Managed Care

 

Hemoglobin A1c Outcomes and Health Care Resource Use In Type 2 Diabetes Mellitus Patients Treated With Combination Oral Antidiabetic Drugs Through Step Therapy And Loose-Dose and Fixed-Dose Combinations

MANAGED CARE July 2012. © MediMedia USA
Peer-Reviewed

Hemoglobin A1c Outcomes and Health Care Resource Use In Type 2 Diabetes Mellitus Patients Treated With Combination Oral Antidiabetic Drugs Through Step Therapy And Loose-Dose and Fixed-Dose Combinations

This study verifies, as defined by current guidelines, that initial treatment with fixed dose combinations (FDC) is associated with a better likelihood of HbA1c goal attainment and lower health care resource use and costs.

Setareh A. Williams, PhD
Director, economics and outcomes research, AstraZeneca LP
Erin K. Buysman, MS
Associate director, OptumInsight
Erin M. Hulbert, MS
Senior research analyst in health economics and outcomes research, OptumInsight
Joette Gdovin Bergeson, PhD
Associate director of health economics and outcomes research, Bristol-Myers Squibb Co.
Bin Zhang, MD
Associate director of health economics and outcomes research, Bristol Myers-Squibb Co.
John Graham, PharmD
Executive director of health services, Bristol Myers-Squibb Co.

Setareh A. Williams, PhD, director, economics and outcomes research, AstraZeneca LP; Erin K. Buysman, MS, associate director, and Erin M. Hulbert, MS, senior research analyst in health economics and outcomes research, OptumInsight; Joette Gdovin Bergeson, PhD, associate director of health economics and outcomes research, Bristol-Myers Squibb Co.; Bin Zhang, MD, associate director of health economics and outcomes research, Bristol Myers-Squibb Co.; John Graham, PharmD, executive director of health services, Bristol Myers-Squibb Co.

Abstract

Purpose: To compare outcomes of type 2 diabetes mellitus (T2DM) patients initiating therapy with FDC vs. those with loose-dose combination (LDC) or step therapy (ST) in a managed care population.

Design: A retrospective claims database analysis.

Methodology: Treatment-naïve T2DM patients who were continuously enrolled in a health plan during 2006–2009 were studied. Eligible patients were assigned to FDC, LDC, or ST cohorts. Glycated hemoglobin goal attainment (HbA1c <7%) was assessed using the American Diabetes Association (ADA) treatment guidelines. Health care resources use and costs, including inpatient, emergency room (ER), and ambulatory visits, were measured during the 12 months after therapy initiation. All-cause and diabetes-related use and costs were assessed.

Principal findings: 21,048 patients met study criteria (FDC n=8,416, ST n=8,407, LDC n=4,225), and 1,926 of these patients had HbA1c results. FDC patients had lower rates of post-index all-cause inpatient stays and ER visits compared with the other cohorts. FDC patients had lower average counts of diabetes-related ambulatory visits (2.7) compared with ST (3.7; p<0.001) and LDC (3.2; p<0.001) and significantly lower average post-index all-cause and diabetes-related costs compared with the other cohorts, with average all-cause costs for FDC, ST, and LDC of $8,445, $10,515, and $9,688, respectively, and diabetes-related costs of $1,641, $2,099, and $1,900, respectively. FDC patients had higher rates of achieving HbA1c goal (61%) compared to ST (48%; p<0.001) or LDC (52%; p=0.015). Differences in outcomes remained following multivariate analyses.

Conclusion: Treatment with FDC was associated with lower health care resources use and costs and better likelihood of HbA1c goal attainment.

Introduction

As of 2010, approximately 23 million people in the United States are diagnosed with T2DM (ADA 2010). In the United States, 12.0 million men and 11.5 million women aged 20 years or older have diabetes. According to the American Diabetes Association (ADA), 1.6 million new cases of diabetes are diagnosed annually. Diabetes was the seventh leading cause of death in the United States in 2006 (ADA 2010). The estimated total cost of diabetes in the United States was $174 billion in 2007, making diabetes a tremendous burden on the nation’s health care system (ADA 2010). The goal of diabetes treatment is to maintain blood glucose at normal levels to reduce the risk of complications (UKPDS 1998). The majority of patients presenting with T2DM are overweight or obese at the time of diagnosis and are unable to achieve or sustain normal glycemic levels without pharmacological treatment. Treatment options for T2DM include lifestyle modifications, weight loss, insulin therapy, glucagon-like peptide-1 analogs, and oral antidiabetic agents (OADs). Classes of OADs include sulfonylureas, meglitinides, biguanides, thiazolidinediones, alpha-glucosidase inhibitors (AGIs), and dipeptidyl peptidase-4 (DPP-IV) inhibitors.

In 2009 and 2011, the ADA published updated guidelines for the Standard of Medical Care in Diabetes, recommending initial treatment with metformin and lifestyle interventions. If metformin alone does not reduce HbA1c levels to less than 7%, another glucose-lowering agent, which may include insulin, may be added. The majority of individuals require combination therapy to effectively control their diabetes (ADA 2009/2011). Because monotherapy is unlikely to be successful in patients with HbA1c levels of 7.6–9.0%, the American Association of Clinical Endocrinologist/American College of Endocrinology (AACE/ACE) states that patients with higher baseline HbA1c levels should be initiated on dual therapy to reach an HbA1C target level of 6.5% (ADA 2010).

