Jurnal Primer Tugas 2 b

Clinical Therapeutics/Volume 38, Number 1, 2016

Clinical and Economic Outcomes Associated With the Timing of Initiation of Basal Insulin in Patients With Type 2 Diabetes Mellitus Previously Treated With Oral Antidiabetes Drugs Philip Levin, MD1; Steve Zhou, PhD2; Emily Durden, PhD3; Amanda M. Farr, MPH3; Jasvinder Gill, MD2; and Wenhui Wei, PhD2 1

MODEL Clinical Research, Towson, Maryland; 2Sanofi US, Bridgewater, New Jersey; and 3Truven Health Analytics, Ann Arbor, Michigan

ABSTRACT Purpose: In patients with type 2 diabetes mellitus (T2DM) not achieving glycemic targets using oral antidiabetes drugs (OADs), studies suggest that timely insulin initiation has clinical benefits. Insulin initiation at the early versus late stage of disease progression has not been explored in detail. This retrospective database analysis investigated clinical and economic outcomes associated with the timing of insulin initiation in patients with T2DM treated with Z1 OAD in a real-world US setting. Methods: This study linked data from the Truven Health MarketScan® Commercial database, Medicare Supplemental database, and Quintiles Electronic Medical Records database. A total of 1830 patients with T2DM were included. Patients were grouped according to their OAD use before basal insulin initiation (1, 2, or Z3 OADs) as a proxy for the timing of insulin initiation. Clinical and economic outcomes were evaluated over 1 year of follow-up. Findings: During follow-up the 1 OAD group, compared with the 2 and Z3 OADs groups, had a greater reduction in glycosylated hemoglobin A1c (
1.7% vs
1.0% vs
0.9%, respectively; P o 0.0001), greater achievement of glycemic target (38.2% vs 26.7% vs 19.6%, respectively; P o 0.0001), and a lower incidence of hypoglycemia (2.7% vs 6.6% vs 5.0%, respectively; P ¼ 0.0002), with no difference in total health care costs ($21,167 vs $21,060 vs $20,133, respectively). Implications: This study shows that early insulin initiation (represented by the 1 OAD group) may be clinically beneficial to patients with T2DM not controlled with OADs, without adding to costs. This supports the call for timely initiation of individualized insulin therapy in this population.

110

(Clin Ther. 2016;38:110–121) & 2016 The Authors. Published by Elsevier HS Journals, Inc. Key words: early insulinization, economic outcomes, insulin initiation, type 2 diabetes.

INTRODUCTION After diagnosis of type 2 diabetes mellitus (T2DM), lifestyle changes and oral antidiabetes drugs (OADs) are recommended.1 If blood glucose targets are not achieved with first-line therapy after approximately 3 months, guidelines from the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD) advocate the addition of another OAD or insulin to the treatment regimen.1 Importantly, the treatment regimen should be tailored to each patient’s needs. However, because of the progressive nature of diabetes, most patients will eventually need insulin therapy to maintain or achieve glycemic targets (generally glycosylated hemoglobin A1c [A1C] o7.0%).1,2 Several clinical studies suggest that timely initiation of insulin therapy is beneficial for patients with diabetes.3–5 However, the differences between initiating insulin when a patient is unable to control hyperglycemia on an OAD or at a later stage of a patient’s disease progression when taking multiple OADs have not been examined fully. This is important because many patients with T2DM with inadequate glycemic control using OADs endure Accepted for publication November 16, 2015. http://dx.doi.org/10.1016/j.clinthera.2015.11.011 0149-2918/$ - see front matter & 2016 The Authors. Published by Elsevier HS Journals, Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Volume 38 Number 1

P. Levin et al. a prolonged glycemic burden before they begin insulin therapy.6,7 Clinical inertia, the failure to intensify treatment despite a patient not meeting glycemic targets, may be a contributing factor.8,9 Both patients and physicians are responsible for clinical inertia: concerns about hypoglycemia, weight gain, and injectable agents contribute to resistance to the initiation of insulin therapy.10 However, the association between the timing of insulin initiation and health care outcomes in real-world settings, including the economic impact, remains to be explored. This analysis assessed the clinical and economic outcomes associated with the timing of basal insulin initiation in US patients with T2DM previously treated with Z1 OAD.

METHODS Study Design and Patients This was a retrospective database analysis of combined data from 3 databases: 2 Truven Health MarketScan databases (Claims and Encounters [Commercial] and Medicare Supplemental and Coordination of Benefits [Medicare]) and the linked Quintiles Electronic Medical Records (EMR) database. These 3 sources provided data for patients insured commercially or through Medicare and information from their EMR. The MarketScan Commercial database contains the health care experience of  39.5 million employees and their dependents, covered under a variety of fee-forservice and managed-care health plans. The MarketScan Medicare database contains the health care experience of  3.4 million retirees with Medicare supplemental insurance paid for by employers. Both the MarketScan Commercial and Medicare databases provide detailed cost, use, and outcomes data for health care services performed in the inpatient and outpatient settings. The Quintiles EMR database contains ambulatory clinical data from 49000 providers and covers 413.5 million unique patient lives. A hybrid deterministic-probabilistic approach was used to link a subset of patients in the MarketScan databases to the Quintiles EMR database at the patient level. The deterministic match required an exact match on several variables from both sources, including the 3-digit ZIP code of residence, sex, and month and year of birth. The probabilistic match involved finding exact matches for Z3 physician visits. Physician visits were selected as the attribute for matching due to likely agreement between medical

