Medicaid Medical Costs Associated with Hypertension by Diabetes Status among Women in Alabama
Though a high proportion of Medicaid population in Alabama are women, little is known about their economic burdens of diabetes and hypertension. We used Alabama Medicaid claims data of 16,107 female enrollees aged 19–64 years to estimate per-capita total annual medical costs of hypertension by diabetes status. Hypertension prevalence was 60.0% and 17.3% among those with and without diabetes. The estimated annual medical cost for enrollees with hypertension was $6,689 (in 2017 $), of which $2,369 was associated with having hypertension. The hypertension-associated excess costs were $2,646 and $2,378 for enrollees with and without diabetes. All subgroups such as Blacks and those with Charlson Comorbidity Index ≥ 1, had higher medical costs when they had a combination of hypertension and diabetes compared with having diabetes without hypertension. Hypertension and diabetes increased medical costs substantially, and the findings can inform decision makers about effective resource utilizations for prevention and treatment strategies.
Health care expenditure, high blood pressure, diabetes, vulnerable population
In 2016, heart disease was the leading cause of death in the United States and diabetes was the 7th leading cause of death.1 Hypertension, a major risk factor for heart disease, and diabetes share several biological, behavioral, and environmental risk factors2 and, as a result, frequently present as comorbidities. Among the U.S. Medicaid population, which is relatively young with ∼73% younger than 45 years old in 20133 and disproportionately affected by chronic conditions than their uninsured counterparts,4 the prevalence of diabetes and hypertension averages approximately 8%–13% and approximately 17%–27%, respectively.5 Recent analyses of the Medical Expenditure Panel Survey (MEPS) on noninstitutionalized U.S. adults aged 18 years or older found that approximately 27% of adults with hypertension have diabetes,6 and 78% of adults with diabetes have hypertension.7
The economic burden of hypertension and diabetes is substantial. For the Medicaid population, the total annual medical costs per person with hypertension ranged from approximately $5,000 to $20,000, and the estimated per-person annual costs associated with diabetes ranged from approximately $3,200 to $4,600.5 Recent studies using MEPS found that costs for adults with both hypertension and diabetes were $4,400 more than for adults with diabetes alone.7 When compared with costs associated with having hypertension alone, costs were $5,000 greater among adults with both hypertension and diabetes.6
Morbidity, mortality, and associated medical costs vary among state Medicaid programs because of different demographic characteristics, coverage benefits, and policies. A previous study documented that per-person annual medical costs associated with diabetes varied greatly among state Medicaid programs, ranging from $5,000 in Alabama to $15,000 in New York.8 To understand and monitor state-level economic burden of chronic conditions more effectively, it was suggested that the data be examined for the Medicaid population in individual states.8
As a public health insurance program that is administered by states and funded jointly by states and the federal government, Medicaid plays an important role in providing health care coverage for about 66 million people in the United States, including eligible low-income adults, children, pregnant women, elderly adults, and people with disabilities.9 The coverage benefits and eligibility criteria vary widely across states. For example, for parents of dependent children (in a family of three), income eligibility limits as a percent of the federal poverty level ranged from 201% in Connecticut to 18% in Alabama and Texas.10
Overall, women account for a higher proportion of the Medicaid population than men.3 Among the states, Alabama has one of the highest proportions of adult female Medicaid enrollees (over 60% in 2013).3 In general, women have a higher prevalence of multiple chronic conditions than men.11 More specifically, hypertension among individuals with diabetes is more common and more severe among females. Furthermore, females with diabetes are more likely to use medical services than males with the same condition,12–14 and thus incurring greater medical costs. A focus on prevention among the Medicaid population, especially for women, may lessen costs associated with hypertension, diabetes, and related comorbidities. For example, appropriate antihypertensive therapy reduces the risk and incidence of cardiovascular disease among adults with diabetes.15–17 Additionally, we also note a recent position statement by the American [End Page 524] Diabetes Association emphasizing the great importance of hypertension treatment among people with diabetes.18
Socioeconomic status and race/ethnicity both play important roles as risk factors for hypertension and diabetes.4,19,20 Low-income populations and Black adults are disproportionately burdened, and many of them are insured by state Medicaid programs.4,19,20 Among Medicaid enrollees in Alabama, about 45% are Black.3 The Medicaid program in Alabama pays for medical services for residents of the state in need of health care assistance who have generally low/very low incomes.21 In 2018, the annual family income threshold to be eligible for parents in a family of three was 18% of federal poverty level, or $3,740 in Alabama.10
To our knowledge, data about the effects of hypertension on medical cost by diabetes status is lacking, especially among Medicaid enrollees at the state level. Such information is important for informing public health decision makers who make resource allocation decisions, and practitioners who develop prevention and treatment strategies. This study addresses this gap by estimating hypertension-associated medical costs by diabetes status among female enrollees aged 19–64 years living in Alabama using Alabama Medicaid claims data.
