Disparities in Pancreas Cancer Care
Click here to load reader
-
Upload
elizabeth-b -
Category
Documents
-
view
218 -
download
1
Transcript of Disparities in Pancreas Cancer Care
ORIGINAL ARTICLE – PANCREATIC TUMORS
Disparities in Pancreas Cancer Care
Anasooya Abraham1, Waddah B. Al-Refaie1,2, Helen M. Parsons1, Vikas Dudeja1, Selwyn M. Vickers1,
and Elizabeth B. Habermann1,3
1The Minnesota Surgical Outcomes Workgroup, University of Minnesota Surgical Outcomes Research Center, University
of Minnesota, Minneapolis, MN; 2Minneapolis Veterans Affairs Health Care System, Minneapolis, MN; 3Department of
Surgery, University of Minnesota, Minneapolis, MN
ABSTRACT
Background. Prior literature shows demographic differ-
ences in patients surgically treated for pancreatic cancer
(PC). We hypothesized that socioeconomic disparities also
exist across all aspects of PC care, in both surgically and
non-surgically treated patients.
Methods. We identified a cohort of patients with Ameri-
can Joint Committee on Cancer (AJCC) stage I–IV PC in
the 1994–2008 California Cancer Registry. We used mul-
tivariate logistic regression to examine the impact of race,
sex, and insurance status on (1) resectability (absence of
advanced disease), (2) receipt of surgery, and (3) receipt of
adjuvant/primary chemotherapy (?/– radiotherapy).
Results. Among 20,312 patients, 7,585 (37 %) had
resectable disease; 40 % who met this definition received
surgery (N = 3,153). On multivariate analysis, males were
less likely to present with resectable tumors [odds ratio
(OR) 0.91, 95 % confidence interval (CI) 0.85–0.96], but
sex did not otherwise predict treatment. Black patients
were as likely as White patients to show resectable disease,
yet were less likely to receive surgery (OR 0.66, 95 % CI
0.54–0.80), and adjuvant (OR 0.75, 95 % CI 0.58–0.98) or
primary chemotherapy ?/– radiation. Compared with
Medicaid recipients, non-Medicare/Medicaid enrollees
were more likely to receive surgery (OR 1.7, 95 % CI 1.4–
2.2), and the uninsured were less likely to receive adjuvant
therapy (OR 0.54, 95 % CI 0.30–0.98).
Conclusions. Though Black patients appear to present
with comparable rates of resectability, they receive care
that deviates from current guidelines. Insurance status is
associated with inferior profiles of resectability and treat-
ments. Future policies and research should identify
effective strategies to ensure receipt of standard care.
While over 40,000 new cases of pancreatic cancer are
diagnosed each year, 5-year survival is nominal and has
remained low for 30 years.1–3 Prior studies show that the
epidemiology, treatment, and survival of this malignancy
vary by demographics such as race and socioeconomic
status (SES).4,5 Black males suffer the highest incidence of
pancreatic cancer, yet Blacks are less likely to receive
specialty referral, surgical evaluation, cancer-directed sur-
gery, and treatment at high-volume centers compared with
Whites.1,6–8 Similarly, several studies have demonstrated
that pancreas cancer patients with lower SES have worse
postoperative mortality and survival.6,9,10 These disparities
also exist in access to and participation in clinical trials.
While much of the existing literature examines disparities
in the subset of patients with locoregional PC amenable to
surgery, nearly 75 % of pancreatic cancer patients present
with advanced disease.1 We build upon previous disparities
research and examine differences in treatment of pancreatic
cancer patients presenting with all disease stages.
Current National Comprehensive Cancer Network rec-
ommendations for locoregional disease include resection
and adjuvant therapy. Unresectable patients with good
performance status are offered chemoradiation. We
hypothesized that sociodemographic differences (age, sex,
race, insurance status) in the management of pancreatic
cancer exist along all points of care, from presentation to
treatment (Appendix 1). We examined variations by race,
sex, and insurance status on:
1. Resectability at presentation
2. Surgical resection in those with resectable disease
3. Chemotherapy receipt (with and without radiation)
after pancreatectomy
� Society of Surgical Oncology 2013
First Received: 10 February 2012;
Published Online: 12 April 2013
E. B. Habermann
e-mail: [email protected]
Ann Surg Oncol (2013) 20:2078–2087
DOI 10.1245/s10434-012-2843-z
4. Chemotherapy receipt [with and without radiotherapy
(RT)] as primary treatment in nonresectable PC
(locally advanced/metastatic)
The delivery of guideline-recommended care remains
paramount as more effective therapies are explored in
clinical trial settings to promote a combined approach to
improving population-based outcomes for pancreatic can-
cer in the USA.
METHODS
Study Design and Database
We used the 1994–2008 California Cancer Registry (CCR)
to conduct our retrospective, observational study. The CCR is
a statewide, population-based cancer surveillance system.