January 2016

records and claims data. In total, 1,348,279 patients were included in the linked MarketScan and Quintiles database. Patients included in the study were 18 years of age or older and received a diagnosis of T2DM between July 1, 2004 and December 31, 2011. Eligible patients were defined as having had Z1 inpatient visit or Z2 non-inpatient visits (Z30 days apart) with a primary or secondary diagnosis of T2DM (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] diagnosis code 250.x0 or 250. x2).11 Patients initiated on a basal insulin (neutral protamine Hagedorn [NPH] insulin, insulin glargine, or insulin detemir) between July 1, 2004 and December 31, 2010 (index date) and who were also receiving Z1 class of OADs (metformin, sulfonylureas, thiazolidinediones, α-glucosidase inhibitors, meglitinides, dipeptidyl peptidase-4 inhibitors) at baseline were included. Patients were required to have had continuous health care coverage for 6 months before (baseline) and 12 months after (follow-up) initiation of insulin. Baseline A1C values and weight measurements were available for all patients. Patients were excluded from the study if they had filled any prescription claims for insulin during the baseline period other than rapid-acting insulin claims within 15 days before insulin initiation, which could be initiated in addition to basal insulin as part of a basal bolus insulin treatment regimen. The study population was stratified by the number of baseline OADs (1, 2, or Z3) as a proxy for the timing of insulin initiation relative to diabetes disease progression. This was based on previous12 and current1 ADA guidelines on pharmacological therapy for hyperglycemia in T2DM in which the recommendation is to start with OAD monotherapy (metformin or a sulfonylurea), and insulin may be added to either an OAD or subsequently to a combination of OADs.

Baseline Measures Patient demographic and clinical variables comprised sex, age, weight, body mass index (BMI), duration of diabetes, comorbidities, Deyo-modified Charlson Comorbidity Index,13 hypoglycemic events, and A1C values for the 6-month baseline period or 15 days after the index date. If there were multiple measurements during the baseline period, the value from the test closest to the index date was used.

111

Clinical Therapeutics Health care resource utilization was described by place of service (outpatient office visits, emergency department [ED] visits, and inpatient admissions) for the 6-month baseline. Diabetes-related health care resource utilization included claims with a primary or secondary diagnosis of diabetes (ICD-9-CM diagnosis code 250.xx). Health care costs were computed as total paid amounts of adjudicated claims during the 6-month baseline and were annualized for comparability. Diabetes-related health care costs included costs from medical claims with a primary or secondary diagnosis of diabetes (ICD-9-CM diagnosis code 250. xx), antidiabetes medications, and glucose meters and test strips. All costs were adjusted to 2011 US dollars using the medical care component of the US consumer price index.14

Endpoint Measures Clinical outcomes were evaluated over 1 year of follow-up. They included A1C reduction from baseline and the proportion of patients achieving A1C o7.0%, and changes in body weight and BMI; values reported within 3 months of the end of the 12-month follow-up period were included; if there were multiple measurements the value from the test closest to the end of follow-up was used. Hypoglycemia in the follow-up period was also reported; hypoglycemia was defined as a health care encounter with a primary or secondary diagnosis of hypoglycemia (ICD-9-CM diagnosis codes 251.0, hypoglycemic coma; 251.1, other specified hypoglycemia; or 251.2, hypoglycemia, unspecified)15 and other relevant codes using an algorithm developed by Ginde et al.16 The setting of a hypoglycemic event (outpatient office, inpatient admissions, or ED) was used as a proxy for its severity. Treatment persistence was defined as a patient remaining on study drugs during the followup period without discontinuation or switching after study drug initiation.17 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 in patients with Z1 refill).17 Patients who restarted their initial medication after a period without it during follow-up were considered nonpersistent. Health care resource utilization was described by place of service (outpatient office, ED, inpatient

112

admission) during the 1-year follow-up. Diabetesrelated health care resource utilization included claims with a primary or secondary diagnosis of diabetes (ICD-9-CM diagnosis code: 250.xx). Health care costs were computed as total amounts of adjudicated claims during the 1-year follow-up. Diabetes-related health care costs included costs from medical claims with a primary or secondary diagnosis of diabetes (ICD-9-CM diagnosis code: 250.xx), antidiabetes medications, and glucose meters and test strips.

Statistical Analysis Confounding factors are an issue in nearly all observational research in which patients are not randomly assigned to treatment cohorts, and, therefore, certain patient characteristics may be associated with both exposure and outcome. Such issues were addressed using the inverse probability-of-treatment weighting method,18–21 which adjusts for differences in baseline characteristics between study cohorts, rendering groups more comparable. A multinomial logistic regression model was used to predict the propensity of timing of insulin initiation (after 1, 2, or Z3 OADs) given the demographic and clinical characteristics of the patient; the variables included in this model were age group, sex, region, urbanicity, index year, index insulin type (basal or basal plus short acting), baseline high-density cholesterol, provider specialty (specialist, primary care, unknown), duration of diabetes, baseline hypoglycemia, baseline obesity, baseline depression, baseline cerebrovascular disease, baseline diabetes-related outpatient office visit, and baseline diabetes-related other outpatient services. The inverse of the propensity was applied as the probability weight to outcome measures to produce weighted results. Baseline results are presented unweighted, whereas outcomes were weighted using the inverse probability-of-treatment weighting method. Global tests (χ2 tests for categorical variables and ANOVA for continuous variables) were used to determine statistically significant differences between the patient groups for unweighted baseline characteristics and weighted study outcomes. Pairwise statistical comparisons between the 1, 2, and Z3 OADs groups were conducted using χ2 tests for categorical variables and t tests for continuous variables in unweighted and weighted results.

