Examining Part D Beneficiaries' Medication Use in the Doughnut Hole

Publication
Article
AJPB® Translating Evidence-Based Research Into Value-Based Decisions®Spring 2009
Volume 1
Issue 1

The authors found that the Part D doughnut hole significantly reduced the use of medications for treating potentially disabling and life-threatening conditions, but increased out-of-pocket costs for Medicare beneficiaries.

The Medicare Part D prescription drug benefit was introduced on January 1, 2006. The Part D program provides an outpatient prescription drug benefit for beneficiaries over the age of 65 years and individuals under age 65 years with certain disabilities.1-3 The program has significant potential to improve access to prescription medications.4-8 For many beneficiaries who previously lacked prescription drug coverage, Part D represents a new opportunity to obtain financial assistance for increasingly costly medications.9,10

Although the Part D program is voluntary, penalties apply for eligible individuals who sign up after the registration deadline. Part D benefits can be administered through stand-alone prescription drug plans, managed care Medicare Advantage prescription drug plans, or commercial administrators. This creates a potential for biased risk selection.11

Under the standard Medicare Part D program, beneficiaries are subject to cost-sharing requirements including deductibles, coinsurance, and copays. In 2006, there was a $2850 “doughnut hole,” or coverage gap, for prescription spending between $2250 and $5100. While in the coverage gap, beneficiaries are responsible for the full cost of their drugs, up to $5100. Beneficiaries pay $3600 in true out-ofpocket (OOP) costs until the coverage gap ends. Once a beneficiary reaches $5100 in total prescription drug costs, catastrophic coverage begins. Beneficiaries then pay 5% of any additional covered drug costs for the rest of the year.12,13 The calculations begin anew every year.

It is clearly beneficial for enrollees to stay out of the doughnut hole, especially for those who have partial coverage or no coverage in the doughnut hole. Previous studies have shown that many Part D beneficiaries who reached the coverage gap in prescription drug plans with capped benefits reduced their prescription spending.14-20 Reducing the use of essential medications or discontinuing long-term medications may put beneficiaries at greater risk for poor health outcomes.21-33 Hsu et al found that, compared with those whose benefits were not capped, Medicare beneficiaries with caps on drug benefits were more likely to be nonadherent to long-term drug therapy and had poorer control of blood pressure, lipid levels, and glucose levels, and higher relative rates of emergency department visits, hospitalizations, and death.21

Our study adds to existing literature on how discontinuities in the drug coverage affect beneficiary spending. This research builds on earlier studies that examined the impact of Medicare Part D on overall prescription drug utilization and spending.34-38 The impact is likely to vary by drug class, disease state, severity of disease, and nature of disease (ie, acute, chronic, or asymptomatic).

Assessing the implications of coverage gaps on drug utilization among various therapeutic classes may aid policymakers and benefit administrators in designing new benefits or modifying existing structures. It also may assist providers in helping patients achieve better health outcomes. Such an assessment also may enhance understanding of spending behavior of beneficiaries in the doughnut hole, particularly as the coverage gap is projected to grow by 9% to 10% per year to a level of $5066 in 2013.39 Understanding the effects of such a coverage gap is a necessary step toward fully understanding the implications of the Part D program benefit design.7,40,41

Relatively few studies have examined the impact of price differentials between Medicare beneficiaries and nonbeneficiaries. Using a large claims database from a national pharmacy benefit manager, we assessed the use of medication by therapeutic class, as measured by days of therapy, OOP spending, and generic drug utilization rates in a population affected by the Part D coverage gap as compared with a population enrolled in plans without a coverage gap.

METHODS

Data Source and Inclusion Criteria

Deidentified prescription claims data for the period from January 1, 2006, through December 31, 2006, were obtained from a large pharmacy benefit management database. The study was conducted in full compliance with Health Insurance Portability and Accountability Act regulations.42 Subjects included in the study were 65 years of age or older as of January 1, 2006, were continuously enrolled in either the Part D plan or a commercial plan, and had total prescription drug spending of more than $2250 in 2006.

