The Effects of Prescription Drug Cost Sharing: A Review of the Evidence

November 1, 2005
Teresa B. Gibson, PhD
Teresa B. Gibson, PhD

,
Ronald J. Ozminkowski, PhD
Ronald J. Ozminkowski, PhD

,
Ron Z. Goetzel, PhD
Ron Z. Goetzel, PhD

Volume 11, Issue 11

Objectives: To determine whether patients respond toincreased cost sharing by substituting less expensive alternativesfor medications with higher levels of copayments or coinsurance,and to examine the body of evidence on the relationships betweencost sharing and use of essential or maintenance medications,health outcomes, process-of-care measures (such as medicationadherence and discontinuation), and costs.

Study Design: Literature review.

Methods: Healthcare reference databases and key journalswere searched to identify peer-reviewed empirical studies thatexamined the effects of variation in the amount of prescription drugcopayments or coinsurance on healthcare utilization patterns.Thirty studies met our search criteria.

Results: Higher levels of prescription drug cost sharing generallyproduce intended effects, namely, decreasing the consumptionof prescription drugs and steering patients away from nonpreferredto preferred brand-name drugs. However, patients do not alwaysswitch to generic drugs. Although not consistently reported, themost troublesome effects associated with higher levels of cost sharingare treatment disruptions (such as lower levels of treatmentadherence, continuation, and initiation) for chronically ill patients.At times, higher levels of cost sharing can affect the use of essentialmedications and outcomes of care.

Conclusions: Cost sharing reduces the consumption of prescriptiondrugs but may have unintended effects on the process andoutcomes of therapy. Further research is warranted in this area. Thecentral question for health plan managers and policy makers iswhether we will continue to use cost sharing as is or make modificationsto reduce unintended effects.

(Am J Manag Care. 2005;11:730-740)

Prescription drug expenditures are one of thefastest growing components of national healthexpenditures.1 To control prescription drug costs,health plans and employers have increased prescriptiondrug cost-sharing amounts for patients.2 Copaymentsfor enrollees in employer-sponsored plans have risenconsiderably. From 2001 to 2004, the mean copaymentsfor generic drugs increased 42.9% (from $7 to $10),while copayments for preferred brand-name drugsrose 61.5% (from $13 to $21) and copayments for nonpreferredbrand-name drugs increased 94.1% (from$17 to $33).2

In addition to increasing the cost-sharing amounts,health plans continue to move away from 1-tier plansthat charge the same amount for all types of drugs andaway from 2-tier plans that charge a lower cost-sharingamount for generic drugs and a higher cost-sharingamount for brand-name drugs.2 As a result, 3-tier plansthat assess a third, higher amount for nonpreferredbrand-name drugs are now the dominant type of prescriptiondrug benefit; 3-tier plans in 2004 applied toalmost two thirds of workers.2 Although less common(representing 3% of workers), some health plans areintroducing plans that assign an even higher fourth tierto cover lifestyle or very expensive medications.2-4In this review, we synthesize and summarize thestate of knowledge about the effects of increased prescriptiondrug cost sharing on use, expenditures, andoutcomes. First, we address the following question: Dopatients respond to increased cost sharing by substitutingless expensive medications or delivery methods (eg,mail order) for medications with higher levels of copaymentsor coinsurance?

Second, concerns have been expressed about theadverse effects of cost sharing on health outcomes andthe process of care.5,6 In light of these concerns, weextend previous reviews of the literature and examinethe growing body of evidence on the relationshipsbetween cost sharing and the use of essential or maintenancemedications, health outcomes, process-of-caremeasures (such as medication adherence and discontinuation),and costs.

Previous reviews of the prescription drug cost-sharingliterature summarized evidence related to theeffects of changes in cost sharing on prescription druguse and expenditures, but the results of these studiesare dated7-9 or have a focus that is different from thatof this review, such as the effect of cost sharing on seniors10,11or on vulnerable populations.12,13 Other reviewsevaluated the effects of cost sharing within the broadercontext of pharmacy benefit management tools10,14 orare specific to multitiered formularies.15 This review isnot intended to replicate the results from prior reviewsbut rather to provide new insights, including an assessmentof the effects of changes in prescription drugcopayments and coinsurance on specific measures ofuse, outcomes, and expenditures.

SELECTING STUDIES FOR REVIEW

prescription drugs,

copayments

cost sharing

multitier formulary

multitiered

formulary.

direct costs, indirect costs, adherence,

compliance, income, socioeconomic status,

Medicaid.

Studies were selected for this review on the basis ofseveral MEDLINE searches, covering 1974 through April2005. The first set of searches was based on the phrasewhich was paired with or or or These terms were searched in various combinationswith andStudies were excluded that were not inEnglish or were not based on study populations fromthe United States or Canada. The published studies thatwere identified were supplemented with studies fromour files and with other studies that were identified inreference lists of selected publications. We selectedempirical studies using claims-based data sources inwhich cross-sectional or longitudinal variation in theamount of copayments or coinsurance occurred, permittingan examination of the relationships betweenvariation in these types of cost sharing and use patterns.

We identified 30 studies that met our criteria. Morethan one third (11/30) of the studies reviewed hereinwere not addressed in prior reviews. Most of the unreviewedstudies postdated prior reviews or did not fit thecriteria for more recent reviews.10-12,15 Each study wasassessed to determine whether the findings revealedthat cost sharing had a significant (nonzero) effect, noeffect, or mixed effects (some nonzero and some not significantlydifferent from zero) on measures of prescriptiondrug use.

