Download How Prescription Drug Use Affects Health Care Utilization and

HowPr
e
s
c
r
i
pt
i
onDr
ugUs
e
Affe
c
t
sHe
a
l
t
hCa
r
eUt
i
l
i
z
a
t
i
ona
nd
Spe
ndi
ngbyOl
de
r
Ame
r
i
c
a
ns
:
ARe
v
i
e
woft
heLi
t
e
r
a
t
ur
e
Ci
ndyPa
r
k
sThoma
s
,
Ph.
D.
Br
a
nde
i
sUni
v
e
r
s
i
t
y
Sc
hne
i
de
rI
ns
t
i
t
ut
ef
orHe
a
l
t
hPol
i
c
y
How Prescription Drug Use
Affects Health Care Utilization and
Spending by Older Americans:
A Review of the Literature
Cindy Parks Thomas, Ph.D.
Brandeis University
Schneider Institute for Health Policy
AARP’s Public Policy Institute informs and stimulates public debate on the issues we
face as we age. Through research, analysis and dialogue with the nation’s leading experts,
PPI promotes development of sound, creative policies to address our common need for
economic security, health care, and quality of life.
The views expressed herein are for information, debate, and discussion, and do not
necessarily represent official policies of AARP.
2008-04
April 2008
© 2008, AARP
Reprinting with permission only
AARP, 601 E Street, NW
Washington, DC 20049
http:///www.aarp.org/ppi
Acknowledgments
The author would like to thank Stanley S. Wallack, Christine E. Bishop, and Karen
Richards Tyo of Brandeis University Schneider Institute for Health Policy for comments
on earlier versions, Susan Raetzman and Keith Lind of AARP for guidance on this
project, and several anonymous reviewers for their valuable contributions to this paper.
ii
FOREWORD
Prescription drugs are intended to have beneficial clinical effects. One would hope their
use might also reduce patients’ need to use other health care services. This effect is often
observed in the literature (Chrischilles, Dasbach, Rubenstein et al., 2001; Sokol,
McGuigan, Verbrugge, & Epstein, 2005). Sometimes these cost offsets are offered to
justify prescription drug prices, particularly in the case of brand-name drugs
(Lichtenberg, 2002). Despite several studies demonstrating that specific prescription
drugs and/or therapeutic classes can produce offsets in utilization of non-drug medical
services and may even result in net savings, evidence is sparse on the effects of
expanding prescription drug coverage and changing prescription cost sharing for patients.
The 2006 implementation of the Medicare prescription drug benefit has sparked interest
in the effects of such coverage expansions.
While some prescription drug use may not produce cost offsets (e.g., some patients may
live longer and incur higher health care costs), in other cases, prescription drug use may
result in some degree of cost offsets and possibly net savings (e.g., some medications
may prevent or delay serious progression of disease and associated costs). The nature and
extent to which studies have found these effects is the subject of this paper.
AARP’s Public Policy Institute commissioned this paper to review and summarize
evidence in the professional literature regarding the impact of expanding prescription
drug coverage on non-drug utilization and associated cost offsets, particularly for older
Americans. As used in this paper, the concept of “cost offsets” is intended to mean some
level of cost reduction in non-drug spending that is associated with prescription drug use.
Net savings is used to mean that prescription drug costs are more than offset by reduced
spending on non-drug services.
Although some of the most important effects of prescription drug use include its impact
on quality of life and longevity, this paper does not explore these effects fully, largely
because they need to be better described in the literature.
While none of the studies reviewed for this paper included findings from the Medicare
prescription drug program because such data have not yet been released, the findings
have obvious relevance and implications for the Medicare program. To the extent that
expanded Medicare coverage for prescription drugs produces cost offsets in other parts of
the Medicare program, overall Medicare spending may be affected favorably.
We hope this paper will serve as a resource for analysts, advocates, policy makers and
stakeholders who are interested in and concerned about this topic to: (1) establish a
baseline regarding areas of basic agreement in study findings; (2) identify areas where
consensus is lacking; (3) identify gaps in the literature; and (4) provide guidance on
issues that deserve further investigation.
Keith D. Lind
Strategic Policy Advisor
AARP Public Policy Institute
iii
THIS PAGE INTENTIONALLY LEFT BLANK
iv
TABLE OF CONTENTS
ACKNOWLEDGMENTS ................................................................................................... ii
FOREWORD.................................................................................................................... iii
EXECUTIVE SUMMARY ................................................................................................ vii
Background.............................................................................................................. vii
Purpose..................................................................................................................... vii
Methodology............................................................................................................ vii
Findings-in-Brief ..................................................................................................... vii
I.
INTRODUCTION...................................................................................................... 1
II.
CONCEPTUAL FRAMEWORK ............................................................................... 2
III.
IDENTIFYING GAPS IN THE LITERATURE........................................................... 4
IV.
METHODS FOR LITERATURE SELECTION AND REVIEW ................................. 6
V.
FINDINGS ................................................................................................................ 7
A. Use of Individual Prescription Drugs .................................................................. 7
B. Overview of Studies Reviewed............................................................................ 8
C. Prescription Drug Coverage, Health Status, and Non-Drug
Medical Spending ................................................................................................ 8
D. Prescription Drug Benefit Design...................................................................... 11
1. Administrative Controls and Benefit Limits................................................ 12
2. Prescription Drug Cost Sharing and Spending for Non-Drug Services....... 14
3. Reference Pricing......................................................................................... 16
E. Newer versus Older Drugs................................................................................. 16
VI.
NATURE AND QUALITY OF THE EVIDENCE ..................................................... 16
VII. CONSENSUS FINDINGS ...................................................................................... 18
VIII. POLICY IMPLICATIONS ....................................................................................... 19
IX.
RECOMMENDATIONS FOR FUTURE RESEARCH ............................................ 19
X.
CONCLUSION ....................................................................................................... 21
REFERENCES................................................................................................................ 23
APPENDIX...................................................................................................................... 33
v
THIS PAGE INTENTIONALLY LEFT BLANK
vi
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
EXECUTIVE SUMMARY
Background
Prescription drugs play a central role in the health and well-being of older Americans. In
spite of the expectation that greater use of prescription drugs should lead to improved
health outcomes, fewer complications, decreased disability, and lower overall health care
costs, the impact of prescription drug insurance coverage on medical spending has not
been well documented. While the literature generally agrees that providing prescription
drug coverage increases use of medications, estimates of the overall effect on heath and
medical costs of providing such coverage have ranged widely. This is particularly true for
studies that look at program-level outcomes for Medicare and Medicaid and at broad
populations.
Purpose
This paper assesses the professional literature on the impact of prescription drug
insurance and the use of prescription drugs on utilization and spending by the elderly on
prescription drugs and non-drug health care services, with a particular focus on the
contribution of newer research.
Methodology
This paper systematically reviews peer-reviewed and comparable professional literature
on the impact of prescription drugs, prescription drug insurance coverage, and benefit
design on drug and non-drug medical care utilization and spending in the older
population. For a list of studies reviewed, see the appendix and references.
Findings-in-Brief
The studies we reviewed have examined primarily the impact of providing prescription
drug insurance and specific benefit design features, such as various cost-sharing
approaches. With certain caveats described below, the review finds support for the
following:
ƒ
Prescription drug coverage can produce cost offsets from reductions in non-drug
services, such as hospitalizations and emergency visits.
ƒ
Studies that incorporate increased longevity into spending projections suggest that
cost offsets may diminish over time.
ƒ
Strict benefit limits of all kinds (spending caps or number of prescriptions) decrease
prescription drug use and increase use of other medical services, including acute and
long-term care services.
ƒ
Increased prescription drug cost sharing decreases the use of essential, as well as nonessential, classes of medications.
ƒ
In general, increased patient cost sharing reduces use and spending for prescription
drugs, with a 10% increase in drug cost sharing being associated with a decrease in
drug use of approximately 1%–6%.
vii
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
ƒ
Appropriate use of prescription drugs leads to improved health outcomes for many
chronic conditions in the elderly.
ƒ
Prescription drug coverage increases drug use by as much as 20% and improves
patient adherence to medication regimens.
ƒ
The effects of acquiring prescription drug coverage and increased cost sharing on
prescription drug use, use of other medical services, and health outcomes are often
delayed, so accounting for them warrants at least a 12-month follow-up.
A considerable lack of consensus remains in important areas:
ƒ
The magnitude of cost offsets associated with broad changes in prescription drug
coverage and/or use has not been determined reliably.
ƒ
Whether cost offsets that arise produce net savings after costs associated with
increased prescription drug used are taken into account.
ƒ
The marginal effects of various drug benefit design features (e.g., caps, cost sharing,
formularies, etc.) on cost offsets to non-drug services are not conclusive.
Finally, although it was not the central focus of this paper, the review found a dearth of
evidence regarding the effects of broad changes in prescription drug coverage on
disability or quality of life.
Several caveats apply to these findings, including:
ƒ
Many previous studies, particularly older ones, have relied on small sample sizes and
have not used robust methodologies. For instance, many studies have relied on crosssectional rather than longitudinal data and have not used adequate controls for
unobserved factors.
ƒ
Few studies take into account the dynamic relationship among prescription drugs,
health status, demand for insurance, and spending over time.
ƒ
While recent studies have begun to address these dynamic relationships, most studies
remain limited by a lack of large, longitudinal databases and by methodological
limitations, such as potential selection bias due to difficulty controlling for
unobserved factors.
ƒ
None of the studies reviewed for this paper included findings from the Medicare
prescription drug program because such data have not yet been released.
The consensus findings of this review suggest some important policy implications,
including:
ƒ
Appropriate prescription drug use should be encouraged, particularly among the
elderly with chronic conditions.
ƒ
When patient adherence to prescription drug regimens is less than optimal, it can be
improved by reducing prescription drug prices for patients by adding coverage for the
viii
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
uninsured and reducing cost sharing for those who have prescription drug coverage.
However, increased adherence may also increase prescription drug spending.
ƒ
One may need to target cost sharing for prescription drugs by type of medication and
patient condition (through programs such as value-based insurance design) to
minimize unintended consequences in other areas, such as non-drug spending and
health outcomes.
ƒ
We may reduce unintentional increased use and spending for non-drug services by
avoiding strict benefit limits on prescription drug use.
ƒ
We should view with skepticism study findings that do not take delayed effects into
account.
In addition to the policy implications of these study findings, further clarification of the
impact of prescription drug coverage and benefit design would enable a range of
stakeholders (e.g., patients, payors, and policy makers) to make more informed choices.
Questions that need to be addressed include the following:
ƒ
What are the effects of high cost sharing and significant gaps in prescription drug
coverage, such as those found in high-deductible health plans (i.e., health savings
accounts) and the Medicare Part D drug benefit (i.e., the donut hole)?
ƒ
What are the effects of prescription drug coverage and patient adherence, not only on
cost offsets from other medical services, but also on health status, quality of life, and
productivity?
ƒ
What are the effects of targeted coverage designs such as differential cost sharing for
particular drugs or classes to encourage more use of effective drugs in the patients
who would most benefit?
To understand these issues better, researchers should consider undertaking future studies
that combine the sophisticated analytic techniques used in some of the recent studies
reviewed in this paper with prospective studies and/or data sets representing natural
experiments (where legal or policy changes have produced substantial changes in the
scope of prescription drug coverage, such as the recent addition of a Medicare
prescription drug benefit). While study designs using natural experiments may still suffer
from methodological limitations (i.e., bias, limited generalizability, etc.), they do offer
important opportunities to sort out the effects of changes in prescription drug coverage
and, in particular, different benefit designs.
With implementation of a prescription drug benefit as part of the Medicare program in
2006, analysis of Medicare data may make it possible to better identify longer-term
effects of prescription drug coverage on health outcomes as well as on costs (both drug
costs and cost offsets from other services). However, Medicare studies will require
linking Part D drug data with hospital claims data from Part A and physician claims data
from Part B. We may facilitate sorting out the effects of various drug benefit designs by
linking Medicare data with non-Medicare data sets, such as claims from Medicare
supplemental insurance (i.e., Medigap) and commercial claims for people before they
became Medicare eligible. To fully explore the data and parse these and other questions
ix
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
with important policy implications, government and private researchers will need to have
timely access to longitudinal data sets.
x
THIS PAGE INTENTIONALLY LEFT BLANK
xi
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
I. INTRODUCTION
Prescription drugs play a critical role in the health and well-being of individuals. In
particular, for older members of the population, medications are a foundation for
treatment of chronic disease: nearly 90% of Americans age 65 and older take prescription
medications, with half taking three or more medications regularly (National Center for
Health Statistics [NCHS], 2004). Medications, especially those used to treat chronic
diseases, are now used routinely to prevent disease progression and avoid complications.
Insurance for prescription drugs, while by no means assuring that individuals will receive
needed medications, offers the potential to improve younger and older Americans’ access
to medications. The literature has shown quite consistently that having prescription drug
coverage is associated with increased use of medications. In spite of the very reasonable
expectation that increased use of prescription drugs could lead to fewer complications,
fewer instances of disability, improved health outcomes, and lower overall health care
costs, the link between these factors is not well established empirically. This is
particularly true for studies that look at program-level outcomes for Medicare and
Medicaid and at broader populations.
There has been a lack of consensus in professional literature on the extent to which
providing insurance coverage for prescription drugs reduces or avoids the use of other
medical services. Questions also remain regarding which benefit design features of
prescription drug insurance can lead to improved adherence to drug regimens and
whether such adherence leads to better health and outcomes for patients.
Since before passage of the Medicare Modernization Act of 2003, there has been a
pressing need for, and interest in, understanding and documenting whether providing
prescription drug coverage leads to improved health outcomes and/or offsets in utilization
and cost of other medical services. Research to assess the aggregate program effects of an
expansion in coverage across all therapeutic drug classes is critical for budget projections
and program cost effects, but may mask strong effects in particular disease areas.
Research focusing on specific clinical areas and on specific insurance design features,
such as cost sharing or administrative controls, is also necessary to determine optimal
design of coverage.
This paper assesses the literature through 2007 to determine what past and future research
can offer both methodologically and for policy purposes. It critically examines research
on the impact of prescription drug use and coverage design on health care utilization and
costs, to highlight underlying relationships and suggest considerations for future studies.
While the clinical benefits of individual prescription drugs are well documented, the
focus of this review is the impact of older adults’ prescription drug coverage and use on
non-drug health care utilization and costs. However, because the demand for and use of
prescription drugs over time is largely intertwined with health status, this relationship
cannot be ignored, and studies that address these important links are highlighted. This
paper addresses the following questions:
1. Conceptually, what is the effect of prescription drug coverage on the broader health
care system?
1
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
2. How do changes in older persons’ prescription drug utilization affect their use of and
spending for other health services?
3. What is the nature and quality of the evidence regarding these questions? How
methodologically rigorous are the studies that produced it? To what extent are the
study results relevant to the Medicare population? Is evidence from recent studies
methodologically stronger than that from previous studies?
4. Based on the evidence, can one project potential cost offsets or savings from expanded
access to prescription drugs under Medicare’s new prescription drug benefit?
5. What further research would be useful?
To address these questions, this paper first provides a conceptual framework for analysis.
Second, it describes the literature regarding direct effects of prescription drug insurance
on use and spending and identifies gaps in the literature. After describing methods used
for literature selection and review, this paper describes the literature regarding effects of
specific drugs and drug classes on use and spending, followed by a discussion of the
impact of prescription drug coverage and benefit design on non-drug spending. Finally,
the paper summarizes the nature and quality of the evidence, areas of consensus and lack
of consensus, policy implications, and recommendations for future research.