Real-world evidence evaluating the outcomes associated with earlier initiation of dual therapy in T2DM is limited. Previous studies have focused on the impact of FDC and LDC on adherence. Previous studies have shown improved adherence in patients treated with FDC therapy vs. patients treated with loose dose combination of the same agents, regardless of whether patients initiated with FDC or switched to FDC from LDC of the same treatments.

In one study, 7,570 FDC users had 1.8% higher medication possession ratios (MPR) in the post-index period compared with 14,762 patients on LDC (Cheong 2008). In another study, within a managed care organization, previously treated patients receiving monotherapy with an oral antidiabetic medication who required additional therapy exhibited significantly greater adherence when they were switched to FDC therapy compared with LDC therapy. Patients receiving LDC therapy who were switched to FDC therapy exhibited significantly greater adherence after the switch (Melikiun 2002).

The impact of earlier initiation of dual combination therapy in T2DM and the benefits of FDC on health outcomes independent of adherence are not known. The purpose of this study was to compare health outcomes among patients who initiated with FDC vs. LDC and ST in a population naïve to diabetes treatment.

Methods

Study design

This was a retrospective database analysis of health care claims data of patients enrolled in a large, geographically diverse, managed health care plan in the United States. The claims database contained eligibility information, pharmacy claims, and medical claims for approximately 18 million patients enrolled in the plan during the study period. Personal identifiers were removed before data analyses, and data were accessed in a manner that was compliant with the HIPAA Privacy Rule of 1996. Institutional Review Board approval was not needed.

Data sources

Medical claims, pharmacy claims, and laboratory test results were identified for commercially insured health plan members with prescription fills for oral antidiabetic (OAD) agents between Jan. 1, 2007 through Oct. 31, 2008. The date of the first qualifying OAD prescription fill was defined as the index date. The 6-month period prior to the index date were the pre-index period and the 12-month period following the index date was the post-index period.

Subject identification

Inclusion criteria included: age ≥18 as of the index year; ≥2 filled prescriptions for an OAD during the post-index period (including the index fill); no filled prescriptions for OADs during the pre-index period; no filled prescriptions for non-OAD medications (GLP-1 analogs, pramlintide, or insulin) during the pre-index period or on the index date; evidence of medical claims for T2DM (ICD-9-CM diagnosis codes 250.x0 or 250.x2) during the pre-index or post-index periods; and continuous enrollment in a commercial health plan with pharmacy and medical benefits during the pre-index and post-index periods. Additional inclusion criteria for the analysis of HbA1c outcomes included: ≥1 HbA1c laboratory value during (index date: 182 days) through index date; and ≥1 HbA1c laboratory value during (index date + 91 days) through (index date + 365 days).

Patients were assigned to study cohorts based on the OADs filled during the post-index period. Patients with fills for triple therapy (either FDC + third agent or 3 agents in loose-dose form) or with fills for GLP-1 or insulin on the index date were excluded, as were patients with evidence of gestational diabetes (ICD-9-CM diagnosis codes 648.8x or HCPCS S9214) during the pre-index or post-index periods.

Cohorts

The FDC cohort included patients with 2 or more prescription fills for an FDC therapy. Examples of FDC therapy include Metaglip (metformin plus glipizide) or Actoplus MET (metformin plus pioglitizide), which are available in one pill. The ST cohort included patients with prescription fills for one of the (non-FDC) medications on the index date who augmented therapy with a second OAD or GLP-1 analog and had at least two distinct 15-day overlaps of the 2 medications included in the ST regimen. GLP-1 was the only injectable allowed for ST inclusion.

The LDC cohort (LDC) included patients with fills for 2 non-FDC OAD classes on the index date who had at least two distinct 15-day overlaps in the 2 medications included in the LDC therapy regimen. Examples of ST or LDC therapy include metformin with Amaryl (glimipiride) or metformin with Onglyza (saxagliptin), which are taken as 2 separate pills. It should be noted that medications that are available in fixed-dose form may also be taken as LDC or ST therapy if the 2 medications are taken as 2 separate pills, whereas FDC therapy is available for limited number of combinations. Patients with evidence of monotherapy throughout the post-index period and FDC patients with evidence of previous OAD prescription fills were initially identified and considered for the analyses. The monotherapy group was excluded from the final study sample because they had better HbA1c control at index, appeared to be less progressed in their diabetes, and were more likely to reach goal during follow-up. Non-naïve FDC patients were excluded from the study because the use of OADs in the baseline period resulted in better HbA1c control at index. Therefore, the non-naïve patients were not comparable to patients in other treatment-naïve cohorts.

