Mail-order pharmacy use and medication adherence among Medicare Part D beneficiaries with diabetes

Abstract

Objective:

To examine medication adherence among Medicare Part D beneficiaries initiating oral anti-diabetic medications and explore whether there is any association of using mail-order pharmacy (vs. retail pharmacy) with better adherence in this patient population.

Research design and methods:

Using administrative pharmacy claims data, we conducted a retrospective cohort study on Medicare Part D beneficiaries who newly initiated oral anti-diabetic treatment between July 1, 2008 and December 31, 2008. Mail-order pharmacy users were matched to retail pharmacy users via propensity scoring, controlling for patient demographic and clinical characteristics. Adherence with oral anti-diabetic medications during the benefit year of 2009 was assessed using the proportion of days covered (PDC). Comparison of medication adherence between the mail-order pharmacy group and retail pharmacy group was conducted in the propensity matched sample using the paired t-tests and McNemar’s tests.

Results:

A total of 22,546 patients who initiated oral anti-diabetic medications were identified. The average PDC was 0.60 and only 41.6% of the study population attained good adherence (defined as PDC ≥ 0.8) with oral anti-diabetic medications during calendar year 2009. The matched sample included 1361 patients in each of the mail-order and retail pharmacy cohorts. Compared with the retail pharmacy group, mail-order pharmacy users demonstrated a significantly higher PDC (0.68 vs. 0.61; P < 0.001) throughout the benefit year. More patients in the mail-order pharmacy group (49.7%) attained good adherence with their oral anti-diabetic medications compared to 42.8% in the retail pharmacy group (P < 0.001).

Limitations:

The study was subject to limitations inherent in retrospective claims database analysis.

Conclusions:

Adherence with oral anti-diabetic medications among Medicare Part D beneficiaries is suboptimal. Patients using mail-order pharmacy had better adherence to oral anti-diabetic medications than those who used retail pharmacies. However, the causal relationship between mail-order pharmacy use and adherence should be further examined in a randomized study setting.

Key words::

• Adherence
• Diabetes
• Mail-order pharmacy
• Medicare Part D

Background

Despite proven clinical benefits of pharmacotherapy and recommendations of treatment guidelines, medication adherence remains an issue among patients with chronic diseases1,2. Current studies reveal that only about half of chronic disease patients are adherent with their medications. The average adherence to oral hypoglycemic agents ranges from 36 to 93% and more than half of the patients who initiate statin treatment discontinue their medication within the first year of treatment3,4.

Medicare Part D offers prescription drug coverage for Medicare beneficiaries through stand-alone prescription drug plans (PDP) and Medicare Advantage prescription drug (MAPD) plans. Since its enactment in 2006, there has been considerable interest in understanding medication adherence among the Medicare Part D enrollees. However, empiric studies examining medication adherence among the Part D enrollees with chronic conditions such as diabetes are rare5. Recent studies found that medication adherence was improved after Medicare Part D, but the level of adherence is still suboptimal5,6. For Medicare Part D beneficiaries with chronic diseases such as diabetes, medication adherence could be challenging as these patients usually have high medication burden due to disease-related complications and other comorbidities. They may still experience cost-related non-adherence due to the donut hole in the current Part D benefit design7,8. In addition, Medicare Part D beneficiaries are also likely to have limited mobility because of chronic conditions, which might further jeopardize medication adherence in beneficiaries who have to pick up their prescription refills at a retail pharmacy. Hence, medication adherence remains a concerning issue in this patient population9,10. Poor adherence to chronic medications can worsen disease outcomes and increase the use of healthcare resources such as physician visits, nursing homes, hospitals, and healthcare costs11,12. For healthcare policy makers, health plans and pharmacy benefit management companies that manage Part D patients, controlling escalating healthcare costs by enhancing medication adherence in this population has been a matter of great importance.

