Health Affairs, 23, no. 1 (2004): 227-236 doi: 10.1377/hlthaff.23.1.227 © 2004 by Project HOPE	New Online   Obama's Success & Next Steps   3 Health Policy Options for President Obama   A Hybrid Proposal   Bipartisan Compromise

This Article Abstract Reprint (PDF) Submit a response to this article Alert me when this article is cited Alert me when Comments are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (14) Citing Articles via Google Scholar Google Scholar Articles by Kamal-Bahl, S. Articles by Briesacher, B. Search for Related Content PubMed PubMed Citation Articles by Kamal-Bahl, S. Articles by Briesacher, B. Related Collections Pharmaceuticals Health Spending

DataWatch

How Do Incentive-Based Formularies Influence Drug Selection And Spending For Hypertension?

Abstract

 
This study examined the association between incentive-based formularies and antihypertensive drug selection and spending. We compared the use of drugs from five drug classes by the number of tiers and copayment differentials. We found that raising copayments within a single-tier formulary system had a relatively modest impact on use of antihypertensives, compared with raising them in multi-tier systems. Likelihood of using ACE inhibitors and angiotensin II receptor blockers was lower among two-tier plans with generic/brand differentials of $10 relative to flat-copayment plans. Incentive formularies were associated with lower total antihypertensive spending by plans, but enrollees paid more out of pocket.

We know that tiered drug plans shift medication costs from health plans to patients, but how this translates to changes in medication choices is unclear. Economic theory specifies that the usual relationship between the price of a product and the quantity demanded becomes distorted for medical care for two reasons. First, the presence of health insurance generally insulates patients from the true costs of care, so they act with regard for only the out-of-pocket price. Second, the choice of medical care is actually made by physicians acting on behalf of patients. In the case of multi-tier drug plans, the distortion is theoretically lessened by giving patients incentives to select lower-cost medications (either generic drugs or the "preferred" brand-name product). However, this impact is still moderated by physicians. Theoretically, physicians have no personal reason to consider patients’ out-of-pocket costs except as a way to satisfy and retain patients. In practical terms, many plans that adopt tiered formularies also implement other measures that influence physicians’ compliance with the formulary.

The multi-tier copayment structure is a fairly recent innovation, so there is limited empirical research showing the extent of reductions in medication use and costs. Previous research has mostly assessed the impact of the institution of a flat copayment or increasing copayment amounts.⁵ In general, we know that even small increases in drug copayments result in decreases in drug use and spending. Switching from a flat to a two-tier copayment structure has been shown to reduce mean drug spending in a employer-sponsored population by 6–19 percent, although no significant increases in out-of-pocket costs were reported.⁶ Studies on three-tier drug formularies indicate lower health plan drug spending accompanied with an increase in patients’ out-of-pocket costs.⁷ Most studies have focused on overall drug use and spending rather than specific therapeutic drug classes.

The purpose of this paper is to examine the influence that multi-tier drug benefits exert on drug selection and spending. We expect that compared with rising copayment amounts within flat-copayment plans, multi-tier plans will have a greater influence, particularly where the copayment differentials are largest and create the greatest incentives for lower-tier use. We reduce some of the complexities of this relationship by examining a patient population facing a similar array of drug choices: people diagnosed with hypertension and using antihypertensive agents. Drug therapies for hypertension typically span multiple copayment tiers, including generic drugs, multisource brands, and the newest single-source brands.⁸ Our research offers one of the first examinations of incentive-based formularies at the level of therapeutic drug class.

Data And Methods

Top Editor's Notes Data And Methods Study Findings Conclusions And Policy... Notes

 
Data source. This study used Medstat’s 1999 MarketScan database, which consists of medical care claims, drug claims, and encounter data from approximately forty-five large employers, health plans, and public organizations. The MarketScan database collectively refers to the following three databases: Commercial Claims and Encounter, Medicare Supplemental and Coordination of Benefits (COB), and Benefit Plan Design. We used all three. The first two databases capture person-specific clinical use and spending data across inpatient, outpatient, prescription drug, and carve-out services for more than two million active employees as well as retirees with employer-sponsored Medicare supplement plans. The Benefit Plan Design data set contains abstracts from the plan booklets of participating plans in the MarketScan data set about the type of insurance plan (comprehensive plan; preferred provider organization, or PPO; or health maintenance organization, or HMO), mail-order pharmacy options, generic incentives, and copayments, including differential amounts for preferred and nonpreferred brand-name drugs.

