The association between use of mealtime insulin pens versus vials and healthcare charges and resource utilization in patients with type 2 diabetes: a retrospective cohort study

Abstract

Objective:

To compare all-cause and diabetes-related resource utilization and healthcare charges among adults with type 2 diabetes mellitus who initiated therapy with mealtime insulin disposable pens or vials.

Methods:

Data were obtained from the Innovus inVision™ database from January 1, 2006 through June 30, 2010. Generalized linear models with a gamma distribution and log link estimated the association between medical charges and use of mealtime insulin pens vs vials in the 1 year post-index date, while generalized linear models with a negative binomial distribution estimated resource utilization.

Results:

Controlling for patient characteristics, general health, and patient copayments, insulin therapy initiation with disposable pens, compared to vials, was associated with significantly fewer all-cause hospitalizations (1.45 vs 1.66; p < 0.0001) as well as a significantly shorter hospital length of stay (2.16 days vs 3.53 days; p < 0.0001). Pen use, compared to vials, was also associated with significantly fewer diabetes-related hospitalizations (1.36 vs 1.47; p < 0.0001), and shorter hospital length of stay (1.12 days vs 1.72 days; p < 0.0001). Despite higher diabetes-related drug charges ($3593 vs $2755; p < 0.0001) associated with the use of pens, results showed significantly lower all-cause total healthcare charges ($42,150 vs $53,340; p < 0.0001) and significantly lower diabetes-related total healthcare charges ($12,722 vs $14,540; p < 0.0001) for patients who initiated therapy on mealtime insulin with pens compared to vials.

Limitations:

Data were drawn from administrative claims and included only patients with medical and outpatient prescription drug benefit coverage. Hence, the results may not be generalizable. The retrospective analyses relied on diagnostic codes to identify patients, assess patient general health, and determine other values, rather than formal, clinical assessments. The analyses did not include indirect healthcare costs.

Conclusions:

The administration of mealtime insulin via disposable pens, compared to vials, was associated with a significant reduction in all-cause and diabetes-related resource utilization and total healthcare charges.

Keywords::

• Type 2
• Diabetes
• Pens
• Vials
• Charges
• Resource utilization

Abbreviations::

• ER, emergency room – HbA1c, hemoglobin A1c, HIPAA, Health Insurance Portability and Accountability Act – T2DM, type 2 diabetes mellitus

Introduction

A metabolic disease characterized by insufficient insulin sensitivity or secretion leading to hyperglycemia1, diabetes mellitus is the leading cause of kidney failure, lower-limb amputations (non-trauma related), and new cases of blindness among adults in the US, as well as one of the foremost causes of heart disease, stroke, and death2. The majority of diabetes mellitus cases (90–95%) are of type 2 (T2DM)1. In the US, T2DM affects between 23.2–24.5 million individuals, or between 7.5–7.9% of the current US population of 309 million2,3; these numbers are predicted to double over the next 25 years4. In 2007, the US economic burden of T2DM was estimated to be $159.5 billion annually, with direct costs of medical treatment totaling $105.7 billion5.

Treatment guidelines for T2DM recommend oral anti-diabetic drugs as first line pharmacotherapy6. For patients who fail to respond adequately or experience worsening glycemic control, insulin therapy may be initiated7. Patients are generally prescribed basal insulin and then prescribed rapid-acting insulin for use at mealtimes if the patient’s fasting glycemia is within the target range, but overall hemoglobin A1c (HbA1c) ≥7% after 2–3 months on the basal dose8. Research has shown that use of insulin therapy can improve insulin sensitivity and reduce cardiovascular mortality9. However, despite these benefits, evidence suggests that the proportion of patients with T2DM who use insulin has decreased from 24.2% to 16.4% in the US in the year 2000, compared to use in the prior decade10. Furthermore, research has also shown that, over the same time period, the percentage of individuals who achieved glycemic control, as measured by HbA1c level <7%, declined from 44.5% to 35.8%10.

While treatment guidelines emphasize the importance of prescribing the right type of insulin at the appropriate time8, recent studies have indicated that the method used to supply insulin to the body may also affect patient outcomes11–17. The conventional way that exogenous insulin has been supplied to the bloodstream is via subcutaneous injection, using a syringe and vial11. However, patient aversion to self-injection using the vial method has been shown to have a negative impact on adherence18,19. To make insulin therapy less complicated and fear-inducing, the insulin pen, which combines the insulin reservoir and syringe into one unit, was introduced in 198514,20,21. The results of a number of studies suggest that patients prefer pens to vials22.

