JGIM
ORIGINAL RESEARCH
Breast Density Notification Legislation and Breast Cancer Stage
at Diagnosis: Early Evidence from the SEER Registry
Ilana Richman, MD1,2, Steven M. Asch, MD MPH1,3, Eran Bendavid, MD MS2,3,
Jay Bhattacharya, MD PhD2, and Douglas K. Owens, MD MS1,2
1
Center for Innovation to Implementation, Palo Alto VA Health Care System, Palo Alto, USA; 2Center for Primary Care and Outcomes Research/
Center for Health Policy, Stanford University School of Medicine, Stanford, CA, USA; 3Division of General Medical Disciplines, Stanford University
School of Medicine, Stanford, CA, USA.
BACKGROUND: Twenty-eight states have passed breast
density notification laws, which require physicians to inf o r m w om e n of a f i n d i n g o f d e n s e b r e a s t s o n
mammography.
OBJECTIVE: To evaluate changes in breast cancer stage
at diagnosis after enactment of breast density notification
legislation.
DESIGN: Using a difference-in-differences analysis, we
examined changes in stage at diagnosis among women
with breast cancer in Connecticut, the first state to enact
legislation, compared to changes among women in control
states. We used data from the Surveillance, Epidemiology,
and End Results Program (SEER) registry, 2005–2013.
PARTICIPANTS: Women ages 40–74 with breast cancer.
INTERVENTION: Breast density notification legislation,
enacted in Connecticut in October of 2009.
MAIN MEASURE: Breast cancer stage at diagnosis.
KEY RESULTS: Our study included 466,930 women,
25,592 of whom lived in Connecticut. Legislation was
associated with a 1.38-percentage-point (95 % CI 0.12 to
2.63) increase in the proportion of women in Connecticut
versus control states who had localized invasive cancer at
the time of diagnosis, and a 1.12-percentage-point (95 %
CI −2.21 to −0.08) decline in the proportion of women with
ductal carcinoma in situ at diagnosis. Breast density notification legislation was not associated with a change in
the proportion of women in Connecticut versus control
states with regional-stage (−0.09 percentage points, 95 %
CI −1.01 to 1.02) or metastatic disease (−0.24, 95 % CI
−0.75 to 0.28). County-level analyses and analyses limited to women younger than 50 found no statistically significant associations.
LIMITATIONS: Single intervention state, limited followup, potential confounding from unobserved trends.
CONCLUSIONS: Breast density notification legislation in
Connecticut was associated with a small increase in the
proportion of women diagnosed with localized invasive
breast cancer in individual-level but not county-level
analyses. Whether this finding reflects potentially beneficial early detection or potentially harmful overdiagnosis is
Electronic supplementary material The online version of this article
(doi:10.1007/s11606-016-3904-y) contains supplementary material,
which is available to authorized users.
Received May 9, 2016
Revised September 23, 2016
Accepted October 12, 2016
Published online November 14, 2016
not known. Legislation was not associated with changes
in regional or metastatic disease.
KEY WORDS: cancer screening; breast cancer; health policy.
J Gen Intern Med 32(6):603–9
DOI: 10.1007/s11606-016-3904-y
© Society of General Internal Medicine 2016
INTRODUCTION
Among women undergoing routine screening mammography,
about 43 % have radiographically dense breasts.1 Dense
breasts are considered normal, but the finding has two important implications. First, breast density may be an independent,
moderate risk factor for breast cancer.2 Second, the sensitivity
and specificity of mammography are reduced among women
with dense breasts, although newer digital techniques partially
mitigate this effect.3
Given the limitations of mammography, some have
advocated supplemental screening with magnetic resonance imaging (MRI) or ultrasound in women with dense
breasts. Indeed, each of these modalities can detect additional cases of breast cancer after a negative mammogram
in women with dense breasts. Ultrasound, for example,
detects about 5 cancers per 1000 women screened, while
MRI can detect between 3 and 33 cases of cancer.4 Supplemental screening most commonly detects localized invasive cancers rather than ductal carcinoma in situ (DCIS)
or late-stage disease.5
Supplemental screening tests, however, have poor positive
predictive value, prompting concerns that widespread use
among low-risk women could generate a considerable number
of false positives and subsequent invasive testing.5–9 Furthermore, there are no randomized trials or observational studies
evaluating the effect of adjunctive screening on morbidity or
mortality.4,10 The US Preventive Services Task Force 2016
guideline on screening for breast cancer determined that there
was insufficient evidence to recommend for or against supplemental screening.11
Despite uncertainty about the optimal approach to screening
women with dense breasts, 28 states have passed breast density notification laws. These laws vary by state, but generally
require that physicians disclose a finding of dense breasts to
603
604
Richman et al.: Breast Density Legislation and Breast Cancer Stage at Diagnosis
women undergoing mammography.12 A number of states also
require that physicians inform women with dense breasts that
they may benefit from adjunctive screening with ultrasound or
MRI, and some states (including Connecticut) require that
private insurers cover supplemental screening for women with
dense breasts. Although the number of states with breast
density notification legislation has grown steadily, no study
has examined the effect of this kind of legislation on health
outcomes for women.