Abstractors collect data from 10 regional registries encom-
passing 58 counties according to documented standards on
demographics, cancer type, extent of disease, treatment, and
survival. Information on first course of therapy is provided but,
regarding chemotherapy, does not include specific agents,
dosages or duration/completion of therapy.
Patients
We limited our cohort to patients with adenocarcinomas
of the pancreatic head, body, and tail, International Clas-
sification of Disease site codes C250–C252; histology
codes 8010, 8020–8022, 8000, 8070–8078, 814–815 8210–
8211, 8230, 8260–8263, 8470–8471, 8480–8481, 8490,
8500–8503, 8560, 8562, 8570, 8574–8575, 8262, 8310,
8323, 8440, 8571–8573, 8576.
We excluded patients younger than 18 and older than
95 years at diagnosis (as there are no published guidelines
on age thresholds for offering therapies). Our upper and
lower age limits were chosen to allow appropriate and
reportable cell size for analysis and to preserve anonymity.
The race variable used in this study codes Hispanics as
White. As our focus was on differences in comparison with
Blacks, we felt this was an appropriate classification. We
excluded patients who had more than one prior cancer (any
site) and those with unknown disease extent, surgery
receipt, nodal status or chemotherapy status. Patients
treated with radiation alone were excluded due to inade-
quate cell size for analysis. Cases identified via death
certificates, nursing home records or coroner’s reports were
excluded. CCR data were censored on December 31, 2008.
Definition of Resectable Disease
In CCR, extent of disease is abstracted from pre- and/or
postoperative clinical and pathological data. We adhered to
National Comprehensive Cancer Network and American
College of Surgeons definitions of resectable disease.
While we cannot be sure that every clinician abides by
these definitions, they are generally established. Though all
data to achieve these definitions are not provided in CCR,
we utilized what was available to approximate the concept
of resectability. Accordingly, we defined tumors involving
major blood vessels (aorta, celiac axis, superior mesenteric
artery or vein), liver or diaphragm as unresectable. All
other localized disease or that involving stomach, spleen,
omentum, urologic organs, adrenal glands, colon or small
bowel was classified as resectable.
Adjuvant Chemotherapy and Radiation Definitions
CCR defines the first course of therapy as all therapy
received before disease progression or treatment failure; if
disease progression or treatment failures are not docu-
mented, first course is considered to be any treatment
received 1 year from diagnosis. Radiation and chemo-
therapy receipt are coded separately. Radiation was defined
as none versus beam/implants/isotopes/not otherwise
specified. Radiation alone was excluded from the study
since few persons received only radiation.
If data indicated that chemotherapy was (1) contraindi-
cated, (2) recommended, but not given, or (3) refused, or
(4) the patient died prior to administration, patients were
coded as not having received chemotherapy. Chemother-
apy and radiation receipt were examined separately among
patients who had and had not undergone surgery in order to
evaluate trends in adjuvant therapy versus primary che-
moradiotherapy receipt.
Constructed Variables
Surgery was defined as any pancreatectomy or local
excision. Local excision in CCR could include any partial
pancreatic excision and therefore could include the stan-
dard, curative pancreaticoduodenectomy.
We categorized age as 18–65 (a non-Medicare-eligible
subgroup), 65–69, 70–74, 75–79, and 80–95 years and race
as Black, White, and other. We classified insurance as
Medicare, Medicaid, non-Medicare/Medicaid [including
health maintenance organization (HMO), preferred pro-
vider organization (PPO), other managed care, Veterans
Affairs, TRICARE, Military, Indian/Public Health Service,
and county insurance], unknown, and no insurance.
Another method of managing unknown insurance status
would have been to exclude these cases. However, this
would result in loss of valuable information regarding
demographics and treatment (e.g., patients for whom race
and treatment factors were known but insurance status was
unknown). Unknown insurance status represents a distinct
Pancreas Cancer Care 2079
entity which we felt should be included in statistical
analysis.
Well-differentiated and moderately well-differentiated
tumors were classified as low grade, while poorly differ-
entiated and undifferentiated tumors were classified as high
grade. Finally, we classified disease extent as peripancre-
atic, beyond the pancreas or metastatic.
To examine therapy trends, patients were classified as
having undergone surgery alone, adjuvant chemotherapy
only, or both adjuvant chemotherapy and radiotherapy. As
radiation alone is not standard treatment for pancreatic cancer
and represented too small a group for analysis, this group was
excluded. Non-surgically treated patients were classified by
chemotherapy or chemoradiotherapy receipt. CCR does not
provide information on specific indications for chemoradia-
tion therapy in metastatic disease. However, these modalities
are frequently used as palliation and so were included.
Finally, in the analysis of chemoradiation receipt (as
adjuvant and primary therapy), persons who received
chemotherapy only or radiation only were classified as
having received no chemoradiation.