Volume 38 Number 1

P. Levin et al.

Patients with ≥ 1 prescription claim for basal insulina from July 1, 2004 through December 31, 2010 N = 26,092

Patients with T2DM diagnosed from July 1, 2004 through December 31, 2011 n = 23,794

Patients who initiated basal insulin from July 1, 2004 through December 31, 2010 n = 16,910

Patients aged ≥ 18 years at insulin initiation n = 16,688

Patients with continuous enrollment in a health plan with both medical and pharmacy coverage for 6 months prior to insulin initiation through 12 months after insulin initiation n = 7561

Patients with ≥ 1 OAD prescription claim during the 6 months prior to insulin initiation n = 5572

Patients meeting selection criteria n = 5572

Patients with non-missing weight values during the 6 months prior to and through 15 days after insulin initiation n = 2540

Patients with non-missing A1C test results during the 6 months prior to and through 15 days after insulin initiation (included in the study) n = 1830

Figure 1. Data attrition. *Insulin glargine, insulin detemir, or NPH insulin. A1C ¼ glycosylated hemoglobin A1c; EMR ¼ electronic medical record; OAD ¼ oral antidiabetes drug; T2DM ¼ type 2 diabetes mellitus.

RESULTS Baseline Patient Characteristics A total of 1830 eligible patients were included in the analysis; patient attrition details are shown in Figure 1. Of the included patients, 24.6% used 1 OAD at baseline, 40.3% used 2 OADs, and 35.1% used Z3 OADs (Table I). In the 1 OAD group, 46.0% of the patients were previously treated with metformin and 34.4% with a sulfonylurea. In the 2 OADs group, the most commonly used OADs were metformin (80.9%), sulfonylureas (74.4%), and thiazolidinediones (24.0%). In the Z3 OADs group, the most commonly used OADs were metformin (95.6%), sulfonylureas (93.1%), thiazolidinediones (71.3%), and dipeptidyl peptidase-4 inhibitors (46.3%).

January 2016

The mean age was 57 years, 47% were female, the mean A1C was 9.2%, the mean weight was 99.5 kg with a mean BMI of 34.5 kg/m2, and 3% of patients had experienced Z1 hypoglycemic event. Table I shows the baseline clinical and demographic data for the patients grouped by number of OADs at baseline. In these unweighted data, there were significant differences between the 2 OADs and Z3 OADs groups compared with the 1 OAD group for sex (P ¼ 0.008) and duration of diabetes at insulin initiation (P ¼ 0.002).

Health Care Resource Utilization Outcomes (Baseline) At baseline, patients in the 1 OAD group had significantly higher all-cause inpatient admissions

113

Clinical Therapeutics

Table I. Baseline characteristics (all data are unweighted). 1 OAD (n ¼ 450) Age, y, mean (SD) 56.7 (13.9) Female, no. (%) 234 (52.0) Weight, kg, mean (SD) 99.2 (26.3) 34.7 (8.9) BMI, kg/m2, mean (SD) A1C, %, mean (SD) 9.3 (2.3) Duration of diabetes at insulin initiation, no. (%) o1 years 118 (26.2) 1–2 years 7 (1.6) 3–4 years 3 (0.7) Z5 years 4 (0.9) Unknown 318 (70.7) OADs, no. (%) Sulfonylureas 155 (34.4) Metformin 207 (46.0) Thiazolidinediones 46 (10.2) α-Glucosidase inhibitors 1 (0.2) Meglitinides 9 (2.0) Dipeptidyl peptidase-4 inhibitors 32 (7.1) 1.9 (1.7) Deyo-CCI score,13 mean (SD) Comorbid conditions, no. (%) Retinopathy 11 (2.4) Nephropathy 0 Neuropathy 51 (11.3) Hypertension 205 (45.6) Dyslipidemia 137 (30.4) Depression 22 (4.9) Hypoglycemia, no. (%) Z1 Hypoglycemic event 12 (2.7) All-cause health care utilization, no. (%) Z1 Inpatient admission 95 (21.1) Z1 ED visit 125 (27.8) Z1 Endocrinologist visit 63 (14.0) Diabetes-related health care utilization, no. (%) Z1 Inpatient admission 57 (12.7) Z1 ED visit 27 (6.0)

2 OADs (n ¼ 738)

Z3 OADs (n ¼ 642)

P*

57.4 347 98.8 34.4 9.2

(12.1) (47.0) (23.2) (7.4) (2.0)

57.4 273 100.4 34.3 9.1

(11.4) (42.5) (22.6) (7.1) (1.8)

0.550 0.008 0.480 0.746 0.146

113 14 8 10 593

(15.3) (1.9) (1.1) (1.4) (80.4)

117 10 7 12 496

(18.2) (1.6) (1.1) (1.9) (77.3)

0.002

549 597 177 3 32 118 1.8

(74.4) (80.9) (24.0) (0.4) (4.3) (16.0) (1.7)

598 614 458 22 48 297 1.7

(93.1) (95.6) (71.3) (3.4) (7.5) (46.3) (1.5)

o0.001 o0.001 o0.001 o0.001 0.001 o0.001 0.051

35 6 62 332 223 39

(4.7) (0.8) (8.4) (45.0) (30.2) (5.3)