Study measures
Outcomes

The HbA1c outcomes were evaluated only for the subset of patients with HbA1c values available in the laboratory data. HbA1c outcomes were measured from index date + 91 days through the end of the 12-month follow-up period (index date + 365 days). For the subset of patients whose HbA1c levels were above 7% at index, an indicator variable was created to identify whether the patient achieved HbA1c goal at any time during the follow-up period (from index date + 91 days through index date + 365 days). For patients with multiple HbA1c laboratory values in the post-index period, any laboratory value that met the defined goal was sufficient to qualify as “goal attainment.” The 2011 ADA Clinical Practice Recommendations (ADA 2010) set a goal of 7%. The time from the index date to HbA1c goal attainment was measured. For patients who did not attain goal, this variable was set to 365 days.

All health care resources usage and cost outcomes were measured in the overall sample in the post-index period, inclusive of the index date. All-cause and diabetes-related health care resource usage included ambulatory visits (office and outpatient), ER visits, and inpatient admissions. The number of ambulatory visits was also calculated for each patient. Usage was defined as attributable to diabetes if the claim had a primary diagnosis code for T2DM (ICD-9-CM 250.x0 or 250.x2). All-cause and diabetes-related health care costs were computed as the combined health plan and patient paid amounts in the post-index period. Costs were calculated as total, medical, and pharmacy costs. Costs were adjusted using the annual medical care component of the consumer price index (CPI) to reflect inflation between 2006 and 2009 (CPI 2009). Costs were defined as attributable to diabetes if the claim had a primary diagnosis code of T2DM or was a pharmacy claim for OADs, GLP-1 analogs, pramlintide, or insulin.

Demographic characteristics

Age was defined as of the index year, and gender was determined based on information from enrollment files. Patients were assigned to either the 18–64 or ≥65 age group. The geographic region in which the study subject was enrolled in a health plan was identified as Northeast, Midwest, South, or West.

Clinical characteristics

The index diabetes treatment regimen was identified as sulfonylureas (SU), metformin (MET), thiazolidinediones (TZD), DPP-IV inhibitors (DPP-IV), β-glucosidase inhibitors (AGI), meglitinide derivatives, and/or GLP-1 (GLP-1 for ST only). For patients in the FDC and LDC cohorts, both OADs filled as part of the fixed- or loose-dose combination on the index date were identified; for patients in the ST cohort, both agents filled as part of the ST regimen were identified. The Charlson-Quan comorbidity score was calculated for each patient based on the presence of diagnosis codes on medical claims in the pre-index period (Quan 2005). General comorbid conditions were defined using the Clinical Classification software managed by the Agency for Healthcare Research and Quality (2009). This measure generates indicator variables for specific disease conditions based on ICD-9-CM diagnoses. Comorbidities of interest as confounders included hypertension, disorders of the lipid metabolism, cancers and neoplasms, and conditions affecting mental health status. A count of non-diabetes-related medications filled in the pre-index period was calculated. Flags for in-patient stays and ER visits and a count of ambulatory visits, both all-cause, and diabetes related, were calculated in the pre-index period. CPI-adjusted, pre-index, all-cause and diabetes-related health care costs were computed as the combined health plan and patient paid amounts in the pre-index period. An MPR was calculated based on the MPR for each OAD filled during the post-index period, weighted by the length of time that the patient was supposed to be taking the medication. In addition, pre-index HbA1c values during (index date: 182) days through the index date were determined in the subset of patients with available HbA1c results.

Analysis

Data extraction and statistical analyses were performed using SAS version 9.2 (SAS Institute Inc., Cary, N.C.) and Stata version 10.0 (StataCorp, College Station, Texas). All patient characteristics and study outcomes were compared descriptively between the FDC cohort and the ST and LDC cohorts using t-tests or chi-squared tests. Outcomes analyzed using multivariate methods included: HbA1c goal attainment, post-index all-cause inpatient/ER visits, post-index count of diabetes-related ambulatory visits, post-index all-cause total costs, and post-index diabetes-related total costs. HbA1c goal attainment and occurrence of all-cause post-index inpatient or ER visit were modeled using logistic regression models; counts of diabetes-related ambulatory visits were modeled using a negative binomial regression with a log link; and cost measures were modeled using generalized linear models with a gamma distribution and a log link, which avoids potential difficulties introduced by transformation and retransformation of the dependent variable (Manning 1998).

For ease of interpretation, the average predicted number of diabetes-related ambulatory visits and all-cause and diabetes-related costs were calculated for each cohort. In addition to cohort, each model included age, gender, health plan geographic region, pre-index Charlson-Quan comorbidity score, flags for hypertension, disorders of lipid metabolism, cancers and neoplasms, conditions affecting mental health status, pre-index count of medications, pre-index total all-cause costs (in 1,000s of dollars), pre-index inpatient stays, pre-index ER visits, pre-index count of diabetes-related ambulatory visits, and weighted MPR. In addition, the pre-index HbA1c value was included in the model of HbA1c goal attainment, but was not included in the other models for which laboratory data were not available for all patients.

