Prepared Witness Testimony

The House Committee on Energy and Commerce

Designing a Twenty-First Century Medicare Prescription Drug Benefit.

Subcommittee on Health
April 8, 2003
10:00 AM
2123 Rayburn House Office Building 

Mr. Roger Feldman Ph.D.
Health Services Rsch/Policy
University of Minnesota
15 -210 PWB
516 Delaware Street SE
Minneapolis, MN, 55455

It is my pleasure to appear before you this morning to discuss a vital issue for the Medicare program: the provision of an outpatient prescription drug benefit for fee-for-service Medicare.  For the past 20 years, I have studied the private health insurance industry from the vantage of a university researcher.  From 1995 to 2000, I assisted the Centers for Medicare and Medicaid Services in designing a competitive pricing demonstration for Medicare.  These experiences have been instrumental in shaping the ideas that I wish to share with you today.   

As you well know, the Medicare entitlement provides only limited coverage of outpatient prescription drugs.  This is in sharp contrast to private insurance plans that cover most Americans under 65.  Virtually all analysts agree that prescription drug coverage should be part of Medicare.  Without it, the elderly not only must spend unbearably large sums of money out-of-pocket for drugs, but they may forego cost-effective treatments.  As President Bush said on March 4, "Medicare will pay a doctor to perform a heart bypass operation, but it will not pay for drugs that could prevent the surgery." 

Despite this apparent consensus, it has been extremely difficult to fashion a Medicare drug coverage plan.  Much of the difficulty centers on debate whether the program should be organized by the public sector through a single government-administered entity, or offered by the private sector through competing pharmacy benefit plans.  In my opinion, this debate has polarized the discussion and has obscured the solution that we should be working toward.  That solution includes both the public and private sectors.  The role for private sector plans is to operate the drug program, and the government's role is to set the rules under which those private plans operate.   

Why the Private Sector?

The private sector should run the program because it is better than the government at discovering and implementing innovations to reduce the cost and improve the quality of health insurance benefits.  Let me describe one example of private sector innovation.  As you know, pharmaceutical manufacturers have long promoted their products through the use of "detailing" representatives who visit physicians.  Blue Cross and Blue Shield of Minnesota, a not-for-profit health benefit plan, was concerned that detailing was not in the interest of patients, employers and insurers because it advocates the use of drugs with high profit margins at the expense of alternatives that might be less expensive and more beneficial.  So within each drug therapeutic class, Blue Cross declared that certain drugs would be preferred, whereas others would be given a yellow cautionary flag and some would be deemed as red agents - the least cost effective.  Blue Cross conducted classes for physicians and sent pharmacists to visit clinics with messages promoting the plan's favored drugs.  Andrea De Vries, a student from our doctoral program at the University of Minnesota, found that there was a consistent increase in prescribing preferred agents among clinics with more pharmacist visits.[1]  She concluded that it is possible to have a positive effect on drug utilization that is not driven by a financial incentive. 

This example illustrates the private sector's ability to recognize a problem, devise a solution, and implement it effectively.  Pharmacy benefit management (PBM) companies that run prescription drug benefit programs for most insured Americans under age 65 are another example of private sector innovation.  PBMs control the cost of prescription drug programs by targeting the behavior of pharmacists, drug manufacturers, consumers, and prescribing physicians.  One study found that PBMs obtained discounts of 13.2% below the average wholesale price of drugs, as well as manufacturer rebates of about 5%.[2]  Researchers from the RAND Corporation reported that aggressive management through private PBMs has been shown to reduce drug expenditures by 15% or more.[3] 

The private sector can adapt quickly when the need arises.  For example, in response to rising drug costs, many employers have introduced multi-tiered drug benefit plans where employees have to pay more for non-preferred drugs.  The use of "3-tier" pricing arrangements (lowest payment for generic drugs, middle payment for formulary or preferred brands, and highest payment for non-formulary brands) nearly doubled from 29% of covered workers in 2000 to 57% in 2002 according to a survey by the Kaiser Family Foundation.[4]  An additional 28% of workers had "2-tier" drug benefits with a lower payment for generics and a higher payment for brand name drugs. 