Study sample. We used diagnoses codes from the inpatient and outpatient service claims to identify all people age eighteen or older with a diagnosis of hypertension (ICD-9-CM 401.xx–405.xx). We examined prescription service claims to identify users of antihypertensive drugs. People were counted as using antihypertensives if at any time during 1999 they had at least one claim from any of the following drug classes: diuretics, beta-blockers, angiotensin-converting enzyme (ACE) inhibitors, calcium-channel blockers, angiotensin II receptive blockers (ARBs), or others (centrally acting sympathetic agonists, peripherally acting sympathetic antagonists, and vasodilators). We excluded people who did not use antihypertensives, had no drug benefits or had a plan with insufficient detail on the drug benefit, or changed plans during the year.⁹ Our final sample comprised 149,243 adults enrolled in forty-two drug benefit plans (single-tier plans, n = 18; two-tier plans, n = 20; and three-tier plans, n = 4).

Study variables. Dependent variables were use of antihypertensive drugs and their costs. We examined antihypertensive drugs by therapeutic class and generic or brand-name status. For spending on antihypertensives, we computed per person monthly costs by dividing the total spending on antihypertensive medications by months of enrollment for each person. Drug costs were annualized by multiplying the average monthly estimate by twelve. We also measured the portion of antihypertensive drug costs paid for by the health plan or the patient.

The main independent variable was copayment tiers. We grouped the forty-two plans into seven broad categories defined by number of tiers and copayment levels, to reflect the wide range of copayments across plans with the same number of tiers. In assigning plans into tier categories, we considered both the absolute copayment amounts and the generic/brand differential.

Other independent variables were personal demographics, selected comorbidities, and type of health plan. Demographic characteristics included age, sex, geographic region of residence, and employment status (active full time, early retiree, Medicare retiree, other). We also controlled for other medical conditions that could increase the likelihood of using antihypertensive therapy (congestive heart failure, diabetes, arrhythmia, myocardial infarction, angina pectoris, and other coronary heart disease). We included measures for the type of health insurance plan (basic/major medical, comprehensive, HMO, PPO, and point-of-service, or POS, plan), since each could have its own characteristics in controlling drug use and costs, and whether the plan had a mail-order pharmacy option, since cost sharing is lower for this mode of purchase.

Statistical analysis. Basic descriptive statistics were computed to assess the relationship between antihypertensive use and copayment tiers. We conducted logistic regressions to estimate the likelihood of antihypertensive use by class as a function of copayment tiers, controlling for other person- and plan-level covariates. For drug expenditures, we used log-transformed ordinary least squares (OLS) models to derive conditional estimates of the influence of copayment tiers on antihypertensive spending given any antihypertensive use. The OLS regression models were also used to predict changes in antihypertensive spending as enrollees with flat-copayment plans moved into plans with two or three tiers. We simulated the new drug spending levels for enrollees in the most generous single-tier plan by altering only their drug copayment design while holding their demographic, health, and other plan characteristics constant. The unit of analysis in the study was the individual; however, we also adjusted for correlation among people belonging to the same plan. We used the robust cluster estimation commands to specify that the individual observations were independent across plans (clusters) but not within plans. All multivariate analyses were conducted with statistical software STATA 7.0.

Study Findings

Top Editor's Notes Data And Methods Study Findings Conclusions And Policy... Notes

 
Exhibit 1 presents the distribution of plans and enrollees and average annual spending for antihypertensives in 1999, by tier and copayment level.¹⁰ About half (51.6 percent) of the study sample was enrolled in two-tier plans, 33.2 percent belonged to single-tier plans, and the rest (14.2 percent) were in three-tier plans. Most enrollees faced a copayment of about $5, although almost a fifth could not purchase generic or brand-name antihypertensives for less than $10. Just over 60 percent had no incentive or only a modest one (less than $5) to use generics over branded products. Average annual use of antihypertensives ranged from 18.9 to 10.8 prescriptions per person, with the lowest levels of use occurring in the two-or three-tier plans with higher copayments. Mean spending for a year’s worth of antihypertensives ranged from $485 to $255 per person, and again the lowest amounts were closely associated with plans requiring the highest copayments.