II. CONCEPTUAL FRAMEWORK
To better understand the dynamics through which prescription drug insurance contributes
to changes in use of drugs and other medical services, it is important to review the basic
factors that contribute to prescription drug use and spending. These can be grouped into
categories adapted from several health and social behavioral models (Anderson, 1995;
Piette, Heisler, Horne, & Alexander, 2006):
ƒ
Patient characteristics: Demographics, health status, medical need, sociocultural
influences, perceived benefits of the prescription, income, health literacy, insurance
status (medical and/or prescription drug), and risk preferences in the demand for
insurance.
ƒ
Characteristics/insurance benefit design: Insurance influences use of drugs by
changing individuals’ out-of-pocket costs. These costs vary by type of insurance
(limited or expansive networks, physician services access); source (public or private);
insurance benefit controls (formularies, drug utilization management, benefit limits);
cost-sharing structures; and availability/use of other services both previously and
concurrently.
ƒ
Health care system factors and complementary inputs: Availability of providers;
organization of care; payment systems; available benefits and coverage, including
pharmacy assistance and discount programs; associated non-drug costs of obtaining
medications (travel, etc.); and advertising pressures.
ƒ
Provider characteristics: Physician knowledge of prescription therapy and its costs;
practice style; therapeutic preferences; provider prescribing incentives; pharmacist
availability; and knowledge.
2
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
ƒ
Medication characteristics: Price; ease of administration; associated costs of
administration; perceived and actual effectiveness; availability of substitutes;
associated risks and potential side effects; and advertising influences.
The above factors contribute to the supply and demand for medical care and the rate at
which drugs and other services are consumed. In economic terms, these factors,
particularly past and concurrent medical care utilization and health status, contribute to
the demand for health care services that produce health (known as the health production
function [Grossman, 1972]). However, the interaction among these factors is complex, as
many of them change over time or cannot be observed from available data (e.g.,
individual preferences for insurance, undocumented health care events, individual
budgets and spending preferences).
Figure 1 illustrates the relationship among prescription drug insurance, prescription
drug consumption, and related medical services. These relationships are the organizing
framework for the current literature review.
Economic theory predicts that as insurance lowers a medication’s out-of-pocket cost,
use of that medication will increase. The type and features of insurance benefit design
influence the demand for prescription drugs and other services. At the same time, the
demand for insurance itself is dependent on health status, which, in turn, contributes
to changes in the demand for prescription drugs and other medical services. This
relationship is particularly strong for prescription drug insurance; if individuals could
predict short-term medication use, they could purchase insurance as a prepayment
mechanism for their medications.
Prescription drugs may be viewed as substitutes for some services (offsetting the need
for some services, such as hospitalizations and emergency care) but complementary to
other services (resulting in increases in physician or other services associated with
prescribing and managing treatment and adverse events). To the extent that adherence to
drug regimens improves health outcomes and longevity (Gowrisankaran & Town, 2005;
Heisler, Langa, Eby, & Fendrick, 2005), health care costs incurred during added years of
life, in turn, may offset short-term savings from reductions in use and spending for other
services.
These dynamics mean that the effect of prescription drugs on health should be accounted
for in estimates of future health care spending. Further complicating matters, many of the
relevant factors in this process are unobservable and may change over time—for instance,
prior health status, risk preferences, provider practice style, health system patterns,
natural progression of disease, and the interaction between medical care and health status
over time. If the study design and data analysis do not address these issues adequately,
the impact of prescription drug coverage on the health care system will be overestimated
for some services and underestimated for others.
3
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
Figure 1
Relationships of Prescription Drug Insurance, Prescription Drug Use,
and Use of Non-Drug Medical Services
Controlling factors:
Patient
Health System
Physician
Patient
need/
demand
Selection
Drug (and MD)
insurance coverage
Type
Substitute Services:
hospital
emergency
other outpatient
long-term care
Utilization
Cost sharing
mgt.
Purchase meds:
full vs. actual
compliance
Change in
health status/
mortality
Net effect on
health/other
services and
costs over
time
Complementary Services:
physician and
other related to adverse
drug events or drug
management
Dynamic health effects
When estimating the impact of policies on the broader health care system, researchers
should take into account such factors as:
ƒ
differences between insured and uninsured populations (observed and unobserved
selection);
ƒ
changes in health status;
ƒ
improvements in longevity;
ƒ
insurance benefit design characteristics; and
ƒ
other observed and unobserved characteristics of patients.
Predictive models (both those that estimate future program costs and those that estimate
the impact of particular design features) that do not take these risk factors into account
are likely to produce biased estimates of the effects of prescription drugs and insurance
coverage on health and medical care utilization.
III. IDENTIFYING GAPS IN THE LITERATURE
Numerous studies have shown that having prescription drug insurance (versus having
none) is associated with increased use of prescription drugs. Most studies of older
Americans have used data from the Medicare Current Beneficiary Survey (MCBS), other
national surveys, or retiree health care claims. These studies usually have been based on
cross-sectional data (analyses at a fixed time), adjusting for measurable health status, and
4
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
other indicators. Such studies show a moderate to strong insurance effect on the
likelihood of prescription drug use (insured persons are 6%–17% more likely to use drugs
[Gianfresco & Baines, 1994; Stuart & Grana 1998]), with generally larger differences
between insured and uninsured for particular drug classes (Adams 2001a; Blustein, 2000;
Doshi, Brandt, & Stuart, 2004; Federman, Adams, Ross-Degnan, Soumerai, & Ayanian,
2001; Stuart, Doshi, Briesacher, Wrobel, & Baysac, 2004). Longitudinal studies (time
series analysis of individual patient records or aggregate population data) that have
followed populations as they gain insurance suggest that it is important to observe this
dynamic over time (Briesacher, 2005; Khan, Kaestner, & Lin, 2007; Stuart & Coulson
1993; Stuart et al., 2004).
Numerous studies have also established that changes in beneficiary cost sharing for
prescription drugs affects drug utilization and spending by insured populations (Gibson,
Ozminkowski, & Goetzal, 2005; Gruber, 2006; Lexchin & Grootendorst 2004; Maio et
al., 2005; Rice & Matsuoka 2004). These studies confirm that higher drug cost sharing
results in lower utilization of and spending for drugs, but the magnitude of this effect
varies. Studies that report price elasticity of demand for drugs (an economic measure of
consumer responsiveness to prices) generally range from -0.1 to -0.6 (meaning that a
10% increase in consumer cost sharing results in a 1%–6% decrease in medication
utilization) (Goldman, Joyce, & Zheng, 2007; Gruber, 2006; Shea, Terza, Stuart, &
Briesacher, 2007). Additional findings from this research include:
ƒ
Levels of cost sharing and plan generosity are strongly related to prescription drug
consumption (Artz, Hadsall, & Scholdelmeyer, 2002; Doshi & Polsky, 2007;
Huskamp, 2003; Joyce, Escarce, Soloman, & Goldman, 2002; Stuart & Zacker, 1999).
ƒ
Effects of insurance coverage design on utilization vary across drug classes (Goldman
et al., 2004; Stuart & Grana 1998).
ƒ
Having insurance for physician visits is critical to using a drug benefit (Coulson &
Stuart 1992; Grootendorst, O’Brien, & Anderson, 1997; Rogowski, Lillard, &
Kington, 1997).
ƒ
Beneficiary cost sharing has a strong impact on drug adherence and/or health
outcomes for those with chronic diseases (Cole, Norman, Weatherby, & Walker,
2006; Gibson et al., 2005; Goldman et al., 2004).
Studies that have attempted to distinguish between essential and non-essential drug use
have shown that even though non-essential drug use may be more responsive to higher
cost sharing, both categories are affected to some extent (Goldman et al., 2004; Johnson,
Goodman, Hornbrook, & Eldredge, 1997b; Tamblyn, Laprise, Hanley, Abrahamowicz
et al., 2001).
Studies that consider insurer type (e.g., Medicaid, Medigap, private managed care) have
shown that:
ƒ
Medicaid drug coverage is associated with higher drug consumption and spending
compared to other types of supplemental insurance (Adams, 2001a; Rice &
Matsuoka, 2004), and
5
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
ƒ
managed care populations use drugs differently from individuals with other types of
insurance, including greater use of chronic care drugs, newer drugs, and generic
drugs, and the effects of cost sharing may be lower than they are in other populations
(Hillman et al., 1999; Lyles & Palumbo, 1999; Stafford et al., 2003; Wallack, Martin,
Thomas, & Ryan, 2007).
While studies have clearly established a direct relationship among prescription drug
coverage, drug utilization, and spending, the relationship between prescription drug
coverage and use of other health care services has not been as well established. For this
reason, we undertook this review.
IV. METHODS FOR LITERATURE SELECTION AND REVIEW
Relevant studies for this review included those of the impact of drug coverage on use
of non-drug medical services. Specific outcomes of interest included both overall
health care utilization and its various components (i.e., hospitalization, physician visits,
emergency department visits, and skilled nursing and other long-term care services)
and costs associated with these services. We also discuss relevant examples of research
examining the effect of optimal use of particular drug classes on the cost of other
medical services.
We conducted a systematic literature review following standard approaches (Cochrane
Collaborative Center for Reviews and Dissemination, 2006; Fink, 2004). We searched
articles through online sources, including National Library of Medicine (PubMed,
Medline), EconLit, Web of Science, and Google Scholar, using both standard searches
and “reverse” searches of articles citing relevant sources. In addition to the standard
journals, we looked for government website publications and published foundation
reports that exhibited sufficient methodological rigor. In particular, several of those
published by the National Bureau of Economic Research (NBER) fell into this category.
Further, we queried outside experts to identify the most influential papers on the topic
and methodologies to ensure that we identified all relevant articles.
We considered articles for review only if they directly investigated either the impact of
prescription drug use in particular classes on medical or long-term care services, or the
impact of prescription drug insurance programs and benefit design (cost sharing,
coverage caps, formularies, etc.) on drug use and other medical services.
While an extensive literature exists on the clinical benefits of particular drugs and drug
classes in controlled trials, together with a more limited number of studies of their cost
effectiveness (Neumann, 2005), we have not attempted to summarize these results for a
number of reasons. Most of these studies demonstrate the clinical or cost effectiveness of
using a particular prescription drug for a particular patient population. In addition to the
narrow scope of these studies, other concerns have been raised regarding the
generalizability of these studies to broad expansions of prescription drug insurance
coverage (e.g., publication bias—the tendency to publish favorable results and funding
from industry sources); limited generalizability outside of trial populations; and lack of
benefit design considerations (Congressional Budget Office [CBO], 2002). Nevertheless,
we have included selected studies with particular relevance to older populations or
chronic disease.
6
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
To be included in this review, studies had to meet one of the following criteria: peer
reviewed or equivalent; outcomes as noted (acute medical service use and spending,
long-term service use and spending, economic evaluations); international studies if
comparable and relevant to the U.S. health care system (e.g., Canadian); computer
simulation models for forecasting costs and health outcomes; adequate sample size;
and transparent methods. We also included studies that examined older populations,
exclusively or as a subgroup, or had particular relevance to chronic disease or the
Medicare population, as well as relevant methodological papers. Observational studies
based on claims data or patient reporting of objective measures, such as diagnoses, were
included only if they controlled for patient demographics. We excluded opinion surveys
that report subjective information on health status or cost related to prescription drug.
V. FINDINGS
A. Use of Individual Prescription Drugs
There is ample evidence from clinical trials that, used correctly in the right patients,
individual prescription drugs and drug classes can improve health status. This literature
has been reviewed elsewhere (e.g., O’Brien, 2003). The evidence from these studies is
mixed on the extent of cost offsets from reductions in other medical service utilization
that may be associated with appropriate medication therapy and tends to vary by clinical
condition and drug type. As we discuss further below, these findings have led some
analysts to attribute substantial increases in the population’s health status over the past
few decades to the availability of new prescription drugs (Lichtenberg, 2006). Some have
inferred that increased spending associated with improvements in health care technology,
including prescription drugs, has been well worth the cost (Cutler, Rosen, & Vijan,
2006). Others have shown through cost-benefit analysis that the use of some drugs
produces considerable savings in some areas, but this varies greatly by type of
medication, disease, and population (Neumann, 2005).
For many chronic diseases, medication use has produced significant health care cost
savings in controlled settings that eliminate the insurance effect. For instance, research
has shown that AIDS anti-retroviral therapy can produce savings in less hospital use and
lower hospital spending (Bozzette et al., 2001). Other studies have suggested that annual
savings from such interventions may be offset by increased longevity (Goldman et al.,
2004; Lichtenberg, 2006). The use of drugs in Alzheimer’s disease has also shown
sizeable cost offsets through fewer hospitalizations and less residential care in patients
with moderate to severe disease (Feldman, Gauthier, Hecker et al., 2004; Hill, Futterman,
Mastey, & Fillit, 2002; Lu & Fillit, 2005). In osteoporosis, a randomized controlled trial
found that drug therapy reduced the incidence and health care costs associated with
fractures by 35% (Chrischilles et al., 2001). Additional conditions for which adequate
adherence to medications has been strongly associated with lower health care costs and
hospitalization rates include asthma, diabetes, hypertension, high cholesterol, and heart
disease (Balkrishnan, Christensen, & Bowton, 2002; Balkrishnan, Rajagopalan, &
Camacho, 2003; Fischer & Avorn, 2004; Gibson et al., 2006; Hepke, Martus, & Share,
2004; Sokol et al., 2005; Stephens, Botteman, & Hay, 2006; Stukel, Lucas, & Wennberg,
2005). On the other hand, several economic analyses of prescription drug use for
treatment of mental health conditions have shown improved clinical outcomes but mixed
7
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
evidence of health care cost offsets (Frank, McGuire, & Normand, 2006; Lave, Frank,
Schulberg, & Kamlet, 1998; McCombs, Nichol, Johnstone et al., 2000).
Such studies suggest the potential for prescription drugs prescribed and consumed under
controlled circumstances to lead to health care cost offsets, if not net savings. However,
because insurance coverage and other factors strongly influence prescription drug use, the
remainder of this review focuses on the literature dealing with the effect of insurance
coverage and benefit design on prescription drug use and related offsets in use and cost of
non-drug health care services.
B. Overview of Studies Reviewed
A large volume of published articles address various aspects of prescription drug
coverage and health care costs, but only a limited number use robust methods and apply
to older populations. We found 10 studies (peer-reviewed or equivalent) that assessed the
impact of having or gaining prescription drug coverage. A number of these assessed the
effect of payor type (managed care versus other models, and public versus private
insurance) on drug and other medical costs for specific clinical conditions.
Twenty-nine studies assessed the effect of insurance benefit design on use of medical and
other non-drug services, particularly for older adults. Studies examined the impact of
either administrative or financial controls (benefit caps, formularies, and prior
authorization) or patient cost sharing (copayments, coinsurance, progressive cost-sharing
tiers, or reference pricing). The following section briefly summarizes findings from older
studies, then discusses findings from newer studies in more detail.
C. Prescription Drug Coverage, Health Status, and
Non-Drug Medical Spending
Literature examining the impact of prescription drug coverage and use on non-drug
spending is limited, but increasing evidence suggests that providing prescription drug
coverage has a small but significant effect on other medical services, at least in the short
run. Most studies have used observational data (most often the MCBS, which combines
claims data and patient reporting), comparing the insured to the uninsured, to estimate the
impact of drug coverage on non-drug services and on the overall costs of the Medicare
program. Five studies published since 2004 estimate the program-wide impact of adding
drug benefit coverage in terms of cost offsets, either for the Medicare program alone or
for the entire population.
All nine studies found that an increase in prescription drug use was associated with
insurance coverage, in both probability of prescription drug use and prescription drug
spending. Six of these studies showed drug coverage to decrease costs of Part A and/or
Part B services significantly. The remaining three showed little or no long-term cost
offset from drug coverage. The methods used to estimate drug cost offsets varied in
amount and type of data, analytical methods, and follow-up period, as discussed in more
detail below.