We analyzed the Centers for Medicare and Medicaid Services (CMS) administrative claims data, also known as Medicaid Analytic eXtract (MAX) files, which include 100% of fee-for-service (FFS) enrollees.22 The files contain enrollment information, medical service use, and cost of services for FFS Medicaid enrollees at the individual level. All health care service uses, costs, and enrollment information were linked through encrypted enrollee identifiers.22
Our study sample included 16,107 adult female Medicaid enrollees aged 19–64 years who were enrolled in a FFS plan for the entire calendar year of 2012, which is the latest available data at the time of the study for Alabama. We created a sub-group based on enrollees' diabetes status, defined as having at least one inpatient or two outpatient claims at least 30 days apart on the basis of a primary or secondary diagnosis of International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes for diabetes (250.xx, 357.2x, 362.0x, 366.41).23 In each group, enrollees were identified as having hypertension if they had at least one inpatient or two outpatient claims on the basis of a primary or secondary diagnosis of ICD-9-CM codes for hypertension (401.x to 405.x, 362.11, 437.2).24
We excluded (a) those with dual eligibility status (eligible for both Medicare and Medicaid) and disability-based eligibility status;25 (b) those in long-term care facilities, excluded because diagnosis codes reflected enrollees' diagnoses at admission to the facility, and therefore their health care uses may not have been fully captured;22,26 (c) pregnant women, excluded because they have unique medical or treatment needs;27 and (d) those who did not identify as Black or White, excluded because of the small proportions (approximately 1%) (Figure 1).
Our main outcome for this analysis was [End Page 525]
medical cost for total health care services, which included the costs for inpatient stays, outpatient visits, and prescription medications paid for by the Alabama Medicaid program. We adjusted all costs to 2017 U.S. dollars using the U.S. Consumer Price Index (https://www.bls.gov/cpi/).
We first calculated means for continuous variables and proportions for categorical variables of the Medicaid enrollees by diabetes status. For characteristics of enrollees, we compared proportions between enrollees with and without diabetes by using the chi-squared test. We compared per-capita, unadjusted mean annual medical costs between those with and without hypertension by using the t-test. We conducted the test for the total sample and the samples of enrollees with and without diabetes.
The cost distribution of the data is right-skewed, and 1.4% of enrollees have no medical costs. To account for this we used a generalized linear model with log link and gamma distribution to estimate medical costs associated with hypertension and diabetes.28 Three generalized linear models were performed (one for the total sample, one for the sample of enrollees with diabetes, and one for the sample without diabetes). For the total sample, we controlled for several covariates, including a categorical variable for hypertension (1 = with hypertension, 0 = without hypertension) and diabetes (1 = with diabetes, 0 = without diabetes) and their interaction term/variable.29 Other covariates were age group, race (since only Whites and Blacks were included in the study, therefore, race was used instead of race/ethnicity), hypertension-related conditions, the Charlson Comorbidity Index (CCI) (the inclusion of hypertension-related conditions and CCI was to adjust for potential cofounders), and interaction terms/variables [End Page 526] between age group, race, hypertension-related conditions, and CCI with hypertension and diabetes, (Appendix, Table A1, available from the authors upon request). We identified enrollees as having hypertension-related conditions (a categorical variable) if they were diagnosed with myocardial infarction, congestive heart failure, peripheral vascular disease, chronic pulmonary disease, renal disease, or stroke on the basis of ICD-9-CM codes.24,30 The CCI was created and adapted on the basis of ICD-9-CM codes developed by Deyo and colleagues.30 For the samples of enrollees with and without diabetes, we controlled for several covariates such as hypertension, age group, race, hypertension-related conditions, CCI, and interaction terms/variables between age group, race, hypertension-related conditions, and CCI with hypertension (Appendix Table A2 and A3, available from the authors upon request).
