? www.ajmc.com ??
? CLINICAL ?
(IV) infused treatment. Oncolytic medications that can be adminis-
tered orally are a relatively new addition to cancer treatment and
provide patients with the benefits of ease of use and convenience.1 Uti-
lization of oral oncolytics is expected to increase.2 Recent reports have
suggested they account for approximately 25% of the current oncology
Because of costs associated with these newer oral oncolytic agents,
pharmacy benefit plans may implement cost-containment mechanisms
such as increased patient cost sharing through placement in higher co-
payment tiers. The degree to which increased cost sharing influences
access to and utilization of oral oncolytics may be an important factor
affecting patient adherence to prescribed cancer therapy.
Traditional adherence studies of oncology medication track patient
utilization longitudinally by assessing refill rate, self-reported compli-
ance with prescribed therapies, continuous dose observations, pill count-
ing, and administrative claims analysis.1,4 Studies of nonadherence have
found that barriers to appropriate care include cost-sharing requirements,
agent toxicity, patient- and disease-related factors, social issues, and fi-
The purpose of our study was to assess abandonment of newly initiat-
ed oral oncolytics. Specifically, we were interested in examining patients
who had been prescribed oral oncolytic therapy, submitted prescriptions
for their first oral oncology medications to pharmacies, and then reversed
claims after adjudication (ie, initial approval of claim). If patients did
not follow up with a subsequent oncolytic agent, we noted their reversed
claim as an abandoned prescription.
Our objective was to calculate reversal and abandonment rates of
newly initiated oral oncolytic medications using an approach similar to
those found in the literature.5-7 Building on previous research, we inves-
tigated the degree to which abandonment is affected by patient and plan
characteristics, specifically cost-sharing requirements, patient income,
concurrent prescription activity, and insurance type.
atients diagnosed with cancer require timely access to appro-
priate treatments to achieve optimal outcomes. Until recently,
drug therapy for patients with cancer consisted of intravenous
We acquired administrative
claims data from the Wolters Klu-
Patient and Plan Characteristics Affecting Abandonment of
Oral Oncolytic Prescriptions
Sonya Blesser Streeter, MPP, MPH; Lee Schwartzberg, MD; Nadia Husain, ScM;
and Michael Johnsrud, PhD
In this issue
Take-Away Points / SP39
Full text and PDF
Objective: To calculate the abandonment rate of
oral oncolytic medications and identify factors
that may affect likelihood of abandonment.
Study Design: Cross-sectional cohort study using
administrative claims data.
Methods: We analyzed a nationally representa-
tive pharmacy claims database and identified
10,508 patients with Medicare and commercial
insurance for whom oral oncolytic therapy was
initiated between 2007 and 2009. We calculated
the abandonment rate for the initial claim, in
which abandonment was defined as reversal
of an adjudicated pharmacy claim without a
subsequent paid claim for any oncolytic (oral or
intravenous) within the ensuing 90 days. We as-
sessed likelihood of abandonment using bivari-
ate and multivariate logistic regression analyses
including patient demographics, plan type, drug
type, cost sharing, and concurrent prescription
Results: The abandonment rate of newly initiated
oral oncolytics was 10.0%. Unadjusted bivariate
analyses found that high cost sharing, increased
prescription activity, lower income, and Medicare
coverage were associated with a higher aban-
donment rate (P <.05). In the logistic regression
model, claims with cost sharing greater than
$500 were 4 times more likely to be abandoned
than claims with cost sharing of $100 or less
(odds ratio [OR], 4.46; P <.001). Patients with 5
or more prescription claims processed within in
the previous month had 50% higher likelihood
of abandonment than patients with no other
prescription activity (OR, 1.50; P <.001).
Conclusion: Abandonment of newly prescribed
oral oncolytic therapy is not uncommon, and the
likelihood increases for patients enrolled in plans
with pharmacy benefit designs that require high
cost sharing. Increased concurrent prescription
activity was also associated with a higher aban-
donment rate. These factors should be taken into
account when considering likely adherence to
(Am J Manag Care. 2011;17(5 Spec No.):SP38-SP44)
For author information and disclosures,
see end of text.
