating marked ethnic and socioeconomic differences in tumor biology and disease patterns.
      The second group,
      • Mandelblatt J.S.
      • Cronin K.A.
      • Bailey S.
      • Berry D.A.
      • de Koning H.J.
      • Draisma G.
      • Feuer E.J.
      • et al.
      Breast Cancer Working Group of the Cancer Intervention and Surveillance Modeling Network. Effects of mammography screening under different screening schedules: Model estimates of potential benefits and harms.
      , modeled various screening schedules and age ranges to determine each models' relative efficiency in decreasing mortality. Although higher screening rates decreased mortality, in each model the benefit for each additional mammogram eventually became marginal. This point was deemed the “efficiency frontier,” and the authors concluded that programs that screen biennially from age 50 to 69, 74, or 79 years were among the most efficient. Like Nelson et al., however, the authors acknowledged that “the models…do not capture differences in outcomes among certain risk subgroups, such as women with BRCA1 or BRCA2 genetic susceptibility mutations…or black women who seem to have more disease at younger ages than white women.”
      These analyses focused on age as the primary determinant of cancer risk and concluded that screening biennially, starting at age 50, was most efficient from a national payer perspective. However, both African-American ethnicity and low socioeconomic status have been associated with inferior cancer outcomes, including younger age at diagnosis and increased mortality at the same age and stage (
      • Aragon R.
      • Morgan J.
      • Wong J.H.
      • Lum S.
      Potential impact of USPSTF recommendations on early diagnosis of breast cancer.
      ,
      • Cunningham J.E.
      • Montero A.J.
      • Garrett-Mayer E.
      • Berkel H.J.
      • Ely B.
      Racial differences in the incidence of breast cancer subtypes defined by combined histologic grade and hormone receptor status.
      ,
      • DuBard C.A.
      • Schmid D.
      • Yow A.
      • Rogers A.B.
      • Lawrence W.W.
      Recommendation for and receipt of cancer screenings among Medicaid recipients 50 years and older.
      ,

      Howlader N, Noone AM, Krapcho M, Neyman N, Aminou R, Waldron W, …, Edwards BK (eds.). SEER Cancer Statistics Review, 1975-2008, National Cancer Institute. Bethesda, MD, Available: http://seer.cancer.gov/csr/1975_2008/, based on November 2010 SEER data submission, posted to the SEER web site, 2011.

      ,
      • Gabram S.G.
      • Lund M.J.
      • Gardner J.
      • Hatchett N.
      • Bumpers H.L.
      • Okoli J.
      • Brawley O.W.
      • et al.
      Effects of an outreach and internal navigation program on breast cancer diagnosis in an urban cancer center with a large African-American population.
      ,
      • Klassen A.C.
      • Smith K.C.
      The enduring and evolving relationship between social class and breast cancer burden: A review of the literature.
      ,
      • Komenaka I.K.
      • Martinez M.E.
      • Pennington Jr., R.E.
      • Hsu C.H.
      • Clare S.E.
      • Thompson P.A.
      • Goulet Jr., R.J.
      • et al.
      Race and ethnicity and breast cancer outcomes in an underinsured population.
      ,
      • Lobb R.
      • Ayanian J.Z.
      • Allen J.D.
      • Emmons K.M.
      Stage of breast cancer at diagnosis among low-income women with access to mammography.
      ,
      • McBride R.
      • Hershman D.
      • Tsai W.Y.
      • Jacobson J.S.
      • Grann V.
      • Neugut A.I.
      Within-stage racial differences in tumor size and number of positive lymph nodes in women with breast cancer.
      ,
      • O'Brien K.M.
      • Cole S.R.
      • Tse C.K.
      • Perou C.M.
      • Carey L.A.
      • Foulkes W.D.
      • Millikan R.C.
      • et al.
      Intrinsic breast tumor subtypes, race, and long-term survival in the Carolina Breast Cancer Study.
      ,
      • Smith-Bindman R.
      • Miglioretti D.L.
      • Lurie N.
      • Abraham L.
      • Barbash R.B.
      • Strzelczyk J.
      • Kerlikowske K.
      • et al.
      Does utilization of screening mammography explain racial and ethnic differences in breast cancer?.
      ,
      • Ward E.M.
      • Fedewa S.A.
      • Cokkinides V.
      • Virgo K.
      The association of insurance and stage at diagnosis among patients aged 55 to 74 years in the national cancer database.
      ,
      • Yang R.
      • Cheung M.C.
      • Franceschi D.
      • Hurley J.
      • Huang Y.
      • Livingstone A.S.
      • Koniaris L.G.
      African-American and low-socioeconomic status patients have a worse prognosis for invasive ductal and lobular breast carcinoma: Do screening criteria need to change?.
      ). Evidence suggests that such factors may leave low-income African-American women at disproportionate risk of negative consequences from later, less frequent screening. Grady Memorial Hospital is a safety net hospital with more than 900 beds in downtown Atlanta, Georgia, serving predominantly lower-income, publically insured African-American patients whose outcomes may provide evidence to address this question. We modeled the potential impact of strict adherence to ACS or USPSTF screening recommendations on stage at diagnosis and expected survival for the women diagnosed with breast cancer at our AVON Foundation Comprehensive Breast Center. Given that previous studies have demonstrated variation in cost of cancer care with stage at diagnosis (
      • Subramanian S.
      • Trogdon J.
      • Ekwueme D.U.
      • Gardner J.G.
      • Whitmire J.T.
      • Rao C.
      Cost of breast cancer treatment in Medicaid: Implications for state programs providing coverage for low-income women.
      ,
      • Taplin S.H.
      • Barlow W.
      • Urban N.
      • Mandelson M.T.
      • Timlin D.J.
      • Ichikawa L.
      • Nefcy P.
      Stage, age, comorbidity, and direct costs of colon, prostate, and breast cancer care.
      ), we also analyzed potential differences in health care costs associated with any overall stage migration.

      Methods

      Patient Data Collection

      All patients diagnosed with breast cancer in the calendar years 2008 through 2010 were identified from the Grady Memorial Hospital tumor registry (n = 447). Exclusion criteria included in situ disease (n = 73), metastatic disease at diagnosis (n = 49), ipsilateral cancer recurrence (n = 23), duplicate entries of bilateral disease (n = 19), male breast cancer (n = 2), and missing information that prevented confirming the date of diagnosis (n = 7). For women with cancer diagnosed in both breasts during the study period, the earlier diagnosis was included as one occurrence.
      The final sample totaled 274 patients. Charts of each patient were reviewed for age at diagnosis, insurance status, ethnic group, clinical tumor stage based on the American Joint Committee on Cancer, 7th edition (AJCC-7), radiographic screening history and reasons for screening, pathologic tumor characteristics, tumor histology, and hormone receptor (estrogen receptor, progesterone receptor, and her2/neu) status. Study variable definitions are clarified below:

      Clinical stage (“observed stage”)

      Staging was primarily based on radiographic reports at the time of diagnosis and defined using the AJCC-7 TNM classification. Tumor size (T) was the largest single dimension on diagnostic ultrasonography. If not performed, or inaccurate owing to a very large lesion, diagnostic mammogram was used. Nodal status (N) was based on radiology reports from diagnostic mammography and ultrasonography, and confirmed with multidisciplinary conference notes, pathology reports of node aspiration, positron emission tomography scans, and/or physical examination findings (breast, radiation oncology, or medical oncology clinic notes). The clinical stage determined in this manner provided the “observed stage” for each patient.

      Screening history and reasons for screening

      The date and type of imaging performed during the screening and diagnosis process were recorded as well as the interval from last imaging to current mammogram. The reason for the patient's current mammogram was abstracted from the radiology report, including routine screening, workup of symptoms, or other.

