Comparing Insulin Pen versus Vial Use in the Elderly Medicare Part D Population for Long-Acting Basal Insulin Adherence

Since insulin was discovered almost nine decades ago, the technology involving insulin delivery has come a long way from the original crude and inconvenient metal syringes.(1) Today’s technology delivers insulin via devices with fine needles for almost painless administration. A recent advancement is the insulin pen, which came to market in 1985 and is unique in that the insulin container and syringe are combined into a single unit. Pens are available for both bolus and basal insulin. Bolus insulin is given at mealtimes to keep glucose levels under control immediately following meals, whereas basal insulin is a longer-acting insulin form given once to twice daily to keep glucose levels stable during fasting phases throughout the day. Basal insulin is available in the market in intermediate- and long-acting forms. The long-acting basal insulin products were available as both vial and pen: Lantus and Levemir. The pen’s lifestyle flexibility makes sense for bolus insulin and intermediate-acting basal insulin injections due to the potential need for insulin administration during public outings. It is inconspicuous and provides lifestyle flexibility for on-the-go individuals, reducing the social embarrassment that can be associated with self-administration(1- 8) and thereby, it is hoped, increasing medication adherence. Lifestyle flexibility is less of an issue for long-acting basal insulin because injections may occur in the privacy of one’s home during the morning or evening hours. If lifestyle flexibility is not a factor for long-acting basal insulin administration, then pen devices may not improve adherence as compared with vials.

The primary aim of this study was to determine whether pens result in improved adherence for long-acting basal insulin administration as compared with vials in elderly Medicare patients. Our hypothesis was that pen devices lead to no improvement in adherence with long-acting basal insulin. Our second aim was to identify characteristics that are associated with adherence for long-acting basal insulin.

Methods

Data

Medicare claims data from calendar years 1990 and 2010 were obtained through the CMS’s Chronic Condition Data Warehouse (CCW) and Oracle. The First Databank therapeutic classification system was used to identify basal insulin by linking the National Drug Code to the medication name using the First Databank classification system (proprietary to First Databank; license held by CMS). These data were used to perform the analysis.

Sample

The study population was elderly beneficiaries (≥65 years of age) enrolled in the Medicare Part D program for all 12 months of the respective calendar year (i.e., either year 2008 or 2010). Beneficiaries had to have at least two medication fills for the specific insulin device and to have obtained insulin solely through use of the specific device during the calendar year. Beneficiaries with a nursing home stay or skilled nursing facility stay were removed from the sample to avoid insulin being administered by professional staff at these facilities, because that would eliminate patient freedom of choice regarding medication adherence.

Data from calendar year 2010 were used to perform the beneficiary-level logistic regression analyses to address both aims. For diabetes type, Part A or Part B data from calendar years 1990 to 2010 were used to identify type 1 or type 2 diabetes. Diabetes severity was allocated following Young et al.’s approach to assignment based on diagnosis codes in the Parts A and B data.(9) Diabetes severity was coded as a categorical variable: high, medium-high, medium-low, and low.

Measures

Determining adherence with insulin can be problematic, because the product is available in bulk packaging, and the amount of product used daily is beneficiary specific. A single product (whether pen or vial) may last less than one month for some beneficiaries but longer than a month for other beneficiaries. Such differences can lead to inaccuracies in the “days’ supply” reported by the pharmacy for payment. Additionally, because package sizes differ between pen versus vial devices (i.e., the package size for pens is 3 mL per pen, packaged as five pens for a total of 15 mL, whereas the package size for vials is a single 10-mL vial), the calculation of days’ supply becomes more problematic.

To address this concern, we followed the approach of Buysman et al. and created an adjusted days’ supply for each prescription claim based on whether a pen or vial was used.(10) The adjustment factor was calculated as the ratio of the median time between insulin claims divided by the median pharmacy reported days’ supply. This adjustment factor was then multiplied by the actual reported days’ supply for each prescription fill to give the adjusted days’ supply. For year 2008, the adjustment factor was 1.50 for pens and 1.20 for vials; for year 2010, the adjustment factor was 1.47 for pens and 1.17 for vials.