Statistical Analyses
We compared patient- and tumor-related factors by
resectability, surgery, and chemotherapy with or without
RT using the v2 test. The Cochran–Armitage method was
used to test trends. We used multivariate logistic regression
to predict the following: (1) resectability at diagnosis, (2)
receipt of surgery in those classified as resectable, (3)
adjuvant chemotherapy receipt with and without radiation,
and (4) chemotherapy (with or without RT) in those with
unresectable PC (locally advanced/metastatic), while
adjusting for covariates [age category, sex, race, insurance
status, grade, extent of disease, disease site (head, body,
tail), tumor size, and nodal positivity].
When main effects were significant, interactions for all
binary combinations of age group, sex, race, and insurance
status were tested. Interactions were considered significant
at p = 0.1 level, and all other results were significant at
p = 0.05 level using two-sided tests, when applicable. All
analyses were performed using SAS version 9.2 (SAS
Institute, Cary, NC). The study was approved by the
institutional Human Subjects Committee of the University
of Minnesota and by the California Cancer Registry.
RESULTS
Cohort Characteristics
During our study period, 20,312 patients with stage I–IV
pancreatic adenocarcinoma were identified (Table 1 and
Appendix 2). Median age at diagnosis was 70 years, and 51 %
were female. The majority of our population was White (82 %)
and insured (88 %). Up to 46 % of cases were metastatic.
Resectability at Diagnosis and Pancreatic Resection
While 37 % (N = 7,585) of the cohort were identified to
have resectable disease, only 42 % (N = 3,153) of these cases
underwent surgery. In other words, only 15 % (N = 3,153) of
the cohort underwent surgery. On bivariate analyses, we found
no significant differences in resectability across patient race
(38 % White, 37 % Black, 36 % other, p = 0.412) (Table 1).
However, a higher proportion of Whites underwent pancreatic
resection compared with Blacks (42 vs 36 %, p = 0.002).
After adjusting for covariates, increasing age, female sex,
non-Medicare/Medicaid insurance, and pancreatic head
tumors predicted presentation with resectable disease. Race
did not predict resectable disease (Table 2).
Among those with resectable disease, White race,
younger age, and non-Medicare/Medicaid and Medicare
insurance predicted undergoing pancreatectomy. Blacks
were 34 % less likely to undergo pancreatic resection as
compared with Whites (OR 0.66, 95 % CI 0.54–0.80).
However, sex did not predict surgery receipt (Table 3).
Adjuvant Chemotherapy (With or Without Radiation)
Among surgically treated PC patients, excluding those
who received radiation alone, 51 % (N = 1,999) received
chemotherapy alone and 32 % (N = 1,260) received che-
moradiotherapy (Table 4).
After adjusting for covariates, age, race, extent of dis-
ease, and nodal positivity all continued to predict
chemotherapy receipt. However, Black patients were 25 %
less likely to receive adjuvant chemotherapy (OR 0.75,
95 % CI 0.58–0.98). Sex was not a significant predictor of
receiving this treatment (Table 5).
Black patients were 30 % less likely to receive adjuvant
chemoradiation than Whites (OR 0.71. 95 % CI 0.53–
0.95). Furthermore, uninsured patients were half as likely
to receive adjuvant therapy. Again, sex did not predict this
treatment (Table 6).
Unresectable PC: Chemotherapy With or Without
Radiation, as Primary Treatment
Among persons with unresectable PC (N = 11,986),
42 % (N = 5,008) received chemotherapy and 10 %
(N = 1,140) received chemoradiotherapy. Whites received
primary chemotherapy and chemoradiation more fre-
quently than Blacks (42 vs 37 %, p = 0.001; 10 vs 6.0 %,
p \ 0.001). Only 32 % of uninsured patients received
primary chemotherapy compared with 50 % of non-
Medicare/Medicaid patients (p \ 0.001), and only 3.7 % of
2080 A. Abraham et al.
uninsured patients received primary chemoradiotherapy
compared with 12 % of non-Medicare/Medicaid patients
(p \ 0.001). Table 7 summarizes demographic and tumor
features of the unresectable, non-surgically treated cohort
receiving chemotherapy and chemoradiation.
Black patients were 30 % less likely to receive primary
chemotherapy compared with White patients (OR 0.69,
95 % CI 0.60–0.80). Additionally, we found a significant
age–sex interaction (p = 0.005) in the model predicting
primary chemotherapy (unresectable, non-surgically treated
patients). Stratification by age demonstrated that the oldest
males (age C80 years) were 1.5 times as likely (95 % CI
1.2–1.9) as females to receive primary chemotherapy. In
70–74-year-olds, males were 1.3 times as likely (95 % CI
1.1–1.6) and in 74–79-year-olds, males were also 1.3 times
as likely (95 % CI 1.0–1.5) to receive primary chemother-
apy. For those aged less than 65 years and 65–69 years, this
association was not significant. Table 8 summarizes the
adjusted odds of primary chemotherapy receipt.
Black patients were 50 % less likely to receive primary
chemoradiation, and uninsured patients were 70 % less
likely to receive primary chemoradiation than Medicaid
recipients (OR 0.31, 95 % CI 0.17–0.58). Increasing age
was associated with decreasing odds of both adjuvant and
primary chemoradiation. Sex did not predict primary che-
moradiation (Table 9).