Antihypertensive drug selection. Exhibit 2 presents the results from a set of logistic regressions examining the association of copayment tiers with use of medications from the five antihypertensive drug classes.¹² Drug-class models are ordered according to the relative costliness of the medications, with diuretics as the least expensive and ARBs as the most. The models are identically specified except for the dependent variable. The reference group is the same for all models—enrollees in single-tier plans with the lowest copayment amounts.

Predicted shifts in antihypertensive spending. Exhibit 3 shows predicted antihypertensive spending for enrollees in low-copayment single-tier plans under alternative copayment tiers. The main purpose of this exhibit is to see how total and out-of-pocket spending could shift for the same group of people as their cost sharing increases from single- to three-tier plans. These estimates are illustrative only, as the cross-sectional design of this study does not allow for modeling of true behavioral responses to policy changes. The first two rows display the predicted effects of imposing a doubling or tripling increase of copayments for all formulary medications. Doubling the copayment seems to double the portion paid out of pocket—a change that mildly depresses overall spending (about a 9 percent drop); tripling copayments more than triples the share paid out of pocket (from 5 percent to 28 percent of the total), and that shift in personal burden is accompanied by a drop in total spending of almost 20 percent. Looking next at the use of copayment incentives for generic use, in two-tier plans total spending on antihypertensives appears sensitive to the brand-name copayment levels despite the availability of lower-cost generics. Doubling or tripling the brand-name copayments in two-tier plans displays relatively similar associations as those seen in single-tier plans. The generic tier becomes influential only when the brand-name copayments reach a nearly fourfold increase and the generic/brand differential is at least $10. This considerable incentive to select generics is associated with a 40 percent drop in overall spending. The last row of Exhibit 3 shows the expected result of adding further incentives for using preferred brands. Adding a third tier is associated with a large shift in costs to patients: Under a three-tier plan, enrollees paid 42 percent of the total costs for their antihypertensives, as out-of-pocket spending rose more than 600 percent. Total and personal spending on antihypertensives seems closely related to the brand-name copayment levels that now include three- to fivefold increases for drugs in the uppermost tier.

Another limitation is that we could not control for enrollees’ income, a factor that could affect how enrollees’ drug selection decisions and spending vary with copayment levels. Additionally, our data set had only a few examples of three-tier plans. Hence, we could not study in detail the magnitude of increasing copayment incentives for preferred brands. Finally, our study did not examine the impact of tiered benefit design features on use and costs of medical services.

Conclusions And Policy Implications

Top Editor's Notes Data And Methods Study Findings Conclusions And Policy... Notes

 
This study is the first to examine the association between incentive-based formularies and antihypertensive drug selection and spending. We found that raising copayments within a single-tier formulary has a relatively modest impact on use of antihypertensives compared with doing so in multi-tier plans, especially those with higher brand/generic differentials. Hence, not only the absolute copayment amount but also the presence and extent of brand/generic differentials are associated with drug use and spending. We also found evidence that incentive plans might be associated with getting certain types of antihypertensives. The likelihood of receiving the more expensive ACE inhibitors and ARBs was significantly lower among enrollees in two-tier plans with the highest copayment differentials than for those in single-tier plans with the lowest copayments. Access to these classes of drugs is important because they are the first-line treatment of choice for patients with hypertension and concomitant heart failure or diabetes with proteinuria. The newer ARBs are similar to ACE inhibitors but also have a more desirable side-effect profile. Our findings indicate that patients in tiered cost-sharing plans might be using less costly drug classes rather than the ACE inhibitors and ARBs. Whether these less costly drug classes were inappropriate medication choices for these patients and their long-term adverse medical consequences should be the subject of future research. Finally, we found that incentive formularies are associated with lower total antihypertensive spending on the part of health plans and employers, not patients. In our simulations of new drug spending levels for enrollees in the most generous single-tier plan, their out-of-pocket spending increased three- to sevenfold as their antihypertensive agents moved into tiered formularies, while health plans saw decreases in outlays of 18–52 percent. Future research using longitudinal data should explore these observations to rule out potential confounders inherent in the population. More recent data would also provide more examples of three-tier plans to study and much larger brand/generic copay differentials. Finally, subsequent work should examine if these findings are reproducible among other disease populations dependent upon other therapeutic drug classes.

