METHODSData from the 2011–2014 Medical Expenditure Panel Survey (MEPS) were used for this analysis. MEPS is a nationally representative survey of the U.S. civilian non-institutionalized population that provides comprehensive information on healthcare utilization and costs under the U.S. healthcare system. MEPS data have been used extensively for studying medical expenditures for various health conditions, including hypertension and cardiovascular disease (CVD).18–22 The MEPS data on medical conditions were used to define hypertension and comorbidities based on ICD-9 codes. The MEPS full-year consolidated data were employed to define medical expenditures and patient characteristics.
Study PopulationAdults with hypertension were included if they met all of the following criteria: any diagnosis codes for hypertension (ICD-9 codes: 401–405); age ≥18 years; and were not pregnant during the study period. Among pregnant women, hypertension and gestation hypertension might affect the use and cost of medical services differently from the general population. A total of 26,049 persons were included.
MeasuresVarious comorbidity indices have been developed to measure important coexisting medical conditions. The Elixhauser Comorbidity Index (EI) is one of the most frequently used and validated comorbidity indices, and the original EI includes a set of 30 conditions that are coded as present or absent to produce 30 binary variables.23 In addition, the score version of the EI has also been validated.24,25 The score version of the EI was used to measure comorbidity because it is better than the original version for exploring individual comorbidity status and the impact of individual comorbidity on medical expenditures. In addition, using the score version of the EI has been validated by its ability to outperform the Charlson Comorbidity Index, the most commonly used comorbidity index.26–34
The authors expanded the EI to better analyze the impact of hypertension-related comorbidities on medical expenditures by including a broader range of CVD-related conditions than the original EI. First, 27 conditions from the original version of the EI indicators were obtained. These modifications included excluding hypertension; combining two types of diabetes (diabetes with complication and diabetes without complication) into one condition (diabetes); and combining two types of anemia (blood loss anemia and deficiency anemia) into one condition (anemia). Second, three conditions were added: coronary heart disease (CHD); stroke; and hyperlipidemia. Finally, two conditions were redefined: renal failure to renal disease, and arrhythmia to heart rhythm disorders. In sum, a total of 30 possible comorbidities were included, as follows:
CVD and hypertension-related conditions: congestive heart failure (CHF), heart rhythm disorders, valvular disease, CHD, stroke, hyperlipidemia, pulmonary circulation disorders, peripheral vascular disorders, diabetes, renal disease, obesity, and depression; and
other conditions: paralysis, neurologic disorders, hypothyroidism, liver disease, peptic ulcer, HIV/AIDS, lymphoma, metastatic cancer, solid tumor, rheumatoid arthritis, coagulopathy, weight loss, fluid and electrolyte disorders, anemia, alcohol abuse, drug abuse, psychoses, and chronic pulmonary disease.
The modified EI score was defined as the sum of each of the 30 comorbidities, with a possible range of 0–30. Depression was included in the CVD and hypertension-related conditions based on the results of previous meta-analyses on the association of hypertension with depression.35,36
Appendix Table 1 (available online) shows the list of the comorbidities and the ICD-9 code for each condition.
The outcome variable was annual total medical expenditures that were not specifically related to hypertension diagnosis or management. The expenditures included those for inpatient hospitalizations, outpatient visits, emergency department visits, office-based physician visits, home healthcare visits, and prescription medications. The annual medical expenditures included the total of all expenditures paid by all payers, including Medicare, Medicaid, other public insurance, private insurance, and payments made by the adult with hypertension or that person’s family during the calendar year. In terms of cost of illness studies, the perspective for this study was that of the healthcare sector as a whole, with expenditures used as a proxy for costs. All costs were adjusted to 2014 U.S. dollars using the personal consumption expenditure index–health by function. Ten covariates were included: age group, gender, race/ethnicity, U.S. region, insurance type, marital status, education level, poverty level, smoking status, and health status.
Statistical AnalysisBivariate analyses were conducted to assess the associations between personal characteristics and the EI. Chi-square tests were used for the categorical variables for the population’s characteristics, and ANOVA was used for the continuous variables. Descriptive statistics were employed to describe the frequencies of comorbidities.
A generalized linear model regression (gamma distribution with a log link function) was used to estimate the associations between the modified EI and total medical expenditures while controlling for covariates. A log link and gamma distribution was selected because the distribution of medical expenditures was positively skewed, and 99% of the sample had positive medical expenditures. Mean annual medical expenditures by the EI score were predicted using the average marginal effects method from the model. All analyses were performed using SAS, version 9.3, and Stata/SE, version 14.
RESULTSA total of 26,049 patients were included in this study. Table 1 shows the characteristics of U.S. hypertensive adults by the modified EI. The weighted average estimate of the number of U.S. hypertensive adults was 64,147,867 per year, and the weighted percentage of hypertension in the population of interest was 27.8%.