Our goal was to evaluate the relationship between breast
density notification legislation and stage at diagnosis among
women with breast cancer. Breast density notification legislation may lead to an increased use of supplemental screening,
and supplemental screening, in turn, might lead to an increase in
early-stage breast cancer diagnosis. Thus, we evaluated changes
in cancer stage at the time of diagnosis among women living in
Connecticut, the first state subject to breast density notification,
compared to control populations.
As a secondary goal, we used a county-level analysis to
examine changes in stage-specific incidence associated with
breast density notification legislation. If legislation enabled
early detection of cancers that would have eventually progressed to late-stage cancer, legislation should be associated
with increased incidence of early-stage cancer and decreased
incidence of late-stage cancer. In contrast, if legislation facilitated detection of DCIS or early invasive cancers that did not
progress and become harmful (i.e., overdiagnosis), legislation
would be associated with increased DCIS or early-stage invasive cancer without a concomitant drop in late-stage cancer.
METHODS
Study Design
We used two complementary approaches to examine the association between breast density notification legislation and
changes in breast cancer stage at diagnosis. In our primary
approach, we evaluated the proportion of women with DCIS,
localized, regional, or metastatic disease at the time of diagnosis. We used a difference-in-differences design to compare
changes in the proportion of women diagnosed at each stage in
Connecticut, the first state subject to breast density notification
legislation, to changes among women in states without such
legislation.
In a second, complementary difference-in-differences analysis, we evaluated changes in stage-specific incidence in
counties in Connecticut and control counties. The goal of this
analysis was to describe changes in incidence associated with
breast density notification legislation and to evaluate for evidence of overdiagnosis associated with this policy. Overdiagnosis refers to detecting a cancer, usually by screening, that
would never have been clinically apparent. In the case of
breast cancer, screening contributes to overdiagnosis of both
in situ and early-stage invasive cancers.13
Stage-specific incidence patterns over time can provide evidence of an effective or ineffective screening program. As noted,
JGIM
an effective screening program should result in an increase in the
incidence of early-stage disease as early-stage cancers are
detected, and a decline in late-stage disease as advanced disease
is averted by early diagnosis.14 In contrast, an increased incidence
of early-stage disease without a concomitant decline in late-stage
disease suggests that the screening program detects only clinically insignificant cancers and contributes to overdiagnosis.15 Using
this framework, we evaluated changes in stage-specific incidence
in counties in Connecticut and control counties.
Data Sources, Variables, and Participants. We analyzed
incident breast cancer cases from the Surveillance,
Epidemiology, and End Results Program (SEER) Registry.16
SEER is a population-based registry that captures cancer incidence and mortality in specific geographies. SEER covers 13 US
states (Alaska, California, Connecticut, Georgia, Hawaii, Kentucky, Iowa, Louisiana, Michigan, New Jersey, New Mexico,
Utah, Washington). Among these, two states, Connecticut and
California, passed breast density notification legislation during
the period for which data were available. Connecticut enacted its
legislation in October of 2009,17 and California in April of
2013.18 Because we had less than 1 year of follow-up data for
California, and because we would not expect to see a substantial
effect from legislation in this short time, we excluded data from
the post-legislation period in California in our main analyses. In a
sensitivity analysis, we included all California data.
