Incremental cardiovascular costs and resource use associated with diabetes: an assessment of 29,863 patients in the US managed-care setting

Background Patients with type 2 diabetes are at increased risk of cardiovascular events, and there is an associated economic burden attached to this risk. We conducted a retrospective claims database analysis to evaluate incremental cardiovascular costs in diabetic versus non-diabetic patients hospitalized for a cardiovascular event. Methods Patients hospitalized for a cardiovascular event between January 1, 2001 and June 30, 2005 were identified from a large US managed-care population. Diabetic patients were identified by evidence of type 2 diabetes in the 12 months prior to the index hospitalization. Direct medical costs and resource use - including inpatient expenditures (for the index and first recurrent hospitalizations), as well as outpatient, laboratory, and pharmacy expenditures (during the 3-year follow-up period) - were determined for patients with or without diabetes. Results Of the 29,863 patients identified with a cardiovascular hospitalization, 5,501 patients (18.4%) had a history of diabetes in the pre-index period (mean age, 57.8 years; 42.1% female). The overall mean follow-up period was 22.8 months. The incidence of subsequent cardiovascular events in the first year of follow-up was significantly higher for patients with diabetes compared with non-diabetic patients for all types of cardiovascular events except angina. Compared with non-diabetic patients, patients with diabetes had similar mean direct medical costs per patient for the index cardiovascular hospitalization ($17,435 versus $16,917; P = 0.09), and the first recurrent cardiovascular hospitalization ($18,488 versus $17,481; P = 0.2), yet higher mean total direct medical costs per patient for cardiovascular events during follow-up years (Year 1: $8,805 versus $6,982; Year 2: $13,860 versus $10,056; Year 3: $16,149 versus $12,163; all P ≤ 0.0002). The cost difference between diabetic and non-diabetic patients remained significant after adjusting for age, gender and other potential confounders in multivariate regression analysis. The mean (SD) total period of inpatient cardiovascular hospitalization after 3 years of follow-up was 3.3 (12.4) days for patients with diabetes compared with 1.8 (5.8) days for non-diabetic patients (P < 0.0001). Conclusion Diabetic patients hospitalized for a cardiovascular event incur higher costs for cardiovascular care than their non-diabetic counterparts. This analysis of the incremental cardiovascular cost and resource use provides the basis for greater accuracy and precision when modeling the economic value of initiatives aimed at reducing cardiovascular morbidity in patients with diabetes mellitus.


Introduction
Approximately 17.5 million people in the United States have recognized diabetes, with an estimated 1 million new cases diagnosed each year [1]. Diabetes is a wellestablished risk factor for future cardiovascular events [2], including coronary heart disease (CHD), ischemic stroke, and peripheral vascular disease. Of the ~284,000 deaths in 2007 attributed to diabetes and its associated complications, nearly 65% were due to cardiovascular or cerebrovascular causes [1]. Indeed, the presence of diabetes can increase the risk of CHD and stroke by as much as 5-fold [3][4][5][6], in addition to conferring a worse prognosis for survival from a cardiovascular event [7][8][9][10]. As such, diabetes is currently regarded as a CHD risk equivalent [11,12] and recommendations for aggressive risk factor management in the diabetic population reflect this high-risk status [13].
Given the prevalence of diabetes and the morbidity and mortality associated with it, the burden the disease imposes on the US economy is considerable: the total cost of diabetes in 2007 was estimated at $116 billion in medical expenditures and an additional $58 billion in lost productivity [1]. Patient-level estimates of the cost of cardiovascular care in diabetic populations are critical for pharmacoeconomic modeling and guiding decisions on disease management in this cohort. Although previous studies have investigated the impact of cardiovascular disease (CVD) on health-care costs for diabetes [14][15][16][17][18][19][20], few contemporary analyses have determined incremental cardiovascular costs among diabetic versus non-diabetic patients with pre-existing CVD, particularly those costs associated with specific types of cardiovascular events such as coronary artery bypass graft (CABG) procedures, myocardial infarction (MI), and ischemic stroke. Therefore, using claims data representative of a large US managed-care population, we assessed the direct medical costs and resource use associated with initial and subsequent cardiovascular episodes in patients with or without diabetes hospitalized for a cardiovascular event.