Copyright © 2011 by American Society of Clinical Oncology
and Managed Care & Healthcare Communications, LLC.
VOL. 17, Special Issue
? THE AMERICAN JOURNAL OF MANAGED CARE ??
Abandonment of Oral Oncolytic Prescriptions
wer Dynamic Claims Lifecycle Database, a
source of nationally representative pharmacy
utilization data, for 20,607 patients for whom
at least 1 pharmacy claim was adjudicated for
1 of the oral oncolytic agents included in our
study between January 1, 2007, and June 30,
2009. The data set included pharmacy claims
and demographic data for all patients as well
as medical claims for a subpopulation of pa-
tients. We studied patients using capecitabine,
imatinib, sorafenib, lenalidomide, sunitinib,
erlotinib, temozolomide, and lapatinib. These agents repre-
sented widely available oral oncolytic agents at the time of
our analyses. We then identified a subset of adjudicated oral
oncolytic claims between May 1, 2007, and March 31, 2009
(defined as the index period) to determine if a patient had
newly initiated therapy with a study drug. For each claim,
we looked back 120 days in the patient’s claim history from
first adjudication of an oral oncolytic claim during the index
period to exclude patients with previous oral or IV oncolytic
treatment. To confirm that the data set included prescription
data for each patient both before and after the newly initi-
ated oral oncolytic, we restricted the sample to those patients
who had at least 1 claim for any type of medication at least
120 or more days before and at least 90 days or more after the
first oral oncolytic claim, a methodology employed to assess
patient eligibility during the observation period.8,9
To calculate the portion of newly initiated oral oncolyt-
ics that were ultimately abandoned, we defined adjudication
status of the oral oncolytic as paid, reversed with follow-up,
or abandoned. Patients with claims defined as reversed with
follow-up had a successfully paid IV or oral oncolytic claim
within 90 days after submission date of the reversed newly
initiated oncolytic. Patients with claims defined as abandoned
had reversed the newly initiated oncolytic but did not have a
paid oncolytic claim within 90 days of submission date of the
We required the sample to include only patients insured
by a non-Medicare commercial or Medicare plan, including
prescription drug plans and Medicare Advantage Prescription
Drug plans. We also restricted the sample to claims with com-
plete patient- and claim-level data for all variables utilized in
the regression model.
To conduct bivariate and multivariate analyses, we iden-
tified patient-level characteristics such as age (<40, 41 to 65,
66 to 80, or >81 years), sex, and geographic region (North-
east, South, Midwest, or West). A variable for annual pa-
tient income was included in the data set and categorized as
less than $40,000, between $40,000 and $75,000, or more
than $75,000. Cost-sharing amounts were collected for each
paid and reversed claim and were grouped into the following
categories: $0 to $100, $101 to $150, $151 to $200, $201
to $250, $251 to $350, $351 to $500, or more than $500.
We also created a variable to measure concurrent prescrip-
tion activity for each patient based on the number of claims
processed within 30 days before the submission date of the
newly initiated oral oncolytic. The prescription activity
variable was defined as follows: no claims, 1 claim, 2 to 3
claims, 4 to 5 claims, or more than 5 claims. Variables were
created to control for the specific oral oncolytic agent that
the patient was prescribed.
?2 analyses were conducted to compare abandonment
rates across the independent variables previously described.
A logistic regression model was constructed using abandon-
ment status (0 = no, 1 = yes) within 90 days after the initially
adjudicated oral oncolytic claim as the dependent variable.
Covariates included all independent variables described to
identify significant predictor variables (P <.05). Odds ratios
and 95% confidence intervals [CIs] were calculated for all pre-
dictors in the model using selected references for comparison.
Of the initial 20,607 patients in the data set, we identified
10,508 patients who met the final inclusion criteria for our
study. Sixty-seven percent of the claims were adjudicated and
paid for by the patient and/or pharmacy benefit plan, whereas
33% of claims were reversed. Of the total number of reversed
claims, 23% of patients followed up with another oncolytic
agent within 90 days. The remaining 10% of patients had
no follow-up and thus abandoned the oral oncolytic agent.
