Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Jun 1.
Published in final edited form as: Clin Breast Cancer. 2019 Sep 30;20(3):e261–e280. doi: 10.1016/j.clbc.2019.09.005

EFFECTS OF PRIOR MEDICATION REGIMEN FACTORS AND BIPOLAR AND PSYCHOTIC DISORDERS ON BREAST CANCER ENDOCRINE THERAPY ADHERENCE

Cole B Haskins 1,2, Joan M Neuner 3, Bradley D McDowell 4, Ryan M Carnahan 1, Jess G Fiedorowicz 1,5, Robert B Wallace 1, Brian J Smith 4,6, Elizabeth A Chrischilles 1
PMCID: PMC7103521  NIHMSID: NIHMS1543639  PMID: 32139273

Abstract

Background:

Endocrine therapy adherence remains a barrier to optimal estrogen receptor positive (ER+) breast cancer outcomes. We theorized that experience navigating difficult medication regimen factors, such as route of administration complexity, may improve subsequent adherence following stressful cancer diagnoses, but not in patients with bipolar and psychotic disorders at-risk for poor access and non-adherence.

Methods:

We included 21,894 women aged 68 or older at their first surgically treated stage I-IV ER+ breast cancer (2007-2013) from the SEER-Medicare dataset, 5.8% having bipolar and psychotic disorders. We required continuous fee-for-service Medicare (Parts A and B) for at least 36 months before and 18 after cancer diagnosis. “Medication regimen factors” in the Part D claims four months prior included the number of all medications used, pharmacy visits, and administration complexity (Medication Regimen Complexity Index subscale). Cox regression was used to model time-to-initiation and discontinuation, with longitudinal linear regression for adherence to endocrine therapy.

Results:

Women with more frequent prior medication use and pharmacy visits were more likely to initiate [“4+ Meds & 2+ Visits” vs “No Meds” HR 1.47 (95% CI 1.33-1.63)], adhere [+6.0% (95% CI 4.3-7.6)], and continuously use endocrine therapy [discontinuation HR 0.48 (95% CI 0.39-0.59)]. Medication administration complexity had modest effects. Difficult medication regimens were more common among patients with bipolar and psychotic disorders but had no statistically significant effects.

Conclusions:

Experience with frequent prior medication use and pharmacy visits may increase likelihood of endocrine therapy use in most patients, but not in those with bipolar and psychotic disorders.

MicroAbstract:

Breast cancer endocrine therapy non-adherence can worsen survival and recurrence risk. We examined patient medication regimens prior to breast cancer diagnosis, which may impact readiness for future adherence. Patients who used more medication and frequently visited pharmacies before cancer were more likely to adhere, but subgroups with behavioral risk factors (bipolar and psychotic disorders) saw no significant benefit.

Introduction

Oral endocrine therapies are used to treat estrogen receptor positive (ER+) breast cancer, reducing recurrence and mortality, but few patients fully adhere to therapy [1-5]. Prior research has examined sociodemographic, treatment, and cost factors related to non-adherence. Medication regimen characteristics prior to cancer have been less-well studied, especially among patients with serious mental illness, such as bipolar and psychotic disorders, who often have limited social support, access to care, debilitating symptoms, and general non-adherence [6-9]. Patients with these barriers may not easily navigate difficult regimens, impacting readiness for future endocrine therapy adherence.

Treatment burden is a multidimensional concept describing factors associated with health care utilization [10]. While taking numerous medications could be burdensome, this experience may promote skills for managing future cancer treatments. In a study of women aged 65 or older with breast cancer, those using multiple medications at baseline were more likely to complete tamoxifen therapy [11]. However, difficult medication administration routes, measured with the medication regimen complexity index (MRCI), may decrease chronic illness treatment adherence [12, 13]. The physical act of obtaining medication may impact adherence; reducing necessary pharmacy visits is associated with improved cardiovascular medication adherence in Medicare patients [14].

Normalization Process Theory has been applied to the treatment burden concept, describing how experience taking medication can impact organization and formation of health behaviors [15, 16]. Experience with difficult regimens may normalize medication use, altering long-term adherence. Medication regimen factors may differentially affect patients with mental illness. Negative mood before treatment was associated with endocrine therapy non-adherence [17]. Negative mood effects were amplified when combined with disease stage and symptoms, financial concerns, and complex medication regimens [17].

We previously examined preexisting mental illness and endocrine therapy adherence in Medicare enrolled women; initiation rates were suboptimal with 20% of women never initiating, and those with bipolar and psychotic disorders at increased risk [HR 0.93 (95% CI 0.87-0.99)] [18]. In the present study, we examined the relationship between prior medication regimen utilization factors (all prescription medications) and endocrine therapy initiation/adherence/discontinuation in a cohort of women from the Surveillance, Epidemiology and End Results (SEER)-Medicare database, including subgroup analyses of patients with bipolar and psychotic disorders. We defined “medication regimen factors” using burden-relevant claims-based variables likely to impact adherence, including number of medications and pharmacy visits, and ease of administration [12, 19].

Methods

Study Population

We used the SEER-Medicare database to identify 21,894 women with ER+ breast cancer, surgically treated and diagnosed between 2007 and 2013 at age 68 or older. The SEER cancer registries include approximately 28% of the US population [20]. We were able to identify cancer diagnoses and characteristics, and detailed fee-for-service Medicare claims including (International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM)), clinical procedures, and prescription drug claims (National Drug Codes (NDC)) [21].

This study included women with continuous fee-for-service Parts A and B Medicare coverage for at least 36 months prior to invasive breast cancer diagnosis and 18 months after, as well as Part D coverage at least four months before and 18 months after. Breast cancer diagnosis can take place over several health care encounters, so multiple diagnoses in a three-month span were defined by the earliest diagnosis date and highest stage (4.7% of cases). All breast cancers were diagnosed before death and SEER and Medicare files listed the same death date, as applicable. Patients with potentially inaccurate endocrine therapy Part D data were excluded, defined as greater than a 10:1 ratio of prescription fill size to days supplied (n=154). Censoring occurred at end of continuous Medicare coverage or claims, new breast cancer diagnosis, use of hospice care, or death.

Measures

SEER data provided baseline demographics and tumor characteristics (AJCC 6th edition stage, receptor status, diagnosis date) [21, 22]. General comorbidity was assessed in Medicare claims using a modified NCI index (excluding dementia) for a 36 month baseline period before breast cancer and reassessed every six months for 36 months after [23]. We measured function-related indicators of comorbidity in the year prior, such as mobility limitations, which have been associated with poor health outcomes in older adults [24, 25]. Patient home rurality was determined using county-level 2013 Rural-Urban Continuum codes [26]. Because the original study was aimed at exploring the effects of mental illness, we used the Diagnostic and Statistical Manual of Mental Disorders-IV to guide selection of groups such as bipolar and psychotic disorders (bipolar depression, schizophrenia, and other psychotic disorders) [18, 27]. We adjusted for anxiety, substance use, and cognitive-impairing dementia disorders because these were previously associated with endocrine therapy initiation, discontinuation, or adherence [18]. In the 36-month baseline period, we identified mental illness comorbidity using relevant ICD-9 diagnosis codes appearing at least once in an inpatient claim, or twice in outpatient claims 30 or more days apart.

