Association Between Self-Reported Hypertension and Antihypertensive Medication Use and Cardiovascular Disease–Related Events and Expenditures Among Patients Diagnosed With Hypertension

Abstract

Objectives:

Research suggests that persons who are aware of the risk factors for cardiovascular disease (CVD) are more likely to engage in healthy behaviors than persons who are not aware of the risk factors. We examined whether patients whose insurance claims included an International Classification of Diseases, Ninth Revision (ICD-9) code associated with hypertension who self-reported high blood pressure were more likely to fill antihypertensive medication prescriptions and less likely to have CVD-related emergency department visits and hospitalizations (hereinafter, CVD-related events) and related medical expenditures than patients with these codes who did not self-report high blood pressure.

Methods:

We used a large convenience sample from the MarketScan Commercial Database linked with the MarketScan Health Risk Assessment (HRA) Database to identify patients aged 18-64 in the United States whose insurance claims included an ICD-9 code associated with hypertension and who completed an HRA from 2008 through 2012 (n = 111 655). We used multivariate logistic regression analysis to examine the association between self-reported high blood pressure and (1) filling prescriptions for antihypertensive medications and (2) CVD-related events. Because most patients with hypertension will not have a CVD-related event, we used a 2-part model to analyze medical expenditures. The first part estimated the likelihood of a CVD-related event, and the second part estimated expenditures.

Results:

Patients with an ICD-9 code of hypertension who self-reported high blood pressure had a significantly higher predicted probability of filling antihypertensive medication prescriptions (26.5%; 95% confidence interval, 25.7-27.3; P < .001), had a significantly lower predicted probability of a CVD-related event (0.6%, P < .001), and on average spent significantly less on CVD-related events ($251, P = .01) than patients who did not self-report high blood pressure.

Conclusion:

This study affirms that self-knowledge of high blood pressure, even among patients who are diagnosed and treated for hypertension, can be improved. Interventions that improve patients’ awareness of their hypertension may improve antihypertensive medication use and reduce adverse CVD-related events.

Keywords

hypertension heart disease health economics

Hypertension, or high blood pressure, is a major preventable risk factor for cardiovascular disease (CVD) that affects 1 in 3 adults in the United States.¹ Of the estimated 36 million patients who had uncontrolled hypertension during 2003-2010, 14 million, or 40%, were unaware and did not receive treatment.²

The proportion of patients who are unaware that they have hypertension is higher among patients who do not have a usual source of health care than among patients who do (64% vs 36%) and among patients who do not have health insurance than among patients who do (52% vs 37%).² Hypertension is a subclinical condition; as such, it is not surprising that patients, many of whom have multiple chronic conditions, do not understand or cannot recall that they have hypertension.³ Furthermore, patients with low levels of health literacy may not be engaged in their care and may have trouble understanding the technical terms physicians use to describe hypertension and related conditions.^4

-9

Treatment strategies to control high blood pressure include antihypertensive medications and lifestyle modification, both of which rely on patient or caregiver management of care. Fewer than half of patients with uncontrolled hypertension report taking medications, and studies report that 40% to 60% of patients do not adhere to antihypertensive medications.^2,10

-13 Lack of awareness may contribute to medication nonadherence and, ultimately, to complications that are preventable.

Hypertension awareness has typically been studied in the context of a cascade of events, in which awareness is followed by treatment, which is followed by blood pressure control. Researchers who use survey data to study awareness are generally also limited to self-reported information on outcomes.¹⁴ The literature suggests that persons who are aware of CVD risk factors are more likely to engage in healthy behaviors than persons who are unaware of CVD risk factors. To our knowledge, however, no studies have examined the association between awareness and medication use to treat hypertension or related adverse CVD outcomes (eg, myocardial infarction, stroke) and costs.^15,16

We explored 2 questions in this study: (1) Are patients who are aware that they have hypertension, based on diagnosis and self-reported high blood pressure, more likely to fill prescriptions for antihypertensive medications than patients who receive a diagnosis of high blood pressure but do not report high blood pressure? and (2) Do patients who are aware of their hypertension have fewer CVD-related emergency department (ED) visits and hospitalizations (ie, CVD-related events) and medical costs than patients who are unaware of their hypertension? Previous studies of hypertension awareness used self-reported survey data. We used data on self-reported hypertension awareness from a health risk assessment (HRA) linked to hypertension diagnosis and treatment information from commercial claims data.

