Impact of Insulin Delivery Systems in Elderly Patients With Type 2 Diabetes

, , ,
AJPB® Translating Evidence-Based Research Into Value-Based Decisions®, September/October 2015, Volume 7, Issue 5

This real-world study suggests that improved clinical outcomes are possible in the difficult-to-treat elderly patients with type 2 diabetes group without affecting healthcare costs, by initiating insulin therapy with a pen device.

In the United States, an estimated 20% of individuals 65 years or older have diabetes—most commonly, type 2 diabetes (T2D).1,2 The prevalence of diabetes in the elderly is expected to reach more than 30% by 2050, affecting over 25 million people 65 years or older,3 due to increasing rates of overweight and obesity. Diabetes in the elderly is a risk factor for earlier death, greater physical disability, and cardiovascular comorbidities.1 A recent survey showed that the prevalence of diabetes-related comorbidities is, in fact, higher among older patients.4 Mean Charlson comorbidity index (CCI) scores in patients with T2D aged 65 to 74 years and ≥75 years were both significantly higher than in those aged 18 to 64 years. Patients aged 65 to 74 years also had the highest rates of hypertension (69%) and hypercholesterolemia (67.4%).4

Elderly patients with diabetes are also at increased risk of hypoglycemia. A recent study among patients in long-term care (LTC) facilities showed that hypoglycemia and diabetes are associated with higher mortality and resource utilization, including emergency department (ED) and hospital transfers.5 Furthermore, diabetes is associated with institutionalization among elderly patients, with the prevalence of diabetes among patients in LTC facilities estimated to be at 34.2%.6

Current American Geriatrics Society guidelines for the care of older adults with T2D suggest a target for glycated hemoglobin (A1C) of 7.5% to 8% in most cases; a lower target A1C (7%-7.5%) might be appropriate in healthy patients with few comorbidities, whereas higher targets (8%-9%) are appropriate for those with multiple comorbidities, poor health, and limited life expectancy.6 Unless contraindicated, metformin is the preferred first-line therapy for T2D. After the use of metformin, American Diabetes Association guidelines recommend that glucose-lowering therapy should be individualized.1 As T2D is a progressive disease, insulin therapy is eventually indicated for many patients.1

Pen devices offer an appealing alternative to traditional vial-and-syringe delivery in elderly patients7 who have T2D, because these present particular challenges that can complicate patient management. In addition to the risks described above, diabetes in the elderly is also a risk factor for geriatric syndromes, including depression and cognitive impairment, and for polypharmacy.1 Furthermore, physical impairment and problems with vision or manual dexterity are common in elderly patients with diabetes.8,9 With an aging diabetes population, these are increasingly important considerations in assessing the ease and safety of insulin administration. Although the safety of insulin use in healthy older adults—together with diabetes education, careful monitoring, and ongoing cognitive assessment—is supported by guidelines,6 patients and healthcare providers tend to avoid insulin therapy because of fear of adverse effects such as weight gain and hypoglycemia, and concerns relating to dosing errors and injections.10-13

Several studies have shown benefits either for initiating insulin treatment with a pen device or switching from vial-and-syringe delivery to pen delivery. Compared with vial-and-syringe delivery, patient-reported benefits include increased convenience, preference, and satisfaction, as well as superior discretion, ease of use, ease of reading the dose, ease of dose selection, and delivery of small doses.7,14-17 Furthermore, among commercially insured patients, studies have demonstrated the health-economic benefits of using insulin pen devices.18-22 Improved adherence, lower rates of hypoglycemia, improved A1C control, and potential cost advantages were shown for insulin pen devices. However, these studies were not conducted specifically in older patients, and evidence is needed regarding the impact of the use of pen devices versus vial-and-syringe delivery systems in this growing population of patients with T2D. Therefore, the present study was conducted to examine the differences in clinical, healthcare utilization, and costs outcomes associated with insulin delivery by pen device or vial and syringe among insulin-naïve Medicare patients with T2D.