Medicare Part D prescription drug claims were used to measure endocrine therapy initiation, occurring at the first claim following cancer diagnosis. Endocrine therapies included tamoxifen and toremifene (selective estrogen receptor modulators, SERMs), anastrozole, exemestane, and letrozole (aromatase inhibitors, AIs). We examined “medication regimen factors”, including the number of all prescription medications used, pharmacy visits, and medication administration complexity before breast cancer. A four-month lookback was selected to increase detection of prescriptions with greater than a one-month supply. Medications were identified by generic name, with unique medications counted once. We measured pharmacy visits by the number of unique dates of service, except dates within a three-day window of each other were considered the same pick-up event and counted only once. We measured administration complexity using the Medication Regimen Complexity Index (MRCI) subscale “A”, with oral tablet forms given the minimum score of “1” and higher scores assigned to difficult methods such as injections [13] (Appendix 1). We controlled for financial expenditure using total out-of-pocket medication costs.

Model Selection

Medication regimen factors influence on endocrine therapy initiation was assessed using Cox regression with follow-up starting at cancer diagnosis. Adjusted models also included stage, age, race, ethnicity, NCI comorbidity score, function-related indicators, rurality, mental illness, and year of diagnosis. The number of medications and pharmacy visits were combined into a summary variable representing these related concepts. In creating this “combined medication & visit score”, individual responses with significant hazard ratios were weighted and combined [“1-3 Meds”, “4+ Meds or 2+ Visits”, and “4+ Meds and 2+ Visits” (highest weight)], with “no medication use” as reference (Appendix 2). We performed a test for trend across medication regimen factor strata and subgroup analyses of patients with and without bipolar and psychotic disorders.

Endocrine therapy discontinuation was assessed using Cox regression models, with follow-up starting at initiation and concluding at a maximum five years of use. Discontinuation was defined as the day of no available medication, with no more fills in the remaining 90 or more follow-up days [29]. Adherence was analyzed using longitudinal linear regression models fit with generalized estimating equations (normal distribution, identity link, unstructured covariance matrix), with annually repeated Proportion Days Covered (PDC = days covered/days in follow-up) adherence [30]. If patients switched from SERM to AI endocrine therapies, residual medication was disregarded once the new medication was filled. Patient follow-up spent in skilled nursing facilities or a hospital was not analyzed, as staff dispense medication and claims should not be available in Medicare Part D [31]. Adherence values in a given year were excluded if a patient was in such facilities for over six months. Patients with less than one year of endocrine therapy follow-up did not contribute to adherence analyses. Sensitivity analyses were performed to assess confounding risk from inclusion of high stage breast cancer (limiting analysis to patients with stage I-III), and treatment delays related to adjuvant chemotherapy use (adjustment for treatment receipt, based on HCPCS claims). Statistical analyses were conducted using SAS v9.4 statistical software (SAS Institute Inc, Cary, NC).

Results

We identified 21,894 female patients with stage I-IV ER+ breast cancer diagnosed from 2007 to 2013 who met eligibility criteria. Most patients were age 68-84 years old at diagnosis (86.6%), White (89.0%), non-Hispanic (94.7%), lived in metropolitan areas (82.5%), had progesterone receptor positive (85.8%) stage I tumors (60.3%), with no function-related indicators (59.0%). The majority used four or more medications in the four months prior to breast cancer (69.8%), had five or more pharmacy visits (62.5%), and had medication administration complexity in the lowest strata (54.8%) (Table 1).

Table 1:

Characteristics of breast cancer patients diagnosed 2007-2013, according to bipolar and psychotic disorder status

Bipolar and psychotic disorder diagnosed
Covariate Level All N=21,894 No N=20,635 (94.2%) Yes N=1,259 (5.8%) p
value*
Age at Diagnosis 68-74 9,308 42.5% 8,732 42.3% 576 45.8% 0.006
75-84 9,659 44.1% 9,133 44.3% 526 41.8%
85-94 2,831 12.9% 2,685 13.0% 146 11.6%
95+ 96 0.4% 85 0.4% 11 0.9%
Race Asian a-- a-- 910 4.4% a-- a-- 0.19
Black 1,315 6.0% 1,233 6.0% 82 6.5%
Other a-- a-- 140 0.7% a-- a--
White 19,480 89.0% 18,352 88.9% 1,128 89.6%
Ethnicity Hispanic 1,166 5.3% 1,090 5.3% 76 6.0% 0.25
Non-Hispanic 20,728 94.7% 19,545 94.7% 1,183 94.0%
Rurality Metro 18,053 82.5% 16,957 82.2% 1,096 87.1% <.001
Non-metro, non-rural 3,361 15.4% 3,220 15.6% 141 11.2%
Rural a-- a-- a-- a-- a-- a--
Unknown a-- a-- a- a-- a-- a--
Year of Diagnosis 2007 1,961 9.0% 1,855 9.0% 106 8.4% 0.90
2008 2,968 13.6% 2,801 13.6% 167 13.3%
2009 3,144 14.4% 2,954 14.3% 190 15.1%
2010 3,131 14.3% 2,953 14.3% 178 14.1%
2011 3,300 15.1% 3,119 15.1% 181 14.4%
2012 3,492 15.9% 3,279 15.9% 213 16.9%
2013 3,898 17.8% 3,674 17.8% 224 17.8%
Stage at Diagnosis I 13,194 60.3% 12,503 60.6% 691 54.9% <.001
II 6,534 29.8% 6,119 29.7% 415 33.0%
III 1,516 6.9% 1,397 6.8% 119 9.5%
IV a-- a-- 195 0.9% a-- a--
Unknown a-- a-- 421 2.0% a-- a--
Receptor Status ER+PR+ 18,789 85.8% 17,718 85.9% 1,071 85.1% 0.43
ER+PR- 3,105 14.2% 2,917 14.1% 188 14.9%
Endocrine Therapy Class AIs 13,659 62.4% 12,889 62.5% 770 61.2% <.001
Both 1,903 8.7% 1,819 8.8% 84 6.7%
None 4,382 20.0% 4,074 19.7% 308 24.5%
SERMs 1,950 8.9% 1,853 9.0% 97 7.7%
Adjusted NCI Comorbidity Mean (Median) 2.1 (2.0) 2.0 2.0 3.1 3.0 <.001
Function-related Indicators 0 12,921 59.0% 12,741 61.7% 180 14.3% <.001
1 5,159 23.6% 4,718 22.9% 441 35.0%
2 1,995 9.1% 1,722 8.3% 273 21.7%
3+ 1,819 8.3% 1,454 7.0% 365 29.0%
Anxiety Disorder 2,086 9.5% 1,673 8.1% 413 32.8% <.001
Dementia Disorder 1,017 4.6% 742 3.6% 275 21.8% <.001
Substance Use Disorder 1,006 4.6% 846 4.1% 160 12.7% <.001
Number of Medicationsd 0-3 6,617 30.2% 6,474 31.4% 143 11.4% <.001
4-8 10,358 47.3% 9,833 47.7% 525 41.7%
9+ 4,919 22.5% 4,328 21.0% 591 46.9%
Number of Pharmacy Visits 0-1 2,463 11.2% 2,414 11.7% 49 3.9% <.001
2-4 5,755 26.3% 5,602 27.2% 153 12.1%
5-10 9,799 44.8% 9,234 44.7% 565 44.9%
11+ 3,877 17.7% 3,385 16.4% 492 39.1%
MRCI-A Route Complexity 0-1 12,006 54.8% 11,488 55.7% 518 41.1% <.001
2-4 7,252 33.1% 6,789 32.9% 463 36.8%
5+ 2,636 12.0% 2,358 11.4% 278 22.1%
Combined Medication & Visit Score No Meds 1,207 5.5% 1,182 5.7% 25 2.0% <.001
1-3 Meds 1,132 5.2% 1,115 5.4% 17 1.3%
4+ Meds or 2+ Visits 4,402 20.1% 4,294 20.8% 108 8.6%
4+ Meds & 2+ Visits 15,153 69.2% 14,044 68.1% 1,109 88.1%
Out-Of-Pocket Medication Cost $0 < $1 2,266 10.3% 2,007 9.7% 259 20.6% <.001
$1 < $50 5,882 26.9% 5,596 27.1% 286 22.7%
$50 < $500 11,874 54.2% 11,353 55.0% 521 41.4%
$500+ 1,872 8.6% 1,679 8.1% 193 15.3%
Number Who Initiated 17,512 80.0% 16,561 80.3% 951 75.5% <.001
Days to Initiation Mean (Median) 156.6 (120.0) 156.7 120.0 156.6 120.0 0.68
Number Who Discontinued 4,503 25.7% 4,248 255
bDays to Discontinuation Mean (Median) 701.5 (521.0) 704.8 523.0 647.1 458.0 0.28
Number with 1+ Years 14,517 66.3% 13,739 778
cAverage Adherence Mean (Median) 0.84 (0.90) 0.84 0.91 0.84 0.91 0.23
Follow-up Days Mean (Median) 1377.9 (1280.0) 1,383.3 1,280.0 1,288.3 1,157.0 <.001
Number Censored at Death 1,376 6.3% 1,242 6.0% 134.0 10.6% <.001
Open in a new tab