Methods

Data Source

We drew sample data for the study from the MarketScan Commercial Database 2007-2012 linked with the MarketScan Health Risk Assessment Database 2008-2012.^17,18 The MarketScan Commercial Database is one of the largest commercial claims databases in the United States. It represents about 38% of persons with employer-based health insurance and includes data on claims associated with inpatient stays, outpatient and ED visits, and prescription drugs (unpublished data: E. Danielson, Health Research Data for the Real World: The Thomson Reuters MarketScan Databases, white paper, 2014; Hansen LG, Chang S. Health Research Data for the Real World: The Thomson Reuters MarketScan Databases, white paper, 2011).¹⁹ Beneficiaries who completed a voluntary HRA (ie, appeared in the MarketScan HRA subsample) represented 3% to 5% of the beneficiaries in the MarketScan Commercial Database. The design, content, and questions in the HRAs varied by employers and health plans that chose to provide HRAs. We provided exact questions where available and otherwise used the language provided by MarketScan to describe the variables. Beneficiaries completed HRAs throughout the calendar year. We used the most recent data available at the time of the analysis.

Data in the MarketScan databases are deidentified; as such, this research did not involve human subjects, and institutional review board review was not required.

Inclusion and Exclusion Criteria

We included beneficiaries aged 18-64 who were continuously enrolled with drug coverage for at least 24 months and who had a diagnosis of hypertension from 2007 through 2011 using the following International Classification of Diseases, Ninth Revision (ICD-9) codes: 401xx (essential hypertension), 402xx (hypertensive heart disease), 403xx (hypertensive chronic kidney disease), 404xx (hypertensive heart and chronic kidney disease), 405xx (secondary hypertension), 362.11 (hypertensive retinopathy), 437.2x (hypertensive encephalopathy).²⁰ We excluded pregnant women and enrollees in capitated health plans or health maintenance organizations because claims from these plans may not represent the full cost of care.

We further limited the sample to patients who also appeared in the MarketScan HRA Database in the year after hypertension diagnosis, by using only the first HRA completed by a patient (excluding additional HRAs within the same year or meeting the criteria in future years). We excluded patients who completed the HRA but were not asked about high blood pressure. Because of concerns about data validity, we also excluded patients who were extremely short or tall (ie, <57 inches or >79 inches) or extremely underweight or overweight (<100 lb or ≥475 lb).

The sample selection process resulted in a pooled cross-sectional study sample (n = 111 655; Figure). Although we observed patients for 2 years, we used data from patients’ first year to determine eligibility for inclusion in the sample and data from patients’ second year to measure key dependent and independent variables.

Figure.

Study sample of patients who completed a health risk assessment (HRA) who were diagnosed with hypertension in the previous year. Data sources: 2008-2012 MarketScan Health Risk Assessment Database¹⁸ and 2007-2011 MarketScan Commercial Database.¹⁷ Any outpatient, inpatient, or emergency department visit for hypertension was determined by using the International Classification of Diseases, Ninth Revision (ICD-9) codes 401xx, 402xx, 403xx, 404xx, 405xx, 362.11, or 437.2x.²¹ Hypertension diagnosis were identified in the calendar year before HRA completion. A total of 157 600 enrollees who were not asked about high blood pressure were excluded, and 5323 enrollees who were extremely tall or short or extremely underweight or overweight were also excluded because of concerns regarding data validity. Abbreviations: HMO, health maintenance organization.