Study Design and Patients

This was a retrospective database study using Humana medical, pharmacy, and enrollment data linked to laboratory results. Humana’s administrative claims database contains data for more than 12 million members of Medicare, commercial, and Medicaid health plans. The database includes claims information from more than 5 million Medicare members through Medicare Advantage plans, as well as prescription drug coverage data from throughout the United States. The full study period was from January 1, 2007, to December 31, 2012, which included an identification period from January 1, 2008, to December 31, 2011. The study included a 12-month, fixed-length, pre-index period before the index date and a post index period after the index date. The index date was defined as the first date of prescription of insulin glargine pen device or vial and syringe. The study design is depicted in

Figure 1


Data were included from patients 65 years or older on the index date they were diagnosed with T2D (defined as ≥1 inpatient or 2 physician visits dated ≥30 days apart with a primary or secondary diagnosis of T2D using International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 250.x0 or 250.x2) and who initiated treatment with insulin glargine between January 1, 2008, and December 31, 2011 (identification period), delivered by either pen device (pen initiators) or vial-and-syringe delivery systems (vial-and-syringe initiators). Included patients were also required to have medical and pharmacy Medicare Advantage coverage with continuous enrollment from 12 months before the index date (pre-index period) to 12 months after the index date (post index period); ≥1 prescription for ≥1 oral antidiabetes drug or glucagon-like peptide-1 receptor agonist in the pre-index period; and A1C results available in the pre-index period or within 15 days after the index date. Data were excluded from patients who had any insulin use during the pre-index period or initiated with both insulin pen device and vial-and-syringe on the same day.

Outcomes Analyses

Propensity score matching (PSM) (1:1 ratio) using logistic regression was used to match pen initiators with vial-and-syringe initiators and to adjust for observed differences in pre-index demographic and clinical characteristics. Standardized differences for pre-index characteristics were examined to verify the success of PSM. Variables included in the PSM model were determined following review of the pre-matching descriptive analysis of patient characteristics and other pre-index measures and are presented in

eAppendix Table 1

(available at Pre- and post index data were compared between pen initiators and vial-and-syringe initiators.

Pre-index patient demographic and disease characteristics were evaluated. Comorbidities of interest and Deyo CCI were examined using ICD-9-CM codes. Pre-index antidiabetes treatment and healthcare resource utilization and costs were also examined.

Primary post index outcome measures were A1C levels, change from pre-index to post index A1C level, hypoglycemia, treatment persistence, and treatment adherence. Secondary outcomes included daily average consumption (DACON) of insulin, healthcare resource utilization, and healthcare costs. Hypoglycemic events were defined as healthcare encounters (outpatient, inpatient, or ED visit) with a primary or secondary ICD-9-CM diagnosis code for hypoglycemia (ICD-9-CM code 250.8—diabetes with other specified manifestations; 251.0–hypoglycemic coma; 251.1–other specified hypoglycemia; or 251.2–hypoglycemia, unspecified).23 Treatment persistence was defined as a patient remaining on the study drug during the post index period without discontinuation or switching after study drug initiation.18,21,24-26 Study medication was considered discontinued if the prescription was not refilled within the expected time of medication coverage (the 90th percentile of the time, stratified by the metric quantity supplied, between first and second fills among patients with ≥1 refill).21,24,26 Patients who restarted their initial medication after a period without it during the post index period were considered nonpersistent. Sensitivity analyses were also conducted using 75th and 95th percentiles of the expected time of medication coverage.

Treatment adherence was measured by the adjusted medication possession ratio (MPR), which takes into account the difference in insulin device package size during the 1-year post index period,25 and defined as adjusted MPR greater than or equal to 0.8. The adjusted MPR was calculated as: traditional MPR of index insulin device × (average days between prescription refills for index insulin devices/average days of supply for patients using index insulin devices); the traditional MPR is the total days’ supply of all filled study drug prescriptions in the analysis period divided by the number of days in the analysis period. Insulin utilization, measured by DACON, was calculated as the total number of units dispensed before the last refill of the study drug divided by the total number of days between initiation and last refill during the post index period.

Healthcare resource utilization included physician office visits, outpatient visits (categorized as any outpatient visits not classified as a physician office visit), ED visits, inpatient admissions, and inpatient length of stay (days); diabetes-related healthcare resource utilization included claims with a primary or secondary diagnosis of diabetes (ICD-9-CM code 250.xx). Healthcare costs were computed as paid allowed amounts of adjudicated claims, including insurer and health-plan payments, co-payments, and deductibles (regardless of diagnosis) in the follow-up period.

All costs were presented as allowed amounts. Diabetes-related healthcare costs included costs from medical claims with a primary or secondary diagnosis of diabetes (ICD-9-CM code 250.xx), antidiabetes medications, glucose meters, and test strips. Adjustment for inflation (to 2012 US$) was made by multiplying each year’s cost by the Medical Care Consumer Price Index.

Statistical Analyses

Among matched patients, pre-index characteristics, clinical outcomes, and economic parameters were summarized and compared, with P values provided by the paired t test for continuous variables and the McNemar’s test for categorical variables. Statistical significance was determined when P <.05. Multivariate ordinary least squares (OLS) regression analysis was performed on A1C reduction for the subgroup of patients with A1C records in both pre- and post index periods. All analyses of data were conducted using SAS version 9.1 (SAS Institute, Cary, North Carolina).