“Combined Medication & Visit” variable created from each variable’s significant response levels in separate adjusted models (Appendix 2).

a

Values under 11 and nearest strata censored to maintain confidentiality.

b

Discontinuation analysis censored after five years continuous use.

c

Adherence measured with the proportion of days covered (PDC), for patients with one or more years of observation with over half of this time outside hospitals and skilled nursing facilities.

d

Medication variables describe utilization in the four months prior to breast cancer diagnosis.

*

Parametric p value calculated by chi-square test, Wilcoxon for continuous variables.

Compared to patients without, the 5.8% of patients with bipolar and psychotic disorders were more likely to be younger at diagnosis (age 68-74 45.8% vs 42.3%, p=0.006), diagnosed after stage I (stage II 33.0% vs 29.7%, p<0.001), live in metropolitan areas (87.1% vs 82.2%, p<0.001), and have more function-related indicators (“3+” 29.0% vs 7.0%, p<0.001). Patients with bipolar and psychotic disorders more frequently used medication in the four months before breast cancer (“9+” medications 46.9% vs 21.0%, p<0.001), had eleven or more pharmacy visits (39.1% vs 16.4%, p<0.001), and administration complexity in the highest strata (22.1% vs 11.4%, p<0.001) (Table 1).

In multivariable adjusted models, patients with more medication and pharmacy visits remained more likely to initiate endocrine therapy [“4+ Meds & 2+ Visits” vs “No Meds” HR 1.47 (95% CI 1.33-1.63)], and those with moderate medication administration complexity [MRCI-A “2-4” HR 0.96 (95% CI 0.93-0.99)] were less likely. Patients with bipolar and psychotic disorders [HR 0.92 (95% CI 0.86-0.98)] were less likely to initiate after adjustment for these factors. Patients with more medication use and pharmacy visits were less likely to discontinue [HR 0.48 (95% CI 0.39-0.59)] and had 6.0% higher average adherence rates (p<0.001) (Table 2).

Table 2:

Bivariable and adjusted multivariable models: medication regimen factors and endocrine therapy initiation, discontinuation, adherence

