Health Economics Perspective of the Components of the Cardiometabolic Syndrome

Abstract

J ClinHypertens (Greenwich). 2010;12:549–555. © 2010 Wiley Periodicals, Inc.

The components of the cardiometabolic syndrome (CMS) increase the risk of coronary artery disease (CAD). The authors compared 12‐month costs of subjects with different number of components of the CMS. In claims data from a large health benefits company, 383,420 individuals with the first International Classification of Diseases, Ninth Revision codes for hypertension, diabetes, lipid abnormalities, and obesity were identified. Patients were stratified according to presence of CAD and the number of components of the CMS. Twelve‐month costs were added after the identification of the risk factor. Mean annual costs increased with the number of components of CMS both in those with and without CAD, even after adjusting for age, sex, and comorbidities (P<.01). Similar trends were seen for medical and pharmacy costs. The adjusted total annual health care cost in those with an isolated component of the CMS was $5564 (95% confidence interval: $5491–$5631) while in those with 4 components was $12,287 (95% confidence interval: $11,987–$12,587). Individuals with accumulating components of the CMS have higher health care costs regardless of age, sex, and other comorbidities.

Components of the cardiometabolic syndrome (CMS) predispose individuals to coronary artery disease (CAD) and diabetes. ¹ , ² Patients with CAD and diabetes have shorter life expectancy, lower quality of life, and higher health care costs. ³ , ⁴ Previous Medicare reports have shown that patients with lower cardiovascular (CV) risk earlier in life have lower expenses at the end of their lives. ⁵

CAD is the leading cause of death in the United States and is an important contributor to health care expenses. However, the impact of an increasing number of components of the CMS while adjusting for the presence of CAD on medical and pharmacy costs is still unclear. This question has particular recent relevance in light of the increasing prevalence of components of the CMS in younger populations. ⁶ , ⁷ It is likely that in the near future the prevalence of multiple risk factors in older generations will increase significantly in all age groups. Therefore, it is imperative to determine how the components of the CMS can impact the health care economy.

Our hypothesis was that costs increased as components of the CMS were added. The aim of this study was to evaluate the 12‐month health care costs of subjects with different number of components of the CMS. A secondary objective was to evaluate the cross‐sectional association between the International Classification of Diseases, Ninth Revision (ICD‐9) codes for CAD and the number of components of the CMS.

Methods

Setting

This was a retrospective cohort from a database of a large health benefits company encompassing approximately 3.5 million covered members who were enrolled in a health maintenance organization (HMO), preferred‐provider organization (PPO), Medicare or Medicaid plans.

Three electronic files, a member file containing demographic information for each member per encounter (age, sex, type of insurance, and geographical region); a medical file containing up to 5 recorded ICD‐9 codes per encounter; and a pharmacy file containing all Generic Product Identifier numbers of pharmacy‐dispensed medications per claim were merged.

The study was approved by the University of Miami Institutional Review board.

Study Population

We identified all members, age 18 years and older, continuously enrolled in a health plan for at least 12 months and with a medical claim between January 1, 2003 and May 31, 2004 for at least one of the following components of the CMS (with the corresponding ICD‐9 codes): diabetes (250.xx), hypertension (401–405.xx), abnormal lipid panel (272.xx), and obesity (278, 278.0, 278.00, 278.01, 278.1). The date of the member’s first recorded ICD‐9 code was considered the index date. The identification of the components of the CMS has shown in administrative databases a 96% specificity and 95% positive predictive value. ⁸

Assessment of Coronary Disease and Hypertensive Heart Disease

We stratified health care costs for those with CAD and hypertensive heart disease. CAD was based on the presence of the following ICD‐9 codes: atherosclerosis (440.xx), myocardial infarction (410.xx–412.xx), and coronary atherosclerosis (414.xx). To evaluate the effects on health care costs of other forms of CV heart disease we included hypertensive heart disease defined as the presence of ICD‐9 code 402.xx.

Definitions

Mutually exclusive groups of ICD‐9 codes were defined according to the number of components of the CMS present in a given member. Thus, a subject with obesity and diabetes was categorized into the group of 2 components of the CMS whereas a subject with diabetes, obesity, hyperlipidemia, and hypertension was categorized into the group of 4 components of the CMS. This categorization strategy based on the number of risk factors rather than on each specific component prevents double counting of health care costs on single members. At the same time, the individual contribution of each component of the CMS to health care costs was evaluated.