The purpose of the present retrospective, naturalistic analysis was to examine the impact of initiation with either the pen or the vial method of insulin delivery on healthcare resource utilization and medical charges. To this end, this study compared the insurance claims of a nationwide sample of patients with T2DM who initiated mealtime insulin therapy with either the pen or the vial delivery method. The results of this analysis will, therefore, inform healthcare providers and payers as to the relative benefits associated with insulin delivery method.

Patients and methods

The Innovus InVision™ Data Mart provided the data for this study. A database of retrospective claims from a diverse group of US health plans, the Innovus InVision™ Data Mart, contains person-specific information, including patient characteristics, enrollment dates, inpatient and outpatient medical claims, and outpatient pharmaceutical claims. For a sub-set of these persons, the database also provides laboratory test results. The database is fully compliant with the Health Insurance Portability and Accountability Act (HIPAA).

To be included in this study, an individual had to have filled a prescription for a mealtime insulin analog (insulin aspart, insulin glulisine, insulin lispro, insulin lispro mix 75/25 (75% insulin lispro protamine suspension [ILPS], 25% insulin lispro), insulin lispro mix 50/50 (50% ILPS, 50% insulin lispro, or biphasic insulin aspart 70/30 (70% insulin aspart protamine suspension and 30% insulin aspart injection)) via pen or vial, with the date of the first such prescription identified as the index date. Furthermore, individuals were required to have received at least two diagnoses of T2DM over the time period from 6 months prior to the index date (i.e., the pre-period) through 1 year post-index date (i.e., the post-period). Patients were excluded from the study if they switched method of insulin delivery (i.e., from pens to vials or vice versa) at any time in the post-period or if they filled a prescription for cartridge mealtime insulin, 3 mL vials, inhaled insulin, or an insulin pump at any time over the study period. Patients were also excluded if they received a diagnosis of type 1 diabetes, secondary diabetes, or gestational diabetes over the study period. Finally, patients were required to have continuous insurance coverage over the study period and to be at least 18 years of age on the index date.

The study examined two classes of patient-level, health economic outcomes over the 1 year post-period: healthcare resource utilization and direct healthcare charges. Resource utilization included the number of hospitalizations and length of stay (days), number of outpatient visits, and number of emergency room (ER) visits. Resource use was analyzed as all-cause and diabetes-related, with the latter defined as a resource utilization occurrence with an accompanying diagnosis of T2DM. The healthcare charges analyses considered charges incurred over the post-period for inpatient, outpatient, ER and pharmacy charges. These analyses also examined all-cause charges as well as diabetes-related charges, where diabetes-related charges were defined as charges with an accompanying diagnosis of T2DM, use of an anti-diabetic medication, or use of diabetic supplies. All charges were converted to 2010 dollars using the medical component of the Consumer Price Index.

In order to examine the outcomes of interest while controlling for other possible influencers, multivariate analyses were estimated. Specifically, generalized linear models with a negative binomial regression were employed to estimate resource utilization, while generalized linear models with a gamma distribution and log link were employed to estimate charges. For the models of inpatient and ER charges, two-part models were utilized due to the large number of individuals who did not have such charges. In these cases, the first part utilized a logistic regression model to estimate the probability of being hospitalized or having an ER visit, respectively, while the second part estimates charges for those patients who actually utilized the service under consideration23. While examining the association between method of delivery and patient outcomes, the analyses controlled for patient characteristics, initial prescription information, patient general health, and method of delivery. Patient characteristics included age, sex, region of residence, and type of insurance coverage, while initial prescription information included the copayment associated with the initial prescription and whether the prescribing physician was an endocrinologist or not. Patient general health was assessed over the pre-period and was proxied by the number of unique medications prescribed, whether the patient was hospitalized, and the Charlson comorbidity index24,25. Also, insulin method of delivery (pens or vials) was included in all analyses. All analyses were conducted using SAS, version 9.2, and a significance level of <0.05 was considered to be statistically significant.

Results

The study inclusion/exclusion criteria resulted in a sample of 8374 individuals—4429 of whom initiated therapy with mealtime pens and 3945 of whom initiated therapy with mealtime vials. Table 1 illustrates how each inclusion/exclusion criterion affected sample size.