A DEFINITION OF COST SHARING ANDTHE DEMAND FOR MEDICAL CARE

Cost Sharing

Cost sharing is defined as the direct charge to apatient at the time a prescription is filled. Cost sharingrepresents the price of the prescription drug to theinsured patient, while insurance covers the remainderof the cost. This review addresses the effects of the 2most common forms of prescription drug cost sharing,namely, copayments, a flat fee assessed per prescription(eg, $10), and coinsurance, a fixed fraction of each dollarof cost (eg, 20%).2

Economic Framework

Economic theory states that, when a patient isassessed the full price of a prescription drug and hasenough information to assess the drug's benefits andadverse effects, he or she will consume an optimalamount of the drug, given his or her preferences andincome constraints. The theory assumes that rationalpatients will weigh the costs and benefits of drugs vsother methods of producing health and will consumecombinations of these that maximize their health, subjectto their income constraints.

Having prescription drug insurance motivatespatients to consume more drugs than they would normallyconsume, because the price to the patients islower than the full price.16,17 Raising the price of thedrug via higher levels of cost sharing is expected to havethe following economic effects, although this list is notexhaustive:

Changes in Consumption.

Higher prices are expectedto move patients up the demand curve and closer tothe economically optimal amount, resulting in a reductionin consumption.

Substitution.

Patients are likely to search for lessexpensive substitutes as the prices of prescription drugsrise. Therefore, if they discover a good substitute,patients are likely to consume smaller quantities of prescriptiondrugs and larger quantities of the substitute.

Value.

A price increase would decrease the likelihoodthat drugs of low value, for which the cost exceedsthe benefit, would be used.16,17 Conversely, patientswould be price insensitive for high-value drugs, such asthose that are life sustaining, and would be expected tocontinue to fill prescriptions. However, this assumesthat consumers have adequate information to evaluatethe benefits and costs of drugs, which has not beenestablished.18

Previous studies8,12,17,19 demonstrated the first effect,that higher levels of cost sharing result in reductions inprescription drug use. These studies found that, similarto most healthcare services, the demand for prescriptiondrugs is insensitive to price changes. Most estimatesof price elasticity suggest that a 10% increase inprice, for example, would decrease use by less than that,ranging from 1% to 4%. However, the price elasticity ofdifferent medication classes can vary widely.20,21

We devote the next sections of this review to ananalysis of several issues. First, we determine whetherpatients use larger quantities of drug substitutes as aresult of cost sharing. Next, we evaluate whether patientsdistinguish between high-and low-value drugs by examining studies that relate to the use of various classificationsof prescription drugs (eg, essential and nonessentialmedications). On a related note, we determine ifpatients are making well-informed decisions by assessingwhether higher levels of cost sharing are associated withdeclines in patient health, patient care, and medical outcomes.Finally, we summarize the effects of higher levelsof cost sharing on healthcare expenditures.

DRUG SUBSTITUTION

During the past 2 decades, health plans introduceddifferent formularies that provided economic incentives,via higher patient cost-sharing amounts, forpatients to use lower-cost prescription drugs or nondrugsubstitutes. Many studies addressed herein estimatedthe effects of incentive-based formularies on the use ofsubstitutes, such as preferred brand-name drugs,generic drugs, and over-the-counter drugs. Severalstudies also estimated the effects of incentive-basedformularies on the use of mail-order pharmacies, whichare typically marketed as being less expensive thanretail pharmacies.

We review the evidence in these areas to determinewhether the use patterns of potential drug substitutesincreased as cost sharing rose. We conclude that patientsappear to be responding to some, but not all, financialcost-sharing incentives to switch to close drug substitutes(Table 1). To date, we found no studies thataddress the use of nonmedical substitutes, such as physicalexercise, in response to higher levels of cost sharing.

Nonpreferred vs Preferred Brand-name Drugs

In accord with financial incentives, all of the studies22-27 reviewed showed that adding a third tier for nonpreferredbrand-name drugs resulted in a decrease inthe use of these drugs and an increase in the use of preferredbrand-name drugs, although the classifications ofpreferred and nonpreferred drugs differed among thestudies. The extent of substitution varies by medicationclass25 and can be significant. For example, in a study26of enrollees in independent practice health plans, themean net increase in the use of preferred brand-namedrugs for plans with differential copayments for preferredand nonpreferred brand-name drugs was 13.3%for angiotensin-converting enzyme (ACE) inhibitors,8.9% for proton pump inhibitors, and 6% for statins.

Generic Substitution

In contrast, we found little evidence of generic substitutionin plans introducing or increasing a generic vsbrand cost-sharing differential. Few studies20,28,29 reportedan increase in the number of generic drugs dispensedas a result of higher generic vs brand price differentials.Conversely, in a recent study, Christian-Herman andcolleagues30 evaluated the effects of a switch to a generic-only benefit from a generic vs brand benefit in aMedicare health maintenance organization (HMO) andreported a 20% rise in generic prescriptions per person.However, the effect was attenuated considering the concurrent13.7% rise in generic prescriptions in a comparisongroup of Medicare HMO enrollees who retained thegeneric vs brand coverage. Because health plans havelong attempted to steer patients to less expensive generic drugs via differential cost sharing, these findings aresomewhat surprising and bear further investigation.

Several studies reported the effects of higher levels ofcost sharing on the generic fill rate, which is the percentageof drugs dispensed as generic. Two studies19,24reported no effect on the generic fill rate. Five studies25,27,29-31 reported increases in the generic fills.Caution should be exercised when interpreting the findingsof these studies. An increase in the generic fill rateis not a clear indication of generic substitution becausethe generic fill rate can be affected by a change in thenumber of generic drugs dispensed (in the numeratorand the denominator), by a change in the number ofbrand-name drugs dispensed (in the denominator), orby a combination of both.