Initiation
n=21,894		 aDiscontinuation
n=17,512		 bAdherence
n=17,512
Parameter Unadjusted HR
(95% CI)		 Adjusted
Multivariable HR
(95% CI)		 Unadjusted HR
(95% CI)		 Adjusted
Multivariable HR
(95% CI)		 Unadjusted Estimate
(95% CI)		 Adjusted Multivariable
Estimate (95% CI)
cIntercept - - - - - **0.773 (0.754, 0.793)
MRCI-A Route Complexity 2-4 0.99 (0.96-1.03) *0.96 (0.93-0.99) 1.01 (0.95-1.08) 1.07 (1.00-1.14) 0.004 (−0.002, 0.009) 0.001 (−0.004, 0.007)
(ref= 0-1) 5+ 1.02 (0.98-1.07) 1.00 (0.95-1.05) 1.05 (0.95-1.15) *1.12 (1.02-1.24) **0.017 (0.009, 0.025) **0.014 (0.006, 0.023)
Overall variable p value 0.57 0.08 0.62 0.04 <0.001 0.002
Combined Medication & Visit Score 1-3 Meds **1.30 (1.19-1.43) **1.34 (1.19-1.50) 0.87 (0.74-1.03) *0.71 (0.56-0.89) 0.005 (−0.011, 0.022) *0.024 (0.005, 0.044)
(ref= No Meds) 4+ Meds or 2+ Visits **1.27 (1.18-1.37) **1.36 (1.23-1.51) **0.74 (0.65-0.84) **0.56 (0.45-0.68) *0.022 (0.009, 0.036) **0.049 (0.032, 0.065)
4+ Meds & 2+ Visits **1.30 (1.22-1.40) **1.47 (1.33-1.63) **0.72 (0.64-0.81) **0.48 (0.39-0.59) **0.029 (0.016, 0.041) **0.060 (0.043, 0.076)
Overall variable p value <0.001 <0.001 <0.001 <0.001 <0.001 <0.001
Out-of-Pocket Medication Costs $1 < $50 **1.19 (1.13-1.26) 0.98 (0.91-1.06) **0.82 (0.74-0.92) 1.17 (0.99-1.39) 0.008 (−0.001, 0.017) **−0.021 (−0.032, −0.009)
(ref= $0 < $1) $50 < $500 **1.17 (1.11-1.23) 0.96 (0.89-1.03) 0.93 (0.84-1.03) *1.31 (1.11-1.55) −0.007 (−0.016, 0.001) **−0.042 (−0.053, −0.031)
$500+ 1.03 (0.96-1.10) *0.86 (0.79-0.94) 0.92 (0.81-1.06) *1.25 (1.03-1.51) 0.000 (−0.012, 0.012) **−0.037 (−0.050, −0.023)
Overall variable p value <0.001 <0.001 <0.001 <0.001 <0.001 <0.001
Stage at Diagnosis II **1.13 (1.10-1.17) **1.14 (1.10-1.18) **0.81 (0.76-0.87) **0.80 (0.75-0.85) −0.001 (−0.007, 0.004) −0.001 (−0.006, 0.005)
(ref= I) III *1.07 (1.01-1.13) 1.04 (0.99-1.10) **0.74 (0.65-0.84) **0.72 (0.64-0.82) *−0.016 (−0.027, −0.006) *−0.014 (−0.024, −0.004)
IV **1.50 (1.29-1.74) **1.57 (1.36-1.82) 1.24 (0.96-1.61) 1.13 (0.87-1.47) **−0.060 (−0.092, −0.029) *−0.048 (−0.078, −0.017)
Unknown **0.82 (0.73-0.92) *0.87 (0.77-0.97) 0.87 (0.70-1.09) 0.84 (0.67-1.05) −0.006 (−0.025, 0.014) −0.003 (−0.022, 0.017)
Overall variable p value <0.001 <0.001 <0.001 <0.001 <0.001 0.003
Race Asian *1.11 (1.04-1.20) *1.11 (1.03-1.19) *0.81 (0.69-0.94) *0.82 (0.70-0.95) *0.015 (0.004, 0.026) 0.010 (−0.002, 0.021)
(ref= White) Black 0.99 (0.93-1.05) 0.95 (0.89-1.01) **0.72 (0.63-0.83) **0.74 (0.64-0.86) *−0.012 (−0.023, −0.001) *−0.015 (−0.026, −0.004)
Other 1.02 (0.85-1.22) 0.98 (0.82-1.17) 0.87 (0.59-1.29) 0.91 (0.61-1.35) −0.033 (−0.067, 0.001) *−0.039 (−0.073, −0.005)
Overall variable p value 0.03 0.01 <0.001 <0.001 0.001 0.002
Ethnicity
(ref= Non-Hispanic)		 Hispanic 1.03 (0.97-1.10) 0.99 (0.93-1.06) *0.80 (0.69-0.92) *0.82 (0.71-0.95) −0.004 (−0.015, 0.007) −0.008 (−0.019, 0.003)
Age at Diagnosis 75-84 **0.91 (0.88-0.94) **0.92 (0.89-0.95) **1.24 (1.16-1.32) **1.27 (1.19-1.35) −0.003 (−0.008, 0.002) −0.003 (−0.008, 0.003)
(ref= 68-74) 85-94 **0.64 (0.60-0.67) **0.65 (0.61-0.68) **1.50 (1.36-1.65) **1.57 (1.42-1.73) *0.010 (0.001, 0.018) *0.012 (0.004, 0.020)
95+ **0.44 (0.33-0.59) **0.45 (0.34-0.60) *2.33 (1.40-3.87) **2.68 (1.61-4.48) −0.036 (−0.099, 0.027) −0.026 (−0.091, 0.039)
Overall variable p value <0.001 <0.001 <0.001 <0.001 0.03 0.01
Adjusted NCI Comorbidity Score
(6-month time varying)		 0.99 (0.99-1.00) 1.00 (0.99-1.01) 1.01 (0.99-1.02) 1.01 (1.00-1.03) **0.005 (0.004, 0.006) −0.001 (−0.002, 0.000)
Function-related Indicators 1 0.98 (0.94-1.02) 1.00 (0.96-1.03) 1.06 (0.99-1.14) 1.03 (0.96-1.11) *−0.010 (−0.017, −0.004) **−0.011 (−0.017, −0.005)
(ref= 0) 2 **0.87 (0.82-0.92) **0.91 (0.86-0.96) *1.19 (1.07-1.31) *1.13 (1.01-1.26) −0.003 (−0.012, 0.006) −0.005 (−0.014, 0.005)
3+ **0.91 (0.86-0.96) 0.97 (0.91-1.03) 1.02 (0.91-1.14) 0.92 (0.81-1.05) 0.005 (−0.004, 0.014) 0.003 (−0.007, 0.013)
Overall variable p value <0.001 0.01 0.009 0.04 0.004 0.002
Bipolar and Psychotic Disorder *0.90 (0.84-0.96) *0.92 (0.86-0.98) 1.13 (0.99-1.28) 1.12 (0.98-1.28) 0.001 (−0.010, 0.013) −0.002 (−0.013, 0.010)
Anxiety Disorder 1.00 (0.95-1.05) 0.99 (0.94-1.05) **1.24 (1.13-1.37) **1.23 (1.11-1.36) −0.004 (−0.013, 0.005) −0.007 (−0.016, 0.003)
Dementia Disorder **0.82 (0.76-0.88) *0.89 (0.82-0.97) 1.09 (0.94-1.28) 1.02 (0.86-1.20) 0.008 (−0.005, 0.021) 0.005 (−0.009, 0.018)
Substance Use Disorder 1.00 (0.93-1.07) 0.98 (0.91-1.05) **1.41 (1.24-1.61) **1.44 (1.26-1.64) **−0.026 (−0.040, −0.012) **−0.024 (−0.038, −0.010)
Home Rurality Metro **0.82 (0.74-0.91) **0.84 (0.76-0.92) 1.21 (0.99-1.49) 1.22 (0.99-1.50) **−0.032 (−0.046, −0.018) **−0.032 (−0.045, −0.018)
(ref= Rural) Non-metro *0.84 (0.76-0.93) *0.86 (0.77-0.95) 1.15 (0.92-1.42) 1.15 (0.93-1.43) *−0.022 (−0.037, −0.006) *−0.020 (−0.035, −0.006)
Unknown 1.59 (0.40-6.26) 1.54 (0.38-6.20) 3.38 (0.47-24.26) 3.33 (0.46-23.97) −0.046 (−0.162, 0.071) −0.040 (−0.143, 0.064)
Overall variable p value <0.001 0.003 0.11 0.11 <0.001 <0.001
Year of Diagnosis 2008 0.99 (0.93-1.06) 0.98 (0.92-1.04) 0.95 (0.86-1.05) 0.95 (0.86-1.06) 0.002 (−0.009, 0.014) 0.002 (−0.009, 0.013)
(ref= 2007) 2009 1.01 (0.95-1.08) 1.01 (0.94-1.07) 0.91 (0.82-1.00) 0.91 (0.82-1.00) **0.021 (0.010, 0.032) **0.023 (0.012, 0.033)
2010 **1.13 (1.06-1.20) *1.10 (1.04-1.18) *0.86 (0.77-0.97) *0.86 (0.77-0.96) **0.049 (0.038, 0.059) **0.058 (0.048, 0.069)
2011 **1.18 (1.10-1.25) **1.16 (1.09-1.23) 0.91 (0.81-1.02) 0.91 (0.81-1.02) **0.049 (0.038, 0.059) **0.069 (0.059, 0.080)
2012 **1.17 (1.10-1.25) **1.15 (1.09-1.22) 0.92 (0.82-1.05) 0.92 (0.81-1.04) **0.048 (0.037, 0.058) **0.084 (0.073, 0.094)
2013 **1.21 (1.14-1.29) **1.18 (1.11-1.26) 0.94 (0.82-1.08) 0.93 (0.81-1.06) **0.024 (0.012, 0.035) **0.082 (0.071, 0.094)
Overall variable p value <0.001 <0.001 0.24 0.22 <0.001 <0.001
cYear of Endocrine Therapy Use 2 - - - - **0.061 (0.057, 0.065) **0.067 (0.063, 0.071)
(ref= 1) 3 - - - - **0.079 (0.074, 0.084) **0.089 (0.084, 0.094)
4 - - - - **0.100 (0.095, 0.106) **0.119 (0.113, 0.125)
5 - - - - **0.102 (0.095, 0.109) <0.001 **0.131 (0.123, 0.139)
Overall variable p value - - - - <0.001 <0.001
Open in a new tab

P value *<0.05, **<0.001, 95% confidence interval. Overall variable p value calculated for categorical variables with 3 or more strata. “Combined Medication & Visit” variable created from each variable’s significant response levels in separate adjusted models (Appendix 2).

a,b

Follow-up began the day of endocrine therapy initiation, concluding at censoring or a maximum five years of endocrine therapy use. Discontinuation was assessed in Cox regression.

a

Discontinuation analysis censored after five years continuous use.

b

Adherence measured with the proportion of days covered (PDC), for patients with one or more years of observation with over half of this time outside hospitals and skilled nursing facilities. Estimate of 0.01 corresponds to 1%.

c

Applicable to adherence only.