Over the past decade, there has been a dramatic increase in using mail-order pharmacy for chronic medication refills. Mail-order has been shown to not only increase savings and safety13, but may also lead to improved medication adherence compared to using retail pharmacies. Limited evidence from recently published studies indicated that patients who received prescriptions by mail were more likely to have good adherence in comparison to patients who obtained medication refills at local pharmacies14,15. However, so far no study has examined the impact of mail-order pharmacy use on medication adherence among the Medicare Part D beneficiaries.

The purpose of our study was two-fold: (1) to examine medication adherence specifically among Medicare Part D beneficiaries with diabetes; and (2) to explore whether there is any association of using mail-order (vs. retail pharmacy) with better adherence in this patient population.

Patients and methods

Study population

This retrospective cohort study used the electronic pharmacy claims data from a large pharmacy benefit management (PBM) company based in the western United States. The study was reviewed by an external institutional review board (Individual Review Consulting, Inc., Corte Madera, CA, USA).

The study population included Medicare Part D patients who newly initiated oral anti-diabetic treatment between July 1, 2008 and December 31, 2008. Patients were required to be continuously enrolled during the study period of January 1, 2008 through December 31, 2009 and not have prescription claims for any anti-diabetic medication between January 1, 2008 and June 30, 2008. Patients were excluded from the study if they used insulin or exenatide during the study period, or if they were low income subsidies (LIS) recipients, or if they were receiving antidepressant or antidementia agents during the study period. Due to the specific benefit design of Medicare Part D allowing for the possibility of a ‘coverage gap’ or ‘donut hole’ to occur once a spending threshold was reached, the index date for the study was set as January 1, 2009, the starting date of the following benefit year rather than the date of initiation of anti-diabetic treatment. Patients were followed from the index date until December 31, 2009.

Two study groups were defined. The mail-order group included patients who exclusively used the PBM’s mail-order pharmacies for oral anti-diabetic medication refills throughout calendar year 2009. The retail pharmacy group included patients who exclusively used retail pharmacies for oral anti-diabetic medication refills during the same time period.

Variables

Several demographic and clinical variables were used in the analysis. Patient demographic characteristics included age and gender. Chronic Disease Score was used to assess disease burden for each patient using pharmacy claims data during the 6 month period prior to the index date16. Patients were categorized by their type of health plan: Medicare Advantage Prescription Drug (MAPD) plan vs. stand-alone Prescription Drug Plan (PDP). The total number of distinct medications with a days supply value that fell into the month of January, 2009 was obtained for each patient. Type of index medication (‘generic’ or ‘brand’) was determined based on the first prescription of oral anti-diabetic medication filled in calendar year 2009. If multiple prescriptions were filled on the same day, as long as there was a brand medication, then the index medication was counted as a brand medication. For patients with cross-over claims (i.e., prescription claims during calendar year 2008 that had days supply crossing into calendar year 2009), this was determined based on the first oral anti-diabetic prescription in calendar year 2009. For patients who did not have any claims for oral anti-diabetic medications in 2009, this was determined based on their last prescription fill of oral anti-diabetic medication in calendar year 2008. Duration of anti-diabetic treatment was defined as the number of months between the first prescription fill date for oral anti-diabetics in 2008 and December 31, 2008.

Propensity score matching

Although use of mail-order pharmacy among the PBM members is on a voluntary basis, mail-order pharmacy users could still potentially differ from retail pharmacy users on a number of patient-level characteristics. Hence, we employed the propensity score matching technique to adjust for the potential selection bias17.

Patients in the retail pharmacy group were matched 1:1 to patients in the mail-order group based on their propensity score which was derived using a logistic regression model adjusting for the following study covariates: age, gender, chronic disease score, type of health plan (MAPD vs. PDP), number of distinct medications, type of index medication (generic vs. brand), and the duration of oral anti-diabetic treatment in calendar year 2008.

Outcomes measure

Medication adherence with oral anti-diabetic medications during calendar year 2009 was measured using the proportion of days covered (PDC). PDC reflects the proportion of the actual number of days being covered by medication during a specified time interval. PDC is a common measure of medication adherence and has been used in many retrospective studies based on administrative pharmacy claims databases2,5,9,18,19. Recently, PDC has been endorsed by the National Quality Forum as an indicator of quality in drug therapy management20.

