J. Samantha Dougherty, PhD

 To assess the extent to which medication adherence in congestive heart failure (CHF) and diabetes may serve as a measure of physician-level quality.

 A retrospective analysis of Medicare data from 2007 to 2009, including parts A (inpatient), B (outpatient), and D (pharmacy).

 For each disease, we assessed the correlation between medication adherence and health outcomes at the physician level. We controlled for selection bias by first regressing patient-level outcomes on a set of covariates including comorbid conditions, demographic attributes, and physician fixed effects. We then classified physicians into 3 levels of average patient medication adherence—low, medium, and high&mdash;and compared health outcomes across these groups.

 There is a clear relationship between average medication adherence and patient health outcomes as measured at the physician level. Within the diabetes sample, among physicians with high average adherence and controlling for patient characteristics, 26.3 per 1000 patients had uncontrolled diabetes compared with 45.9 per 1000 patients among physicians with low average adherence. Within the CHF sample, also controlling for patient characteristics, the average rate of CHF emergency care usage among patients seen by physicians with low average adherence was 16.3% compared with 13.5% for doctors with high average adherence.

 This study’s results establish a physician-level correlation between improved medication adherence and improved health outcomes in the Medicare population. Our findings suggest that medication adherence could be a useful measure of physician quality, at least for chronic conditions for which prescription medications are an important component of treatment.

In both the diabetes and congestive heart failure disease spaces, average levels of patient medication adherence (calculated at the physician level) are significant predictors of patient health.

Average rates of hospitalization, emergency care, and comorbidities are lower among patients treated by physicians with high average adherence rates; this correlation persists after controlling for individual patient characteristics.

The utility of average medication adherence as a potential measure of physician quality should be examined.

Measuring the quality of care offered by healthcare providers is an important tool for effectively promoting value in the US healthcare system. New quality metrics are constantly being developed and refined, and they are increasingly used in performance-based contracting. From the standpoint of providing incentives to increase the value of care, ideally these measures would directly observe the incremental value of care—in economic terms, the “marginal product of care”&mdash;and link reimbursement to the difference between incremental value and incremental cost. However, this direct effect is not easy to measure. Quality metrics sometimes attempt to do so by capturing various elements of healthcare delivery, including process, outcomes, and patient satisfaction.

More recently, quality measures have been developed to evaluate the prescribing patterns, use, and consequences of prescription medications. Some of this growth reflects the increasing availability of medications to treat and prevent chronic illnesses and the growing consensus that adherence is an important part of disease management. Study results routinely show that proper adherence can lead to lower net healthcare costs, particularly for patients with cardiovascular disease.

 Recent work also demonstrates that some health plans are systematically associated with higher medication adherence and better patient outcomes.

 This evidence coincides with the growing use of adherence measures in pay-for-performance systems such as the Medicare Advantage Star Rating System.

There is also a growing effort to increase quality monitoring for individual physicians. As part of the 2010 Affordable Care Act (ACA), CMS reports information on individual physician quality through the Physician Compare Initiative.

 More recently, the Medicare Access and CHIP Reauthorization Act of 2015 created the Quality Payment Program, through which some participating physicians receive reimbursement modifications based on reported quality metrics, including some based on adherence. However, despite the use of these metrics, it is not well established whether adherence varies systematically across physicians or correlates with the quality of medical care provided. Although the association between adherence and outcomes has been established generally, it is possible that at the individual physician level the signal to noise ratio is too weak to make it a useful reimbursement tool. More work is needed to understand if some physicians are better than others in terms of promoting medication adherence and, if so, what this means for patient outcomes.

This study compares average patient medication adherence with select outcomes-based quality measures at the physician level using Medicare claims data. We do this for all patients in a nationally representative sample of Medicare data who have diabetes or congestive heart failure (CHF), 2 expensive chronic conditions that require extensive medication management to treat. We examine whether there are systematic differences in patient adherence across physicians and how those differences are associated with improved patient outcomes. Although this analysis cannot indicate a causal relationship between physician behavior and medication adherence and patient outcomes, it can evaluate whether any association exists between systematic physician-level differences in medication adherence and patient outcomes. Establishing that such a relationship exists is a necessary first step toward validating the use of physician-level measures of medication adherence in incentive-based reimbursement schemes.

This study uses linked Medicare parts A (inpatient), B (outpatient), and D (pharmacy) claims data from a nationally representative, randomly selected 20% sample of Medicare beneficiaries 65 years and older with fee-for-service coverage of Medicare parts A, B, and D. Additionally, beneficiary enrollment, demographic information, and vital status come from the Medicare Denominator File. The inpatient data provide information on all hospital stays, including length of stay, the diagnosis-related group associated with the stay, and up to 10 individual procedure codes and diagnostic codes. The outpatient data include information on outpatient hospital visits, home hospice care, and the use of durable medical equipment. It also includes all claims submitted by physicians and other health providers, including physician office visits. The Part D prescription drug information provides information on prescription drug events, including the National Drug Code, number of days supplied, and date of service. These Part D data are linked with individual beneficiaries’ vital status from the Denominator File, which contains geographic identifiers, date of birth, date of death, gender, race, and age. Because Part D did not begin until 2006, we restrict our analysis to patients from 2007 through 2009. Although these data are older, they provide a useful perspective because they are based on a period prior to the passage of the ACA, when physician prescribing patterns were likely unaffected by the recent drive toward improved quality metrics.

We provide a brief summary of our analytical methods in this paragraph. Complete details are provided in the technical 

). We assess the relationship between adherence and several measures of patient health (including rates of hospitalization, emergency department [ED] care, and comorbidities), all calculated at the physician level. The full set of outcomes considered is listed in 

. To accomplish this, we first used claims data to generate a measure of patient-level adherence, the proportion of days covered (PDC). We next executed a series of patient-level regressions, in which 2 classes of outcomes (medication adherence and quality measures based on patient health) were modeled as a function of patient characteristics (including age, gender, and comorbid conditions) and physician fixed effects. We then predicted physician-level adherence and quality values from these regressions, calculated at the population mean of those patients. This produced measures of physician-level adherence and quality that controlled for observable differences in physician patient populations. Next, we regressed the generated physician-level quality measures on the generated physician-level measures of medication adherence. Because we employed a 2-stage model, we used bootstrap methods to estimate standard errors. Finally, we examined the relative extent to which physician fixed effects and individual patient comorbidity profiles predict patient health outcomes, using analysis of variance.

, we report differences in patient characteristics, aggregated to the physician-year level. The table shows average medication adherence for patients with diabetes and patients with CHF, patient outcomes, number of patients per physician-year, and mean total annual medical expenditures. Our measure of medication adherence is higher in the diabetes sample than the CHF sample, with a mean of 40.2% PDC (interquartile range [IQR], 30.8%-50.0%) for patients with diabetes compared with 36.2% for patients with CHF (IQR, 27.3%-44.7%). The private-sector patients considered by Seabury et al

 showed similar results in that adherence was higher for patients with diabetes than those with CHF, but the average adherence levels were much higher in the private-sector data, perhaps reflecting the sicker nature of patients in Medicare. The average medical expenditures per patient-year are high: $39,954 for patients with diabetes and $54,723 for patients with CHF. Complication rates and spending levels in this analysis are higher than those seen in many other studies because this sample is composed of patients who saw a physician at least once in the year, restricted to physicians who see a relatively large number of patients with diabetes or CHF.