Study Design and Outcomes Measures

A retrospective pre—post with control group cohort design was used. The study group included beneficiaries who had no supplemental prescription drug coverage when they reached the coverage gap and who did not reach the threshold for catastrophic coverage in 2006. The control group included those who were enrolled in commercial prescription plans. The preperiod was defined as the time period before beneficiaries reached the doughnut hole (total prescription drug spending <$2250); the postperiod was the time period after beneficiaries reached the coverage gap (total prescription drug spending >/=$2250).

The top 10 therapeutic classes, by their share of overall total spending in 2005, were included in this study: antihyperlipidemics, antihypertensives, antidepressants, antiulcerants, antidiabetic drugs, antiasthmatics, anti-inflammatory analgesics, opioid analgesics, anticonvulsants, and antivirals. Additionally, 2 therapeutic classes of special interest were included in the analysis: antipsychotics and antineoplastics, which represent essential treatments for psychoses and cancers, respectively. Average prescription days of therapy per utilizing beneficiary per month, OOP costs per utilizing beneficiary per month (OOP costs did not include premium payments), and generic drug utilization rates were the targeted outcomes measures.

Statistical Analysis

Descriptive analysis was performed to examine the distribution patterns of demographic characteristics and clinical conditions within the study groups. Disease states were inferred based on Generic Code Number sequence numbers from First Data Bank. Direct comparisons of the use of medications by therapeutic classes were made between the preperiod and postperiod. A difference-indifference (DID) study approach38,43,44 was used to estimate how the Medicare Part D prescription drug coverage gap affected the use of medications among therapeutic classes, by adjusting for demographic characteristics, disease states, and secular trends in the control group. The DID model was specified as follows:

Yjit = α0 + α1*dt + α2*dj + β*djt + δ*zjit + εjit

In the model, Yjit represents the outcome of interest (days of therapy, OOP costs, or generic utilization rate) at time t (preperiod: t = 0; postperiod: t = 1) and group j (study group: j = 1; control group: j = 0); zjit represents individual i’s characteristics (age, sex, and disease states); dt is a dummy for time: dt = 1 if t = 1 and 0 otherwise; dj is a dummy for group: dj = 1 if j = 1 and 0 otherwise; djt is a dummy: djt = 1 if t = 1 and j = 1, and 0 otherwise; α1 represents time effect (pre and post); α2 represents fixed effect of the group; δ captures the effect of demographics; β is the DID estimate accounting for difference owing to time (α1) and group variation (α2); and εjit is an error term. All analyses were performed using SAS version 9.1 (SAS Institute, Cary, NC).

RESULTS

Descriptive Statistics

A total of 138,612 beneficiaries met the selection criteria and were included in this study (Table 1). The mean age (±SD) was 75.91 (±7.41) years, and 63.79% were women. On average, beneficiaries filled 8.41 (±3.53) total maintenance medications and had 4.90 (±1.71) chronic disease states. Overall, 70.98% beneficiaries had a nonspecific cardiovascular condition, 67.18% had hyperlipidemia, 43.36% had hypertension, 35.23% had pain and inflammation, and 33.21% had gastric disorders. As compared with beneficiaries in the control group, those in the study group were older (76.18 vs 73.76 years; P <.001), and had lower total medication costs ($3317 vs $3650; P <.001) but higher OOP expenses ($1743 vs $832; P <.001). A higher proportion of beneficiaries were women (65.10% vs 53.26%; P <.001). In the study group, a higher proportion of beneficiaries had a nonspecific cardiovascular condition (71.18% vs 69.37%; P <.001), but the proportions of beneficiaries were lower for hyperlipidemia (66.80% vs 70.16%; P <.001), hypertension (43.15% vs 45.01%; P <.001), pain and inflammation (35.11% vs 36.16%, P = .011), and gastric disorders (32.75% vs 36.86%; P <.001) (Table 1).