Two of the studies also assessed disaggregated genericvs brand use. The first, a study29 of commerciallyinsured adults in 2 health plans that raised 2-tier copaymentsand increased the generic vs brand differential,reported that the generic fill rate rose as a result of adecrease in brand-name use, although generic useremained unchanged. In the second study,30 after introductionof a generic-only benefit in a Medicare HMO,the authors found evidence of a large increase (15%) inthe generic percentage dispensed, although much ofthis resulted from a reduction in brand-name use. Theexact size of the effect on the generic fill rate isunknown, as patients had little incentive to fill brandnamedrugs within the plan and the authors were unableto track out-of-plan purchases of brand-name drugs.

It is possible that the copayment differential betweengeneric and brand-name drugs has been too small, aslarger generic vs brand differentials appear to motivatepatients to consume more generic drugs.27,30,31Another possible reason for the observed lack of movementtoward generic drugs may be that patients perceivedifferences between the quality of generic drugsand brand-name drugs that motivate them to avoidgeneric equivalents. These perceptions may be real.Lichtenberg32 found that mortality rates, work loss days,and nondrug medication costs were significantly lowerfor patients taking newer drugs (which are more likelyto be brand name) than older drugs (which are morelikely to be generic).

Over-the-counter Drugs

We found 2 studies that addressed substitution ofover-the-counter drugs for prescription drugs, withmixed results, but the findings were inconclusive. In theRAND Health Insurance Experiment, in which costsharingtiers were randomly assigned to families, thenumber of over-the-counter drugs purchased rose as thelevel of cost sharing decreased, suggesting that over-thecounterdrug purchases are a complement to othermedical care services.33 However, this study was basedon purchases of over-the-counter drugs, which are oftenbought and stockpiled for later use. Conversely,Goldman and colleagues20 estimated that, if cost sharingwas doubled, the decrease in the use of medications forchronic conditions was more than twice as large (32%)when a close over-the-counter substitute was availablethan when an over-the-counter substitute was not available(15%), but no data were available to verify thatpatients consumed more over-the-counter drugs.

Mail-order Pharmacy

Finally, we discovered evidence from a study27 thatpatients with higher levels of retail pharmacy cost sharingare more likely to use mail-order pharmacies.Additional evidence in this area is warranted.

VALUE OF PRESCRIPTION DRUGS

In the next sections, we first summarize the effectsof higher levels of cost sharing on the use of essential ormaintenance medications and assess whether patientsreduce the consumption of low-value prescriptiondrugs instead of those that may affect health, outcomes,or the process of care. We then summarizethe effects of higher levels of cost sharing on medicalcosts. Detailed information on these studies in anappendix is available from the author or at http://www.medstat.com/1research/gibson2.mht.

Essential Medications

"More essential" medications are generally consideredto be necessary to maintain or improve health.These drugs include maintenance medications that aretaken regularly over a long period (eg, antihypertensivemedications) and medications that are taken sporadically(eg, bronchodilators). As the medical need formedication rises, we would expect that higher levels ofcost sharing are likely to have a smaller effect on the useof medications that are essential to health and wellbeing.To examine the evidence, we focused on studiesthat examined the relationship between cost sharingand the use levels of essential medications. We summarizethe evidence in Table 2.

Although the definitions of essential drugs differed,most studies20,21,23,24,30,31,34-38 found that higher levels ofprescription drug cost sharing were associated with areduction in the consumption of essential medications.Two exceptions examined the effects of the 1996 introductionof a 25% coinsurance policy with an income-basedcap in Quebec (with a small change in costsharing because of the cap). One study39 focused on theuse of 4 classes of medications in older patients, and theother study40 focused on cardiac medication use amongolder patients discharged after acute myocardial infarction(MI). Both studies relied on a pre-post design withouta comparison group to assess changes in use, withthe authors identifying the effects of the cost-sharingincrease by comparing changes in prepolicy and postpolicyuse. Because prescription use has been risingrapidly over time, a comparison with contemporaneoustrends instead of historical trends may have produceddifferent results. In addition, Pilote et al40 investigatedthe utilization of services for patients recently hospitalizedfor acute MI, who may have been reluctantto reduce cardiac drug use even after an increase incost sharing.

Most of the studies20,23,35,37 that found a significantassociation between cost sharing and a reduction inthe use of essential medications focused on a broadpopulation of acutely and chronically ill beneficiaries.Other studies21,24,30,31,34,36,39 found smaller reductionsin use for chronically ill patients or active users ofessential medications, who are less likely to be pricesensitive. One exception to this finding was the studyof regular users of essential medications among olderand welfare recipients in Quebec who were subject tothe 1996 coinsurance increase. In this study, Tamblynand colleagues38 reported reductions in the use ofessential medications. Unlike most of the other studies,which examined the effects on the use within individualclasses of medications, this study examined alarge number of medications grouped into a singleclass, which may have had cumulative results.

Less Essential and More Essential Medications

When the price of a drug rises, patients who are armedwith full information about the benefits and risks of medicationswould be expected to decrease the consumptionof less essential medications to a greater extent thanmore essential medications. We examined the results ofstudies that included less essential and more essentialmedication classes. In the studies21,35,39 examining therelationships between cost sharing and the use of individualtherapeutic classes of medications, there were noclear trends for less essential compared with more essentialmedications. However, larger reductions in the use ofless essential medications were reported in 3 studies29,36,38in which individual therapeutic classes were notexamined and medications were aggregated into the 2classes of more essential and less essential medications.Patients perhaps make substitutions across individualclasses of essential medications, although further investigationinto the effects of higher levels of cost sharing onessential medications is warranted.

HEALTH OUTCOMES

If patients are fully informed about the benefits andrisks of medications, we would expect that higher levelsof cost sharing, and any associated reduction in prescriptiondrug use, would not have an effect on healthoutcomes. Patients would reduce consumption at themargin, with little to no effect on quality and health outcomes.In this section, we review studies that examinethe effects of an increase in cost sharing on indicators ofhealth outcomes in the categories of (1) healthcareservice utilization and (2) health status and mortality.