      Tumor Growth Model and Predicted Stage Distributions

      • Peer P.
      • Dijck J.
      • Hendriks J.
      • Holland R.
      • Verbeek A.
      Age dependent growth rate of primary breast cancers.
      have previously published tumor doubling times based on measurements from serial mammograms of women in The Netherlands. The results provide the average number of days for a breast lesion to double in size based on the woman's age. Given that tumor biology is also associated with rate of growth, we further subdivided the doubling times to allow specification based on both age and tumor grade, with older age and lower grade tumors correlating with slower rates of growth (Table 1).
      Table 1Peer et al. Mean Tumor Doubling Times Adjusted for Tumor Grade (
      • Lund M.J.
      • Butler E.N.
      • Bumpers H.L.
      • Okoli J.
      • Rizzo M.
      • Hatchett N.
      • Gabram S.G.
      • et al.
      High prevalence of triple-negative tumors in an urban cancer center.
      )
      Age (y)Days, n (Range)
      Grade 1Grade 2Grade 3
      <50131 (114–147)96 (79–113)61 (44–78)
      50–70190.5 (177–204)163 (150–176)135 (121–149)
      >70243.5 (237–250)158 (179–237)149 (120–178)
      A linear regression model, previously reported by this study's authors (
      • Habtes I.
      • Friedman D.
      • Raskind-Hood C.
      • Adams K.
      • Becker E.
      • D'Orsi C.
      • Gabram S.
      • et al.
      Determining the impact of the United States mammography screening guidelines on patient survival in a predominantly African-American population treated in a public hospital during 2008.
      ), was derived from the estimates of doubling times and used to both “rewind” and “fast forward” patterns of tumor growth in each patient based on their lesion's size at the time of diagnosis. Using this equation, tumor size at 1-month intervals was calculated for 4 years retrospectively (predicted sizes had tumor been detected earlier) and prospectively (predicted sizes had tumor been detected later) for each patient. For patients less than 50 years old, the tumor size at age 50 was also separately calculated.
      Each patient's documented screening history was compared with that recommended by the ACS and the USPSTF, examining her age, the interval between mammograms, and the reason for testing. This was then used to simulate a ‘new time of diagnosis' and in turn, a corresponding stage at diagnosis ‘as if’ she had followed the ACS or USPSTF guidelines.
      For a woman diagnosed with breast cancer between the ages of 50 and 74, when both organizations agree on routine screening, the time since her last mammogram would suggest an advantage for one set of guidelines or the other. If the last screening mammogram occurred, for instance, 19 months before the current one, hypothetical adherence to ACS guidelines would prescribe screening 7 months earlier (at 1 year from the prior mammogram), whereas hypothetical adherence to the USPSTF guidelines would delay screening an additional 5 months (at 2 years from the prior mammogram). Thus, if the lesion was found on the basis of routine screening, the “observed stage” would be the clinical stage at the patient's actual time of diagnosis according to her medical record. Her “ACS stage” would be the stage had she been diagnosed 7 months earlier, and her “USPSTF stage” would be the stage had she been diagnosed 5 months later. If the lesion was found at the same interval owing to symptoms instead, it could hypothetically be diagnosed earlier by ACS guidelines. However, under USPSTF guidelines the time of diagnosis would be unchanged (rather than later), because the symptoms would have led to investigation before the next mammogram was due. In that case, “observed stage” and “USPSTF stage” would be the same, although “ACS stage” might be earlier. In contrast, in a woman diagnosed through either routine screening or symptomatic workup who had not undergone mammography in more than 2 years, adherence to either set of guidelines would be expected to provide an advantage. For women younger than 50 or older than 75, only ACS guidelines recommend routine screening. In these cases, depending on her mammography interval, a woman might be diagnosed earlier based on simulated ACS adherence, at the same time (found on symptoms), or later (found on screening) based on simulated USPSTF adherence. We assumed that there was no provider-recommended earlier screening for women under 40 years old following USPSTF guidelines.
      Once the “new time of diagnosis” according to each simulated set of guidelines was determined, expressed as months before or after the observed diagnosis, the calculated tumor size was selected based on the regression model. The threshold for mammographic detection was assumed to be 10 mm (
      • Pisano E.D.
      • Gatsonis C.
      • Hendrick E.
      • Yaffe M.
      • Baum J.K.
      • Acharyya S.
      • Rebner M.
      • Digital Mammographic Imaging Screening Trial (DMIST) Investigators Group
      • et al.
      Diagnostic performance of digital versus film mammography for breast-cancer screening.
      ).
      Although some percentage of tumors can be expected to manifest on imaging below this cutoff, it was chosen as a conservative estimation for modeling purposes. If the calculated tumor size at the simulated time of diagnosis was detectable, AJCC-7 T classification was assigned. Because of the inability to predict regional metastasis by this model, the nodal or N classification was set as zero. The threshold at which tumors become palpable, and therefore also assumed to be “symptomatic” and detectable without need for screening mammography, was set at 20 mm (
      • Güth U.
      • Huang D.J.
      • Huber M.
      • Schötzau A.
      • Wruk D.
      • Holzgreve W.
      • Zanetti-Dällenbach R.
      • et al.
      Tumor size and detection in breast cancer: Self-examination and clinical breast examination are at their limit.
      ) or stage IIA. These stages were recorded as “ACS stage” or “USPSTF stage” and comprised the two distributions of predicted stages at diagnosis.

      Survival, Cost, and Statistical Analyses

      Nationally observed percentages of patients surviving at 1 to 5 years from the time of diagnosis for stages I, II, and III breast cancer were requested from the National Cancer Database (

      American College of Surgeons. National Cancer Database survival report 2003. Online via Datalinks Portal for Commission on Cancer-accredited cancer programs. Available: https://datalinks.facs.org. Accessed by Sherita Hearn ( [email protected] ), tumor registry coordinator. Accessed May 11, 2011.

      ). Each patient's yearly likelihood of survival within each screening scenario was then tabulated based on their observed or predicted stage at diagnosis. The survival estimates for each group of patients was averaged by year to create three 5-year survival curves for the cohort: one each for their observed, ACS-predicted, and USPSTF-predicted distributions of stages at diagnosis.
      Cost measurements were derived from Georgia Comprehensive Cancer Registry case files that linked to a woman in the Georgia Medicaid administrative claims data for the study years. Eligibility for inclusion in these calculations was based on 18-month continuous enrollment in Medicaid. Total costs were calculated for breast diagnoses and procedures only as well as for costs including all claims. The estimates were stratified by AJCC-7 stage at diagnosis. As a result, the total cost for all the women in each stage distribution (observed, ACS, and USPSTF) could be obtained by multiplying the average cost of treatment for a given stage of breast cancer by the number of patients diagnosed at that stage within the distribution. This sum was averaged over the number of patients to provide per-patient cost estimates for all screening scenarios, with incremental savings or additional expenses per patient reported.
      Significance of the differences between comparison groups for stage at diagnosis, survival by year, and per-patient costs was assessed using Pearson χ2 contingency statistics.