Data source and claims
This retrospective cohort analysis assessed transactional billing records from the PharMetrics Patient-Centric Database, which contains fully adjudicated medical and pharmaceutical claims for > 50 million unique patients from > 90 health-care plans across the United States.
The database includes both inpatient and outpatient diagnoses (based on International Classification of Diseases, 9th Revision, Clinical Modification [ICD-9-CM] codes) and procedures (based on Current Procedural Terminology 4  and Health Care Financing Agency Common Procedure Coding System [HCPCS] codes), in addition to both retail and mail-order prescription records (including Generic Product Identifier [GPI] codes). Both paid and charged amounts are available for all services rendered, as well as dates of service for all claims. Additional data elements include demographic variables (eg, age, gender, geographic region), product type (eg, Health Maintenance Organization, Preferred Provider Organization), payor type (eg, commercial, selfpay), provider specialty, and start and stop dates for plan enrollment. Only health-care plans that submit data for all plan members are included in the database, ensuring complete data capture and representative samples. Contributions are also subjected to a series of data quality checks to ensure a standardized format and minimal error rates.

Study period
The overall study period was from January 1, 2001 to June 30, 2006. The patient identification period was from January 1, 2001 to June 30, 2005. The first hospitalization with relevant diagnosis or procedure codes for a cardiovascular event during the identification period was defined as the index hospitalization. The pre-index period was defined as the 12 months prior to the index hospitalization admission date. The post-index (follow-up) period began 1 day after the discharge date of index hospitalization. The follow-up period was stratified to define 3 cohorts of patients with 1, 2, or 3 years of continuous enrollment after the index hospitalization discharge date. Follow-up terminated at the earliest of patient disenrollment or June 30, 2006. Patients who died during the study were included in the analysis regardless of length of follow-up.

Identification of study cohorts
The patient selection procedure is shown in Figure 1. Patients with a complete hospitalization (admission and live discharge within the identification period) for a cardiovascular event between January 1, 2001 and June 30, 2005 were identified from the claims database on the basis of ICD-9-CM diagnosis codes and CPT-4 procedure codes (see Additional File 1). Inclusion in the final study cohort required patients to be continuously enrolled for the entire pre-index and follow-up periods, to not have a cardiovascular-related claim in the pre-index period, to have an index hospitalization stay ≤ 27 days, and to be ≥ 18 years of age at index hospitalization. Diabetic patients were identified by evidence of type 2 diabetes in the preindex period (diagnosis of type 2 diabetes or antidiabetic medication use; Additional file 1).

Cost analyses and resource utilization
All medical, laboratory, and pharmacy claims were assessed for the identified patients for the period from January 1, 2000 to June 30, 2006. Direct medical costs included any cost incurred during an inpatient hospitali- The study cohort identification procedure with the final study cohort stratified by history of type 2 diabetes in the 12 months prior to hospitalization. CV, cardiovascular Figure 1 The study cohort identification procedure with the final study cohort stratified by history of type 2 diabetes in the 12 months prior to hospitalization. CV, cardiovascular. Figure provides an overview of the patient identification procedure with reasons for inclusion/exclusion through the selection process. Subsequent cardiovascular events following discharge from the index hospitalization were assessed by identifying complete hospitalizations during the follow-up period with the relevant diagnosis or procedure codes for a cardiovascular event (Additional file 1) to calculate the incidence of subsequent cardiovascular events during the first year of follow-up.

Statistical methods
Mean and SD were calculated for continuous variables; frequency was calculated for categorical variables. Continuous outcome variables were compared using a t-test of means, and categorical outcome variables were compared using a Chi Square test. Multivariable regression analysis was performed using generalized linear models (GLM) with total cardiovascular-related costs after 1-year followup as the dependent variable, adjusting for age, gender, geographic region, insurance type, physician specialty and relevant co-morbidities, including diabetes status. Statistical analyses were conducted using the SPSS ® (SPSS Inc, Chicago, IL) and SAS ® (SAS Institute Inc, Cary, NC) software suites.

Study cohorts
A total of 198,961 patients were identified as having a hospitalization for a cardiovascular event between January 1, 2001 and June 30, 2005 ( Figure 1); of these, 29,863 patients were eligible for inclusion in the final study cohort. Exclusion from the study was due primarily to the requirement of continuous enrollment for the entire preindex period and for 360 days in the post-index period, and to not have a cardiovascular-related claim in the pre-index period. Within this overall study population, 5,501 patients (18.4%) were identified as having a history of type 2 diabetes in the 12 months prior to hospitalization ( Figure 1).