In multivariable tests for trend, patients with more medications and pharmacy visits were more likely to initiate [HR 1.11 (95% CI 1.08-1.13) per strata], less likely to discontinue [HR 0.82 (95% CI 0.79-0.86)], and had higher daily adherence [estimate +1.6% (95% CI 1.2, 1.9)]. As out-of-pocket costs increased, initiation decreased [HR 0.97 (95% CI 0.95-0.99)], discontinuation increased [HR 1.07 (95% CI 1.02-1.12)], and adherence decreased [estimate - 1.4% (95% CI −1.8, −1.1)]. Increasing administration complexity was associated with discontinuation [HR 1.06 (95% CI 1.01-1.11)], and adherence [estimate +0.6% (95% CI 0.2, 1.0)] (data not shown).

Among patients without bipolar and psychotic disorder diagnoses, prior medication use and pharmacy visits remained positively associated with endocrine therapy initiation [“4+ Meds & 2+ Visits” vs “No Meds” HR 1.47 (95% CI 1.33-1.64)], continuous use [discontinuation HR 0.49 (95% CI 0.39-0.61)], and adherence [estimate +5.4% (95% CI 3.7, 7.2)]. In contrast, patients with these mental illness diagnoses had no significant associations among these factors [“4+ Meds & 2+ Visits” vs “No Meds” HR 1.07 (95% CI 0.64-1.78); discontinuation HR 0.64 (95% CI 0.23-1.73); adherence +6.3% (95% CI −5.1, 17.8)], but high out-of-pocket costs were detrimental to initiation [“$500+” vs “$0 < $1” HR 0.73 (95% CI 0.57-0.93)] and adherence [−6.5% (95% CI −10.1, −3.0)] (Table 3).

Table 3:

Prior medication regimen effects on endocrine therapy adherence, by subgroups with and without bipolar and psychotic disorders

Initiation aDiscontinuation bAdherence
Bipolar and
Psychotic Disorder
n=1,259		 No Diagnosis
n=20,635		 Bipolar and
Psychotic Disorder
n=951		 No Diagnosis
n=16,561		 Bipolar and
Psychotic Disorder
n=951		 No Diagnosis
n=16,561
Parameter Hazard Ratio
(95% CI)		 Hazard Ratio
(95% CI)		 Hazard Ratio
(95% CI)		 Hazard Ratio
(95% CI)		 Estimate
(95% CI)		 Estimate
(95% CI)
cIntercept - - - - **0.711 (0.576, 0.846) **0.777 (0.757, 0.797)
MRCI-A Route Complexity 2-4 1.00 (0.85-1.16) *0.96 (0.93-1.00) 0.94 (0.69-1.27) *1.08 (1.00-1.16) 0.007 (−0.018, 0.031) 0.001 (−0.005, 0.006)
(ref= 0-1) 5+ 0.97 (0.81-1.17) 1.00 (0.95-1.06) 1.00 (0.69-1.44) *1.14 (1.02-1.26) 0.016 (−0.013, 0.046) *0.014 (0.005, 0.022)
Overall variable p value 0.95 0.06 0.89 0.03 0.56 0.005
Combined Medication & Visit Score 1-3 Meds 0.88 (0.42-1.85) **1.34 (1.19-1.51) 1.01 (0.25-4.05) *0.71 (0.56-0.90) −0.035 (−0.206, 0.136) 0.020 (0.000, 0.040)
(ref= No Meds) 4+ Meds or 2+ Visits 0.87 (0.50-1.50) **1.37 (1.23-1.52) 0.78 (0.27-2.22) **0.56 (0.45-0.70) 0.032 (−0.089, 0.153) **0.044 (0.026, 0.061)
4+ Meds & 2+ Visits 1.07 (0.64-1.78) **1.47 (1.33-1.64) 0.64 (0.23-1.73) **0.49 (0.39-0.61) 0.063 (−0.051, 0.178) **0.054 (0.037, 0.072)
Overall variable p value 0.42 <0.001 0.60 <0.001 0.19 <0.001
Out-of-Pocket Medication Costs $1 < $50 1.04 (0.83-1.29) 0.98 (0.90-1.07) 1.18 (0.73-1.90) 1.16 (0.96-1.40) −0.021 (−0.051, 0.009) *−0.016 (−0.029, −0.004)
(ref= $0 < $1) $50 < $500 0.96 (0.78-1.16) 0.96 (0.89-1.04) 1.43 (0.93-2.18) *1.28 (1.07-1.54) **−0.069 (−0.098, −0.039) **−0.036 (−0.048, −0.024)
$500+ *0.73 (0.57-0.93) *0.88 (0.80-0.97) 1.36 (0.83-2.23) 1.22 (1.00-1.50) **−0.065 (−0.101, −0.030) **−0.029 (−0.044, −0.015)
Overall variable p value 0.02 0.01 0.36 0.006 <0.001 <0.001
Open in a new tab

P value *<0.05, **<0.001, 95% confidence interval. Overall variable p value calculated for categorical variables with 3 or more strata. “Combined Medication & Visit” variable created from each variable’s significant response levels in separate adjusted models (Appendix 2). Models adjusted for stage, age, race, ethnicity, NCI comorbidity score, function-related indicators, rurality, mental illness, year of diagnosis (Appendix 3).

a,b

Follow-up began the day of endocrine therapy initiation, concluding at censoring or a maximum five years of endocrine therapy use.

a

Discontinuation analysis censored after five years continuous use.

b

Adherence measured with the proportion of days covered (PDC), for patients with one or more years of observation with over half of this time outside hospitals and skilled nursing facilities. Estimate of 0.01 corresponds to 1%.

c

Applicable to adherence only.

In sensitivity analyses restricted to stage I-III breast cancer and adjusting for receipt of adjuvant chemotherapy, the statistically significant medication regimen associations in our primary adjusted initiation/discontinuation/adherence models remained significant with comparable effects in their corresponding sensitivity analyses (Table 4). Only one variable's statistical significance changed in sensitivity analyses: the association between medication cost and discontinuation in the subgroup with bipolar and psychotic disorders (out-of-pocket costs "$50 < $500"; primary analysis HR 1.43 p=0.10; sensitivity analysis HR 1.57 p=0.04). The use of adjuvant chemotherapy was independently associated with endocrine therapy initiation, (HR 0.48, p<0.001).

Table 4:

Sensitivity analysis: medication regimen effects in multivariable models limited to stage I-III breast cancer patients with adjustment for adjuvant chemotherapy use, compared to primary analysis results