The average physician-year in the data covered 52 patients in the diabetes sample and 34 patients in the CHF sample. These figures are skewed by a subset of physicians who see a large number of patients; the physician in the 90th percentile saw 103 patients with diabetes and 65 with CHF in a year.

 compares the differences in health outcomes with average adherence across physician-years in the diabetes sample. The top panel reports the unadjusted association between outcomes and adherence, whereas the bottom panel reports the association between outcomes and adherence adjusted for patient characteristics, including age, sex, and comorbidities. For each outcome, the complication rate is averaged across physicians with low, moderate, or high average medication adherence among their patients. The results show a clear relationship between the average adherence and outcomes of a physician’s patients with diabetes. In the top “unadjusted” panel, physicians whose patients have low adherence have 49.6 patients per 1000 with uncontrolled diabetes compared with 22.1 per 1000 among physicians with high medication adherence. A similar relationship, in terms of both direction and magnitude, holds for ED visits and short-term and long-term complications. The magnitude of the effect diminishes following adjustment for patient characteristics: 45.9 per 1000 patients of physicians with low average adherence have uncontrolled diabetes compared with 26.3 per 1000 patients of physicians with high average adherence. In all cases, the bootstrap tests found that groupwise differences were statistically significant (

 reports analogous results for the CHF sample. As with the diabetes sample, patients of physicians with better average adherence also experienced better average outcomes. The unadjusted CHF hospitalization rate of patients of the low-adherence physicians was 51.8% compared with 45.7% for patients of physicians with high average adherence, a difference of 6.1 percentage points or 11.8%. In terms of CHF-related ED visits, the rate was 16.4% for patients of physicians with low adherence compared with 13.5% for physicians with high average adherence, a 17.8% difference. These results are consistent in terms of direction and magnitude for hospitalizations and ED visits for the other conditions. Comparing the top and bottom panels indicates that the regression adjustment for age, gender, and comorbidities has very little effect on the magnitude of the difference across physicians with low-adherence or high-adherence patients. As in the diabetes sample, in all cases the differences are statistically significant at the 1% level (

 compares the proportion of the variance in the regression models of each medication adherence measure and outcome that is explained by physician fixed effects or patients’ own comorbidity profiles. Physician fixed effects consistently account for a much larger portion of the explained variation than do individual comorbidities. In the diabetes sample, physician fixed effects account for about 15% to 20% of the explained variation in medication adherence and 3% to 5% in outcomes. This is considerably more than patient comorbidities, which generally account for less than 1% of the explained variation in outcomes. For the CHF sample, physician fixed effects account for somewhat less of the explained variation in adherence (about 15%-18%) but more of the explained variation in outcomes (about 8%-11%). However, patient comorbidities still account for very little of the explained variation, typically 2% or less. One possible explanation is that the patient comorbidity profile is a relatively weak predictor of patient health, although some comorbidities (eg, depression) are strongly indicative of worse outcomes. In any case, the analyses from Tables 3 through 5 suggest that there are unobserved characteristics of physicians that systematically predict medication adherence and, further, that unobserved characteristics are predictive of patient outcomes. Of course, which unobserved characteristics led to better patient adherence and which led to better patient outcomes remains unknown.

Our analysis finds a strong correlation between medication adherence and patient outcomes at the physician level. This suggests that measuring the medication adherence of a physician’s patients can provide a signal of provider performance as measured by patient outcomes. More effort should be given to developing physician-level and/or disease-specific quality indications based on adherence and to evaluating whether the use of these measures could improve care and outcomes for patients with chronic diseases.

Although the relationship between improved adherence and better outcomes is well established at the patient level, it has not been established that the same relationship holds for individual physicians.

 Recent work shows that adherence can be a similarly useful measure of health plan performance.

 Prior work also shows that physicians with different measured quality also vary systematically in their prescribing patterns. For example, patients with type 2 diabetes of physicians who routinely prescribe metformin as frontline therapy have better outcomes than those of physicians who prescribe sulfonylureas.

 However, in this study we explicitly documented a correlation between improved patient outcomes and improved adherence to medication in the case of both diabetes and CHF.

Unfortunately, our data do not cast light on 

 some physicians have patients with better medication adherence. Evidence suggests that many physicians feel untrained or ill equipped to promote medication adherence.

 If that is true, it is possible that the systematic differences in medication adherence across physicians could reflect something intrinsic about the physician—patient relationship. Past work has found that good physician–patient relationships were important for promoting medication adherence.

 Other work has found that physician characteristics, such as job satisfaction and propensity to schedule follow-up visits, were associated with improved adherence.

 At the very least, the existing evidence is clear that there is an association between a physician’s engagement with patients and better medication adherence by patients. Future work should better establish the activities or qualities of physicians that promote medication adherence and use these to develop physician guidelines to help promote these more broadly. Additionally, there is likely correlation between average adherence at the physician level and average adherence at the pharmacy level; future work should investigate which of these levels is better able to influence adherence.

 Our findings suggest that medication adherence measures have the potential to be a useful quality measurement tool, at least for chronic conditions for which prescription medications are an important component of treatment. There are many potential advantages to using medication adherence as a quality metric. For one, although medication use is imperfect, there are commonly accepted and objective metrics that can be constructed using pharmacy claims data. Also, medication use is much more common than many types of complications, so it can be more accurately measured and is potentially more informative than a performance metric based on outcomes.

Although more needs to be done to establish a causal result, the cost implications could be large. Seabury et al

 demonstrated that in addition to predicting outcomes, plan-level medication adherence is a strong signal of costs. In other words, plans whose patients had better adherence also had lower total average expenditures among their patients with diabetes and CHF. Although it is unclear how much of the difference between high- and low-adherence physicians is actually due to factors within a physician’s control, our findings suggest the potential for large savings if we can sufficiently train and incentivize physicians to help their patients better adhere to their medication regimens. In the United States in 2014, there were more than 330,000 ED admissions for hyperglycemia or hypoglycemia and about 1 million for CHF.

 Prior studies have estimated the cost of these admissions at around $2000 each.

 If improved adherence reduced the incidence of these events by even a small amount, the cost savings would be large. For example, a 5% reduction in each type of admission would yield an estimated cost savings of almost $130 million. This suggests a need for careful, rigorous evaluation of new initiatives such as the Quality Payment Program to better understand what impact, if any, incentivizing physicians to improve quality has on patient outcomes.

This study has several important limitations. First, our sample was large, but it was limited to Medicare patients. Although we expect the relationship between adherence and outcomes to generalize to commercial patients, this should be demonstrated in future work. Also, the quality measures were limited by the fact that our data are claims based and lack information about disease severity. Some outcomes that we considered are relatively rare, and many physicians had no patients with these complications. Thus, although the relationship between adherence and outcomes is informative in the aggregate, patient outcomes may provide limited information about the performance of any single provider. Further studies might consider how closely average medication adherence predicts more clinically relevant measures of provider performance (eg, average glycated hemoglobin levels). Also, although we defined good adherence as having a PDC above 80%, the level chosen by CMS, it would be useful to evaluate the predictive ability of different threshold values.

Another limitation is that we only measured an association between average adherence and patient outcomes, and this relationship may not be causal. If healthier patients systematically select to certain physicians, it could overstate the correlation between adherence and outcomes due to what is known as the “healthy adherer” effect. For example, Dormuth et al

 found that patients with better adherence also display many other positive health habits, such as safer driving. Although it is not clear that patients systematically select to physicians according to unobserved health factors, if they do, it would significantly impact the interpretation of our findings. If the association that we estimated represents a true causal relationship, then we would expect that tying reimbursement to adherence could improve outcomes for patients by providing physicians with enhanced incentives for better performance. If, on the other hand, high-adherence patients self-select toward some physicians more than others, reimbursing physicians based on adherence would simply reward physicians for something they have limited impact on. This could even generate adverse incentives, if it promoted “cream skimming” and gave physicians incentives to avoid sick patients. We find some evidence to suggest that the selection across providers is relatively minor, at least according to noncardiovascular comorbid conditions. Nevertheless, this accentuates the importance of understanding the mechanisms through which physicians influence adherence, as this will help policy makers devise quality metrics and reimbursement schemes to promote the right incentives to providers and unlock value for patients.