Healthcare Service Utilization

Although healthcare service utilization is not a directmeasure of quality of care or health status, an increasein the utilization ofservices (outpatientmedical visits, emergencydepartmentvisits, inpatientadmissions, or hospitalreadmissions)coincident with higherlevels of cost sharingmay indicate pooroutcomes or theoccurrence of adverseevents (Table 3).

In most cases,higher levels of costsharing were not associated with changes in the utilizationof low-intensity outpatient medical services, suchas physician office visits,22,24,41 outpatient visits,40 andhome health visits.41 However, these studies assessedsmall changes in prescription drug cost sharing. As costsharing continues to rise, it is plausible that the utilizationof outpatient visits may change as patients seekphysician advice for medication management and otherservices.

Two studies reported an increase in high-intensityhealth services, such as inpatient visits, as cost sharingrose. In the first study, Christian-Herman and colleagues30found that inpatient visits rose after ageneric-only drug plan, comprising a large change incost sharing, was initiated among Medicare HMOenrollees. However, the findings were not adjusted fordifferences in the composition of the treatment andcomparison groups because the authors did not haveaccess to detailed demographic data. In addition, anincrease in admissions was not detected across all ofthe diagnostic groups that were studied. Patients withcongestive heart failure or coronary artery disease didnot have a significant change in admission rates, butpatients with diabetes mellitus had significantly moreadmissions. In the second study, which measured theeffects of the implementation of the 25% coinsurancecharge up to an income-based cap, a minimal changein cost sharing) among older persons and welfarerecipients in Quebec, Tamblyn and colleagues38 foundthat patients who decreased the consumption of essentialdrugs had more adverse events (including firstacute hospitalization, long-term care admissions, anddeath). Patients reducing the consumption of lessessential medications did not have a significant changein adverse events.

Four studies reported no association between higherlevels of cost-sharing and utilization rates for higher-intensityservices, such as inpatient admissions,22,24,41emergency department visits,22,24,40,41 and readmissionsamong older patients hospitalized with complicationsafter acute myocardial infarction.40 The lack of associationwith higher levels of cost sharing may be explainedby the patient populations, in which (for many of theprivately insured22,24 and older41 enrollees who werestudied) prescription drug costs may represent smallpercentages of incomes. Furthermore, in the study byPilote et al,40 the utilization of services was investigatedamong patients recently hospitalized for acute myocardialinfarction, whose demand for prescription drugs andother medical services may have been price inelastic.

Health Status and Mortality

We found no studies measuring the effects of costsharing on direct measures of health status, such asself-reported health status and empirical measures ofclinical health status (eg, laboratory readings). However,Johnson and colleagues21 reported a significantlylarge decline in a claims-based score of health status(based on the Chronic Disease Score and the DiagnosticCost Group) for a Medicare HMO with copaymentincreases from $1 to $3 to $5 but not in a MedicareHMO with a copayment increase from 50% with a $25maximum to 70% with a $30 maximum.

Finally, the study40 focusing on the introduction ofthe 25% coinsurance charge in Quebec reported thathigher levels of cost sharing had no effect on mortalityrates among patients discharged after acute MI. Futurestudies that add to the evidence of the effects on healthstatus and mortality are warranted.

MAINTENANCE MEDICATIONS ANDPROCESS OF CARE

Many health conditions require adherence to a regularregimen of maintenance prescription drugs to slowthe progression of disease, to maintain health, or toimprove health.

While an increase in adverse health outcomes maybe indicative of inadequate treatment, changes in theprocess of care may similarly be indicative of treatmentdisruptions. We herein review the evidence of thelinks between higher levels of cost sharing andprocess-of-care measures for maintenance medications,including adoption of therapy, drug discontinuation,and adherence.

Adoption of Therapy

Because many chronic conditions are managed withprescription drug therapy, initiation of appropriate drugtreatment can be essential to manage the progression ofdisease. Two cross-sectional studies revealed that higherlevels of cost sharing may lessen the rate of adoption ofexpensive, newer therapies, such as disease-modifyingdrugs for multiple sclerosis42 and cyclooxygenase-2inhibitors.43 For patients taking a large number of concurrentmedications, we found no studies that addressedthe relationship between cost sharing and cost-relatedtrade-offs between medications. Future studies evaluatingwhether higher levels of cost sharing delay initiation oftherapy will aid in treatment and benefit design forpatients with chronic and other serious illnesses.

Drug Discontinuation

Drug discontinuation is perhaps the most extremeresponse to a prescription drug price increase. Drugdiscontinuation is typically identified by an absence ofrefills within the same or a similar medication class(after allowing for switching to related medications).We found evidence in the literature that patients do,at times, discontinue medications when cost sharingrises, but the results varied across studies and medicationclasses (Table 4).

Ellis and colleagues44 reported that, compared withpatients with copayments less than $10, discontinuationrates of statins for new and incident statin patientswere more than 4 times higher for patients with acopayment of $20 or more and 1.4 times higher forpatients with a copayment between $10 and $20. In thisstudy, discontinuation was defined as a 7-day gap intreatment, so patients who stretched out refills mayhave been classified as discontinuers. Because this mayoccur more frequently in groups with higher copayments,this could have inflated the discontinuation rate;however, it raises the issue of whether short gaps areacceptable treatment patterns.

In a study23 of enrollees from 2 employers thatmoved to a 3-tier pharmacy benefit, the employer movingfrom a 1-tier plan to a 3-tier plan and increasingcopayments in all tiers had significantly higher rates ofdrug discontinuation (no refills in the 6 months afterthe copayment increase) than a comparison group forthe 3 classes of medications that were analyzed (ACEinhibitors, proton pump inhibitors, and statins). Thesecond employer instituted less extreme cost-sharingchanges, moving from a 2-tier to a 3-tier plan andincreasing copayments in the third tier only; here thediscontinuation rates for proton pump inhibitors andstatins were no different from those in the comparisongroup. Somewhat surprisingly, enrollees of this employer'shealth plan had significantly lower rates of discontinuationfor ACE inhibitors than the comparison group.