In contrast, changing the delivery of Medicare benefits for any reason has proven to be excruciatingly difficult.  For example, I know of only one instance where the government has used its purchasing power to contract with selected providers for Medicare services. That is the demonstration of competitive pricing for durable medical equipment (DME).  Congress actually used its power to block a demonstration of competitive pricing for Medicare M+C plans, despite the support of the CMS Administrator and a direct mandate from Congress to conduct such a demonstration.[5]     

Why Multiple Choices?              

A closely related question is whether Medicare beneficiaries should have a choice of pharmacy benefit plans.  Advocates of a single plan point to several supposed drawbacks of multiple choice, including increased administrative costs, adverse selection, and the burden of protecting beneficiaries from the consequences of making bad choices. 

Before dealing with these specific criticisms, let me say that choice of medical benefits should be good thing because Americans value choice.  "One-size" benefits do not fit everyone.  In the market for employer-based health insurance, for example, it has been demonstrated beyond a doubt that employers offer multiple health insurance plans because employees want to have choices.[6]  Even countries with national health insurance systems allow some individuals to opt out (Germany) or to purchase private insurance when they need to fill gaps in the government plan (Britain and Canada). 

Having choices also improves quality.  An evaluation of a health insurance purchasing coalition operated by large employers in Minneapolis found that introduction of multiple choices (as many as 15 separate provider-controlled delivery systems) in 1997 was associated with improvement in technical quality of care for patients with diabetes.[7] Rates of use of preventive services either remained stable or improved after the introduction of choice.        

What about the extra administrative cost of offering multiple choices?  This is a true cost that can't be ignored.  We know that large employers are more likely than small ones to offer multiple health insurance plans because they can spread the administrative cost over more enrollees.  But it is easy to make too much of this problem.  We could minimize the cost of buying automobiles if there were only one auto dealer in each city, and the administrative cost of grocery stores would be lower if there were only one of them.  But we value choice of auto dealers and grocery stores quite highly, despite the extra administrative costs.  Based on my discussions with a large PBM, I estimate that the minimum efficient size for low administrative costs is several hundred thousand enrollees.  This estimate suggests that statewide or regional bidding areas composed of multiple states would be large enough to offer multiple choices with low administrative costs. 

Whenever beneficiaries have multiple choices, the sicker ones may be more attracted to some health plans than to others.  This phenomenon is called "adverse selection," and it can have implications for the efficiency of the health benefit program.  Plans may try to avoid high risks by skimping on quality and cutting services that attract them.  Adverse selection could be a problem for a multiple-choice prescription drug benefit in Medicare.  However, it is useful (as it was for administrative costs) to put the problem in perspective.  My colleagues and I recently completed a study of adverse selection in the M+C program, which will be published in the Health Care Financing Review.  We estimated that M+C plans that offered drug benefits with an annual cap above $800 in 1999 attracted enrollees who cost 3.6% more than average.[8]  This is an upper limit on the cost of adverse selection, because adverse selection from offering drug benefits at all has to be larger than selection from tinkering with quality and services once the benefit is offered. 

The Role of Government

If this amount of adverse selection is unacceptable, it can be reduced further by "risk adjusting" the payments to the drug benefit plans.  Risk adjustment means that payments are increased to account for the adverse selection experienced by a plan.  A discussion of risk adjustment leads directly to the question of the government's role in a Medicare prescription drug plan.  In my opinion, the government should set the rules for the program and make sure those rules are enforced. 

One of the most important rules is how to share risk with the participating plans.  As I understand the proposals currently on the table, the possibilities range from government bearing all of the risk except for the plans' fees, to the private sector bearing most of the risk subject to risk-sharing corridors.  For example, plans might be at risk for a target level of spending, plus or minus 5%.  If costs exceed the upper limit, the government pays the extra amount, and conversely plans keep the extra savings if costs are less than the lower limit.  

