Inside the Doughnut Hole

It's Midsummer, And across the country pharma sales managers are participating in what's become an annual ritual. "What's happening to sales?" the rep's boss asks. "Sales are off," the manager replies. "Consumers are hitting the doughnut hole (or gap). My reps tell me that patients are dropping off therapy because now they have to pay full price." Or: "We need a new patient assistance program to fill in the gap."

The "Doughnut Hole"—the coverage gap built into the Medicare Part D benefit design—was originally intended as a means of shifting the cost of this new program back onto consumers, and limiting the budget impact of a prescription benefit for the federal government. But it has also created significant uncertainty for pharmaceutical companies, which need to know how this shift to a defined contribution toward annual drug spending ($2,400 in 2007) will affect sales. In addition, it has created virtually unanswerable excuses for sales reps and managers.t does the doughtnut hol reallyUntil now, it has been difficult to verify these claims. The impact of the coverage gap on patient behavior has largely been a mystery.

The question is: What does the doughnut hole really do to drug sales?

Until recently, it's been impossible to tell. But over the last year, Amundsen Group has worked with healthcare data provider Verispan to demystify the impact of the coverage gap on Part D patient behavior. Specifically, Amundsen used Verispan's anonymous longitudinal patient data (ALPD) to create a representative sample of more than 3 million Medicare Part D beneficiaries who might have had exposure to a gap in branded or generics coverage, tracking their behavior as their share of prescription costs changed from a standard Part D copay or coinsurance to paying full price. For the 2006 and 2007 benefit years, this extensive sample was used to help a number of clients answer pressing questions such as: How many of my patients reach the coverage gap? How do they behave when they reach it? Do they stop taking my brand or shift to a generic? If they reduce their adherence at the end of the year, do they come back when their cumulative drug spend resets in January?

The conclusions so far:

1) It is possible, although very difficult, to measure the impact of a coverage gap on specific therapeutic classes and brands When you do so, it is quickly apparent that far fewer prescriptions require patients to pay the full price than most manufacturers expect, thanks to employer supplemental coverage, state and manufacturers' patient assistance programs, other third party coverage, and the Low Income Subsidy (LIS) program. (LIS patients use about half of the drugs prescribed under Part D and pay almost nothing.)

2) The industry has been asking the wrong question "What happens when my patient crosses the coverage gap threshold?" is not the only relevant question. Rather, how does the existence of a gap in branded or generics coverage affect overall patient adherence in Part D? It turns out that the largest impact of the coverage gap is often found among patients who avoid the coverage gap by stretching their prescriptions or dropping some medications—thus never reaching the threshold.

3) The impact of the coverage gap will be very different by therapeutic class and brand It depends on whether or not a widely accepted generic or an OTC alternative is available.

Measuring the Impact

By tracking patients longitudinally, and ensuring that we have captured all of their prescriptions, Amundsen has been able to mimic the Part D insurers' calculation of cumulative drug spend and out-of-pocket cost sharing for a given patient, thereby having a reasonable estimate of when they should have reached the $2,400 limit in 2007 ($2,250 in 2006).

From the 3 million patients in the Verispan/Amundsen coverage gap database, it was estimated that approximately 72 percent of nonsubsidy patients and 83 percent of total enrollees do not have a coverage gap, or do not reach the cumulative drug spend that would put them in the gap. LIS patients, who had very distinctive branded copays of $3.10 and $5.35 in 2007, and who have no exposure to the coverage gap, account for about 38 percent of enrollees and frequently are a higher proportion of the prescriptions in any given class.

Of the remaining 18 percent of patients who actually reach the $2,400 threshold, about half are actually paying full price for all their drugs. The remaining half have either full coverage or some gap coverage. Plans with some gap coverage may be "generics only," or have identified a number of branded drugs that are apparently exempted. The number having full coverage in the gap remains a bit of a surprise, and the explanation for its importance varies by therapeutic class and Part D region.

These percentages will be very different for any cohort of patients who might be taking a particular class of drugs. Take branded proton pump inhibitors (PPIs) as an example. Because branded PPIs have relatively high negotiated prices, and because PPI patients in Part D have many comorbid conditions, a higher percentage of PPI patients reached the coverage gap in 2007—29 percent, compared with 18 percent in the group as a whole. Of the nonsubsidy PPI patients who reach the threshold, more than one-third will have coverage for virtually all of their medications. Another group will have some gap coverage for generics only, or for some branded products.

To measure how much impact the coverage gap has had on each of the cohorts that experience exposure to the coverage gap, Amundsen used a standard methodology that benchmarks the persistence, compliance, adherence, and generics substitution of each exposed cohort against a group that should not be affected—those who have full coverage in the gap. For example, we would expect to see a reduction in the share of branded PPIs as plans and consumers put more emphasis on generic omeprazole. Both the LIS cohort and those PPI patients with full coverage increased their utilization of generics by about 5 market share points over the 12 months of 2007. Those in the exposed cohorts, on the other hand—those with no gap coverage, some branded coverage, or generics coverage—shift 10 to 15 market share points to generics over the same time period.

Asking the Wrong Question

What pharmaceutical executives really want to know is how many days of therapy they might be losing from Part D patients as a result of this feature of the benefit design. In seeking this information, most marketing and sales people in the industry have focused their attention solely on those who may have exceeded the coverage threshold for a given year.

But to truly understand the impact of the coverage gap, you have to consider other groups as well—particularly those who don't reach the $2,400 limit, LIS patients who have no gap in coverage, and the sickest Part D patients who have spent $3,850 out of pocket and thus qualify for catastrophic coverage with lower copays. For each patient cohort you want to know:

How many? Number of patients, average number of prescriptions per year

How much? Percentage of days of therapy lost after reaching the coverage gap?

For how long? Number of months affected

In analysis of several therapeutic classes, the group that shows the most change has often been the cohort that never reaches the coverage gap. For most therapeutic classes, this is the largest cohort. They cut back refills the most, reducing their days of therapy. And they start to change their behavior earlier in the calendar year. The "avoidance cohort," as it is called, typically start out taking many different drugs each month, and the majority of these patients spend more than $200 a month in the first half of the year. By mid-year, however, these patients change their behavior significantly. They drop therapy or switch to an OTC alternative (which can't be tracked in retail Rx data). They reduce their compliance and stretch out their prescriptions. They are also more likely to switch to a generic if it is available in the classes of drugs they are using.

Patients who reached the $2,400 threshold in 2007 got there for a reason. They were compliant with their medications, and filled enough prescriptions to get into the gap early in the year. In terms of days of therapy lost, however, they can be less important than the avoidance cohort. For one thing, there are far fewer of them. And on average, they reduce usage less and are affected for the shortest periods of time.

Patients who started the year on a branded PPI product, but didn't spend enough on all prescriptions to reach the coverage limit, were just as likely to drop off therapy as were patients who reached the coverage gap and faced full prices. Patients with coverage in the gap—from their LIS status or some other coverage—were far more likely to be persistent.

When you calculate—How many? How much impact? How long?—for each of these cohorts, you can determine the overall impact of the gap in coverage on the nonsubsidy patients in a therapeutic class. In the PPI class, for example, about 2 percent of days of branded therapy were lost among nonsubsidy Part D patients as a result of the coverage gap.

Variation by Therapeutic Class and Brand

Although this analysis has been conducted for a wide range of therapeutic classes and products, we have been unable to predict, a priori, which patients will be most impacted by the coverage gap. However, the coverage gap can be less of an issue in classes in which LIS enrollees account for a disproportionately high percentage of the Part D prescriptions. For example, LIS patients may account for more than 75 percent of prescriptions filled for atypical antipsychotics, such as Zyprexa (olanzapine). Conversely, the presence of a widely accepted generics alternative, such as omeprazole or simvastatin, is a good predictor of a class that will be impacted more significantly. Other factors that may influence the importance of the coverage gap include the existence of manufacturers' patient assistance programs, the legacy effects of state patient assistance programs that might have granted preferred status to a particular product, and the availability of OTC alternatives that would be acceptable to the consumer—such as allergy products.

Will patients who hit the coverage gap in 2007 come back to the brand when their cumulative drug spend resets to $0? Will patients who were exposed to the coverage gap last year learn how to manage around it and get smarter every year? Will manufacturers find ways to respond to the coverage gap in a cost-effective way?

Like our clients, we want to lift another veil from the coverage gap mystery, and will do so as 2008 data become available.

Mason Tenaglia and Lauri Mitchell are directors of the Amundsen Group. They can be reached at MTenaglia@Amundsengroup.com and LMitchell@Amundsengroup.com. Keith Mandia is senior director of managed care product management for Verispan. He can be reached at Keith.Mandia@Verispan.com