PDC was computed as the sum of days of supply for all fills of oral anti-diabetic medication in calendar year 2009 divided by the total number of calendar days in 2009. Days covered by more than one drug were only counted once. For prescriptions extending beyond the end of follow-up period, their days of supply were truncated on December 31, 2009. For cross-over claims, the part of days of supply from a fill in 2008 that crossed into 2009 was counted towards the PDC calculation. PDC for oral anti-diabetic medication was determined as both a continuous variable and a categorical variable respectively. ‘Good adherence’ was attained if a patient had PDC greater than or equal to 0.82,5,18,19,21,22.

Statistical analysis

All analyses were conducted using SAS version 9.1 (SAS Institute Inc, Cary, NC, USA). Descriptive analyses were performed on patient demographic and clinical characteristics. Statistical tests were conducted using t-tests, Wilcoxon rank sum tests, and chi-square tests where appropriate. For the matched study population, between-group comparisons were assessed using paired t-tests, signed rank tests, and McNemar’s tests respectively. All reported P-values are two-sided.

Results

A total of 22,546 patients with diabetes who met the study selection criteria were identified. Among them, there were 18,854 retail pharmacy users and 1496 mail-order users. The majority of patients (89%) were 65 years or older. Most of the patients (77.5%) were enrolled in a stand-alone Prescription Drug Plan. Compared with patients who used retail pharmacies for medication refills, patients in the mail-order pharmacy group were more likely to be over 65 years old (95.3% vs. 87.8%; P < 0.001) and enrolled in a Medicare Advantage Prescription Drug plan (30.7% vs. 21.7%; P < 0.001). Characteristics of the study population are displayed in Table 1.

Table 1.  Characteristics of the overall study sample, mail-order group, and retail pharmacy group before propensity score matching.

Study Variable	Overall Study Sample	Mail-order	Retail Pharmacy	P-Value
	(N = 22,546)	(N = 1496)	(N = 18,854)	(Mail-order vs. Retail)
Age, mean (SD)	72.3 (9.6)	73.0 (7.0)	72.2 (10.0)	0.884
≥65	20,067 (89.0%)	1425 (95.3%)	16,550 (87.8%)	<0.001
Gender female	11,512 (51.1%)	732 (48.9%)	9706 (51.5%)	0.058
Health plan type				<0.001
PDP	17,469 (77.5%)	1037 (69.3%)	14,758 (78.3%)
MAPD	5077 (22.5%)	459 (30.7%)	4096 (21.7%)
CDS, mean (SD)	4.7 (2.4)	4.5 (2.2)	4.6 (2.4)	0.855
0–3	10,639 (47.2%)	765 (51.1%)	8883 (47.1%)	<0.001
4–5	3800 (16.9%)	187 (12.5%)	3298 (17.5%)
≥6	8107 (36.0%)	544 (36.4%)	6673 (35.4%)
Index medication				0.964
Generic	19,299 (85.6%)	1277 (85.4%)	16,102 (85.4%)
Brand	3247 (14.4%)	219 (14.6%)	2752 (14.6%)
Number of distinct medications, mean (SD)	5.1 (2.9)	5.2 (2.5)	5.0 (3.0)	<0.001
Duration of anti-diabetic treatment (in months), mean (SD)	2.7 (1.7)	2.9 (1.7)	2.7 (1.7)	<0.001

Values are number (%) unless noted otherwise. SD = standard deviation; PDP = stand-alone Prescription Drug Plan; MAPD = Medicare Advantage Prescription Drug plan; CDS = Chronic Disease Score.

Table 2 shows the baseline characteristics for the propensity matched population. After matching, a total of 1361 patients from each group were included in the analysis. Patients’ baseline characteristics were well balanced after matching.

Table 2.  Characteristics of the mail-order users and retail pharmacy users after propensity score matching.