We estimated per-capita annual mean medical costs by hypertension status with parameter estimates from the model and values of the covariates. The mean medical costs associated with hypertension were calculated as the difference in the predicted annual mean medical costs between enrollees with and without hypertension. We repeated this process for the total sample and samples of enrollees with and without diabetes. In addition, we also estimated mean medical costs associated with hypertension for Black and White enrollees, for those with one or more hypertension-related conditions and those without, and for those with one or more CCI and those with no CCI. We conducted all statistical analyses using SAS Enterprise Guide 7.1 (SAS software. Copyright © 2017 SAS Institute Inc. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc., Cary, NC, USA.) and Stata/MP 14.2. (StataCorp. 2015. Stata Statistical Software: Release 14. College Station, TX: StataCorp LP)
Overall hypertension prevalence was 20.5%, with prevalence of 60.0% and 17.3% among enrollees with and without diabetes, respectively (Table 1). In addition, nearly 19% of the enrollees had at least one hypertension-related condition (such as myocardial infarction, congestive heart failure, peripheral vascular disease, chronic pulmonary disease, renal disease, or stroke), with a greater percent of enrollees with diabetes having at least one hypertension-related condition than those without diabetes (32.8% and 17.8%, respectively). In addition, enrollees with diabetes were older, more likely to be Black, and had higher CCI than those without diabetes (P < .01).
Per-capita, unadjusted, mean annual medical costs were $7,869 for enrollees with hypertension, and $3,875 for those without hypertension (Table 2). Among enrollees with diabetes, unadjusted mean annual medical costs were $10,629 and $7,171 for those with hypertension and those without hypertension, respectively. Among enrollees without diabetes, the costs were $7,091 and $3,745, for those with hypertension and those without hypertension, respectively. White enrollees had higher unadjusted mean annual medical costs than Black enrollees regardless of hypertension and diabetes status (Table 2). Similarly, those with one or more hypertension-related conditions had higher unadjusted mean annual medical costs than those without, regardless of hypertension and diabetes status. Although not the focus of the study we also calculated per [End Page 527]
capita, unadjusted, mean annual medical costs for adult male Medicaid enrollees by diabetes and hypertension status. (Appendix, Table A4, available from the authors upon request).
After controlling for covariates, the estimated annual cost for enrollees with hypertension was $6,689, of which $2,369 was associated with having hypertension (Table 3). The hypertension-associated excess costs were $2,646 and $2,378 for enrollees with and without diabetes, respectively (all P < .01). Among enrollees with hypertension, having [End Page 528]
[End Page 529] one or more hypertension-related conditions increased medical costs by $4,150 and $2,637 for those with and without diabetes (P < .01). Total medical costs were higher regardless of hypertension and diabetes status for White enrollees compared with Black enrollees (P < .01, Table 3 and Appendix, Table A1-A3, available from the authors upon request). These estimates may lead to over $12 million associated with either condition, or about 16% of the $77 million in medical costs incurred among enrollees in our studied population (Appendix, Table A5, available from the authors upon request).
Hypertension and diabetes are frequently occurring and costly conditions. Hypertension is one of the most frequent comorbidities among people with diabetes and vice versa.6,7 Little is known about the single and joint economic burdens of diabetes and hypertension among the Medicaid population.31 In our sample of non-disability-based eligibility, non-pregnant, female Alabama Medicaid enrollees aged 19–64 years, one in five had hypertension, three-fifths of those with diabetes had hypertension. We found that excess medical costs associated with having hypertension were substantial, regardless of diabetes status. However, having hypertension in addition to diabetes was associated with $2,646 in excess cost compared with having diabetes alone, and resulted in $10,233 in overall annual health care spending per enrollee. Furthermore, all subgroups had higher medical costs when they had a combination of hypertension and diabetes compared with having hypertension without diabetes. Overall, about 16% of the total medical cost was associated with either hypertension or diabetes in the population we studied (Appendix, Table A5, available from the authors upon request). Because of the frequency of these conditions co-existing and the excess cost associated with having them, information about the epidemiologic and economic burdens of diabetes and hypertension among Medicaid enrollees has important public health implications such as assisting decision-makers to realize the potential savings gained through their prevention.