An earlier study by Lingle, Kirk, and Kelley (1987) showed that low-income seniors
enrolled in a state pharmacy assistance program, and thus more likely to fill prescriptions,
had lower hospital costs. However, the age of the study, inability to infer a causal
8
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
relationship, and restriction of the study population to low-income seniors limit the
validity and relevance of these findings.
Several MCBS-based studies provide important insights into medical cost offsets from
prescription drug coverage, but they have limited generalizability beyond the study
population, due to their methods and small sample size. Briesacher and associates (2005)
showed non-significant decreases in hospital spending for all Medicare beneficiaries who
had acquired drug coverage in several two-year periods. This study adjusted for
differences in health status that might have encouraged sicker patients to gain drug
coverage preferentially, and our review revealed no other obvious evidence of selection
bias. This study estimated that acquiring prescription drug coverage increased drug
spending by up to 66% after the first year. Although the study found some lagging cost
offsets from lower hospital and physician spending, these effects were not statistically
significant. As the study’s authors noted, the results are not generalizable and should not
be considered definitive, but they do suggest that, if this trend continues and becomes
significant, the impact of drug coverage may continue well into the second or third year
after drug coverage is acquired.
Stuart and associates (2004) examined the impact of drug coverage for Medicare
beneficiaries with chronic obstructive pulmonary disease (COPD). A propensity score
approach (i.e., matching or weighting insured and uninsured individuals on observable
characteristics to better account for unobserved differences between the two groups)
based on 1999 morbidity was used to correct for selection bias. Those with drug coverage
paid 61% more for COPD drugs, had 29% less physician spending, and had no significant
differences in hospital spending in 2000, after adjusting for covariates. As in the previous
study, Stuart and associates (2004) relied on a small sample (384 beneficiaries with
coverage and 78 controls), so this study should be considered exploratory.
Atherly (2004) examined the impact of Medicare supplemental drug coverage on
asthmatics, using cross-sectional data from MCBS 2001. While drug coverage was
associated with 21% greater physician spending for all beneficiaries, for those with
asthma, physician spending was actually 7% less. This suggests that drug coverage may
have a differential impact on certain types and severity of disease. Although the analysis
was adjusted for covariates, its findings are still limited because it used cross-sectional
data.
In another estimate of the overall cost impact of adding a prescription drug benefit for
an older population, Gilman, Gage, and Mitchell (2004) used data from the Vermont
Medicaid pharmacy waiver program, which provided prescription drug benefits to nonMedicaid, low-income seniors. This study, which used a design based on data from one
year before and one year after implementation, found drug coverage for low-income
seniors to be associated with an average increase in Medicare spending of $1,000 during
the enrollment year. The following year, the population experienced a return to baseline
spending, suggesting that a change in health status may have led to enrollment in
this program, but that overall spending was not affected. The study also showed a
simultaneous statewide increase in the number of seniors enrolling in Medicaid, but one
should not infer a causal relationship with drug coverage due to the study’s methodology.
Instead, this finding was more likely the result of low-income individuals signing up for
the state pharmacy assistance program and also enrolling in the state Medicaid program.
9
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
Using a very different methodology that relies heavily on assumptions, Rosen and
associates (2005) simulated the insurance effect of an anti-hypertensive medication for
diabetics, allowing for disease progression and modeling the effect of expected increased
drug adherence due to insurance coverage. Based on epidemiological and costeffectiveness data, this study used no direct patient data; instead, it relied on sensitivity
analyses to model the expected patient response to drug coverage. Based on computer
simulation of first dollar coverage and estimates of how the addition of drug coverage
might increase use of the drug in question, this study found that prescription drug
coverage led to lower annual Medicare costs ($1,606 less) and increased longevity (0.23
quality-adjusted life years [QALYs]). However, on a program-wide basis, the study also
found that annual savings would be offset by increased use of other health care services
during added years of life. In a related sensitivity analysis based on more limited
coverage available under the standard Medicare drug benefit, the study found that, in
spite of annual savings from fewer hospitalizations, coverage for the drug in question
would result in net costs to Medicare, although coverage would be relatively cost
effective (i.e., less than $1,000 for the expected increase in QALYs per diabetic).
Several other recent studies also used computer simulation to model future health
spending associated with providing drug benefits through the Medicare program.
Recognizing the importance of the dynamics of health status and demand for drugs in
specifying the impact of drug coverage, Yang, Gilleskie, and Norton (2004) used
multiyear MCBS data (1992–1998) to estimate the effect of drug coverage, taking into
account the dynamics of insurance choice, health status changes, and unobserved patient
characteristics. In doing so, they incorporated the following assumptions:
ƒ
Patient preferences for insurance coverage depend on health status.
ƒ
Use of different types of medical care correlates with illness.
ƒ
Current medical care consumption influences future health and future consumption.
ƒ
Past medical care consumption influences current consumption (Yang et al., 2004;
Yang Gilleskie, & Norton, forthcoming).
Using longitudinal data, this study addressed several of the limitations of static, crosssectional analyses and demonstrated the extent to which other study estimates may have
been biased by not adjusting adequately for earlier medical use. Based on this model, the
authors found that drug coverage would increase drug spending by 12%–17% over five
years, with smaller cost offsets from less use of other services. However, increased drug
consumption would increase longevity, which would add to health care spending. As a
result, drug coverage was predicted to result in small increases in Medicare Parts A and B
spending after five years due to improved longevity. These estimates of the impact of
drug coverage on drug spending were slightly lower than those of other study findings.
Updating MCBS data through 2001 and using the same methods, the authors found a
similar (7%–27%) increase in drug spending over a five-year interval and similar effects
on long-term health and spending (e.g., increased longevity and greater use of other
services) (Yang et al, forthcoming). In a separate paper using the same techniques, the
authors estimated the impact of different sources of drug coverage, showing a small but
10
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
significant subsequent year savings to Medicare from providing drug coverage (Yang &
Norton, 2006).
In another study, also using multiple years of MCBS data as the basis for computer
simulation, Shang (2005) estimated that Medigap drug coverage increased drug spending
by 22%, but was offset by reductions in Medicare Part A (hospital spending) of 10%–
13% ($277–$348 per capita per year). This study found the net effect to be that a $1
increase in drug spending was associated with an estimated $1.60–$2.00 decrease in
annual Medicare spending. This study went on to model the cost effectiveness of drug
treatment for hypertension on overall costs, assuming full patient adherence and
accounting for changing morbidity and mortality over five years. In this costeffectiveness analysis, the study found that, depending on drug price, the increased cost
of drug therapy for hypertension was offset entirely by lower Medicare spending on other
services, resulting in net annual savings to Medicare and increased longevity of
beneficiaries. However, as in some of the other studies described, this study found that
the long-term impact on total long-term Medicare spending was small due to greater
spending for health care services during increased longevity. In addition to these findings,
this study provides important methodological lessons (using a discrete factor econometric
model) by demonstrating that adjusting for unobserved as well as observed characteristics
of beneficiaries results in lower estimates of the cost of drug coverage, in contrast to
models using only observable characteristics.
Using longitudinal MCBS data (1992–2000), Khan and associates (2007) estimated the
effect of prescription drug coverage for the elderly on prescription drug use and health
status to compare those who changed insurance status to those who did not. This study
found that gaining public insurance increased drug use by 14% in the following year,
while acquiring employer-sponsored insurance or HMO coverage increased drug use by
6%. Treating insurance coverage as unrelated to health (i.e., random) and adjusting for
health status, they did not find drug coverage to be associated with health status,
disability, or hospitalization rates in the year following acquisition of coverage.
In summary, available studies of the impact of prescription drug coverage for older
Americans on non-drug services have produced a range of findings. However, there is
general agreement that prescription drug coverage can produce cost offsets from
reductions in non-drug services. Studies that incorporate increased longevity into
spending projections suggest that cost offsets may diminish over time. On the other hand,
longevity is a positive impact that may not translate to cost offsets or net savings.
Several of these recent studies make sophisticated efforts to account for selection bias as well
as changes in health status and demand for insurance, drugs, and other services over time
(Khan et al., 2007; Shang, 2005; Yang et al., 2004; Yang et al., forthcoming). By contrast,
some studies do not correct adequately for selection bias or time-variant effects, and, as a
result, they may overstate the impact of drug insurance on drug spending while understating
drug insurance’s impact on other services (Briesacher, 2005; Stuart et al., 2004).
D. Prescription Drug Benefit Design
Numerous reviews have reviewed the literature on the impact of prescription drug
insurance benefit design on drug use and spending (Adams, 2001b; Gibson et al., 2005;
Goldman et al, 2007; Gruber, 2006; Lexchin & Grootendorst, 2004; Maio et al., 2005;
11
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
Rice & Matsuoka, 2004) and of administrative controls (Aeserud et al., 2006; Dewa &
Hoch, 2003; Hoadley, 2005; Lyles & Palumbo, 1999; Maio et al., 2005; Soumerai, RossDegnan, Fortress, & Abelson, 1993; Soumerai, 2004). These studies have generally
agreed on the following:
ƒ
Increased cost sharing leads to lower drug use and spending.
ƒ
Benefit limits, such as caps on drug spending and the number of prescriptions, reduce
prescription drug use.
ƒ
Hospitalizations, nursing home use, and emergency department visits increase with
limits to drug coverage, particularly in frail individuals and those with chronic
conditions.
We summarize studies of the impact of benefit design in the following section with
respect to particular design elements: administrative controls, benefit limits, cost sharing,
and reference pricing.
1. Administrative Controls and Benefit Limits
Administrative controls are important mechanisms by which public and private payors
control the type and amount of prescription drug use by patients. Common administrative
controls, found in many drug benefit designs, include benefit limits, formulary design,
and drug utilization review (prior authorization for drugs, step therapy, dosage unit limits,
etc.). Recent less widely adopted administrative controls include financial incentives for
providers and pharmacists as well as medication therapy management (a pharmacist’s
periodic review and coordination of prescription drugs).
Soumerai et al. (1993) extensively reviewed drug utilization management policies used
by state Medicaid programs and concluded that few well-designed studies applied
adequate research methods to evaluate prescribing controls. This review found that
administrative controls, such as prior authorization in particular, substantially decrease
use of the targeted medications, but evidence was lacking on the effect of reduced drug
use on health or other medical services. A recent update of the literature found that more
research is still needed in this area (Soumerai, 2004).
Studies of prescription benefit caps have concluded that, although they limit use and cost
of drug therapy, costs from increased use of other services outweigh savings associated
with reduced drug use, at least for such vulnerable populations as the frail elderly. In a
series of seminal studies assessing a limit on the number of prescriptions, Soumerai and
associates found that restricting drug benefits with a monthly prescription limit reduced
use of essential and nonessential drugs and increased overall Medicaid spending
(Soumerai, Avorn, Ross-Degnan, & Gortmaker, 1987), and increased use of nursing
homes, physician services, and emergency room visits (Soumerai, Ross-Degnan, Avorn,
& McLaughlin, 1991)—particularly for mental health services (Soumerai, McLaughlin,
Ross-Degnan, Casteris, & Bollini, 1994). Research on other Medicaid programs
confirmed many of these findings (Cromwell, Bass, Steinberg et al., 1998; Martin &
MacMillan, 1996). Experts often cite studies by Soumerai et al. (1987, 1991, 1994) as
evidence that drug utilization controls lead to adverse outcomes and increase acute and
long-term care use and spending. However, these studies were limited to frail elderly,
12
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
who were at the highest risk of hospitalization and, thus, may not be generalizable to a
broader elderly population.
Drug expenditure limits have been evaluated as well. In a methodologically robust study
(regression-adjusted, cross-sectional comparison), Hsu, Price, Huang, and Brand (2006)
found that an expenditure cap in a health maintenance organization (HMO) was
associated with 31% lower drug costs, together with higher emergency department use,
more non-elective hospitalizations, and higher death rates. In the group with drug
expenditure limits, non-adherence rates were higher for patients with hypertension,
diabetes, and high cholesterol, and physiological outcomes were worse. Savings from
reduced prescription drug use were offset by increased spending in other areas, and total
medical spending did not differ significantly among groups. An important feature of this
study was the availability of clinical data to link patient non-adherence to prescribed drug
regimens in the group with expenditure caps to indicators of poorer clinical outcomes
(higher blood cholesterol levels, higher blood pressure readings, and higher blood sugar
levels in diabetics) in the group with expenditure caps. Opinion surveys of patients
subject to drug benefit caps, as well as an analysis of insurance claims data, confirm the
above findings that patients discontinue their medications after reaching a cap on drug
benefits (Cox & Henderson, 2002; Fortess, Soumerai, McLaughlin, & Ross-Degnan,
2001; Joyce, Goldman, Karaca-Mandic, & Zheng, 2007; Schulz, Lingle, Chubon, &
Coster-Schulz, 1995; Tseng, Brook, Keeler, & Mangione, 2003; Tseng, Brook, Keeler,
Steers, & Mangione, 2004). While these findings may not be directly applicable, they
have potentially important implications for the design of Medicare’s prescription drug
benefit, which includes a gap in coverage, also known as the “donut hole.”
Formularies (lists of covered drugs) are another feature of most drug benefit plans. In a
few controlled studies, closed formularies (those that limit which drugs are covered,
versus open formularies, which impose no restrictions) were associated with lower drug
expenditures, but the impact on non-drug services has been inconclusive, and generally
these studies have not been well controlled for patient selection and differences across
health plans. One study of working insured adults (Motheral & Henderson, 1999)
indicated that closed formularies did not result in increased spending on other services.
However, other studies did show increasing use of medical services in privately insured
populations (Horn, Sharkey, & Phillips-Harris, 1998; Murawski & Abdelgawad, 2005)
and decreasing use of hospital care in Medicaid (Kozma, Reeder, & Lingle, 1990).
Christian-Hermann, Emons, and George (2004) found that a generic-only formulary
decreased health plan drug use and costs across several chronic disease categories,
increased member out-of-pocket spending, and resulted in an increase in hospital
admissions. This study has particular relevance to Medicare because many Part D drug
plans provide generic-only coverage, no brand-name drug coverage, in the coverage gap.
Clearly, more work is needed to evaluate the impact of formulary design as it affects use
of other health care services and outcomes. In summary, a number of systematic studies
have assessed particular administrative controls of drug use on health care spending.
Almost all of these studies indicate that broad administrative controls lower drug
spending, but do not adequately discriminate between types of drug spending that may be
offset by reduced spending in other areas or health effects (either favorable or adverse),
especially for individuals with chronic disease or who have low incomes.
13
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
2. Prescription Drug Cost Sharing and Spending for Non-Drug Services
Early study findings have varied regarding the effects of changes in copayments or
coinsurance for prescription drug coverage on spending for other health care services
(Balkrishnan, Byerly, Camacho, Shrestha, & Anderson, 2001; Fairman, Motheral, &
Henderson, 2003; Johnson, Goodman, Hornbrook, & Eldredge, 1997a; Johnson et al.,
1997b; Motheral & Fairman, 2001; Pilote, Beck, Richard, & Eisenberg, 2002; Tamblyn
et al., 2001). Recent studies suggest that savings sought by payors from higher patient
cost sharing for prescriptions drugs may be offset by higher medical costs (Chandra,
Gruber, & McKnight, 2007; Cole et al., 2006; Dor & Encinosa, 2004; Gaynor, Li, &
Vogt, 2006; Goldman et al., 2004; Goldman, Joyce, & Karaca-Mandic, 2006; Li, Guh,
Lacaille, Esdaile, & Anis, 2007; Zeber, Grazier, Valenstein, Blow, & Lantz, 2007).
Studies that focus on higher drug cost sharing by patients with particular chronic diseases
have found cost offsets from increased hospitalizations (Cole et al., 2006; Goldman et al.,
2006; Zeber et al., 2007).