In all, 14.0% of U.S hypertensive adults did not have any comorbidity (n=4,073), whereas 23.0% had one (n=6,135); 24.4% had two (n=6,310); and 38.7% had three or more (n=9,531). Figure 1 shows the percentages of the 15 most frequent comorbidities in U.S hypertensive adults, which were hyperlipidemia, diabetes, rheumatoid arthritis, depression, chronic pulmonary disease, CHD, hypothyroidism, renal disease, solid tumor, heart rhythm disorders, stroke, neurologic disorders, fluid and electrolyte disorders, CHF, and valvular heart disease.
Table 2 shows the estimated total medical expenditures by personal descriptors and the number of comorbidities in U.S. hypertensive adults. Overall, the mean annual medical expenditure was $9,841 (95% CI=$9,464, $10,218). The comorbidity score was positively associated with the estimated mean annual total medical expenditures. Estimates were $3,914 (95% CI=$3,456, $4,372) for those without any comorbidity; $5,798 (95% CI=$5,384, $6,213) for those with one; $8,333 (95% CI= $7,821, $8,844) for those with two; and $13,920 (95% CI=$13,166, $14,674) for those with three or more comorbidities. Appendix Table 2 (available online) describes generalized linear model estimates and associated p-values for total medical expenditures.
Figure 2 shows the estimated medical expenditures by the presence of nine hypertension-related comorbidities among the 15 most frequent comorbidities. The impact of comorbidities on medical expenditures varied by type, with CHF associated with the largest impact on expenditures. Annual total medical expenditures were $17,063 (95% CI=$16,224, $17,902) in persons with CHF versus $9,683 (95% CI=$9,313, $10,053) in persons without CHF, a difference of $7,380. The increased medical expenditures from specific comorbidities were $6,069 for stroke, $6,046 for CHD, $5,093 for heart rhythm disorders, $4,851 for hyperlipidemia, and $4,453 for valvular disease. Diabetes, renal disease, and depression each increased expenditures by ≅$5,000. Appendix Table 3 (available online) describes the differences in estimated annual medical expenditures by the presence of comorbidities.
DISCUSSIONThe present study used a nationally representative sample of the U.S. population to estimate medical expenditures in hypertensive adults according to comorbidity status. To the authors’ knowledge, this is the first study that provides insight into the economic burden of hypertension by comorbidity status. Overall, as expected, the study found that a higher comorbidity score was associated with higher total medical expenditures in adults with hypertension. In addition, the differences in expenditures varied by type of comorbidity.
Also as expected, having incrementally more total comorbidities was associated with increasing medical expenditures. In non-hypertensive adults without any of the listed comorbidities, annual total medical expenditures were $1,656 (authors’ calculation from the 2011–2014 MEPS), which could reflect the use of preventive services or medical services for other conditions that were not included in the modified EI. Thus, having hypertension alone added $2,258 per year, because annual medical expenditures were $3,914 in hypertensive adults without any comorbidities included in the EI, and this figure increased to $13,920 in those with three or more comorbidities included in the EI. The findings of the present study extend this work to pinpoint even further which American adults with hypertension generally incur the highest expenses for medical care.
Hyperlipidemia, CHD, heart rhythm disorders, CHF, valvular disease, and stroke were classified in this study as comorbidities related to CVD. Notably, all six CVD-related comorbidities emerged in the list of the 15 most frequent comorbidities in the population of interest.
Hyperlipidemia was the most frequent comorbidity in adults with hypertension in this study, indicating more than half of adults with hypertension had hyperlipidemia. Similarly, a study using 2009 MEPS data also found that hypertension and hyperlipidemia was the most frequent dyad.37 Based on the 2012 edition of the chartbook, 52.9% of Medicare beneficiaries had both hypertension and hyper-lipidemia.38 The present study found that having hyper-lipidemia added $4,851 in medical expenditures per year at the individual level, albeit this marginal increase was lower than it was for the other CVD-related comorbidities except for valvular disease ($4,453). This is a substantial economic burden to society because of the high prevalence of hypertension and hyperlipidemia. Because hypertension and hyperlipidemia are major CVD risk factors and often coexist, adults with hypertension and hyperlipidemia should adopt healthier lifestyles and adhere to prescribed medications to reduce their risk of CVD.39
In this study, having CHF was associated with the highest incremental medical expenditure in hypertensive adults for the 15 most common comorbidities. Annual total medical expenditures were $17,063 in persons with CHF and $9,683 in those without CHF, a difference of $7,380. Stroke and CHD were ranked as the third and fourth most costly comorbidities among the 15 most frequent, each adding approximately $6,000. Among Medicare beneficiaries, hypertension, hyperlipidemia, and ischemic heart disease, together, was the most prevalent triad with a prevalence of 33.7%.38 In addition, approximately 55% of Medicare beneficiaries with stroke and heart failure had five or more additional chronic health conditions. As such, a patient with stroke or heart failure also had many other conditions contributing to the overall difference in expenditures.38 Hypertension is an independent risk factor for both stroke and CHD. Moreover, appropriate hypertension treatment has been found to reduce the incidence of CHF by >50%, of stroke by 30%–40%, and of myocardial infarction by 20%–25%.40–42 Thus, the appropriate management of hypertension, by itself, can reduce the economic burdens to society. Furthermore, such management can decrease the prevalence of these CVDs and reduce overall medical expenditures incurred.