A recent study of U.S. adults aged 18 years or older found that over 78% of people with diabetes also had hypertension, whereas about 30% of persons without diabetes had hypertension.6,7 They estimated that the excess medical costs associated with having hypertension among people with and without diabetes was around $4,400 and $2,200, respectively,7 compared with $2,646 and $2,378, respectively, reported in this study. The differences in prevalence and excess costs among these studies can be explained, in part, by the demographic characteristics and insurance coverage differences of the samples. Particularly, their study population was much older with 55% of them being ≥ 45 years old compared to our study population of only 11%, which might affect those with diabetes more with aging-related comorbidities (and additional medical needs and costs) than those without diabetes. Therefore, the excess medical costs associated with having hypertension among people with diabetes was much higher than our study, but relatively close for those without diabetes. Nevertheless, both studies demonstrated that hypertension adds a substantial health and economic burden to individuals, regardless of diabetes status.
Analyzing the combined burden of diabetes and hypertension has merit because [End Page 530]
[End Page 531] there is a large overlap in the modifiable risk factors.15,16 Both biological and epidemiological data suggest that diabetes and hypertension may have a common metabolic pathway and are major independent risk factors for microvascular and macrovascular diseases.32 Previous studies, such as those assessing the health effects of using the Dietary Approaches to Stop Hypertension (DASH) diet33 or participating in the Diabetes Prevention Program,34 have shown the great potential of lifestyle approaches for jointly tackling diabetes and hypertension.33,35
Our data show that after controlling for covariates, the total medical costs for White enrollees were higher than those for Black enrollees, regardless of diabetes or hypertension status. Similar differences in the health care costs and use between racial/ethnic groups have been reported previously.36–38 Explanations for this phenomenon range from (a) geographic and institutional variation in care facilities, (e.g., it has been shown that Black adults live in areas where access to health care or specialty care are not readily available and lack services needed for their care) to (b) providers' biases with unequal treatments which resulted in differences in the receipt of services (e.g., it has been found that Black adults are less likely to receive high-quality treatments than Whites).36–38 Assuming similar treatment needs among those with and without diabetes and hypertension, our data suggest that Black enrollees might receive less frequent treatment, which might result in an increased risk for preventable, burden-some, and costly complications.
To limit the heterogeneity of our sample, we excluded disability-based eligibility enrollees, those with dual status, pregnant women, and those in long-term care facilities; the remaining group of enrollees predominantly was composed of females aged younger than 44 years (Table 1). These enrollees represent a young, low-income, vulnerable population that could not afford health care otherwise; therefore, prevention strategies are urgently needed to potentially reduce the health and economic burdens on the state.
As state Medicaid policies differ widely in their eligibility criteria and their covered services, our findings are likely not generalizable to the female U.S. population or to female enrollees in other state Medicaid programs. However, detailed state-specific economic estimates become increasingly important to judge the long-term cost-effectiveness of primary prevention programs for chronic diseases.39 Because of the limitations of administrative claims data, we were not able to distinguish type 1 from type 2 diabetes reliably. However, our data likely predominantly represent enrollees with type 2 diabetes because 90–95% of all cases of diabetes are type 2.40 In addition, the administrative claims data have only limited socioeconomic information, thus the number of controlled variables was small. Finally, the diabetes and hypertension status was based on claim information. Thus, undiagnosed diabetes or hypertension may lead our estimates being conservative.
In closing, we found a high prevalence of co-occurrences of diabetes and hypertension among female Medicaid enrollees in Alabama, and diabetes, hypertension and hypertension-related conditions increased medical costs substantially. The information can be used to inform decision-makers about effective resource utilizations for hypertension and diabetes prevention and treatment. This may assist in reducing the health burden and medical costs among this vulnerable population. [End Page 532]
The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.
BOON PENG NG is affiliated with University of Central Florida, College of Nursing & Disability, Aging and Technology Cluster. GUIJING WANG, MATTHEW D. RITCHEY, and JING FANG are affiliated with the Centers for Disease Control and Prevention, Division for Heart Disease and Stroke Prevention. MICHAEL LAXY is affiliated with the Institute of Health Economics and Health Care Management, Helmholtz Zentrum Munchen; German Center of Diabetes Research (DZD), Neuherberg-Munich; Global Diabetes Research Center, Rollins School of Public Health, Emory University; Technical University of Munich, Department of Sport and Health Science, Munich. CHANHYUN PARK is affiliated with the Northeastern University, Bouvé College of Health Sciences, School of Pharmacy. ROBIN E. SOLER, BRYCE D. SMITH, and PING ZHANG are affiliated with the Centers for Disease Control and Prevention, Division of Diabetes Translation.