An additional limitation is that it is difficult to precisely measure adherence to injectable medications in claims data, due to the lack of information on titration instructions. This adds some uncertainty to our findings for the diabetes population, but that uncertainty is partially offset by the parallel conclusions in the CHF population. Also, some patients are seen by more than 1 physician in a given year and thus may contribute to the adherence and outcome fixed effects for multiple physicians. We assume that no single patient could create significant correlation across physicians, based on the fact that we required a minimum of 10 patients per physician.

Lastly, we were unable to adjust for the effect of social determinants of health (SDH). Several studies have demonstrated that SDH and patient outcomes are closely related. However, the observational claim-based Medicare database provides very limited information on SDH, which may result in unmeasured confounding.

We sought to assess the potential for utilization of physician-level measures of medication adherence as quality indicators. For both CHF and diabetes and controlling for other factors, we found that the patients of physicians with higher average rates of adherence have better health outcomes than do patients of physicians with low average rates of adherence. This suggests that medication adherence may be a useful indicator of quality of care at the physician level.

 Keck School of Medicine, University of Southern California (SAS), Los Angeles, CA; Pharmaceutical Research and Manufacturers of America (PhRMA) (JSD), Washington, DC; Precision Health Economics (JS), Boston, MA.

 Dr Seabury is a consultant for Precision Health Economics, a life-sciences consulting company. Mr Sullivan is an employee of and holds stock options in Precision Medicine Group, which consults for pharmaceutical companies. The remaining author reports no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

 Concept and design (SAS, JSD); analysis and interpretation of data (JSD, JS); drafting of the manuscript (SAS); critical revision of the manuscript for important intellectual content (SAS, JSD, JS); and statistical analysis (SAS, JS).

 Seth A. Seabury, PhD, USC Schaeffer Center, VPD Suite 414, Los Angeles, CA 90087. Email: seabury@usc.edu.

Crossing the Quality Chasm: A New Health System for the 21st Century

. Washington, DC; National Academies Press; 2001.

2. Tamblyn R, Laprise R, Hanley JA, et al. Adverse events associated with prescription drug cost-sharing among poor and elderly persons. 

. 2001;285(4):421-429. doi: 10.1001/jama.285.4.421.

3. Heisler M, Langa KM, Eby EL, Fendrick AM, Kabeto MU, Piette JD. The health effects of restricting prescription medication use because of cost. 

. 2004;42(7):626-634. doi: 10.1097/01.mlr.0000129352.36733.cc.

4. Goldman DP, Joyce GF, Lawless G, Crown WH, Willey V. Benefit design and specialty drug use. 

. 2006;25(5):1319-1331. doi: 10.1377/hlthaff.25.5.1319.

5. Hsu J, Price M, Huang J, et al. Unintended consequences of caps on Medicare drug benefits. 

. 2006;354(22):2349-2359. doi: 10.1056/NEJMsa054436.

6. Goldman DP, Joyce GF, Zheng Y. Prescription drug cost sharing: associations with medication and medical utilization and spending and health. 

. 2007;298(1):61-69. doi: 10.1001/jama.298.1.61.

7. Thornton Snider J, Seabury S, Lopez J, McKenzie S, Goldman DP. Impact of type 2 diabetes medication cost sharing on patient outcomes and health plan costs. 

8. Seabury SA, Lakdawalla DN, Dougherty JS, Sullivan J, Goldman DP. Medication adherence and measures of health plan quality. 

9. Physician Compare. Medicare.gov website. medicare.gov/physiciancompare. Accessed January 10, 2016.

10. Ghosh A, Simon K, Sommers BD. The effect of state Medicaid expansions on prescription drug use: evidence from the Affordable Care Act [NBER working paper no. 23044]. National Bureau of Economic Research website. nber.org/papers/w23044. Published January 2017. Accessed June 1, 2017.

11. Mulcahy AW, Hollands S, Duffy EL, Strong A, Wynn BO. Modeling the economic impact of a California workers’ compensation formulary. RAND Corporation website. rand.org/pubs/research_reports/RR1971.html. Published 2017. Accessed June 1, 2017.

12. Bonakdar Tehrani A, Cunningham PJ. Closing the Medicare doughnut hole: changes in prescription drug utilization and out-of-pocket spending among Medicare beneficiaries with Part D coverage after the Affordable Care Act. 

. 2017;55(1):43-49. doi: 10.1097/MLR.0000000000000613.

13. Bognar K, Bell KF, Lakdawalla D, et al. Clinical outcomes associated with rates of sulfonylurea use among physicians.

. 2013;19(spec no. 5):SP182-SP186. ajmc.com/journals/evidence-based-diabetes-management/2013/2013-1-vol19-sp5/clinical-outcomes-associated-with-rates-of-sulfonylurea-use-among-physicians. Accessed January 28, 2019.

Articles

Physician-level medication adherence is a strong predictor of patient health and should be considered as a measure of physician quality.

Medication Adherence as a Measure of the Quality of Care Provided by Physicians

To test if offering zero generic co-pays for oral antidiabetic drugs (OADs) and statins increases generic dispensing for low-income subsidy (LIS) recipients with diabetes enrolled in Medicare Part D.

 We analyzed a natural experiment in which LIS recipients were randomized to Part D plans in 2008. Some plans placed selected generic OADs and statins on zero co-pay tiers whereas others did not. Randomization eliminated selection effects which could bias the study findings.

 We analyzed a 5% random sample of Medicare beneficiaries with diabetes from the Chronic Condition Data Warehouse using Part D claims, formulary provisions, and co-pay tiers together with a special file prepared by CMS that identified all randomly assigned LIS recipients in 2008. We calculated proportions using generic drugs in the 2 classes and annual days’ supply among users in plans with and without zero co-pay tiers for the country as a whole and California (where zero co-pay plans were particularly popular).

We found that the demand for generic OADs was not significantly different in plans with and without zero co-pay tiers. By contrast, a large difference was observed in the percent of LIS recipients using generic statins in plans with zero co-pay tiers (61.4% vs 54.6%; P <.01). However, the difference disappeared once we controlled for formulary restrictions on the most popular brand statin at the time (Lipitor).

This cautionary tale suggests that policy makers should give greater consideration to formulary provisions when evaluating the effects of free generics in value-based insurance designs.

An increasingly popular strategy in value-based health insurance designs, including some Part D plans, is to have a preferred generic tier with no co-pays for selected medications offered on the presumption this will shift demand away from expensive brand name drugs and produce savings for insurers and customers alike. 

We tested whether low versus zero generic co-pays in Part D plans had any significant impact on generic dispensing of antidiabetic drugs and statins and found none. 

Plans used both the pull of free drugs and the push of formulary restrictions to achieve higher rates of generic use, and it was the push rather than the pull that proved effective in the case of statins. 

An increasingly popular strategy in value-based insurance design (VBID) is a preferred generic co-pay tier in which certain medications are available free of charge on the presumption this will shift demand away from expensive brand name drugs and produce savings for insurers and customers alike. Behavioral economists find that the demand for free goods tends to be much higher than for the same goods offered at very low prices. According to Anderson: “Free has the effect of bending the demand curve—demand shoots up in a very nonlinear fashion.”

 Shampan’er and Ariely argue that this occurs because moving to free not only reduces cost, but also confers special benefits to consumers.

 But does this phenomenon hold for prescription drugs? The answer to this question has obvious relevance to private payers using VBID precepts to set drug co-payment levels, and it is also important for public payers including Medicare, as evidenced by Medicare Payment Advisory Commission’s recommendation that co-payments for beneficiaries receiving low-income subsidies (LIS) be raised for brands but reduced to zero for generics.

 would appear to support the wisdom of that policy based on analyses of Medicare beneficiaries’ use of statins, antidepressants, and antidiabetic drugs in Part D plans with and without zero co-pay tiers. However, neither study could account for formulary exclusions and both used cross-sectional designs that are potentially sensitive to selection bias (ie, if beneficiaries with a preference for generic drugs are drawn to plans offering them at no cost, this unobserved behavior will confound the true impact of zero co-pays on generic utilization). We designed the current study to be free of both sources of bias.

