eAppendix. Details of multivariate multiple regression models.
Blumberg DM, Prager AJ, Liebmann JM, Cioffi GA, De Moraes CG. Cost-Related Medication Nonadherence and Cost-Saving Behaviors Among Patients With Glaucoma Before and After the Implementation of Medicare Part D. JAMA Ophthalmol. 2015;133(9):985-996. doi:10.1001/jamaophthalmol.2015.1671
Understanding factors that lead to nonadherence to glaucoma treatment is important to diminish glaucoma-related disability.
To determine whether the implementation of the Medicare Part D prescription drug benefit affected rates of cost-related nonadherence and cost-reduction strategies in Medicare beneficiaries with and without glaucoma and to evaluate associated risk factors for such nonadherence.
Design, Setting, and Participants
Serial cross-sectional study using 2004 to 2009 Medicare Current Beneficiary Survey data linked with Medicare claims. Coding to extract data started in January 2014 and analyses were performed between September and November of 2014. Participants were all Medicare beneficiaries, including those with a glaucoma-related diagnosis in the year prior to the collection of the survey data, those with a nonglaucomatous ophthalmic diagnosis in the year prior to the collection of the survey data, and those without a recent eye care professional claim.
Effect of the implementation of the Medicare Part D drug benefit.
Main Outcomes and Measures
The change in cost-related nonadherence and the change in cost-reduction strategies.
Between 2004 and 2009, the number of Medicare beneficiaries with glaucoma who reported taking smaller doses and skipping doses owing to cost dropped from 9.4% and 8.2% to 2.7% (P < .001) and 2.8%, respectively (P = .001). However, reports of failure to obtain prescriptions owing to cost did not improve in the same period (3.4% in 2004 and 2.1% in 2009; P = .12). After Part D, patients with glaucoma had a decrease in several cost-reduction strategies, namely price shopping (26.2%-15.2%; P < .001), purchasing outside the United States (6.9%-1.3%; P < .001), and spending less money to save for medications (8.0% to 3.5%; P < .001). Using a multivariate analysis, the main independent risk factors common to all cost-related nonadherence measures were female sex, younger age, lower income (<$30 000), self-reported visual disability, and a smaller Lawton index.
Conclusions and Relevance
After the implementation of Part D, there was a decrease in the rate that beneficiaries with glaucoma reported engaging in cost-saving measures. Although there was a decline in the rate of several cost-related nonadherence behaviors, patients reporting failure to fill prescriptions owing to cost remained stable. This suggests that efforts to improve cost-related nonadherence should focus both on financial hardship and medical therapy prioritization, particularly in certain high-risk sociodemographic groups.
A particularly critical issue in glaucoma treatment is the cost of therapy.1,2 Among patients with glaucoma, cost is a significant barrier to adherence,1,3 which is a frequent driver of progressive visual loss. Rossi et al4 found that patients with a stable visual field during a 1-year period had a median adherence rate of 85% while patients with progressive visual field changes had a median adherence rate of 21%. Using a medication-event monitoring system, Slaeth and colleagues5 found that patients who were less than 80% adherent were significantly more likely to develop more severe visual field defects.
Quiz Ref IDPrior to 2006, there was no uniform drug coverage for Medicare beneficiaries and only 65% to 80% of Medicare beneficiaries had prescription drug coverage.6 In 2006, the Medicare Modernization Act implemented the Part D prescription drug benefit, which allowed adults covered with Medicare to purchase insurance for prescription drugs. The purpose of the Part D program is to increase the availability of prescription medications to Medicare beneficiaries and, thus, decrease cost-related nonadherence (CRN) in the general Medicare population. However, patients with glaucoma may be particularly vulnerable to CRN because of the chronic and asymptomatic nature of the disease, the difficulty patients experience in self-administering eyedrops, and the perception that topical medications are less important than systemic medications because they are not dosed orally.
To date, little is known about how recent policy changes have affected the ability of US patients with glaucoma to acquire their medications. In this study, we assessed CRN among a nationally representative sample of Medicare beneficiaries with glaucoma before and after the implementation of Medicare Part D (2004-2009). In particular, we tested the hypothesis that implementation of Medicare Part D decreased rates of CRN both in beneficiaries with and without glaucoma and we investigated risk factors for self-reported CRN and cost-saving behaviors.