Before discussing the details of my proposal risk-sharing proposal (which is only one among many), I want to emphasize that the purpose of risk bearing is to give private Medicare plans better incentives for cost containment.  In order to accomplish this goal, as a general rule, private plans should bear the risk for events that they control, but they should not bear risk for events they do not control.  In this context, private plans do not control the number of prescriptions that providers write within a therapeutic drug class.  Thus, they should not be at full risk.  However, private plans do control the choice of specific drugs to fill those prescriptions and the prices of those drugs.  Thus, they should be at risk for the cost of drug management and drug prices.  You could think of the cost equation as the product of three elements: 

DRUG COST PER PERSON =

(1) NUMBER OF THERAPEUTIC PRESCRIPTIONS PER PERSON *

(2) SPECIFIC DRUGS USED TO FILL PRESCRIPTIONS *

(3) PRICES OF SPECIFIC DRUGS 

To implement a payment system based on this formula, the government could specify the number of prescriptions in each therapeutic class that are written for a "standard beneficiary," who could be the average fee-for-service Medicare beneficiary for illustration.  In a standard population of 1,000, there might be 50 prescriptions for ACE Inhibitors, 25 for Lipotriptics, and 10 for H2 Blockers per month.  Using the preferred drugs on its formulary and its prices, a private plan might bid $100 per month to cover one standard enrollee.  After all the participating plans have submitted bids, the government could use the bids to determine the enrollee's out-of-pocket premium contribution.  For example, the government might pay 100% of the premium up to the higher of the median bid or the enrollment-weighted average bid, as we proposed for the M+C competitive pricing demonstration.  Premium competition would provide powerful incentives for plans to submit low bids. 

Next, enrollees would sign up for the drug benefit program.  Because of the large subsidy (e.g., the government pays 100% of the premium of at least one plan), enrollment would be nearly universal, so adverse selection between enrollees and the general Medicare population would be minimized.  Additional restrictions such as a penalty for delayed enrollment in the drug benefit program could also be imposed.[9]    

After joining the program, enrollees would have an opportunity at regular intervals to select a particular private drug benefit plan, and the government would observe how many prescriptions were written in each therapeutic class for the plan's enrollees.  This system requires the private plans to submit claims for each prescription, but that is their standard procedure for commercial business, so it is not an additional cost for them.  If the plan that bid $100 for a standard enrollee attracts sicker beneficiaries who use twice the standard amount of prescriptions, it would be paid $200 per month.  Because the plan is not at risk for the health of the enrollees it attracts, it does not have an incentive to skimp on covering drugs that attract them.  However, if the plan overestimates its ability to manage the types of drugs used, or if it is overly aggressive in estimating its ability to get low prices, it must eat those extra costs. 

This is not an ideal bidding system.  Ideally, all plans, including fee-for-service Medicare, should submit bids to cover all services.  Consumers could then choose among all plans based on premiums, amenities, and their preferences for the plans' medical management styles.  We proposed that system for the competitive pricing demonstration, although the demonstration eventually was restricted to M+C plans only. 

I am confident that the government - either CMS or a new, single-purpose agency - is technically capable of running a competitive pricing system for Medicare drug benefits.  Our experience in Denver proved conclusively that CMS could issue an RFP for a complete package of Medicare services in a very short time, and it could evaluate the plans' responses.  Bidding for one piece of the benefit package must be easier than that experience.  Later demonstration efforts in Phoenix and Kansas City explored the design of formularies, co-payments for single- and multiple-source drugs, expenditure caps, and which prices to apply against the cap.[10]  Solutions that were acceptable to the local stakeholders were found for all of those design issues.  