Results

Sample identification

Figure 1 outlines the sample selection process. Nearly 17.8 million patients were enrolled in this commercial health plan between Jan. 1, 2007 and Oct. 31, 2008. Of these, 580,681 patients had evidence of OAD use, and nearly all of these patients were 18 or older; 78,359 patients met criteria to be included in the FDC, ST, or LDC cohorts — 23,024 of these patients (29%) met the continuous enrollment criteria and another 1,976 patients were excluded because they did not meet T2DM criteria or they had gestational diabetes. This resulted in a final sample of 21,048 patients for the analysis of cost and health care usage outcomes (8,416 [40%] FDC, 8,407 [40%] ST, and 4,225 [20%] LDC). Of these, pre- and post-index HbA1c results were available for 1,926 patients. Patients with evidence of OAD monotherapy (44,940) and patients not naïve to OADs who switched to FDC therapy (10,690) were also considered for the study, but it was determined, through the course of the analysis, that these patients were not comparable to the other groups because they had lower baseline HbA1c values and were thought to have less severe diabetes compared with patients in the other cohorts. Therefore, our analysis focused on the remaining 3 cohorts.

Figure 1. Sample selection and attrition

Sample selection and attrition

Table 1 presents the demographics of the final study sample. FDC patients were significantly younger (mean age: 53.3 years) than LDC or ST patients (ST: 55.8; LDC: 55.0; p<0.001 in all cohorts). FDC patients had a higher proportion of males (59%) compared with ST (54%; p<0.001), but a similar proportion compared to LDC patients (58%; p=0.498). The health plan region distribution was significantly different among the cohorts, although a majority of patients in all cohorts were from the South. FDC (1.13) and LDC patients (1.15) had similar average Charlson-Quan comorbidity scores (p=0.451), whereas ST (1.21) patients had a significantly higher score (p<0.001).

Table 1 presents the clinical characteristics of the final study sample. Metformin was the most commonly prescribed OAD in the index regimen for all cohorts but was more common in FDC (95%) patients, compared with ST (86%) and LDC (87%) patients. This is probably due to the fact that metformin is included as part of nearly all available FDC products, whereas LDC and ST can combine any two agents, regardless of their availability in FDC form. Sulfonylureas were prescribed more frequently as part of ST or LDC therapy compared with FDC, whereas TZDs and DPP-IVs were more often prescribed as part of an FDC regimen. AGIs and meglitinide derivatives were not available in an FDC and were rarely prescribed in the other cohorts. GLP-1 analogs were only considered as part of ST and were filled by 7% of patients. FDC patients filled, on average, 3.4 non-diabetes medications during the pre-index period, while ST patients filled 3.2 such medications (p=0.001) and LDC patients filled 2.3 (p<0.001).

Table 1. Patient demographics and clinical characteristics
1. FDC treatment (N=8,416) 2. Step therapy (N=8,407) 3. LDC (N=4,225) Cohort 2 vs. cohort 1 p-value Cohort 3 vs. cohort 1 p-value
Mean age (SD) 53.34 (10.86) 55.82 (11.72) 54.98 (11.39) <0.001 <0.001
Male gender N (%) 4,971 (59) 4,559 (54) 2,469 (58) <0.001 0.498
Geographic region N (%)
Northeast 526 (6) 467 (6) 202 (5) 0.056 <0.001
Midwest 1,137 (14) 1,515 (18) 709 (17) <0.001 <0.001
South 6,201 (74) 5,630 (67) 2,901 (69) <0.001 <0.001
West 552 (7) 795 (9) 413 (10) <0.001 <0.001
Mean Charlson-Quan Comorbidity Score (SD) 1.13 (1.03) 1.21 (1.20) 1.15 (1.13) <0.001 0.451
Index treatment regimen n (%)
Metformin 7,967 (95) 7,202 (86) 3,656 (87) <0.001 <0.001
Sulfonylurea 3,660 (43) 5,308 (63) 3,114 (74) <0.001 <0.001
Thiazolidinediones 3,607 (43) 2,257 (27) 1,263 (30) <0.001 <0.001
DPP-IV inhibitors 1,598 (19) 1,195 (14) 320 (8) <0.001 <0.001
β-glucosidase inhibitors 0 (0) 39 (0) 15 (0) <0.001 <0.001
Meglitinide derivatives 0 (0) 207 (2) 82 (2) <0.001 <0.001
GLP-1 0 (0) 606 (7) 0 (0) <0.001
Mean count of pre-index non-diabetes medications (SD) 3.41 (3.65) 3.22 (3.91) 2.28 (3.11) 0.001 <0.001
Mean count of pre-index non-diabetes medication dispensing (SD) 6.99 (9.65) 6.71 (10.41) 4.45 (7.91) 0.076 <0.001
Pre-index medications: Medications, not including antidiabetic medications, filled in the pre-index period; Pre-index medication dispensing: Dispensing for all medications, not including antidiabetic medications, filled in the pre-index period; GLP-1: GLP-1 was the only injectable allowed for step therapy inclusion

Table 2 presents the pre-index usage by the final study sample. Patients with FDC regimens had significantly lower rates of all-cause inpatient or ER visits compared with patients in the other cohorts. Diabetes-related inpatient visits were low in general, with approximately 0.8% of FDC patients, 1.4% of ST (p<0.001), and 3.6% of LDC (p<0.001) patients having inpatient stays. In addition, FDC patients had a significantly lower rate of diabetes-related ER visits compared with LDC patients. The count of pre-index diabetes-related ambulatory visits was significantly different between FDC and the other cohorts, although the actual mean number of visits was similar: FDC (1.0); ST (0.9); and LDC (0.8).