Several observational studies have used a natural experimental design with comparison
groups, adjusted for observable patient-related characteristics, to assess the impact of
patient cost sharing. Tamblyn and associates (2001) found that introducing 25%
coinsurance for an elderly, low-income Canadian population led to a decrease in use of
both essential medications (down 9%) and less essential medications (down 14%), while
adverse events doubled. Another Canadian study of post-heart attack patients found no
significant effects from a modest increase in drug cost sharing on drug use or other
services after one year (Pilote et al., 2002). However, the authors note that this finding
may be limited to patients who have experienced a serious medical event, such as a heart
attack.
A study of elderly HMO enrollees who received prescription drug coverage under two
different formularies found that across-the-board limits on patient drug spending were
associated with a significant increase in total HMO costs. On the other hand, the
formulary that expanded access to generic drugs but did not limit patient drug spending
per se was associated with a 6% decline in total HMO costs. However, the results may
have been confounded because the formulary changes were implemented at different
times, and the study lacked adequate controls (Balkrishnan et al., 2001). Fairman and
associates (2003) and Motheral and Fairman (2001) found the greatest effect of a threetier formulary (e.g., increasing copayments for generic, preferred, and non-preferred
brand-name drugs) was shifting drug costs to health plan enrollees. This study found no
change in inpatient or outpatient spending after one year or three years; however, costsharing increases were quite small.
Recently, Chandra and associates (2007) analyzed the differential impact of prescription
drug cost sharing on the Medicare program and private supplemental (Medigap) insurers.
This study compared the impact of increased cost sharing for California enrollees in
Medicare Advantage plans (HMOs and preferred provider organizations [PPOs]) who
also had supplemental drug coverage from their former employer. This study found that
demand was highly responsive to cost sharing across drug categories (i.e., drugs for
acute, chronic, and discretionary use), particularly for HMO enrollees. At the same time,
higher patient cost sharing resulted in a net saving for Medigap insurers. While this study
had non-trivial limitations (simultaneous increases in cost sharing across services, and
analysis at the plan level rather than the level of individual members), it is the only one to
14
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
compare the impact of prescription drug cost sharing on the Medicare program and
Medigap insurers. The findings from this study have important policy implications for
Medicare, which incurred higher program costs from hospitalizations associated with
decreased prescription drug use.
Other studies have shown a strong response to prescription drug cost sharing and a strong
offset effect on medical service use. Using cross-sectional data, Goldman and associates
(2004) found that doubling copayments for working-age, insured adults decreased
prescription drug use by one-third to one-half, but increased emergency visits by 17%
and hospital days by 10%. The authors cautioned that copayments for other services may
have changed simultaneously with drug copayments, which could have confounded the
study results.
In a study based on a large multiemployer database of prescription drug and medical
claims (MarketScan), adjusted for changing individual health status and other differences,
Gaynor and associates (2006) found that, for insured individuals under age 65, a $1
increase in drug cost sharing reduced drug spending (through fewer prescriptions) by
$23.62 in the first year, and $32.57 by the second year. However, because patients
substituted outpatient care for drugs, total medical spending decreased by less than that
amount: $20.88 in the first year and $21.23 by the second. In the long run, drug savings
from higher cost sharing were largely offset by increases in spending for outpatient care.
The authors conclude that higher drug cost sharing may not control health spending;
however, this finding was revealed only through lagged (i.e., time series) analysis.
Dor and Encinosa (2004) used cross-sectional data from a large insured population to
compare the differential impact of drug coverage with flat copayments (a fixed-dollar
amount regardless of total price per prescription) to coinsurance (a fixed percent of the
full drug price). In an insured adult diabetic population, the study found that coinsurance
decreased patient compliance (i.e., reduced refills of diabetes medication) to a greater
degree than did fixed-dollar copayments. Drawing on other research, the authors
estimated that a $6–$10 increase in drug copayments would result in $125 million per
year contemporaneous savings in national spending on diabetes drugs, but would increase
spending for other medical services by $360 million per year, for a net national cost of
$235 million annually.
In a Canadian study, Li et al. (2007) found that cost sharing by seniors with rheumatoid
arthritis had relatively little impact on prescription drug use (i.e., price elasticity for drugs
was low), but physician visits increased because, in Canada, such visits require no cost
sharing. The authors concluded that there was a strong incentive to substitute physician
services for prescription drugs.
In a computer simulation, Goldman et al. (2006) used Medicare (MCBS) data to estimate
the effect of “evidence-based copayments” on treatment of high cholesterol. They found
that a “value-based insurance design” that eliminated copayments for medium- and highrisk individuals, and increased copayments to $22 for low-risk individuals, would result
in net savings by avoiding 80,000 hospitalizations and 31,000 emergency department
visits per year.
Other studies of health plans and employer groups have also reported savings from such
value-based insurance designs (Chernew, Rosen, & Fendrick, 2007; Mahoney, 2005), but
15
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
these have not been rigorous or well controlled. One controlled time-series study assessed
the impact of value-based insurance design on drug adherence in a disease management
environment (Chernew et al., 2008). Although this study was limited to two employer
groups, the authors found that lower copayments for chronic disease treatment resulted in
greater adherence to diabetes medications and several categories of heart disease
medications. This study did not measure the impact of such a plan design on medical
services, but it does suggest that this approach to coverage promotes better adherence to
medications for patients most at risk and may prevent disease complications.
3. Reference Pricing
Reference pricing, a cost-sharing approach in which one or several medications within
each class is fully covered, but individuals must pay the extra cost of higher-priced drugs
in the class, has been evaluated in a series of Canadian studies for anti-hypertensives,
anti-ulcer drugs, anti-angina drugs, and non-steroidal anti-inflammatory drugs (Hazlet &
Blough, 2001; Grootendorst et al., 2005; Grootendorst & Stewart, 2006; Marshall et al.,
2002; Schneeweiss, Dormuth, Grootendorst, Soumerai, & Maclure, 2004; Schneeweiss,
Walker, Glynn, Maclure, Dormuth, & Soumerai, 2002). The studies that assessed the
impact of reference pricing on non-drug health services found that it resulted in net
savings to the province of British Columbia, with either no change in spending for other
services (Hazlet & Blough, 2001) or a transitory spending increase in physician and
hospital care that subsided after two months (Schneeweiss et al., 2002, 2004).
E. Newer versus Older Drugs
In a series of studies, Lichtenberg (2001, 2002) has attempted to demonstrate that newer
(patent-protected, brand-name) drugs that generally cost more than older (brand-name
and generic) drugs are associated with greater offsets in non-drug spending. These studies
generally have found that newer drugs are more effective, have fewer side effects, and are
associated with lower overall medical expenditures than are older drugs. However, these
studies suffer from several methodological weaknesses that have raised concerns
regarding these findings, including confounding from a concurrent slowdown in overall
health care spending, changes in prescribing patterns, and differences in disease severity.
Raising additional questions about these findings, other studies using individual-level
data of prescription drug use for mental health conditions have found that new
antipsychotic medications had the opposite effect on Medicaid populations—that is,
while newer drugs may have offered clinical advantages, they did not reduce spending on
other medical services (Del Paggio, Finley, & Cavano, 2002; Duggan 2005; Frank et al.,
2006). In a literature review, Mintzes and Lexchin (2005) concluded that only 5% of
drugs introduced in Canada from 1996 to 2000 offered improved effectiveness over
older, less expensive drugs and, thus, credited no net savings to overall health spending.
Our review found no studies of potential cost offsets that may be associated with the
relatively new category of specialty drugs, known as biologics.
VI. NATURE AND QUALITY OF THE EVIDENCE
Studies regarding cost offsets that may be associated with broad-based prescription drug
coverage and expanded utilization employ a wide range of methodologies to analyze
existing data sets retrospectively. While many researchers hold out randomized,
controlled trials (RCTs) as the “gold standard” for establishing causation and replicable
16
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
results, there is a notable absence of such rigorous studies in the literature regarding
prescription drug cost offsets. Further, for a host of reasons (e.g., time, expense, difficulty
randomizing, and questions about generalizability), RCTs on this topic are unlikely to be
forthcoming anytime soon. While some might find observational studies more reliable if
performed prospectively (depending on a variety of factors, such as sample size, size of
expected effects, etc.), this review found no such prospective studies of prescription drug
coverage or use.
The literature on the topic consists almost entirely of retrospective observational studies
based on claims data and/or self-reported data from a variety of sources. These types of
studies are quicker and less expensive to perform than are prospective studies. Also,
studies that rely on existing data sources are more likely to describe real-world behavior
than are prospective studies, which often create artificial conditions. In addition,
retrospective studies have the advantage of being able to derive statistical power from
large data sets and, thus, are more likely to detect more subtle effects than are prospective
studies, which often rely on smaller sample sizes.
As described above, methodological issues, particularly selection bias, associated with
this type of research are substantial. Studies that compare analytic techniques indicate
that inadequately controlling for selection bias tends to produce overestimates of the
impact of insurance on drug use and spending. Several sources of selection bias include
incomplete information regarding the extent to which:
ƒ
sicker people are more likely to seek drug insurance coverage than are healthier
people;
ƒ
people are likely to select benefit design features in ways that correlate with their
health status;
ƒ
changes in health status are likely to influence demand for prescription drugs; and
ƒ
use of prescription drugs and non-drug medical services influence each other.
More recent studies have used enhanced methods, such as indirect econometric
corrections, to better account for biases and data limitations. Recent studies have revealed
the importance of accounting for delayed responses (i.e., lagged effects) to prescription
drug coverage and use (Briesacher, 2005; Chandra et al., 2007; Gaynor et al., 2006;
Zeber et al., 2007), and unobserved factors, such as health status (Chandra et al., 2007;
Yang & Norton, 2006). However, even the most recent observational studies do not
entirely agree on the size of cost offsets that may be associated with prescription drug
coverage and use—for example, Shang (2005) found significant cost offsets, whereas
Yang and associates (2004, forthcoming) found minimal overall cost offsets.
Some recent studies have produced promising results by combining claims data with
results from clinical trials in computer simulations that model health and spending effects
of prescription drugs on specific clinical conditions (Goldman et al., 2006; Rosen et al.,
2005; Shang, 2005). However, examining the impact of drug use on a disease-specific
basis has important limitations, since many older Americans have a number of
conditions. While some studies of prescription drug effects have simulated the effects of
increased longevity and disability, as well as cost offsets (Joyce, Keeler, Shang, &
17
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
Goldman, 2005; Yang et al., 2004), these studies have not taken fully into account the
value of related benefits (e.g., improved physical and emotional health status, quality of
life, and productivity).
VII. CONSENSUS FINDINGS
Based on more rigorous studies, researchers generally agree on and support the following
broad findings:
ƒ
Prescription drug coverage can produce cost offsets from reductions in non-drug
services, such as hospitalizations and emergency visits.
ƒ
Studies that incorporate increased longevity into spending projections suggest that
cost offsets may diminish over time.
ƒ
Strict benefit limits of all kinds (spending caps or number of prescriptions) decrease
prescription drug use and increase use of other medical services, including acute and
long-term care services.
ƒ
Increased prescription drug cost sharing decreases the use of essential, as well as
nonessential, classes of medications.
ƒ
In general, increased patient cost sharing reduces use and spending for prescription
drugs, with a 10% increase in drug cost sharing being associated with a decrease in
drug use of approximately 1%–6%.
ƒ
Appropriate use of prescription drugs leads to improved health outcomes for many
chronic conditions in the elderly.
ƒ
Prescription drug coverage increases drug use by as much as 20% and improves
patient adherence to medication regimens.
ƒ
The effects of acquiring prescription drug coverage and increased cost sharing on
prescription drug use, use of other medical services, and health outcomes are often
delayed, so accounting for them warrants at least a 12-month follow-up.
A considerable lack of consensus remains in important areas:
ƒ
The magnitude of cost offsets associated with broad changes in prescription drug
coverage and/or use has not been determined reliably.
ƒ
Whether cost offsets that arise produce net savings after costs associated with
increase prescription drug used are taken into account.
ƒ
The effects of various drug benefit design features (e.g., caps, cost sharing,
formularies, etc.) on cost offsets to non-drug services.
Finally, although it was not the central focus of this paper, the review found a dearth of
evidence regarding the effects of broad changes on prescription drug coverage on
disability or quality of life.
18
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
VIII. POLICY IMPLICATIONS
Broadly based consensus findings from this review suggest some important policy
implications, including the following:
ƒ
Appropriate prescription drug use should be encouraged, particularly among the
elderly with chronic conditions.
ƒ
When patient adherence to prescription drug regimens is less than optimal, it can be
improved by reducing prescription drug prices for patients by adding coverage for the
uninsured and reducing cost sharing for those who have prescription drug coverage.
However, increased adherence will increase prescription drug spending.
ƒ
One may need to target cost sharing for prescription drugs may by type of medication
and patient condition (through programs such as value-based insurance design) to
minimize unintended consequences in other areas, such as non-drug spending and
health outcomes.
ƒ
We may reduce unintentional increased use and spending for non-drug services by
avoiding strict benefits limits on prescription drug use.
ƒ
We should view with skepticism study findings that do not take delayed effects into
account.
In addition to the policy implications of these study findings, further clarification of the
impact of prescription drug coverage and benefit design would enable a range of
stakeholders (e.g., patients, payors, and policy makers) to make more informed choices.
Questions that need to be addressed include the following:
ƒ
What are the effects of high cost sharing and significant gaps in prescription drug
coverage, such as those found in high-deductible health plans (i.e., health savings
accounts) and the Medicare Part D drug benefit (i.e., the donut hole)?
ƒ
What are the effects of prescription drug coverage and patient adherence, not only on
cost offsets from other medical services, but also on health status, quality of life, and
productivity?
ƒ
What are the effects of targeted coverage designs such as differential cost sharing for
particular drugs or classes to encourage more use of effective drugs in the patients
who would most benefit?
IX. RECOMMENDATIONS FOR FUTURE RESEARCH
To understand these issues better, researchers should consider undertaking future studies
that combine the sophisticated analytic techniques used in some of the recent studies
reviewed in this paper with prospective studies and/or data sets representing natural
experiments (where legal or policy changes have produced substantial changes in the
scope of prescription drug coverage, such as the recent addition of a Medicare
prescription drug benefit). While study designs using natural experiments may still suffer
from methodological limitations (i.e., bias, limited generalizability, etc.), they do offer
important opportunities to sort out the effects of changes in prescription drug coverage
19
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
and, in particular, different benefit designs. Future studies that rely on retrospective
observational data should account for the following:
ƒ
trends over time based on longitudinal data (preferably based on individual patient
records, rather than on aggregate data);
ƒ
interactions between use of medical services and unobserved variables, such as health
status;
ƒ
interactions among changes in prescription drug coverage, drug use, and use of and
spending for non-drug services; and
ƒ
delayed or lagged effects of changes in prescription drug coverage, drug use and its
effects on health, disability (and function), longevity, quality of life, and productivity
(based on follow-up for at least 12 months).
Implementation of Medicare Part D’s prescription drug benefit under the Medicare
Modernization Act of 2003 offers a potentially rich database and new challenges for
understanding the impact of prescription drug use in real-world settings. However, data
from Medicare Part D which was implemented in 2006 have not been released. In
addition, wide variations in benefit designs allowed under Part D (e.g., different cost
sharing, coverage gaps, formularies, and medication therapy management) will make it
challenging to sort out the effects on prescription drug use and cost offsets from the
effects on other non-drug services.
Future studies will need to assess the impact of the new Part D drug benefit on:
ƒ
access of Medicare beneficiaries to different types of prescription drugs,
ƒ
effects on subpopulations within Medicare, and
ƒ
effects of different health care delivery arrangements (i.e., HMOs versus fee-forservice, freestanding drug plans versus integrated health plans).
Responses to these questions will have important implications for policy and budget
purposes. In addition, Part D data may make it possible to better identify longer-term
effects of prescription drug coverage on health outcomes, such as disability and quality of
life, as well as costs (both drug costs and cost offsets from other services). However,
Medicare studies will require linking drug data from Part D with hospital claims data
from Part A and physician claims data from Part B. Sorting out the effects of various
benefit designs may be facilitated by linking Medicare data with non-Medicare data sets,
such as Medicare supplemental insurance (i.e., Medigap) and commercial claims for
patients before they became Medicare eligible.