Initiation aDiscontinuation bAdherence
Full Cohort
n=21,244		 Bipolar and
Psychotic Disorder
n=1,225		 Full Cohort
n=17,020		 Bipolar and
Psychotic Disorder
n=928		 Full Cohort
n=17,020		 Bipolar and
Psychotic Disorder
n=928
Parameter Sensitivity Analysis
Hazard Ratio
(Primary Analysis)		 Sensitivity Analysis
Hazard Ratio
(Primary Analysis)		 Sensitivity Analysis
Hazard Ratio
(Primary Analysis)		 Sensitivity Analysis
Hazard Ratio
(Primary Analysis)		 Sensitivity Analysis
Estimate
(Primary Analysis)		 Sensitivity Analysis
Estimate
(Primary Analysis)
MRCI-A Route Complexity 2-4 *0.96 (*0.96) 0.98 (1.00) 1.07 (1.07) 0.93 (0.94) 0.002 (0.001) 0.008 (0.007)
(ref= 0-1) 5+ 0.99 (1.00) 0.98 (0.97) *1.12 (*1.12) 1.03 (1.00) **0.015 (**0.014) 0.019 (0.016)
Combined Medication & Visit Score 1-3 Meds **1.29 (**1.34) 0.78 (0.88) *0.72 (*0.71) 0.96 (1.01) *0.028 (*0.024) −0.031 (−0.035)
(ref= No Meds) 4+ Meds or 2+ Visits **1.33 (**1.36) 0.75 (0.87) **0.56 (**0.56) 0.76 (0.78) **0.052 (**0.049) 0.040 (0.032)
4+ Meds & 2+ Visits **1.43 (**1.47) 0.92 (1.07) **0.49 (**0.48) 0.60 (0.64) **0.062 (**0.060) 0.065 (0.063)
Out-of-Pocket Medication Costs $1 < $50 1.02 (0.98) 1.14 (1.04) 1.16 (1.17) 1.23 (1.18) **−0.022 (**−0.021) −0.021 (−0.021)
(ref= $0 < $1) $50 < $500 1.00 (0.96) 1.05 (0.96) *1.30 (*1.31) *1.57 (1.43) **−0.043 (**−0.042) **−0.069 (**−0.069)
$500+ *0.91 (*0.86) *0.78 (*0.73) *1.23 (*1.25) 1.47 (1.36) **−0.039 (**−0.037) **−0.063 (**−0.065)
Adjuvant Chemotherapy Prior to Endocrine Therapy **0.48 (n/a) **0.52 (n/a) **0.83 (n/a) 0.80 (n/a) −0.003 (n/a) −0.018 (n/a)
Open in a new tab

P value *<0.05, **<0.001. Models adjusted for same variables in primary analysis: stage, age, race, ethnicity, NCI comorbidity score, function-related indicators, rurality, mental illness, year of diagnosis.

a,b

Follow-up began the day of endocrine therapy initiation, concluding at censoring or a maximum five years of endocrine therapy use.

a

Discontinuation analysis censored after five years continuous use.

b

Adherence measured with the proportion of days covered (PDC), for patients with one or more years of observation with over half of this time outside hospitals and skilled nursing facilities. Estimate of 0.01 corresponds to 1%.

Discussion

Overview:

In this cohort of Medicare-enrolled women with stage I-IV ER+ breast cancer, 20.0% never initiated endocrine therapy. Patients with any prior medication use in the four months prior to breast cancer diagnosis were more likely to initiate, continuously use, and adhere, with maximum effects in those with frequent medication use and pharmacy visits. Complex prior medication administration was associated with modestly detrimental effects to initiation and discontinuation, with some potential benefit to adherence. Despite high rates of medication use and pharmacy visits in patients with bipolar and psychotic disorders, we did not detect any potentially protective medication effects, in contrast to the remainder of the cohort.

Prior Medication Regimen Factors

Patients using any medication in the four months prior to cancer were more likely to initiate endocrine therapy, consistent with reports suggesting tamoxifen adherence increases with more medications used before breast cancer [11]. Implementation and embedding of behavior can lead to sustained change. Therefore, experience using multiple medications may increase readiness to adhere [15]. While medication administration complexity has been identified as a non-adherence risk factor [12], in our study patients with the highest strata of prior regimen complexity had an 1.4% increase in adherence rates. Administration complexity during cancer treatment may be burdensome, whereas experiences prior to cancer may normalize chronic regimen behaviors to some extent. Because oral endocrine therapies are simple to administer, experience with complex medications may have a limited effect.

While experience with medication administration complexity did not improve endocrine therapy initiation, patients with many pharmacy visits were more likely to initiate. Prior work suggests “synchronization”, the ability to pick up all medications at the same time, may improve adherence by reducing travel and simplifying management [32]. The current study is not designed to assess synchronization, but frequent visits prior to cancer may reflect the ability and readiness to travel and initiate treatment. Additionally, pharmacist-patient interactions may increase initiation likelihood. In a study of adherence to diabetes mellitus medication and initiation of concomitant therapies, patients with supplemental pharmacist counseling and follow-up contact were 38% more likely to initiate [33]. These interactions and education may influence key beliefs, such as medication distrust and uncertainty of efficacy, which are associated with non-initiation of osteoporosis treatment [34]. Among breast cancer patients in the Detroit and Los Angeles SEER registries, those reporting inadequate endocrine therapy education were significantly less likely to initiate [35]. Interventions providing education and counseling may be valuable in improving endocrine therapy initiation rates.

Medication Regimen Factors in Patients with Bipolar and Psychotic Disorders

Although patients with bipolar and psychotic disorders were less likely to initiate endocrine therapy, and often struggle with general daily medication adherence [36], we observed high rates of prior medication utilization in this subgroup and no significant associations between these medication regimen factors and initiation, discontinuation, or adherence. The trend in discontinuation across levels of prior medication and pharmacy use mirrored that in patients without these conditions to some extent, but analytic power was limited because most patients with bipolar and psychotic disorders had high medication and pharmacy use.

Bipolar and psychotic disorder symptoms may compromise the ability to acquire beneficial health behaviors. Alternatively, social factors including stigma, limited access to care, and poor provider-patient relationships may affect receipt of guideline concordant cancer and general health care [6-8]. Medication regimens for patients with bipolar and psychotic disorders include more psychiatric medications, compared to patients without these conditions (Appendices 4 and 5). Therefore, prior medication use may capture a different behavioral concept in patients with mental illness (e.g., increased psychiatric treatment, not general medication adherence). These patients were diagnosed at later stages, consistent with evidence suggesting patients with mental illness have delays in initial breast cancer diagnosis and primary treatment [37]. Interventions to improve endocrine therapy use should consider unique approaches for these and other at-risk patients with individual and societal challenges.

Sensitivity Analyses

Our sensitivity analyses were performed to address confounding risk, including late stage cancer diagnosis as it may relate to symptom severity which could influence both baseline medication use and future endocrine therapy adherence. After excluding stage IV cancer patients, our medication regimen findings remained significant, with minimal change to the magnitude of effects. This suggests prior medication regimen factors are associated with endocrine therapy use, even after acknowledging potential treatment differences in patients with stage IV cancer.

We also examined potential confounding from adjuvant chemotherapy use. Adjuvant chemotherapy may be more frequently used in healthier patients able to tolerate treatment, who may have used fewer medications in the past. Also, patients may delay endocrine therapy initiation until adjuvant treatment has concluded. In our adjusted models, adjuvant chemotherapy was strongly associated with a delay in endocrine therapy initiation, however medication regimen effects remained significantly associated with outcomes. We can conclude that adjuvant chemotherapy is related to delays in endocrine therapy initiation, but this relationship does not meaningfully confound the independent effects of prior medication regimen factors.

Other Patient Characteristics

Patients with an intermediate number of function-related indicators of comorbidity (Appendix 6) were less likely to initiate endocrine therapies. A cohort study of myocardial infarction patients found those with more function-related indicators were less likely to use preventive medications [25]. Patients with some limitations in daily functioning may minimize travel, reducing pharmacy access, but patients with a greater functional burden may receive family or social assistance, reducing the impact of individual limitations. Patients in rural counties were more likely to initiate therapy, which may represent practice-level, cultural, or other unmeasured factors in these patients.

High endocrine therapy costs have been associated with non-adherence in commercially insured women [38]. We found that high out-of-pocket costs for any medication were associated with reduced endocrine therapy initiation and adherence rates. This is consistent with a more general body of research indicating that increased patient costs are associated with lower medication initiation rates [39]. Although conceptually interrelated, the detrimental effects associated with medication cost and protective effects of prior medication use and pharmacy visits remained independently significant in adjusted models. The negative impact of high cost on initiation and adherence remained significant among patients with bipolar and psychotic disorders, even though other medication regimen factors were not. Medicaid patients with schizophrenia have reduced antipsychotic utilization and increased emergency service use when monthly drug reimbursement is capped [40]. Excessive costs may contribute to symptom exacerbations, further reducing likelihood of endocrine therapy initiation and adherence.