Adherence was defined using the medication possession ratio (MPR) modified measure,(11,12) which indicates the proportion of days the beneficiary possessed a filled prescription. To construct this measure, for each beneficiary, the total adjusted days’ supply (covered days) was the numerator. The denominator (observation period) consisted of the number of days from the first dispensing of insulin to the date of last dispensing plus the number of days’ supply obtained at the last dispensing. Any time spent for a hospitalization was subtracted during the period. This ratio was multiplied by 100 to express the MPR as a percentage.

We considered beneficiaries to be adherent with their insulin regimen if they had an MPR of 80% or higher—a cutoff score commonly used in the literature for many medication studies to define adherence.(13-16) MPR was calculated based on whether the delivery system was pen or vial. For enrollee use, we assessed the percentage of long-acting basal insulin enrollees who used a pen versus those who used a vial during the calendar years of 2008 and 2010.

Population characteristics included gender, age, race/ethnic origin, Medicare and Medicaid dual eligibility status, disease burden, Part D plan type, metropolitan area, and region. Age, race/ethnic origin, and disease burden were coded as categorical variables. Age was divided into three groups: 65 to 74 years; 75 to 84 years; and 85 and over. Race/ethnicity was categorized into the groups of Black, Hispanic, white, and other. Disease burden was assessed based on the CMS Hierarchical Condition Category risk score data and divided into the groups of high, medium-high, medium-low, and low. Part D enrollment was divided into the groups of full-year enrollment in a Medicare Advantage Prescription Drug Plan (MA-PD) or full-year enrollment in a stand-alone prescription drug plan (PDP).

Analysis

Separate cross-sectional descriptive analyses at the enrollee level were performed for calendar years 2008 and 2010, addressing medication adherence and enrollee use (Table 1) and enrollee characteristics (Table 2). Chi-square analyses were used to test for statistical differences. Table 1 addresses the primary aim of the study; Table 2 provides general descriptive information regarding beneficiaries with long-acting basal insulin use. Multiple logistic regression was performed to assess which characteristics were associated with long-acting basal insulin adherence during calendar year 2010 (Table 3). Table 3 addresses the secondary aim of our study. Analysis of the data was performed using SAS 9.3 software (proprietary to SAS Institute, Inc., license agreement held by CMS). Results were reported as odds ratios (OR) along with p values.

Results

Overall adherence for long-acting basal insulin increased approximately 4% for each device type between year 2008 and 2010 (Table 1). Vial use showed a slightly higher medication adherence rate than pen use for each respective year. The percentage of enrollees using pen devices increased by approximately 14% during the period (from 24% to 38%), whereas vial use decreased by the same amount, indicating that the market is moving toward increased pen use.

Regarding characteristics associated with increased pen use from year 2008 to 2010, we found a greater uptick in long-acting basal pen use among women, the youngest elderly (i.e., age 65 to 74), whites, dual enrollees, PDP enrollees, urban area, and southern region (Table 2). We also noted that pen use decreased among non-dual enrollees. This finding may be a function of the cost associated with pens as compared with vials. Some plans may require a higher copay for the pen device.(17)