The purpose of this study was to understand the effect of Medicare Part D on cost-related nonadherence and cost-saving measures among Medicare beneficiaries with glaucoma.
After the implementation of Medicare Part D, the number of beneficiaries with glaucoma who reported taking smaller doses and skipping doses decreased.
Failure to obtain prescriptions owing to cost remained stable.
The percentage of beneficiaries with glaucoma engaging in cost-saving measures, such as price shopping or purchasing outside the United States, declined after the implementation of Medicare Part D.
The Medicare Current Beneficiary Survey (MCBS) is a nationally representative panel survey of both institutionalized and noninstitutionalized Medicare beneficiaries. Survey data were obtained from face-to-face interviews by trained interviewers.7,8 Coding to extract data started in January 2014 and analyses were performed between September and November of 2014. The MCBS sample was drawn from an enrollment list of all persons entitled to Medicare at the beginning of 2004 to 2009 and included a national sample of approximately 12 000 to 16 000 participants each year. The MCBS study links interviews, Medicare claims, and detailed self-reported use data (regardless of payer), including filled prescriptions. Respondents provide information on filled prescriptions during in-depth interviews at 4-month intervals, during which respondents are asked to bring and review pill bottles and other medication containers. Medicare claims provide detailed information, including diagnostic codes for services from medical professionals.
Primary and secondary diagnostic codes included in Medicare claims were used to identify the subset of MCBS survey respondents who received a diagnosis of glaucoma from one of their professionals during a health care encounter. Control patients without glaucoma were selected from the overall MCBS population of community-dwelling beneficiaries who had any ophthalmic International Classification of Diseases, Ninth Revision code from the year prior and were determined not to have a glaucoma-related diagnosis. Each year of follow-up was considered a separate observation. The dependence between observations was taken into account using generalized estimating equations in all statistical analyses.
For all beneficiaries, demographic characteristics were described using annual survey responses, including age, self-reported race/ethnicity, sex, education, current employment status, source of Medicare eligibility, self-reported visual acuity/health, and income, which included pension, social security, and retirement benefits. Functional independence in activities of daily living were measured from survey data and calculated into the instrumental activities of daily living by Lawton and Brody.9 The Charlson Comorbidity Index, a weighted sum of 17 conditions predictive of 1-year mortality, was derived from International Classification of Diseases, Ninth Revision codes and used to measure health status.
All survey participants were Medicare beneficiaries; however, approximately 80% also had supplemental drug coverage. Because our survey years straddled the transition to Part D in 2006, sources of drug coverage among our survey respondents shifted. Potential sources of prescription coverage from 2004 to 2006 included private plans, Medicare health maintenance organizations, and Medicaid. In 2007 to 2009, sources of coverage were categorized as private plans, Medicare Part D, Medicaid, and other public plans, including state-sponsored supplements.
Cost-related nonadherence was mapped to the following 4 validated10,11 survey questions: did the respondent not fill a prescription because it was too expensive, skip doses to make the medicine last longer, take less medicine than prescribed to make the medicine last longer, or delay filling prescriptions owing to cost. For all measures, responses were obtained from the entire cohort and were categorized as yes if often or sometimes and no if never.
In addition, several cost-saving behaviors measured in the MCBS were included in our analyses. These behaviors included generic drug use, purchasing prescriptions via mail/Internet/outside the United States, receiving prescription samples from a physician, shopping for the best prices, and spending less on other basic needs to afford medications. Responses were categorized as yes if often or sometimes and no if never.
First, the demographic and clinical characteristics of Medicare beneficiaries in the entire cohort and in those with and without glaucoma were determined by year. Next, the proportions of the 4 CRN measures and the 6 cost-saving measures described earlier were evaluated in participants with glaucoma before and after the implementation of Part D (2004 and 2009) using a χ2 test.
We further performed 2 multivariate multiple regression models (ie, more than 1 outcome variable and multiple predictors) to identify factors associated with the 2 sets of outcomes in the entire MCBS cohort12 (for additional details, see the eAppendix in the Supplement). For these analyses, glaucoma was considered an independent predictor.