In our main analyses, we included women residing in all
SEER areas who were between the ages of 40 and 74 and who
had a diagnosis of invasive breast cancer or DCIS. We excluded
women with a diagnosis of lobular carcinoma in situ (LCIS), as
LCIS is typically not detected during screening.19 We also excluded women who had unstaged cancer. We excluded data from
counties affected by Hurricane Katrina in 2005, which are not
considered part of the standard SEER data set. We used US
Census county populations to calculate incidence. The Area
Resource File for 2013–2014 and 2014–2015 provided countylevel covariates.20 We limited our analysis to the period from
2005 to 2013.
Our main outcome variable was stage at diagnosis. We used a
summary staging variable prepared by SEER which collapses
stage at diagnosis into four basic categories: in situ, localized,
regional, and distant/metastatic.21 Localized disease includes
cancer confined to the breast, and regional disease includes local
extension outside the confines of the breast and to regional lymph
nodes. In a sensitivity analysis, we used an alternate summary
staging variable which uses a more stringent definition of metastatic disease.
Statistical Analysis
For our individual-level analysis, we used a multinomial logistic regression model with stage at diagnosis as the outcome.
This model included an indicator variable for women residing
in Connecticut, an indicator for time period pre- and postlegislation (01/2005–09/2009, 10/2009–12/2013), and the
JGIM
605
Richman et al.: Breast Density Legislation and Breast Cancer Stage at Diagnosis
interaction between the two, which captures effects associated
with breast density notification legislation. We also adjusted
for race, ethnicity, age, and the post-legislation time period in
California. We expressed results as predicted probabilities. We
assessed for pre-legislation trends in our main analysis by
interacting time (year) with treatment group during the prelegislation period.
For our county-level analysis, we fit a series of mixedeffects negative binomial models of change in stage-specific
incidence over time. As in our individual-level models, we
included a binary term to represent counties in Connecticut, a
binary term for the pre/post-legislation time period, and the
interaction. Here, we defined the pre- and post-legislation
periods as 2005–2009 and 2010–2013, respectively. Models
included an exposure term to account for county population
(total population of women in the age range of interest in the
county) and county-specific random effects to account for
repeated measures on each county. We incorporated timevarying covariates derived from the Area Resource File including median household income, percentage of the female
population that is white, percentage uninsured, percentage of
women aged 40–64, and physicians per capita.
Subgroup Analyses
For both analytic approaches, we evaluated stage at diagnosis
among women ages 40–49, given that dense breasts are more
common among younger women.1 We also evaluated women
ages 50–74, because while dense breasts are less common,
screening overall is more common in this population.22
Sensitivity Analyses. Connecticut enacted breast density
notification legislation in late 2009. In a sensitivity analysis,
we excluded the first year post-legislation in our individuallevel model, as it was a transitional period. We also estimated a
model that included post-legislation data from California in
2013.
Power Calculation. For our county-level analyses, we used a
simulation-based method to estimate the minimum effect size
detectible with ≥80 % power for our effect of interest, the
change in incidence in counties in Connecticut compared to
control counties, in our county-level model.
RESULTS
Our individual-level analysis included 504,376 women ages
40–74 who were diagnosed with breast cancer between 2005
and 2013. We excluded 5515 women who did not have staging
data available, 13,616 women with LCIS, and 18,315 who
were diagnosed in California between April and December
2013. Our final analysis included 466,930 women with a
breast cancer diagnosis. Of these, 25,592 lived in Connecticut
at the time of diagnosis. Characteristics of the women in our
sample are reported in Table 1. Women in Connecticut were
more likely to be white, and a larger proportion had DCIS at
diagnosis than did women in other states. Figure 1 depicts the
proportion of women diagnosed at each stage in Connecticut
and in control states by year.
Our individual-level analysis examined changes in the proportion of women diagnosed at each stage over time among
women in Connecticut and control states. We found that the
proportion of women with a localized cancer increased from
50.0 % to 52.4 % for women in Connecticut during the postlegislation period. We observed a smaller increase in control
states, from 50.6 % to 51.6 %. Overall, we found a 1.38percentage-point (95 % CI 0.12 to 2.63) increase in the proportion of women diagnosed with localized breast cancer in
Connecticut post-legislation, after accounting for trends in
non-legislation states. We also found a 1.12-percentage-point
decrease (95 % CI −2.21 to −0.08) post-legislation in the
proportion of women diagnosed with DCIS in Connecticut,
after accounting for changes in control states. Changes in the
proportion of women diagnosed with regional-stage disease
(−0.09 percentage points, 95 % CI −1.01 to 1.02) and metastatic disease (−0.24 percentage points, 95 % CI −0.75 to 0.28)
did not differ between Connecticut and control states (Table 2).
