Article

Effect of Prescription Drug Coupons on Statin Utilization and Expenditures: A Retrospective Cohort Study

Abstract and Figures

Importance: Drug coupons are widely used, but their effects are not well understood. Objective: To quantify the effect of coupons on statin use and expenditures. Design: Retrospective cohort analysis of IMS Health LRx LifeLink database. Setting: U.S. retail pharmacy transactions. Participants: Incident statin users who initiated branded atorvastatin or rosuvastatin between June 2006 and February 2013. Main outcomes and measures: Monthly statin utilization (pill-days of therapy), switching (filling a different statin), termination (failure to refill statin for 6 mo), and out-of-pocket and total costs. Results: Of 1.1 million incident atorvastatin and rosuvastatin users, 2% used a coupon for at least one statin fill. At 1 year, compared with noncoupon users, those who used a statin coupon on their first fill were dispensed an equal number of monthly pill-days (23.7 vs 23.8), were less likely to switch statins (14.4% vs 16.3%), and were less likely to have terminated statin therapy (31.3% vs 39.2%). At 4 years, coupon users were more likely to have switched (45.5% vs 40.8%) and less likely to have terminated statin therapy (50.6% vs 61.1%) compared with noncoupon users. Those who used greater numbers of coupons were substantially less likely to switch and terminate statin therapies. Monthly out-of-pocket costs were lower among coupon than noncoupon users at 1 year ($9.7 vs $15.1), but total monthly costs were qualitatively similar ($115.5 vs $116.9). At 4 years, monthly out-of-pocket costs among coupon users remained lower ($14.3 vs $16.6) compared with noncoupon users. Sensitivity analyses supported the main results. Conclusions: Coupons for branded statins are associated with higher utilization and lower rates of discontinuation and short-term switching to other statin products.
Content may be subject to copyright.
EFFECT OF PRESCRIPTION DRUG COUPONS ON STATIN
UTILIZATION AND EXPENDITURES: A RETROSPECTIVE
COHORT STUDY
Matthew Daubresse, MHS1,2, Martin Andersen, PhD3, Kevin R. Riggs, MD, MPH1,4,5, and G.
Caleb Alexander, MD, FACP1,2,4
1Center for Drug Safety and Effectiveness, Johns Hopkins University, Baltimore, Maryland
2Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore,
Maryland 3Department of Economics, University of North Carolina at Greensboro, Greensboro,
North Carolina 4Division of General Internal Medicine, Department of Medicine, Johns Hopkins
Medicine, Baltimore, Maryland 5Johns Hopkins Berman Institute of Bioethics, Baltimore, Maryland
Abstract
Importance—Drug coupons are widely used, but their effects are not well understood.
Objective—To quantify the effect of coupons on statin use and expenditures
Design—Retrospective cohort analysis of IMS Health LRx LifeLink database
Setting—U.S. retail pharmacy transactions
Participants—Incident statin users who initiated branded atorvastatin or rosuvastatin between
June 2006 and February 2013
Main Outcomes and Measures—Monthly statin utilization [pill-days of therapy], switching
[filling a different statin], termination [failure to refill statin for 6 months], and out-of-pocket and
total costs
Results—Of 1.1 million incident atorvastatin and rosuvastatin users, 2% used a coupon for at
least one statin fill. At one year, compared to non-coupon users, those who used a statin coupon on
their first fill were dispensed an equal number of monthly pill-days (23.7 vs. 23.8), were less likely
to switch statins (14.4 vs. 16.3%), and were less likely to have terminated statin therapy (31.3 vs.
39.2%). At 4 years, coupon users were more likely to have switched (45.5 vs. 40.8%) and less
likely to have terminated statin therapy (50.6 vs. 61.1%) compared to non-coupon users. Those
who used greater numbers of coupons were substantially less likely to switch and terminate statin
Correspondence: G. Caleb Alexander, MD, MS, Johns Hopkins Bloomberg School of Public Health, Department of Epidemiology,
615 N. Wolfe Street W6035, Baltimore, MD 21205, Phone: 410 955 8168; Fax: 410 955 0863; galexand@jhsph.edu.
Disclosures
Dr. Alexander is Chair of the FDA’s Peripheral and Central Nervous System Advisory Committee, serves as a paid consultant to IMS
Health and serves on an IMS Health scientific advisory board. This arrangement has been reviewed and approved by Johns Hopkins
University in accordance with its conflict of interest policies. The statements, findings, conclusions, views, and opinions contained and
expressed in this article are based in part on data obtained under license from the following IMS Health Incorporated information
service(s): IMS Health LifeLink (2006–2013). The statements, findings, conclusions, views, and opinions contained and expressed
herein are not necessarily those of IMS Health Incorporated or any of its affiliated or subsidiary entities.
HHS Public Access
Author manuscript
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
Published in final edited form as:
Pharmacotherapy
. 2017 January ; 37(1): 12–24. doi:10.1002/phar.1802.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
therapies. Monthly out-of-pocket costs were lower among coupon than non-coupon users at 1 year
($9.7 vs. $15.1), but total monthly costs were qualitatively similar ($115.5 vs. $116.9). At 4 years,
monthly out-of-pocket costs among coupon users remained lower ($14.3 vs. $16.6) compared to
non-coupon users. Sensitivity analyses supported the main results.
Conclusions—Coupons for branded statins are associated with higher utilization and lower rates
of discontinuation and short-term switching to other statin products.
INTRODUCTION
Although 86% of prescriptions filled in the United States in 2013 were generics, payers and
patients spent $232 billion on branded medications, accounting for 71% of all prescription
drug costs.1 Pharmaceutical companies employ a variety of promotional strategies to
encourage the use of these single-source branded medications. One such strategy is the use
of drug coupons, which are widely available at physicians’ offices and on the Internet, and
can be used to decrease patient copays for certain medications.2,3 Between 2009 to 2011, the
number of coupons offered in the United States increased 260% and approximately 11% to
13% of branded prescriptions were associated with a copay coupon.4
Debate surrounding the appropriateness and “moral hazard”5 of coupons mirrors that of
other promotional activities, such as direct-to-consumer advertising6,7 and the use of free
samples.8 Proponents argue that coupons lower patients’ out-of-pocket costs, reduce cost-
related nonadherence, and provide a safer alternative to drug samples by requiring
dispensing through licensed pharmacists.3,9 Opponents argue that coupons incentivize
patients to initiate and adhere to expensive branded therapies, increasing out-of-pocket and
third party spending that ultimately drives higher premiums for coupon-users and non-users
alike.3,9,10,11
Despite their increasing prevalence, there is remarkably little evidence regarding the effect
of coupons on prescription drug utilization or expenditures. One retrospective cohort study
used commercial pharmacy claims from incident statin patients to examine the impact of
coupons on brand-name statin utilization and spending. The authors found that coupon users
had higher rates of adherence and substantially higher total statin costs than those who
initiated generic statins.3,9 Despite these insights, the authors only examined prescription
fills, restricted their follow-up to a single observation at 12 months, and used a cross-
sectional study design that limited causal inference.
We conducted a retrospective cohort study comparing the effect of coupon use on utilization
and expenditures among incident statin-users. We focused on statins because the indications
for statin use consist of prevalent and costly chronic conditions, and because there are
multiple statins on the market, some of which have been heavily marketed and promoted
with drug coupons.
METHODS
We examined data from used the IMS Health LifeLink LRx Anonymized Longitudinal
Prescription database, consisting of prescriptions from retail, food store, independent, and
Daubresse et al. Page 2
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
mass merchandiser pharmacies, all of which represent approximately two-thirds of retail
prescriptions dispensed in the United States.12 These data include detailed information about
the quantity, form, and dose of medications dispensed as well as annual flags to indicate
individuals who utilized a mail-order pharmacy. The data are generated on a daily basis at
pharmacies and are then automatically transmitted to IMS Health through weekly feeds. The
prescription data are also linked to the above-noted database using a patented algorithm
based on 16 different fields such as the patient’s first name, last name, and date of birth.
Each prescription claim contains information about the retail transaction (days supply,
number of refills), patient (year of birth, sex), product (National Drug Code [NDC]), and the
payer and prescription drug plan. We used payer and plan variables to identify statin claims
associated with copay coupons.
Setting and Participants
We derived a closed cohort of incident statin users from a larger extract that contained all
prescriptions from January 2006 through August 2013 for any patient who filled two or
more prescriptions for an opioid in one of eleven states over any 1-year period during that
time. This extract, derived for a separate study, consisted of 5.3 billion retail transactions
from more than 50 million patients, 1.5 million prescribers and 52,000 pharmacies.
Participants were incident statin users, defined by evidence of no prior statin use for at least
a 6-month period with evidence of other prescription claims activity, who initiated branded
atorvastatin or rosuvastatin. Medicare and Medicaid prohibit coupon use, so we excluded
individuals over the age of 65 or who otherwise used Medicaid or Medicare. We categorized
patients into four mutually exclusive groups: (1) coupon used on first statin fill (initial
coupon users); (2) coupon used on statin fill, but not the first (subsequent coupon users); (3)
coupon used on non-statin fills (other coupon users); or (4) no coupons used (non-coupon
users). Our final dataset included approximately 700,000 coupons for atorvastatin (82%) or
rosuvastatin (18%), which together accounted for 87% of all statin coupons observed.