      Results

      The observed distribution of stage at diagnosis among the initial group of 396 women is shown in Table 2. Stages excluded in the main analysis are presented for comparison with national statistics. Demographic, socioeconomic, and clinical characteristics of the included cohort (n = 274) are shown in Table 3. The average age at diagnosis was 56 years (range, 28–99 years). The cohort was predominantly African American (n = 226/274; 82.5%) and most were publically insured (Medicaid 33.6%/Medicare 24.1%) or uninsured (38.3%). Overall, about 32% of women were diagnosed before the age of 50. Hispanic/Latina women seem more likely to present before the 5th decade (72.2%) and to be uninsured (88.9%); however, the sample size is very small (n = 18).
      Table 2Complete Distribution of Observed Stages at Diagnosis in Study Cohort
      AJCC-7 Stagen%SEER Stagen%
      07318.4In situ7318.4
      19423.7Local15739.6
      212130.6Regional10526.5
      34711.9Distant4912.4
      44912.4
      Missing/unknown123.0Missing/unknown123.0
      Total396100Total396100
      Abbreviations: AJCC-7, American Joint Committee on Cancer, 7th edition; SEER, Surveillance Epidemiology and End Results.
      Table 3Demographic, Social, and Clinical Characteristics of Study Cohort
      CharacteristicEntire CohortEthnic Group
      n%African-American (%)Caucasian (%)Hispanic/Latina (%)Other (%)
      Age at diagnosis (y)
       Mean565757.646.951.1
       <508832.128.827.372.236.8
       50–7415857.761.572.722.236.8
       ≥75228.09.35.6
       Missing62.20.426.4
       Total274100100100100.0100.0
      Ethnic group
       African American22682.5
       Caucasian114.0
       Hispanic/Latina186.6
       Other196.9
       Total274100
      Insurance status
       Private82.93.19.1
       Medicare6624.127.918.25.6
       Medicaid9233.635.827.35.636.8
       None/self-pay10538.333.245.588.947.4
       Unknown/missing31.115.8
       Total274100100100100100
      Tumor size
       Mean size at diagnosis (mm)30.630.222.736.121.9
       <2 cm10839.442.954.516.722.2
       2–5 cm10739.138.945.561.122.2
       >5 cm3914.214.216.7
       Missing/unknown207.34.05.655.6
       Total274100.0100100100100
      Tumor stage
       AJCC-7
      19434.335.845.522.215.0
      212144.243.854.561.120.0
      34717.216.816.75.0
      Missing/unknown124.43.560.0
      Total274100100100100100
       SEER
      Local15757.358.063.661.130.0
      Regional10538.338.536.438.910.0
      Missing/unknown124.43.560.0
      Total274100100100100100
      Tumor grade
       13111.311.99.116.7
       213649.650.936.455.641.7
       39333.934.154.527.825.0
       Missing/unknown145.13.133.3
       Total274100100100100100
      Hormone receptor status
       ER and/or PR+15255.554.472.766.741.7
       Her2/neu + only186.67.19.18.3
       Triple negative8631.433.218.227.88.3
       Unknown/missing186.65.35.641.7
       Total274100100100100100
      Abbreviations: AJCC-7, American Joint Committee on Cancer, 7th edition; ER, estrogen receptor; PR, progesterone receptor; SEER, Surveillance Epidemiology and End Results.
      The average size of lesions at presentation was approximately 3 cm. Again, Hispanic/Latina women seemed to be at a disadvantage, with more 2- to 5-cm tumors (61.1% of lesions) than other ethnic groups (22.2%–45.5%). Most tumors presented at AJCC-7 stage 2 (44.2%) and demonstrated intermediate grade (49.6%). African-American women showed the smallest proportion of estrogen or progesterone receptor positivity (54.4% vs. 72.7% of Caucasian women and 66.7% of Hispanic/Latina women) as well as the heaviest burden of triple negative (estrogen, progesterone, and her2/neu receptor negative) disease (33.2% vs. 27.8% for Hispanic/Latina women and 18.2% in Caucasian women).
      Unfortunately, the most common screening history across all ethnicities (Table 4) was “no prior mammography” (29.2% overall); only 30.6% of women had a documented mammogram within the last 2.5 years. Most women were diagnosed after symptom onset (53.3% overall) rather than routine screening, a trend consistent across all ethnicities.
      Table 4Screening History and Reasons for Workup
      Prior ScreensEntire CohortEthnic Group
      n%African American (%)Caucasian (%)Hispanic/Latina (%)Other
      Mammogram >30 mo ago4717.217.727.322.2
      Mammogram in the last 2 years (18–29 mo)2810.210.69.15.616.7
      Mammogram in the last year (0–17 mo)5620.422.627.35.68.3
      No prior mammography8029.227.936.427.841.7
      Not documented6323.021.238.933.3
      Total274100100100100100
      Average interval since last mammogram (mo)33.232.439.1/25.7
      Alternate value when one outlier of 120 mo excluded.
      47.7/33.2
      Alternate value when one outlier of 120 mo excluded.
      13.7
      Reason for imaging
       Routine screening8832.135.827.322.28.3
       Symptoms14653.353.554.561.158.3
       Other reason176.27.19.1
       Unknown238.43.59.116.733.3
       Total274100100100100100
      Alternate value when one outlier of 120 mo excluded.

      Tumor Growth Model and Predicted Stage Distributions

      When the tumor growth model and predicted stage algorithms were applied to the observed stages at diagnosis, significant stage migrations resulted (Figure 1; Table 5). Simulating ACS guidelines yielded the highest proportion of stage I tumors (43.1%) versus the proportion of stage I tumors observed (34.3%) and predicted by simulating the USPSTF guidelines (23.0%), as well as decreased stage II and III disease. Hypothetically following USPSTF guidelines resulted in a substantial increase in predicted stage II disease (53.3%) compared with ACS (38.3% stage II) and observed (44.2% stage II).
      Figure thumbnail gr1
      Figure 1Stage migrations by alternative simulated mammography schedules. ACS, American Cancer Society; USPSTF, U.S. Preventive Services Task Force.
      Table 5Stages and Statistical Comparison of Observed Versus Simulated Mammography Schedules
      AJCC-7 StageScreening Schedules
      ObservedACSUSPSTF
      n%n%n%
      19434.311843.16323.0
      212144.210538.314653.3
      34717.2279.94115.0
      Missing124.4248.8248.8
      Total274100274100274100
      ComparisonChi Squarep Value
      Observed vs. ACS261.1<.0001
      Observed vs. USPSTF303.3<.0001
      ACS vs. USPSTF230.9<.0001
      Abbreviations: ACS, American Cancer Society; AJCC-7, American Joint Committee on Cancer, 7th edition.

      Survival Analysis

      Five-year survival estimations based on each strategy are presented in Figure 2. The ACS simulation predicted the highest survival at each of the first 5 years after diagnosis relative to both the observed stages at diagnosis (intermediate survival estimates) and the USPSTF stages (lowest survival estimates). Differences between scenarios were small (0.4%–2.3%) but significant.
      Figure thumbnail gr2
      Figure 2Predicted 5-year survival of observed and simulated cohorts. ACS, American Cancer Society; USPSTF, U.S. Preventive Services Task Force.

      Cost Analysis

      Calculated costs by stage are presented in Table 6. Both breast-related and all-claims costs were found to increase by $20,000 to $25,000 with each increasing stage of disease. When stage-specific costs were applied over distributions of stages according to observed, ACS-predicted, and USPSTF-predicted schema, simulated ACS guideline outcomes showed significant per-patient savings relative to both observed (savings of $3,745 per patient) and USPSTF costs (savings of $5,528 per patient). Simulating following the USPSTF guidelines generated the highest costs.
      Table 6Observed Costs by Stage and Predicted Incremental Costs by Screening Cohort
      Breast Procedures, Diagnoses OnlyAll Claimsp Value
      AJCC-7 stage ($USD per patient)
       122,95938,104
       241,79959,217
       366,19086,148
      Scenario comparison (Net $USD per patient)
       ACS vs. observed−37,45−4,171<.0001
       USPSTF vs. observed1,9522,188.0002
       ACS vs. USPSTF−5,528−6,173<.0001
      Abbreviations: ACS, American Cancer Society; USPSTF, U.S. Preventive Services Task Force.

      Discussion

      This modeling approach has been validated by the same authors in an initial study examining only changes to stage at diagnosis in women diagnosed at Grady Memorial Hospital in 2008 (
      • Habtes I.
      • Friedman D.
      • Raskind-Hood C.
      • Adams K.
      • Becker E.
      • D'Orsi C.
      • Gabram S.
      • et al.
      Determining the impact of the United States mammography screening guidelines on patient survival in a predominantly African-American population treated in a public hospital during 2008.
      ). The current analyses and conclusions are significantly different and support recommendations outside the scope of our previous work. The initial manuscript modeled data for 84 patients; the endpoint was stage at diagnosis and a significant advantage was found for ACS guidelines. We published this fragment both to validate the model and to provide the backing of the peer review process to the recommendations included in the text, which we wished to apply in our institution. Conclusions emphasized the importance of screening discussions with a primary care provider for lower socioeconomic status, African-American women. However, this manuscript builds on the previous in both power and scope, increasing to 274 patients and applying the survival and cost analyses. Not only do these data add insight from a national payer perspective, but allow us to give specific recommendations for populations similar to ours (rather than patient–provider discussion alone).
      Substantial evidence supports wide ethnic and socioeconomic variations in breast cancer patterns. Despite lower incidence, African-American women are more likely than Caucasian women to be diagnosed at younger ages (

      Howlader N, Noone AM, Krapcho M, Neyman N, Aminou R, Waldron W, …, Edwards BK (eds.). SEER Cancer Statistics Review, 1975-2008, National Cancer Institute. Bethesda, MD, Available: http://seer.cancer.gov/csr/1975_2008/, based on November 2010 SEER data submission, posted to the SEER web site, 2011.