Among women ages 40–49 and women ages 50–74, we did
not observe statistically significant changes in the proportion
of women diagnosed at each stage in Connecticut compared to
control states, though effect sizes were similar to what we
observed in our main analysis (Table 2).
When assessing pre-legislation trends, we found that the
proportion of women with metastatic disease at the time of
diagnosis declined slightly in Connecticut while remaining
stable in control states (approximate 0.30 percentage-point
decline per year, p = 0.02). In our main analysis, however,
we did not observe a decline in the proportion of women
diagnosed with metastatic disease in Connecticut compared
to control states, even though these divergent trends could
have biased us toward such a finding. We did not observe
Table 1 Characteristics of Women with Breast Cancer, 2005–2013
Characteristic
Connecticut
(n = 25,592)
Other states*
(n = 441,338)
p
value
Age at diagnosis,
mean (SD)
Race (%)
White
African American
Native American
Asian
Unknown
Ethnicity (%
Hispanic)
Stage (%)
DCIS
Localized
Regional
Metastatic
57 (9.3)
58 (9.2)
<0.001
89
7.8
0.11
1.8
1.2
7.1
79
11
0.59
8.7
0.6
10.1
<0.001
24
52
20
4
20
51
24
5
<0.001
<0.001
*Other states: AK, CA, CT, GA, HI, KY, IA, LA, MI, NJ, NM, UT, WA
606
Richman et al.: Breast Density Legislation and Breast Cancer Stage at Diagnosis
JGIM
Figure 1 Stage at diagnosis among women in Connecticut and control states. Panel A depicts the distribution of stage at diagnosis among
women ages 40–74 with breast cancer in control states. Panel B depicts the distribution of stage at diagnosis among women ages 40–74 with
breast cancer in Connecticut. *Breast density notification legislation was enacted in October of 2009.
divergent pre-legislation trends for the other three disease
stages.
In a sensitivity analysis, we used an alternate staging system
and found similar results (Online Appendix Table 1). Excluding the first post-policy year did not alter our results substantively (Online Appendix Table 2). In an analysis in which we
included post-legislation data from California, we observed a
small decline in the proportion of women with metastatic
disease at the time of diagnosis among women in states with
breast density notification legislation (−0.37 percentage
points, 95 % CI −0.59 to −0.14; Online Appendix Table 3).
This finding was only observed in models that included postlegislation data from California.
Our county-level analysis examined changes in stagespecific incidence and included 611 counties over a 9-year
period. Eight of these counties were in Connecticut and thus
were subject to breast density notification legislation from late
2009 onward. Online Appendix Table 4 describes the counties
in our sample.
Figure 2 depicts age-adjusted breast cancer incidence by
stage in Connecticut and in control states from 2005 to 2013.
Stage-specific incidence appears relatively flat in the postpolicy period in both Connecticut and control states.
In adjusted models, we found no statistically significant
changes in the incidence of DCIS (incidence rate ratio [IRR]
0.96, 95 % CI 0.90 to 1.03), localized disease (IRR 1.02, 95 %
CI 0.98 to 1.01), regional-stage disease (IRR 0.98, 95 % CI
0.92 to 1.04), or metastatic disease (IRR 0.98, 95 % CI 0.86 to
1.12) in counties in Connecticut compared to control counties
(Table 3). We estimated that we had 80 % power to detect a 23
Richman et al.: Breast Density Legislation and Breast Cancer Stage at Diagnosis
JGIM
607
Table 2 Individual-Level Analysis of Distribution of Stage at Diagnosis Among Women with a New Breast Cancer Diagnosis
Ages 40–74*
(n = 466,930)
DCIS
Localized
Regional
Metastatic
Ages 40–49*
(n = 105,301)
DCIS
Localized
Regional
Metastatic
Ages 50–74*