Baseline demographics and clinical characteristics
The baseline demographics and clinical characteristics of the study cohorts are shown in Table 1. Overall, the patient population primarily consisted of commercially insured patients belonging to either a Health Maintenance Organization or Preferred Provider Organization health-care plan. The mean (SD) follow-up period was 22.8 months. Compared with non-diabetic patients and the study population overall, the cohort of patients with diabetes was slightly older, comprised more women, and had a higher incidence of comorbidity on the basis of the Charlson Comorbidity Index score and prevalence of hypertension and dyslipidemia ( Table 1). As such, patients with diabetes were more likely to be taking lipidlowering and antihypertensive medications. The cardiovascular event type at index hospitalization was similar across the study cohorts, with the exception of MI (lower incidence in diabetic patients) and heart failure (higher incidence in diabetic patients).

Impact of diabetes on incidence of subsequent cardiovascular events
The proportion of patients in each study cohort experiencing a subsequent cardiovascular event during the first year of follow-up is shown in Table 2. Patients with diabetes had a higher incidence of each specific cardiovascular event at 1 year of follow-up than patients without diabetes ( Table 2). The incidence of subsequent cardiovascular events in the first year of follow-up was significantly higher for patients with diabetes compared with non-diabetic patients for all types of cardiovascular event excluding angina (all P ≤ 0.03).  Table 3). An analysis of the specific type of cardiovascular event underlying the index hospitalization showed that, compared with those without diabetes, patients with diabetes had higher costs for those hospitalizations associated with CABG procedures, MI, angina, other ischemic heart disease (e.g. coronary atherosclerosis), and peripheral vascular disease ( Mean total cumulative direct medical costs per patient for cardiovascular events across 3 years of follow-up are shown in Table 5 and Figure 2. With the exception of costs associated with CABG procedures in the second year of follow-up, diabetic patients incurred higher costs for follow-up care for all cardiovascular event types assessed than patients without diabetes ( Table 5). The highest follow-up costs for both study cohorts were incurred for MI and other cardiovascular procedures. Follow-up costs were generally driven by inpatient hospitalization costs (LOS) as well as costs associated with outpatient care (eg, physician office visits, imaging tests), particularly in dia-

Impact of diabetes on health-care resource use
Patients with diabetes had a mean (SD) LOS for the index hospitalization of 4.8 (3.5) days compared with 4.3 (3.0) days for non-diabetic patients (P < 0.0001) ( Table 3). For both study cohorts, the longest periods of index hospitalization were associated with CABG procedures (diabetics: 10.0 [3.9] days; non-diabetics: 9.0 [3.6] days), followed by hospitalizations with a diagnosis of peripheral vascular   (Table 3).
For the first recurrent cardiovascular hospitalization, mean (SD) LOS was 5.6 (12.6) days for patients with diabetes, compared with 4.6 (7.3) days for non-diabetics (P = 0.0008) ( Table 4). Recurrent hospitalizations with the longest periods of inpatient stay for both study cohorts were those with a diagnosis of transient ischemic attack or other cerebrovascular accident and those involving CABG procedures (Table 4).
Mean LOS for cardiovascular hospitalizations across 3 years of follow-up are shown in Table 5. With the exception of inpatient stays with a diagnosis of peripheral vascular disease in the third year of follow-up, where data on only 2 patients with diabetes were available, diabetic patients had longer periods of follow-up hospitalization for all cardiovascular event types assessed than patients without diabetes ( Table 5). The longest periods of followup hospitalization for both study cohorts were those associated with a diagnosis of heart failure (Table 5). Overall, the total period of inpatient cardiovascular hospitalization after 3 years of follow-up was 3.3 (12.4) days for patients with diabetes compared with 1.8 (5.8) days for non-diabetic patients (P < 0.0001).