      ,
      • Yang R.
      • Cheung M.C.
      • Franceschi D.
      • Hurley J.
      • Huang Y.
      • Livingstone A.S.
      • Koniaris L.G.
      African-American and low-socioeconomic status patients have a worse prognosis for invasive ductal and lobular breast carcinoma: Do screening criteria need to change?.
      ). They present with more advanced-stage disease, and are more likely to die from breast cancer when matched for age and stage (

      Howlader N, Noone AM, Krapcho M, Neyman N, Aminou R, Waldron W, …, Edwards BK (eds.). SEER Cancer Statistics Review, 1975-2008, National Cancer Institute. Bethesda, MD, Available: http://seer.cancer.gov/csr/1975_2008/, based on November 2010 SEER data submission, posted to the SEER web site, 2011.

      ,
      • Komenaka I.K.
      • Martinez M.E.
      • Pennington Jr., R.E.
      • Hsu C.H.
      • Clare S.E.
      • Thompson P.A.
      • Goulet Jr., R.J.
      • et al.
      Race and ethnicity and breast cancer outcomes in an underinsured population.
      ,
      • Yang R.
      • Cheung M.C.
      • Franceschi D.
      • Hurley J.
      • Huang Y.
      • Livingstone A.S.
      • Koniaris L.G.
      African-American and low-socioeconomic status patients have a worse prognosis for invasive ductal and lobular breast carcinoma: Do screening criteria need to change?.
      ). Part of the disadvantage may be owing to increased hormone receptor-negative and triple-negative tumors (
      • Cunningham J.E.
      • Montero A.J.
      • Garrett-Mayer E.
      • Berkel H.J.
      • Ely B.
      Racial differences in the incidence of breast cancer subtypes defined by combined histologic grade and hormone receptor status.
      ,
      • Komenaka I.K.
      • Martinez M.E.
      • Pennington Jr., R.E.
      • Hsu C.H.
      • Clare S.E.
      • Thompson P.A.
      • Goulet Jr., R.J.
      • et al.
      Race and ethnicity and breast cancer outcomes in an underinsured population.
      ,
      • O'Brien K.M.
      • Cole S.R.
      • Tse C.K.
      • Perou C.M.
      • Carey L.A.
      • Foulkes W.D.
      • Millikan R.C.
      • et al.
      Intrinsic breast tumor subtypes, race, and long-term survival in the Carolina Breast Cancer Study.
      ) or larger, more lymph node–positive tumors (
      • McBride R.
      • Hershman D.
      • Tsai W.Y.
      • Jacobson J.S.
      • Grann V.
      • Neugut A.I.
      Within-stage racial differences in tumor size and number of positive lymph nodes in women with breast cancer.
      ). Some effects have been reproduced in low-income populations demonstrating higher breast cancer risk in the 4th decade of life (
      • Lobb R.
      • Ayanian J.Z.
      • Allen J.D.
      • Emmons K.M.
      Stage of breast cancer at diagnosis among low-income women with access to mammography.
      ), cancers of worse prognosis (
      • Klassen A.C.
      • Smith K.C.
      The enduring and evolving relationship between social class and breast cancer burden: A review of the literature.
      ), and more advanced disease at diagnosis (
      • Ward E.M.
      • Fedewa S.A.
      • Cokkinides V.
      • Virgo K.
      The association of insurance and stage at diagnosis among patients aged 55 to 74 years in the national cancer database.
      ). Finally, differences in mammography utilization may affect profoundly the calculation of numbers invited to screening versus lives saved in women of minority ethnic groups or lower socioeconomic strata (
      • Aragon R.
      • Morgan J.
      • Wong J.H.
      • Lum S.
      Potential impact of USPSTF recommendations on early diagnosis of breast cancer.
      ,
      • DuBard C.A.
      • Schmid D.
      • Yow A.
      • Rogers A.B.
      • Lawrence W.W.
      Recommendation for and receipt of cancer screenings among Medicaid recipients 50 years and older.
      ,
      • Gabram S.G.
      • Lund M.J.
      • Gardner J.
      • Hatchett N.
      • Bumpers H.L.
      • Okoli J.
      • Brawley O.W.
      • et al.
      Effects of an outreach and internal navigation program on breast cancer diagnosis in an urban cancer center with a large African-American population.
      ,
      • Smith-Bindman R.
      • Miglioretti D.L.
      • Lurie N.
      • Abraham L.
      • Barbash R.B.
      • Strzelczyk J.
      • Kerlikowske K.
      • et al.
      Does utilization of screening mammography explain racial and ethnic differences in breast cancer?.
      ).
      Although breast cancer screening may be moving toward a more personalized model based on individual risk (
      • Schousboe J.T.
      • Kerlikowske K.
      • Loh A.
      • Cummings S.R.
      Personalizing mammography by breast density and other risk factors for breast cancer: Analysis of health benefits and cost-effectiveness.
      ), our findings reiterate the importance of integrating existing knowledge of ethnic and socioeconomic disease patterns into any national screening recommendations. They also support the continued use of ACS screening schedules in low-income African-American women treated in urban public hospitals, and in that regard, may inform patient navigation programs for similar populations.
      Our study cohort has a much greater proportion of African-American women than the major randomized, controlled trials and national statistical databases. The ages at diagnosis fall between two studies examining ethnic breakdown in large cancer registries. One Massachusetts group (
      • McBride R.
      • Hershman D.
      • Tsai W.Y.
      • Jacobson J.S.
      • Grann V.
      • Neugut A.I.
      Within-stage racial differences in tumor size and number of positive lymph nodes in women with breast cancer.
      ) reported 45% of patients diagnosed before age 50; however, this population had a larger percentage of Hispanic/Latina women (45%), who seemed to be diagnosed earlier in our population as well and, if better represented, may have increased under-50 diagnosis. In another study from the Florida Cancer Registry (
      • Yang R.
      • Cheung M.C.
      • Franceschi D.
      • Hurley J.
      • Huang Y.
      • Livingstone A.S.
      • Koniaris L.G.
      African-American and low-socioeconomic status patients have a worse prognosis for invasive ductal and lobular breast carcinoma: Do screening criteria need to change?.
      ), subset analysis showed that 22.4% of African-American patients were diagnosed before age 50, which is similar to the 28.8% of our patients. Furthermore, that study found a younger average age at diagnosis, 56.6 years, for African-American women compared with Caucasian women. This closely approximates our average age of diagnosis, although we found no Black–White disparity (57 years in African-Americans vs. 57.6 years in Caucasians). However, our Caucasian subset is too small (n = 11) to truly analyze. The similarity in presentation of our patients and the Florida registry patients may be owing to their roughly equivalent socioeconomic status and access to care, a fact that has eliminated outcomes disparities in other studies (
      • Chu Q.D.
      • Smith M.H.
      • Williams M.
      • Panu L.
      • Johnson L.W.
      • Shi R.
      • Glass J.
      • et al.
      Race/Ethnicity has no effect on outcome for breast cancer patients treated at an academic center with a public hospital.
      ). Patients in our study were, however, much more likely to be uninsured (vs. receiving Medicaid or Medicare) than the Florida cohort (only 8.8% of African-American women).
      Tumor characteristics in our population were worse compared with the Florida study overall (
      • Yang R.
      • Cheung M.C.
      • Franceschi D.
      • Hurley J.
      • Huang Y.
      • Livingstone A.S.
      • Koniaris L.G.
      African-American and low-socioeconomic status patients have a worse prognosis for invasive ductal and lobular breast carcinoma: Do screening criteria need to change?.
      ). We found fewer tumors less than 2 cm (39.4% in our population vs. 58%) and a lower proportion of local disease (Table 2). However, compared with Florida's African-American patients alone, the relationships among tumor size categories are similar (42.9% less than 2 cm/38.9% 2–5 cm/14.2% greater than 5 cm our population vs. 47.3%/38.8%/13.9% Florida), whereas those among local, regional, and distant disease are worse (50.5% local/33.8% regional/15.8% distant vs. 52.4%/41.7%/5.9%). Our population also has a greater shift toward distant disease compared with national averages for African-American women in SEER (51.9%/37.0%/8.1%;

      Howlader N, Noone AM, Krapcho M, Neyman N, Aminou R, Waldron W, …, Edwards BK (eds.). SEER Cancer Statistics Review, 1975-2008, National Cancer Institute. Bethesda, MD, Available: http://seer.cancer.gov/csr/1975_2008/, based on November 2010 SEER data submission, posted to the SEER web site, 2011.