28 4 61 269 226 36

(4.4) (0.6) (9.5) (41.9) (35.2) (5.6)

0.131 0.173 0.247 0.391 0.100 0.873

25 (3.4)

18 (2.8)

0.728

133 (18.0) 155 (21.0) 89 (12.1)

89 (13.9) 119 (18.5) 106 (16.5)

0.006 0.001 0.060

93 (12.6) 45 (6.1)

65 (10.1) 38 (5.9)

0.284 0.990

A1C ¼ glycosylated hemoglobin A1c; BMI ¼ body mass index; CCI ¼ Charlson Comorbidity Index; ED ¼ emergency department; OAD ¼ oral antidiabetes drug. * A χ2 test was used for categorical variables and ANCOVA was used for continuous variables.

114

Volume 38 Number 1

P. Levin et al. (21.1%; P ¼ 0.006) and ED visits (27.8%; P ¼ 0.001) than those in the 2 and Z3 OADs groups (2 OADs: 18.0% and 21.0%, respectively; and Z3 OADs: 13.9% and 18.5%, respectively) (Table I). There were no significant differences between the groups for diabetesrelated health care utilization at baseline (Table I).

Health Care Costs (Annualized Baseline) Total all-cause health care costs differed between the groups at baseline, with higher costs for the group treated with 2 OADs than the Z3 OADs group ($23,998 vs $19,517, respectively; P ¼ 0.048), but there was no significant difference between the 1 OAD and 2 or Z3 OADs groups (Figure 2A). Non-office outpatient costs ($6088 vs $4019, respectively; P ¼ 0.008) and outpatient office costs ($1173 vs $1044, respectively; P ¼ 0.012) were higher for the 1 OAD group versus the Z3 OADs group. Prescription charges were higher for the Z3 OADs group ($5535) than for the 1 OAD group ($3363; P o 0.0001) and the 2 OADs group ($3859; P o 0.0001). There were no other significant differences between the groups. There were no differences in total diabetes-related health care costs between the groups at baseline (Figure 2B). However, prescription costs were higher at baseline for the Z3 OADs group ($2330) compared with the 1 OAD ($400) and 2 OADs ($909) groups (P o 0.0001 for both). Baseline outpatient office costs were also higher for the Z3 OADs group ($473) versus the 1 OAD ($395) and 2 OADs ($413) groups (P ¼ 0.0007 and P ¼ 0.003, respectively). There were no other differences between the groups.

Clinical Outcomes and Treatment Persistence (1-Year Follow-Up) Weighted results show that, at follow-up, compared with the 2 OADs and Z3 OADs groups, the 1 OAD group had a significantly greater reduction in A1C, a higher proportion of patients achieving A1C o7.0%, and fewer patients experiencing Z1 hypoglycemic event (Table II). Treatment persistence was significantly lower in the 1 OAD group than in the other 2 groups (51.8% vs 58.0% for the 2 OADs group [P ¼ 0.004] and 68.7% for the Z3 OADs group [P o 0.001]). There was no significant difference in the duration of persistence between the groups. Stratified by baseline BMI, the greatest reductions in A1C were among patients in the 1 OAD group

January 2016

when baseline BMI was r30 kg/m2 or 31 to 35 kg/m2 (P o 0.01) (Figure 3). Among patients who were severely obese at baseline (BMI 435 kg/m2), there were no significant differences in A1C outcome at follow-up between treatment groups. The weightadjusted daily average consumption of insulin per 100 lb of bodyweight did not differ in the BMI r30 kg/m2 strata (15, 14, and 15 U for the 1, 2, and Z3 OADs groups, respectively) or in the BMI 31 to 35 kg/m2 strata (12, 13, and 13 U, respectively); in the BMI 435 kg/m2 strata, it was significantly lower in the 1 OAD group (11 U) compared with the Z3 OADs group (14 U) (P ¼ 0.0036), but not the 2 OADs group (12 U). No significant differences were observed for change in weight (þ1.2 kg vs þ1.6 kg vs þ1.0 kg) or BMI (þ0.5 kg/m2 vs þ0.6 kg/m2 vs þ 0.4 kg/m2) between the 1 OAD, 2 OADs, and Z3 OADs groups, respectively.

Health Care Resource Utilization Outcomes (1-Year Follow-Up) Weighted results show that, during follow-up, a significantly higher proportion of patients in the 1 OAD group had all-cause inpatient admissions (26.1%) compared with the 2 OADs (20.0%; P o 0.001) and Z3 OADs (13.0%; P o 0.0001) groups. The proportion of patients with all-cause ED visits was similar across the 3 groups (26.2%, 27.7%, and 24.1%, respectively). A significantly lower proportion of patients in the 1 OAD group had diabetesrelated outpatient office visits (87.9%) compared with the 2 OADs group (94.1%) and the Z3 OADs group (93.4%) (P o 0.0001 for both). A significantly higher proportion of patients in the Z3 OADs group (95.5%) compared with patients in the 1 OAD group (91.1%; P ¼ 0.0001) and 2 OADs group (92.7%; P ¼ 0.0072) had Z1 diabetes supply claim.