Regression results (Table 3) indicate that medication adherence was lower when long-acting basal insulin was administered using pens than using vials (OR = 0.94, p <.001). Men were slightly more likely to be adherent than females (OR = 1.04, p <.001). The youngest elderly (age 65-74) were 27% more likely to be adherent as compared with the eldest elderly (age 85 and over; OR = 1.27, p <.001). The middle-aged elderly (75-84) were 15% more likely to be adherent than were the eldest elderly (OR = 1.15, p <.001). Blacks were 20% less likely to be adherent as compared with whites (OR = 0.80, p <.001), whereas Hispanics were slightly more likely to be complaint as compared with whites (OR = 1.03, p <.001). Dual enrollees were 32% more likely to be adherent than non-duals (OR = 1.32, p <.001). Higher disease burden had a slight negative impact on adherence, with enrollees with the highest disease burden being 4% less likely to be adherent as compared with enrollees with the least disease burden (OR = 0.96, p <.001). Enrollees in PDPs were slightly more likely to be adherent as compared with MA-PD enrollees, at 3% (OR = 1.03, p <.001). Urban residents were 18% more likely to be adherent (OR = 1.18, p <.001). Enrollees in Midwest, Southern, and Western regions were less likely to be adherent as compared with enrollees in the Northeast region, with enrollees in the South region having the highest likelihood of being non-adherent, at 17% (OR = 0.83, p <.001). Patients newly diagnosed with diabetes were 31% more likely to be adherent (OR = 1.31, p <.001). Type 2 diabetics were more likely to be adherent, at 9% (OR = 1.09, p <.001). Diabetes severity had no statistical impact on adherence with long-acting basal insulin.

Discussion

This study addressed whether pens result in improved adherence for long-acting basal insulin administration as compared with vials in elderly Medicare patients. The results indicate that adherence is not improved when long-acting basal insulin is administered using pens versus vials. One possible explanation may be that social stigma is not applicable to this insulin type, considering administration is more likely to occur in a private setting such as the patient’s home during the early morning or late evening hours. Another explanation may be that elderly persons are comfortable with vials and have no difficulty with their use. As time progresses and the younger generation ages into the elderly category, our findings may change. We noted that the youngest elderly had the greatest increase in pen use during the study period as compared with the remaining age groups. Other characteristics associated with increased pen use were female gender, white race/ethnic origin, dual enrollment, PDP enrollment, urban residency, and living in the South.

Holding all other variables constant in the regression analysis, we observed that dual enrollees were more likely than non-dual enrollees to be adherent. This finding may reflect lower cost-sharing due to the low income subsidy benefit. Blackwell et al.(18) reported a similar finding with cardiovascular medication use among dual enrollees in the Part D population.

The regression analysis also found that elderly enrollees newly diagnosed with diabetes were more likely to be adherent. Similar results have been found regarding recently initiated oral hypoglycemic medications.(19) The increased diligence may be due to the newness of the diagnosis. Future research could address whether adherence patterns over time differ between users of oral agents and long-acting basal insulin.

The regression results also specified that the eldest elderly (age 85 and over) were less adherent than their younger counterparts. This finding is surprising, considering past research targeting the elderly has identified old age as being associated with better cardiovascular medication adherence.(18, 20) Further research may be warranted to study whether elderly beneficiaries have a more difficult time remembering to take their long-acting basal insulin medication because it is not typically a daily event as compared with the meal-time bolus insulin. If research were to show that was the case, individualized mechanisms or processes could be devised to improve adherence.

Limitations

This study had several limitations:

• First, the age of our data may be a limiting factor.

• Second, a beneficiary was considered to be adherent if he or she had an MPR of 80% or higher. However, there also may be “hidden” improvement in those who scored below the 80% level. For example, adherence may be increasing significantly, from 20% to 65% for example, but since this improvement does not cross the threshold of 80%, it was not measured as an increase in adherence.

• Third, our findings are based on cross-sectional data and, therefore, are not suitable for identifying causative relationships.

• Fourth, the study evaluated enrollees having sole use of the specific device during the study timeframe; enrollees switching between device types were not addressed in the analysis. However, our approach was necessary in order to identify the impact of device type on adherence with long-acting basal insulin use.

Conclusion

Although previous research has shown that insulin pen use is associated with improvement in adherence, our study findings show that, when focusing specifically on long-acting basal insulin use, adherence for pen use among elderly enrollees may not be any better than vial use. Further analysis assessing the cost–benefit ratio of pen versus vial use among this particular user group may be warranted given these results, and will be of value to clinicians, health insurers, patients, and other stakeholders.

References

1. Baruah M. Insulin pens: the modern delivery devices. J Assoc Physicians India. 2011 Apr;59 Suppl:38-40.

2. Asche CV, Shane-McWhorter L, Raparia S. Health economics and compliance of vials/syringes versus pen devices: a review of the evidence. Diabetes Technol Ther. 2010;12(1):S101-108.