Discussion
In this retrospective claims database analysis, we provide quantitative documentation of the differential costs associated with managing patients with diabetes hospitalized for a cardiovascular event within US managed-care settings. Our observations that the diabetic population incurs higher direct medical costs for cardiovascular care during the initial hospitalization and follow-up period than their non-diabetic counterparts is not new; however, in light of advances in medical care and the pertinence of its setting (within a US managed care population), this contemporary assessment is relevant. We confirm that patients with diabetes also experience longer periods of inpatient cardiovascular hospitalization than those without diabetes. The higher costs and resource use in the diabetic cohort likely reflect the combination of a higher incidence of subsequent cardiovascular events observed in this population compared with patients without diabetes as well as a higher overall comorbidity index. In addition to evaluating the incremental costs and resource use associated with cardiovascular events overall, this study is the first of its kind to provide an economic assessment of this nature for specific cardiovascular event types such as CABG procedures, MI, and ischemic stroke.
Excess medical costs for the initial cardiovascular hospitalization in the diabetic cohort were primarily driven by large cost differentials for those hospitalizations associated with CABG procedures, other ischemic heart disease (eg, coronary atherosclerosis), and peripheral vascular disease -CVD types commonly associated with diabetes.
During the follow-up period, diabetic patients experienced a higher incidence of subsequent cardiovascular events than the non-diabetic cohort for each event type  ( Table 2). As such, medical costs for cardiovascular care during the follow-up period were consistently higher in the diabetic cohort. All cardiovascular event types examined in this analysis contributed to excess follow-up costs in patients with diabetes, with cost differentials between the diabetic and non-diabetic cohorts increasing with each successive year of follow-up. Inpatient hospitaliza-tion costs and costs associated with outpatient care, such as physician office visits and imaging tests, were the primary drivers of follow-up cardiovascular costs irrespective of diabetic status.
Taking into account expenditures for the initial cardiovascular hospitalization plus all cardiovascular events during  [14][15][16]19,20]. For example, a recent model of the lifetime costs of complications associated with diabetes found that macrovascular disease accounted for 85% of cumulative costs in the first 5 years and 52% of costs over 30 years [19].
Furthermore, we and others [14,15,18] have shown that the cost of CVD is notably higher in diabetic patients compared with similar patients without diabetes, likely due to the severity of cardiovascular events experienced by diabetic individuals as indicated by a worse prognosis for survival [7][8][9][10]. A recent study [18] assessing annual medical-care costs in diabetic patients from a large managed-care organization found that patients with both diabetes and CVD had higher medical-care costs than non-diabetic patients with CVD ($10,172 versus $6,396 per patient per year). These estimates are somewhat lower than the costs obtained in our analysis, where annual costs per patient for follow-up cardiovascular events were up to $16,149 for diabetic patients and $12,163 for non-diabetic patients in the third year of follow-up. The higher costs reported in our analysis in part reflects the fact that costs are from 2006 (versus 1999 costs in the previous study). Also, our analysis is based on actual costs paid by healthcare plan providers in the PharMetrics database, as opposed to applying standard unit costs to resource-use profiles in a not-for-profit health maintenance organization, which may have the effect of underestimating the cost of care of these patients relative to our setting. Another study estimating direct medical costs associated with an initial event and 1 year of follow-up care for specific cardiovascular complications in diabetic patients found that the total event cost was $30,364 for MI; $6,024 for angina; $40,209 for ischemic stroke; and $3,874 for TIA [20]. Although closer to our assessment of cardiovascular costs in diabetic patients, these costs are still gener-Total mean direct medical costs per patient over 3 years of follow up for all cardiovascular events (left-hand graph) and for selected cardiovascular event types (right-hand graph) ally lower than those obtained in our analysis, again likely due to the use of standard unit costs from the year 2000.
As a retrospective analysis of a health-care plan claims database, this study is not without limitations. The use of diagnosis, procedure, and medication codes to identify patients and assess downstream costs and resource use relies on the accurate assignment of these codes to patient records in order to faithfully capture an individual's medical history. Contributions to the database used in this analysis are subjected to data quality checks to ensure minimal error rates. Furthermore, only health-care plans that submitted data for all plan members are included in the database, ensuring complete data capture and representative samples. Another potential limitation of this study is that the analysis of direct medical costs focuses the interpretation of the results to a payor's perspective and to US managed-care populations. However, the records in this database are representative of the national, commercially-insured population on a variety of demographic measures including age, gender, and plan type. The data are also longitudinal, with an average member enrollment time of 2 years. Hence, the results of this analysis may be generalized to similar managed-care populations at a nation-wide level.

Conclusion
The results of the current analysis demonstrate that the incidence of subsequent events in the first year following hospitalization for an initial cardiovascular episode is significantly higher for patients with diabetes compared with non-diabetic patients. As a consequence of this increased cardiovascular burden, diabetic patients hospitalized for a cardiovascular event incur higher costs for cardiovascular care than their non-diabetic counterparts. The real-world cost estimates described here will aid the development of future economic models that assess the impact of healthcare initiatives aimed at this growing diabetic population.