A sensitivity test to extend the follow-up window from 90
to 120 days did not significantly affect the portion of claims
defined as abandoned.
Table 1 provides demographics and plan characteristics
for patients included in our study. Approximately half of the
patients in our sample were younger than 65 years. Patients
were slightly more likely to be female, more likely to have an
income between $40,000 and $75,000, and most commonly
Oral oncolytic therapy is an increasingly important aspect of cancer care, and ad-
herence to treatment is critical to deriving benefit.
?? Our study found that 10% of patients abandoned their anticancer medicine, and
another quarter had some delay in initiating another oncolytic.
?? Pharmacy plan design (cost-sharing amount) and complexity of patients’ drug
therapy (prescription activity) are significant drivers of abandonment of oral on-
?? As the structure of Medicare Part D and commercial plans is modified and health
reform initiatives evolve, policy makers and stakeholders should be aware of the
impact of benefit structure on adherence and access to vital oncology therapy.
? www.ajmc.com ??
? CLINICAL ?
?? Table 1. Adjudication Status of Newly Initiated Oral Oncolytic Claims
Total Paid or Reversed With Follow-Up Abandoned
Annual household incomea
Patient cost-sharing amounta
Prescription activity, No. of claimsa
No.% No. %No.%
10,508100.09455 90.01053 10.0
a ?2; P <.05.
VOL. 17, Special Issue
? THE AMERICAN JOURNAL OF MANAGED CARE ??
Abandonment of Oral Oncolytic Prescriptions
from the South. Most patients were insured by a commercial
Bivariate analyses comparing the combination of paid
claims and reversed claims with follow-up versus abandoned
claims indicated significant differences across age groups and
rate of abandonment. Thirteen percent of patients older than
80 years abandoned their first oral oncolytic compared with
10% of patients 40 years or younger (P <.05). Insurance sta-
tus was also a significant variable, with abandonment rates of
16% for Medicare claims versus 9% for commercial insurance
claims (P <.05).
Lower annual household income was associated with
higher abandonment rates. Patients with incomes of less than
$40,000 per year had an abandonment rate of 11%, decreasing
to 10% for incomes between $40,000 and $75,000 and 9% for
incomes above $75,000 (P <.05).
?? Table 2. Results of Logistic Regression of Likelihood of Abandonment of Newly Initiated Oral Oncolytic Claims
Independent VariableOdds Ratio 95% CIP
Age group, years (reference, 0-40)
41-650.82 0.55 to 1.23.346
66-800.71 0.47 to 1.07.102
>81 0.800.52 to 1.23 .313
Sex (reference, female)
Male0.99 0.86 to 1.14.899
Annual household income (reference, >$75,000)
<$40,0001.19 0.99 to 1.41.058
$40,000-$75,0001.13 0.96 to 1.32 .142
Geographic region (reference, Midwest)
Northeast1.15 0.95 to 1.38.157
South 0.910.76 to 1.08 .279
West 1.010.81 to 1.25 .937
Patient cost-sharing amount (reference, $0-$100)
$101-$1501.84 1.23 to 2.75.003
$151-$2001.51 0.97 to 2.34.066
$201-$250 2.30 1.31 to 4.04.004
$251-$350 2.311.59 to 3.36<.001
$351-$500 3.28 2.20 to 4.88<.001
>$5004.46 3.80 to 5.22 <.001
Prescription activity, No. of claims (reference, 0)
1 1.02 0.80 to 1.30.870
2-3 1.26 1.03 to 1.53.023
4-51.27 1.02 to 1.57 .029
>5 1.50 1.24 to 1.81<.001
Study drug (reference, capecitabine)
Imatinib2.09 1.68 to 2.60<.001
Sorafenib4.87 3.74 to 6.34<.001
Lenalidomide1.040.79 to 1.38 .759
Sunitinib1.63 1.21 to 2.21.001
Erlotinib1.471.20 to 1.81 <.001
Temozolomide 1.110.85 to 1.47.445
Lapatinib2.15 1.39 to 3.33.001
CI indicates confidence interval.