Table 1. Inclusion/exclusion criteria and sample size.

Criteria	Pens	Vials	Total
Receipt of mealtime insulin, with first such date identified as index date	60,466	88,238	148,704
Receipt of at least two diagnoses of type 2 diabetes from pre-period through post-period	50,955	66,798	117,753
No switching of method of delivery in the post-period	40,886	52,832	93,718
No use of cartridge mealtime insulin or 3 mL vials from pre-period through post-period	40,886	52,809	93,671
No diagnosis of type 1 diabetes in pre- or post-period	21,107	20,047	41,154
No diagnosis of secondary diabetes in pre- or post-period	20,954	19,933	40,887
No diagnosis of gestational diabetes in pre- or post-period	20,459	19,488	39,947
No receipt of inhaled insulin in pre- or post-period	20,412	19,461	39,873
No receipt of insulin pump in pre- or post-period	20,249	19,098	39,347
Continuous insurance coverage from start of pre-period through end of post-period	4437	3957	8394
Age at least 18 on index date	4429	3945	8374

Table 2 compares the descriptive statistics of patients who initiated therapy with a pen and those who initiated therapy with a vial. Results indicated that patients who initiated therapy with a pen were significantly younger, more likely to reside in the Northeast or Midwest regions of the US, more likely to be insured via a health maintenance organization, and more likely to have received their initial prescription from an endocrinologist. In contrast, the pen cohort was significantly less likely to reside in the South or West or to be insured via a preferred provider organization. Patients in the pen cohort were also found to have significantly higher copay associated with their initial prescription. They were also in better general health: although having a slightly higher number of prescriptions in the pre-period, a lower percentage of pen patients were hospitalized, and such patients had a lower mean Charlson score compared to individuals who initiated with vials.

Table 2. Descriptive statistics—entire cohort.

Characteristic	Pens (n = 4429)		Vials (n = 3945)		p-value
	n or Mean	% or SD	n or Mean	% or SD
Age (Mean; SD)	53.99	10.74	55.58	12.07	<0.0001
Sex					0.7260
Female	1978	44.66	1796	45.53
Male	2450	55.32	2148	54.45
Unknown	1	0.02	1	0.03
Region					<0.0001
Northeast	290	6.55	198	5.02
Midwest	1434	32.38	949	24.06
South	2299	51.91	2256	57.19
West	404	9.12	541	13.71
Other	2	0.05	1	0.03
Insurance					<0.0001
Exclusive Provider Org	578	13.05	489	12.40
Health Maintenance Org	674	15.22	394	9.99
Independent	158	3.57	231	5.86
Other	7	0.16	5	0.13
Point of Service	2689	60.71	2349	59.54
Preferred Provider Org	323	7.29	477	12.09
Prescribing Physician					<0.0001
Endocrinologist	427	9.64	228	5.78
Primary Care Physician/Other	4002	90.36	3717	94.22
Copayment (Mean; SD)	34.54	27.51	24.93	24.25	<0.0001
Pre-period variables
% Hospitalized	1017	22.96	1412	35.79	<0.0001
Charlson Score (Mean; SD)	2.02	1.62	2.31	1.87	<0.0001
# of Medications Rx (Mean; SD)	9.66	5.83	9.10	6.49	<0.0001

Table 3 shows results for the association between mealtime insulin delivery system and healthcare resource utilization, controlling for patient characteristics, general health, and initial prescription information. For the all-cause resource utilization results, all differences between the cohorts achieved statistical significance. Specifically, compared to the vial cohort, the pen cohort had a lower mean number of hospitalizations (1.45 (SD = 0.40) vs 1.66 (SD = 0.62)), shorter hospital length of stay (2.16 days (SD = 3.43) vs 3.53 days (SD = 5.16)), and fewer outpatient visits (19.51 (SD = 10.94) vs 21.58 (SD = 13.18)) during the post-period. However, the results showed that the pen cohort had a greater mean number of ER visits compared to the vial cohort (0.46 (SD = 0.33) vs 0.43 (SD = 0.33)). When examining diabetes-related resource utilization, results revealed that use of pens, compared to vials, on average was associated with fewer ER visits (0.056 (SD = 0.13) vs 0.062 (SD = 0.13); p = 0.0473), fewer hospitalizations (1.36 (SD = 0.26) vs 1.47 (SD = 0.31); p < 0.0001), and shorter hospital length of stay (1.12 days (SD = 1.24) vs 1.72 days (SD = 1.85); p < 0.0001). There was no significant difference between the two cohorts with regards to the number of diabetes-related outpatient visits.