In brief, our approach exploited a natural experiment in which Part D enrollees receiving the LIS were randomized to benchmark plans in their region. Some of these plans offered zero generic co-pays and others did not. Randomization ensures equivalent beneficiary characteristics among all plans within a given region and eliminates selection bias. Any observed differences in generic utilization across plans must thus be due to plan policies, including free generics and differences in formulary restrictions. We focused the analyses on Medicare beneficiaries with diabetes who used oral hypoglycemic agents (ie, OADs) and/or statins in 2008.

The study used 2008 data from a 5% random sample of the Medicare population from the CMS Chronic Condition Data Warehouse (CCW). The files included basic enrollment information by service type (Parts A, B, C, and D); Part D LIS status, including plan assignment method (beneficiary selection or CMS random assignment); paid claims records for all Medicare-covered services; and a special file provided us by CMS that contained complete formulary information on covered drugs and restrictions (prior authorization [PA], step therapy, and quantity limits) for every Part D plan.

We selected the study population as a subset of a sample of Medicare beneficiaries with diabetes drawn from a previously published study.

 The inclusion criteria for the original study cohorts were: 1) Medicare beneficiaries enrolled throughout 2008 with continuous Part A, B, and D coverage; 2) dual-eligible LIS recipients with incomes less than or equal to the federal poverty level and continuously enrolled in CMS-assigned benchmark prescription drug plans (PDPs); 3) diagnosed with diabetes based on 

International Classification of Diseases, Ninth Revision, Clinical Modification

 codes 250.xx, 357.2, 362.01, 362.02, or 366.41 on hospital inpatient and medical claims prior to 2008; and 4) filled at least 1 OAD and/or statin prescription in 2008.

Beneficiaries enrolled in Medicare Advantage prescription drug plans (MAPDs) were excluded because CMS does not randomly assign LIS recipients to MAPDs. LIS enrollees in long-term care nursing facilities were excluded because medications are centrally managed and available with no cost sharing. Finally, beneficiaries residing in American territories and Puerto Rico were excluded as we did not have information regarding generic substitution laws in these areas, which might affect generic utilization rates.

For the current study, we created 2 overlapping cohorts of randomized LIS recipients who used OADs and/or statins in 2008. Each cohort was subdivided into 2 groups based on whether their assigned Part D plan offered any generic OADs or statins free of charge. Plans with zero generic co-pay tiers were identified using the following criteria. First, we limited the analysis to plans with at least 30 drug users in each cohort to assure there were multiple beneficiaries taking the most common generic drugs in each class. Next, we identified plans in which non-LIS enrollees had zero co-pays for all prescriptions filled for at least 1 generic OAD or statin over the year. We then checked to determine whether all LIS recipients enrolled in these same plans also paid zero co-pays for the same generic medications. Plans meeting both criteria were considered zero co-pay plans. Nonzero co-pay tier plans were identified based on evidence that no generic OAD or statin was filled at zero co-pay tiers except for prescriptions filled by dual-eligible LIS recipients during the catastrophic phase of the Part D benefit. Individuals assigned to plans that did not offer free generics paid statutory co-pays of $1.05 for generics and $3.10 for brands.

Exploratory analyses showed that among all PDP regions, 49% of LIS enrollees in plans offering zero co-pays for some generic OADs and 59% of enrollees in plans offering zero co-pays for generic statins were California residents. Given the high popularity of zero co-pay tiers in California PDPs, we decided to estimate models both at the national level and for the state.

 presents a flowchart showing how the study samples were selected, beginning with the entire CCW 5% sample of about 2.6 million beneficiaries and ending with the subsamples of LIS recipients randomized to California PDPs with and without zero co-pay provisions for OADs and statins.

Although random plan assignment would ensure the balance of enrollee characteristics within each Part D region, it would not ensure balance across regions. To address that issue, we identified a large number of personal factors that could plausibly influence rates of generic prescribing, including demographic characteristics, diabetes duration and severity, diabetes management activities, comorbidities, and utilization of Medicare services other than drugs. Individual variables in each of these domains are shown in the left-hand column of 

Randomization also does not control for plan policies, other than zero co-pay tiers, that might influence utilization of generic OADs and statins. The most obvious candidate here is plan formulary design given that Part D sponsors have been shown to use these drug lists to drive market demand.

 We used the First DataBank drug dictionary to identify all drugs within our 2 classes of interest and then matched that against the CMS formulary file.

 This allowed us to identify all OADs and statins that were included and excluded from each plan formulary by name and brand/generic status. The formulary file also permitted us to identify all products with significant access restrictions requiring either PA or step therapy.

The 2 most common operational measures of formulary restrictions in the literature are the percentage of individuals subject to specific types of restrictions

 and a dummy variable indicating whether at least 1 drug in the class is subject to restriction.

 Neither approach captures the full impact of formulary designs on individual behavior because restrictions on popular drugs can have a much bigger effect than those placed on seldomly used products. Although our study was not focused primarily on the impact of formulary design, we were interested in whether formulary restrictiveness was correlated with plan policies for zero co-pays. We tested this association for all individual OADs and statins and discovered a strong positive correlation coefficient (r = 0.5) between restrictions on Lipitor (primarily formulary exclusions) and the offer of zero co-pays for generic statins. No other restrictions on commonly used drugs had correlation coefficients greater than 0.25. Based on these findings, we added a Lipitor restriction variable to our multivariate model together with count variables indicating the total number of OAD and statin drugs subject to formulary exclusion, PA, or step therapy.

Finally, because some Part D regions include several states, we controlled for state-level differences in generic substitution laws. There were 3 broad types of laws: 1) allows for generic substitution by pharmacists if “brand only” not indicated by physician, allows for brand if requested by patient, and mandates brand only if indicated by physician; 2) mandates generic substitution by pharmacists if “brand only” not indicated by physician, allows for brand if requested by patient, and mandates brand only if indicated by physician; and 3) mandates generic substitution if “brand only” not indicated by physician, and mandates brand if indicated by physician.

We tested the relationship between zero co-pay tiers and generic utilization rates using a standard 2-part model design. The first equation in each set of models estimated the relationship between the availability of zero generic co-pays and the proportion of all drug users filling any generic prescription. The second equation estimated the impact of zero co-pays on annual days’ supply of generics filled by generic users. All models were estimated using ordinary least squares (OLS) regression, with the full set of covariates listed in Table 1. In the national models, we controlled for regional differences with 33 region dummies, with 1 region (32 for California) as the reference.

The study was approved by the University of Maryland Institutional Review Board.

Table 1 profiles the characteristics of our study cohorts of LIS recipients using OADs. At the national level, we identified 43 plans offering zero co-pays for generic OADs with a combined 3682 LIS enrollees meeting study inclusion/exclusion criteria. The most common free OADs in these plans were glipizide, glipizide XL, glyburide, and glyburide-metformin (

). The comparison cohort included 11,007 LIS recipients enrolled in 242 plans that offered no free OADs. The 2 national cohorts exhibited somewhat different demographic characteristics, but had similar rates for diabetes severity, comorbidities, and Medicare utilization. The biggest difference between the 2 plan types was in the number of OADs subject to formulary exclusion or major restriction (12 in zero co-pay plans vs 9 in nonzero co-pay plans).