      ). Our stages are very similar to those presented by Komenaka et al., who examined stage at diagnosis among underinsured African-American women. Our findings for the proportion of triple-negative tumors also match Komenaka's (31.9% our population vs. 32%), as do percentages of estrogen receptor–negative tumors (38.9% our population vs. 41%). Triple-negative disease comparisons between ethnicities match previously published studies in which African-American women face the highest burden, Hispanic/Latina women the next highest, and Caucasian women the lowest (
      • Ray M.
      • Polite B.N.
      Triple-negative breast cancers: A view from 10,000 feet.
      ). They are also consistent with previously published data by these authors on the high prevalence of triple-negative tumors in the Grady Memorial Hospital population (
      • Lund M.J.
      • Butler E.N.
      • Bumpers H.L.
      • Okoli J.
      • Rizzo M.
      • Hatchett N.
      • Gabram S.G.
      • et al.
      High prevalence of triple-negative tumors in an urban cancer center.
      ).
      Compared with the Komenaka et al. study, our population was less likely to have undergone screening mammography in the 2 years before diagnosis (30.6% our population vs. 51%). Our percentage undergoing mammography in the last year is also substantially less than the 60.6% of African-American women reported in 2007 from the U.S. Medical Expenditure Panel Survey (
      • Miranda P.Y.
      • Tarraf W.
      • González H.M.
      Breast cancer screening and ethnicity in the United States: Implications for health disparities research.
      ), but those data were self-reported and thus subject to recall and social acceptability biases. Only 32.1% of lesions in our population were found on routine screening, in contrast with 59% nationally (
      • Breen N.
      • Yabroff K.R.
      • Meissner H.I.
      What proportion of breast cancers are detected by mammography in the United States?.
      ). That so few tumors were detected on mammography may bear strongly on our outcomes given differences in behavior between mammography-detected tumors and tumors detected by symptoms or self-examination (
      • Shen N.
      • Hammonds L.S.
      • Madsen D.
      • Dale P.
      Mammography in 40-year-old women: What difference does it make? The potential impact of the U.S. Preventative Services Task Force (USPSTF) mammography guidelines.
      ). We are not able to ascertain, however, whether these are missed or true interval tumors (
      • Kirsh V.A.
      • Chiarelli A.M.
      • Edwards S.A.
      • O'Malley F.P.
      • Shumak R.S.
      • Yaffe M.J.
      • Boyd N.F.
      Tumor characteristics associated with mammographic detection of breast cancer in the Ontario breast screening program.
      ).
      Our cost findings are similar to those reported by
      • Barron J.J.
      • Quimbo R.
      • Nikam P.T.
      • Amonkar M.M.
      Assessing the economic burden of breast cancer in a US managed care population.
      and
      • Janjan N.
      • Nattinger A.
      • Young M.
      First-year treatment charges for early-stage vs locally advanced breast cancer.
      and feature increments between stages similar to previous reports in comparable populations (
      • Subramanian S.
      • Trogdon J.
      • Ekwueme D.U.
      • Gardner J.G.
      • Whitmire J.T.
      • Rao C.
      Cost of breast cancer treatment in Medicaid: Implications for state programs providing coverage for low-income women.
      ). Overall, the data suggest that although cost effectiveness is an important national goal, using variables specific to a low-income minority population may yield modified guidelines on which to base ‘best’ screening practices. There have been numerous proponents to reevaluate the interpretation of the data behind the 2009 USPSTF recommendations (
      • Helvie M.A.
      • Petitti D.B.
      • Calonge N.
      • Melnyk B.M.
      • Wilt T.J.
      USPSTF erroneously understated life-years-gained benefit of mammographic screening of women in their 40s.
      ,
      • Hendrick R.E.
      • Helvie M.A.
      United States Preventive Services Task Force screening mammography recommendations: Science ignored.
      ). One study tested the idea of incorporating disparity-specific variables into outcome models for cervical cancer. The authors found that different scenarios provided reductions in disparities among ethnic groups at increased or equal cost effectiveness (
      • Goldie S.J.
      • Daniels N.
      Model-based analyses to compare health and economic outcomes of cancer control: Inclusion of disparities.
      ).
      The study is limited by restrictions to the tumor model. Because we cannot predict local invasion or metastasis, stages 0 and 4 were excluded. Similarly, we were unable to assign an “N” status to the predicted TNM staging, which may have underestimated simulated stage 3 and 4 disease and led to a bias toward screening. In contrast, our threshold mammographic detection was purposefully conservative at 10 mm; any bias caused by this decision would point away from the benefits of screening, as would the assumption that all tumors become palpable at 20 mm. Additionally, there is a potential for significant variation in tumor biology among the Peer et al. population of White, non-Hispanic, European women and our studied population of African-American women. However, the doubling time values published in the Peer et al. paper are roughly in accordance with the estimates of breast cancer growth found in a number of other published studies relying on serial mammographic measurements (
      • Kuroishi T.
      • Tominaga S.
      • Morimoto T.
      • Tashiro H.
      • Itoh S.
      • Watanabe H.
      • Ishida T.
      • et al.
      Tumor growth rate and prognosis of breast cancer mainly detected by mass screening.
      ,
      • Lundgren B.
      Observations on growth rate of breast carcinomas and its possible implications for lead time.
      ,
      • Michaelson J.
      • Satija S.
      • Moore R.
      • Weber G.
      • Halpern E.
      • Garland A.
      • Kopans D.B.
      Estimates of breast cancer growth rate and sojourn time from screening database information.
      ,
      • Peer P.
      • Dijck J.
      • Hendriks J.
      • Holland R.
      • Verbeek A.
      Age dependent growth rate of primary breast cancers.
      ,
      • Spratt J.A.
      • von Fournier D.
      • Spratt J.S.
      • Weber E.E.
      Mammographic assessment of human breast cancer growth and duration.
      ,
      • Spratt Jr., J.S.
      • Spratt T.L.
      Rates of Growth of Pulmonary Metastases and Host Survival.
      ,
      • von Fournier D.
      • Weber E.
      • Hoeffken W.
      • Bauer M.
      • Kubli F.
      • Barth V.
      Growth rate of 147 mammary carcinomas.
      ).
      The tumor model focuses on a small sample size of lower socioeconomic status, African-American women in an urban public hospital. Although the results are significant in that subset, they cannot be generalized to the overall United States population or to all African-American women across socioeconomic boundaries. We further focused on specific tumor characteristics within that population (age of patient and size and grade of tumor) as a reasonable but limited model of behavior. Additional information, such as family history or hormone receptor status, would admittedly influence tumor behavior as well, but falls outside the scope of this model.
      Finally, some evidence suggests that earlier screening results in lead time bias rather than actual reduction of advanced stage disease, and that other factors are responsible for reducing mortality after widespread screening (
      • Autier P.
      • Boniol M.
      • Middleton R.
      • Doré J.F.
      • Héry C.
      • Zheng T.
      • Gavin A.
      Advanced breast cancer incidence following population-based mammographic screening.
      ). Because we cannot measure these effects directly in our hypothetical cohort, we cannot conclusively rule out this possibility. However, other papers in similarly underserved populations (
      • Elting L.S.
      • Cooksley C.D.
      • Bekele B.N.
      • Giordano S.H.
      • Shih Y.C.
      • Lovell K.K.
      • Theriault R.
      • et al.
      Mammography capacity impact on screening rates and breast cancer stage at diagnosis.
      ,
      • Norman S.A.
      • Russell Localio A.
      • Weber A.L.
      • Coates R.J.
      • Zhou L.
      • Bernstein L.
      • Nadel M.R.
      • et al.
      Protection of mammography screening against death from breast cancer in women aged 40-64 years.
      ) and indirect observations in the Grady Memorial Hospital population (
      • Gabram S.G.
      • Lund M.J.
      • Gardner J.
      • Hatchett N.
      • Bumpers H.L.
      • Okoli J.
      • Brawley O.W.
      • et al.
      Effects of an outreach and internal navigation program on breast cancer diagnosis in an urban cancer center with a large African-American population.
      ) suggest that screening in such women may indeed result in a decreased incidence of advanced stage presentation as a mechanism for decreased mortality. Adjustment of the model for age and grade of tumor should further mitigate against lead time bias.
      Although USPSTF guidelines allow for high-risk women to be screened before age 50 and after age 75 based on discussions with their primary care provider, we did not attempt to account for these scenarios in the model. Data from the University of Pennsylvania in 2005 suggest that African-American women at risk of BRCA 1 or 2 mutations are almost five times less likely than Caucasian women to receive counseling about genetic testing, even though all patients had seen a primary care physician within the 3 years before the study period (
      • Armstrong K.
      • Micco E.
      • Carney A.
      • Stopfer J.
      • Putt M.
      Racial differences in the use of BRCA1/2 testing among women with a family history of breast or ovarian cancer.
      ). The researchers pointed out the concentration of African-American patients among relatively few primary care physicians; moreover, these providers are less likely to be board certified and report more difficulty in delivering high-quality care relative to primary care physicians seen by Caucasian patients (
      • Bach P.B.
      • Pham H.H.
      • Schrag D.
      • Tate R.C.
      • Hargraves J.L.
      Primary care physicians who treat blacks and whites.
      ). In addition, Medicaid recipients have been shown to have lower rates of cancer screening compared with the general population. In one study of those aged 50 and over eligible for cancer screening, only 60% of women were referred for mammography by their primary care provider (
      • Smith-Bindman R.
      • Miglioretti D.L.
      • Lurie N.
      • Abraham L.
      • Barbash R.B.
      • Strzelczyk J.
      • Kerlikowske K.
      • et al.
      Does utilization of screening mammography explain racial and ethnic differences in breast cancer?.
      ). Finally, the success at Grady Memorial Hospital of direct community outreach and patient navigation programs in increasing mammography utilization and schedule adherence (
      • Gabram S.G.
      • Lund M.J.
      • Harrison C.
      • Koenig E.
      • Bumpers H.L.
      • Okoli J.
      • Brawley O.W.
      • et al.
      Improving access to care in an AVON Comprehensive Breast Center (AVON-CBC) serving a predominantly urban African American (AA) patient population.
      ,
      • Gabram S.G.
      • Lund M.J.
      • Gardner J.
      • Hatchett N.
      • Bumpers H.L.
      • Okoli J.
      • Brawley O.W.
      • et al.
      Effects of an outreach and internal navigation program on breast cancer diagnosis in an urban cancer center with a large African-American population.
      ) suggests that many of our patients may be guided more strongly by information received outside of the primary care context. These factors provide evidence that for underinsured African-American women, a provision leaving extended screening at the discretion of a patient–provider conversation may not be sufficient to ensure screening outside of the routine recommendation.
      Directions for future research include analysis of serial mammograms within our population to better understand tumor growth within this subset of women. Also, given the striking pattern of screening behavior and age at diagnosis observed among Hispanic/Latina women, directed inquiry in a larger sample size might elucidate whether these women are truly at such a diagnostic disadvantage.