To fully explore the data and parse these and other questions with important policy
implications, academic and other private researchers will need to have timely access to
longitudinal combined Medicare data sets.
20
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
X. CONCLUSION
Overall, it appears reasonable to conclude that prescription drug coverage improves
health outcomes for many chronic conditions in the elderly and increases drug spending.
It also appears to produce cost offsets from reduced utilization of non-drug services, at
least to some extent. However, many questions remain about the extent and focus of these
effects and how best to increase desired effects and minimize unintended and undesirable
consequences.
If finding answers to relevant questions in this area is left to uncoordinated, underfunded
research, studies are likely to produce piecemeal results and unlikely to provide definitive
answers to many questions of interest to researchers and other stakeholders. As a result,
policy makers and payors should consider funding research specifically designed to better
estimate the broad population effects of prescription drug use on overall spending and
health.
21
THIS PAGE INTENTIONALLY LEFT BLANK
22
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
REFERENCES
Adams A. (2001a). Use of antihypertensive drugs by Medicare enrollees: Does type of
drug coverage matter? Health Affairs, 20(1), 276–286.
Adams A. (2001b). The case for a Medicare drug coverage benefit: A critical review of
the empirical evidence. Annual Review of Public Health, 22, 49–61.
Aeserud, M, Dahlgren, A. T., Kosters, J. P., Oxman, A. D., Ramsay, C., & Sturm, H.
(2006). Pharmaceutical policies: Effects of reference pricing, other pricing, and
purchasing policies. Cochrane Database of Systematic Reviews. Retrieved from
http://www.thecochranelibrary.com
Anderson, R. M. (1995). Revisiting the behavioral model and access to medical care:
Does it matter? Journal of Health and Social Behavior, 36, 1–10.
Artz, M. B., Hadsall, R. S., & Schondelmeyer, S. W. (2002). Impact of generosity level
of outpatient prescription drug coverage on prescription drug events and
expenditure among older persons. American Journal of Public Health, 92,
1257–1263.
Atherly, A. (2004). The effect of prescription drug coverage on the cost of care to
Medicare beneficiaries with asthma. Expert Review of Pharmacoeconomics
Outcomes and Research, 4, 421–428.
Balkrishnan, R. M., Byerly, W. G., Camacho, F. T., Shrestha, A., & Anderson, R. T.
(2001). Effect of prescription benefit changes on medical care utilization in a
Medicare HMO population. American Journal of Managed Care, 7, 1093–1100.
Balkrishnan, R. M., Christensen, D. B., & Bowton, D. L. (2002). Self-reported health
status, prophylactic medication use, and health care costs in older adults with
asthma. Journal of the American Geriatrics Society, 50, 924–929.
Balkrishnan, R. M., Rajagopalan, R., & Camacho, F. T. (2003). Predictors of medication
adherence and associated health care costs in an older population with type 2
diabetes mellitus: A longitudinal cohort study. Clinical Therapeutics, 25,
2958–2971.
Blustein, J. (2000). Drug coverage and drug purchases by Medicare beneficiaries with
hypertension. Health Affairs, 19(2), 219–225.
Bozzette, S. A., Joyce, G., McCaffrey, D. F., Leibowitz, A. A., Morton, S. C., Berry, S. H.,
Rastegar, A., Timberlake, D., Shapiro, M. F., & Goldman, D. P.; HIV Cost and
Services Utilization Study Consortium (2001). Expenditures for the care of HIVinfected patients in the era of highly active antiretroviral therapy. The New
England Journal of Medicine, 344(11), 817-823.
23
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
Briesacher, B. (2005). Medicare beneficiaries and the impact of gaining prescription drug
coverage on inpatient and physician spending. Health Services Research, 40[5
part 1], 1279–1296.
Chandra, A., Gruber, J., & McKnight, R. (2007, March). Patient cost sharing,
hospitalization offsets, and the design of optimal health insurance for the elderly.
National Bureau of Economic Research Working Paper 12972. Retrieved from
http://www.nber.org/papers/12972
Chernew, M. E., Rosen, A. B., & Fendrick, A. M. (2007). Value-based insurance design.
Health Affairs, 26(2), w195–203.
Chernew, M. E., Shah, M. R., Wegh, A., Rosenberg, S. N., Juster, I. A., Rosen, A. B.,
Sokol, M. C., Yu-Isenberg, K., & Fendrick, A. M.(2008). Impact of decreasing
copayments on medication adherence within a disease management environment.
Health Affairs 27(1), 103-112
Chrischilles, E. A., Dasbach, E. J., Rubenstein, L. M., et al. (2001). The effect of
alendronate on fracture-related healthcare utilization and costs: The fracture
intervention trial. Osteoporosis, 12(8), 654–660.
Christian-Hermann, J., Emons, M., & George, D. (2004). Effects of generic-only drug
coverage in a Medicare HMO. Health Affairs, web exclusive, w4-455–468.
Retrieved December 1, 2006, from http://www.healthaffairs.org
Cochrane Collaborative Center for Reviews and Dissemination. (2006). Undertaking
systematic reviews of research on effectiveness; CRD’s guidance for those
carrying out or commissioning reviews. Retrieved July 15, 2006, from
http://www.york.ac.uk/inst/crd/report4.htm
Cole, J. A., Norman, H., Weatherby, L. B., & Walker, A. M. (2006). Drug copayment
and adherence in chronic heart failure: Effect on cost and outcomes.
Pharmacotherapy, 26(8), 1157–1164.
Congressional Budget Office (CBO). (2002, October). Issues in designing a prescription
drug benefit for Medicare. Washington, DC: Author.
Coulson, N. E., & Stuart, B. (1992). Persistence in the use of pharmaceuticals by the
elderly. Evidence from annual claims. Journal of Health Economics, 11, 360–364.
Cox, E. R., & Henderson, R. R. (2002). Prescription use behavior among Medicare
beneficiaries with capped prescription benefits. Journal of Managed Care
Pharmacy, 8, 360–364.
Cromwell, D. M., Bass, E. B., Steinberg, E. P., et al. (1998). Can restrictions on
reimbursement for anti-ulcer drugs decrease Medicaid pharmacy costs without
increasing hospitalizations? Health Services Research, 33, 1593–1601.
24
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
Cutler, D. M., Rosen, A. B., & Vijan, S. (2006). The value of medical spending in the
United States, 1960–2000. The New England Journal of Medicine, 355, 920–927.
Del Paggio, D., Finley, P. R., & Cavano, J. M. (2002). Clinical and economic outcomes
associated with olanzapine for the treatment of psychotic symptoms in a county
mental health population. Clinical Therapeutics, 24(5), 803–817.
Dewa, C. S., & Hoch, J. S. (2003). Fixed and flexible formularies as cost-control
mechanisms. Expert Reviews in Pharmacoeconomics Outcomes Research, 3,
303–35.
Dor, A., & Encinosa, W. (2004, August). Does cost sharing affect compliance? The case
of prescription drugs. National Bureau of Economic Research Working Paper No.
w10738. Retrieved from http://www.nber.org/papers/w10738
Doshi, J., Brandt, N., & Stuart, B. (2004). The impact of drug coverage on Cox-2
inhibitor use in Medicare. Health Affairs, web exclusive, w4-94-w4-105.
Accessed November 12, 2006, from http://www.healthaffairs.org
Doshi, J. A., & Polsky, D. (2007). Drug benefit generosity and essential medication use
among Medicare eligible retirees. American Journal of Managed Care, 13, 425–
431.
Duggan, M. (2005). Do the new prescription drugs pay for themselves? The case of
second-generation antipsychotics. Journal of Health Economics, 24, 1–31.
Fairman, K. A., Motheral, B. R., & Henderson, R. R. (2003). Retrospective, long-term
follow-up study of the effect of a three-tier prescription drug copayment system
on pharmaceutical and other medical utilization and costs. Clinical Therapeutics,
25, 3147–3161.
Federman, A. D., Adams, A. S., Ross-Degnan, D., Soumerai, S. B., & Ayanian, J. Z.
(2001). Supplemental insurance and use of effective cardiovascular drugs among
elderly Medicare beneficiaries with coronary heart disease. Journal of the
American Medical Association, 286(14), 1732–1739.
Feldman, H., Gauthier, S., Hecker, J., et al. (2004). Economic evaluation of donepezil in
moderate to severe Alzheimer disease. Neurology, 63(4), 644–650.
Fink, A. (2004). Conducting literature reviews (2nd ed.). London: Sage Publications.
Fischer, M. A., & Avorn, J. (2004). Economic implications of evidence-based prescribing
for hypertension. Journal of the American Medical Association, 291, 1850–1856.
Fortress, E. E., Soumerai, S. B., McLaughlin, T. J., & Ross-Degnan, D. (2001).
Utilization of essential medications by vulnerable older people after a drug benefit
cap: Importance of mental disorders, chronic pain, and practice setting. Journal of
the American Geriatrics Society, 49, 793–797l
25
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
Frank, R., McGuire, T. G., & Normand, S-L. (2006, October). Cost-offsets of new
medications for treatment of schizophrenia. National Bureau of Economic
Research Working Paper No. 12643. Cambridge, MA: National Bureau of
Economic Research.
Gaynor, M., Li, J., & Vogt, W. (2006, December). Is drug coverage a free lunch? Crossprice elasticities and the design of a prescription drug benefits. National Bureau of
Economic Research Working Paper No. 12758. Retrieved from
http://www.nber.org/papers/w12758
Gianfresco, F. D., & Baines, A. P. (1994). Utilization effects of prescription drug benefits
in an aging population. Health Care Financing Review, 15, 113–120.
Gibson, T. B., Mark, T. L., Axelsen, K., Baser, O., Rublee, D. A., & McGuigan, K. A.
(2006). Impact of statin copayments on adherence and medical care utilization
and expenditures. American Journal of Managed Care, 12(special issue), SP11–
SP19.
Gibson, T. B., Ozminkowski, R. J., & Goetzel, R. Z. (2005). The effects of prescription
drug cost sharing: A review of the evidence. The American Journal of Managed
Care, 11, 730–740.
Gilman, B., Gage, B., & Mitchell, J. B. (2004). Final report: Evaluation of Vermont’s
pharmacy assistance programs for low income Medicare beneficiaries. RTI
International, February 28, 2003. Retrieved from
http://www.cms.hhs.gov/Reports/downloads/gilman_2003_5.pdf
Goldman, D. P., Joyce, G. F., Escarce, J., Pace, J. E., Soloman, M., Lauori, M.,
Landsman, P., & Teutsch, S. (2004). Pharmacy benefits and the use of drugs by
the chronically ill. Journal of the American Medical Association, 291(19), 2344–
2350.
Goldman, D. P., Joyce, G. F., & Karaca-Mandic, P. (2006). Varying pharmacy benefits
with clinical status: the case of cholesterol-lowering therapy. American Journal of
Managed Care, 12(1), 21–28.
Goldman, D. P., Joyce, G. F., & Zheng, Y. (2007). Prescription drug cost sharing.
Associations with medication use and medical utilization and spending in health.
Journal of the American Medical Association, 298(1), 61–69.
Gowrisankaran, G., & Town, R. J. (2005). Drug benefits, managed care and mortality:
An analysis of elderly mortality. National Bureau of Economic Research Working
Paper No. w10204. Retrieved from http://www.nber.org/papers/w10204
Grootendorst, P. V., Marshall, J. K., Holbrook, A. M., Dolovich, L. R., O’Brien, B. J., &
Levy, A. R. (2005). The impact of reference pricing of nonsteroidal antiinflammatory agents on the use and costs of analgesic drugs. Health Services
Research, 40:5(Part 1), 1297–1317.
26
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
Grootendorst, P. V., O’Brien, B. J., & Anderson, G. M. (1997). On becoming 65 in
Ontario: Effects of drug plan eligibility on use of prescription drugs. Medical
Care, 35, 386–398.
Grootendorst, P. V., Solovich, L. R., O’Brien, B. J., Holbrook, A. M., & Levy, A. R.
(2001). Impact of reference-based pricing of nitrates on the use and costs of antianginal drugs. Canadian Medical Association Journal, 165, 1011–1019.
Grootendorst, P., & Stewart, D. (2006). A re-examination of the impact of reference
pricing on anti-hypertensive drug plan expenditures in British Columbia. Health
Economics, 15, 735–742.
Grossman, M. (1972). On the concept of health capital and the demand for health.
Journal of Political Economy, 80(2), 223–255.
Gruber, J. (2006, October). The role of consumer copayments for health care: Lessons
from the RAND health insurance experiment and beyond. Menlo Park, CA: Henry
J. Kaiser Family Foundation.
Hazlet, T. K., & Blough, D. K. (2001). Health services utilization with reference pricing
of histamine 2 receptor antagonists in British Columbia elderly. Medical Care, 40,
640–649.
Heisler, M., Langa, K., Eby, D., & Fendrick, M. (2005). The health effects of restricting
prescription medication use because of cost. Medical Care, 42(7), 626–634.
Hepke, K. L., Martus, M. T., & Share, D. A. (2004). Costs and utilization associated with
pharmaceutical adherence in a diabetic population. American Journal of Managed
Care, 10, 144–151.
Hill, J. W., Futterman, R., Mastey, V., & Fillit, H. (2002). The effect of donepezil therapy
on health costs in a Medicare managed care plan. Managed Care Interface, 15,
63–70.
Hillman, A. L., Pauly, M. V., Escarce, J., Ripley, K., Gaynor, M., Louse, J. C., & Ross R.
(1999). Financial incentives and drug spending in managed care. Health Affairs,
18(2), 189–200.
Hoadley, J. (2005, March). Cost containment strategies for prescription drugs: Assessing
the evidence in the literature. Menlo Park, CA: Henry J. Kaiser Family
Foundation.
Horn, S. D., Sharkey, P. D., & Phillips-Harris, C. (1998). Formulary limitations and the
elderly: Results from the managed care outcomes project. American Journal of
Managed Care, 4, 1105–1113.
Hsu, J., Price, M., Huang, J., & Brand, R. (2006). Unintended consequences of caps on
Medicare drug benefits. The New England Journal of Medicine, 354(22), 2349–
2359.
27
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
Huskamp, H. (2003). The effect of incentive-based formularies on prescription drug
utilization and spending. The New England Journal of Medicine, 349(23), 2224–
2232.
Johnson, R. E. Goodman, M. J., Hornbrook, M. C., & Eldredge, M. B. (1997a). The
effect of increased prescription drug cost-sharing on medical care utilization and
expenses of elderly health maintenance organization members. Medical Care, 35,
1119–1131.
Johnson, R. E., Goodman, M. J., Hornbrook, M. C., & Eldredge, M. B. (1997b). The
impact of increasing patient prescription drug cost sharing on therapeutic classes
of drugs received and on the health status of elderly HMO members. Health
Services Research, 32(1), 103–122.
Joyce, G., Escarce, J., Soloman, M., & Goldman, D. (2002). Employer drug benefit plans
and spending on prescription drugs. Journal of the American Medical Association,
288(14), 1733–1739.
Joyce, G. F., Goldman, D. P., Karaca-Mandic, P., & Zheng, Y. (2007). Pharmacy benefit
caps and the chronically ill. Health Affairs, 26(5), 1333–1344.
Joyce, G., Keeler, E. B., Shang, B., & Goldman, D. (2005, September 25). The lifetime
burden of chronic disease among the elderly. Health Affairs, web exclusive:w5R18-W5R29. Retrieved November 12, 2006, from http://www.healthaffairs.org
Khan, N., Kaestner, R., & Lin, S. (2007, January). Prescription drug insurance and its
effect on utilization and health of the elderly. National Bureau of Economic
Research Working Paper No. 12848. Retrieved from
http://www.nber.org/papers/w12848
Kozma, C. M., Reeder, E., & Lingle, E. W. (1990). Expanding Medicaid drug formulary
coverage. Effects on utilization of related services. Medical Care, 28, 963–977.
Lave, J., Frank, R., Schulberg, H. C., & Kamlet, M. (1998). Cost effectiveness of
treatments for major depression in primary care practice. Archives of General
Psychiatry, 55, 645–651.
Lexchin, J., & Grootendorst, P. (2004). Effects of prescription drug user fees on drug and
health services, and on health status in vulnerable populations: A systematic
review of the evidence. International Journal of Health Services, 34(1), 101–122.
Li, X., Guh, D., Lacaille, D., Esdaile, J., & Anis, A. (2007). The impact of cost sharing of
prescription drug expenditures on health care utilization by the elderly: Own- and
cross-price elasticities. Health Policy, 82, 340–347.
Lichtenberg, F. (2001). Are the benefits of newer drugs worth their cost? Evidence from
the 1996 MEPS. Health Affairs, 20(5), 241–251.
28
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
Lichtenberg, F. (2002, June). Benefits and costs of newer drugs: An update. National
Bureau of Economic Research Working Paper No. 8996.Cambridge, MA:
National Bureau of Economic Research.
Lichtenberg, F. (2006, August). The impact of increased utilization of HIV drugs on
longevity and medical expenditures: An assessment based on aggregate time
series data. National Bureau of Economic Research Working Paper No. 12406.
Retrieved from http://www.nber.org/papers/w12406
Lingle, E. W., Kirk, K. W., & Kelly, W. R. (1987). The impact of outpatient drug
benefits on the use and costs of health care services for the elderly. Inquiry, 24,
203–211.
Lu, S., Hill, J., & Fillit, H. (2005). Impact of donezepil use in routine clinical practice on
health care costs in patients with Alzheimer’s disease and related dementias
enrolled in a large Medicare managed care plan: A case control study. American
Journal of Geriatric Pharmacotherapy, 3(2), 92–102.
Lyles, A, & Palumbo F. (1999). The effect of managed care on prescription drug costs
and benefits. Pharmacoeconomics, 15, 129–140.
Mahoney, J. J. (2005). Reducing patient drug acquisition costs can lower diabetes health
claims. American Journal of Managed Care, 11(5 supplement), S170–S176.
Maio, V., Pizzi, L., Roumm, A., Clarke, J., Goldfarb, N., Nash, D., & Chess, D. (2005).
Pharmacy utilization and the Medicare Modernization Act. The Milbank
Quarterly, 83(1), 101–130.
Marshall, J. K., Grootendorst, P. V., O’Brien, B. J., Dolovisch, L. R., Holbrook, A. M., &
Levy, A. R. (2002). Impact of reference based pricing for histamine 2 receptor
antagonists and restricted access for proton pump inhibitors in British Columbia.
Canadian Medical Association Journal, 166, 1655–1662.
Martin, B. C., & MacMillan, J. A. (1996). The impact of implementing a more restrictive
prescription limit on Medicaid recipients. Medical Care, 34, 686–701.
McCombs, J. S., Nichol, M. B., Johnstone, B. M., et al. (2000). Antipsychotic drug use
patterns and the cost of treating schizophrenia. Psychiatric Services, 51, 525–527.
Mintzes, B., & Lexchin, J. (2005). Do higher drug prices lead to better health? Canadian
Journal of Clinical Pharmacology, 12, e10–e21.
Motheral, B., & Fairman, K. (2001). Effect of at three tiered prescription copay on
pharmaceutical and other medical utilization. Medical Care, 39(12), 1293–1304.
Motheral, B., & Henderson, R. (1999). The effect of a copay increase on pharmaceutical
utilization, expenditures, and treatment continuation. The American Journal of
Managed Care, 5, 1383-1394.
29
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
Murawski, M. M., & Abdelgawad, T. (2005). Exploration of the impact of preferred drug
lists on hospital and physician visits and the costs to Medicaid. American Journal
of Managed Care, 11, sp35–sp42.
National Center for Health Statistics (NCHS). (2004). Health, United States, 2004.
Hyattsville MD.
Neumann, P. J. (2005). Medicare and cost-effectiveness analysis. The New England
Journal of Medicine, 353, 1516–1522.
O’Brien, E. (2003, May). Will prescription drug coverage for the low-income elderly pay
for itself? A review of the literature. Prepared for the Kaiser Commission on
Medicaid and the Uninsured. Washington DC: Kaiser Family Foundation.
Piette, J. D., Heisler, M., Horne, R., & Alexander, G. C. (2006). A conceptually based
approach to understanding chronically ill patients’ responses to medication cost
pressures. Social Science and Medicine, 62, 846–857.
Pilote, L., Beck, C., Richard, H., & Eisenberg, M. J. (2002). The effects of cost sharing
on essential drug prescriptions, utilization of medical and outcomes after acute
myocardial infarction in elderly patients. Journal of the Canadian Medical
Association, 167, 246–252.
Popovian, R., Johnson, K. A., Nichol, M. B., & Liu, G. (1999). The impact of
pharmaceutical capitation to primary medical groups on the health care
expenditures of Medicare HMO enrollees. Journal of Managed Care Pharmacy,
5, 414–419.
Rice, T., & Matsuoka, K. (2004). The impact of cost sharing on appropriate utilization
and health status: A review of the literature on seniors. Medical Care Research
and Review, 61(4), 415–452.
Rogowski, J., Lillard, L. A., & Kington, R. (1997). The financial burden of prescription
drug use among elderly persons. The Gerontologist, 37, 475–482.
Rosen, A. B., Hamel, M. B., Weinstein, M. C., Cutler, D. M., Fendrick, A. M., & Vijan,
S. (2005). Cost-effectiveness of full Medicare coverage of angiotensin-converting
enzyme inhibitors for beneficiaries with diabetes. Annals of Internal Medicine,
143, 89–99.
Schneeweiss, S., Dormuth, C., Grootendorst, P., Soumerai, S. B., & Maclure, M. (2004).
Net health plan savings from reference pricing for angiotensin converting enzyme
inhibitors in elderly British Columbia residents. Medical Care, 42, 653–660.
Schneeweiss, S., Walker, A., Glynn, R., Maclure, M., Dormuth, C., & Soumerai, S.
(2002). Outcomes of reference pricing of rangiotensin-converting enzyme
inhibitors. The New England Journal of Medicine, 346(11), 822–829.
30
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
Schulz, R. M., Lingle, E. W., Chubon, S. J., & Coster-Schulz, M. A. (1995). Drug use
behavior under the constraints of a Medicaid prescription cap. Clinical
Therapeutics, 17, 330–340.
Shang, B. (2005). The cost and health effects of prescription drug coverage and
utilization in the Medicare population. Dissertation research for the Pardee Rand
Graduate School. Retrieved from
http://www.rand.org/pubs/rgs_dissertations/2005/RAND_RGSD197.pdf
Shea, D. G., Terza, J. V., Stuart, B., & Briesacher, B. (2007). Estimating the effects of
prescription drug coverage for Medicare beneficiaries. Health Services Research,
42(3, Part I), 933–949.
Sokol, M. C., McGuigan, K. A., Verbrugge, R. R., & Epstein, R. S. (2005). Impact of
medication adherence on hospitalization risk and health care cost. Medical Care,
43, 521–530.
Soumerai, S. (2004, March). Benefits and risks of increasing restrictions on access to
costly drugs in Medicaid. AARP Public Policy Institute. Retrieved from
http://www.aarp.org/research/health/drugs/aresearch-import-852.html
Soumerai, S., Avorn, J., Ross-Degnan, D., & Gortmaker, S. (1987). Payment restrictions
for prescription drugs under Medicaid. Effects on therapy, cost, and equity. The
New England Journal of Medicine, 317(9), 550–556.
Soumerai, S., McLaughlin, T., Ross-Degnan, D., Casteris, C. S., & Bollini, P. (1994).
Effects of a limit on Medicaid drug reimbursement on the use of psychotropic
agents and acute mental health services by patients with schizophrenia. The New
England Journal of Medicine, 331, 650–655.
Soumerai, S., Ross-Degnan, D., Avorn, T., & McLaughlin, T. (1991). Effects of
Medicaid drug payment limits on admissions to hospitals and nursing homes. The
New England Journal of Medicine, 325(15), 1072–1077.
Soumerai, S. B., Ross-Degnan, D., Fortress, E. E., & Abelson, J. (1993). A critical
analysis of studies of state drug reimbursement policies: Research in need of
discipline. The Milbank Quarterly, 71, 217–252.
Stafford, R. S., Davidson, S. M., Davidson, H., Miracle-McMahill, H., Crawford, S. L.,
& Blumenthal, D. (2003). Chronic care disease medication use in managed care
and indemnity insurance plans. Health Services Research, 38(2), 595–612.
Stephens, J. M., Botteman, M. F., & Hay, J. W. (2006). Economic impact of antidiabetic
medications and glycemic control on managed care organizations: A review of the
literature. Journal of Managed Care Pharmacy, 12(2), 130–142.
Stuart, B., & Coulson, N. E. (1993). Dynamic aspects of prescription drug use in an
elderly population. Health Services Research, 28, 237–264.
31
How Prescription Drug Use Affects Health Care Utilization and
Spending by Older Americans: A Review of the Literature
Stuart, B., Doshi, J., Briesacher, B., Wrobel, M., & Baysac, F. (2004). Impact of
prescription coverage on hospital and physician costs: A case study of Medicare
beneficiaries with chronic obstructive pulmonary disease. Clinical Therapeutics,
26(10), 1688–1699.
Stuart, B., & Grana, J. (1998). Ability to pay and the decision to medicate. Medical Care,
36(2), 202–211.
Stuart, B., & Zacker, C. (1999). Who bears the burden of Medicaid drug copayment
policies? Health Affairs, 18, 201–212.
Stukel, T. A., Lucas, F. L., & Wennberg, D. E. (2005). Long term outcomes of regional
variation in invasive versus medical management of Medicare patients with acute
myocardial infarction. Journal of the American Medical Association, 293, 1095–
1104.
Tamblyn, R., Laprise, J., Hanley, M., Abrahamowicz, M., et al. (2001). Adverse events
associated with prescription drug cost sharing among poor and elderly persons.
Journal of the American Medical Association, 285(4), 421–429.
Tseng, C., Brook, R., Keeler, E., & Mangione, C. (2003). Impact of an annual dollar limit
or “cap” on prescription drug benefits for Medicare patients. Journal of the
American Medical Association, 290(2), 222–227.
Tseng, C., Brook, R. H., Keeler, E., Steers, W. N., & Mangione, C. M. (2004). Cost
lowering strategies used by Medicare beneficiaries who exceed drug benefit caps
and have a gap in drug coverage. Journal of the American Medical Association,
292, 952–960.
Wallack, S. S., Martin, T., Thomas, C. P., & Ryan, A. (2007). Differences in prescription
drug use in HMOs and self-insured health plans. Medical Care Research and
Review, 64(1), 98–16.
Yang, Z., Gilleskie, D., & Norton, E. (2004, December). Prescription drugs, medical care,
and health outcomes: A model of health dynamics. NBER Working Paper No.
w10964. Retrieved from http://www.nber.org/papers/w10964
Yang, Z., Gilleskie, D., & Norton, E. (forthcoming). Health insurance, medical care, and
health outcomes: A model of health dynamics. Journal of Human Resources.
Yang, Z., & Norton, E. C. (2006). How much would a Medicare prescription drug benefit
cost? Offsets in Medicare Part A cost by increased drug use. Journal of
Pharmaceutical Finance Economics and Policy, 15(2), 97–119.
Zeber, J. E., Grazier, K., Valenstein, M., Blow, F., & Lantz, P. M. (2007). Effect of a
medication copayment increase in veterans with schizophrenia. American Journal
of Managed Care, 13(6), 335–346.
32
APPENDIX
33
Summary Table of Studies Reviewed
Reference
Population /
Location
Rx Drug or Class
Outcome Measures /
Statistical Method
Findings
Effects of Prescription Drug Coverage
Atherly, A.(2004). The effect of
prescription drug coverage on the
cost of care to Medicare beneficiaries
with asthma. Expert Review of
Pharmacoeconomics Outcomes and
Research, 4, 421-428.
U.S., Medicare,
N/A
elderly
34
Briesacher, B. Medicare beneficiaries and
the impact of gaining prescription drug
U.S., Medicare
coverage on inpatient and physician
spending. (2005) Health Services
Research, 40 [5 part 1], 1279-1296.
N/A
Gilman, et al. (2004). Final report:
Evaluation of Vermont’s pharmacy
assistance programs for low income
Vermont SPAP:
Medicare beneficiaries. RTI International,
N/A
elderly, poor
February 28, 2003. Retrieved from
http://www.cms.hhs.gov/Reports/
downloads/gilman_2003_5.pdf.
Outcome variable:
medical expeditures.
Linear probability
models; instrumental
variables; MCBS data
source.
Outcome variables: total
Rx drug expenditures,
Medicare hospital
expenditures, & Medicare
MD expeditures;
estimated by a series of
fixed-effects, 2-year
panel models (1995 and
2000 MCBS data),
multivariate models.
Outcome variables: Rx
drug spending; hospital
inpatient, hospital
outpatient, and MD
expenditures. Medicare
and Medicaid claims
data; difference-indifference regression
model.
For individual plans, there was unobserved adverse
selection of plans with drug benefits and unobserved
favorable selection of plans without. Individual
supplemental plans without Rx drugs increased
Medicare expenditures by $914 annually; those with
drugs increased Medicare expenditures by $491.
Employer policies also significantly increased
Medicare expenditures ($207 without drug coverage
and $447 with drug coverage), but the increase was
less than that associated with individual policies.
Persons gaining drug coverage at any point spent 20%
more on drugs than those with none; a subsample
gaining year-end coverage (and followed up longer)
spent 66% more after the first year with coverage,
compared to 8% more for those with no coverage. No
consistent evidence that drug coverage either increases
or reduces spending for hospital or MD services.
Findings from small sample, only suggestive.
Average adjusted Medicare spending for hospital
inpatient, outpatient, and MD services increased by
nearly $1,000 during the initial year of enrollment
compared to the spending trend amount nonenrolled
beneficiaries.
Summary Table of Studies Reviewed (continued)
Reference
Population /
Location
Rx Drug or Class
Outcome Measures /
Statistical Method
Findings
35
Effects of Prescription Drug Coverage (continued)
MCBS 1992 - 2000;
Suggests that change in drug coverage status is
Khan, et al. (2007, January). Prescription
Fixed-effects model,
random (exogenous) the use of Rx drugs (public
drug insurance and its effect on utilization
longitudinal panel,
estimates the causal
coverage by 14% and employer by 6%, with no
and health of the elderly. National Bureau
Medicare
N/A
discernable effect on hospital admissions or health.
of Economic Research, working paper no.
effects of Rx drug
12848. Retrieved from
coverage on drug use, use HMO drug coverage associated with improved
of other medical services, functional status.
http://www.nber.org/papers/w12848.
and health of the elderly.
Outcome variables:
total utilization & cost
for Medicare services,
NJ Medicare recipients used, on average, $238.50
Lingle, et al. (1987). The impact of
U.S., Medicare,
total PAA program drug less in inpatient hospital care under the Rx assistance
outpatient drug benefits on the use and
elderly, NJ &
N/A
costs, total PAA program (PAA) program than did their counterparts in eastern
costs of health care services for the
PA
administrative costs.
PA. who did not have access to the program. Two
elderly. Inquiry, 24, 203-211.
Quasi-experimental
points in time with limited causal evidence.
design; regression
analysis.
Outcome variables: costs,
Compared with current practice, first-dollar coverage
Rosen, et al. Cost-effectiveness of full
QALYs, life-years, and
of ACE inhibitors saved both lives and money (0.23
Medicare coverage of angiotensinincremental cost
QALYs gained and $1,606 saved per Medicare
U.S., Medicare, ACE inhibitors in
effectiveness. Cost
converting enzyme inhibitors for
beneficiary). Compared with the new Medicare drug
diabetics
diabetics
beneficiaries with diabetes. (2005).
effectiveness analysis,
benefit, first dollar coverage remained a dominant
Annals of Internal Medicine, 143, 89-99.
Markov simulation
strategy (0.15 QALYs gained, $922 saved).
modeling.
Shang, B. (2005).The cost and health
Outcome variables:
effects of prescription drug coverage and
Medigap Rx drug coverage significantly increases
Rx drug spending,
utilization in the Medicare population.
drug spending by $170 or 22%, and reduces Part A
U.S., Medicare,
Anti-hypertensives Medicare Part A
Dissertation research for the Pardee
spending by $277 - $348 or 10% - 13%. Results for
elderly
spending, Medicare
RAND Graduate School. Retrieved from
Medicare Part B spending were insignificant.
Part B spending
http://www.rand.org/pubs/rgs_dissertations/
2005/RAND_RGSD197.pdf.
Summary Table of Studies Reviewed (continued)
Reference
Population /
Location
Rx Drug or Class
Outcome Measures /
Statistical Method
Findings
36
Effects of Prescription Drug Coverage (continued)
Outcome variables:
Stuart, et al.(2004) Impact of prescription
After adjustment, drug coverage was associated with
drug spending,
coverage on hospital and physician costs:
61% higher spending on medications and 29% lower
hospitalizations, MD
A case study of Medicare beneficiaries
spending on MD services (both, P < 0.05). Hospital
U.S, Medicare,
COPD
services. MCBS data.
with chronic obstructive pulmonary
costs appeared slightly lower for those with drug
elderly
Logistic regression with
disease. Clinical Therapeutics, 26,
benefits, but the difference was not statistically
propensity scoring to
1688-99.
significant.
adjust for selection.
With Medicare drug coverage, Rx drug expenditures
Using MCBS 1992 in our sample would rise 12.2% - 17.5% over 5 years,
1998/ simulations of
Yang, et al. (2004) Prescription drugs,
while inpatient care and MD services would increase
future Medicare costs,
only slightly. Long-run survival probabilities increase,
medical care, and health outcomes:
using dynamic model.
U.S., Medicare,
leading to larger proportions of elderly survivors with
A model of elderly health dynamics.
N/A
Outcomes: Rx drug use,
National Bureau of Economic Research, elderly (MCBS)
functional limitations. These projections, however,
Medicare Part A
are smaller than those produced by extrapolating
Working paper no. 10964. Retrieved from
spending, Medicare Part
static models that fail to incorporate the dynamic
http://www.nber.org/papers/w10964.
B spending, functional
consequences of increased drug use on health
status, mortality.
outcomes and other Medicare-covered services use.
July 2007 update of above study incorporates 1992 Yang, et al. (Forthcoming) Health
2001 MCBS and differences in insurance coverage
insurance, medical care, and health
U.S., Medicare,
See above, 1992 - 2001
N/A
type, finding a 7% - 27% increase in Rx drug
outcomes: A model of health dynamics. elderly (MCBS)
update.
spending, with a similar effect on outcomes and health
Journal of Human Resources.
spending.
Short term effects of providing prescription drug
Using MCBS
coverage to Medicare beneficiaries: for every $100
1992 - 1998/ simulation spent on Rx drugs, hospital expenditures decrease by
Yang, Z., & Norton, E. C. (2006). How
estimating the effect of
an average of $4.40 the following year, or an offset
much would a Medicare prescription
total outpatient
drug benefit cost? Offsets in Medicare
U.S., Medicare,
rate of 4.4 percentage points. Offset rates differ by age
N/A
and comorbidity, with highest marginal benefit for
prescription drug
Part A cost by increased drug use.
elderly (MCBS)
spending in the previous those with chronic diseases. For the most severely ill
Journal of Pharmaceutical Finance
year to predict subsequent or those with functional limitations, higher Rx use
Economics and Policy,15(2), 97–119.
can not restore function but may prolong life, and
hospital care cost.
thus may not reduce Part A spending over long term.
Summary Table of Studies Reviewed (continued)
Reference
Population /
Location
Rx Drug or Class
Outcome Measures /
Statistical Method
Findings
37
Effects of Prescription Drug Cost Sharing and Other Benefit Design Features on Insured Populations
Bivariate and multivariate
An evidence-based formulary with expanded generic
Balkrishnan,et al. (2001). Effect of
analyses. Outcome
coverage was associated with a significant decrease of
prescription benefit changes on medical
variables: health care
27% in Rx costs, a 4% decrease in MD visits, and a
U.S., Medicare,
N/A
services, Rx costs, total
care utilization in a Medicare HMO
6% decrease in total costs. A formulary with coverage
HMO
population. American Journal of
annual costs. Panel data,
limits resulted in a 29% increase in Rx costs and a
Managed Care, 7, 1093-1100.
random-effects regression
38% increase in total costs for the HMO.
model.
Plan-level data.
MD visits and Rx drugs are highly price-sensitive
Regression-adjusted
effects of increased cost among the elderly in both types of plans regardless of
Chandra et al. (2007). Patient costRetired Public
sharing on MD services, policy changes. Higher cost sharing for Rx and MD
sharing, hospitalization offsets, and
visits in HMO increased hospital use, especially for
Employees,
pharmaceuticals and
the design of optimal health insurance
California
hosptializations/ separate those with chronic illness or high previous medical
for the elderly. National Bureau of
N/A
(CalPERS)
impact for Medicare;
spenders. Supplemental (Medigap) insurer received
Economic Research, Working paper
Medicare
continuously enrolled for savings due to cost sharing and decreased use while
no. 12972. Retrieved from
Medicare had net cost increase due to hospitalization
population
three years, 1/2000 http://www.nber.org/papers/12972.
9/2003. HMO and PPO increases. Results are confounded by simultaneous
design changes analyzed change in cost sharing for and decrease in MD visits.
separately.
Out-of-pocket spending increased for case group by
$16.60 between 2001 and 2002 (from $30.90 - $47.50)
Christian-Herman, et al. (2004).
Outcome variables: outbut only by $14.22 for control group (from $26.27 Effects of generic-only drug coverage
of-pocket costs, hospital
$40.49) (p <.0001). In 2002, the case group had an
in a Medicare HMO. Health Affairs,
admissions, generic Rxs.
U.S. Medicare
increase of 3.02 admissions/1,000 while the control
N/A
HMO, CA
web exclusive, w4-455-68. Retrieved
Case group/control group
group had a decrease of 0.22 admissions/1,000. Both
December 1, 2006 from
using administrative
the case and control groups increased their use of
http://www.healthaffairs.org.
HMO data.
generic medications between 2001 and 2002, but the
increase was larger in the case group.
Summary Table of Studies Reviewed (continued)
Reference
Population /
Location
Rx Drug or Class
Outcome Measures /
Statistical Method
Findings
38
Effects of Prescription Drug Cost Sharing and Other Benefit Design Features on Insured Populations (continued)
ACE inhibitors (anti-hypertensives): $10 increase
Outcome variables:
in copayment led to 2.6% decrease in adherence,
Cole, et al. (2006). Drug copayment
Congestive heart
risk of hospitalization,
U.S. insured
0.8% decrease in medical costs, 6.1% increase in
and adherence in chronic heart failure:
failure (CHF) - ACE
medical costs. Two-stage risk of hospitalization. Beta blockers: $10 increase
population,
Effect on cost and outcomes.
inhibitors and beta
regression, one-year
in copayment led to 1.8% decrease in adherence,
2002 - 2003
Pharmacotherapy, 26(8), 1157-1164.
blockers
post-implementation.
2.8% decrease in medical costs, and 8.7% increase
in risk of hospitalization.
In the coinsurance model, an increase in the
coinsurance rate from 20% to 75% increased noncompliance by 9.9%, and reduced full compliance
Outcome variables:
by 24.6%. For copayments, an increase in copay
full, partial, or nonDor, A. & Encinosa, W. (2004, August).
from $6 to $10 resulted in a 6.2% increase in nevercompliance with refill
Does cost sharing affect compliance?
U.S., Adults age
compliers and a 9% reduction in full compliers.
for diabetic medication,
Using available aggregate estimates of the cost of
The case of prescription drugs. National 18+, diabetics,
Diabetes
using MarketScan drug
Bureau of Economic Research, Working commercial
diabetic complications, a $6 - $10 increase in
claims data and Employer
paper no.10738. Retrieved from
insurance
copayment would have the direct effect of reducing
Eligibility data. Crossnational drug spending for diabetes by $125 million.
http://www.nber.org/papers/w10738.
sectional data, ordered
However, the increase in non-compliance rates is
logit model.
expected to increase the rate of diabetic complications,
resulting in an additional $360 million in treatment
costs (net cost = $235 million annually).
The 610% increase in Medicaid spending on antipsychotic drugs during the study period caused by
Duggan, M. (2005). Do the new
Outcome variables: cost the shift to three new treatments has not reduced
U.S., CA,
spending on other types of medical care. Because of
prescription drugs pay for themselves?
of antipsychotic drugs,
Anti-psychotics in
The case of second-generation
total inpatient/outpatient data limitations, the findings for health outcomes are
Medicaid,
schizophrenia
age 18+
speculative but suggest that the new medications
antipsychotics. Journal of Health
Medicaid spending.
Economics, 24, 1-31.
Logistic regression used. have increased the prevalence of diabetes while
reducing the prevalence of extrapyramidal symptoms
among the mentally ill.
Summary Table of Studies Reviewed (continued)
Reference
Population /
Location
Rx Drug or Class
Outcome Measures /
Statistical Method
Findings
39
Effects of Prescription Drug Cost Sharing and Other Benefit Design Features on Insured Populations (continued)
Outcome variables: total
drug cost; net insurer cost Relative to the comparison group (n = 4132), the
(drug cost - copayment); intervention group (n = 3577) showed less growth in
net cost and lower use of third-tier (non-formulary)
number of Rx claims;
Fairman, K.A. et al. (2003).
medications (P < 0.001 and P < 0.01, respectively).
number of MD visits,
Retrospective, long-term follow-up
U.S.,
All, including
inpatient hospitalizations, The intervention and comparison groups did not differ
study of the effect of a three-tier
commercially
estrogens, antiand ER visits; and rates significantly in numbers of office visits, ER visits, or
prescription drug copayment system
inpatient hospitalizations. Rx continuation rates were
insured PPO
hypertensives, and
of continuing with
on pharmaceutical and other medical
lower for the intervention than for the comparison
members
anti-hyperlipidemics chronic medication
utilization and costs. Clinical
group at 6 months for oral contraceptives (P < 0.05),
therapy.
Therapeutics, 25, 3147-3161.
but chronic Rx continuation rates did not differ
Quasi experimental,
pre-post with comparison significantly at any other time or for estrogens,
group; Logistic and Cox anti-hypertensives, or anti-hyperlipidemics.
regression.
$1 increase in copay leads to $24 decrease in drug
Outcome variables:
Gaynor et al. (2006, December). Is drug
spending after 1 year and an additional $9 decrease
drug use, drug spending,
Insured U.S.
after 2 years. Savings on Rx drugs are partially offset
coverage a free lunch? Cross-price
outpatient use, and
working
elasticities and the design of prescription
by increases in outpatient spending. Total medical
spending lagged with
population
drug benefits. National Bureau of
N/A
spending falls by $20.88 in the first year and $21.23
two years' follow-up,
to age 64
in the second. Long-run reductions in drug spending
Economic Research, Working paper no.
1997 - 2003. Accounted
(MarketScan
12758. Retrieved from
partially offset by consequent increases in outpatient
for dynamic demand for
Data)
spending; no association with inpatient spending.
http://www.nber.org/papers/w12758.
drugs.
Study underscores importance of lagged response.
Summary Table of Studies Reviewed (continued)
Reference
Population /
Location
Rx Drug or Class
Outcome Measures /
Statistical Method
Findings
40
Effects of Prescription Drug Cost Sharing and Other Benefit Design Features on Insured Populations (continued)
Doubling copayments was associated with reductions
in use of 8 Rx classes. The largest decreases occurred
for NSAIDs (45%) and anti-histamines (44%).
Anti-inflamatories
Reductions in overall days supplied of antiOutcome variable:
(NSAIDs),
hyperlipidemics (34%), anti-ulcerants (33%),
relative change in
anti-histamines,
Goldman et al. (2004). Pharmacy benefits U.S., adults age
anti-asthmatics (32%), anti-hypertensives (26%),
drug days supplied.
anti-hypertensives,
anti-depressants (26%), and anti-diabetics (25%)
Retrospective study,
and the use of drugs by the chronically ill. 18 - 64,
anti-asthmatics,
were also observed. Among patients diagnosed with
pharmacy claims data.
Journal of the American Medical
commercial
anti-ulcerant,
Association, 291(19), 2344-2350.
insurance
a chronic illness and receiving ongoing care, use
2-part model: probit
anti-diabetic,
was less responsive to copayment changes. Use of
plus generalized linear
anti-hyperlipidemic,
anti-depressants declined by 8%; use of antimodel regression.
anti-depressant
hypertensives decreased by 10%. Larger reductions
were observed for NSAIDs (27%), anti-histamines
(31%), and anti-diabetes drugs (23%).
Outcome variables:
medication possession
Goldman, et al. (2006).Varying pharmacy
Varying copayments for cholesterol-lowering therapy
ratio, ED visits, number
U.S., adults age
benefits with clinical status: The case of
by therapeutic need would reduce hospitalizations and
20+, commercial Anti-hyperlipidemic of hospitalizations.
cholesterol-lowering therapy. American
ER use—with total savings of more than $1 billion
insurance
Ordered logit, pharmacy,
Journal of Managed Care, 12(1), 21-28.
annually.
and medical claims from
88 health plans.
Outcome variables:
medical services.
Five therapeutic
Longitudinal generalized
Hazlet, T.K. & Blough, D.K. (2001).
Pharmacare
For Rxs, MD office visits, lab, ER visits,
classes of drugs,
linear model, controlling
Health services utilization with reference beneficiaries,
including
hospitalizations, and hospital length of stay, no
drug pricing of histamine 2 receptor
British
for age, sex, and Rxs in
worsening of health outcomes associated with
histamine-2 receptor
unique drug classes; trend
antagonists in British Columbia elderly. Columbia,
implementing the reference pricing policy.
antagonists
lines in each of these time
Medical Care, 40, 640-649.
Canada
(H-2 RAs)
series were compared for
3 periods.
Summary Table of Studies Reviewed (continued)
Reference
Population /
Location
Rx Drug or Class
Outcome Measures /
Statistical Method
Findings
41
Effects of Prescription Drug Cost Sharing and Other Benefit Design Features on Insured Populations (continued)
Outcome variables: Rx
costs, total medical costs, Subjects whose benefits were capped had pharmacy
ED visit rate, non-elective costs for drugs that were 31% lower than those of
hospitalization rate, death subjects whose benefits were not capped but had total
Hsu, et al. (2006). Unintended
rate. Prospective cohort medical costs that were only 1% lower. Subjects
consequences of caps on Medicare drug U.S., Medicare,
Multiple
design. Medicare+Choice whose benefits were capped had higher relative rates
benefits. The New England Journal of
HMO
claims data, 2-part model of visits to the ER (relative rate, 1.09, non-elective
Medicine, 354 (22), 2349-2359.
with logistic regression hospitalizations (relative rate, 1.13), and death
for probability of any cost (relative rate, 1.22); difference, 0.68 per 100 personand linear regression of years (0.30 to 1.07).
costs.