(n = 361,629)
DCIS
Localized
Regional
Metastatic
Connecticut
pre-legislation %
(95 % CI)
24.4 (23.6 to 25.1)
50.0 (49.1 to 50.8)
21.4 (20.7 to 22.2)
4.16 (3.82 to 4.51)
Connecticut
pre-legislation %
(95 % CI)
26.7 (25.2 to 28.8)
44.0 (42.4 to 45.7)
25.7 (24.2 to 27.1)
3.59 (2.95 to 4.23)
Connecticut
pre-legislation %
(95 % CI)
23.6 (22.8 to 24.5)
51.8 (50.8 to 52.8)
20.2 (19.4 to 21.0)
4.36 (3.94 to 4.77)
Connecticut
post-legislation %
(95 % CI)
23.5 (22.7 to 24.2)
52.4 (51.6 to 53.3)
19.7 (19.0 to 20.4)
4.39 (4.03 to 4.75)
Connecticut
post-legislation %
(95 % CI)
26.0 (24.4 to 27.6)
46.2 (44.4 to 48.0)
24.1 (22.5 to 25.6)
3.72 (3.02 to 4.42)
Connecticut
post-legislation %
(95 % CI)
22.7 (21.8 to 23.5)
54.3 (53.3 to 55.3)
18.5 (17.7 to 19.2)
4.60 (4.18 to 5.01)
Other states
pre-legislation %
(95 % CI)
19.6 (19.4 to 19.8)
50.6 (50.4 to 50.8)
24.6 (24.4 to 24.8)
5.21 (5.12 to 5.30)
Other states
pre-legislation %
(95 % CI)
20.7 (20.4 to 21.1)
45.8 (45.4 to 46.2)
28.8 (28.4 to 29.2)
4.66 (4.49 to 4.84)
Other states
pre-legislation %
(95 % CI)
19.3 (19.1 to 19.5)
51.9 (51.7 to 52.2)
23.4 (23.2 to 23.6)
5.37 (5.26 to 5.48)
Other states
post-legislation %
(95 % CI)
19.8 (19.7 to 20.0)
51.6 (51.4 to 51.8)
22.9 (22.7 to 23.1)
5.67 (5.57 to 5.77)
Other states
post-legislation %
(95 % CI)
21.3 (20.9 to 21.7)
46.1 (45.6 to 46.5)
27.4 (27.0 to 27.8)
5.22 (5.01 to 5.40)
Other states
post-legislation %
(95 % CI)
19.4 (19.2 to 19.6)
53.2 (53.0 to 53.5)
21.6 (21.4 to 21.8)
5.80 (5.69 to 5.91)
Difference in
differences %
(95 % CI)
−1.12 (−2.21 to −0.08)
1.38 (0.12 to 2.63)
−0.09 (−1.01 to 1.02)
−0.24 (−0.75 to 0.28)
Difference in
differences %
(95 % CI)
−1.27 (−3.50 to 0.97)
1.93 (−0.60 to 4.47)
−0.24 (−2.45 to 0.98)
−0.43 (−1.42 to 0.55)
Connecticut
pre-legislation %
(95 % CI)
−1.08 (−2.30 to 0.15)
1.17 (−0.27 to 2.62)
0.10 (−1.05 to 1.25)
−0.19 (0.79 to 0.41)
p value
0.04
0.03
0.99
0.37
p value
0.26
0.14
0.83
0.39
p value
0.08
0.11
0.87
0.53
*All models adjusted for age, race, ethnicity, and period from April 2013 on when data from California were excluded
% change in incidence of DCIS in counties in Connecticut
compared to other counties, a 13 % change in localized disease, a 16 % increase in regional-stage disease, and a 27 %
change in metastatic disease at an alpha = 0.05 level.
DISCUSSION
To the best of our knowledge, ours is the first study to evaluate
the effect of breast density notification legislation on breast
cancer stage at diagnosis. We found that passage of legislation
in Connecticut was associated with a small increase in the
proportion of women diagnosed with localized invasive cancer, after accounting for trends in states without breast density
notification legislation. We also observed a small decline in
the proportion of women diagnosed with DCIS. Although not
Figure 2 Age-adjusted breast cancer incidence by stage over time.
The graph depicts age-adjusted stage-specific breast cancer incidence in Connecticut and control states from 2005 to 2013. The
vertical bar represents the time of enactment of breast density
notification legislation in Connecticut.
definitive, these findings could be explained by an effect of
breast density legislation on supplemental screening, which
can detect additional early-stage invasive cancers.4 In a sensitivity analysis in which we included data from California, we
observed a small decline in the proportion of women with
metastatic disease on presentation. This finding is driven by
data from California, and is not likely attributable to passage of
breast density notification legislation, as such legislation had
only been in place for a matter of months in California, and
would have been very unlikely to have had an effect on
metastatic disease within that time frame.