Measures

Our key independent variable was awareness of hypertension, as measured by self-reported high blood pressure on an HRA. We assessed the use of antihypertensive medications by using data on prescription drug claims filled at retail pharmacies. We identified antihypertensive medication prescription drug claims based on MarketScan-provided Redbook drug codes for therapeutic class (diuretics, beta-blockers, calcium channel blockers, vasodilators, alpha-beta blockers, angiotensin-converting enzyme inhibitors) and National Drug Codes for angiotensin II receptor blockers. We measured medication adherence among beneficiaries with ≥1 antihypertensive medication claim by using the proportion of days covered. We calculated this proportion as the patients’ supply of medication beginning from the first observed claim in the year until the end of the calendar year divided by the number of days in the same period.²¹ For example, patients who filled 1 prescription for a 90-day supply on January 1 would have a proportion of 90/365, or 24%. For patients who take more than 1 class of antihypertensive medication, we calculated the proportion separately by class and then averaged across the number of classes of antihypertensive medication per patient. We considered patients with ≥80% days covered to be adherent.

The primary outcome variables were (1) a binary variable indicating whether a patient had a CVD-related hospitalization or ED visit during the year and (2) the sum of total annual expenditures for CVD-related hospitalizations or ED visits. We identified these CVD-related events occurring in the year of HRA completion and in the previous year by using ICD-9 code groups for ischemic heart disease and acute myocardial infarction (410-414.xx), cerebrovascular disease (43x.xx, 800-804.xx, 850-854.xx, 997.02, first-listed diagnosis V57.x), and heart failure (first-listed diagnosis 398.91, 402.x1, 404.x1, 404.x3).²⁰ Variables obtained from HRAs included self-reported comorbidities (heart disease, high cholesterol, diabetes), obesity (body mass index >30 kg/m², calculated based on height and weight), and smoking status (“Do you currently smoke cigarettes?”) from the HRA. Expenditures were the sum of payments made by the primary insurer, other insurers, and patients. We inflated costs to 2012 dollars by using the Centers for Medicare & Medicaid Services’ Personal Health Care deflator.²²

Statistical Analysis

We estimated the likelihood of filling a prescription for antihypertensive medications (as indicated by a prescription drug claim), controlling for demographic characteristics (age, sex), whether the patient had a CVD-related event in the previous calendar year (equal to 1 if the patient had a CVD-related event and 0 otherwise), self-reported health risks (eg, current cigarette use), and year (2008-2012). We did not include data on race or ethnicity because they were not available. We estimated the likelihood of a CVD-related event and expenditures among the subsample of patients who filled a prescription for antihypertensive medications.

We used logistic regression analysis to model binary variables such as the use of antihypertensive medications and CVD-related events. We used a 2-part model to analyze CVD-related events to account for the fact that 98% of the patients in the sample did not have a CVD-related event. The first part of the model estimated the probability of having a CVD-related event by using logistic regression analysis. The second part used a generalized linear model with a gamma distribution and a log link to estimate expenditures. The sample for the second part was restricted to patients with nonzero CVD-related event expenditures. Consistent with previous applications of this method, we included the same set of independent variables in the first and second parts of the model.^23,24 The Box Cox test and generalized linear model family (Park) test supported the choice of model. We analyzed data by using Stata SE version 12.²⁵ Because coefficients from the 2-part model were not readily interpretable, we used the results to calculate then average marginal effects (or “predicted marginals”) using the Stata margins, dydx(*) and margins, at commands. The marginal effect was equal to the product of the first-stage prediction (ie, the probability of having a CVD-related event) and the second-stage prediction (ie, expenditures conditional on having ≥1 CVD-related event). We considered P < .05 to be significant.

Results

Of the 111 655 patients in the sample, 48 612 (43.5%) self-reported high blood pressure, 83 454 (74.7%) filled ≥1 prescription for antihypertensive medication, and 2389 (2.1%) had ≥1 CVD-related event in the same year in which they completed their HRA (Table 1). Patients who self-reported high blood pressure were more likely to be female, to self-report comorbidities, to not use cigarettes, and to fill a prescription for antihypertensive medication than patients who did not self-report high blood pressure.

Table 1.