Table 3. Resource utilization and charges—multivariate regression.

	Pens (n = 4429)		Vials (n = 3945)		p-value
	Mean	SD	Mean	SD
All-Cause Resource Utilization
Number of Hospitalizations	1.45	0.40	1.66	0.62	<0.0001
Hospital Length of Stay	2.16	3.43	3.53	5.16	<0.0001
Number of Outpatient Visits	19.51	10.94	21.58	13.18	<0.0001
# of Emergency Room Visits	0.46	0.33	0.43	0.33	<0.0001
Diabetes-Related Resource Utilization
Number of Hospitalizations	1.36	0.26	1.47	0.31	<0.0001
Hospital Length of Stay	1.12	1.24	1.72	1.85	<0.0001
Number of Outpatient Visits	7.35	1.56	7.33	1.72	0.5152
# of Emergency Room Visits	0.056	0.13	0.062	0.13	0.0473
All-Cause Healthcare Charges
Inpatient	11,959.6	14,467.2	19,922.0	21,572.5	<0.0001
Outpatient	20,745.6	30,974.0	25,666.8	34,214.5	<0.0001
Emergency Room	196.3	355.4	190.0	357.6	0.4180
Drug	8694.8	4303.3	8299.7	4692.5	<0.0001
Total	42,149.7	47,011.2	53,339.8	57,756.3	<0.0001
Diabetes-Related Charges
Inpatient	4351.6	2760.5	6451.1	4019.9	<0.0001
Outpatient	4659.6	2299.9	5340.3	2803.3	<0.0001
Emergency Room	91.6	242.3	83.8	251.7	0.1535
Drug	3592.6	793.3	2755.4	656.3	<0.0001
Total	12,721.9	4892.8	14,539.7	6239.8	<0.0001

Resource utilization estimated with generalized linear models with negative binomial distribution. Charges estimated with generalized linear model with gamma distribution and log link. Independent variables include: age, sex, region of residence, type of insurance plan, copayment, prescribing physician, pre-period number of unique medications prescribed, pre-period hospitalization, pre-period Charlson score, and use of pen (vs vial).

Note that all-cause and diabetes-related inpatient and ER charge regressions are two part models where the first part estimates the probability of being hospitalized (or having an ER visit) and the second part estimates charges for patients who utilized the service23.

The analysis also examined the association between method of delivery and medical charges. Results, provided in Table 3, revealed significantly higher diabetes-related drug charges ($3593 (SD = $793) vs $2775 (SD = $656); p < 0.0001) and all-cause drug charges ($8695 (SD = $4303) vs $8300 (SD = $4693); p < 0.0001) associated with the use of pens compared to vials. In contrast, all-cause inpatient, outpatient, and total healthcare charges, as well as diabetes-related inpatient, outpatient, and total healthcare charges, were found to be significantly lower for patients who initiated on pens. There were no significant differences between the two cohorts with respect to diabetes-related ER charges or all-cause ER charges.

As a test of the robustness of the results, the multivariate analyses were re-estimated using a Heckman two-stage sample selection model to control for the possible effects of sample selection bias26,27. Results of such models were generally consistent with the results presented and, hence, were not reported.

Discussion

This large-scale, nationwide, naturalistic study compares the outcomes of T2DM patients initiating mealtime insulin therapy with either a pen or a vial. The results revealed significant differences between the pen and vial users’ patient characteristics, resource utilization, and healthcare charges.

Patient characteristics

In concert with previous research17, the present analysis revealed that patients who initiated with a pen were younger and in better health relative to those who initiated with vials. Pen initiators were also more likely to live in the Northeast or Midwest, a finding in agreement with a previously published examination of predictors of pen vs vial initiation28. Earlier literature has explained the difference in geographical distribution between pen and vial users by noting that ‘pen initiators were more likely to… live in locations where the percentage of college graduates was higher’28, and other research has noted that better educated people tend to ‘be more knowledgeable about their health, and to have more time to attend to their health’29. A previous study has also indicated that younger adults are more likely to have higher health literacy relative to older adults30. Our findings thus suggest that pen users are more informed consumers of healthcare relative to vial users.