TABLE 1 Demographic and
tumor features by resectability
and receipt of surgery
a Among those classified as
resectableb We classified insurance as
Medicare, Medicaid, non-
Medicare/Medicaid (including
HMO, PPO, other managed
care, Veterans Affairs,
TRICARE, Military, Indian/
Public Health Service, and
county insurance), unknown,
and no insurance
N (%), total cohort (N = 20,312)
Variable Resectable
disease, 7,585 (37)
p value Receipt of
surgery,a3,153 (42)
p value
Age (years) \0.001 \0.001
18–65 2,352 (31) 1,433 (45)
65–69 1,036 (14) 545 (17)
70–74 1,221 (16) 556 (18)
75–79 1,214 (16) 392 (12)
80–95 1,762 (23) 227 (7.0)
Sex \0.001 0.001
Female 4,079 (54) 1,624 (52)
Male 3,506 (46) 1,529 (48)
Race 0.412 0.008
Black 602 (7.9) 215 (6.8)
White 6,239 (82) 2,633 (84)
Other 744 (9.8) 305 (9.7)
Insurance status \0.001 \0.001
Non-Medicare/Medicaidb 3,345 (44) 1,603 (51)
Medicare 3,024 (40) 1,044 (33)
Medicaid 427 (5.6) 185 (5.9)
None 146 (1.9) 58 (1.8)
Unknown 643 (8.5) 263 (8.3)
Extent of disease \0.001 \0.001
Peripancreatic 6,646 (88) 2,946 (93)
Beyond the pancreas 939 (12) 207 (7.0)
Metastatic 0 (0) 0 (0)
Tumor size \0.001 \0.001
\2 cm 483 (6.0) 352 (11)
C2 cm 5,395 (71) 2,601 (82)
Unknown 1,707 (23) 200 (6.3)
Grade \0.001 \0.001
Low 2,739 (36) 1,864 (59)
High 1,679 (22) 1,036 (33)
Unknown 3,167 (42) 253 (8.0)
Pancreas Cancer Care 2081
Sensitivity Analyses
We carried out several additional analyses to test whe-
ther our findings resulted from modeling decisions. On
multivariate analysis reported above, we found that age,
sex, and insurance status predicted higher nonadherence to
resection when indicated. Next, as surgery, chemotherapy,
and radiation may not be offered in the context of short life
expectancy or poor postoperative course, we repeated our
analyses excluding those who died within 2 months of
diagnosis. These estimates remained comparable (data not
shown). Finally, we repeated our analyses after adding
‘‘prior cancer’’ as a covariate. Prior cancer was a strong
predictor of presenting with resectable disease (OR 3.3,
95 % CI 2.6–4.2) and of receipt of surgery (OR 2.4, 95 %
CI 1.7–3.4); however, this variable did not alter adjuvant
chemotherapy or chemoradiation therapy receipt after
pancreatectomy or in those with nonresectable PC.
DISCUSSION
In this population-based analysis we have shown that
sociodemographic factors impact pancreatic cancer care
from presentation to therapy. While Black race did not
predict resectable disease, it consistently predicted
decreased rates of pancreatic resection when indicated,
adjuvant chemoradiation, and primary chemoradiation
receipt in unresectable disease. Non-Medicare/Medicaid
insurance predicted resectability, and receipt of surgery and
primary chemoradiation. To our knowledge, the present
study is the first to evaluate disparities in the continuum of
PC care for both those receiving and not receiving surgery.
In agreement with prior studies, our results confirm that
extent of disease at presentation does not vary by patient
race. This has been previously reported using the Alabama
Cancer Registry and Surveillance, Epidemiology, and End
Results (SEER) data, among others.7,11–13 Whether resec-
tion rates vary by race, however, appears debatable. In
work by Shaver et al., race did not predict cancer-directed
surgery;12 however, other studies have reported that Black
race is associated with lower relative odds of resection.7,13
Previous work with SEER and SEER-Medicare demon-
strated that Black patients were 0.64 times as likely (95 %
CI 0.49–0.84) to receive surgery compared with Whites.7
We have validated these inequalities in resection among
Black patients within the context of comparable stage at
presentation.