Our main findings have two potential interpretations, each
with important implications. First, it is possible that because of
breast density notification legislation, more women in Connecticut have been diagnosed with an early-stage cancer and
have been spared an eventual late-stage diagnosis. This earlystage diagnosis may translate into important health benefits, as
these women may avoid the morbidity and mortality associated with a late-stage diagnosis. It is also possible, though, that
at least some of this increase in early-stage diagnosis is due to
overdiagnosis. In this case, the additional cancers detected are
not clinically significant, and finding them does not translate
into an improvement in health.
We pursued a county-level analysis to examine the association between legislation and stage-specific incidence and to
help distinguish these possibilities. Specifically, given stable
underlying disease rates, an increased incidence of early-stage
diagnosis with an associated decrease in late-stage diagnosis
would suggest that legislation promotes successful detection
of early-stage disease and averts late-stage diagnoses.14,15 In
contrast, an increase in incidence of early-stage disease, with
no change in late-stage disease, would suggest that legislation
may contribute to overdiagnosis. We did not detect a reduction
in late-stage cancer in Connecticut, a pattern that is consistent
with the explanation that supplemental screening has led to
608
Richman et al.: Breast Density Legislation and Breast Cancer Stage at Diagnosis
JGIM
Table 3 County-Level Analysis of Stage-Specific Incidence Among Women with a New Breast Cancer Diagnosis
Stage (n = 611 counties)
Ages 40–74*
DCIS
Localized
Regional
Metastatic
Ages 40–49**
DCIS
Localized
Regional
Metastatic
Ages 50–74**
DCIS
Localized
Regional
Metastatic
Connecticut pre- vs.
post-IRR (95 % CI)
Other counties pre- vs.
post-IRR (95 % CI)
Difference-in-differences
IRR (95 % CI)
p value
0.98
1.06
0.92
1.10
(0.91–1.04)
(1.02–1.11)
(0.86–0.98)
(0.97–1.25)
1.01 (0.99–1.03)
1.04 (1.03–1.05)
0.94 (0.93–0.95)
1.11 (1.09–1.15)
0.96
1.02
0.98
0.98
(0.90–1.03)
(0.98–1.07)
(0.92–1.04)
(0.86–1.12)
0.26
0.31
0.48
0.81
1.01
1.06
0.94
1.07
(0.91–1.12)
(0.98–1.15)
(0.84–1.04)
(0.81–1.41)
1.00
0.99
0.93
1.12
(0.97–1.03)
(0.97–1.01)
(0.91–0.96)
(1.05–1.19)
1.01
1.07
1.00
0.96
(0.91–1.12)
(0.99–1.16)
(0.90–1.12)
(0.72–1.27)
0.89
0.07
0.97
0.77
0.94
1.04
0.90
1.08
(0.87–1.00)
(0.99–1.08)
(0.84–0.97)
(0.94–1.25)
0.99
1.02
0.92
1.09
(0.97–1.01)
(1.01–1.04)
(0.91–0.94)
(1.06–1.13)
0.95 (0.88–1.02)
1.01 (–0.97–1.06)
0.97 (0.91–1.05)
0.99 (0.86–1.14)
0.13
0.59
0.47
0.90
*Models adjusted for median household income, percentage white, percentage insured, physicians per capita, percentage of women ages 40–65, and
period during 2013 when data from California were excluded
**Age-stratified models are adjusted for same set of covariates as main model except for age distribution
IRR = incidence rate ratio
overdiagnosis. Our sample, however, had limited power to
detect the small effects we saw in our individual-level analysis.
Furthermore, changes in late-stage disease may take years to
appear, given the natural history of breast cancer, and 4 years
of follow-up may not be sufficient to observe a change. Thus,
whether breast density notification legislation has led to improved health outcomes in Connecticut remains an open
question.
Our findings should be interpreted with caution. First, residual confounding may explain our results. While a
difference-in-differences analysis is robust to confounding
from group differences that are constant over time, changes
that occurred around the time of policy enactment and differentially affected Connecticut could explain our results. We
have controlled for key demographics which may change
differentially over time and could also affect screening rates,
but residual confounding remains possible.