      Conclusion

      Our data support the substantial body of evidence that African-American and lower socioeconomic status women present with markedly different patterns of screening behavior and breast cancer characteristics than those represented by population-wide averages. Awareness of such issues is even more critical in underserved populations, where patient navigator programs provide a large degree of influence in helping women to access and utilize screening and diagnostic services, and when changes to national guidelines may impact the services covered by public insurance.

      Implications for Practice and/or Policy

      Although further research is needed to validate our findings and determine whether our data can be generalized on a more national and/or global level, we believe that low-income African-American women and any navigation programs that serve them, should be encouraged to continue to follow the ACS recommendations for annual mammography beginning at 40 years and continuing as long as life expectancy and functional status allow. Not only could considering alternative screening strategies provide cost-effective means of reducing late-stage disease burden, but reinvestigating the microsimulation models specific to lower income and ethnic minority women may provide valuable insight into reducing the cancer disparities that present from diagnosis to death in this population.

      Acknowledgments

      Imnett Habtes, MD, initiated this study in 2010; she contributed to study design and collected data for 2008. Tumor Registry and National Cancer Database statistics were provided by Sherita Hearn. Tumor size projections, costs, and statistical analysis of scenario comparisons were provided by Cheryl Raskind-Hood, MPH, and Kathleen Adams, PhD. Sheryl Gabram-Mendola, MD, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
      The authors are grateful to and acknowledge support of this work from the Avon Foundation for Women whose mission is to promote charitable, scientific, educational and humanitarian activities, with a special emphasis on those activities that improve the lives of women and their families globally.

      References

      1. American College of Surgeons. National Cancer Database survival report 2003. Online via Datalinks Portal for Commission on Cancer-accredited cancer programs. Available: https://datalinks.facs.org. Accessed by Sherita Hearn ( [email protected] ), tumor registry coordinator. Accessed May 11, 2011.