Characteristics of patients aged 18-64 with an insurance claim that included an ICD-9 code for hypertension during 2007-2011^a and who completed a health risk assessment during the next calendar year (2008-2012),^b by whether or not they self-reported high blood pressure, United States

Characteristic	All,^c No. (%) (n = 111 655)^d	Did Not Report High Blood Pressure, No. (%) (n = 63 034)	Reported High Blood Pressure, No. (%) (n = 48 612)
Year of HRA completion
2008	18 399 (16.5)
2009	25 664 (23.0)
2010	15 586 (14.0)
2011	19 002 (17.0)
2012	33 004 (29.6)
Age, y
18-39	9970 (8.9)	5487 (8.7)	4483 (9.2)
40-54	52 897 (47.4)	28 641 (45.4)	24 256 (49.9)
55-64	48 788 (43.7)	28 906 (45.9)	19 882 (40.9)
Female	38 741 (34.7)	17 952 (28.5)	20 789 (42.8)
Comorbid chronic conditions
Diabetes^e	9544 (9.2)	3229 (5.5)	6315 (14.0)
High cholesterol^f	25 014 (25.3)	5486 (10.8)	19 528 (40.6)
Heart disease^g	3901 (4.0)	1333 (2.4)	2568 (6.0)
Risk factors
Obese (BMI >30 kg/m²)^h	37 800 (46.4)	15 267 (41.6)	22 533 (50.4)
Current cigarette useⁱ	13 543 (16.6)	9879 (24.4)	3664 (8.9)
CVD-related event in same year as HRA completion^j	2389 (2.1)	1602 (2.5)	787 (1.6)
CVD-related event in year before HRA completion^j	3000 (2.7)	2058 (3.3)	942 (1.9)
Filled ≥1 prescription for antihypertensive medication in same year as HRA completion	83 454 (74.7)	41 303 (65.5)	42 151 (86.7)

Abbreviations: BMI, body mass index; CVD, cardiovascular disease; HRA, health risk assessment; ICD-9, International Classification of Diseases, Ninth Revision.

^a Data source: MarketScan Commercial Database 2007-2012.¹⁷

^b MarketScan Health Risk Assessment Database 2008-2012.¹⁸

^c Patients with an insurance claim that included an ICD-9 code²⁰ for hypertension in the year before completing the HRA, 2007-2011. Variables are reported for the same year as HRA completion unless otherwise noted.

^d A total of 111 655 patients responded to the question about high blood pressure, of whom not all patients responded to questions about chronic conditions and risk factors.

^e A total of 103 855 patients in the sample were asked about diabetes, of whom 58 815 did not report high blood pressure and 45 040 did report high blood pressure.

^f A total of 99 004 patients in the sample were asked about high cholesterol, of whom 50 855 did not report high blood pressure and 48 149 did report high blood pressure.

^g A total of 98 659 patients in the sample were asked about heart disease, of whom 55 894 did not report high blood pressure and 42 765 did report high blood pressure.

^h A total of 81 401 patients in the sample were asked about height and weight, of whom 36 668 did not report high blood pressure and 44 733 did report high blood pressure.

ⁱ A total of 81 536 patients in the sample were asked about current cigarette use (yes response to the question, “Do you currently smoke cigarettes?”), of whom 40 379 did not report high blood pressure and 41 157 did report high blood pressure.

^j CVD-related emergency department visits and hospitalizations.

Patients who self-reported high blood pressure were significantly more likely than patients who did not self-report high blood pressure to fill a prescription for an antihypertensive medication (predicted probability = 26.5%; 95% confidence interval [CI], 25.7%-27.3%; P < .001; Table 2). The predicted probability of filling an antihypertensive medication prescription was also significantly higher for patients who were older (vs younger), had a CVD-related event in the same year as an HRA completion or in the year before (vs did not have a CVD-related event), and self-reported heart disease or diabetes (vs did not self-report heart disease or diabetes). Patients who self-reported high cholesterol and current cigarette smokers were less likely to fill a prescription for an antihypertensive medication than patients who did not report high cholesterol and did not smoke.

Table 2.