Health Care Costs (1-Year Follow-Up) Weighted results show that during follow-up, there were no differences in medical costs between the OAD groups. Prescription costs were higher in the Z3 OADs group ($6754) compared with the 1 OAD ($4838) and 2 OADs ($5715) groups (P o 0.0001 and P ¼ 0.0072, respectively) (Figure 2A). At follow-up, the overall diabetes-related health care costs were similar, but with higher prescription costs for the Z3 OADs group ($3121) compared with the 1 OAD ($1759) and 2 OADs ($2090) groups

115

Clinical Therapeutics

A 25,000

Annualized All-Cause Health Care Costs ($)

24,854 23,998

20,000

21,060

21,167

20,133

19,517

ED Costs

Inpatient Costs

15,000

Outpatient Office Costs Non-Office Outpatient Costs Prescription Costs

10,000

*** **

5000 *

***

*

Follow-up

Baseline

0 Baseline

Follow-up

Baseline

Follow-up

≥ 3 OADs

2 OADs

1 OAD

*P < 0.001; **P < 0.01; and ***P < 0.02 vs 1 OAD group

B

9000

Annualized Diabetes-Related Health Care Costs ($)

8326 8272

8000

7503 6581

7000

6754

6257

Diabetes Supply Costs

6000

ED Costs Inpatient Costs

5000

Outpatient Office Costs

4000

**

3000

Non-Office Outpatient Costs Prescription Costs

***

2000

* *

1000

** *

0

Baseline

Follow-up

1 OAD

Baseline

Follow-up

2 OADs

Baseline

Follow-up

≥ 3 OADs

*P < 0.001; **P < 0.01; and ***P < 0.02 vs 1 OAD group

Figure 2. Total all-cause (A) and diabetes-related (B) health care costs per patient at baseline and at 1-year follow-up. Numbers represent total health care costs. Presented baseline data are annualized and unweighted, whereas presented follow-up data are weighted. No statistical comparisons were done on the trend in costs from baseline to follow-up. ED ¼ emergency department; OAD ¼ oral antidiabetes drug.

(P o 0.0001 for both) (Figure 2B). Outpatient office visit costs were also higher in the Z3 OADs group compared with the 1 OAD group ($483 vs $385; P ¼ 0.0015).

116

DISCUSSION In this real-world observational study, initiating basal insulin therapy in patients previously treated with 1 OAD resulted in greater A1C reductions and a lower

Volume 38 Number 1

P. Levin et al.

BMI 31-35 kg/m

BMI > 35 kg/m

Change From Baseline in A1c (%)

0.1210 0.0069 0.0143 0.0065 0.8728 o0.0001

0.6185 0.0020 0.0002 o0.0001

0.5 *

A1C ¼ glycosylated hemoglobin A1c; ED ¼ emergency department; OAD ¼ oral antidiabetes drug. * A χ2 test was used for categorical variables and ANCOVA was used for continuous variables. † A t test was used to compare between groups for continuous variables and a χ2 test was used for categorical variables. ‡ Weighting of the data means that the percentages do not calculate exactly.

o0.0001 0.0102 0.0518

0.0006 o0.0001

January 2016

*

–1.0

–1.5 *

*

–2.0

–2.5

–3.0

No. of patients with valid A1C 12 months 322 527 484 after insulin initiation Change in A1C from baseline, %, mean (SD)
1.7 (5.1)
1.0 (3.0)
0.9 (3.2) o0.0001 124 (38.2) 141 (26.7) 96 (19.6) o0.0001 Patients achieving A1C o7.0%, no. (%)‡ Hypoglycemia, no. (%)‡ Total hypoglycemia 12 (2.7) 49 (6.6) 32 (5.0) 0.0002 Outpatient office visit hypoglycemia 6 (1.3) 22 (3.0) 8 (1.3) 0.005 Inpatient/ED hypoglycemia 4 (0.9) 14 (1.9) 23 (3.7) 0.0001

2 OADs (n ¼ 738) 1 OAD (n ¼ 450)

Table II. Clinical outcomes at 1-year follow-up (all data are weighted).

Z3 OADs (n ¼ 642)

Global test, P*

1 OAD vs 2 1 OAD vs Z3 2 OADs vs Z3 OADs, P† OADs, P† OADs, P†

BMI ≤ 30 kg/m

0.0

1 OAD

2 OADs

≥3 OADs

* P < 0.01 vs 1 OAD group

Figure 3. Change in glycosylated hemoglobin A1c from baseline to 1-year follow-up stratified by baseline BMI. A1c ¼ glycosylated hemoglobin A1c; BMI ¼ body mass index; OAD ¼ oral antidiabetes drug.

risk of hypoglycemia than adding insulin to treatment with Z2 OADs. The greater use of sulfonylureas and thiazolidinediones in the 2 and Z3 OADs groups indicates that, in this study, those who did not intensify treatment with insulin, intensified it with sulfonylureas and thiazolidinediones as the disease progressed. Risks of hypoglycemia and weight gain are the main disadvantages associated with the use of all insulins,1 and concerns about these issues are one of the key barriers to insulin initiation.10 Of note, therefore, were the comparatively lower rates of hypoglycemia among patients previously treated with 1 OAD than with Z2 OADs. Lower hypoglycemia in the 1 OAD group after insulin initiation could be related to the shorter duration of diabetes in this group, resulting in patients less prone to hypoglycemia22 or possibly with less hypoglycemia unawareness because of better glucose counterregulation and less autonomic neuropathy.23 However, baseline neuropathy and nephropathy were not significantly different between the groups, suggesting that this was not the case in this population. Alternatively, the 2 and Z3 OADs groups had a greater sulfonylurea use at baseline. This may partly explain the higher incidence of hypoglycemia because sulfonylurea use has been linked with hypoglycemia.1,24 The difficulty in titrating sulfonylurea