Table 2 shows the changes in average days of therapy for each drug class per utilizing beneficiary per month for both the study group and control group. From the pre- to the postperiod in the study group, antineoplastics had the greatest percentage decrease (from 23.56 to 19.19, or −18.55%; P <.001), followed by antidiabetic drugs (from 40.13 to 33.27, or −17.09%; P <.001), antihyperlipidemics (from 27.77 to 23.31, or −16.06%; P <.001), antihypertensives (from 29.18 to 25.81, or −11.55%; P <.001), antiulcerants (from 20.58 to 18.66, or −9.33%; P <.001), antiasthmatics (from 25.53 to 23.23, or −9.01%; P <.001), and antidepressants (from 24.65 to 23.41, or −5.03%; P <.001). However, the average days of therapy increased in the study group for antivirals (from 3.90 to 5.60, or 43.59%; P = .661), opioid analgesics (from 8.26 to 10.40, or 25.91%; P = .542), anti-inflammatory analgesics (from 14.64 to 15.76, or 7.65%; P <.001), antipsychotics (from 17.19 to 17.61, or 2.44%; P = .114), and anticonvulsants (from 19.51 to 19.75, or 1.23%; P = .183). In the control group, either the decrease in the average days of therapy was smaller than that in the study group or the average days of therapy increased.

Table 3 shows the changes in OOP costs of each drug class per utilizing beneficiary per month included in the analysis. From the pre- to the postperiod, the study group’s average OOP costs increased in all drug classes, with increases ranging from 47.20% to 340.77%. The OOP costs more than doubled for opioid analgesics (from $3.36 to $14.81, or 340.77%; P <.001), antivirals (from $6.78 to $29.88, or 340.71%; P <.001), antipsychotics (from $18.61 to $70.56, or 279.15%; P <.001), and antineoplastics (from $24.78 to $93.40, or 276.92%; P <.001). In the control group, however, OOP costs decreased in most of the classes.

Table 4 shows the changes in the generic utilization rate for each drug class. From the pre- to the postperiod, the study group’s generic utilization rate increased in all classes except antipsychotics and antiasthmatics. The 5 medication classes with the highest percentage increases were antihyperlipidemics (from 14.21% to 31.93%, or 124.70%; P <.001), antivirals (from 25.03% to 36.68%, or 46.54%; P <.001), antidepressants (from 48.87% to 64.68%, or 32.35%; P <.001), anti-inflammatory analgesics (from 52.14% to 62.71%, or 20.27%; P <.001), and antiulcerants (from 27.55% to 32.57%, or 18.22%; P <.001). In the control group, the generic utilization rate also increased for antihyperlipidemics, antidepressants, antivirals, and anti-inflammatory analgesics, but the magnitude of increase was much smaller than that in the study group.

Difference-in-Difference Regression Analysis

Table 5 shows the results from the DID model with dependent variables of days of therapy per utilizing beneficiary per month, OOP costs per beneficiary per month, or generic medication utilization rate by each therapeutic class after controlling for secular trends, demographics (age and sex), and disease states. The Part D coverage gap reduced medication utilization in all classes except antivirals and opioid analgesics. For example, the Part D coverage gap was estimated to have reduced antineoplastics drug utilization by 2.70 days of therapy (P <.001) while raising OOP costs by $70.03 (P <.001) and increasing the generic utilization rate by 4.43 percentage points (P = .120).

DISCUSSION AND IMPLICATIONS

Access to Medications in the Coverage Gap

Implementation of the Medicare Part D program may have improved access to prescription medications for some,6-10 but the program poses significant challenges such as coverage gap, late penalties, and numerous plans and various benefit designs to choose from.45,46 One must bear in mind that because the elderly population has multiple comorbidities, picking a plan is a very difficult proposition. The doughnut hole was introduced into legislation as a means of reducing federal spending. Non-Medicare plans, for the most part, do not have such a large gap in benefits. Therefore, seniors do not expect to fall into the gap when they switch over to Part D. Additionally, premiums paid by beneficiaries while in the doughnut hole do not count toward the OOP limit. The Part D program devotes substantial resources to provide cost-sharing assistance for low-income and former Medicaid beneficiaries.47 The coverage gap impacts middle-class and certain disabled beneficiaries who do not qualify for federal or state government aid, and have fixed incomes.48