Second, we examined four disease stages using two separate modeling strategies, an approach which contributes to an
increased likelihood of a spurious finding. With a correction
for multiple testing, our findings would not be statistically
significant. Still, our main findings are consistent with what
we might expect to observe a priori: if breast density legislation were to have any effect, it would be to increase diagnoses
of localized invasive disease. Additionally, while little is
known about how breast density notification laws have affected practice, there is some indication that some radiologists in
Connecticut have implemented programs to facilitate supplemental screening, and the use of supplemental screening may
have increased.23 Such changes in utilization could lead to a
greater proportion of women diagnosed at an early stage.
An assessment of the effect of breast density legislation
with the data we used has inherent limitations. It is possible
that physicians did not provide notification as mandated or that
many women who received a notification took no action, and
it is also possible that women in control states were notified
even in the absence of legislation. A careful analysis of supplemental screening patterns in Connecticut and control states
could potentially support a link between breast density notification legislation and stage at diagnosis, but such information
is not available from the data source we used.
Finally, given demographic, economic, and cultural differences, findings from Connecticut may not be applicable to
other states. Connecticut’s experience may also represent the
case in which breast density notification legislation would be
most likely to lead to a change in breast cancer detection, as
the law has received considerable attention there, resulting in
widespread awareness.24 In addition, Connecticut has an insurance provision that requires private insurers to cover supplemental screening for women with dense breasts, which
would likely facilitate access if women did indeed wish to
pursue supplemental screening. However, the majority of
states with breast density notification legislation do not mandate insurance coverage for supplemental screening.
In summary, breast density notification legislation is associated with a small increase in the proportion of women
diagnosed with early-stage invasive breast cancer. Given limited follow-up, we cannot determine whether this change has
led to improved health outcomes or reflects overdiagnosis. A
longer observation period and study of the effect of legislation
in other states will help in further assessing the impact of
breast density legislation.
Acknowledgments: Dr. Richman was supported by a VA postdoctoral fellowship in health services research. Drs. Asch and Owens
were supported by the Department of Veterans Affairs. Dr. Bhattacharya was supported by the National Institute on Aging grant R37AG036791. This work was presented at the Society of General
Internal Medicine Annual Meeting in Hollywood, Florida, May 2016.
Corresponding Author: Ilana Richman, MD; Center for Primary
Care and Outcomes Research/Center for Health PolicyStanford
University School of Medicine, 117 Encina Commons, Stanford, CA
94305, USA (e-mail: irichman@stanford.edu).
JGIM
Richman et al.: Breast Density Legislation and Breast Cancer Stage at Diagnosis
Compliance with Ethical Standards:
10.
Conflict of Interest: The authors declare that they do not have a
conflict of interest.
11.
Disclaimer: This work does not necessarily represent the views of the
Department of Veterans Affairs, and the authors are solely responsible
for its content.
12.
13.
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
Sprague BL, Gangnon RE, Burt V, et al. Prevalence of mammographically dense breasts in the United States. J Natl Cancer Inst. 2014;
106(10). doi:10.1093/jnci/dju255.
Boyd NF, Guo H, Martin LJ, et al. Mammographic density and the risk
and detection of breast cancer. N Engl J Med. 2007;356(3):227–36.
doi:10.1056/NEJMoa062790.
Kerlikowske K, Hubbard RA, Miglioretti DL, et al. Comparative effectiveness of digital versus film-screen mammography in community practice
in the United States: a cohort study. Ann Intern Med. 2011;155(8):493–
502. doi:10.7326/0003-4819-155-8-201110180-00005.
Melnikow J, Fenton JJ, Whitlock EP, et al. Supplemental screening for
breast cancer in women with dense breasts: a systematic review for the
U.S. Preventive Services Task Force. Ann Intern Med. 2016. doi:10.7326/
M15-1789.
Berg WA, Zhang Z, Lehrer D, et al. Detection of breast cancer with
addition of annual screening ultrasound or a single screening MRI to
mammography in women with elevated breast cancer risk. JAMA.
2012;307(13):1394–404. doi:10.1001/jama.2012.388.
Kriege M, Brekelmans CTM, Obdeijn IM, et al. Factors affecting
sensitivity and specificity of screening mammography and MRI in women
with an inherited risk for breast cancer. Breast Cancer Res Treat.