Table 2. Pre-index health care usage and costs
1. FDC treatment (N=8,416) 2. Step therapy (N=8,407) 3. LDC (N=4,225) Cohort 2 vs. cohort 1 p-value Cohort 3 vs. cohort 1 p-value
All-cause usage
Inpatient stays n (%) 393 (5) 585 (7) 413 (10) <0.001 <0.001
ER visits n (%) 1,173 (14) 1,338 (16) 716 (17) <0.001 <0.001
Mean ambulatory visit count (SD) 6.05 (7.00) 7.10 (8.12) 5.62 (6.88) <0.001 <0.001
Diabetes-related usage
Inpatient stays n (%) 71 (1) 119 (1) 154 (4) <0.001 <0.001
ER visits n (%) 64 (1) 73 (1) 63 (1) 0.436 <0.001
Mean ambulatory visit count (SD) 1.01 (1.29) 0.92 (1.29) 0.84 (1.17) <0.001 <0.001
All-cause costs
Mean total costs (SD) 2,804 (8,131) 3,556 (14,532) 3,633 (12,455) <0.001 <0.001
Mean medical costs (SD) 2,314 (7,881) 3,094 (14,323) 3,321 (12,138) <0.001 <0.001
Mean pharmacy costs (SD) 491 (1,199) 463 (1,207) 312 (1,167) 0.130 <0.001
Diabetes-related costs
Mean total costs (SD) 269 (3,121) 418 (7,919) 794 (7,321) 0.109 <0.001
Mean medical costs (SD) 269 (3,121) 418 (7,919) 794 (7,321) 0.109 <0.001
Mean pharmacy costs (SD) 0 (0) 0 (0) 0 (0)

Table 2 also presents the pre-index total costs of the final study sample. Pre-index all-cause total costs were $2,804, $3,556, and $3,633 for FDC, ST, and LDC patients, respectively (p<0.001 between FDC vs. ST and LDC). Average diabetes-related pre-index costs were similar between FDC and ST patients ($269 vs. $418, p=0.109), but were significantly lower in FDC patients compared with LDC patients ($269 vs. $794, p<0.001).

The last pre-index HbA1c result of the final study sample with HbA1c results (N=1,926) were similar with average results of 8.74, 8.68, and 8.88 among FDC, ST, and LDC patients, respectively.

Outcomes

Table 3 presents the post-index usage and cost outcomes of the final study sample. Patients in the FDC cohort had lower rates of all-cause inpatient or ER visits compared with patients in the other cohorts. Diabetes-related inpatient and ER visits were low in general with 1.7%, 2.7%, and 2.6% of FDC, ST, and LDC patients having diabetes-related inpatient visits and 1.4%, 2%, and 1.7% of FDC, ST, and LDC patients having diabetes-related ER visits. ST and LDC patients had a significantly higher rate of diabetes-related inpatient stays, and ST patients had a significantly higher rate and count of diabetes-related ER visits compared with FDC. The mean count [SD] of diabetes-related ambulatory visits was higher for ST (3.7 [2.99]; p<0.001) and LDC (3.2 [3.02]; p<0.001) compared with FDC (2.7, [2.93]).

Table 3. Post-index usage and costs
1. FDC treatment (N=8,416) 2. Step therapy (N=8,407) 3. LDC (N=4,225) Cohort 2 vs. cohort 1 p-value Cohort 3 vs. cohort 1 p-value
All-cause usage
Inpatient stays n (%) 757 (9) 1,065 (13) 491 (12) <0.001 <0.001
ER visits n (%) 1,839 (22) 2,207 (26) 1,037 (25) <0.001 <0.001
Mean ambulatory visit count (SD) 13.27 (13.72) 17.34 (15.45) 14.53 (14.48) <0.001 <0.001
Diabetes-related usage
Inpatient stays n (%) 141 (2) 230 (3) 111 (3) <0.001 <0.001
ER visits n (%) 117 (1) 167 (2) 72 (2) 0.003 0.170
Mean ambulatory visit count (SD) 2.73 (2.93) 3.68 (2.99) 3.16 (3.02) <0.001 <0.001
All-cause costs
Mean total costs (SD) 8,445 (18,211) 10,515 (22,096) 9,688 (24,494) <0.001 0.004
Mean medical costs (SD) 5,708 (17,488) 7,293 (20,894) 6,879 (23,333) <0.001 0.004
Mean pharmacy costs (SD) 2,737 (2,898) 3,222 (3,525) 2,810 (3,601) <0.001 0.253
Diabetes-related costs
Mean total costs (SD) 1,641 (5,737) 2,099 (8,287) 1,900 (5,288) <0.001 0.012
Mean medical costs (SD) 663 (5,658) 1,087 (8,218) 936 (5,115) <0.001 0.006
Mean pharmacy costs (SD) 979(856) 1,013 (1,010) 964 (1,292) 0.018 0.506