Assessment of Health Care Costs

Annual health care utilization costs were determined based on the total incurred costs (medical and pharmacy) during the 12 months following the index date. Costs were also stratified by medical and pharmacy claims. The total health care costs represent the costs incurred by the health benefits company and the member’s responsibility.

The Charlson comorbidity score was used to reduce the impact that other comorbidities may have on health care costs, as previously shown in administrative databases. ⁹

Statistical Analysis

Because health care costs are naturally skewed ¹⁰ we log transformed the total health care cost variable and removed extreme outliers, defined as total costs above or below 1.5 times the interquartile range.

Trends in the distributions of components of CMS across age, sex, plan type, geographical region, and Charlson score subgroups were obtained using linear regression.

Logistic regression was used to evaluate the relationship (odds ratios and 95% confidence intervals [CIs]) between components of CMS and history of CAD, after adjusting for differences by age and sex.

The contribution that stratifying components of CMS into mutually exclusive groups have on total health care costs was estimated using generalized linear models with binomial distribution. We simultaneously adjusted for age, sex, type of insurance, geographical region, and the Charlson comorbidity score. To assess the impact of CAD on health care costs we stratified costs by the presence or absence of CAD and hypertensive heart disease.

The fitness of the data was assessed using the deviance ratio. Analyses were performed using SAS 9.0 (SAS Institute, Inc., Cary, NC) and all significance tests were 2‐tailed.

Results

Baseline Characteristics

Selected baseline characteristics of the entire sample are shown by number of components of the CMS in Table I. After the exclusion of outliers (n=17,593) from both tails the sample consisted of 383,420 patients with a mean age of 56 years old, 53% were women, 71% had HMO or PPO insurance, and 55% had abnormal lipids. When the sample was divided by number of risk factors, 48% had 1 component, 38% had 2 components, 12% had 3 components, and 1% had 4. Those patients with 1 and 4 components of the CMS were considerably younger than those with 2 and 3 (P<.01). The proportion of CAD increased from 15% on those subjects with 1 component to 29% on those with 4 components of the CMS (P<.01).

Table I.

 Baseline Characteristics for 383,420 Patients With at Least 1 Component of the CMS

Characteristic	Number of CV Risk Factors					P Valuea
	Entire Sample	1	2	3	4
Number	383,420	185,053	146,899	47,203	4265
Age, mean	56.6±14.8	54.6±16.0	58.6±13.8	58.6±12.3	54.3±10.8	<.01
Female %	53	54	52	51	59	<.01
Type of insurance, %
HMO	33	31	33	38	47	<.01
PPO	38	40	36	35	38
Medicare	24	23	26	24	14
Medicaid	5	6	4	3	1
Region, %
Southeast	31	32	32	27	24	<.01
Southwest	23	22	23	29	31
Midwest	36	34	37	38	44
Puerto Rico	8	9	7	5	1
Other	2	3	2	1	1
Risk factors, %
Diabetes	30	13	32	88	100	<.01
Obesity	9	6	8	20	100	<.01
Abnormal lipids	71	62	76	95	100	<.01
Hypertension	55	19	85	98	100	<.01
Coronary artery disease, %	19	15	22	27	29	<.01
Charlson score, mean	0.30±0.77	0.56±1.21	0.75±1.35	0.92±1.46	1.051±1.5	<.01

Abbreviations: CMS, cardiometabolic syndrome; CV, cardiovascular; HMO, health maintenance organization; PPO, preferred‐provider organization. ^a P value represents the trend in change for each characteristics by number of components of the CMS.

Health Care Costs by Coronary Artery Disease Status

Figure 1 reports the mean total, medical, and pharmacy costs stratified by CAD status. The mean annual pharmacy and medical health care costs were higher for those with 3 or more components of the CMS when compared to those with 1 component (P<.01). The largest differences seen are attributed to medical costs. All reported costs were higher in those subjects identified as having CAD (P<.01).

Figure 1.