Conversely, Huskamp and colleagues37 discoveredthat discontinuation rates for children taking attention-deficit/hyperactivity disorder medications for anemployer moving from a 1-tier to a 3-tier plan were notsignificantly higher after the copayment change. Thisfinding may be a result of the necessity of these medicationsfor the well-being of children, resulting in priceinsensitivity; also, parents may be acting as good economicagents for their children.

Two related studies, by Motheral and Fairman24 andby Fairman et al,22 evaluated drug continuation ratesfor preferred provider organization enrollees who werecurrent users of medications in 4 drug classes. After a 3-tier plan was introduced, there was a decline in 6-month (in both studies) and 11-month (in the firststudy) continuation rates for estrogens. No difference incontinuation rates were noted for the other classes ofmedications (oral contraceptives, antihypertensivemedications, and antihyperlipidemic drugs).

Although patients and professionals may differ intheir perception of necessity, the drugs that were discontinuedin most of these studies were considered tobe clinically essential medications, for which discontinuationwas generally not recommended. Until long-termstudies are undertaken, the ultimate effects of discontinuationon the outcomes of care will remain unknown.Also, all of the studies were performed among privatelyinsured patients, for whom copayments were typically asmall percentage of income. Similar studies of discontinuationbehavior need to be performed on more vulnerablepopulation groups in which cost may be a largerissue and discontinuation behavior may differ from thatof the privately insured.

Adherence

In response to higher prices, patients may exhibitbehaviors that reduce the amount of prescription drugstaken, such as skipping doses and stretching outrefills.45,46 This is a concern because adherence to a regimenof maintenance medication has been, for the mostpart, associated with better outcomes.47-49 While wefound no empirical studies that linked cost sharing toadherence and outcomes (although surveys have linkedcost-related underuse to negative outcomes50-52), severalstudies that addressed the relationship between costsharing and adherence are addressed herein.

Although no commonly accepted standard definitionfor adherence exists, in this review adherence may alsobe considered as compliance or persistence and typicallyrefers to refill compliance (the timing of refills) oradequacy of medication coverage (the percentage ofdays with usable medication on hand). The study byPilote et al40 found no relationship between the introductionof the 25% coinsurance charge (up to a cap) inQuebec and adherence to cardiac medications after dischargefrom the hospital following acute MI. As statedpreviously, patients surviving an acute MI are more likelyto be price insensitive and, consequently, adherent tocardiac medications when facing a price increase.

Three other studies found a relationship between costsharing and adherence. The first study44 analyzed adherenceamong adult patients in a midwestern US managedcare organization who were prescribed statin therapy.Patients with higher copayments were less likely toadhere (based on the percentage of days without adequateprescription drug coverage) than patients withlower copayments. In a second study53 of adults in 9firms taking oral antidiabetic medications, higher levelsof coinsurance or copayment were associated with lowerlevels of refill compliance (based on refilling within 90days after an initial prescription ran out). The thirdstudy,30 which analyzed the effects of the institution ofthe generic-only benefit among Medicare HMO enrollees,found that adherence (based on the mean number ofmonths with an available prescription supply) was significantlyreduced in 4 of 5 disease or medication categories(congestive heart failure [ACE inhibitors orangiotensin II receptor blockers], coronary artery disease[statins], diabetes mellitus [any diabetic agent, ACEinhibitors or angiotensin II receptor blockers, andstatins], and antidepressant use). Adherence in the fifthcategory, the use of epileptic agents, declined but not significantly;however, the sample size for epilepsy wassmall, which may have reduced the statistical power.

DIRECT AND INDIRECT COSTS

Expenditures are a composite of the rate of use andthe cost per prescription (defined as quantity-×-cost perprescription) (Table 5). The demand for prescriptiondrugs is inelastic, and higher levels of cost sharing areexpected to result in small declines in the quantity ofprescription drugs. The costs of drugs also affect theexpenditures. We examined 3 types of expenditures,namely, total expenditures (quantity-×-total amountpaid for a prescription), health plan expenditures(quantity-×-health plan payment), and patient expenditures(quantity-×-cost-sharing amount).

We would expect to see a small decline in totalexpenditures associated with a decline in the drugquantity consumed unless patients switched to morecostly drugs. Consistent with this expectation, studies19,24,27-29,36,54 that estimated the effects of an increasein cost sharing on direct prescription drug costs foundthat higher levels of prescription drug cost sharing wereassociated with a reduction in total prescription drugexpenditures.

Three studies22,55,56 that reported no effects or inconsistenteffects of higher cost sharing were based on theexperience of privately insured individuals, who tend tohave higher incomes and may be less price responsive.In a study55 of 212 employer groups in preferredprovider organization and managed indemnity plans,patients with higher levels of cost sharing appeared touse fewer but more expensive medications. In a study56of managed care enrollees in employer-based plans,higher copayments were associated with a reduction inexpenditures for enrollees in independent practiceassociation plans but not network-based plans. Thefindings may be attributed to physician prescribingincentives to eliminate less essential prescriptions.These incentives have had more of an effect on prescribingthan patient copayments. In another study22 ofthe effects of a 3-tier copayment in a preferred providerorganization, the absence of cost savings in the interventiongroup should be interpreted with cautionbecause the cost estimates were based on averagewholesale price and not actual cost.

