Do Cancer Survivors Change Their Prescription Drug Use for
Financial Reasons? Findings From a Nationally Representative
Sample in the United States
Zhiyuan Zheng, PhD
1,2
, Xuesong Han, PhD
1
, Gery P. Guy Jr, PhD, MPH
3
, Amy J. Davidoff,
PhD, MS
4
, Chunyu Li, PhD, MS, MD
3
, Matthew P. Banegas, PhD, MPPH
5
, Donatus U.
Ekwueme, MS, PhD
3
, K. Robin Yabroff, PhD, MBA
6
, and Ahmedin Jemal, DVM, PhD
1
1
Surveillance and Health Services Research Program, American Cancer Society, Atlanta,
Georgia
2
University of Maryland School of Pharmacy, Baltimore, Maryland
3
Division of Cancer
Prevention and Control, Centers for Disease Control and Prevention, Atlanta, Georgia
4
Department of Health Policy and Management, Yale School of Public Health, New Haven,
Connecticut
5
The Center for Health Research, Kaiser Permanente, Portland, Oregon
6
Division of
Cancer Control and Population Sciences, National Cancer Institute, Rockville, Maryland
Abstract
BACKGROUND—There is limited evidence from nationally representative samples about
changes in prescription drug use for financial reasons among cancer survivors in the United States.
METHODS—The 2011 to 2014 National Health Interview Survey was used to identify adults
who reported ever having been told they had cancer (cancer survivors; n = 8931) and individuals
without a cancer history (n = 126,287). Measures of changes in prescription drug use for financial
reasons included: 1) skipping medication doses, 2) taking less medicine, 3) delaying filling a
prescription, 4) asking a doctor for lower cost medication, 5) buying prescription drugs from
another country, and 6) using alternative therapies. Multivariable logistic regression analyses were
controlled for demographic characteristics, number of comorbid conditions, interactions between
cancer history and number of comorbid conditions, and health insurance coverage. Main analyses
Corresponding author: Zhiyuan Zheng, PhD, Surveillance and Health Services Research, American Cancer Society, Inc., 250 Williams
Street, Atlanta, GA 30303; Fax: (410) 706-5394; jzheng@rx.umaryland.edu.
The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers
for Disease Control and Prevention or the US Department of Health and Human Services.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.
AUTHOR CONTRIBUTIONS
Zhiyuan Zheng: Contributed to study concept and design, acquisition of data, statistical analysis, interpretation of data, and writing of
the initial draft. Xuesong Han: Critical revision of the article for intellectual content. Gery P. Guy, Jr: Critical revision of the article
for intellectual content. Amy J. Davidoff: Critical revision of the article for intellectual content. Chunyu Li: Critical revision of the
article for intellectual content. Matthew P. Banegas: Critical revision of the article for intellectual content. Donatus U. Ekwueme:
Critical revision of the article for intellectual content. K. Robin Yabroff: Contributed to study concept and design, writing of the
initial draft, analysis and interpretation of data, critical article revision, study supervision, and guarantor of the overall content.
Ahmedin Jemal: Contributed to study concept and design, writing of the initial draft, analysis and interpretation of data, critical
article revision, study supervision, and guarantor of the overall content.
Additional supporting information may be found in the online version of this article.
HHS Public Access
Author manuscript
Cancer
. Author manuscript; available in PMC 2018 August 07.
Published in final edited form as:
Cancer
. 2017 April 15; 123(8): 1453–1463. doi:10.1002/cncr.30560.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
were stratified by age (nonelderly, ages 18–64 years; elderly, ages ≥65 years) and time since
diagnosis (recently diagnosed, <2 years; previously diagnosed, ≥2 years).
RESULTS—Among nonelderly individuals, both recently diagnosed (31.6%) and previously
diagnosed (27.9%) cancer survivors were more likely to report any change in prescription drug use
for financial reasons than those without a cancer history (21.4%), with the excess percentage
changes for individual measures ranging from 3.5% to 9.9% among previously diagnosed
survivors and from 2.6% to 2.7% among recently diagnosed survivors (
P
< .01). Elderly cancer
survivors and those without a cancer history had comparable rates of changes in prescription drug
use for financial reasons.
CONCLUSIONS—Nonelderly cancer survivors are particularly vulnerable to changes in
prescription drug use for financial reasons, suggesting that targeted efforts are needed.
Keywords
cancer survivors; comorbid conditions; financial burden; high-deductible plan; prescription drugs
INTRODUCTION
The rising cost of cancer drugs imposes a significant financial burden on patients with
cancer and their families.
1,2
The direct medical cost of a new cancer medication per patient
can routinely exceed $100,000 annually.
2,3
Because of rising deductibles, copayments,
coinsurance, and tiered drug formularies, an increasing portion of the cost of cancer drugs is
shifted to the patient as out-of-pocket (OOP) costs.
4,5
Previous studies demonstrated that
cancer survivors faced significantly higher OOP costs than individuals without a cancer
history.
6–8
The financial burden caused by the high OOP costs for cancer drugs may further
impair cancer survivors’ overall well being, adversely affect treatment choices and health
outcomes, and result in higher medical expenditures in the long run.