This study used a commercial population that had private insurance as a comparison group to examine the differential impact of the coverage gap on Medicare beneficiaries by therapeutic classes. Medicare beneficiaries decreased utilization of medications in all therapeutic classes except antivirals and opioid analgesics. Opioid analgesics are prescribed for acute and chronic pain, especially for managing acute pain after surgery.49 Without access to the medical records of these beneficiaries, it is difficult to propose any particular reason for the continued use of this class in the doughnut hole. By far, the significant reduction was seen in cancer, asthma, diabetes, depression, and lipid management drugs (Table 5). Cancer drugs or antineoplastics are very expensive. In the past, most cancer therapies were administered in doctor’s offices and billed as Part B. However, with the advent of newer oral treatments, most oral cancer drugs are now covered under Part D. There are relatively few oral treatments and even fewer generic substitutes to control costs in this category. This class also includes drugs used to counteract the extreme side effects of antineoplastics. Cancer diagnoses often are sudden, and the high cost of these drugs might well become a significant financial burden, especially for beneficiaries with other comorbidities. This possibility is confirmed by the findings in this study that OOP costs for cancer therapies in the Part D coverage doughnut hole versus the previous co-pay coverage increased by $70.03.

In the study group, the patients were slightly sicker, averaging about 4.9 disease conditions compared with nearly 4.8 disease conditions in the control group. The control group was a commercial population, meaning they had either a primary or alternate source of income to help pay for healthcare expenses. By contrast, most Medicare beneficiaries are retired. For many, Social Security payments constitute a large part of their income. The presence of a coverage gap exacerbates their financial burden. Financial barriers to access are significant and may lead beneficiaries to make healthcare decisions based primarily on cost.

Effects of Beneficiary Spending Behavior

The measure used to analyze beneficiary spending behavior was OOP spending, which includes deductibles, copays, and coinsurance, but not premiums. The control group, comprising non—Part D enrollees, spent less OOP than the Part D enrollees. Intuitively, this makes sense because standard benefit Medicare beneficiaries did not have any coverage while in the coverage gap. The non-Medicare beneficiaries had private insurance or equivalent coverage. However, for the Medicare population, the near doubling of OOP costs while in the doughnut hole is of concern. The study results suggest that Medicare beneficiaries were spending less on treatments for potentially disabling and life-threatening conditions—hyperlipidemia, diabetes, asthma, depression, and cancer. However, there was a trend toward higher spending for potentially less life-threatening conditions such as chronic pain. These findings are similar to the findings of Goldman et al regarding cost-related medication decreases in younger patients.32

Financial burden also appeared to be higher in the antipsychotic class, with a reduction in the number of days of supply and concurrent increases in OOP spending. People with mental illnesses tend to have high prescription drug expenditures.50 From our analysis, we can safely conclude that the impact of the Part D coverage gap differs by therapeutic class. Further research is required to ascertain how Medicare beneficiaries select treatments and minimize their OOP costs while in the coverage gap.

Brand-to-Generic Switching

Brand-to-generic switching is perhaps the least studied measure in the Medicare population. Part D beneficiaries were more likely than nonbeneficiaries to receive generic medications while in the coverage gap. The highest rate of generic dispensing was seen with antihyperlipidemics. The 124.70% change may be partly attributed to the availability of generic simvastatin in June 2006. The increase in generic dispensing is indicative of Part D beneficiaries using strategies to lower their OOP costs. Non-Medicare beneficiaries showed little or no brand-to-generic switching during the same period.