We would also expect that an increase in cost sharingresults in pharmaceutical cost savings to a healthplan as its cost per prescription is reduced by theincrease in patient cost sharing. This was consistentlyreported in the literature.22,24,25,27-30,41,56,57 Not surprisingly,another consistent finding was cost shiftingto patients via an increase in out-of-pocket payments.24,25,27-30,41,56,57

It is striking to discover how little we know about thecost effects of an increase in cost sharing, beyond theeffects on prescription drug costs. Johnson et al41 studiedthe effects of higher prescription drug cost sharingin 2 Medicare risk-based HMOs from 1987 to 1991 andfound that changes in cost sharing did not have a significanteffect on medical expenditures.

Further research needs to emphasize whether anincrease in cost sharing is cost saving or results inlong-term or short-term spending. The indirect anddirect effects of cost sharing are important areas offuture study and would move current knowledgebeyond a compartmentalized focus on prescriptiondrugs and into a broader view of the effects on healthand healthcare.

DISCUSSION

Cost sharing was originally intended to curb insurance-related overuse, and research confirms that higherlevels of prescription drug cost sharing generally producesome of the intended effects of decreasing the consumption of prescription drugs and steering patientsaway from nonpreferred brand-name drugs to preferredbrand-name drugs. However, patients do not consistentlyappear to be switching to generic substitutes, whichare considerably less expensive than brand-name drugs.

It is also becoming clear that cost sharing is notalways a benign instrument, and at times it may comeat a price. Although not consistently reported in the literature,the most troublesome effects associated withhigher levels of cost sharing are reports of treatment disruptionfor chronically ill patients who depend on a regularregimen of prescription drugs. In addition, higherlevels of cost sharing can have significant effects on theuse of essential or maintenance medications, the outcomesof care, and the process of care.

Overuse and Underuse

Prescription drug cost sharing is expected to continueto increase in the near future. In the 2004 KaiserFamily Foundation and the Health Research andEducation Trust58 survey of employer-sponsored benefits,38% of small firms and 53% of large firms stated thatthey were likely to increase the amount that employeespay for prescription drugs in the next year. Whethercost sharing will continue to proliferate depends on theextent of overuse vs underuse of prescription drugs. Ifoveruse is the primary problem, then cost sharing couldbe an acceptable cure when patients and providers haveadequate information about the risks and benefits ofmedications. However, actual patient behavior does notreinforce the notion that patients can always discernbetween overuse, underuse, and appropriate use.

Equity and Fairness

One of the clearest conclusions noted in the literatureis that higher levels of cost sharing transfer a largerfinancial burden to the patient. As copayments rise tonew heights, concerns emerge about equity and fairnessbetween different groups of patients, especially thosewith low incomes and those who are chronically ill. Todate, most studies have focused on the effects of higherlevels of cost sharing among homogeneous study populations.A careful analysis of the effects of higher levelsof cost sharing on diverse subgroups of individuals iswarranted and would reveal which subgroups have agreater response to changes in cost sharing.

Demand for Prescription Drugs

Most research has focused on varying the price ofprescription drugs to the patient, but little emphasishas been placed on assessing the level of demand forprescription drugs. There are 2 implicit assumptionsin the current cost-sharing system in which cost-sharingamounts are assessed to the patient at the point ofservice. Given the findings of this review, in certaincases, either or both of the following assumptions maynot be valid and may affect the level of prescriptiondrug use.

The first assumption is that patient demand is valuedappropriately. If patients are undervaluing (or overvaluing)prescription drugs, then the demand is too low (ortoo high) and may result in consumption levels that aretoo low (or too high). If a drug is undervalued, methodsexist to increase the demand for the drug, such aspatient and provider education; these will help patientsto incorporate unrecognized short-term and long-termbenefits (or risks) for the drugs of interest.

The second assumption is that any additional valueto society beyond what patients place on prescriptiondrugs, such as to employers (through increased productivity)and to family members (through reduced time offfrom work), is zero.59-61 If the benefits of prescriptiondrug consumption extend beyond the patients, thenpatients may not incorporate these additional benefitsinto their demand for prescription drugs. Time-testedapproaches, such as subsidies, education, and credits,may be used to increase the demand in such situations.

Future Directions

Although the current system of coinsurance andcopayments is easy to implement, a one-size-fits-allapproach could be replaced with targeted incentives forpatients and providers to use the right medications atlevels of use that are appropriate to their health conditions.Alternatives, such as benefit-based cost sharing,62in which the levels of cost sharing are based on the benefitto the patients, hold promise in this area.

Prescription drug cost sharing has become a widespreadand effective means to control prescription drugcosts among employer-based and publicly funded healthplans. The growing evidence in the literature suggeststhat, at times, unintended effects on the process andoutcomes of therapy may result from cost sharing. Thecentral question for health plan managers and policymakers is whether we will continue to use cost sharingas is or make modifications to reduce unintendedeffects.

From Medstat (TBG) and the Institute for Health and Productivity Studies, CornellUniversity and Medstat (RJO, RZG), Ann Arbor, Mich.

Funding for this work was provided by GlaxoSmithKline, Research Triangle Park, NC.

Address correspondence to: Teresa B. Gibson, PhD, Medstat, 777 E EisenhowerParkway, Ann Arbor, MI 48108. E-mail: teresa.gibson@thomson.com.

Health Aff (Millwood).

1. Smith C, Cowan C, Sensenig A, Catlin A, Team HA. Health spending growthslows in 2003. 2005;24(1):185-194.

Employer Health

Benefits: 2004 Annual Survey.

2. Kaiser Family Foundation and the Health Research and Education Trust.Section 9: prescription drugs and mental health benefits In: Menlo Park, Calif: Kaiser Family Foundation and theHealth Research and Educational Trust; September 2004:111-120.

Managed

Care Magazine.

3. Perlstein S. Four-tier approach injects consumerism into drug benefit. August 2001;10(8):21-24.

4. Traynor K. Drug costs for 2003 may rise 20%. 2002. Available at: www.safemedication.com/meds/ShowArticle.cfm?id=3188. Accessed August 17, 2004.