Another important difference between the pen and vial cohorts was that pen users were more likely to have received their initial prescription from an endocrinologist. This finding supports previous research indicating that an endocrinologist visit within the week before insulin initiation increases the likelihood of initiating with a prefilled pen vs a vial28. It has been hypothesized that endocrinologists may be more likely than other provider types to prescribe newer insulin devices ‘because they may have had more experience with them or because they may have had more support staff to recommend and discuss diabetes care options with patients’28. Other research has indicated that patients whose physicians encouraged pen use were over 100-times more likely to use a pen relative to those whose physicians did not discuss or discouraged pen use31.

Healthcare resource utilization and charges

Results of this study revealed that, in general, initiation on mealtime insulin with a pen was associated with less resource utilization and lower healthcare charges relative to initiation on vials. Specifically, all-cause inpatient, outpatient, and total healthcare charges, as well as diabetes-related inpatient, outpatient, and total healthcare charges, were found to be significantly lower for patients who initiated on pens. These findings were similar to those of previous retrospective studies, which reported that patients who switched from vial use to pen use had lower annualized total all-cause13,16,17 and diabetes-related13,17 healthcare costs. However, those previous studies examined patients who switched to, rather than initiated with, insulin pens, and they examined healthcare costs rather than charges. Thus, comparisons to these earlier studies should be made with caution due to the significant differences in research design.

Consistent with the finding of lower healthcare charges, results of the multivariate analyses also revealed generally lower resource utilization associated with pens compared to vials. Such a finding is consistent with research that has found lower all-cause outpatient visits and hospitalizations, as well as shorter hospital length of stay for patients who switched to insulin pen therapy16. Furthermore, previous research also found shorter hospital length of stay for hypoglycaemia-related visits16,18,19 as well as fewer ER visits for hypoglycemic events or other diabetes-related causes16. Similarly, our research found fewer diabetes-related ER visits; however, in contrast, a small, but statistically significantly greater number of all-cause ER visits associated with pen use, while previous research has found reduced all-cause ER visits associated with switching to pens from vials16. Despite the small difference in all-cause ER visits of the present study, there were no differences in the charges associated with those visits.

Finally, the present study indicated prescription drug charges to be higher for pen users relative to vial users, a finding in concert with previous research which found prescription costs to be higher for pen users17. The lower total all-cause charges and diabetes-related charges of pen users in the present study showed that pen use was associated with overall charge reductions which substantially outweighed additional drug charges. However, the higher charges related to pen use may, at least in part, explain why approximately half of T2DM patients still initiate using a vial and that the overall proportion of pen users is relatively low32.

Limitations

As with any research, the findings presented here should be interpreted within the context of the limitations of the study design. First, this analysis was conducted using an administrative claims database which included only patients with medical and outpatient prescription drug benefit coverage. Furthermore, the analyses only focused on patients with type 2 diabetes. Therefore, the results may not be generalizable across patients with less comprehensive insurance and for all patients with diabetes. Second, retrospective analyses must rely upon diagnostic codes to identify patients, assess patient general health, and determine other values, whereas formal diagnostic assessments may be more reliable. Third, this analysis was only able to measure healthcare charges and not the actual costs reimbursed by third parties. Finally, this analysis was not able to measure any indirect healthcare costs, such as lost productivity or caregiver burden.

Conclusions

In conclusion, this analysis compared the patient demographics, resource utilization, and healthcare charges of a nationwide sample of patients with T2DM who initiated mealtime insulin using either a pen or a vial. After controlling for differences in patient characteristics, general health, and index prescription characteristics, this study revealed that the pen users had generally less resource utilization and lower all-cause and diabetes-related total healthcare charges, despite the fact that they had higher medication charges. The findings of this study suggest that an increase in outpatient pen use may lead to a decrease in the economic burden of diabetes nationwide via the reduction in healthcare charges.

Transparency

Declaration of funding

Eli Lilly and Company was the sponsor of this study. ELE and KSB had primary responsibility for the study design, while MJL had primary responsibility for the analyses and drafting of the manuscript. All authors reviewed and approved the final manuscript.

Declaration of financial/other relationships

ELE and KSB are employees and stockholders of Eli Lilly and Company. MJL is Managing Member of HealthMetrics Outcomes Research, LLC, and was a contractor to Eli Lilly and Company on this project.

Acknowledgments

The authors thank Patricia Platt for her assistance in writing the manuscript.