After adjustment for confounders in a logistic regression model, ST and LDC patients had significantly higher odds of post-index all-cause inpatient or ER visits compared with FDC patients: ST, odds ratio(OR), 1.37; confidence interval (CI), 1.27–1.47; LDC, OR 1.26; CI, 1.16–1.38. Similarly, in a negative binomial model of post-index count of diabetes-related ambulatory visits, the adjusted count of visits was significantly lower among FDC patients (2.8) compared with ST (3.6; p<0.001) and LDC patients (3.2; p<0.001). Post-index all-cause total costs were $8,445, $10,515, and $9,688 for FDC, ST, and LDC patients, and costs were significantly lower in FDC patients compared to ST (p<0.001) and LDC (p=0.004). Average diabetes-related post-index costs were significantly higher in ST ($2,099; p<0.001) and LDC ($1,900; p=0.012) patients compared to FDC ($1,641). After controlling for confounders, mean adjusted all-cause post-index costs were significantly lower among FDC patients ($8,745) compared to ST ($10,352; p<0.001) and LDC patients ($10,294; p<0.001). Similarly, mean adjusted diabetes-related costs were lower in FDC patients ($1,713) compared to ST ($2,005; p=0.001) and FDC patients ($1,925; p=0.050).

Pre- and post-index HbA1c outcomes in the final study sample, with HbA1c results in both the pre-index and post-index periods are shown in Table 4. On average, every cohort had a drop in HbA1c following the index date, with average changes in HbA1c of –1.65, –1.36, and –1.51 for FDC, ST, and LDC patients, respectively. The decrease in FDC patients was significantly larger than the HbA1c decrease observed in ST patients (p=0.013). The rate of goal attainment was significantly higher in FDC patients compared to ST and LDC patients (Table 5). Estimated odds ratios, 95% CI, and p-values from a logistic regression model of post-index HbA1c goal attainment also are provided in Table 5. After controlling for confounders, patients in the ST cohort had 56% lower odds of achieving HbA1c goal (OR: 0.44; CI: 0.34–0.57) and patients in the LDC cohort had 27% lower odds (non-significant) of achieving HbA1c goal (OR: 0.73; CI: 0.53–1.00) in the post-index period compared with those in the FDC cohort.

Table 4. Pre- and post-index HbA1c results
1. FDC treatment (N=873) 2. Step therapy (N=712) 3. LDC (N=341) Cohort 2 vs. cohort 1 p-value Cohort 3 vs. cohort 1 p-value
Mean pre-index HbA1c result (SD) 8.74 (2.36) 8.68 (2.24) 8.88 (2.53) 0.593 0.364
Mean post-index HbA1c result (SD) 7.10 (1.60) 7.32 (1.55) 7.38 (1.70) 0.005 0.008
Mean change in HbA1c from pre-index to post-index (SD) -1.65 (2.33) -1.36 (2.23) -1.51 (2.66) 0.013 0.401
Pre-index HbA1c result ≥ 7.0%; n (%) 615 (70) 520 (73) 241 (71) 0.256 0.938
Post-index HbA1c goal attainment (< 7.0%); n (%) 375 (61) 247 (48) 125 (52) <0.001 0.015
Mean time to post-index HbA1c goal attainment (SD) 261.86 (103.40) 289.98 (95.80) 273.41 (103.98) <0.001 0.142

 

Table 5. Logistic regression model of HbA1c goal attainment
Covariate Odds ratio 95% CI p-value
Step therapy vs. FDC therapy 0.44 (0.34,0.57) <0.001
LDC therapy vs. FDC therapy 0.73 (0.53,1.00) 0.053
Age 1.00 (0.99,1.02) 0.596
Male gender 1.11 (0.88,1.41) 0.386
U.S. census region – Northeast vs. West 0.50 (0.27,0.94) 0.031
U.S. census region – Midwest vs. West 0.52 (0.28,0.96) 0.037
U.S. census region – South vs. West 0.51 (0.32,0.79) 0.003
Pre-index Charlson-Quan Comorbidity Score 0.82 (0.72,0.93) 0.003
Comorbidity – hypertension 0.67 (0.52,0.86) 0.002
Comorbidity – disorders of lipid metabolism 0.83 (0.65,1.07) 0.152
Comorbidity – cancers and neoplasms 1.45 (0.99,2.11) 0.055
Comorbidity – conditions affecting mental health status 1.01 (0.70,1.46) 0.967
Pre-index medication count 1.10 (1.06,1.14) <0.001
Pre-index total cost (in $1000s) 1.05 (1.01,1.09) 0.023
Pre-index Inpatient stay 0.72 (0.30,1.75) 0.474
Pre-index ER visit 0.78 (0.54,1.13) 0.183
Pre-index count of ambulatory visits 0.98 (0.96,1.00) 0.090
Weighted MPR (medication possession ratio) 6.13 (3.68,10.20) <0.001
Last pre-index HbA1c value 0.91 (0.86,0.96) 0.001

 