Med Care.

5. Bierman AS, Bell CM. Penny-wise, pound-foolish: the costs of cost-relatedmedication restriction. 2004;42:623-625.

Med Care.

6. Roemer MI, Hopkins CE, Carr L, Gartside F. Copayments for ambulatory care:penny-wise and pound-foolish. 1975;8:457-466.

Pharmacoeconomics.

7. Levy RA. Prescription drug cost sharing: economic and health impacts andimplications for health policy. 1992;2:219-237.

Pharmacoeconomics.

8. Smith DG, Kirking DM. Impact of consumer fees on drug utilization.1992;2:335-342.

Health Policy.

9. Huttin C. The use of prescription charges. 1994;27:53-73.

Milbank Q.

10. Maio V, Pizzi L, Roumm AR, et al. Pharmacy utilization and the MedicareModernization Act. 2005;83:101-130.

Med Care Res Rev.

11. Rice T, Matsuoka K. The impact of cost-sharing on appropriate utilization andhealth status: a review of the literature on seniors. 2004;61:415-452.

Int J Health Serv.

12. Lexchin J, Grootendorst P. Effects of prescription drug user fees on drug andhealth services use and on health status in vulnerable populations: a systematicreview of the evidence. 2004;34:101-122.

Ann Rev Public Health.

13. Adams AS, Soumerai SB, Ross-Degnan D. The case for a Medicare drug coveragebenefit: a critical review of the empirical evidence. 2001;22:49-61.

Clin Ther.

14. Olson BM. Approaches to pharmacy benefit management and the impact ofconsumer cost sharing. 2003;25:250-272.

Ann Pharmacother.

15. Gleason PP, Gunderson BW, Gericke K. Are incentive-based formulariesinversely associated with drug utilization in managed care? 2005;39:339-345.

Am Econ Rev.

16. Pauly MV. The economics of moral hazard: comment. 1968;58(pt 1):531-537.

Health Aff (Millwood).

17. Pauly MV. Medicare drug coverage and moral hazard. 2004;23(1):113-122.

Dis Manag Health Outcomes.

18. Rizzo JA, Ozminkowski RJ, Goetzel RZ. Prescription to over-the-counterswitching of drugs: methodological issues and implications for non-sedating antihistamines.2005;13:83-92.

Soc Sci Med.

19. Leibowitz A, Manning W, Newhouse JP. The demand for prescription drugsas a function of cost-sharing. 1985;21:1063-1069.

JAMA.

20. Goldman DP, Joyce GE, Escarce JJ, et al. Pharmacy benefits and the use ofdrugs by the chronically ill. 2004;291:2344-2350.

Health Serv Res.

21. Johnson RE, Goodman MJ, Hornbrook MC, Eldredge MB. The impact ofincreasing patient prescription drug cost sharing on therapeutic classes of drugsreceived and on the health status of elderly HMO members. 1997;32:103-122.

Clin Ther.

22. Fairman KA, Motheral BR, Henderson RR. Retrospective, long-term follow-upstudy of the effect of a three-tier prescription drug copayment system on pharmaceuticaland other medical utilization and costs. 2003;25:3147-3166.

N Engl J Med.

23. Huskamp HA, Deverka PA, Epstein AM, Epstein RS, McGuigan KA, FrankRG. The effect of incentive-based formularies on prescription-drug utilization andspending. 2003;349:2224-2232.

Med Care.

24. Motheral B, Fairman KA. Effect of a three-tier prescription copay on pharmaceuticaland other medical utilization. 2001;39:1293-1304.

J Manag Care Pharm.

25. Nair KV, Wolfe P, Valuck RJ, McCollum MM, Ganther JM, Lewis SJ. Effectsof a 3-tier pharmacy benefit design on the prescription purchasing behavior ofindividuals with chronic disease. 2003;9:123-133.

Med Care.

26. Rector TS, Finch MD, Danzon PM, Pauly MV, Manda BS. Effect of tiered prescriptioncopayments on the use of preferred brand medications. 2003;41:398-406.

Health Aff

(Millwood).

27. Thomas CP, Wallack SS, Lee S, Ritter GA. Impact of health plan design andmanagement on retirees' prescription drug use and spending, 2001. Jul/Dec 2002;suppl Web exclusives:W408-419.

JAMA.

28. Joyce GE, Escarce JJ, Solomon MD, Goldman DP. Employer drug benefitplans and spending on prescription drugs. 2002;288:1733-1774.

Am J Manag Care.

29. Motheral BR, Henderson R. The effect of a copay increase on pharmaceuticalutilization, expenditures, and treatment continuation. 1999;5:1383-1394.

Health Aff (Millwood).

30. Christian-Herman J, Emons M, George D. Effects of generic-only drug coveragein a Medicare HMO. Jul/Dec 2004;suppl Web exclusives:W455-468.

Health Aff (Millwood).

31. Kamal-Bahl S, Briesacher B. How do incentive-based formularies influencedrug selection and spending for hypertension. 2004;23(1):227-236.

Health Aff (Millwood).

32. Lichtenberg F. Are the benefits of new drugs worth their cost? evidence fromthe 1996 MEPS. 2001;20(5):241-251.

Med Care.

33. Leibowitz A. Substitution between prescribed and over-the-counter medications.1989;27:85-94.

Health Policy.

34. Blais L, Couture J, Rahme E, LeLorier J. Impact of a cost sharing drug insuranceplan on drug utilization among individuals receiving social assistance.2003;64:163-172.

J Chronic Dis.

35. Foxman B, Valdez RB, Lohr KN, Goldberg GA, Newhouse JP, Brook RH. Theeffect of cost sharing on the use of antibiotics in ambulatory care: results from apopulation-based randomized controlled trial. 1987;40:429-437.