Increased copayment of $1 - $3, from $3 - $5 and
Johnson, et al. (1997a). The effect of
Outcome variables: Rx
from 50% per dispensing to 70% per dispensing with
increased prescription drug cost-sharing
expenses, copayments,
a maximum payment per Rx resulted in lower annual
on medical care utilization and expenses
medical care expenses;
Medicare, HMO Multiple
per capita Rx drug use and expenses. No consistent
of elderly health maintenance
ANCOVA analysis on 2
effect on medical care use (MD visits, ER visits,
organization members. Medical Care, 35,
years of Kaiserhome healthcare visits, hospitalizations) or total
1119-1131.
Permanente claims data.
medical care expenses.
Outcome variables:
See above for design. Relative exposure, annual days
Johnson et al. (1997b). The impact of
annual change in
of use, and Rx drug costs for drugs used in selfdispensing, Rx expenses, limiting conditions and in progressive chronic
increasing patient prescription drug cost
sharing on therapeutic classes of drugs
copayments, medical care conditions were not affected consistently. Health
Medicare, HMO Multiple
received and on the health status of
expenses; ANCOVA
status may have been adversely affected. Larger
analysis on 2 years of
increases in copayments appeared to generate more
elderly HMO members. Health Services
Research, 32(1), 103-122.
Kaiser-Permanente
changes. Small changes in copayments did not appear
claims data.
to affect outcomes substantially.
Summary Table of Studies Reviewed (continued)
Reference
Population /
Location
Rx Drug or Class
Outcome Measures /
Statistical Method
Findings
42
Effects of Prescription Drug Cost Sharing and Other Benefit Design Features on Insured Populations (continued)
Increases were observed in the number of Rxs, MD
visits, and outpatient visits, while the number of
inpatient hospital admissions declined. Similarly,
Outcome variables:
expenditures increased for all services except the
number of Rxs, MD
Kozma, et al. (1990). Expanding
inpatient hospital. The proportion of variance
visits, outpatient visits,
Medicaid drug formulary coverage.
explained by the formulary change was small in all
and inpatient hospital
U.S., Medicaid N/A
service areas, but would be of practical significance
Effects on utilization of related services.
admissions. Repeated
Medical Care, 28, 963-977.
because of the large number of Medicaid recipients
measures design;
affected. From a theoretical perspective, a reduction
multivariate ANOVA.
in inpatient hospital use and expenditures following
the elimination of drug formulary restrictions is
particularly noteworthy.
Estimated own-price
Li et al. (2007). The impact of cost
8,107 Canadian
Price elasticity for Rx drugs was –0.11 (plan A) and
and cross-price
sharing of prescription drug expenditures seniors with
–0.20 (plan A1); cross-price elasticity for MD visits
Rheumatoid arthritis elasticities for Rx drugs
on health care utilization by the elderly: rheumatoid
was positive and significant: 0.04 (plan A) and 0.06
(RA)
and MD visits, using
Own- and cross-price elasticities.
arthritis,
(Plan A1) one year following change. No analysis of
instrumental variables to
Health Policy, 82, 340-347.
2001 - 2002
cost offsets, total savings, or hospitalization.
adjust for RA severity
People consuming newer drugs were significantly
Outcome variables:
less likely to die by the end of the survey and were
mortality, morbidity
significantly less likely to experience work-loss days
Lichtenberg, F. (2001). Are the benefits
indicators, and spending
than were people consuming older drugs. Use of newer
of newer drugs worth their cost?
on non-drug medical
U.S., all ages
New drugs
Evidence from the 1996 MEPS.
drugs tends to lower all types of non-drug medical
events. MEPS data.
spending, resulting in a substantial net reduction in
Health Affairs, 20(5), 241-251.
Pooled data, mulitvariate
the total cost of treating a given condition. Does not
analysis.
adjust for other unrelated system changes over time.
Summary Table of Studies Reviewed (continued)
Reference
Population /
Location
Rx Drug or Class
Outcome Measures /
Statistical Method
Findings
43
Effects of Prescription Drug Cost Sharing and Other Benefit Design Features on Insured Populations (continued)
Outcome variables: total
Rx claims and costs, net
Relative to the comparison group, the intervention
costs (total - copay),
group experienced lower Rx use and expenditures
medication continuation,
Motheral, B. & Fairman, K. (2001).
and reduced net costs. Rx continuation rates were
MD visits, and inpatient
Effect of a three-tier prescription copay U.S.,
lower at 6 and 11 months in one of four chronic
commercial
N/A
and ER use. Quasion pharmaceutical and other medical
therapy classes examined; however, discontinuation
experimental with preutilization. Medical Care, 39(12),
insurance, PPO
could not be clearly linked to tier-three medication
post comparison group.
1293-1304.
use. No significant differences in MD office visits,
Mann-Whitney test.
inpatient, or ER use rates were found.
Logistic, linear and Cox
regression analysis.
Outcome variables:
There was a statistically significant increase in the
cardiovascular-related
number of outpatient hospital visits and MD visits for
inpatient and outpatient the test group compared to the control group in the
Murawski, M.M. & Abdelgawad, T.
hospital visits and
first 6 months after PDL implementation. There was
(2005). Exploration of the impact of
procedures, and MD
a positive but statistically insignificant increase in
preferred drug lists (PDL) on hospital
Cardiovascular
visits and procedures.
the number of inpatient hospital visits. In the second
U.S., Medicaid
and physician visits and the costs to
patients
A regression-based,
6 months, MD visits for the test group increased but
Medicaid. American Journal of
difference-in-differences the increase was not statistically significant. As a
Managed Care, 11, sp35-sp42.
retrospective analysis
result, estimated average Medicaid reimbursement
using patient-level
costs for cardiovascular patients in the state increased
claims data.
during that year.
Summary Table of Studies Reviewed (continued)
Reference
Population /
Location
Rx Drug or Class
Outcome Measures /
Statistical Method
Findings
44
Effects of Prescription Drug Cost Sharing and Other Benefit Design Features on Insured Populations
Outcome variables: Rx
The proportion of patients who received Rxs for βrates for essential cardiac blockers, ACE inhibitors, and lipid-lowering drugs
medications for patients increased over time and, more specifically, did not
Pilote, et al. (2002). The effects of costadmitted before and after appear to decline with the change in drug policy. In
sharing on essential drug prescriptions,
AMI treatment,
the policy reform, and
addition, the policy reform did not appear to affect
utilization of medical care and outcomes
including β-blockers, cardiac procedures,
persistence of drug therapy (the proportion of time for
after acute myocardial infarction (AMI) Canada, age 65+ lipid lowering agents, readmissions for cardiac- which patients were covered by Rxs over the year after
in elderly patients. Journal of the
ACE inhibitors, and related complications,
discharge). There was no within-class shift from more
Canadian Medical Association, 167,
aspirin
outpatient visits to MDs to less expensive drugs. Use of cardiac procedures
246-252.
and ERs, and mortality
increased over time, but this increase was unrelated
rate. Claims data, logistic to the date of the policy reform. Finally, rates of
regressions with
readmission for complications, visits to individual
Bonferroni correction.
MDs and ERs, and mortality rate were unchanged.
Medicare HMO enrollees with two pharmaceutical
capitation MD medical groups (PMGs) had 10%
higher mean total health care costs and 20% higher Rx
Outcome variables: Rx
costs, and a greater percentage of these patients had Rx
Popovian, et al. (1999). The impact of
expeditures; total health
expenditures than did patients in the non-Rx capitation
pharmaceutical capitation to primary
care costs. Retrospective
PMG. The Rx capitation patients were also 70% more
analysis of claims data;
medical groups on the health care
U.S., Medicare,
N/A
likely to incur higher total health care expenditures
expenditures of Medicare HMO
2-part model - logistic
HMO
and 69% more likely to incur higher Rx expenditures.
enrollees. Journal of Managed Care
regression of use of
When controlling for age, gender, and severity of
benefits; linear regression
Pharmacy, 5, 414-419 (not cited in text).
illness, Rx capitation patients had 14% higher total
of amount of costs.
health costs than non-capitated patients (an additional
$376 per patient per year) and 29% higher Rx costs
($110 per patient per year).
Summary Table of Studies Reviewed (continued)
Reference
Population /
Location
Rx Drug or Class
Outcome Measures /
Statistical Method
Findings
45
Effects of Prescription Drug Cost Sharing and Other Benefit Design Features on Insured Populations
The start of reference pricing was followed by a
significant reduction in high-priced CCBs (-150 doses
per month per 10,000 elderly persons), with a
corresponding increase in fully covered CCBs (+116).
Outcome variables: Rx
Overall, anti-hypertensive use did not decline (P =
use, Rx expenditures, MD
.46). Low-income status was a risk factor for
visits, hospitalizations,
Schneeweiss, S., et al. (2003). Clinical
discontinuing treatment; however, this was already
long-term care, and net
and economic consequences of reference Canada, age
Calcium channel
observed to a similar extent 12 months before the start
savings. Quasi65+, British
pricing for dihydropyridine calcium
blockers (CCBs)
of reference pricing. In the overall study cohort, there
experimental longitudinal
channel blockers. Clin Pharmacol Ther, Colombia
was no increase in rates of MD visits, hospitalizations,
study using Pharmacare
74(4), 388-400.
or long-term care admissions. However, the 9% of
claims data; logistic &
patients who switched medications showed an 18%
Poisson regression.
increase (95% CI, 8% to 28%) in MD visits and an
increase in costs of MD visits per patient ($13
Canadian; 95% CI, $3 - $24) compared with nonswitchers during the transition but not afterward.
During the first year after implementation of reference
pricing, savings for continuous users were $6 million
Outcome variable: net
Canadian. Savings for new users were $200,000
health plan savings.
Canadian. Approximately 5/6 of savings were achieved
Modeling approach:
Schneeweiss, et al. (2004). Net health
by use changes and 1/6 by cost shifting to patients.
spending changes were
There were no savings through drug price changes.
plan savings from reference pricing for
Canada, age
ACE inhibitors (anti- identified, quantified
Administering reference pricing cost $420,000
angiotensin-converting enzyme (ACE)
65+, British
through claims analysis,
hypertensive)
inhibitors in elderly British Columbia
Canadian. Overall net savings were estimated to be
Columbia
and arranged into 4 major
$5.8 million Canadian during the first year. The
residents. Medical Care, 42, 653-660.
components similar to
magnitude of these savings is equal to 6% of all
Mason’s work on policy
cardiovascular drug expenditures for seniors in the
cost effectiveness.
province. After 10 years, approximately 50% of
savings will be achieved by new users.
Summary Table of Studies Reviewed (continued)
Reference
Population /
Location
Rx Drug or Class
Outcome Measures /
Statistical Method
Findings
46
Effects of Prescription Drug Cost Sharing and Other Benefit Design Features on Insured Populations (continued)
Reference pricing for ACE inhibitors was not
Outcome variables:
associated with changes in the rates of MD visits,
drug use, drug
hospitalizations, long-term care, or mortality. The
probability of stopping anti-hypertensive therapy
expenditures, MD
Schneeweiss, et al. (2002). Outcomes
visits, hospitalizations,
decreased compared to the probability before the
of reference pricing for angiotensinCanada, age
ACE inhibitors (anti- long-term care, and
change in policy. Eighteen percent of patients who had
converting-enzyme (ACE) inhibitors.
65+, British
been prescribed ACE inhibitors subject to cost sharing
et savings. Quasihypertensive)
The New England Journal of Medicine, Columbia
experimental longitudinal switched to lower-priced alternatives. Compared to
346(11), 822-829.
study using Pharmacare patients who did not switch, those who did had a
moderate transitory increase in the rates of MD visits
claims data; logistic &
Poisson regression.
and hospital admissions through the ER during the 2
months after switching, but not subsequently.
A 35% decline in use of study drugs after the cap was
applied was associated with an increase in rates of
Soumerai, et al. (1991). Effects of
nursing home admissions. No changes were seen in
Survival (defined as
Medicaid drug payment limits on
remaining in the
the comparison cohort (RR = 1.8; 95% CI, 1.2 to 2.6).
admissions to hospitals and nursing
U.S., Medicaid
Limiting reimbursement for effective drugs puts frail,
community) and
homes. The New England Journal of
low-income, elderly patients at increased risk of
time-series analyses
Medicine, 325(15), 1072-1077.
institutionalization in nursing homes and may increase
Medicaid costs.
Soumerai, et al. (1987). Payment
Among 10,734 continuously enrolled patients, the
restrictions for prescription drugs under
limit of 3 Rxs per month caused a sudden, sustained
Interrupted time-series
U.S., Medicaid Multiple Rxs
Medicaid. Effects on therapy, cost, and
drop of 30% in the number of Rxs filled (from 1.10
analysis
equity. The New England Journal of
to 0.77 Rxs per patient per month); no change was
Medicine, 317(9), 550-556.
observed in the comparison state.
Summary Table of Studies Reviewed (continued)
Reference
Population /
Location
Rx Drug or Class
Outcome Measures /
Statistical Method
Findings
47
Effects of Prescription Drug Cost Sharing and Other Benefit Design Features on Insured Populations (continued)
Reimbursement limits were associated with a 15% to
49% reduction in use of mental health drugs, increases
of one to two visits per patient per month to CMHCs,
Soumerai, et al. (1994). Effects of a limit
use of emergency mental health services and partial
on Medicaid drug reimbursement on the
hospitalization (1.2 to 1.4 episodes per patient per
Psychotropic Rx and
use of psychotropic agents and acute
Interrupted time-series
month) but no change in frequency of hospital
U.S., Medicaid MH services for
mental health services by patients with
admission. After the cap was discontinued, use of
analysis
schizophrenia
schizophrenia. The New England Journal
medications and most mental health services reverted
of Medicine, 331, 650-655.
to baseline levels. The estimated average increase in
mental health care costs per patient during the cap
($1,530) exceeded the savings in drug costs to
Medicaid by a factor of 17.
After cost-sharing was introduced, use of essential
Outcome variables: Mean
drugs decreased by 9.12% in elderly persons and by
daily number of essential
14.42% in welfare recipients; use of less essential
and less essential drugs
drugs decreased by 15.14% and 22.39%, respectively.
Tamblyn et al. (2001). Adverse events
per month, ER visits,
The rate (per 10,000 person-months) of serious adverse
associated with prescription drug costhospitalizations, nursing
U.S. & Canada,
events associated with reductions in use of essential
sharing among poor and elderly persons.
Multiple Rxs
home admissions, and
elderly, poor
drugs increased from 5.8 in the pre-policy control
Journal of the American Medical
mortality. Interrupted
cohort to 12.6 in the post-policy cohort in elderly
Association, 285(4), 421-429.
time-series analysis;
persons, and from 14.7 to 27.6 in welfare recipients.
random effects and Cox
Increases in ER visits correlated with reductions in the
hazard modeling.
use of essential drugs.
Pre-post longitudinal
Rx copayment increase from $2 to $7 led to 9% fewer
Data from VA
design with controls.
Rxs overall, and 25% fewer psychotropic Rx refills.
national psychosis
Regression-adjusted
No effect on outpatient visits. Psychiatric hospital
Zeber et al. (2007). Effect of a medication
admissions 5% more likely post-intervention,
registry, 2000 - 2003. impact of increase Rx
copayment increase in veterans with
Veterans with
U.S. Veterans
cost sharing on Rx use
particularly 10 - 20 months after policy change.
schizophrenia. American Journal of
schizophrenia
and cost and health
Estimated $14.7 million net savings for VA from this
Managed Care, 13(6), 335-346.
services use, with 40,000 population by 20 months post copay increase.
receiving medical
care through VA
cases; 20-month follow- However, higher inpatient utilization is troubling in
up.
this high risk population.
AARP Public Policy Institute
Research Report
Public Policy Institute
601 E Street, NW | Washington, DC 20049
202-434-3840 PH | 202-434-6480 F | www.aarp.org/ppi