Psuedo R2 = 0.0922.
? www.ajmc.com ??
? CLINICAL ?
Twenty-three percent of patients submitted more than
5 claims for nononcolytic medications within the previous
month of initiating the oral oncolytic, whereas 29% of pa-
tients had no concurrent prescription activity. Prescription
activity was significantly associated with increased aban-
donment rates. Patients with more than 5 claims in the pre-
vious month had an abandonment rate of 12% compared
with 9% for patients with no claims in the previous month
Seventy-three percent of newly initiated oncolytics had a
cost-sharing amount of $100 or less, although 16% required
an out-of-pocket cost of more than $500. For reference, medi-
an cost-sharing amount in the sample for all patients was $30.
The abandonment rate increased with cost-sharing amount.
Claims with cost sharing above $500 had the highest aban-
donment rate (25%) compared with an abandonment rate of
6% for claims with cost sharing of $100 or less (P <.05).
Among the 8 oral oncolytics in our study, capecitabine ac-
counted for more than one-third of the sample. Imatinib, le-
nalidomide, erlotinib, and temozolomide were also commonly
used medications. The unadjusted abandonment rate for each
oral oncolytic agent varied greatly.
After controlling for multiple factors in the logistic regres-
sion model, we found that patients experiencing higher cost-
sharing amounts were significantly more likely to abandon the
oral oncolytic agent, compared with patients with the lowest
cost-sharing amount (Table 2). Claims with cost sharing over
$500 had more than 4 times the likelihood of abandonment
versus claims with cost sharing of $100 or less (P <.05). Pa-
tients with between 2 and 5 prescription claims and patients
with more than 5 claims in the previous month had a 26%
and 50% higher likelihood to abandon the oral oncolytic
agent, respectively, versus those patients with no concurrent
prescription activity (P <.05). A sensitivity analysis in which
we calculated the prescription activity variable with only paid
claims (not including reversals) produced similar regression
In the multivariate logistic regression, age and annual in-
come (which showed significant bivariate relationships with
abandonment) were no longer significant predictors once we
controlled for other factors. In addition, patients with in-
comes less than $40,000 were 20% more likely to abandon
versus patients with incomes greater than $75,000 (P = .058).
We completed a subanalysis to explore the distribution
of cost-sharing amounts by insurance type (Table 3). Eighty
percent of commercially insured patients paid $100 or less
out of pocket for the first oral oncolytic claim, compared
with 35% of Medicare patients. Only 11% of commercially
insured patients paid more than $500 versus 46% of Medi-
care patients. A ?2 test confirmed that Medicare patients pay
significantly more out of pocket than commercially insured
patients (P <.001).
To estimate the impact of insurance plan type on aban-
donment rates while controlling for other factors, we con-
structed an additional logistic regression model as part of this
subanalysis. This model included the same covariates as our
logistic regression model, with the addition of an insurance-
type variable (commercial vs Medicare). We also restricted
the sample to claims submitted in calendar year 2008 and
added a variable to designate calendar quarter of submission
to control for the impact of the Medicare coverage gap. The
results of this analysis confirmed relationships found in the
main regression analysis. Cost-sharing amount continued to
be a significant predictor of the likelihood of abandoning the
newly initiated oral oncolytic. Insurance type was not a sig-
nificant predictor of abandonment, perhaps because of the
strong relationship between cost sharing and insurance type.
We also found that abandonment rates were significantly
higher between April and December than in the first quarter
of the year (P <.05).
?? Table 3. Distribution of Cost-Sharing Amounts for Newly Initiated Oral Oncolytic Claims by Insurance Type
Commercial Insurance MedicareTotal
Cost-Sharing Requirement No.% No.% No.%
$0-$1007027 80.1 61135.27638 72.7
$101-$1502172.5 54 3.1 2712.6
$151-$200 162 1.996 5.5 2582.5
$201-$25090 1.0 331.9 1231.2
$251-$350192 2.299 5.72912.8
$351-$500 1461.754 3.1 200 1.9
>$500 93710.7 790 45.51727 1.4