        • Aragon R.
        • Morgan J.
        • Wong J.H.
        • Lum S.
        Potential impact of USPSTF recommendations on early diagnosis of breast cancer.
        Annals of Surgical Oncology. 2011; 18: 3137-3142
        • Armstrong K.
        • Micco E.
        • Carney A.
        • Stopfer J.
        • Putt M.
        Racial differences in the use of BRCA1/2 testing among women with a family history of breast or ovarian cancer.
        JAMA. 2005; 293: 1729-1736
        • Autier P.
        • Boniol M.
        • Middleton R.
        • Doré J.F.
        • Héry C.
        • Zheng T.
        • Gavin A.
        Advanced breast cancer incidence following population-based mammographic screening.
        Annals of Oncology. 2011; 22 (Review): 1726-1735
        • Bach P.B.
        • Pham H.H.
        • Schrag D.
        • Tate R.C.
        • Hargraves J.L.
        Primary care physicians who treat blacks and whites.
        New England Journal of Medicine. 2004; 351: 575-584
        • Barron J.J.
        • Quimbo R.
        • Nikam P.T.
        • Amonkar M.M.
        Assessing the economic burden of breast cancer in a US managed care population.
        Breast Cancer Research and Treatment. 2008; 109: 367-377
        • Bjurstam N.
        • Björneld L.
        • Warwick J.
        • Sala E.
        • Duffy S.W.
        • Nyström L.
        • Wahlin T.
        • et al.
        The Gothenburg Breast Screening Trial.
        Cancer. 2003; 97: 2387-2396
        • Breen N.
        • Yabroff K.R.
        • Meissner H.I.
        What proportion of breast cancers are detected by mammography in the United States?.
        Cancer Detection and Prevention. 2007; 31: 220-224
        • Chu Q.D.
        • Smith M.H.
        • Williams M.
        • Panu L.
        • Johnson L.W.
        • Shi R.
        • Glass J.
        • et al.
        Race/Ethnicity has no effect on outcome for breast cancer patients treated at an academic center with a public hospital.
        Cancer Epidemiology, Biomarkers & Prevention. 2009; 18: 2157-2161
        • Cunningham J.E.
        • Montero A.J.
        • Garrett-Mayer E.
        • Berkel H.J.
        • Ely B.
        Racial differences in the incidence of breast cancer subtypes defined by combined histologic grade and hormone receptor status.
        Cancer Causes & Control. 2010; 21: 399-409
        • DuBard C.A.
        • Schmid D.
        • Yow A.
        • Rogers A.B.
        • Lawrence W.W.
        Recommendation for and receipt of cancer screenings among Medicaid recipients 50 years and older.
        Archives of Internal Medicine. 2008; 168: 2014-2021
        • Elting L.S.
        • Cooksley C.D.
        • Bekele B.N.
        • Giordano S.H.
        • Shih Y.C.
        • Lovell K.K.
        • Theriault R.
        • et al.
        Mammography capacity impact on screening rates and breast cancer stage at diagnosis.
        American Journal of Preventive Medicine. 2009; 37: 102-108
        • Gabram S.G.
        • Lund M.J.
        • Gardner J.
        • Hatchett N.
        • Bumpers H.L.
        • Okoli J.
        • Brawley O.W.
        • et al.
        Effects of an outreach and internal navigation program on breast cancer diagnosis in an urban cancer center with a large African-American population.
        Cancer. 2008; 113: 602-607
        • Gabram S.G.
        • Lund M.J.
        • Harrison C.
        • Koenig E.
        • Bumpers H.L.
        • Okoli J.
        • Brawley O.W.
        • et al.
        Improving access to care in an AVON Comprehensive Breast Center (AVON-CBC) serving a predominantly urban African American (AA) patient population.
        Breast Cancer Research and Treatment. 2007; 106: S6041
        • Goldie S.J.
        • Daniels N.
        Model-based analyses to compare health and economic outcomes of cancer control: Inclusion of disparities.
        Journal of the National Cancer Institute. 2011; 103: 1373-1386
        • Green B.B.
        • Taplin S.H.
        Breast cancer screening controversies.
        Journal of the American Board of Family Practice. 2003; 16 (Review): 233-241
        • Güth U.
        • Huang D.J.
        • Huber M.
        • Schötzau A.
        • Wruk D.
        • Holzgreve W.
        • Zanetti-Dällenbach R.
        • et al.
        Tumor size and detection in breast cancer: Self-examination and clinical breast examination are at their limit.
        Cancer Detection and Prevention. 2008; 32: 224-228
        • Habtes I.
        • Friedman D.
        • Raskind-Hood C.
        • Adams K.
        • Becker E.
        • D'Orsi C.
        • Gabram S.
        • et al.
        Determining the impact of the United States mammography screening guidelines on patient survival in a predominantly African-American population treated in a public hospital during 2008.
        Cancer. 2013; 119: 481-487
        • Hellquist B.N.
        • Duffy S.W.
        • Abdsaleh S.
        • Björneld L.
        • Bordás P.
        • Tabár L.
        • Jonsson H.
        • et al.
        Effectiveness of population-based service screening with mammography for women ages 40 to 49 years: Evaluation of the Swedish Mammography Screening in Young Women (SCRY) cohort.
        Cancer. 2011; 117: 714-722
        • Helvie M.A.
        • Petitti D.B.
        • Calonge N.
        • Melnyk B.M.
        • Wilt T.J.
        USPSTF erroneously understated life-years-gained benefit of mammographic screening of women in their 40s.
        Radiology. 2011; 258: 958-959
        • Hendrick R.E.
        • Helvie M.A.
        United States Preventive Services Task Force screening mammography recommendations: Science ignored.
        AJR American Journal of Roentgenology. 2011; 196: W112-W116
      2. Howlader N, Noone AM, Krapcho M, Neyman N, Aminou R, Waldron W, …, Edwards BK (eds.). SEER Cancer Statistics Review, 1975-2008, National Cancer Institute. Bethesda, MD, Available: http://seer.cancer.gov/csr/1975_2008/, based on November 2010 SEER data submission, posted to the SEER web site, 2011.