Med Care.

36. Harris BL, Stergachis A, Ried LD. The effect of drug copayments on utilizationand cost of pharmaceuticals in a health maintenance organization. 1990;28:907-917.

Arch Gen

Psychiatry.

37. Huskamp HA, Deverka PA, Epstein AM, et al. Impact of 3-tier formularies ondrug treatment of attention-deficit/hyperactivity disorder in children. 2005;62:435-441.

JAMA.

38. Tamblyn R, Laprise R, Hanley JA, et al. Adverse events associated with prescriptiondrug cost-sharing among poor and elderly persons. 2001;285:421-429.

J Am Geriatr Soc.

39. Blais L, Boucher JM, Couture J, Rahme E, LeLorier J. Impact of a cost-sharingdrug insurance plan on drug utilization among older people. 2001;49:410-414.

CMAJ.

40. Pilote L, Beck C, Richard H, Eisenberg MJ. The effects of cost-sharing onessential drug prescriptions, utilization of medical care and outcomes after acutemyocardial infarction in elderly patients. 2002;167:246-252.

Med Care.

41. Johnson RE, Goodman MJ, Hornbrook MC, Eldredge MB. The effect ofincreased prescription drug cost-sharing on medical care utilization and expensesof elderly health maintenance organization members. 1997;35:1119-1131.

Clin

Ther.

42. Ozminkowski RJ, Marder WD, Hawkins K, et al. The use of disease-modifyingnew drugs for multiple sclerosis treatment in private-sector health plans. 2004;26:1341-1354.

Arch Intern Med.

43. Briesacher B, Kamal-Bahl S, Hochberg M, Orwig D, Kahler KH. Threetiered-copayment drug coverage and use of nonsteroidal anti-inflammatory drugs.2004;164:1679-1684.

J Gen Intern

Med.

44. Ellis JJ, Erickson SR, Stevenson JG, Bernstein SJ, Stiles RA, Fendrick AM.Suboptimal statin adherence and discontinuation in primary and secondary preventionpopulations: should we target patients with the most to gain? 2004;19:638-645.

JAMA.

45. Tseng CW, Brook RH, Keeler E, Steers WN, Mangione CM. Cost-loweringstrategies used by Medicare beneficiaries who exceed drug benefit caps and havea gap in drug coverage. 2004;292:952-960.

Health Care News.

46. HarrisInteractive. Out-of-pocket costs are a substantial barrier to prescriptiondrug compliance. November 20, 2001;1(32). Available at:http://www.harrisinteractive.com/news/newsletters/healthnews/HI_HealthCareNews2001Vol1_iss32.pdf. Accessed October 6, 2005.

Med Care.

47. Sokol MC, McGuigan KA, Verbrugge RR, Epstein RS. Impact of medicationadherence on hospitalization risk and healthcare cost. 2005;43:521-530.

Dis Manag.

48. Goldfarb N, Weston C, Hartmann CW, et al. Impact of appropriate pharmaceuticaltherapy for chronic conditions on direct medical costs and workplaceproductivity: a review of the literature. 2004;7:61-75.

Ann Pharmacother.

49. Vik S, Maxwell C, Hogan D. Measurement, correlates and health outcomes ofmedication adherence among seniors. 2004;38:303-312.

Med Care.

50. Heisler M, Langa K, Eby EL, Fendrick AM, Kabeto MU, Piette JD. The healtheffects of restricting prescription medication use because of cost. 2004;42:626-634.

Health Aff (Millwood).

51. Mojtabai R, Olfson M. Medication costs, adherence, and health outcomesamong Medicare beneficiaries. 2003;22(4):220-229.

Med Care.

52. Piette JD, Wagner TH, Potter MB, Schillinger D. Health insurance status,cost-related medication underuse, and outcomes among diabetes patients in threesystems of care. 2004;42:102-109.

Does Cost Sharing Affect Compliance? The Case of

Prescription Drugs.

53. Dor A, Encinosa W. Cambridge, Mass: National Bureau of Economic Research Inc;2004. NBER Working Paper Series, No. 10738.

Med Care.

54. Nelson AA, Reeder CE, Dickson WM. The effect of a Medicaid drug copaymentprogram on the utilization and cost of prescription services. 1984;22:724-736.

Inquiry.

55. Smith DG. The effects of copayments and generic substitution on the use andcosts of prescription drugs. 1993;30:189-198.

Health Aff (Millwood).

56. Hillman AL, Pauly MV, Escarce JJ, et al. Financial incentives and drug spendingin managed care. 1999;18(2):189-200.

J Manag Care Pharm.

57. Meissner BL, Moore WM, Shinogle JA, Reeder CE, Little JM. Effects of anincrease in prescription copayment on utilization of low-sedating antihistaminesand nasal steroids. 2004;10:226-233.

Employer Health Benefits: 2004

Annual Survey.

58. Kaiser Family Foundation and the Health Research and Educational Trust.Section 12: employer attitudes and opinions. In: Menlo Park, Calif: Kaiser Family Foundation and the HealthResearch and Educational Trust; September 2004:137-148.

J Occup Environ

Med.

59. Goetzel RZ, Ozminkowski RJ, Meneades L, Stewart M, Schutt DC.Pharmaceuticals: cost or investment? an employer's perspective. 2000;42:338-351.

The Value of Investment in Health Care.

60. MEDTAP International Inc. Bethesda, Md: MEDTAP International Inc; 2004.

Health Aff (Millwood).

61. Kleinke J. The price of progress: prescription drugs in the health care market.2001;20(5):43-60.

Am J Manag Care.

62. Fendrick AM, Smith DG, Chernew ME, Shah S. A benefit-based copay forprescription drugs: patient contribution based on total benefits, not drug acquisitioncost. 2001;7:861-867.