Although our study focused on a population with employer-sponsored drug coverage, we believe that the findings offer caution for applying incentive formularies to more vulnerable groups. The large increases in out-of-pocket expenses for antihypertensives associated with tiered formularies suggests that the personal costs for buying all medications used by an older person with multiple chronic conditions could be considerable. This, coupled with our finding on lower use of ACE inhibitors and ARBs, suggests that people with limited personal resources might be forced to do without necessary medications or use less costly (but probably less effective) ones. Thus, it is important to discourage the adoption of cost containment systems that encourage poor medication choices in the first place. No guidelines exist on what kinds of medications should be available in the lower tiers or how much can be reasonably charged for higher-tier drugs and still be considered formulary medications. These issues are fairly basic to address as Medicare drug benefit bills consider allowing use of tiered formulary designs by private PBMs. A tiered drug copayment design for Medicare beneficiaries can be acceptable only insofar as it encourages less costly use that is still therapeutically appropriate. Policymakers need to evaluate alternatives such as reference-based pricing, wherein insurance covers the cost up to the reference price (defined by low-cost, benchmark drugs), and patients pay the extra cost of more costly drugs within the specific therapeutic cluster of medications considered close substitutes for treating a particular illness.¹³ Recent studies of reference pricing for ACE inhibitors reported a sustained reduction in drug spending with no changes in overall use of antihypertensive therapy, use of medical care, or costs.¹⁴ However, the design of such a system poses its own set of challenges and trade-offs.¹⁵ In the end, whatever form of pharmacy benefit management is chosen, policymakers must bear in mind that if the cost burden on patients results in decreased use of and adherence to lifesaving therapies, then the drug benefit will fail to serve its main purpose: providing affordable access to necessary drugs.

Editor's Notes

Top Editor's Notes Data And Methods Study Findings Conclusions And Policy... Notes

 
Sachin Kamal-Bahl is a doctoral candidate in the Pharmaceutical Health Services Research Program, School of Pharmacy, University of Maryland Baltimore (UMB). Becky Briesacher is an assistant professor and director of research at the UMB’s Peter Lamy Center.

The authors thank Bruce Stuart, Jalpa Doshi, and two anonymous referees for their helpful comments, and Jim Gardner for his analytical help. This work was conducted as part of a predoctoral fellowship funded by Novartis Pharmaceuticals Corporation.

Notes

Top Editor's Notes Data And Methods Study Findings Conclusions And Policy... Notes

 