2006;100(1):109–19. doi:10.1007/s10549-006-9230-z.
Kuhl CK, Schrading S, Strobel K, Schild HH, Hilgers R-D, Bieling HB.
Abbreviated breast magnetic resonance imaging (MRI): first postcontrast
subtracted images and maximum-intensity projection-a novel approach
to breast cancer screening with MRI. J Clin Oncol. 2014;32(22):2304–10.
doi:10.1200/JCO.2013.52.5386.
Sprague BL, Stout NK, Schechter C, et al. Benefits, harms, and costeffectiveness of supplemental ultrasonography screening for women with
dense breasts. Ann Intern Med. 2015;162(3):157–66. doi:10.7326/M140692.
Scheel JR, Lee JM, Sprague BL, Lee CI, Lehman CD. Screening
ultrasound as an adjunct to mammography in women with mammographically dense breasts. Am J Obstet Gynecol. 2015;212(1):9–17.
doi:10.1016/j.ajog.2014.06.048.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
609
Gartlehner G, Thaler K, Chapman A, et al. Mammography in combination with breast ultrasonography versus mammography for breast
cancer screening in women at average risk. Cochrane Database Syst Rev.
2013;4:CD009632. doi:10.1002/14651858.CD009632.pub2.
Siu AL. Screening for Breast Cancer: U.S. Preventive Services Task Force
Recommendation Statement. Ann Intern Med. 2016;164(4):279–96.
doi:10.7326/M15-2886.
Are You Dense :: Twenty-Four State Density Reporting Laws & a Growing
Number of Introduced Bills in 2016. http://www.areyoudense.org/newsevents/twenty-four-state-density-reporting-laws-growing-number-introduced-bills-2016/. Accessed September 27, 2016.
Nelson HD, Pappas M, Cantor A, Griffin J, Daeges M, Humphrey L.
Harms of breast cancer screening: systematic review to update the 2009
U.S. Preventive Services Task Force Recommendation. Ann Intern Med.
2016;164(4):256. doi:10.7326/M15-0970.
Chu KC, Kramer BS, Smart CR. Analysis of the role of cancer prevention
and control measures in reducing cancer mortality. J Natl Cancer Inst.
1991;83(22):1636–43.
Bleyer A, Welch HG. Effect of three decades of screening mammography
on breast-cancer incidence. N Engl J Med. 2012;367(21):1998–2005.
doi:10.1056/NEJMoa1206809.
SEER Data, 1973-2012. http://seer.cancer.gov/data/. Accessed September 27, 2016.
Public Act No. 09-41 for Substitute Senate Bill No. 458. https://www.cga.
ct.gov/2009/ACT/PA/2009PA-00041-R00SB-00458-PA.htm. Accessed
SEptember 27, 2016.
Bill Text - SB-1538 Health care: mammograms. http://leginfo.legislature.
ca.gov/faces/billNavClient.xhtml?bill_id=201120120SB1538. Accessed
September 27, 2016.
Oliveira TMG, Elias J, Melo AF, et al. Evolving concepts in breast
lobular neoplasia and invasive lobular carcinoma, and their impact on
imaging methods. Insights Imag. 2014;5(2):183–94. doi:10.1007/
s13244-014-0324-6.
AHRF: Area Health Resources Files. http://ahrf.hrsa.gov/download.htm.
Accessed September 27, 2016.
RECODED SEER RESEARCH DATA RECORD DESCRIPTION 2015 http://seer.cancer.gov/data/seerstat/nov2014/TextData.FileDescription.pdf. Accessed September 27, 2016.
Block LD, Jarlenski MP, Wu AW, Bennett WL. Mammography use
among women ages 40-49 after the 2009 U.S. Preventive Services Task
Force recommendation. J Gen Intern Med. 2013;28(11):1447–53. doi:10.
1007/s11606-013-2482-5.
Parris T, Wakefield D, Frimmer H. Real world performance of screening
breast ultrasound following enactment of Connecticut Bill 458. Breast J.
19(1):64–70. doi: 10.1111/tbj.12053.
Rhodes DJ, Radecki Breitkopf C, Ziegenfuss JY, Jenkins SM, Vachon
CM. Awareness of breast density and its impact on breast cancer
detection and risk. J Clin Oncol. 2015;33(10):1143–50. doi:10.1200/
JCO.2014.57.0325.