By contrast, the 2 California cohorts had virtually identical characteristics, as would be expected given random assignment (California is a single PDP region). We identified 3 California PDPs that offered free OADs (n = 1790) and 8 that did not (n = 1511). Unlike plans in other regions, the California PDPs offering free OADs actually placed major formulary restrictions on fewer individual products compared with other plans (9.5 vs 12.4).

 presents similar data for our statin user cohorts. Nationally, 30 PDPs with 3334 LIS enrollees offered enrollees free generic statins in 2008 compared with 223 (n = 11,132) that did not. Over half of the zero co-pay plans offered either pravastatin and/or lovastatin free and just under a quarter offered free simvastatin (Table 2). Both types of plans placed formulary restrictions on many statin products. Lipitor was subject to major formulary restrictions for two-thirds of LIS recipients enrolled in zero co-pay plans nationally compared with just 21% in nonzero co-pay plans. In California, about half of the LIS recipients in zero co-pay plans faced restrictions on Lipitor, whereas only 1.8% of enrollees in nonzero co-pay plans did so.

We found that very high percentages of LIS recipients used generic OADs (>86%) in the national-level analyses (

). There was a small, nonsignificant 1.2% difference in the percentage of OAD users who filled any prescription for a generic OAD between plans with and without zero co-pay tiers. Zero co-pay status also had a small 1.7-day nonsignificant effect on the annual days’ supply of OADs among generic users. Adjustment for all covariates, including counts of OADs subject to major formulary restrictions, had no effect on these results (complete results from all regressions are available in the eAppendix [

]). Findings were similar in the OAD analyses restricted to California PDPs.

Our statin analyses produced very different results. Much smaller percentages of the study population used generic statins compared with generic OADs, and there was a marked difference in unadjusted generic statin user rates between plans with and without zero co-pay tiers (61.4% vs 54.6% in the national-level analysis and 53.4% vs 41.1% in the California analyses, respectively; both significant at 

<.01). In the unadjusted comparisons, annual days’ supply of statins was 8.4 days lower in zero co-pay PDPs (

<.01) nationally and 3.4 days (nonsignificant) lower in California.

After adjustment, all significant differences in generic statin use disappeared. To see if Lipitor restrictions were responsible for the dramatic difference in the proportions of generic statin users in zero versus nonzero co-pay plans, we estimated otherwise identical regressions excluding that variable. Without Lipitor restrictions in the models, the adjusted effects associated with zero co-pays were almost as high as in the unadjusted models (results not shown). Formulary restrictions—and Lipitor restrictions in particular&mdash;explain why generic statin use was much higher in PDPs with zero co-pays.

Using a study design with high internal validity based on random assignment, we found little evidence of any bounce in demand for generic OADs or statins among LIS recipients enrolled in plans offering zero co-pays for selected generics in these 2 drug classes. This lack of response is consistent with recent study results showing that low-income individuals are generally not sensitive to small changes in prescription drug prices.

 However, it is not what we expected to find based on other experimental studies of demand for free goods, the results of which have shown a marked spike in demand at zero prices.

 One plausible reason for this null finding is that, unlike other consumer goods, prescription drugs depend on physician orders, and physicians may either be reluctant to change their recommended medications in response to small patient savings or they may be unaware that the drugs are available free.

Our results are also inconsistent with findings published by Hoadley et al

 For example, Hoadley et al found that in 2008 (the same year as in our study), the probability of using any generic statin was 13% higher in plans with zero co-pays compared with plans in which the mean generic co-pay was $5.15. Their analysis excluded LIS recipients, so the price differential between those enrolled in plans with and without a zero co-pay tier was somewhat larger than in our study, but the unadjusted effect size was virtually the same (12.5% difference in our national-level analysis). We determined this difference was attributable solely to formulary restrictions on Lipitor, not zero co-pays.

The CMS Part D formulary file was not available to researchers until 2010, but we had access to an earlier beta version of this file for 2008. Researchers using standard CCW files prior to 2010 had access to flags for PA, step therapy, and quantity limits from the CCW plan characteristics file for drugs filled by at least 1 Part D enrollee included in the researcher’s study sample. However, these flags systematically underestimate the true extent of formulary restrictions, because formulary exclusions are not accounted for and other restrictions are undercounted in cases where no enrollee used drugs subject to utilization management. These effects can be quite large: based on the CMS beta version of the 2008 formulary file, we found that 16.2% of all benchmark plans excluded Lipitor as a covered drug, 7.5% placed it on PA, and 7.1% required step therapy. By contrast, Hoadley et al, using the 2008 plan characteristics file flags, determined that no Part D plan required either PA or step therapy for Lipitor.

Our focus on LIS recipients with diabetes means that we cannot generalize our results to other segments of the Medicare Part D population. However, it is worth noting that LIS recipients account for 40% of the entire Part D enrollment and about 55% of program costs.

 Diabetes is one of the most prevalent chronic conditions in the Medicare population, affecting 28% of all beneficiaries in 2014.

Examining zero co-pay effects on individuals who are otherwise subject to nominal cost sharing represents perhaps the strongest test of the hypothesis that “free” confers special benefits to users of chronic care medications. Based on the economics literature, we had expected a spike in demand at zero price even though the absolute reduction in out-of-pocket cost was minimal; that turned out not to be the case. Moreover, one might have expected that LIS recipients would be sensitive to even small price differences given their low incomes, but that too was not confirmed.

Second, we examined just 2 drug classes and further research is necessary to determine whether the effects we found for OADs and statins also apply to other classes. Both classes included many different products, which should make generic substitution relatively easy. Zero co-pay tiers could have a different impact in classes with fewer medications.

Third, although we were able to control for confounding associated with formulary exclusion and restrictions, we did not have access to other plan policies that could conceivably bias our results. For example, one common aspect of medication therapy management programs is for pharmacists to provide patients with information about possible generic substitution. Disease management and wellness programs might have a similar effect.

Lastly, our data are for 2008 when VBID pricing precepts applied by Part D plans were less developed than today. The fact that zero generic co-pay tiers are much more commonly used today only increases the relevance of our study findings. Part D plan formularies are also much more restrictive than they were in 2008. How these 2 trends interact is an important unanswered question that warrants future research.

Some Part D plans use both the pull of free drugs and the push of formulary restrictions to achieve higher rates of generic use. In this study, the push, rather than the pull, proved effective. This cautionary tale suggests that policy makers should give greater consideration to formulary provisions when evaluating the effects of free generics in VBIDs.

 University of Maryland, Baltimore (BS, FH, JX), Baltimore, MD; PhRMA (JSD), Washington DC.

 Dr Dougherty is employed by PhRMA, a trade organization representing drug manufacturers. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

 Concept and design (BS, FH, JSD); acquisition of data (BS); analysis and interpretation of data (BS, FH, JSD, JX); drafting of the manuscript (BS, FH); critical revision of the manuscript for important intellectual content (BS, JSD, JX); statistical analysis (BS, JX); obtaining funding (BS); and supervision (BS).

 Bruce Stuart, PhD, University of Maryland, Baltimore, 220 Arch St, Baltimore, MD 21201. E-mail: bstuart@rx.umaryland.edu. 

2. Shampan’er K, Ariely D. How small is zero price? the true value of free products. Federal Reserve Bank of Boston website. http://www.bostonfed.org/economic/wp/wp2006/wp0616.pdf. Published October 2006. Accessed October 12, 2015.

3. Report to the Congress: Medicare and the health care delivery system. Medicare Payment Advisory Commission website. http://www.medpac.gov/docs/default-source/reports/jun12_entirereport.pdf?sfvrsn=0. Published June 2012. Accessed October 12, 2015.

4. Hoadley JF, Merrell K, Hargrave E, Summer L. In Medicare Part D plans, low or zero copays and other features to encourage the use of generic statins work, could save billions. 

. 2012;31(10):2266-2275. doi: 10.1377/hlthaff.2012.0019.

The offer of free medications to low-income Medicare beneficiaries with diabetes enrolled in Part D plans has no impact on generic prescribing rates.