1. S. Findlay, "Prescription Drug Expenditures in 2001: Another Year of Escalating Costs" (Washington: National Institute of Health Care Management Research and Education Foundation, May 2002).
2. Scott-Levin, Spring 2001 Managed Care Formulary Drug Audit (Newtown, Penn.: Scott-Levin, June 2000).
3. P. Penna, "Three-Tier Co-Pay Systems and Consumer-Centric Care," Journal of Managed Care Pharmacy 6, no. 5 (2000): 351–353.
4. B. Motheral and K.A. Fairman, "Effect of a Three-Tier Prescription Copay on Pharmaceutical and Other Medical Utilization," Medical Care 39, no. 12 (2001): 1293–1304[CrossRef][ISI][Medline]; and G. Joyce et al., "Employer Drug Benefit Plans and Spending on Prescription Drugs," Journal of the American Medical Association 288, no. 14 (2002): 1733–1739.[Abstract/Free Full Text]
5. R.E. Johnson et al., "The Effect of Increased Prescription Cost-Sharing on Medical Care Utilization and Expenses of Elderly Health Maintenance Organization Members," Medical Care 35, no. 11 (1997): 1119–1131; [CrossRef][ISI][Medline]B.L. Harris, A. Stergachis, and L.D. Reed, "The Effect of Drug Copayments on Utilization and Cost of Pharmaceuticals in a Health Maintenance Organization," Medical Care 28, no. 10 (1990): 907–917[ISI][Medline]; and B.R. Motheral and R. Henderson, "The Effect of a Copay Increase on Pharmaceutical Utilization, Expenditures, and Treatment Continuation," American Journal of Managed Care 5, no. 11 (1999): 1383–1394.[ISI][Medline]
6. Joyce et al., "Employer Drug Benefit Plans."
7. Ibid.;Motheral and Fairman, "Effect of a Three-Tier Prescription Copay"; and C.P. Thomas et al., "Impact of Health Plan Design and Management on Retirees’ Prescription Drug Use and Spending, 2001," 4 December 2002, www.healthaffairs.org/WebExclusives/Thomas_Web_Excl_120402.htm (23 October 2003).
8. A 2000 Scott-Levin report indicated that antihypertensive drug classes including ARBs, calcium-channel blockers, and ACE inhibitors are among the most likely to be included in differential copay systems.
9. Our study was not designed to address the impact of tiered formularies on the likelihood of any antihypertensive use or spending among patients with hypertension. A patient’s decision on whether to use any antihypertensive medication at all is expected to be driven by the lowest copay level—that is, the copay amount for generic antihypertensives. We did not have much variation in the generic copayment across our study health plans; more than 80 percent of all plans had a generic copay of less than $5. Thus, we did not expect to see major differential impacts on any use or spending for antihypertensive drugs. This remains a question to be addressed by future research. People who changed plans were excluded to reduce potential bias from selection into another drug plan based on anticipated use.
10. Because of space limitations, we have not presented characteristics of the study sample. Significant differences were observed across most characteristics when the study sample was stratified by copay tier status and copay levels. These tables can be obtained from Sachin Kamal-Bahl at skama001{at}umaryland.edu.
11. These results held even in the multivariate analyses.
12. A logistic regression was also estimated to examine the association of copayment tiers on use of any brand medication. Enrollees in two-tier plans with the highest brand/generic copay differentials and three-tier plans had lower odds of using any brand medication, compared with those in single-tier plans with the lowest copayments.
13. P. Kanavos and U. Reinhardt, "Reference Pricing for Drugs: Is It Compatible with U.S. Health Care?" Health Affairs (May/June 2003): 16–30.
14. S. Schneeweiss et al., "Impact of Reference-Based Pricing for Angiotensin-Converting Enzyme Inhibitors on Drug Utilization," Canadian Medical Association Journal 166, no. 6 (2002): 737–745[Abstract/Free Full Text]; and S. Schneeweiss et al., "Outcomes of Reference Pricing for Angiotensin-Converting-Enzyme Inhibitors," New England Journal of Medicine 346, no. 11 (2002): 822–829.[Abstract/Free Full Text]
15. Kanavos and Reinhardt, "Reference Pricing for Drugs"; and A. Towse, "The Efficient Use of Pharmaceuticals: Does Europe Have Any Lessons for a Medicare Drug Benefit?" Health Affairs (May/June 2003): 42–45.

                      What's this?

This article has been cited by other articles:

				D. P. Goldman, G. F. Joyce, and Y. Zheng Prescription Drug Cost Sharing: Associations With Medication and Medical Utilization and Spending and Health JAMA, July 4, 2007; 298(1): 61 - 69. [Abstract] [Full Text] [PDF]

				H. A. Huskamp, P. A. Deverka, A. M. Epstein, R. S. Epstein, K. A. McGuigan, A. C. Muriel, and R. G. Frank Impact of 3-Tier Formularies on Drug Treatment of Attention-Deficit/Hyperactivity Disorder in Children Arch Gen Psychiatry, April 1, 2005; 62(4): 435 - 441. [Abstract] [Full Text] [PDF]

				P. P Gleason, B. W Gunderson, and K. R Gericke Are Incentive-Based Formularies Inversely Associated with Drug Utilization in Managed Care? Ann. Pharmacother., February 1, 2005; 39(2): 339 - 345. [Abstract] [Full Text] [PDF]

				F. T. Shaya, S. Blume, and C. D. Mullins Prescription Drug Spending Trends For The Privately Insured In Maryland, 2000-2001 Health Aff., September 1, 2004; 23(5): 226 - 232. [Abstract] [Full Text] [PDF]