 Mean annual total (A), medical care (B), and pharmacy (C) costs by number of risk factors and coronary artery disease (CAD) status. CMS indicates cardiometabolic syndrome.

Similar trend for health care costs increment with increasing number of components of the CMS was observed separately in men and women, and when the medians (instead of means) for total health care costs were used.

Health Care Costs by Hypertensive Heart Disease

Figure 2 reports the mean total, medical, and pharmacy costs stratified by hypertensive heart disease status. The mean annual total, medical, and pharmacy health care costs were higher for those with 3 or more components of the CMS when compared to those with 1 component (P<.01).

Figure 2.

 Mean annual total (A), medical care (B), and pharmacy (C) costs by number of risk factors and hypertensive heart disease (HHD) status. CMS indicates cardiometabolic syndrome.

Multivariate Analysis

The probability of the presence of CAD increased with increasing number of components of the CMS. Compared to those with 1 risk factor and after adjusting for age and sex, the odds ratios were 1.33 (95% CI, 1.30–1.35), 1.88 (95% CI, 1.82–1.93), and 2.87 (95% CI, 2.67–3.08) for those with 2, 3, and 4 components of the CMS, respectively (Table II).

Table II.

 Risk of CAD According to the Number of Components of the CMS

Number of Components of the CMS	Odds Ratioa (95% CI)	Odds Ratio for Men (95% CI)	Odds Ratio for Women (95% CI)
1	Reference
2	1.33 (1.30–1.35)	1.37 (1.33–1.40)	1.27 (1.24–1.31)
3	1.88 (1.83–1.93)	1.88 (1.82–1.95)	1.87 (1.80–1.95)
4	2.87 (2.67–3.08)	2.84 (2.56–3.15)	2.91 (2.64–3.22)

Abbreviations: CAD, coronary artery disease; CI, confidence interval; CMS, cardiometabolic syndrome. ^aAdjusted for age and sex, P<0.01 for trend.

Table III reports the adjusted means of total health care costs. After simultaneously adjusting for age, sex, geographic region, type of insurance, and Charlson comorbidity score the mean total costs for those with 4 components of the CMS was $12,287 compared to those with only 1 risk factor ($5564). A similar trend was observed when adjusted costs were stratified by CAD status. These costs approximately doubled for those subjects with CAD when compared to those without CAD (P<.01), for the 4 groups with different numbers of components of the CMS. Similar trends were seen with hypertensive heart disease (P<.01).

Table III.

 Adjusted Total Health Care Costs and 95% Confidence Intervals by Number of Components of the CMS and Stratified by CAD Status

All Subjects			CAD		No CAD
Number of risk factors	N	Adjusted Total Health Care Costsa	N	Adjusted Total Health Care Costsa	N	Adjusted Total Health Care Costsa
1	185,053	5564 (5497–5631)	28,184	11,894 (11,615–12,172)	156,869	4117 (4059–4176)
2	146,899	6794 (6722–6867)	31,702	13,822 (13,546–14,098)	115,197	4968 (4904–5033)
3	47,203	8438 (8334–8542)	12,904	15,531 (15,188–15,874)	34,299	6173 (607–6209)
4	4265	12,287 (11,987–12,587)	1236	20,153 (19,293–21,014)	3029	9276 (8992–9560)

Abbreviations: CAD, coronary artery disease; CMS, cardiometabolic syndrome. ^aAdjusted for age and sex, plan type, geographic region and Charlson comorbidity score, P<.01 for trend.

The mean total health care costs for those subjects with CAD and specific risk factors were, for those with hypertension ($17,121), obesity ($17,210), abnormal lipids ($14,523), and diabetes ($16,976). The mean total costs for those without CAD showed the same distributions at much lower dollar amounts.

Discussion

These data suggest that annual health care costs have a direct relationship to the number of components of the CMS and that this relationship is independent of age, sex, geographical region, type of insurance, and other comorbidities. Greater annual cost for subjects with increasing components of the CMS was observed both in those with and without CAD. At the same time, subjects with hypertensive heart disease had consistently higher total health care costs when compared to the nonhypertensive heart disease counterparts and the addition of other components of the CMS exponentially increased total costs.

Strengths that lend weight to the results found are its longitudinal cost determination, the large sample size, and the use of actual real world cost data instead of estimation of costs.