Does the Offer of Free Prescriptions Increase Generic Prescribing?

Medicare Part D specialty drug users not qualifying for low-income subsidies (non-LIS beneficiaries) face high and variable cost sharing during the calendar year. We examined their out-of-pocket (OOP) cost patterns under the existing Part D cost-sharing policies and proposed changes to these policies.

Using 100% Medicare claims data from 2012, we examined mean annual and monthly OOP drug costs for Medicare Part D patients who were full-year users of Part D specialty drugs for rheumatoid arthritis (RA) (n = 1063), multiple sclerosis (MS) (n = 2256), or chronic myeloid leukemia (CML) (n = 1135) under existing policy. Using the same data, we simulated costs under both proposed Medicare Payment Advisory Commission (MedPAC) policy recommendations and our own recommendations.

 In 2012, our sample faced mean annual cumulative OOP drug costs (for all medications) of $3949 (RA), $5238 (MS), and $6322 (CML). Mean OOP costs were $977 (RA), $1613 (MS), and $2456 (CML) in January alone. A substantial proportion of total annual OOP prescription spending also occurred during the catastrophic coverage phase (RA: $1229 [31%]; MS: $2456 [47%]; CML: $3546 [56%]). Under proposed MedPAC changes, patients would have faced maximum annual OOP spending of $4700, but mean OOP costs in January and February would have been higher compared with the existing policy. Under our proposed strategy, OOP costs would have been spread evenly over 12 months (≤$392 per month). The potential incremental costs of our proposed strategy would have been $23.55 per non-LIS Part D beneficiary per year.

The existing Part D cost-sharing structure creates a substantial financial burden for specialty drug users, especially early in the year. Implementing both annual and monthly OOP maximum spending limits would result in lower, more consistent OOP costs, potentially increasing patients’ ability to access treatments for life-threatening, chronic, and rare diseases.

Am J Manag Care. 2017;23(3 Suppl):S39-S45

Recent pharmaceutical advances hold the promise of transforming the medical care of serious life-threatening, chronic, and/or rare diseases and significantly improving the lives of those afflicted. Yet, these advances may be out of reach for many Americans because of high out-of-pocket (OOP) costs.

 Financial barriers to treatment access are particularly acute for Medicare Part D patients requiring specialty drug treatments. Since Part D’s creation in 2006, patients who do not receive low-income subsidies (non-LIS beneficiaries) have been subject to highcoinsurance requirements for specialty drugs, which are placed on a “specialty tier.” Unlike co-payments of a fixed dollar amount, coinsurance payments vary based on the cost of the medication. Furthermore, under Part D, the required coinsurance percentage fluctuates across the coverage year (

), with the highest costs concentrated at the beginning of the year. The cycle resets on January 1 of the following year.

 In addition, because Medicare Part D plans are not required to have an annual maximum OOP spending limit, patients who are prescribed continuous specialty drug treatment continue to face substantial OOP costs, even during the catastrophic coverage phase.

 These fluctuating costs have been characterized as a “cost-sharing rollercoaster.”

A growing body of evidence links higher cost sharing with reduced utilization of specialty drugs.

 In the Medicare population, studies of several specialty drug classes have documented that beneficiaries not receiving LIS (ie, those responsible for high cost sharing) have lower rates of treatment initiation, greater incidence of gaps in treatment, and more frequent interruptions in treatment compared with counterparts who receive full LIS (and thus face low, relatively stable costs across the coverage year).

 Although further research is needed, it is likely that the high cost sharing for specialty drugs under Medicare Part D may place patients at risk for compromised treatment outcomes due to reduced or delayed initiation, poor adherence, and higher discontinuation rates.

Strategies to address financial barriers must balance access to treatment with very real financial constraints. Based on the accumulating data, and in keeping with our prior recommendations,

 we propose the introduction of 2 key changes to the Part D cost-sharing requirements that would provide additional protection for Medicare beneficiaries. First, the introduction of an annual OOP maximum spending limit—which is commonplace in private insurance plans and in health insurance exchange plans&mdash;would protect beneficiaries from unmanageable cumulative OOP costs .Indeed, the Medicare Payment Advisory Commission (MedPAC) recently recommended policy changes that would effectively establish an annual OOP maximum for Medicare Part D beneficiaries.

 Second, more stable and consistent timing of OOP spending requirements—effectively introducing a monthly OOP maximum as well&mdash;would avoid the major fluctuations in monthly expenses that characterize the current cost-sharing structure and would be more appropriate for seniors, many of whom are on a fixed income. Previously, we have cited programs that offset the burden of high winter heating bills by distributing energy costs across the calendar year as a model for this approach.

In order to test the impact of our proposed strategies on real-world OOP costs for non-LIS Medicare Part D beneficiaries, we undertook a 3-part investigation. First, using actual 100% Medicare claims from 2012, we illustrated the real-world OOP cost patterns and burden under the existing Part D cost-sharing structure for continuous users of specialty drugs for 3 chronic conditions that rank among the top spending categories for specialty drugs.

 Second, we used the same Medicare claims data on drug utilization and costs to simulate the OOP cost patterns that these specialty drug users would face if recent MedPAC recommendations related to Part D cost sharing were to be implemented. Third, we simulated OOP costs for these patients under our proposed strategies and compared them with those under existing policy and the MedPAC recommendations. Finally, we discuss the incremental costs of our proposed strategies and how they could be funded.

Estimation of OOP Cost Patterns Under the Existing Part D Cost-Sharing Structure

To estimate the impact of current policy, we used a data extract from the 2012 Chronic Conditions Data Warehouse 100% Medicare claims files, which contain data on all fee-for-service (FFS) Medicare beneficiaries. (Part D’s cost-sharing structure has not changed since 2012.) We included non-LIS beneficiaries who had: 1) continuous FFS Medicare and stand-alone Part D plan coverage throughout 2012; 2) a diagnosis of rheumatoid arthritis (RA), multiple sclerosis (MS), or chronic myeloid leukemia (CML); and 3) prescription claims reflecting continuous use of a disease-specific Part D-covered specialty drug (listed in the 

]) over the course of the coverage year (ie, a prescription fill for the specialty drug in January and total days’ supply for any disease-specific specialty drug ≥360 days in 2012). For example, if an individual switched from one disease-specific specialty drug to another but had no gap in treatment (eg, a CML patient had 180 days’ supply of bosutinib and 180 days’ supply of dasatinib), then he or she was classified as a continuous user. After these criteria were applied, the resulting sample consisted of 1063 patients with RA, 2256 patients with MS, and 1135 patients with CML. Using actual 2012 claims data for these disease samples, we calculated annual cumulative OOP drug costs and OOP costs by calendar month and benefit phase for patients in each disease group.

We opted to restrict our sample to continuous users for several reasons. First, in the absence of specific reasons for discontinuation (eg, intolerable side effects, poor response), consistent treatment is typically recommended for each of the conditions we examined.

 Thus, we wished to examine the OOP costs associated with optimal treatment patterns. Second, inclusion of individuals who used the specialty drugs during only part of the year could have artificially depressed our mean estimates of annual and monthly costs. Third, high OOP costs are associated with interruptions in treatment and discontinuation of treatment; thus, including individuals who may have used medications inconsistently due to financial burden would have led to further underestimation of the true OOP costs associated with optimal use of these treatments.

Simulation of OOP Cost Patterns Under Proposed MedPAC Changes to Part D Cost Sharing

Next, using the same drug utilization and cost information from the 2012 Medicare claims data, we simulated the OOP cost patterns for the 3 disease samples if recent MedPAC recommendations were to be applied to the 2012 Part D benefit. This involved simulating the impact of 2 key changes proposed by MedPAC.