Study Variable	Mail-order	Retail Pharmacy	P-Value
	(N = 1361)	(N = 1361)
Age, mean (SD)	72.8 (6.7)	72.9 (6.5)	0.748
Female	672 (49.4%)	665 (48.9%)	0.791
Health plan type			1.000
PDP	998 (73.3%)	998 (73.3%)
MAPD	363 (26.7%)	363 (26.7%)
CDS, mean (SD)	4.4 (2.1)	4.5 (2.4)	0.185
0–3	718 (52.8%)	721 (53.0%)	0.139
4–5	163 (12.0%)	149 (11.0%)
≥6	480 (35.3%)	491 (36.1%)
Index medication			0.330
Generic	1160 (85.2%)	1178 (86.6%)
Brand	201 (14.8%)	183 (13.5%)
Number of distinct medications, mean (SD)	5.1 (2.3)	5.0 (2.4)	0.163
Duration of anti-diabetic treatment (in months), mean (SD)	2.9 (1.7)	2.9 (1.7)	0.259

Values are number (%) unless noted otherwise. SD = standard deviation; PDP = stand-alone Prescription Drug Plan; MAPD = Medicare Advantage Prescription Drug plan; CDS = Chronic Disease Score.

Among the overall study population, 11.7% failed to fill a prescription for an oral anti-diabetic medication in calendar year 2009. The average PDC was 0.60 (SD = 0.35) and only 41.6% of the study population attained good adherence (defined as PDC ≥ 0.8) with oral anti-diabetic medications during calendar year 2009.

Medication adherence was compared between the mail-order group and the retail pharmacy group (Table 3). Before matching, compared with the retail pharmacy group, mail-order pharmacy users demonstrated a significantly higher adherence (0.68 vs. 0.57; P < 0.001) throughout the benefit year. In the propensity matched sample, higher adherence was still observed in the mail-order pharmacy users, though on a slightly smaller magnitude (0.68 vs. 0.61; P < 0.001). More patients in the mail order group attained good adherence with their oral anti-diabetic medications (49.7% vs. 42.8%; P < 0.001).

Table 3.  Adherence to oral anti-diabetic medications.

Adherence	Mail-order vs. retail pharmacy (before matching)			Mail-order vs. retail pharmacy (after matching)
	Mail-Order	Retail Pharmacy	P-value	Mail-Order	Retail Pharmacy	P-Value
	(N = 1496)	(N = 18,854)		(N = 1361)	(N = 1361)
PDC, mean (SD)	0.68 (0.31)	0.57 (0.36)	<0.001	0.68 (0.31)	0.61 (0.34)	<0.001
Proportion of patients who attained good adherence	49.1%	38.5%	<0.001	49.7%	42.8%	<0.001

PDC = proportion of days covered; SD = standard deviation.

Discussion

Among this population of Medicare Part D patients, adherence to oral anti-diabetic medications was suboptimal. Consistent with findings from previous research5,6, our study revealed that among the Medicare Part D beneficiaries with diabetes, there was a high proportion of patients who did not achieve good adherence to oral anti-diabetic medications. The non-adherence rate observed in our study (58.4%) was higher than what was reported in a previous study (35.1%) in the Medicare population5. This difference could partially be explained by the difference in the length of the measurement period. The study by Yang et al. only measured adherence in a 6-month period while our study measured patients’ adherence over an entire benefit year. This could suggest that Medicare beneficiaries may become less adherent to their anti-diabetic medications as the benefit year progresses.

Our study further showed that patients who used mail-order pharmacies for prescription refills had a higher rate of adherence to anti-diabetic medications and this was still observed even after controlling for confounders in the claims database during the matching process. The finding is consistent with previous studies that have demonstrated the association between mail-order pharmacy use and medication adherence14,15. In addition, our study suggests that among the Medicare Part D beneficiaries, for whom medication adherence is a critical issue, using a mail-order pharmacy could potentially help to improve medication adherence. To our knowledge, this is the first study that has examined the impact of mail-order pharmacy use on medication adherence among Medicare Part D beneficiaries.