RESULTSPatient Selection and Pre-Index Characteristics

In total, 108,915 patients with Medicare Advantage coverage who had ≥1 insulin glargine pen or vial-and-syringe prescription were identified; of these, 4876 met all inclusion criteria. Data from 2112 pen initiators and 1977 vial-and-syringe initiators with both pre-index and post index A1C results were available for the analyses. After PSM, the final matched sample contained data from 3202 patients (n = 1601 per cohort). Patient demographic and clinical characteristics were well balanced after PSM (

Table 1

); at pre-index, the overall patient population had a mean ± SD age of 74 ± 6 years, half were men, and the mean ± SD A1C was 8.6% ± 1.8%. Healthcare resource utilization and cost characteristics were also well-balanced between groups at pre-index (Table 1).

Treatment Persistence and Insulin Utilization

Pen initiators were significantly more persistent with treatment (58.5% vs 50.8%; P


for significantly longer periods of time (mean


SD; 307.5

± 83.1

vs 280.7

± 99.1



P <.0001) than vial-and-syringe initiators. The sensitivity analyses conducted using the 75th and 95th percentiles of the expected time of medication coverage yielded similar results (data not shown). Adherence (adjusted MPR ≥0.8) was 71.5% and 63.1% in the pen initiator and vial-and-syringe initiator groups, respectively (P <.0001).

Insulin utilization was similar between the 2 groups (mean DACON in units per day ± SD for pen initiators versus vial-and-syringe initiators: 24.48 ± 25.3 vs 24.62 ± 13.9 units per day, respectively; P

= .8524).

A significantly higher proportion of patients switched from vial-and-syringe to pen delivery (9.5%) than switched from pen to vial-and-syringe delivery (7.1%) during the post index period period (P = .0154).

Glycemic Control

The mean change from pre-index A1C level was significantly greater in pen initiators than vial-and-syringe initiators (−0.81% vs −0.61%; P = .0102) (

Figure 2

). Because a subgroup of patients (~80% of the study population) included in the PSM group had available A1C results in the post index period, a post hoc multivariate analysis was performed to further adjust for any confounding; this demonstrated a 0.13 points greater decrease in A1C for pen initiators than vial-and-syringe initiators (P = .0193).


The proportion of patients with hypoglycemic events of any type was similar between pen initiators and vial-and-syringe initiators (

Figure 3A

). The average number of hypoglycemic events per patient was significantly lower in the pen initiator group than in the vial-and-syringe initiator group (0.29 vs 0.49 events/patient, respectively; P = .0265) (

Figure 3B

). This was mainly driven by the greater number of outpatient events in the vial-and-syringe initiator group versus the pen initiator group (0.23 vs 0.07 events/patient, respectively; P = .0312).

Healthcare Utilization and Costs

The percentage of patients who had an ED visit or hospital admission was lower in the post index period than in the pre-index period (Table 1 and

Table 2

). In general, healthcare resource utilization was similar between groups (Table 2). However, pen initiators had a significantly higher number of patients with all-cause and diabetes-related outpatient visits compared with vial-and-syringe initiators (P = .0011 and P

= .0370,

respectively), but significantly fewer all-cause and diabetes-related outpatient pharmacy claims


P = .0002 and P <.0001, respectively). There was a significantly greater need for diabetes-related glucose meters and supplies among the vial-and-syringe initiators than the pen initiators (P


Although allowed diabetes-related pharmacy charges were significantly higher in the pen initiator group compared with the vial-and-syringe initiators ($2110 vs $1869, respectively


P <.0001), allowed total all-cause healthcare charges and allowed total diabetes-related charges were not significantly different between pen initiators and vial-and-syringe initiators (P = .0716 and P = .2499, respectively) (Table 2). Post index out-of-pocket expenses among pen initiators and vial-and-syringe initiators demonstrated a similar pattern (eAppendix Table 2).


In this retrospective claims analysis of data from US Medicare Advantage-eligible patients with T2D initiating insulin glargine treatment, we found that the use of a pen device for insulin delivery was more advantageous than a vial-and-syringe delivery system. Patients initiating insulin using a pen device had significantly better adherence and persistence than those using vial-and-syringe delivery systems, with greater A1C reduction and fewer hypoglycemic events per patient. These outcomes were achieved with no significant overall cost difference for the pen initiators compared with vial-and-syringe initiators despite higher pharmacy-related costs reported for patients using the insulin pen device. Healthcare utilization and costs showed an overall decrease in both groups compared with pre-index, suggesting a benefit for insulin in better controlling diabetes.