Use of antihypertensive medications, by characteristics and predicted probability of filling a prescription for antihypertensive medication, among patients with an insurance claim that included an ICD-9 code for hypertension during 2007-2011^a and who completed a health risk assessment during the year after diagnosis (2008-2012),^b United States

Characteristic	Total No. of Respondents^c (n = 111 655)	Patients Aged 18-64 Who Filled a Prescription for Antihypertensive Medication, No. (%)	Predicted Probability,^d % (95% CI)	P Value^e
Reported high blood pressure
No	63 034	41 303 (65.5)	1.0 [Reference]
Yes	48 612	42 151 (86.7)	26.5 (25.7 to 27.3)	<.001
Age, y
18-39	9970	5977 (60.0)	1.0 [Reference]
40-54	52 897	38 967 (73.7)	10.9 (9.6 to 12.2)	<.001
55-64	48 788	38 510 (78.0)	15.1 (13.8 to 16.4)	<.001
Sex
Male	72 914	53 816 (73.8)	1.0 [Reference]
Female	38 714	29 638 (76.5)	0.4 (–0.2 to 1.2)	.21
CVD-related event in year before HRA completion^f
No	109 266	81 248 (74.4)	1.0 [Reference]
Yes	2389	2206 (92.3)	7.8 (5.4 to 10.2)	<.001
CVD-related event in same year as HRA completion^f
No	108 655	80 793 (74.4)	1.0 [Reference]
Yes	3000	2662 (88.7)	12.4 (10.1 to 14.7)	<.001
Reported heart disease
No	94 758	70 742 (74.7)	1.0 [Reference]
Yes	3901	3474 (89.1)	10.8 (9.5 to 12.2)	<.001
Reported high cholesterol
No	73 990	54 103 (73.1)	1.0 [Reference]
Yes	25 014	19 971 (79.8)	–2.5 (–3.3 to –1.7)	<.001
Reported diabetes
No	94 311	69 612 (73.8)	1.0 [Reference]
Yes	9544	8071 (84.6)	6.7 (5.7 to 7.7)	<.001
Reported current cigarette use
No	67 993	51 261 (75.4)	1.0 [Reference]
Yes	13 543	10 017 (74.0)	–2.4 (–3.7 to –1.1)	<.001

Abbreviation: CVD, cardiovascular disease; HRA, health risk assessment; ICD-9, International Classification of Diseases, Ninth Revision.

^a Data source: MarketScan Commercial Database 2007-2012.¹⁷

^b MarketScan Health Risk Assessment Database 2008-2012.¹⁸

^d The predicted probability of filling a prescription for antihypertensive medications, as indicated by a prescription drug claim, relative to other category, generated by using the Stata margins command, dydx(*). Controls included all variables listed in the table. Estimates are adjusted for year using dummy variables.

^e P value based on results of the Wald χ² test, with P < .05 considered significant.

^f CVD-related emergency department visits and hospitalizations.

Patients who self-reported high blood pressure had a significantly lower predicted probability of having a CVD-related event than patients who did not self-report high blood pressure (–0.6; 95% CI, –0.9 to –0.3) (Table 3). Additionally, patients who adhered to antihypertensive medications had a lower probability of having a CVD-related event than patients who did not adhere to antihypertensive medications (–1.6; 95% CI, –1.9 to –1.3). Patients who had a CVD-related event in the year before HRA completion (vs did not have a CVD-related event in the year before HRA completion), self-reported heart disease (vs did not self-report heart disease), or were current cigarette smokers (vs not current cigarette smokers) were more likely to have a CVD-related event in the current year.

Table 3.