To account for potential differences between early and late coupon adopters, we excluded
individuals who used a coupon for either statin before coupons were widely available for
that drug, which we defined as the use of a coupon for at least 0.1% of all commercial
claims for a given drug during a particular month. For our primary analyses, we included
only individuals who used pharmacies that consistently reported data to IMS throughout the
study period and who filled at least one prescription for any drug within the first and last 6
months of the study period.
Measures
We examined three measures of utilization and two measures of cost. First, we calculated the
quantity of a prescription medicine sufficient for one day of therapy (pill-day) and then
examined the average monthly number of pill-days supplied. We allowed for unlimited
“stockpiling”13 and thus accounted for inherent differences in prescription quantity across
different pharmacy claims. Second, we calculated switching as the probability of switching
statins from one month to the next. Third, we calculated the proportion of people who
terminated therapy, defined as a 6-month period without any statin utilization. Fourth, we
Daubresse et al. Page 3
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
calculated the average monthly out-of-pocket costs for statins, reflecting the amount of co-
insurance the payer determined is owed by the patient for the transaction, based on an
individual’s benefit design. These values represent the patient’s out-of-pocket costs after any
discount from a coupon was applied. Finally, we quantified patients’ total costs for statins
(the copay plus the amount billed to insurance, i.e., what the pharmacy is paid from all
sources). We excluded claims with missing cost information from our analyses that
examined out-of-pocket and total costs.
In some cases, individuals had multiple statin transactions during a given month that made it
difficult to easily assign patients to a particular statin therapy. In our main analyses, we
dropped individuals who could not be assigned to a single statin therapy based on either the
plurality of their claims in a given month or by comparison of their current month’s use to
their use during the prior or subsequent month.
Statistical Analysis
We used generalized estimating equations (GEE) models, accounting for within-subject
correlations over time to calculate the predicted and marginal effects of coupon use between
coupon users in each group and their counterfactual non-coupon counterparts. We used the
number of months contributed by each person included in the cohort (i.e. person-months) as
our unit of analysis. We controlled for the age and sex of individual respondents and
included flexible specifications for time on drug, measured in months, and controlled for the
year and month that therapy was initiated. Our data did not include diagnoses, so we
controlled for differences in patient comorbidities using the Chronic Disease Score, a
method of quantifying comorbid burden using automated pharmacy claims that has been
validated as a measure of hospitalizations, expenditures, and mortality.14 In order to allow
the effect of time on drug to vary with initial coupon status, we included interactions
between a cubic spline with five knots in time on drug and coupon utilization of the
individual.15 To calculate the effect of coupons, we included dummy variables for coupon
utilization, which indicate whether a coupon was used that month.
We modeled the number of pills dispensed in a given month using a negative binomial
specification; out-of-pocket costs, pharmacy costs, and monthly pill utilization were
modeled using a Poisson distribution. In most analyses, we estimated GEE models with
exchangeable correlation matrices. However, in our constant store sample it was necessary to
assume independence between observations in order for the model to converge. We used
GEE logistic regression, which allows for within subject correlation over time, to examine
the effects of drug coupons on switching and termination. In order to interpret the results of
our models, we used “recycled predictions” to construct a synthetic comparison group for
each of our three coupon categories15,16 and computed the predicted value for each person-
month in each group. This approach provides a method of standardizing the average
predicted values over the distribution of other covariates. We then computed average
marginal differences as the average difference between the coupon group predicted values
and the comparison group predicted values. We computed standard errors for average
predicted values and marginal differences using the delta method from cluster-robust
Daubresse et al. Page 4
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
variance matrices.17 All models were estimated using Stata 13.1 (StataCorp, College Station,
TX).
Sensitivity Analyses
We conducted several sensitivity analyses. First, because we limited our primary analyses to
a closed cohort of individuals, we repeated our analyses, allowing subjects to enter and leave
the cohort. Second, since our data do not capture mail-order medications, we repeated our
analyses excluding individuals who filled any prescriptions by mail order during the analysis
period. Third, since our original extract oversampled opioid users, we repeated our analyses
after limiting them to individuals who had no opioid prescription fills from incident statin
use until termination or censoring. Fourth, we varied rules that allowed for unlimited
stockpiling, allowing patients to stockpile fewer pills from one prescription to a subsequent
prescription. Fifth, we varied our definition of statin termination to include those with no
statin fills for 3 months or 9 months.
RESULTS
Characteristics of coupon and non-coupon users
Our final sample consisted of approximately 1.1 million incident atorvastatin (66%) and
rosuvastatin (34%) users. Of these, 7,839 (.7%) used a coupon on their first statin fill, 12,864
(1.2%) used a coupon on a subsequent statin fill, 11,473 (1.1%) used a coupon for a non-
statin product, and approximately 1 million patients (97%) filled a prescription for an
incident statin without any associated coupon use (Table 1).
Overall, coupon and non-coupon users had similar demographic characteristics, prescription
drug utilization, and comorbid conditions. Coupon users were significantly more likely to
fill claims through an insurer (96% initial coupon users vs. 89% non-coupon users; p<.05),
and other coupon users had significantly higher median copays ($110 vs. $88; p<.05) and
total pharmacy costs ($2354 vs. $1698; p<.05) than non-coupon users.
Appendix Figure 1 depicts trends in branded and generic atorvastatin dispensing during the
study period among coupon users and non-users. In January 2007, there were approximately
223,000 branded prescription transactions without a coupon and a negligible number of
branded sales where a coupon was used. Branded sales remained relatively flat until May
2012, when generic atorvastatin was released. This was associated with a reduction of
approximately 95
%
in branded sales and 75
%
reduction in coupon use over the ensuing 15
months as the generic product took hold. A similar trend for rosuvastatin is shown in
Appendix Figure 2, although a generic formulation of the product was not introduced during
the study period.
Effect of coupons on statin utilization, switching and termination
Table 2 depicts differences in utilization, switching, and termination between coupon users
on atorvastatin or rosuvastatin and their counterfactual non-coupon counterparts within each
group (initial, subsequent, non-statin). Overall, coupon users had similar levels of statin
utilization and switching compared to their non-coupon users. At 1 year, initial coupon users
Daubresse et al. Page 5
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
were dispensed 0.1 fewer average pill days per month, were 1.9% less likely to switch
statins (16.6 vs. 18.9%; p<.05), and were 6.9% less likely to terminate statin therapy (31.3
vs. 39.2%; p<.001) than non-coupon users. Differences in termination amplified over time;
at 4 years, initial coupon users were 11.1% less likely to terminate statin therapy than non-
coupon users (50.6 vs. 60.7%; p<.0001) (Table 2).
Compared to initial coupon users, the association between coupon use and statin utilization
and switching was similar for subsequent and other coupon users. However, both subsequent
and other coupon users were less likely to terminate statin therapy than initial coupon users:
at 2 years, subsequent coupon users were 35.2% less likely to terminate statin therapy than
non-users (15.1 vs. 50.8%; p<.0001), whereas initial coupon users were 8.7% less likely to
terminate treatment (41.8 vs. 50.8%; p<.0001). These patterns continued to persist through 4
years of follow up.
The cumulative probability of statin termination over time among the four groups is shown
in Figure 3. Rates of discontinuation were greatest for non-coupon users. The length of time
until a discontinuation rate of 25% was 10 months for initial coupon users (95% CI: 9–10
months), 35 months for subsequent statin coupon users (95% CI: 33–37 months), 23 months
for other coupon users (95% CI: 23–24), and 7 months for non-coupon users (95% CI: 7–7).
Effect of varying levels of coupon utilization
Higher levels of coupon use resulted in higher utilization and a lower probability of
switching and termination (Table 3). For example, at 3 years, initial coupon users were 5.2%
more likely to switch and 0.6% more likely to terminate than non-coupon users. However, at
3 years, incident statin users who used coupons for five or more fills were 16% less likely to
switch and 28% less likely to terminate than non-coupon users.
Effect of coupon use on out-of-pocket and total costs
All coupon users had consistently lower out-of-pocket costs than non-coupon users (Table
4). At one year, average monthly out-of-pocket costs for statins appeared $5 lower for initial
coupon users than for non-coupon users; however, this difference was not statistically
significant ($9.7 vs. $15.9, NS). Between 2 and 4 years of follow-up, this difference
persisted, ranging from $2–$6.
The association between coupon use and total costs differed from those for out-of-pocket
costs. Overall, there were negligible differences in monthly average total costs between
coupon users and non-coupon users. At 1 month, total costs were approximately $7 higher
for initial coupon users compared to their non-coupon users ($143.3 vs. $136.5, p<0.001).
However, for longer periods of follow-up, this difference decreased and total costs for initial
coupon users were very similar to that of non-coupon users.
Sensitivity analyses
Repeating our analyses stratified by atorvastatin and rosuvastatin (Appendix Tables 1 and 2)
with an open cohort of statin patients, patients with no use of mail-order prescription
services and patients with limited opioid use (Appendix Table 3) did not substantively
Daubresse et al. Page 6
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
impact the results from our main analyses. Similarly, allowing for unlimited stockpiling,
varying the time period defining termination, and using an alternative method for
comorbidity adjustment had little impact on our main results.
DISCUSSION
In this longitudinal study of statin users among commercially insured incident statin users,
those who used a coupon on their first fill were dispensed a similar quantity of pills than
non-coupon users over 1 year, but were less likely to have switched statins or to have
terminated treatment altogether after 12 months of follow-up. There was a dose-response
association present, and these effects increased modestly over time. At one 1year, coupon
users had out-of-pocket costs that were approximately $1–$5/month lower than non-coupon
users but had similar total costs. These results are important because the use of drug coupons
is increasing, and how little is known about the effects of these coupons on patients’
utilization, out-of-pocket costs, and total costs.
Our study contributes to a growing evidence-base regarding the effect of drug coupons on
drug utilization and expenditures. One prior report used a hypothetical insurance program
and publicly available retail prices for statins to suggest that coupons may lead to lower out-
of-pocket costs among patients, but significantly higher costs for insurers due to a reduction
in the use of generic products.3 A second study using commercial pharmacy claims from
incident statin patients suggested that coupon users had more statin fills one year after statin
initiation and both higher out-of-pocket and total statin prescription costs compared to
generic statin initiators and non-coupon users of branded statins.9 In contrast to these
studies, we found that statin coupons were associated with similar levels of utilization and
total pharmacy costs. There are important differences between our approach and these prior
studies that may account for these differences, including our use of longitudinal GEE models
that account for within-subject correlations over time, as well as our analytic approach that
increased comparability across the groups of coupon users and non-users.18
Manufacturers’ use of drug coupons remains a controversial area of pharmaceutical policy.
In addition to historic concerns that are similar to those regarding direct-to-consumer
advertising 6 and the distribution of free medication samples8, there are particular provisions
in payment policy that preclude the use of coupons for services covered by nearly all federal
health care programs.19 Despite safeguards to prevent unauthorized use, a survey
commissioned by the National Coalition on Health Care found that 6% of Medicare
beneficiaries enrolled in Part D were using coupons19, an issue under recent study by the
Office of the Inspector General.20 The practice of providing drug coupons has also been
challenged by groups outside of the federal government. For example, in 2012, a group of
trade union health plans sued eight large drug manufacturers claiming that drug coupon
programs violate federal bribery laws.21 Massachusetts has prohibited drug coupons since
1988, which made it the only state with a complete ban on coupons. However, in 2013, the
Massachusetts legislature created an exception to the law that allowed the use of coupons for
branded drugs with no generic equivalent.22
Daubresse et al. Page 7
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
The rapid growth of specialty drug utilization in the United States, which totaled an
estimated $87 billion in 2012, are projected to reach $400 billion by 202023. This also lends
added urgency to the issue of drug coupons. In one analysis examining the use of coupons
for biologic anti-inflammatory or multiple sclerosis medications among the commercially
insured, coupons offset more than 60% of patients’ out-of-pocket costs. While this
substantially reduced patients’ cost-sharing, it also circumvented payer efforts to constrain
rising health care costs through the use of pharmacy benefits management.24 The use of
coupons in this setting may be increasingly common as payers attempt to manage specialty
costs through higher deductibles as well as changes to pharmacy benefit design, such as the
use of higher cost-sharing tiers in lieu of the standard three-tier design as well as step-
therapy or fail-first programs that steer physicians and patients towards lower cost therapies.
Our analyses had several limitations. First, we were unable to determine the dollar amount
of the coupon used and therefore, the savings to the consumer, after accounting for coupons.
Second, our analysis was limited to individuals filling prescriptions through retail
pharmacies, since our data did not include individual-level claims data for transactions filled
through mail-order services. Third, we assumed that the availability of a drug coupon only
affects individuals who choose to use such a coupon, even though it is possible that the
availability of a coupon affects the broader equilibrium prescription drug prices, formulary
assignment, and out-of-pocket costs. Fourth, our analyses were not designed to account for
additional market complexities, such as atorvastatin’s patent expiry and potential switching
from statin to non-statin lipid lowering therapies that may also have been relevant to the
primary associations of interest. Fifth, these data capture only prescriptions paid for and
given to an individual patient; therefore we were unable to account for prescriptions that
were filled but never picked up. Sixth, we derived our analytic cohort from a larger cohort of
opioid recipients, which may have diminished the generalizability of our findings. However,
restricting our analyses to patients with no opioid fills after their incident statin fill had no
substantive impact on our main results. Finally, our analyses do not allow determination of
whether drug coupons result in lower utilization of generic medications.
CONCLUSIONS
Despite their increasing use, relatively little is known regarding the effect of drug coupons
on consumer behavior. In the case of statins, we found that drug coupons are associated with
greater utilization and lower rates of statin discontinuation and short-term switching. It is
unclear whether coupons have a similar effect when applied in other therapeutic contexts,
and these associations may be of particular interest and importance in the coming decade as
manufacturers continue to design programs that buffer patients from high cost-sharing,
which simultaneously reduce patient’s potential cost-burden while preserving demand for
higher cost, branded products.
Supplementary Material
Refer to Web version on PubMed Central for supplementary material.
Daubresse et al. Page 8
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Acknowledgments
Support and Acknowledgments
Dr. Alexander is supported by the National Heart, Lung and Blood Institute (R01HL107345). Dr. Riggs is
supported by NIH Grant T32HL007180. The funding sources had no role in the design and conduct of the study,
analysis, or interpretation of the data and preparation or final approval of the manuscript prior to publication. The
authors gratefully acknowledge Christine Buttorff for comments on an earlier manuscript draft.
References
1. Aiken, M. [Accessed April 3, 2015] Use and shifting costs of healthcare: A review of the use of
medicines in the U.S. in 2013. IMS Health. Available at: http://www.imshealth.com/portal/site/
imshealth/menuitem.762a961826aad98f53c753c71ad8c22a/?
vgnextoid=2684d47626745410VgnVCM10000076192ca2RCRD
2. Starner CI, Alexander GC, Bowen K, Qiu Y, Wickersham PJ, Gleason PP. Specialty drug coupons
lower out-of-pocket costs and may improve adherence at the risk of increasing premiums. Health
Aff. 2014; 33:1761–9.
3. Grande D. The Cost of Drug Coupons. JAMA. 2012; 307:2375–2376. [PubMed: 22692167]
4. Visante. Pharmaceut Care Manage Assoc. Nov. 2011 How copay coupons could raise prescription
drug costs by $32 billion over the next decade [white paper].
5. Schultz, D. Drug coupons: a good deal for the patient, but not the insurer. Kaiser Health News; Oct.
2012 http://kaiserhealthnews.org/news/drug-coupons/
6. Kornfield R, Donohue J, Berndt ER, Alexander GC. Promotion of prescription drugs to consumers
and providers, 2001–2010. PLoS One. 2013; 8:e55504. [PubMed: 23469165]
7. Bhutada NS, Cook CL, Perri M 3rd. Consumers responses to coupons in direct-to-consumer
advertising of prescription drugs. Health Mark Q. 2009; 26:333–46. [PubMed: 19916098]
8. Alexander GC1, Zhang J, Basu A. Characteristics of patients receiving pharmaceutical samples and
association between sample receipt and out-of-pocket prescription costs. Med Care. 2008; 46:394–
402. DOI: 10.1097/MLR.0b013e3181618ee0 [PubMed: 18362819]
9. Daugherty JB, Maciejewski ML, Farley JF. The impact of manufacturer coupon use in the statin
market. J Manag Care Pharm. 2013; 19:765–72. [PubMed: 24156645]
10. Ross JS, Kesselheim AS. Prescription-drug coupons--no such thing as a free lunch. N Engl J Med.
2013; 369:1188–9. [PubMed: 23984672]
11. Coupons for Patients, but Higher Bills for Insurers. New York Times; Jan 1. 2011
12. IMS Health. Longitudinal LRX Insights. http://www.imsbrogancapabilities.com/en/market-
insights/lrx.html
13. Hess LM, Raebel MA, Conner DA, Malone DC. Measurement of Adherence in Pharmacy
Administrative Databases: A Proposal for Standard Definitions and Preferred Measures. Annals of
Pharmacotherapy. 2006; 40:1280–1288. [PubMed: 16868217]
14. Clark DO, Von Korff M, Saunders K, Baluch WM, Simon GE. A chronic disease score with
empirically derived weights. Med Care. 1995; 33:783–95. [PubMed: 7637401]
15. Korn, EL., Graubard, BI. Analysis of Health Surveys. New York, NY: Wiley; 1999. p. 126-140.
16. Li, Z., Mahendra, G. [Accessed July 28, 2015] Using “Recycled Predicitions” for Computing
Marginal Effects. Available at: http://www.lexjansen.com/wuss/2009/hor/HOR-Li.pdf
17. Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika.
1986; 73:13–22. DOI: 10.1093/biomet/73.1.13
18. Graubard B, Edward L, Korn E. Predictive Margins with Survey Data. Biometrics. 1999; 55:652–9.
[PubMed: 11318229]
19. [Accessed June 10, 2015] Drug companies fend off competition from generics by offering discount
coupons. Available at: http://www.washingtonpost.com/national/health-science/drug-companies-
fend-off-competition-from-generics-by-offering-discount-coupons/2012/10/01/c7a393be-
f05f-11e1-ba17-c7bb037a1d5b_story.html
Daubresse et al. Page 9
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
20. Levinson, DR, Inspector General. Manufacturer Safeguards May Not Prevent Copayment Coupon
Use for Part D Drugs. Department of Health and Human Services, Office of the Inspector General;
Sep. 2014 (OEI-05-12-00540)
21. [Accessed June 10, 2015] Consumer group sues 8 drugmakers over drug coupons. Available at:
http://usatoday30.usatoday.com/money/industries/health/drugs/story/2012-03-07/drug-coupons-
lawsuit/53400686/1
22. [Accessed June 10, 2015] The 189th General Court of the Commonwealth of Massachusetts.
Available at: https://malegislature.gov/Laws/GeneralLaws/PartI/TitleXXII/Chapter175H/Section3
23. CVS Caremark. [Accessed June 10, 2015] Specialty Trend Management: Where to Go Next,
Insights. 2013. Available at: http://www.cvshealth.com/sites/default/files/Insights%202013.pdf
24. Starner CI, Alexander GC, Bowen K, Qui Y, Wickersham P, Gleason PP. Specialty Drug Coupons
Lower Out-of-pocket Costs and May Improve Adherence at the Risk of Increasing Premiums.
Health Affairs. Published online October 6, 2014.
Daubresse et al. Page 10
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Figure 1.
Daubresse et al. Page 11
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Daubresse et al. Page 12
Table 1
Characteristics of coupon users and non-users prior to initiating atorvastatin or rosuvastatin therapy
*
Coupon Users Non-users
Incident statin coupon Subsequent statin coupon Incident non-statin coupon No coupon use
Female, % 55 54 54 57
Age, mean, median, (IQR) 49, 51 (11) 50, 52 (11) 48, 49 (12) 50, 51 (11)
Insurance status,%
Commercially insured 96 96 94 89
None/Cash 4 4 6 11
Number of unique prescribers
**
, mean, median (IQR) 4, 3 (3) 4, 3 (3) 5, 4 (4) 4, 3 (3)
Number of unique pharmacies
**
, mean, median (IQR) 2, 1 (1) 2, 1 (1) 2,1 (2) 2, 1 (1)
Drug utilization
Prescriptions
**
, mean, median (IQR) 22, 18 (18) 27, 22 (24) 30, 23 (27) 26, 21 (24)
Coupon use
Coupons used prior to first statin
**
, mean, median (IQR) 1, 0 (0) 0, 0 (0) 1, 0 (1) ---
Unique therapeutic classes
**
, mean, median (IQR) 8, 7 (6) 9, 8 (7) 10, 8 (8) 9, 8 (7)
Chronic disease score
**
, mean, median (IQR) 2, 2 (2) 2, 2 (2) 2, 2 (2) 2, 2 (2)
Costs
Total copay
**
, mean, median (IQR) 249, 86 (345) 227, 96 (285) 2134, 110 (354) 246, 88 (283)
Total cost
**
, mean, median (IQR) 2422, 1530 (2302) 2905, 1876 (2824) 3760, 2354 (3773) 2864, 1698 (2859)
Total Patients, N 7839 12864 11473 1018739
*
values represent column percents unless otherwise noted
**
based on claims filled 6 months prior to incident statin fill date
Source: IMS Health Lifelink LRx Data, 2007–2013
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Daubresse et al. Page 13
Table 2
Differences in average monthly atorvastatin and rosuvastatin utilization among coupon users and non-users.
*
Time following incident statin fill
Time
1 month 1 year 2 years 3 years 4 years
Statin Utilization (Average pill-days dispensed)
No coupons
Predicted level, pill-days (SE) 29.1 (0.0) 23.7 (0.0) 24.5 (0.0) 24.7 (0.0) 25.4 (−0.1)
Other coupon
Predicted level, pill-days (SE) 28.3 (−0.1) 23.7 (−1.2) 24.7 (−2.8) 24.9 (−4.4) 25.2 (−6.1)
Difference from counterfactual
**
−0.7 (−0.1) 0.1 (−1.2) 0.2 (−2.8) 0.2 (−4.4) −0.1 (−6.1)
Initial statin coupon
Predicted level, pill-days (SE) 28.4 (−0.2) 23.8 (−3.0) 24.3 (−6.7) 24.7 (−10.5) 24.7 (−14.3)
Difference from counterfactual −0.8 (−0.2) −0.1 (−3.0) −0.4 (−6.7) 0.0 (−10.5) −0.6 (−14.3)
Subsequent statin coupon
Predicted level, pill-days (SE) 26.7 (−0.1) 24.0 (−2.1) 24.5 (−4.6) 24.8 (−7.2) 25.1 (−9.9)
Difference from counterfactual −2.2 (−0.1) 0.3 (−2.1) −0.2 (−4.6) 0.0 (−7.2) −0.4 (−9.9)
Statin Switching (Cumulative probability of switching from one statin to another)
No coupons
Predicted level, percent switching (SE) 1.5 (0.0) 16.6 (−0.1) 27.4 (−0.2) 35.6 (−0.3) 42.8 (−0.4)
Other coupon
Predicted level, percent switching (SE) 1.9 (−0.1) 18.9 (−0.4) 31.5 (−0.7) 42.3 (−1.0) 51.4 (−1.2)
Difference from counterfactual 0.3 (−0.1) 1.9 (−0.5) 3.6 (−0.7) 5.9 (−1.0) 7.7 (−1.3)
Initial statin coupon
Predicted level, percent switching (SE) 1.4 (−0.2) 14.4 (−0.9) 25.5 (−1.8) 36.0 (−3.5) 45.5 (−5.6)
Difference from counterfactual −0.1 (−0.2) −1.9 (−1.0) −1.1 (−1.8) 1.8 (−3.5) 4.7 (−5.6)
Subsequent statin coupon
Predicted level, percent switching (SE) 1.4 (−0.2) 15.6 (−0.8) 28.3 (−1.1) 39.3 (−1.4) 48.8 (−1.6)
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Daubresse et al. Page 14
Time following incident statin fill
Time
1 month 1 year 2 years 3 years 4 years
Difference from counterfactual −0.2 (−0.2) −1.4 (−0.8) 0.6 (−1.1) 3.4 (−1.4) 5.6 (−1.7)
Statin Termination (Cumulative Probability of failure to refill for period of at least 6 months)
No coupons
Predicted level, percent terminating (SE) 12.0 (−0.1) 39.2 (−0.1) 50.8 (−0.2) 56.3 (−0.2) 60.7 (−0.2)
Other coupon
Predicted level, percent terminating (SE) 2.3 (−0.1) 14.9 (−0.3) 24.1 (−0.4) 30.2 (−0.5) 35.9 (−0.6)
Difference from counterfactual −9.9 (−0.1) −25.1 (−0.4) −27.6 (−0.5) −27.0 (−0.5) −25.7 (−0.6)
Initial statin coupon
Predicted level, percent terminating (SE) 8.2 (−0.5) 31.3 (−1.0) 41.8 (−1.3) 46.3 (−1.6) 50.6 (−2.1)
Difference from counterfactual −3.5 (−0.5) −6.9 (−1.1) −8.7 (−1.3) −9.9 (−1.6) −10.5 (−2.1)
Subsequent statin coupon
Predicted level, percent terminating (SE) 0.4 (−0.1) 8.3 (−0.5) 15.1 (−0.6) 20.7 (−0.8) 25.9 (−1.0)
Difference from counterfactual −11.6 (−0.1) −30.6 (−0.5) −35.2 (−0.7) −34.9 (−0.8) −33.9 (−1.0)
*
Values represent 12-month averages at varying duration of follow-up;
**
Difference from counterfactual represents the difference in average predicted values between coupon users and their counterfactual non-coupon using counterparts within each group (initial, subsequent,
non-statin);
Source: IMS Health Lifelink LRx Data, 2007–2013
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Daubresse et al. Page 15
Table 3
Association Between Coupons and Utilization Among Incident Statin Coupon Users.
Time following incident statin fill
Time
1 month 1 year** 2 years** 3 years** 4 years**
Statin Utilization (Average pill-days dispensed)
No coupons
Predicted level, pill-days (SE) 26.5 (0.0) 23.7 (0.0) 24.5 (0.0) 24.7 (0.0) 25.4 (−0.1)
Coupon use on first fill only
Predicted level, pill-days (SE) 24.0 (−0.2) 23.0 (−4.0) 23.4 (−8.9) 23.6 (−13.8) 24.6 (−19.5)
Difference from counterfactual
*
−2.5 (−0.3) −0.9 (−4.0) −1.1 (−8.9) −1.0 (−13.8) −0.6 (−19.5)
Coupon use for two fills
Predicted level, pill-days (SE) 26.3 (−0.3) 24.5 (−6.6) 24.3 (−14.0) 24.1 (−21.4) 24.3 (−29.5)
Difference from counterfactual −0.6 (−0.3) 0.4 (−6.6) −0.3 (−14.0) −0.5 (−21.4) −0.8 (−29.5)
Coupon use for three or four fills
Predicted level, pill-days (SE) 29.1 (−0.2) 23.6 (−3.8) 23.6 (−8.1) 25.7 (−13.8) 30.5 (−22.9)
Difference from counterfactual 2.1 (−0.2) −0.6 (−3.9) −1.2 (−8.1) 0.8 (−13.8) 4.2 (−22.9)
Coupon use for five or more fills
Predicted level, pill-days (SE) 30.1 (−0.1) 25.7 (−3.3) 26.8 (−7.2) 27.1 (−11.3) 26.3 (−14.9)
Difference from counterfactual 3.1 (−0.1) 1.5 (−3.3) 1.8 (−7.2) 2.2 (−11.3) 0.9 (−14.9)
Statin Switching (Cumulative probability of switching from one statin to another)
No coupons
Predicted level, percent switching (SE) 3.0 (0.0) 16.7 (−0.1) 27.4 (−0.2) 35.8 (−0.3) 43.0 (−0.4)
Coupon use on first fill only
Predicted level, percent switching (SE) 5.2 (−0.6) 18.8 (−1.5) 30.6 (−2.7) 39.8 (−5.0) 48.3 (−7.5)
Difference from counterfactual 2.2 (−0.6) 2.5 (−1.5) 3.8 (−2.7) 5.2 (−5.0) 7.1 (−7.5)
Coupon use for two fills
Predicted level, percent switching (SE) 2.3 (−0.8) 11.7 (−2.5) 20.8 (−4.3) 35.5 (−6.2) 38.0 (−7.2)
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Daubresse et al. Page 16
Time following incident statin fill
Time
1 month 1 year** 2 years** 3 years** 4 years**
Difference from counterfactual −0.7 (−0.8) −4.5 (−2.5) −5.6 (−4.3) 1.9 (−6.2) −2.4 (−7.2)
Coupon use for three or four fills
Predicted level, percent switching (SE) 0.7 (−0.3) 14.2 (−2.2) 26.4 (−4.2) 40.4 (−7.1) 46.9 (−12.2)
Difference from counterfactual −2.3 (−0.3) −2.2 (−2.2) −0.6 (−4.3) 6.5 (−7.1) 8.8 (−12.1)
Coupon use for five or more fills
Predicted level, percent switching (SE) 0.0 (0.0) 4.7 (−1.3) 13.5 (−2.8) 18.2 (−4.2) 18.3 (−4.3)
Difference from counterfactual −2.9 (−0.1) −11.3 (−1.3) −13.1 (−2.8) −15.5 (−4.2) −22.2 (−4.3)
Statin Termination (Cumulative Probability of failure to refill for period of at least 6 months)
No coupons
Predicted level, percent terminating (SE) 19.1 (−0.1) 39.2 (−0.1) 50.9 (−0.2) 56.4 (−0.2) 60.9 (−0.2)
Coupon use on first fill only
Predicted level, percent terminating (SE) 19.9 (−1.1) 39.6 (−1.4) 51.9 (−1.7) 57.0 (−1.9) 62.2 (−2.2)
Difference from counterfactual 1.1 (−1.1) 1.0 (−1.5) 1.0 (−1.7) 0.6 (−1.9) 0.9 (−2.2)
Coupon use for two fills
Predicted level, percent switching (SE) 8.9 (−1.4) 33.9 (−3.1) 45.7 (−3.9) 52.9 (−5.0) 61.2 (−6.2)
Difference from counterfactual −9.5 (−1.4) −3.9 (−3.1) −4.8 (−3.9) −3.7 (−5.0) −0.2 (−6.2)
Coupon use for three or four fills
Predicted level, percent switching (SE) 2.9 (−0.6) 24.6 (−2.5) 35.4 (−3.4) 45.3 (−5.1) 54.0 (−14.7)
Difference from counterfactual −15.3 (−0.6) −13.1 (−2.5) −15.0 (−3.4) −10.6 (−5.1) −0.3 (−14.7)
Coupon use for five or more fills
Predicted level, percent switching (SE) 0.0 (0.0) 10.4 (−1.7) 20.4 (−2.8) 27.2 (−4.3) 30.2 (−5.9)
Difference from counterfactual −17.9 (−0.1) −26.8 (−1.7) −28.7 (−2.8) −27.9 (−4.3) −29.2 (−5.9)
*
Difference from counterfactual represents the difference in average predicted values between coupon users and their counterfactual non-coupon using counterparts within each group (initial, subsequent,
non-statin).
Source: IMS Health Lifelink LRx Data, 2007–2013
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Daubresse et al. Page 17
Table 4
Differences in average monthly atorvastatin and rosuvastatin expenditures between coupon and non-coupon users.
Time following incident statin fill
Time
1 month 1 year 2 years 3 years 4 years
Average Copayment, $
No coupons
Predicted level (SE) 19.4 (−0.1) 15.9 (−0.1) 17.0 (−0.1) 16.1 (−0.1) 15.2 (−0.2)
Other coupon
Predicted level (SE) 16.1 (−0.2) 13.0 (−3.0) 13.4 (−6.7) 12.6 (−9.7) 12.1 (−12.6)
Difference from counterfactual
*
−3.4 (−0.1) −2.9 (−3.0) −3.6 (−6.7) −3.4 (−9.7) −3.1 (−12.6)
Initial statin coupon
Predicted level (SE) 16.4 (−1.7) 9.7 (−10.2) 11.0 (−25.8) 14.3 (−51.6) 14.3 (−70.0)
Difference from counterfactual −2.9 (−1.7) −5.4 (−10.2) −6.1 (−25.8) −2.8 (−51.6) −2.3 (−70.0)
Subsequent statin coupon
Predicted level (SE) 12.9 (−0.3) 14.7 (−11.0) 15.6 (−25.3) 14.6 (−36.6) 12.2 (−41.6)
Difference from counterfactual −7.6 (−0.3) −1.3 (−11.0) −1.4 (−25.3) −1.6 (−36.6) −3.1 (−41.6)
Average Total Cost, $
No coupons
Predicted level (SE) 128.4 (−0.1) 104.6 (−0.2) 112.4 (−0.2) 115.9 (−0.2) 121.4 (−0.4)
Other coupon
Predicted level (SE) 127.3 (−0.3) 106.7 (−6.2) 114.9 (−14.5) 118.2 (−23.0) 121.6 (−32.0)
Difference from counterfactual −2.5 (−0.3) 1.3 (−6.2) 1.7 (−14.5) 1.9 (−23.0) 0.0 (−32.0)
Initial statin coupon
Predicted level (SE) 143.3 (−1.3) 115.5 (−14.9) 117.3 (−32.7) 121.8 (−52.5) 123.7 (−72.3)
Difference from counterfactual 6.8 (−1.2) 1.4 (−14.9) −2.6 (−32.7) 0.5 (−52.5) −1.5 (−72.3)
Subsequent statin coupon
Predicted level (SE) 118.1 (−0.4) 109.6 (−10.1) 115.6 (−23.3) 121.1 (−37.6) 125.1 (−52.5)
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Daubresse et al. Page 18
Time following incident statin fill
Time
1 month 1 year 2 years 3 years 4 years
Difference from counterfactual −9.9 (−0.3) 3.6 (−10.1) 1.6 (−23.3) 3.7 (−37.6) 2.6 (−52.5)
*
Difference from counterfactual represents the difference in average predicted values between coupon users and their counterfactual non-coupon using counterparts within each group (initial, subsequent,
non-statin)
Source: IMS Health Lifelink LRx Data, 2007–2013
Pharmacotherapy
. Author manuscript; available in PMC 2018 January 01.
... Sixty-three full-text articles were then assessed and eight met all inclusion criteria (Figure 1). Of the eight included studies, [19][20][21][22][23][24][25][26] seven took place in the United States and one took place in Australia, in a variety of settings ranging from single-site medical centers to nationwide samples (Table 1). Seven studies examined FMA for medications treating cardiovascular disease states, while one study assessed FMA for cancer medications. ...
... The percentage of patients who discontinued medication was 7 percentage points lower in users of FMA versus nonusers for cardiovascular disease states. 21,23 Persistence was also reported in 3 of 8 studies. FMA increased odds of (5 of 8) were retrospective observational studies, while one was a prospective cohort study, one was a cluster randomized clinical trial, and one was a post-hoc analysis of the same trial. ...
... Impacts on patient out-of-pocket costs were mixed-two studies reported that out-of-pocket costs were higher for users of a coupon (in those with private insurance) 22 or a voucher (in those with private or public insurance) 25 versus nonusers, while one study (in those with private, public, or other insurance) reported the opposite, 26 and the fourth study reported null effects (in those with either private insurance or no insurance). 23 Among the two studies that reported that out-of-pocket costs were higher among users of FMA versus nonusers, impacts ranged from $25 per year medication persistence for one year (with no more than a gap in medication of 29 days) by a range of 11% to 47%. 24,25 One cancer study reported on the initiation and discontinuation phases of adherence, as well as persistence. ...
Article
BACKGROUND: The prevalence of financial medication assistance (FMA), including patient assistance programs, coupons/copayment cards, vouchers, discount cards, and programs/pharmacy services that help patients apply for such programs, has increased. The impact of FMA on medication adherence and persistence has not been synthesized. OBJECTIVE: The primary objective of this study was to review published studies evaluating the impact of FMA on the three phases of medication adherence (initiation [or primary adherence], implementation [or secondary adherence], and discontinuation) and persistence. Among these studies, the secondary objective was to report the impact of FMA on patient out-of-pocket costs and clinical outcomes. METHODS: A systematic review was performed using MEDLINE and Web of Science. RESULTS: Of 656 articles identified, eight studies met all inclusion criteria. Seven studies examined FMA for medications treating cardiovascular diseases, while one study assessed FMA for cancer medications. Among included studies, FMA had a positive impact on medication adherence or persistence, and most measured this impact over one year or less. Of the three phases of medication adherence, implementation (5 of 8) was most commonly reported, followed by discontinuation (3 of 8), and then initiation (1 of 8). Regarding implementation, users of FMA had a higher mean medication possession ratio (MPR) than nonusers, ranging from 7 to 18 percentage points higher. The percentage of patients who discontinued medication was 7 percentage points lower in users of FMA versus nonusers for cardiovascular disease states. In one cancer study, FMA had a larger impact on initiation than discontinuation, ie, compared to nonusers, users of FMA were less likely to abandon an initial prescription (risk ratio= 0.12, 95% confidence interval [CI]: 0.08-0.18), and this effect was larger than the decreased likelihood of discontinuing the medication (hazard ratio = 0.76, 95% CI: 0.66-0.88). In 3 of 8 studies reporting on medication persistence, FMA increased the odds of medication persistence for one year ranged from 11% to 47%, depending on the study. In addition to adherence, half of the studies reported on FMA impacts on patient out-of-pocket costs and 3 of 8 studies reported on clinical outcomes. Impacts on patient out-of-pocket costs were mixed; two studies reported that out-of-pocket costs were higher for users of a coupon or a voucher versus nonusers, one study reported the opposite, and one study reported null effects. Impacts on clinical outcomes were either positive or null. CONCLUSIONS: We found that FMA has positive impacts on all phases of medication adherence as well as medication persistence over one year. Future studies should assess whether FMA has differential impacts based on phase of medication adherence and report on its longer-term (ie, beyond one year) impacts on medication adherence. DISCLOSURES: This work was sponsored by a grant from Pharmaceutical Research and Manufacturers of America (PhRMA). PhRMA had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Hung reports past employment by Blue Cross Blue Shield Association and CVS Health and a grant from PhRMA outside of the submitted work. Zullig reports research funding from Proteus Digital Health and the PhRMA Foundation. consulting fees from Novartis. Reed reports receiving research support from Abbott Vascular, AstraZeneca, Janssen Research & Development, Monteris, PhRMA Foundation, and TESARO and consulting fees from Sanofi/Regeneron, NovoNordisk, SVC Systems, and Minomic International, Inc. Bosworth reports research grants from the PhRMA Foundation, Proteus Digital Health, Otsuka, Novo Nordisk, Sanofi, Improved Patient Outcomes, Boehinger Ingelheim, NIH, and VA, as well as consulting fees from Sanofi, Novartis, Otsuka, Abbott, Xcenda, Preventric Diagnostics, and the Medicines Company. The other authors have nothing to report. This work was presented as a poster presentation at the ESPACOMP Annual Meeting in November 2020.
... Despite the wide availability of coupons, little is known about why manufacturers choose to offer coupons for specific drugs and not for others. For example, manufacturers may be motivated to reduce patient out-of-pocket spending to improve patient adherence 6 ; alternatively, they may choose to offer coupons to improve their market share within a drug class. Despite previous research on coupon use, little is known about which factors are associated with coupon availability for brandname drugs. ...
... [7][8][9] Others assess individual-level copay reduction and welfare gain associated with coupon use for a specific subset of drugs. 6,[10][11][12][13] In addition, prior work has tended to focus on off-patent drugs with generic alternatives, yet many brand-name drugs face competition from other brand-name drugs with the same mechanism of action or second-or-later-in-class drugs. [14][15][16][17] Because manufacturers' motivation to use coupons may vary by the level of competition each drug faces, it is important to understand the market characteristics when manufacturers choose to offer coupons and when they do not. ...
Article
Full-text available
Importance Drug companies offer coupons to lower the out-of-pocket costs for prescription drugs, yet little is known about why they do so for some drugs but not for others. Objective To examine whether the following factors are associated with manufacturer drug coupon use: (1) patient-cost characteristics (mean per-patient cost per drug, mean patient copay); (2) drug characteristics (generics availability or “later-in-class-entrant” drugs); (3) drug-class characteristics (in-class coupon use among competitors; in-class generic competition; in-class mean cost and copay). Design, Setting, and Participants This was a retrospective cohort analysis of anonymized transactional pharmacy claims sourced from retail US pharmacies from October 2017 to September 2019, supplemented with information derived from Medi-Span, Red Book, and FDA.gov. Data were analyzed from September 2020 to February 2021. Main Outcomes and Measures The primary outcome was availability of a manufacturer’s coupon. The secondary outcome was the mean proportion of transactions in which a coupon was used for each product. Results The sample of 2501 unique brand-name prescription drugs accounted for a total of 8 995 141 claims. Manufacturers offered a coupon for 1267 (50.7%) of these drugs. When the manufacturer offered a coupon, it was used in a mean (SD) 16.3% (20.3%) of the transactions. Within a drug class, higher mean total cost per patient was positively associated with the likelihood of coupon use (odds ratio [OR], 1.03 per 10% increase; 95% CI, 1.01-1.04), but higher mean patient copay was inversely associated (OR, 0.98; 95% CI, 0.97-0.99). For drug characteristics, single-source later-in-class-entrant products were associated with a greater likelihood of coupon use compared with first entrants and multisource brands (OR, 1.44; 95% CI, 1.09-1.89). The intensity of coupon use was associated with later-in-class-entrant products and the class mean per-patient cost (4.16-percentage-point increase; 95% CI, 1.20-7.13; 0.27 per 10% increase; 95% CI, 0.09-0.44). Drugs with a new in-class brand-name competitor had greater mean coupon use compared with drugs without a new competitor (10.2% of claims with a coupon vs 5.9%). Conclusions and Relevance In this cohort study of transactional pharmacy claims, higher mean per-patient total cost within a class was significantly associated with the likelihood of coupon use, but not patient out-of-pocket cost. Manufacturers’ coupons were more likely to be used for expensive later-in-class-entrant products facing within-class competition where coupon use was prevalent.
... There have been a few US studies comparing branded products with these discount programs and their generic competition, which found mixed impacts on adherence and cost. [9][10][11] However, these studies were mostly descriptive and unadjusted, and the economic impact of discount cards remains unclear. Therefore, we used comprehensive pharmacy adjudica tion records to study the use of discount cards and their impact on costs for different payers in Canada. ...
Article
Background: Brand discount cards have become a popular way for patients to reduce out-of-pocket spending on drugs; however, controversy exists over their potential to increase insurers' costs. We estimated the impact of brand discount cards on Canadian drug expenditures. Methods: Using national claims-level pharmacy adjudication data, we performed a retrospective comparison of prescriptions filled using a brand discount card matched to equivalent generic prescriptions between September 2014 and September 2017. We investigated the impact on expenditures for 3 groups of prescriptions: those paid only through private insurance, those paid only through public insurance and those paid only out of pocket. Results: We studied 2.82 million prescriptions for 89 different medications for which brand discount cards were used. Use of discount cards resulted in 46% higher private insurance expenditures than comparable generic prescriptions (+$23.09 per prescription, 95% confidence interval [CI] $22.97 to $23.21). Public insurance expenditures were only slightly higher with cards: an increase of 1.3% or $0.37 per prescription (95% CI $0.33 to $0.41). Finally, out-of-pocket transactions using a card resulted in mean patient savings of 7% or $3.49 per prescription (95% CI -$3.55 to -$3.43). The impact varied widely among medicines across all 3 analyses. Interpretation: The use of brand discount cards increased costs to private insurers, had little impact on public insurers and resulted in mixed impacts for patients. These effects likely resulted from private insurers reimbursing brand drug prices even when generics were available and from discount cards being adjudicated after claims were sent to other insurers in most cases. Patients and their clinicians should recognize that discount cards have mixed impacts on out-of-pocket costs.
Article
BACKGROUND: US health plans are adopting benefit designs that shift greater financial burden to patients through higher deductibles, additional copay tiers, and coinsurance. Prior systematic reviews found that higher cost was associated with reductions in both appropriate and inappropriate medications. However, these reviews were conducted prior to contemporary benefit design and medication utilization. OBJECTIVE: To assess the relationship and factors associated with cost-sharing and (1) medication adherence, (2) clinical outcomes, (3) health care resource utilization (HRU), and (4) costs. METHODS: A systematic review of literature published between January 2010 and August 2020 was conducted to identify the relationship between cost-sharing and medication adherence, clinical outcomes, HRU, and health care costs. Data were extracted using a standardized template and were synthesized by key questions of interest. RESULTS: From 1,995 records screened, 79 articles were included. Most studies, 71 of 79 (90%), reported the relationship between cost-sharing and treatment adherence, persistence and/or discontinuation; 16 (20%) reported data on cost-sharing and HRU or medication initiation, 11 (14%) on costsharing and health care costs, and 6 (8%) on cost-sharing and clinical outcomes. The majority of publications found that, regardless of disease area, increased cost-sharing was associated with worse adherence, persistence, or discontinuation. The aggregate data suggested the greater the magnitude of cost-sharing, the worse the adherence. Among studies examining clinical outcomes, cost-sharing was associated with worse outcomes in 1 study and the remaining 3 found no significant differences. Regarding HRU, higher-cost-sharing trended toward decreased outpatient and increased inpatient utilization. The available evidence suggested higher cost-sharing has an overall neutral to negative impact on total costs. Studies evaluating elimination of copays found either decreased or no impact in total costs. CONCLUSIONS: The published literature shows consistent impacts of higher cost sharing on initiation and continuation of medications, and the greater the cost-sharing, the worse the medication adherence. The evidence is limited regarding the impact of cost-sharing on clinical outcomes, HRU, and costs. Limited evidence suggests increased cost-sharing is associated with more inpatient care and less outpatient care; however, a neutral to no difference was suggested for other outcomes. Although increased costsharing is intended to decrease total costs, studies evaluating reducing or eliminating cost-sharing found that total costs did not rise. Today's growing cost-containment environment should carefully consider the broader impact cost-sharing has on treatment adherence, clinical outcomes, resource use, and total costs. It may be that cost-sharing is a blunt, rather than precise, tool to curb health care costs, affecting both necessary and unnecessary health care use. DISCLOSURES: This study and the development of this article were funded by the National Pharmaceutical Council. Dr Graff and Mr Sils are former employees of the National Pharmaceutical Council. Drs Fusco and Kistler and Ms Ruiz are employees of Xcenda. Xcenda received funding to conduct the literature review.