With respect to chemotherapy and radiation, much of
the existing literature does not distinguish between adju-
vant therapy versus primary therapy for nonresectable (i.e.,
locally advanced or metastatic) PC. Nevertheless, prior
literature would suggest that, overall, Black patients
receive chemotherapy for pancreatic cancer less often than
Whites; For example, Shaver’s group reports an adjusted
odds ratio of 0.61 (95 % CI 0.37–0.95) for chemotherapy
receipt among Black patients compared with White
patients, though no distinction is made as to whether this
therapy was after pancreatic resection or for those with
TABLE 2 Predictors of resectability at diagnosis
Odds ratio 95 % CI p-Value
Race
Black versus white 1.0 0.91–1.1 0.859
Other versus white 0.97 0.88–1.1
Sex
Male versus female 0.91 0.85–0.96 0.001
Insurance status
Medicaid Referent
Non-Medicare/Medicaid 1.1 1.0–1.3 0.045
Medicare 1.1 0.96–1.3 0.017
None 0.89 0.71–1.1 0.334
Unknown 1.1 0.96–1.3 0.155
Age group (years)
\65 Referent
65–69 1.2 1.0–1.3 0.005
70–74 1.2 1.1–1.3 0.001
75–79 1.4 1.2–1.5 \0.001
80–95 1.7 1.5–1.8 \0.001
After adjusting for patient and tumor factors
TABLE 3 Predictors of surgery, in resectable cases
Odds ratio 95 % CI p value
Race
Black versus white 0.66 0.54–0.80 \0.001
Other versus white 0.93 0.78–1.1 0.441
Sex
Male versus female 0.95 0.85–1.1 0.375
Insurance status
Medicaid Referent
Non-Medicare/Medicaid 1.7 1.4–2.2 \0.001
Medicare 1.8 1.4–2.4 \0.001
None 0.83 0.54–1.3 0.386
Unknown 1.8 1.4–2.5 \0.001
Age group (years)
\65 Referent
65–69 0.66 0.55–0.78 \0.001
70–74 0.46 0.39–0.55 \0.001
75–79 0.26 0.22–0.31 \0.001
80–95 0.071 0.059–0.087 \0.001
After adjusting for patient and tumor factors
2082 A. Abraham et al.
locally advanced/metastatic PC.12 Eloubeidi’s group offers
an unadjusted analysis indicating that 27 % of Black
patients compared with 32 % of White patients (p = 0.02)
received chemotherapy across all stages. Further, they
report that this difference persisted in distant disease when
the cohort was stratified by stage.11 In our analysis, how-
ever, we go beyond previous studies and identify that, in
both the adjuvant and primary setting, after adjusting for
covariates, Black race predicts lower chemotherapy receipt
compared with White race.
While the indications for palliative therapy may be
variable, to our knowledge, the present study is also the
first to evaluate disparities in chemoradiation considered as
either combined-modality adjuvant therapy or primary
treatment for those with unresectable PC. Prior research
has suggested that there are no differences in radiation
receipt by race in adjusted and unadjusted analyses.11,12 On
the other hand, recent analysis of the Florida Cancer
Registry suggests that patients from less affluent areas with
higher proportions of Black population were less likely to
receive radiation therapy (14 vs 17 %, p = 0.003).9
However, specific poverty and race interactions and mul-
tivariate modeling of radiation receipt were not reported.
We observed that Black patients and those without insur-
ance coverage receive chemoradiation, as adjuvant therapy
or as primary therapy, less frequently; however, we found
no interaction of insurance with race in our regression
models.
Though surrogates for SES have been studied, there is a
general paucity of literature on the associations of
TABLE 4 Demographic and
tumor features by chemotherapy
and chemoradiation receipt in
surgically treated patients
N (%)
Variable Chemotherapy
receipt,
1,999 (51)
p value Chemoradiation
receipt,
1,260 (32)
p value
Age (years) \0.001 \0.001
\65 1,097 (55) 714 (57)
65–69 359 (18) 228 (18)
70–74 311 (16) 188 (15)
75–79 168 (8.0) 99 (8.0)
80–95 64 (3.0) 31 (3.0)
Sex 0.066 0.212
Female 995 (50) 627 (50)
Male 1,004 (50) 633 (50)
Race 0.383 \0.178
Black 123 (6.2) 73 (6.0)
White 1,683 (84) 1,071 (85)
Other 193 (9.8) 116 (9.0)
Insurance status \0.001 \0.001
Non-Medicare/Medicaid 1,104 (55) 703 (56)
Medicare 598 (30) 342 (27)
Medicaid 128 (6.0) 88 (7.0)
None 34 (2.0) 20 (2.0)
Unknown 135 (7.0) 107 (9.0)
Extent of disease \0.001 \0.001
Peripancreatic 1,460 (73) 980 (78)
Beyond the pancreas 405 (20) 248 (20)
Metastatic 134 (7.0) 32 (3.0)
Tumor size 0.003 \0.093
\2 cm 180 (9.0) 120 (10)
C2 cm 1,689 (84) 1,060 (84)
Unknown 130 (6.0) 80 (6.0)
Grade 0.010 \0.001
Low 1,162 (58) 771 (61)
High 679 (34) 415 (33)
Unknown 158 (8.0) 74 (6.0)
Pancreas Cancer Care 2083
insurance status with pancreatic cancer care. One recent
study reported that uninsured patients were 0.07 times less
likely (95 % CI 0.01–0.49) to receive surgery compared
with insured patients.12 However, only 23 patients in that
study were uninsured, reducing the analytic power of these
results. In comparison, the present study offers 486 unin-
sured patients and identifies differences in adjusted
resectability rates, rates of pancreatic resections, and
adjuvant and primary chemoradiation receipt by insurance
status.