Predicted probability of a cardiovascular disease–related hospitalization or emergency department visit and marginal expenditures^a among patients with an insurance claim that included an ICD-9 code for hypertension^b during 2007-2011 who filled a prescription for antihypertensive medication^c and completed a health risk assessment in the year after diagnosis (2008-2012),^d United States

Characteristics	Predicted Probability,^e % (95% CI)	P Value^f	Marginal Expenditures, $,^a (95% CI)	P Value^f
Reported high blood pressure
No	1.0 [Reference]		1.0 [Reference]
Yes	–0.6 (–0.9 to –0.3)	<.001	–251 (–427 to –74)	.01
Antihypertensive medication adherence^g
1%-79%	1.0 [Reference]		1.0 [Reference]
≥80%	–1.6 (–1.9 to –1.3)	<.001	–610 (–769 to –451)	<.001
Age, y
18-39	1.0 [Reference]		1.0 [Reference]
40-54	1.0 (0.6 to 1.4)	<.001	358 (115 to 602)	.004
55-64	1.5 (1.1 to 1.9)	<.001	394 (159 to 629)	.001
Sex
Male	1.0 [Reference]		1.0 [Reference]
Female	–0.6 (–0.9 to –0.4)	<.001	–340 (–477 to –202)	<.001
CVD-related event in year before HRA completion^h
No	1.0 [Reference]		1.0 [Reference]
Yes	12.6 (10.4 to 14.9)	<.001	2597 (1800 to 3395)	<.001
Reported heart disease
No	1.0 [Reference]		1.0 [Reference]
Yes	4.9 (3.9 to 5.9)	<.001	1838 (1322 to 2353)	<.001
Reported high cholesterol
No	1.0 [Reference]		1.0 [Reference]
Yes	0.3 (–0.1 to 0.6)	.10	–9 (–168 to 149)	.91
Reported diabetes
No	1.0 [Reference]		1.0 [Reference]
Yes	0.4 (0.1 to 0.8)	.03	369 (137 to 602)	.002
Reported current cigarette use
No	1.0 [Reference]		1.0 [Reference]
Yes	0.8 (0.3 to 1.4)	.001	175 (–98 to 449)	.21

Abbreviations: CVD, cardiovascular disease; HRA, health risk assessment; ICD-9, International Classification of Diseases, Ninth Revision.

^a Average marginal expenditures capture the combined effect of the independent variables on the likelihood of having a CVD-related event (the first stage of the 2-part model) and expenditures conditional on having a CVD-related event (the second stage of the 2-part model). The estimates represent the difference in predicted expenditures between enrollees in a given group and enrollees in the reference group. For example, expenditures for enrollees who self-reported high blood pressure were estimated to be $251 lower than expenditures for enrollees who did not self-report high blood pressure.

^b Patients with an insurance claim that included an ICD-9 code²⁰ for hypertension in the year before completing the HRA, 2007-2011. Variables are reported for the same year as HRA completion unless otherwise noted.

^c Data source: MarketScan Commercial Database 2007-2012.¹⁷

^d MarketScan Health Risk Assessment Database 2008-2012.¹⁸

^e The predicated probability of a CVD event, relative to other category, generated by using the Stata margins command, dydx(*). Controls included all variables listed in the table. Estimates are adjusted for year using dummy variables.

^f P value based on results of the Wald χ² test, with P < .05 considered significant.

^g Measured by using proportion of days covered.

^h CVD-related emergency department visits and hospitalizations.

Predicted expenditures among patients who self-reported high blood pressure were $251 lower than predicted expenditures among patients who did not self-report high blood pressure and were $610 lower among patients who adhered to antihypertensive medications than among patients who did not adhere to antihypertensive medications (Table 3). Predicted expenditures were $2597 higher among patients who had a previous CVD-related event than among patients who had no previous CVD-related event and $1838 higher among patients who self-reported heart disease than among patients who did not self-report heart disease.

Sensitivity Analysis

In our sensitivity analysis, medication adherence was a significant predictor of CVD-related events and costs. Including medication adherence did not change the effect of self-reported high blood pressure in either stage of the 2-part model.

When we reestimated the model (not shown) by using various thresholds to define medication adherence, these modifications did not affect the predicted probability for self-reported high blood pressure. When we estimated models that included an interaction that allowed the effect of self-reported high blood pressure to vary by medication adherence, the effect did not differ significantly by level of medication adherence. Including patients with a diagnosis of hypertension regardless of whether they filled ≥1 prescription for antihypertensive medications in the sample did not substantially alter the results.