117

Clinical Therapeutics dose may explain lower hypoglycemia in the 1 OAD group (where sulfonylurea use is less common) as providers can respond to the threat of hypoglycemia by reducing insulin by a few units, whereas sulfonylurea dosing does not have such flexibility because of fewer dose choices. After initiation of insulin treatment, patients in the 1 OAD group were more likely to achieve A1C o7.0% with no increase in hypoglycemic events. Conversely, in the 2 and Z3 OADs groups, a lower proportion of patients achieved A1C o7.0%, yet hypoglycemic events nearly doubled. This suggests that early insulin initiation could reduce unnecessary initiation of sulfonylureas and the higher resultant rate of hypoglycemic events, as seen in those patients on multiple OAD regimens who then add insulin. The clinical benefits associated with early initiation of insulin were achieved despite the 1 OAD group having the lowest treatment persistence rate. The 1year persistence rates (ranging from 52% to 69%) compare favorably with other analyses of basal insulin initiation after OAD treatment in real-world US settings, which range between 44% and 80%.25–29 According to ADA and EASD guidelines, patientcentered approaches to T2DM treatment should consider age, weight, and comorbidities.1 As  80% of patients with T2DM are overweight or obese, key outcomes were stratified by baseline BMI. In severely obese patients (BMI 435 kg/m2), efficacy was not improved by the early addition of insulin. Those with 1 OAD had lower daily average consumption than those in the Z3 OADs group. Although greater insulin resistance among more obese patients may be the cause of this finding, obese and leaner patients have been shown to have similar islet dysfunction.30 More research is required to clarify the underlying reason for this finding.31–33 Together with our results, these data highlight the need for more personalized approaches to the treatment of patients with T2DM, incorporating factors such as weight and their effect on treatment outcomes. Although we compared early or late initiation of insulin (using the number of OADs as a proxy for the timing of initiation relative to disease progression), a recent 10-year systematic review comparing insulin initiation with continuation of OAD therapy found earlier initiation of insulin therapy was associated with greater reductions in A1C levels from baseline.5 The authors proposed a “paradigm shift” in clinical

118

practice toward early initiation of insulin in patients with T2DM. Indeed, the feasibility of initiating insulin and OAD therapy at diagnosis of T2DM was assessed in a small-scale study (n ¼ 58).34 Patients treated with first-line insulin-based therapy showed no greater weight gain or hypoglycemia and no decreased compliance, treatment satisfaction, or quality of life compared with patients treated with conventional 3 OAD therapy. Some studies have explored the economic benefits of insulin initiation in patients with T2DM25,26, 29,32,35–38 ; however, to our knowledge there are no studies directly addressing the economic outcomes of early versus late initiation of insulin. In this study, we show that, at baseline, there was a trend toward the 1 OAD and 2 OADs groups having higher costs compared with the Z3 OADs group, except for prescription costs, which were higher for the Z3 OADs group. After adding insulin to the treatment, there was no overall difference between the groups except for higher prescription costs and outpatient visit costs for the Z3 OADs group. Although a significantly greater proportion of patients in the 1 OAD group had an all-cause hospital admission compared with the 2 and Z3 OADs groups, there was no significant difference between the associated inpatient and the overall all-cause health care costs. Therefore, our data suggest that early insulin initiation has no impact on overall costs. A recently published study sought to investigate health care expenditure stratified by timing of insulin initiation in older adults in a real-world setting.39 This study, which also showed that early insulin initiation was associated with significantly greater reduction in A1C (P o 0.001), suggested no significant differences in hypoglycemia events or health care costs associated with insulin initiation time. This study did not differentiate types of basal insulin or control for the time from diagnosis to when insulin was initiated or consider covariates such as obesity. Here, we demonstrate that the degree of benefit associated with initiating insulin varied according to individual patient characteristics, including BMI and previous OAD regimen as well as time of initiation. A patientcentered approach to treatment, as advocated in the joint ADA/EASD guidelines,1 is supported by these data. As this study is based on real-world data from large national US claims databases combined with EMR

Volume 38 Number 1

P. Levin et al. data, it has richer information than EMR or claims data alone. However, as this was a retrospective, observational study, the data may be subject to selection bias and confounding and do not establish causality of the treatment on observed outcomes. Also, as the analyses were conducted using health care claims data generated mainly for billing purposes, they were potentially subject to coding errors. The time between baseline and final measures may not have been equal for all patients due to the time windows used to define baseline A1C and final A1C. Persistence with therapy was estimated through pharmacy claims data, which reflected prescriptions filled that were not necessarily taken or consumed. The analyses were based on data from a commercially and Medicare-insured US population and limited to those who had merged claims and EMR data, so they may not be representative or generalizable to other patients. Timing of insulin initiation and the clinical rationale behind the therapy choices may not be accurately reflected in the study because information on the duration of diabetes was available only in a subset of patients and the number of OADs served as a proxy for the timing of insulin initiation relative to disease progression. This study included patients who initiated basal insulin between July 2004 and December 2010. In more recent years, however, the pattern of antidiabetes drug use has changed, with less use of sulfonylureas and thiazolidinediones and a greater use of newer therapies.40 As such, this study may not be reflective of patients receiving newer therapies.