We report several limitations to this study inherent in
the database. CCR does not collect information on
comorbidities and performance status, both of which are
variables used in determining oncologic treatment and have
been reported to vary by sociodemographic factors.7,14
Confounding by comorbidity would likely result in over-
estimation of odds ratios reported in this study. Second,
California counties span the extremes of dense urban cities
to rural farmlands. It is possible that, in one or the other,
disparities are more pronounced. Third, prior researchers
have reported demographic differences in patient referral to
high-volume centers for pancreatic resection.15–18 This
element of confounding could also affect our results such
that, potentially, the lower odds of therapy receipt in Black
patients could be accounted for by biased referral to low-
volume centers. Fourth, our conservative definition of
resectable disease may have under- or overestimated those
with resectable cancers. Finally, the diverse population in
the State of California and statewide treatment trends may
not necessarily reflect trends in other regions of the USA.
The current study provides valuable contributions to the
disparities literature in pancreatic cancer care. First, much
of the prior literature has focused on patients with surgi-
cally resectable disease, who represent less than 30 % of
pancreatic cancer patients. Identifying disparities in the
remaining majority is crucial to equitable delivery of care.
Furthermore, the distinction between adjuvant chemoradi-
ation and primary chemoradiation is an important one, as
the populations to which these therapies are prescribed are
fundamentally different. As we reported, disparities exist-
ing in one of these populations do not necessarily exist in
the other. Finally, our relatively recent cohort, spanning
through 2008, reiterates that disparities continue to exist in
the continuum of pancreas cancer care.
The presumption of our work is that treatment dispari-
ties translate to survival disparities, or conversely that the
latter can be explained by the former. Many prior studies
have suggested that race-based survival differences are at
least partly explained by variations in treatment and
SES.13,19–21 The implication, then, is that reducing treat-
ment disparities will reduce survival disparities. In fact, a
lack of association between sociodemographics and cancer
survival under universal access systems perhaps begs this
argument.22,23 Perhaps not coincidentally, these popula-
tions of racial/ethnic minorities and socioeconomically
disadvantaged continue to be underenrolled in clinical tri-
als, calling into question the applicability of trial results to
these populations. Our results highlight specific popula-
tions (Black patients, females, underinsured) in which we
TABLE 5 Predictors of chemotherapy in surgically treated patients
Odds ratio 95 % CI p value
Race
Black versus white 0.75 0.58–0.98 \0.037
Other versus white 0.91 0.73–1.1 0.409
Sex
Male versus female 1.0 0.91–1.2 0.375
Insurance status
Medicaid Referent
Non-Medicare/Medicaid 1.0 0.77–1.4 0.854
Medicare 1.2 0.87–1.7 0.258
None 0.58 0.34–0.99 0.046
Unknown 0.75 0.52–1.1 0.116
Age group (years)
\65 Referent
65–69 0.67 0.55–0.82 \0.0001
70–74 0.48 0.39–0.58 \0.0001
75–79 0.29 0.23–0.37 \0.0001
80–95 0.16 0.12–0.23 \0.0001
After adjusting for patient, tumor, and treatment factors
TABLE 6 Predictors of chemoradiotherapy in surgically treated
patients
Odds ratio 95 % CI p value
Race
Black versus white 0.71 0.53–0.95 0.023
Other versus white 0.87 0.69–1.1 0.262
Sex
Male versus female 1.0 0.88–1.2 0.898
Age group (years)
\65 Referent
65–69 0.76 0.62–0.94 0.010
70–74 0.55 0.45–0.69 \0.001
75–79 0.37 0.28–0.48 \0.001
80–95 0.18 0.12–0.28 \0.001
Insurance status
Medicaid Referent
Non-Medicare/Medicaid 0.91 0.68–1.2 0.524
Medicare 0.93 0.67–1.3 0.647
None 0.54 0.30–0.98 0.044
Unknown 0.96 0.66–1.4 0.821
After adjusting for patient, tumor, and treatment factors
2084 A. Abraham et al.
must ensure not only availability and accessibility of
standard treatments but also the opportunities to participate
in clinical trials.
In conclusion, in this large population-based study, we
found that patient race, insurance status, and sex influence
pancreatic cancer care. Our findings provide insight into
previously observed treatment disparities, call for future
studies to identify barriers to treatment in at-risk popula-
tions, and support the need for specific efforts geared
toward ensuring the equitable delivery of recommended
care.
ACKNOWLEDGMENT This study was supported by Enhancing
Minority Participation in Clinical Trials (Empact), National Institute
on Minority Health and Health Disparities. Project Number:
5RC2MD004797-02. Recipient of the 2011 American Society of
Clinical Oncology Foundation Merit Award and 2011 Pancreas Club
meeting Poster of Note.