9–13
Moreover, the
increasing number of expensive, patient-administrated, oral antineoplastic agents may
increase the likelihood that patients with cancer may delay, skip, or even forgo their
prescription medications for financial reasons.
13–16
A cancer diagnosis is associated with reduced adherence to recommended prescription drugs
for other comorbid conditions, such as diabetes and cardiovascular disease, although, in
some studies, it was unknown whether this was for financial reasons.
17–19
Previous studies
demonstrating that patients with cancer make changes in prescription drug use to defray
OOP costs are limited to small sample sizes, specific cancer sites, single-institutional
experience, or trials without a noncancer control group.
11,20–22
In the current study, we used
nationally representative data to compare the changes in prescription drug use for financial
reasons between cancer survivors and individuals without a cancer history. Because the
intensity and aggressiveness of cancer-related treatments and associated costs are likely to be
greater in nonelderly than in elderly patients and in those diagnosed in the most recent time
period than in the past,
23–25
we stratified the analyses by age group and time since
diagnosis. Moreover, we examined whether high private health insurance deductibles were
associated with changes in prescription drug use for financial reasons among nonelderly
Zheng et al.
Page 2
Cancer
. Author manuscript; available in PMC 2018 August 07.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
cancer survivors. We also examined whether the pattern of changes in prescription drug use
varies according to the number of comorbid conditions among cancer survivors.
MATERIALS AND METHODS
Data Sources
The National Health Interview Survey (NHIS) from 2011 through 2014 was used to identify
cancer survivors and individuals without a cancer history. The NHIS is a cross-sectional
household interview survey conducted annually by the Centers for Disease Control and
Prevention’s National Center for Health Statistics. It is a nationally representative survey of
the civilian, noninstitutionalized population of the United States. The survey collects
information on health status, access to and use of health care services, and comorbid
conditions. The annual NHIS response rate ranged from 73.8% to 82% of eligible
households during our study period.
26
Individual-Level Characteristics and Analytic Sample
In the NHIS, cancer history is self-reported. We defined cancer survivors as those who
reported ever having been told by a physician or other health professional that they had
cancer or a malignancy of any kind. Cancer survivors who reported nonmelanoma skin
cancer or skin cancer with unknown type were excluded.
27,28
Individual-level demographic
and clinical characteristics included, age at the time of the survey (nonelderly: ages 18–49 or
50–64 years; elderly: ages 65–74 or ≥75 years), sex, race/ethnicity (non-Hispanic white,
non-Hispanic black, Hispanic, or other), educational attainment (≤high school graduate or
≥some college), marital status (married or not married), geographic region (Northeast,
Midwest, South, or West), health insurance coverage, and number of comorbid conditions
(0, 1, 2, or ≥3). The health insurance coverage variable was defined differently for the 2 age
groups (nonelderly: any private, other, or uninsured; elderly: Medicare with any private or
other). Comorbid conditions included: arthritis, asthma, diabetes, emphysema, coronary
heart disease, hypertension, stroke, angina pectoris, and heart attack.
Time since diagnosis was calculated using age at cancer diagnosis and age at the time of the
survey. We used 2 years as the cutoff point for categorizing those who were recently
diagnosed and previously diagnosed to be consistent with previous studies.
7,29
In the NHIS,
private health insurance with an annual deductible >$1200 per individual or $2400 per
family was categorized as a high-deductible plan.
26
An indicator for high insurance
deductible status was created among nonelderly cancer survivors who had any private health
insurance coverage during the survey year. The final sample was stratified by age group
(nonelderly population, ages 18–64 years; elderly population, ages ≥65 years) and time since
diagnosis (recently diagnosed, <2 years; previously diagnosed, ≥2 years), including recently
diagnosed (nonelderly, n = 686; elderly, n = 720) and previously diagnosed (nonelderly, n =
3295; elderly, n = 4230) cancer survivors and individuals without a cancer history
(nonelderly, n = 76,096; elderly, n =16,989).
Zheng et al.
Page 3
Cancer
. Author manuscript; available in PMC 2018 August 07.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Measures of Changes in Prescription Drug Use for Financial Reasons
The NHIS included 6 measures of changes in prescription drug use for financial reasons.
These are general questions about all prescriptions drugs and thus are not cancer specific.
Respondents were asked the following questions: “During the past 12 months, were any of
the following true for you: 1) you skipped medication doses to save money, 2) you took less
medicine to save money, 3) you delayed filling a prescription to save money, 4) you asked
your doctor for a lower cost medication to save money, 5) you bought prescription drugs
from another country to save money, 6) you used alternative therapies to save money.” A
summary measure (yes/no) of any of the 6 measures and a categorical count of changes in
prescription drug use for financial reasons (scored as 0, 1, or ≥2) also were also created.