The coverage gap impact differs for various beneficiaries. Those with higher prescription drug spending would be more likely to reach the doughnut hole sooner and move to catastrophic coverage earlier than beneficiaries with lower or moderate drug spending. Any coverage gap or discontinuity in benefits poses a financial burden for beneficiaries, often encouraging individuals to follow cost-minimizing strategies that may not be clinically sound.

It is important for providers to know who is most likely to be at risk. In this study, it appears that patients with major chronic illnesses such as diabetes, hyperlipidemia, asthma, cancer, or depression, or any combination of these conditions, are most likely to fall into the coverage gap and are more likely to try reducing their OOP spending. Clinicians and pharmacists can easily monitor when a patient reaches the coverage gap and work with the patient to reduce the impact of the sudden financial burden. Beneficiaries affected by the coverage gap may be likely to purchase additional prescription drug coverage. These supplemental plans often carry higher premiums to compensate for the reduced cost sharing in the coverage gap. It is understandably difficult to predict how beneficiaries might adjust their spending or whether their choices are likely to be beneficial. However, by being proactive, clinicians and pharmacists may be able to intervene to help ensure that beneficiaries have continued access to their medication, especially by encouraging the use of appropriate drug regimens and maximizing generic utilization.

Results from this study are generally consistent with previous findings from similar studies. In a recent study based on retail pharmacy claims data from IMS Health, Hoadley et al found that 20% of enrollees either stopped taking medications, reduced their medication use, or switched medications when they reached the coverage gap in 2007.51 Due to the major differences in the selected drug classes, outcomes measures, claims databases, and study methods, a direct comparison of the findings cannot be made between the Hoadley et al study and ours.

Study Limitations

This study has several limitations. First, the study was based on beneficiaries enrolled in a pharmacy benefits management program, which may not be wholly representative of the national population. Second, the findings were based on a unilateral measure. An analysis of beneficiaries’ related medical claims information is needed to understand clinical outcomes associated with beneficiaries’ efforts to reduce their OOP spending. Third, the study did not examine each individual’s prescription drug prior authorization and formulary structure, which can affect drug utilization.47,52 To account for confounding through various plan and pricing structures, we applied a DID study approach that controls for the effects of all time-varying determinants of prescription drug use and costs common to the study and control groups.

This study only examined beneficiaries who reached the coverage gap during 2006. Many beneficiaries who did not reach the gap during the year were not included in the study. It remains unclear whether the doughnut hole had any effect on this population. It is possible that these beneficiaries implemented cost-cutting measures similar to those implemented by beneficiaries who did enter the coverage gap. With the coverage gap approaching, they could have decreased medication use or used less costly alternatives to avoid reaching the gap. Nevertheless, the doughnut hole problem is concentrated among a relatively small group of beneficiaries whom policy remedies should focus on.

This study found that standard Medicare Part D beneficiaries in the coverage gap significantly reduced their use of medications for the treatment of potentially disabling and life-threatening conditions such as cancer, asthma, diabetes, depression, and hyperlipidemia. At the same time, beneficiaries did not decrease spending for potentially less life-threatening conditions such as chronic pain. These findings raise concerns about the risk of adverse health events in Part D beneficiaries who are in the coverage gap. With the rising cost of prescription drugs and a widening coverage gap each year, this problem is likely to continue. Standard-benefit beneficiaries are financially overwhelmed, and the choices that they make in the doughnut hole will likely affect their health outcomes. Close monitoring of chronically ill patients with multiple comorbidities is important. Clinicians may need to pay attention to the therapy adherence of their chronically ill Medicare patients. Optimal health outcomes are more difficult to achieve when patients cannot afford medications and reduce therapy or forgo therapy altogether.

It is important for beneficiaries to understand the coverage gap and the importance of continuing medication therapy so that they can move toward the generous catastrophic coverage. Further research is needed to determine whether a coverage gap followed by generous coverage produces better long-term health outcomes than moderate coverage with no coverage gaps at a population level.

Related Videos
Practice Pearl #1 Active Surveillance vs Treatment in Patients with NETs
© 2024 MJH Life Sciences

All rights reserved.