TABLE 7 Demographic and tumor features in inoperable, non-sur-
gically treated patients (N = 11,986)
N (%) (column percents do not add to 100)
Variable Chemotherapy
receipt, 5,008
(41.8)
p value Chemoradiation
receipt, 1,140
(9.5)
p value
Age (years) \0.001 \0.001
\65 2,508 (50) 613 (54)
65–69 811 (16) 171 (15)
70–74 791 (16) 163 (14)
75–79 552 (11) 121 (11)
80–95 346 (7.0) 72 (6.0)
Sex \0.001 0.101
Female 2,256 (45) 540 (47)
Male 2,752 (55) 600 (53)
Race \0.004 \0.001
Black 371 (8.0) 60 (5.0)
White 4,127 (82) 942 (83)
Other 510 (10) 138 (12)
Insurance status \0.001 \0.001
Non-Medicare/
Medicaid
2,727 (54) 637 (56)
Medicare 1,586 (32) 324 (28)
Medicaid 295 (6.0) 84 (7.0)
None 102 (2.0) 12 (1.0)
Unknown 298 (6.0) 83 (7.0)
Extent of disease \0.001 \0.001
Peripancreatic 0 (0) 0 (0)
Beyond the
pancreas
1,520 (30) 775 (68)
Metastatic 3,488 (70) 365 (32)
Tumor size \0.001 \0.001
\2 cm 105 (2.1) 18 (2.0)
C2 cm 3,577 (71) 871 (76)
Unknown 1,326 (26) 251 (22)
Grade \0.001 \0.001
Low 921 (18) 272 (24)
High 892 (18) 197 (17)
Unknown 3,195 (64) 671 (59)
TABLE 8 Predictors of chemotherapy for inoperable, non-surgically
treated patients
Odds ratio 95 % CI p value
Race
Black versus white 0.69 0.60–0.80 \0.001
Other versus white 0.92 0.86–1.1 0.777
Sex
Male versus female 1.1 1.0–1.2 0.001
Age group (years)
\65 Referent
65–69 0.60 0.53–0.68 \0.001
70–74 0.45 0.40–0.51 \0.001
75–79 0.32 0.28–0.36 \0.001
80–95 0.13 0.11–0.15 \0.001
Insurance status
Medicaid Referent Medicaid Referent
Non-Medicare/Medicaid 2.1 1.8–2.5 \0.001
Medicare 2.3 1.9–2.7 \0.001
None 0.70 0.53–0.93 0.014
Unknown 0.97 0.79–1.2 0.749
After adjusting for patient and tumor factors
TABLE 9 Predictors of chemoradiotherapy among inoperable, non-
surgically treated patients
Odds ratio 95 % CI p value
Race
Black versus white 0.53 0.41–0.70 \0.001
Other versus white 1.2 0.98–1.4 0.072
Sex
Male versus female 0.95 0.84–1.1 0.440
Age group (years)
\65 Referent
65–69 0.65 0.53–0.79 0.010
70–74 0.53 0.43–0.65 \0.001
75–79 0.44 0.35–0.55 \0.001
80–95 0.20 0.16–0.27 \0.001
Insurance status
Medicaid Referent
Non-Medicare/Medicaid 1.3 1.0–1.7 0.039
Medicare 1.3 0.97–1.7 0.079
None 0.31 0.17–0.58 0.002
Unknown 0.98 0.71–1.4 0.909
After adjusting for patient and tumor factors
Pancreas Cancer Care 2085
APPENDIX 1
Continuum of pancreas cancer care
APPENDIX 2
See Table 10.REFERENCES
1. Fast Stats: An interactive tool for access to SEER cancer statis-
tics. Surveillance Research Program, National Cancer Institute.
http://seer.cancer.gov/faststats. Accessed 20 Dec 2010.
2. Jemal A, Siegel R, Xu J, Ward E. Cancer statistics, 2010. CACancer J Clin. 2010;60(5): 277–300.
3. Baxter NN, Whitson BA, Tuttle TM. Trends in the treatment and
outcome of pancreatic cancer in the United States. Ann SurgOncol. 2007;14(4): 1320–6.
4. DeLancey JOL, Thun MJ, Jemal A, Ward EM. Recent trends in
black-white disparities in cancer mortality. Cancer EpidemiolBiomark Prev. 2008;17(11): 2908–12.
5. Bilimoria KY, Bentrem DJ, Ko CY, Stewart AK, Winchester DP,
Talamonti MS. National failure to operate on early stage pan-
creatic cancer. Ann Surg. 2007;246(2): 173–80.
6. Cress RD, Yin DX, Clarke L, Bold R, Holly EA. Survival among
patients with adenocarcinoma of the pancreas: A population-based
study (United States). Cancer Causes Control. 2006;17(4): 403–09.
7. Riall T, Townsend CJ, Kuo Y, Freeman J, Goodwin J. Dissecting
racial disparities in the treatment of patients with locoregional
pancreatic cancer: a 2-step process. Cancer. 2010;116(4): 930–9.