Our study, as others previously reported in the literature, found an association between the number of the components of the CMS and the presence of CAD in both men and women. The sex differences are similar to a recent report from the Framingham Heart Offspring study on total coronary heart disease. ² The relationship between the components of the CMS and CAD is partially responsible for the economic impact that increasing numbers of risk factors have on costs; however, it is evident that a clear trend of increasing health care expenditures exists among patients with and without CAD as numbers of risk factors increase.

Our finding of increasing costs with increasing numbers of components of the CMS after stratifying for CAD and adjusting for comorbidities allows us to hypothesize that this common biological process becomes evident through components of the CMS but that the impact on health is broader and more predictive than currently recognized. Alternatively, we can also hypothesize that the increase in costs are explained by health delivery systems that are not appropriately set up to have a global preventive approach to each patient but rather a curative approach where we treat one condition at a time and duplicate use of resources.

Further analysis on the types of medical costs that explain the expenditures increase and also on the increments in costs associated with acute CV events in patients with different numbers of components of the CMS may provide a better insight on this issue. A previous report on health care cost after coronary catheterization has shown an increment from $4000 to $15,000. ¹¹

A recent prospective ⁵ cohort of the Chicago Heart Association Detection Project in Industry (CHA) found $10,000 to $15,000 lower charges in the last year of life comparing subjects with no CV risk factors compared to those with 4 or more risk factors, despite having longer lifespan. However, the study evaluated the contribution of CAD in the last years of life, had a small sample size in the low and high risk strata, and did not group risk.

The fact that the increase in health care costs depends more on medical costs than on pharmacy costs raises several important questions. For instance, why is there a difference in medical costs from the group with 1 risk factor to the group with 2 risk factors? We can argue that according to evidence‐based guidelines preventive health services should not differ significantly between the 2 groups. A closer look to better understand the determinants of the cost difference in medical expenditures between each group is warranted. However, this finding supports the notion that there may be an underlying common process that may influence health status of an individual more globally and that treatment of CV risk factors should have a holistic approach rather than an individual approach since previous publications have supported the fact that diabetes is generally treated more aggressively than hypertension and hyperlipidemia. ¹²

It is unknown if interventions that aim to control earlier biological processes, such as weight reduction and/or insulin sensitizers to treat the underlying causes of metabolic syndrome for example may also have significant impact in the reduction of health care utilization cost. A recent analysis in patients with prediabetes, 52% of them with the metabolic syndrome showed cost‐effective benefit of these interventions. ¹³ Based on our results, the most logical trend will be that by controlling risk factors pharmacy costs will increase, lowering in turn medical costs therefore reducing total health care costs, since medical costs represent the majority of the total costs.

Nonetheless, there are several limitations that deserve mention. First, we had no direct measurement of the metabolic syndrome or the components of the CMS and relied on the surrogate information for obesity, hypertension, dyslipidemia, and hyperglycemia found in claims data. This method could have overestimated subjects with components of the CMS since we could have identified subjects with higher burden of disease who seek medical attention more frequently than the lower burden of disease counterparts, like prehypertensives, individuals with low high‐density lipoprotein cholesterol or overweight subjects. Second, the analysis may have included costs unrelated to these components of the CMS, but rather to other comorbidities present in these subjects; however, adjusting for the Charlson’s comorbidty score as well as the exclusion of extreme outliers decreases the possibility of overestimating costs. Third, the generalizability of the study is limited by the geographic, demographic, and burden of disease of the study population. Finally, even though we used a longitudinal design for cost estimation we do not know if individuals accumulated other components of the CMS over the 1‐year follow‐up period.

Current literature suggests that impact of having multiple chronic conditions is not the expected sum of individual conditions with respect to costs and life expectancy. ¹⁴ Our findings are compatible with that notion and support the importance of developing health policies that attempt to prevent components of the CMS and that mandate a more aggressive treatment of them early in the life cycle of an individual.

Disclosure:  Dr Tamariz, Dr Palacio, and Dr Florez are supported by a grant from the Humana Innovation Center. The study was funded by a grant in aid from Sanofi‐Aventis Pharmaceuticals to the Humana Innovations Center.