Our data add to and support previous findings from comparable analyses of the impact of the pen device compared with vial-and-syringe delivery systems in a range of patients with T2D in a variety of settings.7,14-22 Other reports have similarly described lower rates of hypoglycemia18,22,27 and improved A1C control20,22 with pen devices. The clinical significance of the difference in A1C between pen and vial-and-syringe initiators in our cohort is debatable.28 However, the greater A1C change suggests that patients in the pen initiator group were more likely to achieve the quality goal of an A1C <8%.

Analysis of quality goal attainment (A1C <8%) in a similar cohort of elderly patients suggests goal attainment is associated with a cost savings of $1445 per patient per year.29 This is of particular significance given the increasing move toward the accountable care organization model with its implicit link between attainment of quality goals and reimbursement.30 Furthermore, elderly patients are more vulnerable to hypoglycemia, particularly during treatment intensification, and incidences of hypoglycemia are associated with the exacerbation of existing comorbidities in this population.31,32 The achievement of increased glycemic control alongside a lower rate of hypoglycemia suggests an overall improvement in clinical status in elderly pen initiators.

Benefits of the pen device are further borne out in literature reviews, having been shown to improve persistence and/or adherence compared with vial-and-syringe delivery18-22,27,33 and to decrease healthcare utilization.27,34 Several previous studies have reported findings similar to ours: no overall increase in costs associated with the use of a pen device compared with the delivery of insulin by vial and syringe.18-22 In contrast, 3 studies reported cost-savings with a pen device,19,27,34 although this could be explained by the lower mean age of patients enrolled in these studies. Benefits of the pen device are further borne out in reviews of the data15,35 and by an observed increase in the use of insulin pen devices in the United States among patients with T2D.36

To the best of our knowledge, the present study is the first national study extending the demonstrated advantages of pen devices for insulin delivery to elderly patients with T2D. Our findings are especially relevant due to the growing elderly population, which is a challenging group of patients to treat due to comorbidities and reduced function37 and is, accordingly, an expensive population in which to manage T2D. Pen devices are expected to be helpful for elderly patients as a whole and particularly those with functional and visual impediments who may benefit from the pen’s ease of use.15,35


Limitations include incomplete data or the presence of coding errors due to administrative claims being designed for reimbursement rather than research. Use of pharmacy refills to estimate adherence assumes that patients use medications as prescribed, and do not obtain medications from other sources or waste or hoard supplies. As the study used administrative claims data, the burden of illness is evaluated based on direct medical costs only, excluding costs incurred outside of the healthcare system. As only data from US Medicare Advantage patients in the Humana database were used in the analyses, the results may not be generalizable beyond this population; however, Humana does offer coverage nationwide. As in other claims-data-based studies, certain relevant data are lacking, such as laboratory results on glycemic control for the entire population, weight and health behavior data, and changes in other antidiabetes medications. Although hypoglycemia events resulting in medical encounters were reported, it is possible that not all events were accounted for in billing codes for the visit; therefore, some events (including those not requiring medical encounters) may not have been captured in the analysis. The analysis of changes in A1C levels only in patients with both pre- and post index laboratory results might have led to selection of a population that differed from the matched overall population. Although PSM and multivariate OLS regression modeling were used to minimize selection bias and strengthen causal inferences by controlling for both economic (Medicare plan type, Medicare coverage gap status, low-income subsidy status, dual-eligibility patients) and clinical (eg, comorbidities, A1C, hypoglycemia) factors, it is possible that selection bias based on unobservable variables remained. Changes in use of classes of antidiabetic medications post insulin initiation that have potential to influence the occurrence of study outcomes were not accounted for in these analyses; no conclusions regarding the use of classes of medication and outcomes in the post index period can be made.


Medicare Advantage patients with T2D from a large managed-care organization who initiated insulin therapy with a pen device were found to be more adherent and persistent with therapy, with a significantly greater decrease in A1C levels and less hypoglycemia than those who initiated insulin with a vial-and-syringe delivery system. Despite higher pharmacy costs for patients using the pen device compared with vial-and-syringe delivery system, the overall healthcare costs in both groups were similar. However, these results should be confirmed in larger, long-term studies of the population 65 years or older. These findings extend the previously demonstrated benefits of initiating insulin therapy with a pen device to the elderly population and suggest that improved clinical outcomes are possible in this difficult-to-treat patient group.


The authors received writing/editorial support in the preparation of this manuscript from Rosalie Gadiot, PhD, of Excerpta Medica, funded by Sanofi US, Inc.