The last question raised by multiple choice of drug benefits is whether the information load on seniors would be unbearably difficult, leading them to make bad choices.  The main point I want to make is that the supply of information responds to the demand for it.  If consumers have no reason to care about drug costs, then it follows that they won't demand information and they will remain ignorant.  On the other hand, if they have a reason to care about drug costs, then there is a strong incentive to become informed.  I know this from my own research, in which several colleagues and I surveyed employees of Minneapolis companies that switched from a single-tier to a 3-tier drug benefit last year.  Compared with employees of companies that kept the old benefit, they were more likely to know the correct price of new drugs.[11]  We also found that more formal education was positively correlated with knowing that formularies, generic drugs, and mail order drugs have significant potential to reduce drug costs.  

The second finding highlights the important role of education in informing consumers about drug costs.  Our study looked at formal education, but there is also a role for specific drug education programs directed at seniors.  The government has a major responsibility for providing those programs.  Publishing information on drug benefits and summary measures of consumer satisfaction - as is done for the Federal Employees Health Benefits Program (FEHBP) - is an example of what the government can do for seniors.  The FEHBP also encourages members to request generic drugs instead of brand name drugs and to use their plan's home delivery program if it has one.[12]  The government could also require plans to provide on-line access to their formularies. 

In addition to these information measures, the government could demand that sensible consumer protections be built into every plan.  For example, although the bills under consideration are somewhat different, they all require private plans to include at least one and sometimes two brand-name drugs in each therapeutic class.[13]  An appeals process can offer protection against arbitrary coverage denials.  Finally, the patient's doctor can always write "dispense as written" orders. 

 The Bidding Area and Urban-Rural Differences

I would now like to discuss two related questions that are very important in designing a Medicare drug benefit: Should plans bid on local or regional areas? And should there be adjustments for urban-rural differences? 

The current M+C program allows risk-bearing organizations to designate service areas county-by-county (and even to select smaller areas if there is a significant geographic barrier to covering a whole county).  Although there is some controversy around this definition, it makes sense for M+C plans to serve small areas because medical care markets are local.  This is not the case for pharmacy benefit management.  Prices that PBMs pay for drugs are determined by national volume, and utilization management techniques are national in scope as well.  Because there is no distinct local market, it follows that the size of the bidding area should be determined by the minimum size needed to achieve economies of scale in administration.  As I mentioned earlier, this might be at the statewide or regional level.  Small states could be combined to achieve the critical mass needed for an efficient competitive bidding system.[14] 

Bidding for large regions would solve the problem of access for rural residents, which has plagued the M+C program.  Plans would have to cover all areas, both urban and rural, in the region.  If the cost of dispensing drugs were higher for rural pharmacies than for urban pharmacies (although I know of no evidence that this is the case), then urban residents would implicitly cross-subsidize rural residents in the same bidding region.  If local cost differences were discovered and Congress wished to recognize them, it could add an explicit adjustment factor to the payment system.                     

Concluding Observations on Medicare M+C

I would like to conclude with a few observations on the much-maligned M+C program, which has been accused by some of being unreliable and unstable.[15]  The reason why some HMOs have withdrawn from the M+C program is that the payment system is seriously flawed.  Instead of having HMOs tell the government how much it costs to provide Medicare services through a competitive bidding process, the government - which knows almost nothing about HMOs' costs - tells them how much it will pay.  Despite this flaw, the M+C program was still able to offer 391 separate products with some form of drug coverage in 2001.[16]  288 of those products had coverage limits that exceeded $800 per year, and 177 had unlimited coverage for generic drugs.  The average premium for M+C products with drug coverage was $41 per month.  Finally, about 54% of all Medicare beneficiaries had at least one drug-coverage M+C product available to choose in 2001.  Therefore, although the M+C program is far from perfect, it has provided a choice of drug coverage for many Medicare beneficiaries, an accomplishment that has been beyond the ability of fee-for-service Medicare. 

Thank you for allowing me to present these remarks.

Printer Friendly