CONCLUSIONS Earlier initiation of basal insulin therapy in patients with T2DM previously treated with 1 OAD resulted in better clinical outcomes than adding insulin to OAD treatment later in the course of therapy, with no increase in overall costs. In this setting, the intensification of treatment with insulin as opposed to sulfonylureas and thiazolidinediones could represent a clinical advantage for patients. The early insulin use group has the lowest utilization of sulfonylureas and lowest incidence of hypoglycemia at follow-up. Non-severely obese patients appear to particularly benefit from earlier initiation of basal insulin therapy.

ACKNOWLEDGMENTS This study was funded by Sanofi US, Inc. The authors received writing/editorial support in the preparation

January 2016

of this manuscript from Nicola Truss, PhD, of Excerpta Medica, funded by Sanofi US, Inc. P. Levin, S. Zhou, W. Wei and J. Gill were involved in the study concept and design. E. Durden and A. M. Farr conducted the data collection. P. Levin, S. Zhou, E. Durden, A. M. Farr, W. Wei and J. Gill were all involved in the data analysis and interpretation. S. Zhou and W. Wei participated in drafting the manuscript and P. Levin, E. Durden, A. M. Farr, and J. Gill critically revised the manuscript at each stage. P. Levin, S. Zhou, E. Durden, A. M. Farr, W. Wei and J. Gill all provided final approval of this manuscript.

CONFLICTS OF INTEREST P. Levin reports working on the advisory panel for Sanofi US, Inc. and Novo Nordisk; working as consultant for Novo Nordisk, Eli Lilly and Company, and Sanofi US, Inc.; receiving research support from Eli Lilly and Company, Sanofi US, Inc., Novo Nordisk, Amylin Pharmaceuticals, and Boehringer Ingelheim Pharmaceuticals; and participating in speakers bureaus for Eli Lilly and Company, Novo Nordisk, Amylin Pharmaceuticals, and Boehringer Ingelheim Pharmaceuticals. S. Zhou, J. Gill, and W. Wei, are employees of Sanofi US, Inc. E. Durden and A. M. Farr are employees of Truven Health Analytics, which receives funding from Sanofi US, Inc.

REFERENCES 1. Inzucchi SE, Bergenstal RM, Buse JB, et al. Management of hyperglycemia in type 2 diabetes, 2015: a patientcentered approach: update to the position statement of the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care. 2015;38:140–149. 2. American Diabetes Association. Standards of medical care in diabetes–2015. Diabetes Care. 2015;38(Suppl 1): S5–S87. 3. Ohkubo Y, Kishikawa H, Araki E, et al. Intensive insulin therapy prevents the progression of diabetic microvascular complications in Japanese patients with non-insulindependent diabetes mellitus: a randomized prospective 6-year study. Diabetes Res Clin Pract. 1995;28:103–117. 4. UK Prospective Diabetes Study (UKPDS) Group. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). Lancet. 1998;352:837–853. 5. Asche CV, Bode B, Busk AK, Nair SR. The economic and clinical benefits of adequate insulin initiation and

119

Clinical Therapeutics

6.

7.

8.

9.

10.

11.

12.

13.

intensification in people with type 2 diabetes mellitus. Diabetes Obes Metab. 2012;14:47–57. Chaudhry SI, Berlowitz DR, Concato J. Do age and comorbidity affect intensity of pharmacological therapy for poorly controlled diabetes mellitus? J Am Geriatr Soc. 2005;53:1214–1216. Fu AZ, Qiu Y, Davies MJ, et al. Treatment intensification in patients with type 2 diabetes who failed metformin monotherapy. Diabetes Obes Metab. 2011;13:765–769. Brown JB, Nichols GA, Perry A. The burden of treatment failure in type 2 diabetes. Diabetes Care. 2004;27:1535–1540. Grant RW, Cagliero E, Dubey AK, et al. Clinical inertia in the management of type 2 diabetes metabolic risk factors. Diabet Med. 2004; 21:150–155. Peyrot M, Rubin RR, Lauritzen T, et al. Resistance to insulin therapy among patients and providers: results of the cross-national Diabetes Attitudes, Wishes, and Needs (DAWN) study. Diabetes Care. 2005; 28:2673–2679. Centers for Disease Control and Prevention. Classification of Diseases, Functioning, and Disability. International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). http:// www.cdc.gov/nchs/icd/icd9cm. htm. Accessed October 31, 2014. Nathan DM, Buse JB, Davidson MB, et al. Management of hyperglycemia in type 2 diabetes: A consensus algorithm for the initiation and adjustment of therapy: a consensus statement from the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care. 2006;29:1963–1972. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45:613–619.