Diabetes is closely related to hypertension. Indeed, hypertension and diabetes share several biological and environmental risk factors, and they frequently occur together.43 In this study, diabetes was the second most common comorbidity; 27.3% of hypertensive adults also had diabetes. Similarly, in a study using 2009 MEPS data, diabetes with hypertension and hyperlipidemia was the most frequent triad.37 In addition, a previous study found that the prevalence of diabetes in a hypertensive population was 27%.44 In the whole population, a previous study using the MEPS data reported that the prevalence of diabetes was 8%.45 Thus, the prevalence of diabetes in hypertensive adults seems to be roughly three times that seen in the general population. Because of the higher prevalence of diabetes in persons with hypertension, managing diabetes and hypertension together could be an effective way to reduce the economic burden on society. A community health workers program may be a good example.46
Hypertension and renal disease are interrelated, and hypertension may cause or result from renal disease.47,48 In previous studies, although the prevalence of hypertension varied in patients with renal disease, hypertension was present in up to 90% of such patients.49–51 Management of hypertension is more difficult when there is poorer renal function. This study found that renal disease is not only a frequent comorbidity but also a costly problem in hypertensive adults. Implementing effective awareness and prevention programs for renal disease could help to reduce costs in hypertensive adults with renal disease and respond to public health demand.52
In this study, depression was the fourth most common comorbidity in adults with hypertension, affecting one fourth of this population. This result is quite similar to that obtained by a meta-analysis published in 2015 that found the prevalence of depression to be 26.8% in patients with hypertension.53 Depression is known to have a negative impact on health outcomes in patients with hypertension.21,54,55 The present study found that the presence of depression increased expenditures by just under $5,000. Thus, effective screening for depression and appropriate monitoring of this disorder is important in caring for hypertensive patients. In addition, education in self-management can be a solution to empowering patients to reduce blood pressure and reduce stress and depressive feelings.
Efforts to prevent and control hypertension are critical because hypertension is a major risk factor for heart disease, stroke, and renal disease. In this study, CHF, stroke, CHD, and renal disease were all in the top expenditures category. Hypertension can be better controlled by several approaches, including diet, nutritional supplements, lifestyle modification, and conventional antihypertensive medications.56 The Community Guide has found several strategies to be effective for hypertension prevention and control, including self-measured blood pressure, community health workers, clinical decision support systems, reducing out-of-pocket costs for clinical services, and team-based care.57–59 In addition, managing obesity could be one of the effective ways for controlling of hypertension and other chronic conditions.60 Furthermore, appropriate hypertension management strategies should be developed for hypertensive adults with rheumatoid arthritis, depression, and chronic pulmonary diseases. Although these conditions are not directly associated with hypertension, these are frequent comorbidities in hypertensive adults. Nevertheless, little is known about the evidence on hypertension management strategies with these comorbidities.
This study had several strengths. First, the sample used was large, with a nationally representative database. The large sample enabled exploration into detailed comorbidity status and its impact on medical expenditures. Second, this study used a modified EI, expanding the list of CVD-related diseases to include such disorders as CHD, stroke, and hyperlipidemia. To the authors’ knowledge, this is the only study focusing on issues of comorbidity among hypertensive adults. Finally, a rigorous empirical approach that specified various regression models for the estimation was employed, which were able to control for various covariates.
LimitationsThis study had several limitations. First, the data used were cross-sectional, and thus the authors can only claim that the medical expenditures were disease-associated although acknowledging that they may not have been the treatment costs for each specific disease. Furthermore, the impact of each comorbidity on the total medical expenditures cannot be interpreted as hypertension-associated. The purpose of this study was to determine how much the comorbidity increased the annual total medical expenditures in hypertensive adults. Second, MEPS does not include institutionalized persons in its database. Such people may have many serious comorbidities, and thus this study no doubt underestimated the percentage of hypertensive adults in the U.S. with comorbidities and therefore the associated direct medical costs. Third, because MEPS is a national survey, there is a possibility of recall bias. Respondents might not be able to correctly provide certain types of information, such as medical history. Fourth, this study did not investigate indirect costs, such as caregiver burden and absenteeism. Further research may be needed to examine the impact of comorbidity on productivity loss and the associated indirect costs in adults with hypertension. Fifth, the outcome variable was annual total medical expenditures including any types of medical services. For future studies, analyzing the breakdown of total medical expenditures by the type of services, such as inpatient hospitalizations, outpatient visits, emergency department visits, office-based physician visits, home health-care visits, and prescription medications could provide additional insights about cost burdens and guide intervention efforts. Finally, it was found that being a current nonsmoker was associated with higher annual medical expenditures among hypertensive adults, compared with being a current smoker. However, it is previously known that smoking in hypertensive adults, including former smokers, was associated with increased medical expenditures.61 In MEPS, there is no information about smoking history. The most likely explanation is that hypertensive adults who are sick, with higher expenditures, are more likely to quit smoking.