 The first involved the elimination of the 5% cost-sharing requirement during the catastrophic coverage phase, resulting in an effective annual OOP maximum equal to the annual true OOP (TrOOP) spending limit that triggers catastrophic coverage. In 2012, the TrOOP was $4700. Second, under existing Part D policies, brand name prescription drugs purchased by Part D beneficiaries during the coverage gap phase include a 50% manufacturer discount that is credited toward beneficiaries’ TrOOP spending. Under the MedPAC proposal, these manufacturer discounts would no longer be credited toward patients’ TrOOP spending. Given that movement through the coverage gap is based on TrOOP costs, elimination of the manufacturer discount credit under the MedPAC proposal would effectively extend the time patients spend in the coverage gap and increase the amount they must spend OOP before reaching catastrophic coverage. Then, only patients whose spending was sufficient to trigger entry into the catastrophic coverage phase would benefit from the elimination of cost sharing during that phase.

Simulation of OOP Cost Patterns Under Our Proposed Changes to Part D Cost Sharing

We estimated OOP costs under our proposed policy changes following the same procedure used for the MedPAC simulation, with 2 key differences. First, our proposal does not include a change to the current policy of crediting manufacturer discounts toward patients’ TrOOP spending; we maintained the $4700 spending threshold that triggers catastrophic coverage as our proposed annual OOP spending limit, counting both beneficiary OOP spending and manufacturer discounts. (As in the MedPAC proposal, this spending limit would mean elimination of 5% cost sharing during Part D’s catastrophic coverage phase.) Second, to spread the OOP costs more evenly across the year, we divided the annual OOP maximum spending limit number (ie, $4700) by 12 months to derive a maximum monthly OOP spending limit. This monthly limit also includes both beneficiary OOP spending and manufacturer discounts, meaning that actual beneficiary spending may be lower than the monthly OOP spending limit during the month(s) of the coverage gap.

Estimating the Incremental Cost of Our Proposed Changes and Financing Strategies

To estimate the incremental costs of our proposed policy changes, we identified publicly available data from the year closest to our 2012 utilization data (ie, 2013) to identify: 1) estimates of the total number of non-LIS beneficiaries who entered catastrophic coverage during that year and 2) the average OOP spending among these beneficiaries during the catastrophic coverage period.

 The product of the 2 estimates provides the total OOP costs borne by these beneficiaries that would be foregone (ie, the incremental costs of our proposed policy of instituting an annual OOP spending maximum at the catastrophic coverage threshold level). We then divided the incremental cost of our proposed changes by the total number of non-LIS beneficiaries enrolled in the Part D program in 2013 to identify the per-beneficiary cost of the proposed policy change.

 The per-beneficiary cost is approximate because, whereas the additional cost would be priced out through the Part D plan bidding process, our proposed financing strategy estimates the cost as if it were passed through to beneficiaries directly.

OOP Cost Patterns Under the Existing Part D Cost-Sharing Structure

In 2012, specialty drug users in the RA group faced mean annual cumulative OOP drug costs of $3949 for all their Part D medications, whereas those in the MS and CML groups paid an average of $5238 and $6322, respectively. The majority of these costs were due to specialty drug spending (ie, 88% for the RA group, 91% for the MS group, and 95% for the CML group) (

). A significant proportion of the MS (84%) and CML (87%) groups had annual specialty drug OOP costs of $5000 or more (data not shown). Because OOP spending pushed specialty drug users through the deductible, initial coverage phase, and coverage gap quite quickly in 2012, patients continued to have substantial OOP spending requirements during the catastrophic coverage phase, despite the fact that their coinsurance obligation dropped to 5%. Across disease areas, a substantial proportion of total annual OOP prescription spending occurred during the catastrophic coverage phase (RA: $1229 [31% of annual OOP costs]; MS: $2456 [47%]; CML: $3546 [56%]) (

). At the same time, specialty drug users had to bear large mean OOP costs in January alone (RA: $977; MS: $1613; CML: $2456). Once again, specialty drug costs were driving this OOP spending (see 

). These January OOP costs represented a substantial portion of spending for the entire year (RA: 25% of annual spending; MS: 31%; CML: 40%) (

), and about half of total OOP costs were paid out by February.

OOP Cost Patterns Under Proposed MedPAC Changes to Part D Cost Sharing and Under Our Proposed Strategies

Simulated analyses examining the impact of proposed MedPAC recommendations showed that the policy changes would have mixed effects on beneficiaries' OOP burden (

). Because manufacturer discounts for brand name drugs during the coverage gap phase would no longer be credited toward patients’ TrOOP spending, patients using brand name specialty drugs would have taken longer to reach the TrOOP spending limit, leading them to remain in the coverage gap phase for a greater period of time. Because of this widened coverage gap, they would have been subject to 50% cost sharing until they reached the TrOOP spending limit that triggers entry into the catastrophic coverage phase. As such, only patients whose cumulative annual spending (excluding manufacturer discounts) would have exceeded the designated spending limit ($4700 in 2012) would have seen overall savings from the elimination of 5% cost sharing during the catastrophic coverage phase. Under current policy, manufacturer discounts can effectively reduce OOP spending in the coverage gap by up to half, depending on an individual beneficiary’s drug utilization patterns. This differential impact was apparent across our 3 diagnostic subgroups. Although patients in the MS and CML groups would have seen a decrease in cumulative mean annual OOP costs under the proposed changes, the RA group would have experienced an increase in costs ($4540 under MedPAC vs $3949 in 2012; data not shown) due to fewer beneficiaries reaching the OOP limit.

Given that patients in all disease samples had OOP costs that pushed them into the coverage gap early in the coverage year (often with the first fill of their disease-specific specialty drug), and they would have faced a higher threshold to exit the coverage gap under the MedPAC recommendations, the proposed policy changes would ultimately subject patients to even higher, more concentrated OOP costs during these early months (Figure 5). Patients would have faced maximum annual OOP spending of $4700 in 2012, but this entire OOP cost burden would have to be borne by most patients in the first 3 to 4 months of the calendar year. Furthermore, under the MedPAC changes, the mean OOP costs in January and February would be even higher than under existing policy (eg, $994 under MedPAC vs $977 in January 2012 for RA; $1685 under MedPAC vs $1613 for MS; $2814 under MedPAC vs $2452 for CML) (Figures 4 and 5).

Instituting a monthly OOP maximum based on an annual OOP maximum of $4700—which also counts the 50% manufacturer discounts for brand name drugs in the coverage gap&mdash;under our proposed strategies, in contrast, would result in a maximum of $392 in OOP costs to be borne by the patient in any given calendar month, as indicated by the dotted line in Figure 5.

Incremental Cost of Our Proposed Changes and Financing Strategies

Based on Medicare utilization numbers,18 approximately 700,000 (2.8%) of the 24.2 million non-LIS beneficiaries enrolled in Part D plans reached the catastrophic coverage phase in 2013. They paid mean OOP costs of approximately $814 during that phase, for a total of approximately $569.8 million in patient OOP spending during the catastrophic coverage phase. Dividing that cost among all non-LIS beneficiaries, implementation of our proposal would have cost an additional $23.55 for each non-LIS beneficiary per year (ie, $1.96 per month).

 In practice, the increased costs of $23.55 per beneficiary per year would not be straight pass-through costs via premiums, but would instead be borne out through the Part D plan bidding process, ultimately reducing the amount by which beneficiaries are directly impacted. Nevertheless, the incremental costs of our proposed changes are minimal when averaged over all non-LIS beneficiaries, regardless of the ultimate financing mechanism.

Our analyses demonstrate that Medicare beneficiaries using high-cost specialty drugs face variable and high OOP cost obligations under the existing Part D cost-sharing structure, with a majority of these expenditures concentrated at the beginning of the calendar year. In 2 of our 3 disease samples (MS and CML), the average OOP cost of the first disease-specific specialty drug fill for the calendar year nearly equaled or exceeded the average monthly Social Security benefit.