For each regression model, the same set of predictors was analyzed. Surveys with incomplete data sets were excluded from the analyses. Analyses were conducted using Stata, version 13.1 (StataCorp LP). Given the multiple hypotheses tested, type I error was set at 1% to minimize the chance of false-positive findings.
From a total of 45 222 participants with complete data sets in MCBS from 2004 to 2009, our sample included 20 688 Medicare beneficiaries who had an ophthalmic diagnosis in the prior year. When examined year by year, there were 3743 beneficiaries with ophthalmic diagnoses in 2004, 3681 beneficiaries in 2005, 3611 beneficiaries in 2006, 3512 beneficiaries in 2007, 3197 beneficiaries in 2008, and 2942 beneficiaries in 2009. Among these individuals, the percentage of patients with glaucoma each year ranged from 22% to 26%. Table 1 demonstrates characteristics of MCBS survey patients with glaucoma and patients without glaucoma as well as characteristics of the overall Medicare survey participants and the subpopulation of Medicare participants, with age as a qualifying diagnosis.
As expected in a glaucoma cohort,13 respondents with glaucoma were older (mean age, 77 years) and more likely to be African American (11% vs 9%) than the overall sample of Medicare beneficiaries (both, P < .001; 1-sample test of means and proportions, respectively). The majority of respondents with glaucoma were Medicare eligible based on age, although 7% were eligible based on disability. In contrast to 20% dual eligibilities in the overall Medicare sample, 16% had dual Medicaid and Medicare benefits. Compared with 79% in the general Medicare sample, 97% had some form of prescription drug coverage. Self-reported visual disability on the MCBS, defined as “a lot of trouble seeing” or “no usable vision,” was 33% in the glaucoma subset, whereas 16% reported visual disability in the overall sample. Of the survey respondents with glaucoma, 23% self-reported poor health on the MCBS, which was defined as “fair” or “poor” compared with others of the same age, compared with 14% in the total Medicare survey sample.
The glaucoma sample was stable across time with regard to mean age, racial/ethnic distribution, employment status, Medicare eligibility, and self-reported poor health and visual acuity. Between 2004 and 2009, there was a slight shift toward increasing levels of education (27% of patients with glaucoma finished <high school in 2004 whereas 19% finished <high school in 2009; 23% had some college education in 2004, whereas 29% had some college education in 2009). However, there was no change across time in the percentage of respondents with glaucoma who achieved a college degree or higher.
The proportion of beneficiaries with glaucoma who had some form of prescription drug coverage increased between 2004 and 2009 (94%-99%). Among those with some form of prescription coverage, the proportion with Part D was 48% of beneficiaries in 2006, 47% in 2007, 51% in 2008, and 49% in 2009. During the study period, there was a decrease in the proportion of beneficiaries with glaucoma on Medicaid prescription coverage (23%-18%; P = .03) as more dual eligible patients gained prescription coverage through Part D.14
Despite the implementation of Part D, failure to obtain prescriptions owing to cost remained stable from 2004 to 2009 among beneficiaries with glaucoma (P = .12). By contrast, beneficiaries with glaucoma who reported taking smaller doses and skipping doses owing to cost dropped from 9.4% and 8.2% (P < .001) to 2.7% and 2.8% (P = .001), respectively (Table 2). The delay in filling prescription measure only became available after initiation of Part D; thus, we could not compare changes across the study.
Quiz Ref IDIn 2004, 76% of beneficiaries with glaucoma engaged in some drug cost-saving strategies, most often by asking for generic drugs or free samples. Overall, 76% of beneficiaries with glaucoma continued to engage in at least 1 cost-saving strategy in 2009. There was no change in individuals requesting generic alternatives (47.8%-46.6%; P = .24) or ordering medication by mail (32.0%-34.8%; P = .20; Table 3). There was a decrease in the frequency of beneficiaries asking for samples (52.9%-44.1%; P = .004), price shopping (26.2%-15.2%; P < .001), purchasing outside the United States (6.9%-1.3%; P < .001), and spending less to save money for medications (8.0%-3.5%; P < .001) between 2004 and 2009.