120

14. United States Department of Labor. Bureau of Labor Statistics. Chained Consumer Price Index for all urban consumers (C-CPI-U) 1999–2008, Medical Care. Series ID: SUUR0000SAM. http://data.bls.gov/cgi-bin/ surveymost?su. Accessed October 31, 2014. 15. Zhao Y, Campbell CR, Fonseca V, Shi L. Impact of hypoglycemia associated with antihyperglycemic medications on vascular risks in veterans with type 2 diabetes. Diabetes Care. 2012;35:1126–1132. 16. Ginde AA, Blanc PG, Lieberman RM, Camargo CA Jr. Validation of ICD-9-CM coding algorithm for improved identification of hypoglycemia visits. BMC Endocr Disord. 2008;8:4. 17. Wei W, Pan C, Xie L, Baser O. Real-world insulin treatment persistence among patients with type 2 diabetes. Endocr Pract. 2014;20: 52–61. 18. Curtis LH, Hammill BG, Eisenstein EL, et al. Using inverse probabilityweighted estimators in comparative effectiveness analyses with observational databases. Med Care. 2007;45(10 Suppl 2):S103–S107. 19. Kurth T, Walker AM, Glynn RJ, et al. Results of multivariable logistic regression, propensity matching, propensity adjustment, and propensitybased weighting under conditions of nonuniform effect. Am J Epidemiol. 2006;163:262–270. 20. Austin PC. The relative ability of different propensity score methods to balance measured covariates between treated and untreated subjects in observational studies. Med Decis Making. 2009;29:661– 677. 21. Hirano K, Imbens GW. Estimation of causal effects using propensity score weighting: an application to data on right heart catheterization. Health Serv Outcomes Res Methodol. 2001;2:259–278. 22. Dailey GE, Gao L, Aurand L, Garg SK. Impact of diabetes duration

23.

24.

25.

26.

27.

28.

29.

30.

on hypoglycaemia in patients with type 2 diabetes treated with insulin glargine or NPH insulin. Diabetes Obes Metab. 2013;15: 1085–1092. Seaquist ER, Anderson J, Childs B, et al. Hypoglycemia and diabetes: a report of a workgroup of the American Diabetes Association and The Endocrine Society. Diabetes Care. 2013;36:1384–1395. Hirst JA, Farmer AJ, Dyar A, et al. Estimating the effect of sulfonylurea on HbA1c in diabetes: a systematic review and metaanalysis. Diabetologia. 2013;56: 973–984. Baser O, Wei W, Baser E, Xie L. Clinical and economic outcomes in patients with type 2 diabetes initiating insulin glargine disposable pen versus exenatide BID. J Med Econ. 2011;14:673–680. Xie L, Wei W, Pan C, et al. A realworld study of patients with type 2 diabetes initiating basal insulins via disposable pens. Adv Ther. 2011; 28:1000–1011. Baser O, Tangirala K, Wei W, Xie L. Real-world outcomes of initiating insulin glargine-based treatment versus premixed analog insulins among US patients with type 2 diabetes failing oral antidiabetic drugs. Clinicoecon Outcomes Res. 2013;5:497–505. Davis KL, Tangirala M, Meyers JL, Wei W. Real-world comparative outcomes of US type 2 diabetes patients initiating analog basal insulin therapy. Curr Med Res Opin. 2013;29:1083–1091. Wang L, Wei W, Miao R, et al. Real-world outcomes of US employees with type 2 diabetes mellitus treated with insulin glargine or neutral protamine Hagedorn insulin: a comparative retrospective database study. BMJ Open. 2013; 3:e002348. Ferrannini E, Gastaldelli A, Miyazaki Y, et al. Beta-cell function in subjects spanning the range from

Volume 38 Number 1

P. Levin et al.

31.

32.

33.

34.

35.

36.

37.

38.

normal glucose tolerance to overt diabetes: a new analysis. J Clin Endocrinol Metab. 2005;90:493– 500. Moberly SP, Mather KJ, Berwick ZC, et al. Impaired cardiometabolic responses to glucagon-like peptide 1 in obesity and type 2 diabetes mellitus. Basic Res Cardiol. 2013;108:365. Thayer S, Wei W, Buysman E, et al. The INITIATOR study: pilot data on real-world clinical and economic outcomes in US patients with type 2 diabetes initiating injectable therapy. Adv Ther. 2013; 30:1128–1140. Curtis B, Lage MJ. Glycemic control among patients with type 2 diabetes who initiate basal insulin: a retrospective cohort study. J Med Econ. 2014;17:21–31. Lingvay I, Legendre JL, Kaloyanova PF, et al. Insulin-based versus triple oral therapy for newly diagnosed type 2 diabetes: which is better? Diabetes Care. 2009;32:1789–1795. Fabunmi R, Nielsen LL, Quimbo R, et al. Patient characteristics, drug adherence patterns, and hypoglycemia costs for patients with type 2 diabetes mellitus newly initiated on exenatide or insulin glargine. Curr Med Res Opin. 2009;25:777– 786. Lee LJ, Yu AP, Johnson SJ, et al. Direct costs associated with initiating NPH insulin versus glargine in patients with type 2 diabetes: a retrospective database analysis. Diabetes Res Clin Pract. 2010;87: 108–116. Grabner M, Chu J, Raparla S, et al. Clinical and economic outcomes among patients with diabetes mellitus initiating insulin glargine pen versus vial. Postgrad Med. 2013;125: 204–213. Xie L, Zhou S, Wei W, et al. Does pen help? A real-world outcomes study of switching from vial to disposable pen among insulin glargine-treated patients with type

January 2016

2 diabetes mellitus. Diabetes Technol Ther. 2013;15:230–236. 39. Bhattacharya R, Zhou S, Wei W, et al. A real-world study of the effect of timing of insulin initiation on outcomes in older Medicare beneficiaries with type 2 diabetes

mellitus. J Am Geriatr Soc. 2015;63: 893–901. 40. Turner LW, Nartey D, Stafford RS, et al. Ambulatory treatment of type 2 diabetes in the U.S., 19972012. Diabetes Care. 2014;37: 985–992.

Address correspondence to: Dr. Philip Levin, MODEL Clinical Research, 6535 North Charles St., Suite 400, Towson, MD 21204. E-mail: pal3420@ yahoo.com

121