3. Wright BM, Bellone JM, McCoy EK. A review of insulin pen devices and use in the elderly diabetic population. Clin Med Insights Endocrinol Diabetes. 2010; 3:53-63.

4. Selman JL. Evolution of diabetes insulin delivery devices. J Diabetes Sci Technol. 2010;4:505-513.

5. Asamoah E. Insulin pen: the “iPod” for insulin delivery (why pen wins over syringe). J Diabetes Sci Technol. 2008;2:292-296.

6. Davis EM, Christensen CM, Nystrom KK, Foral PA, Destache C. Patient satisfaction and costs associated with insulin administered by pen device or syringe during hospitalization. Am J Health Syst Pharm. 2008;65:1347-1357.

7. Korytkowski M, Niskanen L, Asakura T. FlexPen: addressing issues of confidence and convenience in insulin delivery. Clin Ther. 2005;27 Suppl B:S80-100.

8. Bohannon N, Jack D. Type II diabetes: tips for managing your older patients. Geriatrics. 1996;51:28-35.

9. Young BA, Lin E, Von Korff M, et al. Diabetes complications severity index and risk of morality, hospitalization, and healthcare utilization. Am J Manag Care. 2008;Jan; 14(1):15-23.

10. Buysman E, Conner C, Aagren M, Bourchard J, Liu F. Adherence and persistence to a regimen of basal insulin in a pre-filled pen compared with vial/syringe in insulin-naïve patients with type 2 diabetes. Curr Med Res Opin. 2011;27:1709-1717.

11. Vanderpoel DR, Hussein MA, Watson-Heidari T, Perry A. Adherence to a fixed-dose combination of rosiglitazone maleate/metformin hydrochloride in subjects with type 2 diabetes mellitus: a retrospective database analysis. Clin Ther. 2004;26:2066-2075.

12. Hess LM, Raebel MA, Conner DA, Malone DC. Measurement of adherence in pharmacy administrative databases: a proposal for standard definitions and preferred measures. Ann Pharmacother. 2006;40,1280-1288.

13. Gary TL, Crum RM, Cooper-Patrick L, Ford D, Brancati F. Depressive symptoms and metabolic control in African Americans with type 2 diabetes. Diabetes Care. 2000;23:23-29.

14. Duncan JC, Rogers R. Medication compliance in patients with chronic schizophrenia: implications for the community management of mentally disordered offenders. J Forensic Sci. 1998;43:1133-1137.

15. Rudd P. The measurement of compliance: medication taking. In: Krasnegor NA, Epstein L, Johnson SF, Yaffe SJ. Development Aspects of Health Compliance Behavior. Hillsdale, NJ: Lawrence Erlbaum Associates; 1993:185-213.

16. Skaer TL, Sclar DA, Markowski DJ, Won JKH. Effect of value-added utilities on prescription refill compliance and Medicaid health care expenditures: a study of patients with non-insulin-dependent diabetes mellitus. J Clin Phar Ther. 1993;18:295-299.

17. Pisano M. Overview of insulin and non-insulin delivery devices in the treatment of diabetes. P T. 2014;39:866-876.

18. Blackwell SA, Baugh DK, Montgomery MA, Ciborowski GM, Waldron CJ, Riley GF. Noncompliance in the use of cardiovascular medications in the Medicare Part D population. Medicare Medicaid Res Rev. 2011 Dec 14;1(4). doi: 10.5600/mmrr.001.04.a05.

19. Garcia-Perez L, Alvarez M, DillaT, Gil-Guillen V, Orozco-Beltran D. Adherence to therapies in patients with type 2 diabetes. Diabetes Ther. 2013;4(2):175-194.

20. Monane M, Bohn RL, Gurwitz JH, Glynn RJ, Levin R, Avorn J. Compliance with antihypertensive therapy among elderly Medicaid enrollees: the roles of age, gender, and race. Am J Public Health. 1996;86:1805-1808.