 This benefit provides a substantial portion of income for many Medicare beneficiaries.

 In all 3 disease areas we examined—RA, MS, and CML&mdash;the required 5% coinsurance payments during the catastrophic coverage phase also resulted in considerable cumulative OOP spending during the remainder of the year, consistent with findings reported elsewhere.

 Our simulation found that changes proposed by MedPAC, which effectively introduce an annual OOP spending maximum via elimination of cost sharing during the catastrophic coverage phase, would provide some relief later in the year for patients facing the highest specialty drug costs. Yet, the changes would actually exacerbate the problem of front-loaded costs at the beginning of the coverage year and fewer beneficiaries would likely benefit from the annual OOP spending limit, as many patients would remain in the coverage gap longer under the proposal to exclude manufacturer discounts for brand name drugs from TrOOP spending. This would exacerbate their total OOP costs and the timing of these costs. Excluding the discounts would also effectively double OOP costs paid by these beneficiaries in the coverage gap by widening the gap; this undermines the goal of better managing the timing and magnitude of OOP costs and ensuring appropriate access to care.

Such high and variable spending requirements are disruptive to monthly budgets, and our data highlight the fact that evaluation of the financial burden related to OOP spending should pay attention to both the magnitude and the timing of OOP expenditures. Our proposed strategies would institute both an annual and a monthly OOP maximum spending limit, which would spread OOP costs more evenly across the year. This would have resulted in OOP costs closer to $400 per month in our 2012 sample (inclusive of manufacturer discounts in the coverage gap); although the amount would increase slightly in 2017 due to an increase in the TrOOP spending limit, it would remain more manageable than existing OOP obligations. This would most likely improve patients’ ability to meet their OOP cost-sharing obligations, especially early in the year. In addition, maintaining the current policy to count manufacturer discounts toward TrOOP would effectively lower overall OOP spending below the $4700 limit for beneficiaries requiring brand name medications. The proposed OOP spending limit would also increase the predictability of monthly OOP obligations, much like how payment options provided by energy companies enable more consistent budgeting despite seasonal variability in consumption. In light of a growing body of evidence that links high cost sharing among Medicare Part D beneficiaries with delayed initiation of treatment, increased gaps in treatment, and reduced adherence (compared with beneficiaries who receive LIS and face more stable, low cost sharing), reducing the OOP burden associated with specialty drug use would likely improve access to and optimal use of these treatments.

It should be noted that our analyses provide a snapshot of the impact of both existing and proposed policies utilizing a sample year of prescription fill data from 2012. This is in line with MedPAC’s own analysis of their proposed changes, which also used a single year of data (from 2013). However, the Part D benefit is dynamic; each year, there are changes to the designated spending limits that trigger entry into each benefit phase, and additional changes related to provisions of the Affordable Care Act (ACA) are scheduled to be implemented over the next 5 years, further impacting OOP costs. Most importantly, the catastrophic coverage threshold is scheduled to increase significantly by 2020, and 2 main factors will lead to additional widening of the coverage gap phase, making it harder for beneficiaries to reach the OOP threshold for catastrophic coverage (and thus, the proposed annual OOP spending maximum).

First, Part D plans will offer more generous plan coverage, which will displace previous beneficiary OOP spending during the coverage gap with payments that will not count toward patients’ TrOOP—thereby indirectly widening the gap. Second, the growth rate of the TrOOP threshold that triggers catastrophic coverage was slowed from 2014 through 2019 under the ACA. As the coverage gap closes, however, the TrOOP threshold for 2020 will be set as if growth had not been artificially slowed.

 This will lead to a $1200 increase in the TrOOP threshold from 2019 to 2020, a phenomenon known as the “OOP cliff.”

 Consequently, the beneficiary OOP burden documented in our analyses will intensify over time. Furthermore, a larger number of beneficiaries are likely to be affected over time. Ongoing developments in pharmaceutical treatments are likely to increase the number of non-LIS beneficiaries who are prescribed medications associated with the highest levels of OOP spending; for example, the past few years have seen a dramatic increase in the use of new treatments for hepatitis C, a trend that began after our 2012 data. This underscores the urgent need for strategies to alleviate OOP costs and burden.

Our analysis had several limitations. First, as with all administrative databases, Medicare claims may be subject to errors or omissions. Second, we examined 2012 data and applied policy changes that took effect under the ACA in that year. As noted above, if the ACA remains in effect, the coverage gap will be phased out by 2020 and the catastrophic coverage limit will be higher; thus, the catastrophic coverage limit (and our proposed annual OOP spending maximum) will increase and the figures presented here will represent an underestimate of patient OOP spending. Our analysis is not able to account for future changes to the ACA that may impact Medicare policy. Third, given the availability of claims data, we limited our sample to FFS Medicare Part D beneficiaries; to illustrate the real-world OOP cost burden in patients prescribed continual treatment throughout the year, we further limited our sample to full-year users of specialty drugs. However, our proposed strategies are recommended for, and our incremental cost and financing calculations apply to, all Medicare Part D beneficiaries (FFS and Medicare Advantage) regardless of whether they were full- or part-year users of specialty drugs. Our calculations do not account for any increases in specialty drug utilization and spending that would occur among part-year users in response to lower and more stable monthly cost sharing under our proposed strategies. Similarly, as the number of beneficiaries eligible for and requiring specialty drug treatments increases over the years, the incremental cost of our proposed strategies would further increase. It is notable that our estimated costs begin at a relatively modest amount (less than $2 per month per beneficiary), however.

As with most aspects of healthcare, the fine-grained logistical details of implementing our proposed strategies will be straightforward in some cases and more complex in others. Redistribution of annual OOP costs in the form of a stable monthly payment is least complicated for individuals who are prescribed medications for continuous use throughout the coverage year (as in our sample), whereas protocols would have to be developed for those who initiate treatment later in the year. For example, if a patient fills a prescription in September that would have carried a $1500 coinsurance payment, he or she may need to continue to pay the remaining balance on that prescription into the following year. Although this does increase the complexity somewhat, much of the rest of the healthcare system bills patients for remaining OOP costs after a service is rendered (eg, imaging tests, surgeries, hospitalizations). Developing adjustments in pharmacy and insurance procedures that are more in step with the advances in pharmaceutical treatments is a worthwhile goal.

Indeed, the expanding role of self-administered pharmaceutical treatments in the management of serious life-threatening, chronic, and/or rare diseases also argues for a less siloed approach and greater attention to overall OOP costs for patients. Our analysis addresses prescription drug OOP costs only, yet patients are also responsible for OOP costs related to premiums, medical deductibles, and medical co-pays and coinsurance. Unlike most employer and health insurance exchange plans, which integrate medical and prescription drug expenses into a combined annual OOP maximum spending limit,

 all Medicare beneficiaries currently lack an annual OOP maximum limit for their Part D prescription drug spending and the majority have no annual maximum for other OOP medical spending. Our proposal represents an important first step toward addressing these issues.

Division of General Internal Medicine, Department of Medicine, Perelman School of Medicine (JAD, PL, VPL), Leonard Davis Institute of Health Economics (JAD, PL), Center for Public Health Initiatives (ARP), University of Pennsylvania, Philadelphia, PA; PhRMA (JSD, AF), Washington, DC.

Medicare claims analyses offer insight into how proposed policy changes would affect out-of-pocket prescription costs for Part D beneficiaries requiring specialty drugs.

J. Samantha Dougherty, PhD

Articles

Medication Adherence as a Measure of the Quality of Care Provided by Physicians

Does the Offer of Free Prescriptions Increase Generic Prescribing?

Reducing Out-of-Pocket Cost Barriers to Specialty Drug Use Under Medicare Part D: Addressing the Problem of "Too Much Too Soon"

Prevalence, Effectiveness, and Characteristics of Pharmacy-Based Medication Synchronization Programs