8. Lucas F, Stukel T, Morris A, Siewers A, Birkmeyer J. Race and
surgical mortality in the United States. Ann Surg. 2006;243(2): 281–6.
9. Cheung M, Yang R, Byrne M, Solorzano C, Nakeeb A, Koniaris L. Are
patients of low socioeconomic status receiving suboptimal manage-
ment for pancreatic adenocarcinoma?. Cancer. 2010;116(3): 723–33.
10. Krzyzanowska M, Weeks J, Earle C. Treatment of locally
advanced pancreatic cancer in the real world: population-based
practices and effectiveness. J Clin Oncol. 2003;21(18): 3409–14.
11. Eloubeidi MA, Desmond RA, Wilcox CM, et al. Prognostic
factors for survival in pancreatic cancer: a population-based
study. Am J Surg. 2006;192(3): 322–29.
12. Shavers VL, Harlan LC, Jackson M, Robinson J. Racial/ethnic
patterns of care for pancreatic cancer. J Palliative Med. 2009;12(7):
623–30.
13. Murphy MM, Simons JP, Hill JS, et al. Pancreatic resection a key
component to reducing racial disparities in pancreatic adenocar-
cinoma. Cancer. 2009;115(17): 3979–90.
14. Arnold L, Patel A, Yan Y, et al. Are racial disparities in pan-
creatic cancer explained by smoking and overweight/obesity?.
Cancer Epidemiol Biomark Prev. 2009;18(9): 2397–405.
15. Chang D, Zhang Y, Mukherjee D, et al. Variations in referral
patterns to high-volume centers for pancreatic cancer. J Am CollSurg. 2009;209(6): 720–6.
16. Riall T, Eschbach K, Townsend CJ, Nealon W, Freeman J, Goodwin
J. Trends and disparities in regionalization of pancreatic resection. JGastrointest Surg. 2007;11(10): 1242–51; discussion 51-2.
TABLE 10 Demographic and tumor characteristics of total cohort
(N = 20,312)
Variable N (%)
Age (years)
18–65 7,218 (36)
65–69 2,887 (14)
70–74 3,306 (16)
75–79 3,006 (15)
80–95 3,895 (19)
Sex
Female 10,404 (51)
Male 9,908 (49)
Race
Black 1,650 (8.0)
White 16,613 (82)
Other 2,049 (10)
Insurance status
Non-Medicare/Medicaida 9,206 (45)
Medicare 7,645 (38)
Medicaid 1,267 (6.2)
None 486 (2.4)
Unknown 1,708 (8.4)
Extent of disease
Peripancreatic 6,646 (33)
Beyond the pancreas 4,336 (21)
Metastatic 9,330 (46)
Tumor size
\2 cm 756 (4.0)
C2 cm 13,989 (69)
Unknown 5,566 (27)
Grade
Low 4,930 (24)
High 4,079 (20)
Unknown 11,303 (56)
a We classified insurance as Medicare, Medicaid, non-Medicare/Medicaid
(including HMO, PPO, other managed care, Veterans Affairs, TRICARE,
Military, Indian/Public Health Service, and county insurance), unknown, and
no insurance
Disease Diagnosis
Stage-Specific First Course of Therapy
Surgical Therapies
Non-Surgical Therapies
Adjuvant Therapy
Survival Death
2086 A. Abraham et al.
17. Epstein A, Gray B, Schlesinger M. Racial and ethnic differences
in the use of high-volume hospitals and surgeons. Arch Surg.2010;145(2): 179–86.
18. Liu J, Zingmond D, McGory M, et al. Disparities in the utiliza-
tion of high-volume hospitals for complex surgery. JAMA.
2006;296(16): 1973–80.
19. Murphy M, Simons J, Ng S, et al. Racial differences in cancer spe-
cialist consultation, treatment, and outcomes for locoregional
pancreatic adenocarcinoma. Ann Surg Oncol. 2009;16(11): 2968–77.
20. Zell J, Rhee J, Ziogas A, Lipkin S, Anton-Culver H. Race,
socioeconomic status, treatment, and survival time among pan-
creatic cancer cases in California. Cancer Epidemiol Biomark.Prev 2007;16(3): 546–52.
21. Komenaka IK, Martinez ME, Pennington RE, et al. Race and
ethnicity and breast cancer outcomes in an underinsured popu-
lation. J Natl Cancer Inst. 2010;102(15): 1178–87.
22. Kuhn Y, Koscielny A, Glowka T, Hirner A, Kalff J, Standop J.
Postresection survival outcomes of pancreatic cancer according to
demographic factors and socio-economic status. Eur J SurgOncol. 2010;36(5): 496–500.
23. Mulligan CR, Meram AD, Proctor CD, Wu H, Zhu K, Marrogi
AJ. Unlimited access to care: effect on racial disparity and
prognostic factors in lung cancer. Cancer Epidemiol BiomarkPrev. 2006;15(1): 25–31.
Pancreas Cancer Care 2087