Extending previous findings [3, 18], patients with stage I cancer and older age at diagnosis were less likely to initiate, even after adjustment for medication regimen factors. Perception of a poor risk-benefit ratio may influence cancer treatment in older women with competing mortality risks and potential endocrine therapy side effects, while stronger provider recommendations may be given to women with advanced cancers [41-43].

Study Strengths and Potential Limitations

We utilized a large SEER-Medicare cohort of older women with stage I-IV breast cancers. Medicare Part D claims allowed for examination of detailed prescription medication regimens, characterizing non-cancer treatments. We measured endocrine therapy initiation, adherence, and discontinuation in three separate models, as appropriate for these related but distinct outcomes. Our findings used variables reflective of clinically accessible data to predict preventable morbidity and mortality due to endocrine therapy non-adherence.

Potential study limitations are related to the nature of administrative claim studies. Medicare patients are not necessarily representative of younger women with other insurance providers, or managed care plans. However, breast cancers are often diagnosed in older women, making the disease burden relevant to the Medicare population [44]. We were unable to assess use of mail-order pharmacy services, which may simplify medication-taking behaviors [45]. While MRCI-A complexity calculations appropriately assess administration complexity, the remaining domains could not be reliably assessed: “additional instructions” are not included in claims and “dosing frequency” can only be partially estimated [13]. Administrative claims-based diagnoses risk potential misclassification and underestimation. Provider-level prescribing behaviors were not assessed but could affect medication utilization and the likelihood of initiation. Finally, our prior medication regimen variables may be correlated with factors that are unavailable or partially unmeasured in these data, for example we cannot tell if low rates of prior medication use are due to fewer medications prescribed, or a general lack of adherence/willingness to fill initial prescriptions.

Conclusions

Despite the established benefits of endocrine therapy, one-fifth of female Medicare beneficiaries with ER+ breast cancer did not initiate treatment. Prior medication regimen factors including frequent medication use and pharmacy visits increased likelihood of initiation, continuous use, and adherence. After adjustment, patients with bipolar and psychotic disorders remained at risk for endocrine therapy non-initiation and were not significantly affected by medication regimen factors. Interventions focused on health behavior development and provider-patient counseling may increase initiation rates in the general population, but patients with bipolar and psychotic disorders may require targeted efforts.

Supplementary Material

1

Clinical Practice Points.

  • Breast cancer endocrine therapy non-adherence is known to reduce survival and increase recurrence risk. Therefore, identification of patients at risk for endocrine therapy non-adherence is critical to improve outcomes.

  • Patients with regimens prior to cancer including more medication and pharmacy visits were significantly more likely to initiate, adhere, and continuously use endocrine therapy.

  • In contrast to the general population, subgroups with bipolar and psychotic disorders had no significant association between extensive prior medication regimens and endocrine therapy adherence.

  • Familiarity with chronic medication regimens may be helpful for the development of adherence behaviors. Providers should consider patients’ prior experience with medication and relevant behavioral illnesses when identifying those at risk for endocrine therapy non-adherence.

Acknowledgements:

This study used the linked SEER-Medicare database. The interpretation and reporting of these data are the sole responsibility of the authors. The authors acknowledge the efforts of the National Cancer Institute; the Office of Research, Development and Information, CMS; Information Management Services (IMS), Inc.; and the Surveillance, Epidemiology, and End Results (SEER) Program tumor registries in the creation of the SEER-Medicare database. The collection of cancer incidence data used in this study was supported by the California Department of Public Health as part of the statewide cancer reporting program mandated by California Health and Safety Code Section 103885; the National Cancer Institute's Surveillance, Epidemiology and End Results Program under contract HHSN261201000140C awarded to the Cancer Prevention Institute of California, contract HHSN261201000035C awarded to the University of Southern California, and contract HHSN261201000034C awarded to the Public Health Institute; and the Centers for Disease Control and Prevention's National Program of Cancer Registries, under agreement # U58DP003862-01 awarded to the California Department of Public Health. The ideas and opinions expressed herein are those of the author(s) and endorsement by the State of California Department of Public Health, the National Cancer Institute, and the Centers for Disease Control and Prevention or their Contractors and Subcontractors is not intended nor should be inferred. The authors acknowledge the efforts of the National Cancer Institute; the Office of Research, Development and Information, CMS; Information Management Services (IMS), Inc.; and the Surveillance, Epidemiology, and End Results (SEER) Program tumor registries in the creation of the SEER-Medicare database.

Funding: This work was supported by the University of Iowa Holden Comprehensive Cancer Center Population Research Core (P30 CA086862), the University of Iowa Medical Scientist Training Program (T32 GM007337), and the National Institute on Minority Health and Health Disparities (R01 MD010728-01).

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Declarations of interest: none

Competing interests: All authors declare that they have no competing interests.

Ethics approval: All work in this study adheres to the ethical and legal guidelines of the University of Iowa IRB and United States of America.

Availability of data: The SEER-Medicare data used in this study are available with approval from the National Cancer Institute (NCI) and were under license for this study. Requests for specific data are available from the authors, if given permission from the NCI. https://healthcaredelivery.cancer.gov/seermedicare/obtain/

References:

  • 1.Partridge AH, et al. , Nonadherence to adjuvant tamoxifen therapy in women with primary breast cancer. J Clin Oncol, 2003. 21(4): p. 602–6. [DOI] [PubMed] [Google Scholar]
  • 2.Hershman DL, et al. , Early discontinuation and nonadherence to adjuvant hormonal therapy in a cohort of 8,769 early-stage breast cancer patients. J Clin Oncol, 2010. 28(27): p. 4120–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Doggrell SA, Adherence to oral endocrine treatments in women with breast cancer: can it be improved? Breast Cancer Res Treat, 2011. 129(2): p. 299–308. [DOI] [PubMed] [Google Scholar]
  • 4.Hayes DF and Lippman ME, Breast Cancer, in Harrison's Principles of Internal Medicine, 20e, Jameson JL, et al. , Editors. 2018, McGraw-Hill Education: New York, NY. [Google Scholar]
  • 5.Davies C, et al. , Long-term effects of continuing adjuvant tamoxifen to 10 years versus stopping at 5 years after diagnosis of oestrogen receptor-positive breast cancer: ATLAS, a randomised trial. Lancet, 2013. 381(9869): p. 805–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Meltzer HY, et al. , Chapter 16. Schizophrenia, in CURRENT Diagnosis & Treatment: Psychiatry, 2e, Ebert MH, et al. , Editors. 2008, The McGraw-Hill Companies: New York, NY. [Google Scholar]
  • 7.Reus VI, Mental Disorders, in Harrison's Principles of Internal Medicine, 19e, Kasper D, et al. , Editors. 2015, McGraw-Hill Education: New York, NY. [Google Scholar]
  • 8.Mestdagh A and Hansen B, Stigma in patients with schizophrenia receiving community mental health care: a review of qualitative studies. Journal of Social Psychiatry and Psychiatric Epidemiology, 2014. 49(1): p. 79–87. [DOI] [PubMed] [Google Scholar]
  • 9.Mitchell A, Why don’t patients take their medicine? Reasons and solutions in psychiatry. Advances in Psychiatric Treatment, 2007: p. 336–346. [Google Scholar]
  • 10.Sav A, et al. , Burden of treatment for chronic illness: a concept analysis and review of the literature. Health Expect, 2015. 18(3): p. 312–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Lash TL, et al. , Adherence to tamoxifen over the five-year course. Breast Cancer Res Treat, 2006. 99(2): p. 215–20. [DOI] [PubMed] [Google Scholar]
  • 12.Ingersoll KS and Cohen J, The impact of medication regimen factors on adherence to chronic treatment: a review of literature. J Behav Med, 2008. 31(3): p. 213–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.George J, et al. , Development and validation of the medication regimen complexity index. Ann Pharmacother, 2004. 38(9): p. 1369–76. [DOI] [PubMed] [Google Scholar]
  • 14.Krumme AA, et al. , Medication Synchronization Programs Improve Adherence To Cardiovascular Medications And Health Care Use. Health Aff (Millwood), 2018. 37(1): p. 125–133. [DOI] [PubMed] [Google Scholar]
  • 15.May C and Finch T, Implementing, embedding, and integrating practices: an outline of normalization process theory. Sociology, 2009. 43(3): p. 535–554. [Google Scholar]
  • 16.Gallacher K, et al. , Understanding patients' experiences of treatment burden in chronic heart failure using normalization process theory. Ann Fam Med, 2011. 9(3): p. 235–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Bender CM, et al. , Influence of patient and treatment factors on adherence to adjuvant endocrine therapy in breast cancer. Oncol Nurs Forum, 2014. 41(3): p. 274–85. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Haskins CB, et al. , Impact of preexisting mental illness on breast cancer endocrine therapy adherence. Breast Cancer Research and Treatment, 2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Ickovics JR and Meisler AW, Adherence in AIDS clinical trials: a framework for clinical research and clinical care. J Clin Epidemiol, 1997. 50(4): p. 385–91. [DOI] [PubMed] [Google Scholar]
  • 20.NIH. SEER-Medicare: SEER Program & Data. NIH Healthcare Delivery Research Program; 2018; Available from: https://healthcaredelivery.cancer.gov/seermedicare/aboutdata/program.html. [Google Scholar]
  • 21.NIH. SEER-Medicare: Medicare Claims Files. NIH Healthcare Delivery Research Program; 2018; Available from: https://healthcaredelivery.cancer.gov/seermedicare/medicare/claims.html#part_d. [Google Scholar]
  • 22.NIH. Breast Schema for 1988+ based on AJCC 6th edition. NIH Surveillance, Epidemiology, and End Results Program; 2018; Available from: https://seer.cancer.gov/seerstat/variables/seer/ajcc-stage/6th/breast.html. [Google Scholar]
  • 23.NIH. NCI Comorbidity Index Overview. NIH Healthcare Delivery Research Program; 2018; Available from: https://healthcaredelivery.cancer.gov/seermedicare/considerations/comorbidity.html. [Google Scholar]
  • 24.Chrischilles E, et al. , Beyond comorbidity: expanding the definition and measurement of complexity among older adults using administrative claims data. Med Care, 2014. 52 Suppl 3: p. S75–84. [DOI] [PubMed] [Google Scholar]
  • 25.Chrischilles EA, et al. , Association Between Preadmission Functional Status and Use and Effectiveness of Secondary Prevention Medications in Elderly Survivors of Acute Myocardial Infarction. J Am Geriatr Soc, 2016. 64(3): p. 526–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.USDA. Rural-Urban Continuum Codes. USDA Economic Research Service; 2018; Available from: https://www.ers.usda.gov/data-products/rural-urban-continuum-codes/. [Google Scholar]
  • 27.Diagnostic and statistical manual IV, ed. A.P. Association. 1994, Washington DC: American Psychiatric Association. [Google Scholar]
  • 28.NIH. SEER-Medicare: Calculation of Comorbidity Weights. NIH Healthcare Delivery Research Program; 2018; Available from: https://healthcaredelivery.cancer.gov/seermedicare/considerations/calculation.html. [Google Scholar]
  • 29.Tajeu GS, et al. , Trends in Antihypertensive Medication Discontinuation and Low Adherence Among Medicare Beneficiaries Initiating Treatment From 2007 to 2012. Hypertension, 2016. 68(3): p. 565–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Hess LM, et al. , Measurement of adherence in pharmacy administrative databases: a proposal for standard definitions and preferred measures. Ann Pharmacother, 2006. 40(7-8): p. 1280–88. [DOI] [PubMed] [Google Scholar]
  • 31.Chang A, A SAS Macro to Calculate the PDC Adjustment of Inpatient Stays. SAS Global Forum Proceedings 2015, 2015. Paper 3560–2015. [Google Scholar]
  • 32.Krumme AA, et al. , Prevalence, effectiveness, and characteristics of pharmacy-based medication synchronization programs. Am J Manag Care, 2016. 22(3): p. 179–86. [PubMed] [Google Scholar]
  • 33.Brennan TA, et al. , An integrated pharmacy-based program improved medication prescription and adherence rates in diabetes patients. Health Aff (Millwood), 2012. 31(1): p. 120–9. [DOI] [PubMed] [Google Scholar]
  • 34.Yood RA, et al. , Patient decision to initiate therapy for osteoporosis: the influence of knowledge and beliefs. J Gen Intern Med, 2008. 23(11): p. 1815–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Friese CR, et al. , Adjuvant endocrine therapy initiation and persistence in a diverse sample of patients with breast cancer. Breast Cancer Res Treat, 2013. 138(3): p. 931–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Lingam R and Scott J, Treatment non-adherence in affective disorders. Acta Psychiatr Scand, 2002. 105(3): p. 164–72. [DOI] [PubMed] [Google Scholar]
  • 37.Iglay K, et al. , Diagnosis and treatment delays among elderly breast cancer patients with pre-existing mental illness. Breast cancer research and treatment 2017. 166(1): p. 267–275. [DOI] [PubMed] [Google Scholar]
  • 38.Sedjo RL and Devine S, Predictors of non-adherence to aromatase inhibitors among commercially insured women with breast cancer. Breast Cancer Res Treat, 2011. 125(1): p. 191–200. [DOI] [PubMed] [Google Scholar]
  • 39.Goldman DP, Joyce GF, and Zheng Y, Prescription drug cost sharing: associations with medication and medical utilization and spending and health. JAMA, 2007. 298(1): p. 61–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Soumerai SB, et al. , Effects of a limiting Medicaid drug-reimbursement benefits on the use of psychotropic agents and acute mental health services by patients with schizophrenia. N Engl J Med, 1994. 331(10): p. 650–5. [DOI] [PubMed] [Google Scholar]
  • 41.Anastrozole, in Micromedex Solutions. Truven Health Analytics: Ann Arbor, MI. [Google Scholar]
  • 42.Exemestane, in Micromedex Solutions. Truven Health Analytics: Ann Arbor, MI. [Google Scholar]
  • 43.Tamoxifen Citrate, in Micromedex Solutions. Truven Health Analytics: Ann Arbor, MI. [Google Scholar]
  • 44.Howlader N, SEER Cancer Statistics Review, 1975-2014. National Cancer Institute, 2017. [Google Scholar]
  • 45.Farias AJ, et al. , Factors Associated with Adherence to Adjuvant Endocrine Therapy Among Privately Insured and Newly Diagnosed Breast Cancer Patients: A Quantile Regression Analysis. J Manag Care Spec Pharm, 2016. 22(8): p. 969–78. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1543639-supplement-1.pdf (299.5KB, pdf)

RESOURCES