Article
Objective. Families affected by asthma report difficulty adhering to care regimens because of high medication costs, coupled with increased cost sharing required by some insurance plans. To inform efforts to support adherence, we conducted a qualitative study to explore how families manage asthma care costs. Methods. We conducted phone interviews with commercially-insured, US adults (n = 59) who had asthma and/or a child with asthma. Our purposive sample included participants with high- and low/no-deductible health plans. We analyzed data using thematic content analysis to identify strategies for managing asthma care costs and to assess strategies’ implications for adherence. Results. Our analysis identified four overarching strategies for managing asthma care costs. First, participants used prevention strategies to avoid costly acute care by minimizing exposure to asthma triggers and adhering strictly to preventive medication regimens. Second, participants used shopping strategies to reduce costs, including by comparing medication prices across pharmacies, using medication coupons or free samples, and switching to lower-cost medications. Third, budgeting strategies involved putting aside funds, including in tax-exempt health savings accounts, or taking on debt to pay for care. Finally, some participants sought to reduce costs by forgoing recommended care, including by skipping medication doses or replacing prescribed medications with alternative therapies. Conclusion. Commercially-insured families use a wide range of strategies to manage asthma care costs, with both positive and negative implications for adherence. Our typology of asthma cost management strategies can inform insurance redesign and other interventions to help families safely reduce costs and maximize adherence to recommended care.
Article
Drug manufacturers sometimes offer co-payment coupons to offset patient out-of-pocket costs. Although coupons can help patients afford necessary medications, they increase overall drug spending by encouraging use of expensive brand-name drugs over less expensive generics.¹,2 Coupons are prohibited by Medicare and Medicaid, but they are available for commercially insured patients. Several states are considering restricting coupon use to promote generic substitution and control drug spending. In October 2017, California passed a law that banned use of co-payment coupons for brand-name drugs once interchangeable generic versions of those products have become available.³ We investigated how generic substitution changed in California after its law took effect in January 2018.
Article
Objective To estimate the prevalence, risk factors, and consequences of cost-related medication nonadherence (CRN) in individuals with chronic liver diseases (CLDs) in the United States. Patients and Methods Using the National Health Interview Survey from January 1, 2014, to December 31, 2018, we identified individuals with CLDs. Using complex weighted survey analysis, we obtained national estimates and risk factors for CRN and its association with cost-reducing behaviors and measures of financial toxicity. We evaluated the association of CRN with unplanned health care use, adjusting for age, sex, race/ethnicity, insurance, income, education, and comorbid conditions. Results Of 3237 respondents (representing 4.6 million) US adults with CLDs, 813 (representing 1.2 million adults, or 25%; 95% CI, 23% to 27%) reported CRN, of whom 68% (n=554/813) reported maladaptive cost-reducing behaviors. Younger age, female sex, low income, and multimorbidity were associated with a higher prevalence of CRN. Compared with patients without CRN, patients experiencing CRN had 5.1 times higher odds of financial hardship from medical bills (adjusted odds ratio [aOR], 5.05; 95% CI, 3.73 to 6.83) and 2.9 times higher odds of food insecurity (aOR, 2.85; 95% CI, 2.02 to 4.01). The CRN was also associated with 1.5 times higher odds of emergency department visits (aOR, 1.46; 95% CI, 1.11 to 1.94). Conclusion We observed a high prevalence of CRN and associated consequences such as high financial distress, financial hardship from medical bills, food insecurity, engagement in maladaptive cost-reducing strategies, increased health care use, and work absenteeism among patients with CLD. These financial determinants of health have important implications in the context of value-based care.
Article
BACKGROUND: There is concern that increasingly common use of patient assistance programs (PAPs), out-of-pocket assistance provided by manufacturers or foundations, distorts our understanding of patient behavior and insurance design incentives. Yet the current literature on prescription drug cost sharing largely overlooks their use. PAPs prevalence and impact on drug demand and price elasticity is a major knowledge gap. OBJECTIVE: To examine the use of PAPs among patients with multiple sclerosis (MS) and the association with drug demand in a specialty pharmacy program within a regional integrated health system that facilitates their use. METHODS: We used pharmaceutical claims data from December 2017 to December 2018 linked to detailed payer information from Kaiser Permanente Washington to characterize the prevalence of PAPs for users of 7 MS specialty drug molecules. We estimated price elasticity of demand (PED) in a two-part model by using the presence of copayment assistance as a source of cost variation. The first part estimated marginal probability of a claim in a given month with a probit model, comparing PAP users and nonusers, whereas the second part estimated days supplied of a medication, given a claim was made as a measure for demand. RESULTS: Of 789 unique patients, 480 (60.7%) used PAPs in at least 1 drug claim during the 13-month time frame, and 248 patients (31.4%) used PAPs for all of their MS drug claims. When used, copay assistance covered 100% of out-of-pocket (OOP) charges for 98% of claims and reduced patient annual OOP cost by $3,493 on average. People who used PAPs had much higher OOP charges, a lower Charlson comorbidity score, and were more likely to have insurance through an exchange. The OOP costs charged to patients was higher for claims where patient assistance was used than claims where assistance was not used ($294 vs $42, P < 0.001). Total claim amount was higher for claims that used assistance ($6,169) than claims that did not ($5,503, P < 0.001). The probability of a patient having a drug claim in a given month was 1.9% higher among those using patient assistance, although this finding was not significant (P = 0.258). An average change in price of -$168.21 with PAP use led to an average change in demand of -0.05 days, for an overall price elasticity of demand (SD = 0.028, P = 0.852) given PAP use of 0.005, indicating that the presence of PAPs did not significantly affect demand. PED estimates were not statistically significant by drug, and the exclusion of Medicare patients did not change this interpretation. CONCLUSIONS: In a mid-size integrated health system in the state of Washington, a program that promotes adherence to specialty drugs via facilitated PAP use was found to reduce patient OOP costs but had no effect on prescription drug utilization. Payers may consider embracing PAPs to remove patient financial barriers to necessary medications and use tools other than cost sharing to influence patient consumption of specialty drugs. DISCLOSURES: This manuscript was funded in part through a Pre-Doctoral Fellowship in Health Outcomes from the PhRMA Foundation awarded to Brouwer for the completion of her dissertation work. Yeung receives some salary support from Kaiser Permanente. The other authors have nothing to disclose.
Article
Objective While several prescription drug-based risk indices have been developed, their design, performance, and application has not previously been synthesized. Study design and setting We searched Ovid MEDLINE, CINAHL and Embase from inception through March 3, 2020 and included studies that developed or updated a prescription drug-based risk index. Two reviewers independently performed screening and extracted information on data source, study population, cohort sizes, outcomes, study methodology and performance. Predictive performance was evaluated using C statistics for binary outcomes and R² for continuous outcomes. The PROSPERO ID for this review is CRD42020165498. Results Of 19,112 articles that were retrieved, 124 were full-text screened and 25 were included, each of which represented a de novo or updated drug-based index. The indices were customized to varied age groups and clinical populations and most commonly evaluated outcomes including mortality (36%), hospitalization (24%) and healthcare costs (24%). C statistics ranged from 0.62 to 0.92 for mortality and 0.59 to 0.72 for hospitalization, while adjusted R² for healthcare costs ranged from 0.06 to 0.62. Seven of the 25 risk indices included used global drug classification algorithms. Conclusions More than two-dozen prescription drug-based risk indices have been developed and they differ significantly in design, performance and application.