Discussion

This analysis of treatment-naïve T2DM patients demonstrated that FDC treatment was associated with better likelihood of HbA1c goal attainment than was LDC or ST. In brief, FDC patients had lower rates of post-index all-cause inpatient stays and ER visits compared with other cohorts and a lower average count of all-cause and diabetes-related ambulatory visits compared with ST and LDC. This cohort also had a lower rate of post-index diabetes-related inpatient stays compared with the other cohorts. FDC patients also had lower average post-index all-cause and diabetes-related total costs compared with ST and LDC patients. FDC patients also had a larger average drop in HbA1c and a higher rate of achieving a <7% HbA1c goal. HbA1c goal attainment sensitivity analysis was conducted using ≤6.5% AACE/ACE guidelines, with similar results (Rodbard 2009). Other studies have shown benefits of intensive glucose control using FDC therapy in patients with T2DM (Chalmers 2006, 2007). In this study, a larger difference was observed for FDC therapy vs. ST, which suggests that there are benefits to earlier initiation of dual therapy per AACE/ACE guidance. Many physicians fail to implement the recommended guidelines regarding frequent follow-up monitoring and dosing titration in the T2DM population (Maclean 2009). Physicians may be slow in advancing therapy, relative to either dosages of medications or switching to a more efficacious therapeutic regimen in a timely manner (Maclean 2009).

One of the most important suggestions of the AACE/ACE guidelines is the recommendation to monitor therapy every 2 to 3 months and to augment therapy until the goal for HbA1c has been achieved (Rodbard 2009). Studies have shown the benefits of glucose monitoring, improving glucose control, and reducing the rate of severe hypoglycemic episodes (Leinung 2010). Dual therapy should begin once a patient naïve to treatment has been identified as having an HbA1c level in the range of 7.6% to 9% because no single medication is likely to bring about a goal of <6.5% (Rodbard 2009). This approach may be more beneficial because it leads to a greater likelihood of successful outcomes.

Study limitations

These findings must be considered within the context of the study limitations. There are several inherent limitations in the use of claims data. First, diagnostic codes may be incorrectly coded or included as rule-out criteria rather than actual disease. However, in this study, the strict diagnostic and medication use criteria used make it unlikely that anyone included did not have T2DM. An additional limitation of the administrative claims data source is the requirement of continuous enrollment in the health plan to ensure complete medication and pharmacy claims data are available for all patients included in the study. The follow-up period was restricted to one year to allow for an adequate sample of patients in the final study population. Laboratory results data were available only for a subset of the overall sample, and results may be incomplete in patients who do have laboratory information. The HbA1c results available were values obtained from laboratory tests done by physicians or lab technicians during patient visits, and results obtained by patients at home or during a visit that was not processed by a laboratory for which data are submitted were not available in the laboratory results data. Therefore, the HbA1c outcomes analysis may not be a complete representation of HbA1c outcomes in patients using OADs.

Although claims data are an excellent source for understanding real-world treatment, data on medication use beyond the patterns of fills observed in the claims are scarce. There is limited information available on the reasons physicians choose to prescribe one type of medication over another in a given patient, or on reasons for discontinuing or switching medications. Several factors may have driven physicians to prescribe a FDC or an LDC medication in a given patient, such as physician or patient preference, patient history, ease of use, costs (e.g, generic vs. branded products, one copay for FDC vs. two copayments for LDC), side effect profiles, perceived or real efficacy or safety concerns, or other factors. It is unknown how much each of these factors may have affected a physician’s propensity to prescribe one type of medication over another.

The data used for this study come from a commercial managed care population and, therefore, the generalizability of the results of this study is limited. Results of this analysis are applicable primarily to the patterns of OAD use in T2DM patients in stable managed care settings. The plans used for analysis, however, are discounted fee-for-service, IPA-network plans rather than capitated or gatekeeper models. These plans also include a wide geographic distribution across the United States, providing the capability for generalization to managed care populations on a national level. Despite the standard limitations, however, administrative claims data continue to be powerful data sources. These data allow for examination of health care usage and expenditure patterns in a real-world setting, away from the controlled environment of clinical trials, and offer the advantage of large sample sizes with diverse medical histories.

Conclusion

Patients initiating OAD therapy with FDC agents may have a better likelihood for HbA1c goal attainment compared with patients initiating OADs with ST or LDC therapy. Patients initiating FDC therapy may also have lower post-index health care usage and costs (all-cause and diabetes-related) compared with patients treated with ST or LDC therapy, which may be driven by a lower number of ambulatory visits and fewer needs for medication adjustments.

In this study, a larger HbA1c improvement was observed for FDC therapy vs. ST, suggesting that there are benefits to earlier initiation of dual therapy per AACE guidance. This finding is also important for health plan decision makers when they develop and implement their preferred drug lists (PDLs), given the potential clinical and economic benefits associated with FDC therapy. In addition, some plans may have prior-authorization rules or other restrictions on the use of FDC therapy that require the failure of monotherapy or LDC therapy before being allowed to begin FDC therapy, although this study found a benefit to starting FDC medication as first-line therapy. Additional research is warranted to evaluate longer-term outcomes of FDC vs. LDC and ST and to assess the impact of specific combination products (especially newer FDC therapies including DPP-IVs) on outcomes.

References

American Diabetes Association. http://www.diabetes.org/diabetes-basics/diabetes-statistics/. Accessed December 6, 2010.