Currently, health plans and pharmacy benefit management companies that manage Part D patients have been exploring various approaches including Medication Therapy Management (MTM) programs with the hope of improving medication adherence among the Part D population, but these programs are highly variable and definitive evidence supporting the effectiveness of many of such interventions is lacking23. Now findings from our study provide the evidence that increasing mail-order pharmacy use among Medicare Part D beneficiaries might be an effective alternative that could potentially improve adherence in this patient population. Health plans and pharmacy benefit management companies may consider developing potential effective strategies to increase mail-order pharmacy use and thus improve medication adherence in this elderly population.

Compared to retail pharmacy, mail-order pharmacy offers the convenience for medication refills by eliminating the time and inconvenience associated with traveling to a local retail pharmacy store. This could be of particular importance to elderly and disabled individuals with limited mobility. In addition, mail-order pharmacy has been shown to be cost-saving for patients through reduced copayment13. Mail-order pharmacy use could also be beneficial to payers. Though few studies found that mail-order pharmacy use might increase drug utilization24, the Federal Trade Commission (FTC) revealed in a 2005 report that PBM-owned mail-order pharmacies offer lower prices on prescription drugs than retail pharmacies and are effective at capitalizing on opportunities to dispense generic medications13, which could help cut down the pharmacy spending. To date, researchers have frequently found improved adherence to be associated with lower health care costs in chronic disease25. One recent study has also directly linked mail-order pharmacy use with reduced total health care cost and disease-related medical cost over time in a commercially insured population15. Although increasing mail-order pharmacy use could potentially benefit the Medicare Part D population through improved adherence, it is important for mail-order pharmacies to build interactive features, such as phone services to offer private prescription medication counseling, into these programs in order to ensure safe medication use among the elderly population.

As some researchers have pointed out, medication adherence is a multi-faceted problem that requires a multi-faceted approach. Mail-order pharmacy use may be just one way to help improve adherence in patients with chronic diseases. Currently, health plans are also experimenting with other interventions but with varying levels of success. Among them, value-based benefit design, which lowers patient copays for high-value medications, could be another useful tool for improving treatment adherence and outcomes. Other interventions that are being applied include medication therapy management, pay-for-performance that rewards physician and patients to improve performance and outcomes, and integrated communication channels that allow healthcare providers to communicate adherence-enhancing messages to patients26.

Several limitations of this study need to be acknowledged. First, this study used administrative claims data. Some variables that could affect medication adherence such as disease severity, income level, and race/ethnicity were not available in the claims data. Although we used propensity score matching to mitigate potential selection bias, this method was not able to control for confounding from unobserved and unmeasured variables in the claims database. Second, self-selection into mail order may potentially bias the study results, although a previous study did not find evidence of unmeasured self-selection14. Third, this study was conducted in Medicare Part D beneficiaries with diabetes and therefore the findings may not be generalizable to other patient populations. Lastly, we used pharmacy claims to assess medication compliance in the study population. Pharmacy refill analysis addresses the question of drug availability and may not reflect the actual consumption of medications. However, it provides crucial insights into patient willingness to comply, and so far, such records may be the most accurate estimate of compliance for large cohorts available today27.

Conclusion

In summary, the findings from this study suggest that adherence to oral anti-diabetic medications among Medicare Part D beneficiaries is suboptimal. Patients using the PBM’s mail-order pharmacy were likely to have better adherence to their oral anti-diabetic medications than those who used retail pharmacies. However, the causal relationship between mail-order and adherence should be further examined in a randomized study setting.

Transparency

Declaration of funding

There was no external funding for this study.

Declaration of financial/other relationships

L.Z., A.Z., K.M.S., A.S.H., B.S.C., and B.K.S. are employees of Prescription Solutions. K.M.S., A.S.H., and B.S.C. have disclosed that they are stock share-holders of the United Health Group.

Acknowledgements

The authors have disclosed that they have received no additional support or contributions from others in the preparation of this manuscript.

This study was presented in part at the 16th Annual Meeting of International Society for Pharmacoeconomics and Outcomes Research, May 21–25, 2011, Baltimore, MD, USA.