Article
Importance Despite ongoing debate regarding the high prices that patients pay for prescription drugs, to our knowledge, little is known regarding the use of coupons, vouchers, and other types of copayment “offsets” that reduce patients’ out-of-pocket drug spending. Although offsets reduce patients’ immediate cost burden, they may encourage the use of higher-cost products and diminish health insurers’ ability to optimize pharmaceutical value. Objective To examine the drugs most commonly covered by offsets, the percentage of out-of-pocket costs covered by offsets, and the characteristics of patients using offsets for retail pharmacy transactions in the United States in 2017 through 2019. Design, Setting, and Participants A retrospective cohort analysis was conducted of a 5% nationally random sample of anonymized pharmacy claims from IQVIA’s Formulary Impact Analyzer, which captures more than 60% of all US pharmacy transactions. This analysis focused on 631 249 individuals who used at least 1 offset between October 1, 2017, and September 30, 2019. Main Outcomes and Measures Offset source, types of drugs covered by offsets, offset dollar value and percentage of out-of-pocket payment covered, and county characteristics of offset recipients. Results The 631 249 individuals in the study (361 855 female participants [57.3%]; mean [SD] age, 45.7 [18.6] years) had approximately 33 million prescription fills, of which 12.8% had an offset used. Of these, 50.2% originated from a pharmaceutical manufacturer, 47.2% originated from a pharmacy or pharmacy benefit manager (PBM), and 2.6% originated from a state assistance program. A total of 80.0% of manufacturer-sponsored offsets were concentrated among 6.2% of unique products, and 79.9% of pharmacy-PBM offsets were concentrated among 4.9% of unique products. Most manufacturer offsets (88.2%) were for branded products, while most pharmacy-PBM offsets were for generic products (90.5%). The median manufacturer offset was $51.00, covering 87.1% of out-of-pocket costs; the median pharmacy-PBM offset was $16.30, covering 39.3% of out-of-pocket costs. There was no meaningful association between offset magnitude and county-level income, health insurance coverage, or race/ethnicity. Conclusions and Relevance In this analysis of patient-level pharmacy claims from 2017 to 2019, approximately half of all offsets involved pharmacy-PBM contractual arrangements, and half were offered by manufacturers. All offsets were associated with a significant reduction in patients’ out-of-pocket costs, were highly concentrated among a few drugs, and were generally not more generous among individuals in counties with lower income or larger Black or uninsured populations.
Article
Full-text available
Pharmaceutical firms heavily promote their products and may have changed marketing strategies in response to reductions in new product approvals, restrictions on some forms of promotion, and the expanding role of biologic therapies. We used descriptive analyses of annual cross-sectional data from 2001 through 2010 to examine direct-to-consumer advertising (DTCA) (Kantar Media) and provider-targeted promotion (IMS Health and SDI), including: (1) inflation-adjusted total promotion spending ($ and percent of sales); (2) distribution by channel (consumer v. provider); and (3) provider specialty both for the industry as a whole and for top-selling biologic and small molecule therapies. Total promotion peaked in 2004 at US$36.1 billion (13.4% of sales). By 2010 it had declined to $27.7B (9.0% of sales). Between 2006 and 2010, similar declines were seen for promotion to providers and DTCA (both by 25%). DTCA's share of total promotion increased from 12% in 2002 to 18% in 2006, but then declined to 16% and remains highly concentrated. Number of products promoted to providers peaked in 2004 at over 3000, and then declined 20% by 2010. In contrast to top-selling small molecule therapies having an average of $370 million (8.8% of sales) spent on promotion, top biologics were promoted less, with only $33 million (1.4% of sales) spent per product. Little change occurred in the composition of promotion between primary care physicians and specialists from 2001-2010. These findings suggest that pharmaceutical companies have reduced promotion following changes in the pharmaceutical pipeline and patent expiry for several blockbuster drugs. Promotional strategies for biologic drugs differ substantially from small molecule therapies.
Article
Full-text available
A study was conducted to understand the influence of coupons and consumers' level of involvement in direct-to-consumer advertising. Consumers exposed to prescription drug advertising with a coupon had significantly more favorable ad and brand-related attitudes, and intention to inquire about the drug to their doctor. However, there was no significant difference in perceived product risk between consumers exposed to the ad with a coupon and consumers exposed to the ad without a coupon. Highly involved consumers had significantly more favorable ad, brand, and coupon-related attitudes, drug inquiry intention, and perceptions about the risks associated with the drug.
Article
Half TitleSeries InfoTitleCopyrightDedicationContentsPreface
Article
A conventional way to estimate the marginal effect of a risk factor is to omit the risk factor (e.g., gender, treatment) from the multivariate model and then use the omitted risk factor as class/group variable to compare observed and predicted outcomes. This can have an undesirable impact on the model since the predicted outcomes could be incorrect if the omitted variable is a significant risk factor. An alternative approach is to use the recycled prediction method to estimate and compare marginal effects without removing the risk factor from the model. While STATA (StataCorp, 2005) and SUDAAN (Research Triangle Institute, 2004) have provided sub-routines for the recycled prediction method, SAS® does not have a dedicated procedure for this purpose. This can be accomplished, however, by using simple SAS programming techniques. In this paper, we use an example to demonstrate a way to perform the recycled predictions within the SAS platform.
Article
Expenditures for specialty drugs account for more than 25 percent of total US drug spending and have been increasing at more than 13 percent annually. We examined insurers' role in maintaining the affordability and accessibility of specialty drugs while maximizing their value. We conducted two analyses: one using an administrative claims database with information on more than ten million commercially insured patients and another using the same database combined with the drug prescription records from a specialty pharmacy. First, we examined the prevalence of specialty drug coupons and the degree to which these reduced patients' out-of-pocket costs, focusing on 264,801 prescriptions. Second, we quantified the association between the magnitude of out-of-pocket costs for specialty drugs and patients' abandonment of their new or restarted therapy, focusing on a group of nearly 16,000 patients. We found that drug coupons accounted for $21.2 million of patients' $35.3 million annual out-of-pocket costs. In the vast majority of cases, coupons reduced monthly cost sharing to less than $250, a point at which patients were far less likely to abandon therapy with biologic anti-inflammatory drugs or with drugs for multiple sclerosis. However, by reducing cost sharing, coupons may also circumvent efforts to encourage patients to use the most cost-effective drugs.
Article
Pharmaceutical manufacturer coupons are a rapidly growing promotional activity intended to encourage initiation and continuing use of brand-name medications, but little is known about impacts on medication adherence and expenditures. To understand which patients use manufacturer coupons and the impact of coupons on brand-name statin (atorvastatin and rosuvastatin) use and expenditures 1 year after initiation of statin therapy. Using commercially available claims data spanning 3 years and representing 340,350 patients, we compared demographics, statin use, and expenditures of patients initiating generic statins, brand-name statins without manufacturer coupons, and brand-name statins with manufacturer coupons. Differences in user groups were tested using chi-squared statistics and Wilcoxon-Mann-Whitney tests. Main outcome measures included statin fills, adherence, and expenditures, including patient out-of-pocket, payer, and total costs. With the exception of population density, there were no significant demographic differences between new to therapy brand-name statin users filling prescriptions with and without coupons. New to therapy patients using generics were younger and lived in less populated areas compared with new to therapy brand-name statin noncoupon users. The number of statin fills in the 12 months following initiation was highest for coupon users, slightly lower for patients initiating generic statins, and lowest for noncoupon users (7.1 vs. 6.3 vs. 5.8; P less than 0.001), with corresponding medication adherence rates (61.1% vs. 60.1% vs. 53.8%; P less than 0.001). Coupon users had higher total statin prescription costs than generic initiators and noncoupon users ($798 vs. $92 vs. $678; P less than 0.001), and higher precoupon out-of-pocket costs ($339 vs. $53 vs. $169; P less than 0.001). Health plan costs for statins excluding rebates were lower for coupon users than noncoupon users ($460 vs. $508; P less than 0.001) but were much higher compared with generic statin initiators ($460 vs. $39; P less than 0.001). Brand-name statin initiators using coupons have higher adherence than patients initiating generic statins or brand-name statins without coupons. While the differences in adherence were statistically significant, they may not be clinically significant. Higher adherence among coupon users appears to occur at the expense of higher out-of-pocket and total statin expenditures.
Article
Visit nearly any official website for a brand-name drug available in the United States and, mixed in with links to prescribing and safety information, you'll find links to drug "coupons," including copayment-assistance programs and monthly savings cards. Most offers are variations on "Why pay more? With the [drug] savings card, you can get [drug] for only $18 per prescription if eligible" or "Get a free 30-capsule trial of [drug] with your doctor's prescription and ask your doctor if [drug] is right for you." Why do manufacturers offer drug coupons? Are they good for patients in the long run? Are they . . .
Article
This paper proposes an extension of generalized linear models to the analysis of longitudinal data. We introduce a class of estimating equations that give consistent estimates of the regression parameters and of their variance under mild assumptions about the time dependence. The estimating equations are derived without specifying the joint distribution of a subject's observations yet they reduce to the score equations for niultivariate Gaussian outcomes. Asymptotic theory is presented for the general class of estimators. Specific cases in which we assume independence, m-dependence and exchangeable correlation structures from each subject are discussed. Efficiency of the pioposecl estimators in two simple situations is considered. The approach is closely related to quasi-likelihood.
Article
Many patients continue to struggle with the costs of pharmaceutical products. For patients with chronic diseases, pharmaceutical costs are an unremitting expense. Nonadherence related to the cost of pharmaceutical agents is a major public health issue. This nonadherence is associated with increased hospitalizations and adverse health effects such as increased risk for stroke and acute myocardial infarction.1,2 Although Medicare Part D has decreased cost barriers to obtaining medications, 20% of Medicare beneficiaries in fair or poor health report cost-related nonadherence.3 Enhanced adherence is an important strategy to improve health outcomes. Drug coupons that discount patient co-pays have emerged as a popular tool to decrease the financial burden of prescription drugs. From 2009 to 2011, the number of drug manufacturer coupons increased markedly, to an estimated 340 individual drug coupon programs.4