American Diabetes Association, 2009. Obtained from: http://care.diabetesjournals.org/content/vol32/Supplement_1/. Accessed February 3, 2010. http://care.diabetesjournals.org/content/34/Supplement_1. Accessed February 15, 2011. American Diabetes Association. Executive Summary: Standards of Medical Care on Diabetes — 2010. Diabetes Care 33 Supplement 1; 2010. pp S4–S10.

Chalmers J, Kengne AP, Joshi R, Perkovic V, Patel A. New insights from ADVANCE. J Hypertens Suppl. 2007;25:S23–30.

Chalmers J, Perkovic V, Joshi R, Patel A. ADVANCE: breaking new ground in type 2 diabetes. J Hypertens Suppl. 2006;24:S22–8.

Cheong, C. Patient Adherence and Reimbursement Amount for Antidiabetic Fixed-Dose Combination Products Compared with Dual Therapy Among Texas Medicaid Recipients. Clinical Therapeutics. 2008;30(10)1893–1907.

Comorbidity software 2009. Available at: http://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp.

Leinung M, Thompson S, Nardacci E. Benefits of continuous glucose monitor use in clinical practice. Endocr Pract. 2010 May–Jun;16(3):371–5.

Maclean JR, Chapman RH, Ferrufino CP, Krishnarajah G. Drug titration patterns and HbA 1c levels in type 2 diabetes. Int J Clin Pract. 2009 Jul;63(7):1008–16.

Manning WG. The logged dependent variable, heteroscedasticity, and the retransformation problem. J Health Econ. 1998;17(3):283–95.

Melikiun, C. Adherence to Oral Antidiabetic Therapy in a Managed Care Organization: A Comparison of Monotherapy, Combination Therapy, and Fixed-Dose Combination Therapy. Clinical Therapeutics. 2002;24(3), 460–67.

Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi JC, Saunders LD, Beck CA, Feasby TE, Ghali WA, Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Medical Care. 2005;43:1130–39.

Rodbard HW, Jellinger PS,, Davidson JA, et al. Statement by an American Association of Clinical Endocrinologists/American College of Endocrinology Consensus Panel on Type 2 Diabetes Mellitus: An Algorithm for Glycemic Control. Endocr Pract. 2009;15:540–59.

UK prospective diabetes study group (UKPDS). 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;352:837–53.

U.S. Department of Labor, Bureau of Labor Statistics. Consumer Price Index. Chained Consumer Price Index for all urban consumers (C-CPI-U) 1999–2009, Medical Care. Series ID: SUUR0000SAM. Washington, D.C.: U.S. Dept. of Labor, Bureau of Labor Statistics, 2009. http://data.bls.gov/cgi-bin/surveymost?su

Disclosure

The authors report that they have no financial, commercial, or industrial relationships to disclose regarding this article. Funding was provided by AstraZeneca and Bristol-Myers Squibb.

Author correspondence

Setareh A. Williams, PhD
Director, health economics and outcomes research
AstraZeneca LP
1800 Concord Pike
Wilmington, DE 19850
E-mail: setareh.williams@astrazeneca.com

Erin K. Buysman, MS
Associate director, health economics and outcomes research
OptumInsight
12125 Technology Drive
Eden Prairie, MN 55344
E-mail:erin.buysman@innovus.com

Erin M. Hulbert, MS
Senior research analyst, health economics and outcomes research
OptumInsight
12125 Technology Drive
Eden Prairie, MN 55344
e-mail: erin.hulbert@innovus.com

Joette Gdovin Bergeson, PhD
Associate director, health economics and outcomes research
Bristol-Myers Squibb Co.
Tampa, FL
E-mail: joette.gdovin@bms.com

Bin Zhang, MD
Associate director, health economics and outcomes research
Bristol Myers-Squibb
777 Scudders Mill Road
Plainsboro, NJ 08536
E-mail: bin.zhang2@bms.com

John Graham, PharmD
Executive director, health services
Bristol Myers-Squibb
777 Scudders Mill Road
Plainsboro, NJ 08536
E-mail: john.graham@bms.com

Meetings

Private Health Insurance Exchanges Conference Washington, D.C. October 7–8, 2014
National Healthcare Facility Management Summit Palm Beach, FL October 16–17, 2014
National Healthcare CFO Summit Las Vegas, NV October 19–21, 2014
National Healthcare CXO Summit Las Vegas, NV October 19–21, 2014
Innovative Member Engagement Operations For Health Plans Las Vegas, NV October 20–21, 2014
4th Partnering With ACOs Summit Los Angeles, CA October 27–28, 2014
2014 Annual HEDIS® and Star Ratings Symposium Nashville, TN November 3–4, 2014
PCMH & Shared Savings ACO Leadership Summit Nashville, TN November 3–4, 2014
World Orphan Drug Congress Europe 2014 Brussels, Belgium November 12–14, 2014
Medicare Risk Adjustment, Revenue Management, & Star Ratings Fort Lauderdale, FL November 12–14, 2014
Healthcare Chief Medical Officer Forum Alexandria, VA November 13–14, 2014
Home Care Leadership Summit Atlanta, GA November 17–18, 2014