Statistical Methods
We compared the distributions of individual-level characteristics between cancer survivors
and those without a cancer history. We used multivariable logistic regression models for
each of the 6 measures and the summary measure to estimate the changes in prescription
drug use for financial reasons for both cancer survivors and individuals without a cancer
history. Among nonelderly cancer survivors with private health insurance, we used
multivariable logistic regressions to compare the patterns of changes in prescription drug use
between high-deductible plans and low-deductible plans. We did not examine this
association among the elderly cancer survivors because of their nearly universal eligibility
for insurance through Medicare. Among all cancer survivors, multivariable ordered logistic
regression was used to investigate associations between the number of comorbid conditions
and changes in prescription drug use for financial reasons. Sensitivity analyses were
conducted to examine the robustness of our findings by evaluating various summary
measures based on a different reduced set of questions about changes in prescription drug
use.
All multivariable regressions adjusted for survey year, age, race/ethnicity, sex, educational
attainment, marital status, geographic region, health insurance coverage, the number of
comorbid conditions, and interactions between cancer history and the number of comorbid
conditions. The NHIS annual sampling weights were adjusted to reflect pooling data across
multiple years and were used in all analyses to provide nationally representative estimates.
30
All adjusted estimates of changes in prescription drug use for financial reasons are presented
as predictive margins.
31
All analytic files were created using SAS version 9.3. Multivariable
logistic and ordered logistic regressions were performed using STATA 13.1. Statistical
comparisons were 2-sided, and significance was defined as
P
< .05.
RESULTS
Distribution of Individual-Level Characteristics and Unadjusted Changes in Prescription
Drug Use for Financial Reasons
Compared with individuals who had no history of cancer, cancer survivors were more likely
to be older, non-Hispanic white, married (nonelderly population only), and privately insured
and to have more comorbid conditions (Table 1). Nonelderly cancer survivors were more
likely to be women, whereas elderly cancer survivors were more likely to be men compared
Zheng et al.
Page 4
Cancer
. Author manuscript; available in PMC 2018 August 07.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
with individuals who had no history of cancer. Among privately insured non-elderly cancer
survivors, 41% of those who were recently diagnosed and 64% of those who were
previously diagnosed enrolled in high-deductible plans. We also observed that both recently
diagnosed and previously diagnosed nonelderly cancer survivors were more likely to report
changes in prescription drug use for financial reasons than individuals without a history of
cancer.
Adjusted Changes in Prescription Drug Use Between Cancer Survivors and Individuals
Without a Cancer History
In the adjusted analyses, we observed that nonelderly cancer survivors were more likely to
report changes in prescription drug use for financial reasons than nonelderly individuals
without a cancer history (see Fig. 1 for total percentage changes and Table 2 for excess
percentage changes). For the summary measure, 31.6% of recently diagnosed and 27.9% of
previously diagnosed cancer survivors reported any change in prescription drug use for
financial reasons, compared with 21.4% of individuals without a cancer history. The excess
changes in prescription drug use for financial reasons associated with cancer were 10.2%
(95% confidence interval [CI], 5.6%–14.8%) and 6.6% (95% CI, 4.4%–8.7%; all
P
< .05)
for recently and previously diagnosed cancer survivors, respectively.
For individual measures among the nonelderly population, we observed that, compared with
the control group: 1) 8.8% of previously diagnosed cancer survivors reported skipping
medication doses (excess change, 2.7%; 95% CI, 1.4%–3.9%); 2) 9.9% of recently
diagnosed and 9.3% of previously diagnosed cancer survivors reported taking less medicine
(excess changes, 3.5% [95% CI, 0.3%–6.6%] and 2.8% [95% CI, 1.6%–4.1%],
respectively); 3) 13.2% of recently diagnosed and 11.7% of previously diagnosed cancer
survivors reported delaying filing a prescription (excess changes, 5.2% [95% CI, 1.4%–
9.1%] and 3.7% [95% CI, 2.3%–5.1%], respectively); 4) 25.8% of recently diagnosed and
21.9% of previously diagnosed cancer survivors reported asking for lower cost medication
(excess changes, 9.9% [95% CI, 5.6%–14.2%] and 6% [95% CI, 4.1%–7.9%], respectively);
and 5) 7.4% of previously diagnosed cancer survivors reported using alternative therapies
(excess change, 2.6%; 95% CI, 1.3%–3.8%; all
P
< .05).
Among the elderly population, we observed that, compared with the control group: 1) 22.6%
of recently diagnosed cancer survivors reported asking for lower cost medication (excess
change, 5% [95% CI, 1%–9.1%]; 2) on average, approximately 4% of previously diagnosed
cancer survivors reported skipping medication doses, taking less medicine, and delaying
filling a prescription (excess change, <1%) compared with individuals without a cancer
history (all
P
< .05).
When the analyses were stratified by the number of comorbid conditions and compared with
the control group (Supporting Table 1; see online supporting information), we observed that
nonelderly cancer survivors were more likely to report changes in prescription drug use for
financial reasons when the number of comorbid conditions was ≤1 for recently diagnosed
patients and ≤2 for previously diagnosed patients (all
P
< .05). Elderly cancer survivors and
those without a cancer history had comparable rates of changes in prescription drug use for
financial reasons when stratified by the number of comorbid conditions.
Zheng et al.
Page 5
Cancer
. Author manuscript; available in PMC 2018 August 07.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript