This study compared the clinical and economic benefits associated with dual-goal achievement, glycated hemoglobin (HbA1c) <7% (53 mmol/mol) and LDL cholesterol (LDL-C) <100 mg/dL, with achievement of only the LDL-C goal or only the HbA1c goal in veterans with type 2 diabetes mellitus (T2DM).
This retrospective cohort analysis evaluated electronic medical records (Veterans Integrated Service Network 16) in adult T2DM patients with two or more measurements of LDL-C and HbA1c between 1 January 2004 and 30 June 2010 (N = 75,646). Cox proportional hazards models were used to compare microvascular and cardiovascular outcomes by goal achievement status; generalized linear regression models were used to assess diabetes-related resource utilization (hospitalization days and number of outpatient visits) and medical service costs.
Relative to achievement of only the LDL-C goal, dual-goal achievement was associated with lower risk of microvascular complications (adjusted hazard ratio [aHR] 0.79), acute coronary syndrome (0.88), percutaneous coronary intervention (0.78), and coronary artery bypass graft (CABG) (0.74); it was also associated with fewer hospitalization days (adjusted incidence rate ratio [aIRR] 0.93) and outpatient visits (0.88), as well as lower diabetes-related annual medical costs (−$130.89). Compared with achievement of only the HbA1c goal, dual-goal achievement was associated with lower risk of the composite cardiovascular-related end point (aHR 0.87) and CABG (aHR 0.62), as well as fewer outpatient visits (aIRR 0.98).
Achieving both HbA1c and LDL-C goals in diabetes care is associated with additional clinical and economic benefits, as compared with the achievement of either goal alone.
The American Diabetes Association recommends that patients with type 2 diabetes mellitus (T2DM) maintain levels of glycated hemoglobin (HbA1c) <7% (53 mmol/mol) and LDL cholesterol (LDL-C) <100 mg/dL (1). The level of HbA1c, an indicator of average glycemia over several months, is a strong predictor of diabetes-related complications (1). Intensive treatments aimed at decreasing HbA1c levels have been associated with a reduced risk of microvascular complications (e.g., nephropathy and retinopathy) in patients with T2DM (1–4). The beneficial effects of decreasing HbA1c levels on the rate of cardiovascular events is less evident, with recent data showing that cardiovascular benefits of tight glycemic control are predominantly observed in patients who are newly diagnosed with diabetes and have minimal comorbidities (5). Conversely, treatments aimed at lowering LDL-C levels in patients with diabetes, especially in individuals with a high baseline cardiovascular risk, have been associated with significantly lower rates of cardiovascular events, including death (6–10). In addition, studies suggest that in diabetes, a condition estimated to incur direct annual costs (2007) of $116 billion in the U.S. (11), treatments aimed at glycemic (12–15) and LDL-C control (16–18) are generally considered cost-effective.
Cardiovascular disease is the most frequent underlying cause of death in diabetic patients (19). Although the cardiovascular benefits of LDL-C control in patients with diabetes have been well established (1), the benefits of achieving the HbA1c goal in addition to achieving the LDL-C goal are not clear. In diabetes, multifactorial interventions (i.e., targeting several risk factors simultaneously, including HbA1c, LDL-C, blood pressure, and lifestyle) have been associated with significant reductions in microvascular and cardiovascular morbidity and mortality when compared with conventional treatments in several studies, including the Steno-2 Study (20,21), and appear to be cost-effective (22,23). However, existing research has not assessed the additional benefits associated with appropriate control of both HbA1c and LDL-C levels versus the control of only one.
Since the cardiovascular benefits associated with HbA1c goal achievement in addition to LDL-C goal achievement are not clear, the primary objective of our study was to assess the clinical and economic benefits associated with the achievement of both HbA1c and LDL-C goals compared with achievement of only the LDL-C goal. Additional objectives included comparisons of clinical and economic outcomes in dual-goal achievers versus HbA1c-only achievers, dual-goal achievers versus no-goal achievers, and each group of single-goal achievers versus no-goal achievers.
RESEARCH DESIGN AND METHODS
Data source
This retrospective observational study was conducted using electronic medical records from the South Central Veterans Affairs Health Care Network (VISN 16), one of the largest of the 23 VISNs in the Veterans Health Administration (VHA). The VHA is a national integrated health care system providing a set of comprehensive services to veterans. As of 2010, ∼23 million veterans were living in the U.S., a large majority (90.6%) of whom were male. About one-third of veterans were enrolled in the VHA.
The VISN 16 data warehouse is an integrated, de-identified, individual-level database representing ∼7.8% of U.S. veterans and covers a geographic region of ∼170,000 square miles, including the states of Arkansas, Louisiana, Mississippi, and Oklahoma, and parts of Alabama, Florida, Missouri, and Texas. It includes records for >445,000 veterans from 10 medical centers and 40 outpatient clinics, with information regarding demographics, vital signs, laboratory results, diagnoses, procedures, inpatient and outpatient services (e.g., admission date, length of stay, and emergency room visits), drug prescriptions, and database enrollment history. As in the national VHA population, patients in VISN 16 are predominantly male (90.1%). All data comply with the Health Insurance Portability and Accountability Act. The study protocol was approved by the institutional review board and research and development committee of the Southeast Louisiana Veterans Health Care Systems.
Sample selection
Adult patients (≥18 years of age) were included in the study if they had two or more diagnoses of T2DM between 1 January 2004 and 30 June 2010. Patients who had more than one diagnosis of T1DM were excluded. Patients had at least one measurement of HbA1c and LDL-C within 30 days of each other (paired measurements) after the first diabetes diagnosis; the earlier date of the HbA1c or LDL-C measurement was defined as the index date. All patients were further required to have at least one more measurement of HbA1c and LDL-C within 1 year after the index date, irrespective of the gap between the measurements. The final sample included patients who were enrolled in the database for at least 12 months after the index date.
Data preparation
Longitudinal data were analyzed by 6-month cycles, starting from the index date. Average HbA1c and LDL-C levels were estimated for each cycle using the area under the curve method (24,25). For each cycle, these estimated averages were used to stratify patients into one of four goal achievement categories: dual goal (average HbA1c <7% [53 mmol/mol] and average LDL-C <100 mg/dL), HbA1c only (average HbA1c <7% [53 mmol/mol] and average LDL-C ≥100 mg/dL), LDL-C only (average LDL-C <100 mg/dL and average HbA1c ≥7% [53 mmol/mol]), or no goal (average HbA1c ≥7% [53 mmol/mol] and average LDL-C ≥100 mg/dL).
Patient characteristics
Patient characteristics as of the first cycle were summarized for the overall population, as well as stratified according to goal achievement status. Demographic information included age on index date, sex, race, BMI, and year of index date. The history of diabetes-related complications (microvascular, macrovascular, and other), comorbidities, and surgical procedures was identified as of the first cycle using ICD-9, Clinical Modification (ICD-9-CM) codes. Medication use during the first cycle was categorized by drug therapeutic class; health care resource utilization during the first cycle was categorized by hospitalization days and outpatient visits.
Characteristics were compared across the four groups according to goal achievement status using the ANOVA method for continuous variables and χ2 tests for categorical variables.
Clinical outcomes
Clinical outcomes were selected a priori and comprised 1) a composite cardiovascular-related end point (cerebrovascular disease [stroke], acute myocardial infarction, or cardiovascular death [defined by a diagnosis of coronary artery disease or cerebrovascular disease on the day of death]), 2) a composite end point for microvascular complications (diabetic retinopathy, nephropathy, or neuropathy), 3) acute coronary syndromes (ACS; acute myocardial infarction or unstable angina), and 4) cardiovascular procedures (percutaneous coronary intervention [PCI] or coronary artery bypass graft [CABG]).
For each clinical outcome, goal achievement and patient characteristics were measured in a given cycle and outcomes were assessed for the following cycle. The time to the first clinical event was evaluated using a Cox proportional hazards model, with goal achievement status as a time-dependent variable, controlling for patient demographics and other potential confounding factors such as cumulative comorbidity history, resource utilization, and medication use. All clinical outcomes were measured from the start of the second cycle until the first event, death, or end of data; for the analyses of specific clinical outcomes, patients were excluded from the analysis if any clinical event defining the particular outcome occurred before the end of the first cycle. Results are reported as adjusted hazard ratios with 95% CI.
Economic outcomes
The numbers of diabetes-related hospitalization days and outpatient visits were estimated for each 6-month cycle. Diabetes-related medical service costs were measured in each cycle using the average cost method (26). Costs were adjusted for inflation to 2011 U.S. dollars according to the medical care services component of the Consumer Price Index (27,28). Utilization and costs were considered diabetes related if they were associated with diagnoses of any of the following: diabetes, macrovascular complications, or microvascular complications.
The associations between goal achievement status in a given study cycle and utilization in the following cycle were assessed using generalized linear regression models (GLMs) with a Poisson distribution; results are reported as adjusted incidence rate ratios with 95% CIs. The associations between goal achievement status in a given cycle and costs in the following cycle were assessed using GLMs with a γ distribution, and adjusted results are reported as annualized incremental cost differences. All longitudinal GLMs accounted for within-patient correlation using a generalized estimating equation approach and controlled for demographics and time-dependent variables such as cumulative comorbidity history, resource utilization, and medication use. SAS software version 9.2 was used to conduct statistical analyses, and a two-tailed α level of 0.05 was used to determine statistical significance.
RESULTS
Patient characteristics
Of the 149,613 patients with at least two recorded diagnoses of T2DM between 1 January 2004 and 30 June 2010, a total of 75,646 patients met the selection criteria and were included in the analysis.
As shown in Table 1, as of the index date, most patients were older than 55 years (84.1%; mean age 64.7 years) and had an average BMI of 31.6 kg/m2. Almost all patients were men (97.4%), and approximately two in three were white (67.4%). During the first cycle, 35.1% of patients achieved both goals (dual-goal achievers), whereas 21.6% achieved only the LDL-C goal (LDL-C achievers), 24.6% achieved only the HbA1c goal (HbA1c achievers), and 18.6% did not achieve either goal (no-goal achievers) (Table 1). Compared with all other groups, dual-goal achievers were older (67.1 vs. 61.4–64.4 years). Rates of microvascular complications were lower for dual-goal (24.1%) and HbA1c achievers (22.4%) than for LDL-C (33.0%) and no-goal achievers (30.2%); similar differences were observed for the usage of insulin (10.4 and 7.5% vs. 34.6 and 30.1%, respectively) and oral antidiabetic drugs (67.4 and 64.0% vs. 82.2 and 82.7%, respectively). A higher rate of macrovascular complications was observed among dual-goal (47.4%) and LDL-C achievers (45.6%) than among HbA1c (33.8%) and no-goal achievers (33.9%) (Table 1).
Clinical outcomes
The median duration of follow-up time was 4.5 years from the index date. After adjusting for demographics, diabetes-related complications, comorbidities, surgical procedures, diabetic medication use, and health care utilization, dual-goal achievement, when compared with achievement of LDL-C goal alone, was associated with a significantly reduced risk of microvascular complications (adjusted hazard ratio 0.79 [95% CI 0.76–0.82]), ACS (0.88 [0.81–0.96]), PCI (0.78 [0.67–0.90]), and CABG (0.74 [0.60–0.92]), but not of the composite cardiovascular-related end point (1.00 [0.94–1.06]) (Fig. 1). Compared with HbA1c achievers, dual-goal achievement was associated with a lower risk of CABG (0.62 [0.49–0.79]) and the composite cardiovascular end point (0.87 [0.81–0.93]).
Goal achievement and clinical outcomes. Dual, patients achieving both LDL-C and HbA1c goals; HbA1c, patients achieving only the HbA1c goal; LDL-C, patients achieving only the LDL-C goal; none, patients achieving neither goal (see text for details).
Goal achievement and clinical outcomes. Dual, patients achieving both LDL-C and HbA1c goals; HbA1c, patients achieving only the HbA1c goal; LDL-C, patients achieving only the LDL-C goal; none, patients achieving neither goal (see text for details).
Compared with no-goal achievers, dual-goal achievers had a significant 19–30% lower hazard of the cardiovascular end point or diabetes-related microvascular or ACS-related events, and a 45–49% lower hazard of undergoing PCI or CABG. In addition, relative to no-goal achievers, both groups of single-goal achievers had significant reductions in hazard for all categories of complications and surgical procedures, except microvascular complications among LDL-C achievers.
Resource utilization
After controlling for demographics, diabetes-related complications, comorbidities, surgical procedures, diabetic medication use, and health care utilization, dual-goal achievers generally used less health care resources compared with single- or no-goal achievers (Fig. 2). In particular, compared with LDL-C achievers, dual-goal achievers had significantly fewer hospitalization days (adjusted incidence rate ratio 0.93 [95% CI 0.87–1.00]) and outpatient visits (0.88 [0.87–0.89]). Compared with HbA1c achievers, dual-goal achievers had significantly fewer outpatient visits (0.98 [0.97–1.00]), but there was no statistical difference in the number of hospitalization days (0.98 [0.89–1.07]). Compared with no-goal achievers, dual-goal achievers had significantly fewer hospitalization days (0.81 [0.72–0.90]) and outpatient visits (0.86 [0.85–0.87]). Similarly, both groups of single-goal achievers also had significantly fewer hospitalization days (LDL-C achievers, 0.87 [0.78–0.96]; HbA1c achievers, 0.83 [0.73–0.94]) and outpatient visits (LDL-C achievers, 0.98 [0.96–0.99]; HbA1c achievers, 0.87 [0.86–0.89]) than patients who did not achieve either goal.
Goal achievement and health care utilization. Dual, patients achieving both LDL-C and HbA1c goals; HbA1c, patients achieving only the HbA1c goal; LDL-C, patients achieving only the LDL-C goal; none, patients achieving neither goal. *P < 0.05; **P < 0.01; ***P < 0.001.
Goal achievement and health care utilization. Dual, patients achieving both LDL-C and HbA1c goals; HbA1c, patients achieving only the HbA1c goal; LDL-C, patients achieving only the LDL-C goal; none, patients achieving neither goal. *P < 0.05; **P < 0.01; ***P < 0.001.
Medical service costs
After adjusting for demographics, diabetes-related complications, comorbidities, surgical procedures, diabetic medication use, and health care utilization, dual-goal achievers incurred significantly lower annualized diabetes-related medical service costs compared with those who achieved only the LDL-C goal (−$130.89; P = 0.016), but no statistically significant difference was observed between dual-goal achievers and HbA1c goal achievers (−$56.17; P = 0.404) (Fig. 3). Diabetes-related medical service costs were significantly lower for dual-goal (−$376.50; P < 0.001), LDL-C (−$245.61; P < 0.001), and HbA1c achievers (−$320.32; P < 0.001) compared with those who did not achieve either goal.
Goal achievement and diabetes-related medical service cost difference. Dual, patients achieving both LDL-C and HbA1c goals; HbA1c, patients achieving only the HbA1c goal; LDL-C, patients achieving only the LDL-C goal; none, patients achieving neither goal. *P < 0.05; ***P < 0.001.
Goal achievement and diabetes-related medical service cost difference. Dual, patients achieving both LDL-C and HbA1c goals; HbA1c, patients achieving only the HbA1c goal; LDL-C, patients achieving only the LDL-C goal; none, patients achieving neither goal. *P < 0.05; ***P < 0.001.
CONCLUSIONS
The study results showed that, compared with the achievement of only the LDL-C goal, achievement of both HbA1c and LDL-C goals is associated with a lower risk of microvascular complications, ACS, and cardiovascular surgeries (PCI or CABG), lower utilization of health care resources, and lower costs of care, but no additional effect was observed for the composite cardiovascular-related end point. In addition, dual-goal achievement relative to HbA1c goal achievement is associated with a lower risk of the composite cardiovascular-related end point, CABG, and outpatient visits.
Clinical outcomes: dual- vs. single-goal achievement
To our knowledge, this is the first study that was designed to quantify the differences in clinical and economic outcomes between dual-goal and single-goal achievement in patients with T2DM. The results from our study are generally consistent with findings from previous studies designed to assess the benefits of treatment paradigms aimed at achieving more than one clinical goal in diabetes. In Steno-2, a Danish open-label, randomized, parallel-group study of patients with established T2DM, patients assigned to receive an intensive treatment targeting tighter goals for blood pressure (systolic <130–140 mmHg; diastolic <80–85 mmHg), HbA1c (<6.5% [48 mmol/mol]), total cholesterol (<175–190 mg/dL), and triglycerides (<150 mg/dL) had a significantly lower risk of cardiovascular disease and microvascular complications over an average follow-up of 7.8 years, compared with patients assigned to antidiabetic treatment in accordance with national guidelines (21). A 5.5-year extension of that same study demonstrated that the multifactorial therapy was associated with sustained lower risk of cardiovascular events or death (20). In the Steno-2 Study, lipid-lowering treatments were suggested to have had the greatest contribution to cardiovascular risk reduction, whereas antiglycemic and antihypertensive treatments were considered to have accounted for the greatest reduction in microvascular complications (29). These results are in alignment with our findings that suggest that there are additional cardiovascular benefits associated with the achievement of both LDL-C and HbA1c goals when compared with only HbA1c goal achievement, while there are additional microvascular benefits associated with the achievement of both HbA1c and LDL-C goals when compared with only LDL-C goal achievement.
In contrast to the Steno-2 Study (21), we did not assess the impact of multiple interventions or goal achievements other than HbA1c and LDL-C. The blood pressure level cutoff recommended by major national and international guidelines for patients with diabetes (<130/80 mmHg) may be difficult to achieve, and the benefits of achieving this blood pressure goal are unclear (30). We decided to await final consensus on the optimal blood pressure goal for patients with diabetes before creating appropriate models to account for blood pressure goal achievements. Current lipid and glycemic goals, however, are relatively easy to achieve, as can be seen from our study, and their effects are therefore easier to take into consideration.
Clinical outcomes: single- vs. no-goal achievement
Our results are also consistent with findings from large randomized trials that evaluated situations analogous to the achievement of single metabolic goals. The 10-year UK Prospective Diabetes Study found that intensive glycemic control reduces the risk of microvascular complications but does not affect the risk of macrovascular disease (31). This lack of association between intensive glycemic control and macrovascular benefits has been observed in other large trials, such as Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation (ADVANCE) (4), Action to Control Cardiovascular Risk in Diabetes (ACCORD) (2,3), and Veterans Affairs Diabetes Trial (VADT) (32), except in the case of patients with newly diagnosed diabetes (5,33). Our study showed a 6% lower hazard of cardiovascular-related end point for HbA1c achievers compared with no-goal achievers. Our findings are in agreement with the results from several large randomized trials that found that the treatments aimed at lowering LDL-C were associated with a reduced risk of cardiovascular events in patients with diabetes (6–10).
In our study, achievement of the HbA1c goal alone and the LDL-C goal alone were each associated with a lower risk of cardiovascular surgeries, which is consistent with other research in patients with diabetes which has shown that patients who undergo CABG (34–36) and PCI (37,38) often have elevated HbA1c levels, and that lowering LDL-C levels is also associated with lower risks of coronary events (6,7,39).
Economic outcomes
This is the first study to examine the differences between dual- and single-goal achievement in terms of health care resource utilization and costs in patients with diabetes. Our results suggest that dual-goal achievement provides additional economic benefits. over LDL-C goal alone, but not over HbA1c goal alone. Our results are consistent with previous studies that show that control of either LDL-C (40) or HbA1c (15) is associated with cost savings. For example, a 2003 study designed to assess cost of statin therapy for the primary prevention of major coronary events in U.S. patients with diabetes and LDL-C levels >100 mg/dL found that among individuals with LDL-C levels of 100–129 mg/dL and ≥130 mg/dL, the annual cost difference between patients with major coronary events with statin treatment versus without statin treatment was $480–950 and $590–1,920, respectively (16). A 2005 analysis conducted to predict costs and outcomes for patients with uncontrolled T1DM and T2DM, compared with patients who remained at HbA1c levels of 7% (53 mmol/mol) or 6.5% (48 mmol/mol), found that efficient targeting of financial resources toward achievement of HbA1c goals in the U.S. would result in $35–72 billion savings over the subsequent 10 years (15).
The design and methods of this study had several strengths. First, the data used for this study include laboratory measurements for patients over time. Second, this study used a longitudinal design that was able to capture the time-varying nature of laboratory measurements. This allowed for better estimation of the association between goal achievement and risk of complications over time, compared with a simple cross-sectional design, which would use baseline laboratory values in regression models. The limitations of the study include the usual caveats associated with retrospective studies. First, due to the retrospective observational design, the analysis may have been affected by unobserved factors that were not taken into account in the model. Second, the electronic medical records did not include information on disease severity, disease duration, lifestyle modifications, or other interventions. Third, although patients enrolled in the VHA do not typically use services outside of the system, any health care services that were administered by a provider outside of the VHA were not included in the electronic records. Fourth, because we used VHA data, the patients in our study were predominantly male. Although our sample was representative of the VHA population, gender imbalance may limit the generalization of our findings. As the VHA population may have characteristics that are distinct from those in the general population, similar studies in the general population should be performed. Finally, studies on the clinical and economic benefits associated with triple goal achievement of HbA1c, LDL-C, and blood pressure goals may shed additional light on the appropriate management of patients with T2DM.
In conclusion, this retrospective study among U.S. veterans suggests that the achievement of both LDL-C and HbA1c goals is associated with additional clinical and economic benefits, compared with the achievement of either goal alone. These findings may facilitate decision making when considering the health and pharmacoeconomic benefits of various treatment strategies to target multiple treatment goals for individuals with diabetes.
A slide set summarizing this article is available online.
Acknowledgments
This study was sponsored by Daiichi Sankyo, Inc. X.Y. is an employee at Daiichi Sankyo, Inc. M.L., E.Q.W., H.S., and D.T. are employees at Analysis Group, Inc. and have received funding from Daiichi Sankyo, Inc. No other potential conflicts of interest relevant to this article were reported.
The contents of this article do not represent the views of the Department of Veterans Affairs or the U.S. Government.
L.S., M.L., E.Q.W., and H.S. conceived and designed the study, analyzed and interpreted data, and reviewed and edited the manuscript for important intellectual content. X.Y. conceived and designed the study and reviewed and edited the manuscript for important intellectual content. D.T. analyzed and interpreted data and reviewed and edited the manuscript for important intellectual content. V.A.F. conceived and designed the study, reviewed and edited the manuscript for important intellectual content, and supervised the study. L.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Parts of this study were presented in abstract form at the 73rd Scientific Sessions of the American Diabetes Association, Chicago, Illinois, 21–25 June 2013, and at the 18th Annual International Meeting of the International Society for Pharmacoeconomics and Outcomes Research, New Orleans, Louisiana, 18–22 May 2013.
The authors thank the Department of Veterans Affairs VISN 16 data warehouse for the de-identified dataset and the Southeast Louisiana Veterans Health Care System for additional resources. Dr. Vojislav Pejović, Dr. Adam Ruth, and Dr. Michael Miller, Prescott Medical Communications Group, Chicago, Illinois, provided medical writing and editorial assistance.