Quiz Ref IDIn 2009, there was a decrease in the number of MCBS beneficiaries who reported taking smaller doses, skipping doses, and not filling prescriptions owing to cost when compared with 2004 (odds ratio = 0.92-0.95; P < .01). Multivariate analyses for the original 3 CRN behaviors (described earlier) and for a fourth CRN behavior (delay filling prescription owing to cost) added in 2007 are shown in Table 4. Independent risk factors common to all CRN measures were lower income (<$30 000), younger age, self-reported visual disability, and limitations in instrumental activities of daily living (smaller Lawton and Brody index). Individuals without any form of prescription drug coverage or with more comorbidities were more likely to engage in CRN. Compared with their peers without glaucoma, glaucoma status was not associated with any of the tested CRN measures.
In 2009, there was a decrease in the rate that MCBS beneficiaries reported engaging in all 6 cost-saving behaviors compared with 2004 (odds ratio = 0.89-0.98; P < .001). The results of the multivariate analysis for the 6 cost-saving behaviors are shown in Table 5. Overall, a higher education level, lack of drug coverage or less comprehensive supplemental private coverage (ie, Medigap), younger age, unemployment, and more comorbidities tended to say yes to most cost-saving behavior measures. Patients with glaucoma were more likely to ask for free samples when compared with their peers without glaucoma.
Poor adherence to glaucoma medications is common and results in increased risk of visual impairment and blindness.2,15,16 Although multiple reasons for poor adherence exist, there are limited data evaluating prescription drug coverage and CRN. The Medicare Modernization Act, which instituted Medicare Part D, added prescription drug coverage to traditional Medicare. The Medicare Modernization Act represents the largest expansion of the program since its inception. Despite many challenges, the Part D drug benefit is widely hailed as a cost-effective entitlement owing to its high approval rating among seniors and lower-than-expected costs.17,18
Quiz Ref IDTo our knowledge, this is the first study in a nationally representative sample of Medicare beneficiaries to evaluate changes in CRN among beneficiaries with glaucoma after the implementation of Part D. First, prescription drug coverage among beneficiaries diagnosed as having glaucoma was much higher than that of the general Medicare population even before the implementation of Part D. In 2004, 94% patients diagnosed as having glaucoma had prescription drug coverage whereas in 2009, 99% had prescription drug coverage. However, we theorized that patients with glaucoma may have gained better or more complete prescription coverage through the drug benefit and that such improved coverage would translate to changes in CRN. Our results suggested that there was a decrease in certain CRN behaviors, such as taking smaller doses, delaying prescription refills, and spending less on basic needs to afford medications. In contrast, the number of beneficiaries who reported failing to fill a medication owing to cost remained stable. These findings suggest that specific CRN behaviors where the patient intends to use medications but suffers from financial hardship may be more sensitive to cost changes resulting from the implementation of Part D. By comparison, failure to fill a medication considers more complex patient behavior subject to influences beyond financial hardship, namely prioritizing medication. Further CRN research is warranted in exploring methods to engage patients in their medical care in addition to reducing out-of-pocket expenditures.
Data are conflicting in the general literature on the effectiveness of the Part D implementation with regard to CRN.19 Using MCBS data, Madden et al20 found evidence for a small but significant overall decrease in CRN after the inception of the drug benefit. However, the authors reported no net decrease in CRN among the sickest beneficiaries who continued to experience higher rates of CRN. Kennedy and colleagues21 reported that subpopulations at particular risk for CRN included women, nonwhite beneficiaries, working-age beneficiaries, dual eligible beneficiaries, and beneficiaries with multiple chronic conditions, with drug costs and lack of prescription coverage as the primary reasons for failing to fill prescriptions. Using the National Health Interview Survey, Levine and colleagues22 found no improvement in CRN after the implementation of Part D among stroke survivors whereas Medicare beneficiaries with significant depressive symptoms did not demonstrate an improvement in CRN after the drug benefit implementation. These findings and others suggest that Part D may be inadequate in certain subpopulations, particularly the sickest Medicare beneficiaries who have intensive medicine needs and cost-related barriers to access. In parallel, our results suggest that for patients with glaucoma, visual disability and systemic comorbidities are both predictors of CRN behavior.
There are limited data comparing consistency of adherence behavior between ophthalmic and systemic medications. Furthermore, there are few reports identifying which factors related to adherence are generalizable both to glaucoma and systemic conditions. For example, cost has previously been shown to be a significant barrier to adherence in both patients with glaucoma3 and patients without glaucoma.21 However, to our knowledge, there are no published reports to date comparing CRN between ocular hypotensive medications and systemic medications and it is an area for future research. Studies have found that patients with asthma have better adherence to oral medications compared with inhaled medications,23 suggesting that adherence rates vary by route of administration. This may offer insight into adherence for topical ophthalmic medications relative to oral medications because eyedrops are usually more difficult for patients to administer24 and may be perceived as not being real medications because they are not dosed orally.25 Furthermore, evidence suggests that chronic and (initially) asymptomatic diseases, such as glaucoma, may be associated with poorer adherence26,27 because patients’ perceived necessity for medications and concerns about potential adverse effects affect adherence.28
In the present population sample, the specific key risk factors for CRN were younger age, an annual income less than $30 000, higher Charlson Comorbidity Index, and visual disability. These findings of CRN in younger patients, lower-income patients, and patients with greater systemic comorbidity are consistent with previous reports in other fields of medicine29,30 but studies on sex and CRN are inconsistent.19,31,32 Similarly, the glaucoma adherence literature is inconsistent, reporting that women may have lower rates of adherence compared with men,33 whereas other studies have found no difference between the sexes.34 Consistent with prior reports of poor adherence and glaucoma-related visual disability,5 our study found higher rates of nonadherence regardless of cause among patients with visual disability.
Medication cost plays an important role on nonadherence in glaucoma and other fields of medicine. The implementation of Medicare Part D was, in part, fostered to minimize nonadherence by tackling one of the causing variables. Although our data show that the vast majority of Medicare beneficiaries with glaucoma now have some type of drug coverage (approximately 99%), it does not imply that CRN is no longer an issue among those beneficiaries or that overall nonadherence decreased in that period. First, our data simply show a decrease in some of the behaviors we were able to test. There are numerous other CRN behaviors not collected in the MCBS (eg, borrowing money from family members or friends). Second, even the behaviors we were able to test have the inherent limitations of self-reported information, which may be biased by inaccurate recall or vulnerable to social desirability.35 Although self-reported nonadherence may be limited by underreporting,36 the magnitude of underreporting appears to be consistent across time in the MCBS data set37 and supports the validity of our results. Third, nonadherence behaviors unrelated to cost in that period warrant further investigation. The MCBS data set does not provide any objective measurement of overall nonadherence, which would have been useful to address some of the earlier points. Fourth, although MCBS is rich in patient-level characteristics, it lacks sufficient detail about CRN with regard to intraocular pressure–lowering medications; instead, analyses were based on general CRN among beneficiaries with glaucoma. Moreover, the data set did not include information about disease severity because beneficiaries with glaucoma were detected based exclusively on claims data. Lastly, we selected only community-dwelling noninstitutionalized Medicare beneficiaries and results may only be generalized to this population.
The results of this study highlight the continued barriers to adherence among Medicare beneficiaries with glaucoma even after the implementation of Medicare Part D. Future studies should explore targeted efforts to minimize CRN in certain high-risk sociodemographic groups.
Corresponding Author: Dana M. Blumberg, MD, MPH, Columbia University, Department of Ophthalmology, Harkness Eye Institute, 635 W 165th St, New York, NY 10032 (email@example.com).
Submitted for Publication: January 15, 2015; final revision received April 13, 2015; accepted April 19, 2015.
Published Online: June 4, 2015. doi:10.1001/jamaophthalmol.2015.1671.
Author Contributions: Dr Blumberg had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Blumberg, Cioffi, De Moraes.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Blumberg, Prager, Cioffi, De Moraes.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Blumberg, Prager, De Moraes.
Obtained funding: Blumberg, Cioffi.
Administrative, technical, or material support: Blumberg.
Study supervision: Blumberg, Liebmann, Cioffi, De Moraes.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Funding/Support: This study was supported by an unrestricted grant from the Irving A. Hansen Memorial Foundation (Blumberg).
Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: We acknowledge Matt Neidell, PhD, Mailman School of Public Health, Columbia University Medical Center, for technical guidance. He received a consulting fee.