        • Janjan N.
        • Nattinger A.
        • Young M.
        First-year treatment charges for early-stage vs locally advanced breast cancer.
        Breast Disease. 1993; 6: 23-27
        • Kirsh V.A.
        • Chiarelli A.M.
        • Edwards S.A.
        • O'Malley F.P.
        • Shumak R.S.
        • Yaffe M.J.
        • Boyd N.F.
        Tumor characteristics associated with mammographic detection of breast cancer in the Ontario breast screening program.
        Journal of the National Cancer Institute. 2011; 103: 942-950
        • Klassen A.C.
        • Smith K.C.
        The enduring and evolving relationship between social class and breast cancer burden: A review of the literature.
        Cancer Epidemiology. 2011; 35 (Review): 217-234
        • Komenaka I.K.
        • Martinez M.E.
        • Pennington Jr., R.E.
        • Hsu C.H.
        • Clare S.E.
        • Thompson P.A.
        • Goulet Jr., R.J.
        • et al.
        Race and ethnicity and breast cancer outcomes in an underinsured population.
        Journal of the National Cancer Institute. 2010; 102: 1178-1187
        • Kuroishi T.
        • Tominaga S.
        • Morimoto T.
        • Tashiro H.
        • Itoh S.
        • Watanabe H.
        • Ishida T.
        • et al.
        Tumor growth rate and prognosis of breast cancer mainly detected by mass screening.
        Japanese Journal of Cancer Research. 1990; 81: 454-462
        • Lobb R.
        • Ayanian J.Z.
        • Allen J.D.
        • Emmons K.M.
        Stage of breast cancer at diagnosis among low-income women with access to mammography.
        Cancer. 2010; 116: 5487-5496
        • Lundgren B.
        Observations on growth rate of breast carcinomas and its possible implications for lead time.
        Cancer. 1977; 40: 1722-1725
        • Lund M.J.
        • Butler E.N.
        • Bumpers H.L.
        • Okoli J.
        • Rizzo M.
        • Hatchett N.
        • Gabram S.G.
        • et al.
        High prevalence of triple-negative tumors in an urban cancer center.
        Cancer. 2008; 113: 608-615
        • Mandelblatt J.S.
        • Cronin K.A.
        • Bailey S.
        • Berry D.A.
        • de Koning H.J.
        • Draisma G.
        • Feuer E.J.
        • et al.
        Breast Cancer Working Group of the Cancer Intervention and Surveillance Modeling Network. Effects of mammography screening under different screening schedules: Model estimates of potential benefits and harms.
        Annals of Internal Medicine. 2009; 151 (Erratum in: Ann Intern Med. 2010 Jan 19; 152(2):136): 738-747
        • Mariotto A.B.
        • Yabroff K.R.
        • Shao Y.
        • Feuer E.J.
        • Brown M.L.
        Projections of the cost of cancer care in the United States: 2010-2020.
        Journal of the National Cancer Institute. 2011; 103 (Erratum in: J Natl Cancer Inst. 2011 Apr 20; 103(8):699): 117-128
        • McBride R.
        • Hershman D.
        • Tsai W.Y.
        • Jacobson J.S.
        • Grann V.
        • Neugut A.I.
        Within-stage racial differences in tumor size and number of positive lymph nodes in women with breast cancer.
        Cancer. 2007; 110: 1201-1208
        • Michaelson J.
        • Satija S.
        • Moore R.
        • Weber G.
        • Halpern E.
        • Garland A.
        • Kopans D.B.
        Estimates of breast cancer growth rate and sojourn time from screening database information.
        Journal of Women's Imaging. 2002; 5: 11-19
        • Miller A.B.
        • To T.
        • Baines C.J.
        • Wall C.
        The Canadian National Breast Screening Study-1: Breast cancer mortality after 11 to 16 years of follow-up. A randomized screening trial of mammography in women age 40 to 49 years.
        Annals of Internal Medicine. 2002; 137: 305-312
        • Miranda P.Y.
        • Tarraf W.
        • González H.M.
        Breast cancer screening and ethnicity in the United States: Implications for health disparities research.
        Breast Cancer Research and Treatment. 2011; 128: 535-542
        • Moss S.M.
        • Cuckle H.
        • Evans A.
        • Johns L.
        • Waller M.
        • Bobrow L.
        Trial Management Group. Effect of mammographic screening from age 40 years on breast cancer mortality at 10 years' follow-up: A randomised controlled trial.
        Lancet. 2006; 368: 2053-2060
        • Nelson H.D.
        • Tyne K.
        • Naik A.
        • Bougatsos C.
        • Chan B.K.
        • Humphrey L.
        U.S. Preventive Services Task Force. Screening for breast cancer: An update for the U.S. Preventive Services Task Force.
        Annals of Internal Medicine. 2009; 151 (W237-42. Review): 727-737
        • Norman S.A.
        • Russell Localio A.
        • Weber A.L.
        • Coates R.J.
        • Zhou L.
        • Bernstein L.
        • Nadel M.R.
        • et al.
        Protection of mammography screening against death from breast cancer in women aged 40-64 years.
        Cancer Causes & Control. 2007; 18: 909-918
        • O'Brien K.M.
        • Cole S.R.
        • Tse C.K.
        • Perou C.M.
        • Carey L.A.
        • Foulkes W.D.
        • Millikan R.C.
        • et al.
        Intrinsic breast tumor subtypes, race, and long-term survival in the Carolina Breast Cancer Study.
        Clinical Cancer Research. 2010; 16: 6100-6110
        • Peer P.
        • Dijck J.
        • Hendriks J.
        • Holland R.
        • Verbeek A.
        Age dependent growth rate of primary breast cancers.
        Cancer. 1993; 71: 3547-3551
        • Pisano E.D.
        • Gatsonis C.
        • Hendrick E.
        • Yaffe M.
        • Baum J.K.
        • Acharyya S.
        • Rebner M.
        • Digital Mammographic Imaging Screening Trial (DMIST) Investigators Group
        • et al.
        Diagnostic performance of digital versus film mammography for breast-cancer screening.
        New England Journal of Medicine. 2005; 353 (Erratum in: N Engl J Med. 2006 Oct 26; 355(17):1840): 1773-1783
        • Ray M.
        • Polite B.N.
        Triple-negative breast cancers: A view from 10,000 feet.
        Cancer Journal. 2010; 16 (Review): 17-22
        • Salzmann P.
        • Kerlikowske K.
        • Phillips K.
        Cost-effectiveness of extending screening mammography guidelines to include women 40 to 49 years of age.
        Annals of Internal Medicine. 1997; 127 (Erratum in: Ann Intern Med 1998 May 15;128(10):878): 955-965
        • Schousboe J.T.
        • Kerlikowske K.
        • Loh A.
        • Cummings S.R.
        Personalizing mammography by breast density and other risk factors for breast cancer: Analysis of health benefits and cost-effectiveness.
        Annals of Internal Medicine. 2011; 155: 10-20
        • Shapiro S.
        • Strax P.
        • Venet L.
        Periodic breast cancer screening in reducing mortality from breast cancer.
        JAMA. 1971; 215: 1777-1785
        • Shen N.
        • Hammonds L.S.
        • Madsen D.
        • Dale P.
        Mammography in 40-year-old women: What difference does it make? The potential impact of the U.S. Preventative Services Task Force (USPSTF) mammography guidelines.
        Annals of Surgical Oncology. 2011; 18: 3066-3071
        • Smith-Bindman R.
        • Miglioretti D.L.
        • Lurie N.
        • Abraham L.
        • Barbash R.B.
        • Strzelczyk J.
        • Kerlikowske K.
        • et al.
        Does utilization of screening mammography explain racial and ethnic differences in breast cancer?.
        Annals of Internal Medicine. 2006; 144: 541-553
        • Smith R.A.
        • Cokkinides V.
        • Brooks D.
        • Saslow D.
        • Shah M.
        • Brawley O.W.
        Cancer screening in the United States, 2011: A review of current American Cancer Society guidelines and issues in cancer screening.
        CA: A Cancer Journal for Clinicians. 2011; 61: 8-30
        • Spratt J.A.
        • von Fournier D.
        • Spratt J.S.
        • Weber E.E.
        Mammographic assessment of human breast cancer growth and duration.
        Cancer. 1993; 71: 2020-2026
        • Spratt Jr., J.S.
        • Spratt T.L.
        Rates of Growth of Pulmonary Metastases and Host Survival.
        Annals of Surgery. 1964; 159: 161-171
        • Subramanian S.
        • Trogdon J.
        • Ekwueme D.U.
        • Gardner J.G.
        • Whitmire J.T.
        • Rao C.
        Cost of breast cancer treatment in Medicaid: Implications for state programs providing coverage for low-income women.
        Medical Care. 2011; 49: 89-95
        • Tabár L.
        • Vitak B.
        • Chen T.H.
        • Yen A.M.
        • Cohen A.
        • Tot T.
        • Duffy S.W.
        • et al.
        Swedish two-county trial: Impact of mammographic screening on breast cancer mortality during 3 decades.
        Radiology. 2011; 260: 658-663
        • Taplin S.H.
        • Barlow W.
        • Urban N.
        • Mandelson M.T.
        • Timlin D.J.
        • Ichikawa L.
        • Nefcy P.
        Stage, age, comorbidity, and direct costs of colon, prostate, and breast cancer care.
        Journal of the National Cancer Institute. 1995; 87: 417-426
        • U.S. Preventive Services Task Force (USPSTF)
        Screening for breast cancer: U.S. Preventive Services Task Force recommendation statement.
        Annals of Internal Medicine. 2009; 151 (W-236. Erratum in: Ann Intern Med. 2010 May 18; 152(10):688. Ann Intern Med. 2010 Feb 2; 152(3):199-200): 716-726
        • von Fournier D.
        • Weber E.
        • Hoeffken W.
        • Bauer M.
        • Kubli F.
        • Barth V.
        Growth rate of 147 mammary carcinomas.
        Cancer. 1980; 45: 2198-2207
        • Ward E.M.
        • Fedewa S.A.
        • Cokkinides V.
        • Virgo K.
        The association of insurance and stage at diagnosis among patients aged 55 to 74 years in the national cancer database.
        Cancer Journal. 2010; 16: 614-621
        • Yang R.
        • Cheung M.C.
        • Franceschi D.
        • Hurley J.
        • Huang Y.
        • Livingstone A.S.
        • Koniaris L.G.
        African-American and low-socioeconomic status patients have a worse prognosis for invasive ductal and lobular breast carcinoma: Do screening criteria need to change?.
        J Am Coll Surg. 2009; 208 (discussion 869–70): 853-868

      Biography

      Clara Farley, MD, is a first-year surgical resident, Emory University School of Medicine. She is a recent graduate of Emory Medical School with an interest in Breast Cancer and improving access to care across socioeconomic levels.
      Danielle Friedman, MD, is a second-year surgical resident at Albert Einstein College of Medicine.
      Imnett Habtes, MD, is an Internist in Mineola, New York, with an interest in pulmonology.
      Cheryl Raskind-Hood, MPH, is a Senior Research Associate at Emory University Rollins School of Public Health, Epidemiology. She has over 20 years of experience as a Biostatistician and Epidemiologist. Interests include maternal/child and women's health, surveillance, statistical methods; data linkage.
      E. Kathleen Adams, PhD, is Professor of Health Policy and Management, Rollins School of Public Health at Emory University. Her research focuses on health care and policy issues related to low-income, vulnerable populations with an emphasis on the Medicaid insured.
      Edmund R. Becker, PhD, is Professor of Health Policy and Management, Rollins School of Public Health at Emory University. His research focuses on health care provider performance and issues related to patient-centered care.
      Carl D'Orsi, MD, FACR, is Professor of Radiology and Oncology/Hematology, Emory University School of Medicine. He is Director Emeritus of Breast Imaging and Director of Breast Imaging Research at Winship Cancer Institute of Emory University.
      Kathleen Gundry, MD, is Assistant Professor of Radiology and Imaging Sciences, Emory University School of Medicine, Director of Breast Imaging at Grady Memorial Hospital. Her interests are in issues relating to the of care underserved populations and in effective patient care.
      George Birdsong, MD, is Professor of Pathology and Laboratory Medicine, Emory University School of Medicine. His interests include cytopathology, breast pathology, and clinical and pathology informatics with an emphasis on cancer reporting.
      Sheryl Gabram-Mendola, MD, MBA, FACS, is Professor of Surgery, Emory University School of Medicine and Principal Investigator Avon Foundation Grant. Her research focuses on decreasing disparities of breast cancer by providing high quality access to care in the Grady Health System.