Patients who did not fill ≥1 prescription for antihypertensive medication were at a lower risk for a CVD-related event than patients who ever filled a prescription for antihypertensive medication. The sample included 138 patients who had been identified as having hypertension on the basis of only 1 diagnosis (and did not fill a prescription for antihypertensive medication) and may not have had hypertension. However, these patients represented fewer than half of 1% of patients who did not fill prescriptions for antihypertensive medication, and their inclusion did not explain this result.

We also looked at the timing of HRAs and CVD-related events in the year of HRA completion and excluded patients who had a CVD-related event before completing their HRA. We found that the levels of predicted probabilities of having a CVD-related event and expenditures on such events were lower overall than the results presented (a direct result of excluding patients with CVD-related events from the sample). The predicted probability of having a CVD-related event was 2.1% among patients who were unaware of their hypertension and 1.5% among patients who were aware of their hypertension. The difference in predicted expenditures was $249 more for patients who were unaware of their hypertension than for patients who were aware of their hypertension.

Discussion

We found that awareness of hypertension among patients with an insurance claim that included an ICD-9 code for hypertension and prescription medication claims for antihypertensive medications was associated with fewer adverse CVD-related events and lower medical costs than among patients who were treated but not aware. Previous studies have considered the relationship between knowledge of CVD risk factors and self-reported health behaviors. One study examined patient knowledge of all cardiovascular risk factors and found modest (<10%) relationships between knowledge of the influences of risk factors on CVD with patients’ self-reported engagement in healthy behaviors.¹⁵ Another study found that women with knowledge of heart disease as the leading cause of death were more likely to self-report engaging in actions such as additional physical activity and losing weight than women without this knowledge. However, the study did not find that women who perceived themselves to be at high risk of heart disease were more likely to engage in healthy behaviors than women who did not perceive themselves to be at high risk of heart disease.¹⁶ Both studies focused on knowledge related to CVD rather than awareness of one’s own health status, which was the focus of our study.

Our study adds to the literature by using a novel data source that links a measure of hypertension awareness with health insurance claims data to evaluate the effect of hypertension awareness on related health use and costs. These outcomes may be salient to policy makers who are evaluating interventions to improve patient self-management and engagement in their health care.

In our study, fewer than half of patients with an insurance claim that included an ICD-9 code for hypertension were aware that they had hypertension. This observation, combined with our results that suggest an association between lack of hypertension awareness and more CVD-related events, indicates that better physician-patient communication could improve health outcomes.

Simple interventions, such as an automated provider notification for uncontrolled blood pressure, may improve patients’ treatment-seeking behavior and medication adherence. Edwards²⁶ found that among Health and Retirement Survey respondents who did not know they had high blood pressure, receiving notification of high blood pressure after measurement raised the prevalence of self-reported diagnosis and medication usage by about 20 percentage points after 2 years.

Patients in our sample had access to health care and had used health care at least once in the previous year. Although access to health care is important in controlling hypertension,^27

-30 access alone is not sufficient to control blood pressure.^14,31
-33 We found that most patients in our sample had filled ≥1 prescription for antihypertensive medication (75%), but not all patients filled enough prescriptions to be adherent to their antihypertensive medications. Adherence to medication plays an important role in controlling hypertension and preventing CVD outcomes, and adherence can be improved.³⁴

Limitations

This study had several limitations. First, our sample relied on employed persons with employer-sponsored health insurance who completed an HRA. When comparing similar beneficiaries who did not complete an HRA, beneficiaries who completed an HRA were more likely to be active, full-time employees who were employed in manufacturing and less likely to be in the transportation, communications, or utilities industries. As such, our results may not be generalizable to patients without access to health care.

Second, our analysis did not consider whether the HRAs were completed before or after antihypertensive medication prescriptions were filled, because both could occur at any time during the course of a year. It is possible that patients may be more aware of their high blood pressure after filling their prescriptions than before and more likely to report high blood pressure in an HRA after filling their prescriptions than before filling their prescriptions.

Third, results were subject to confounding if patients who were aware of their hypertension differed from patients who were unaware of their hypertension based on characteristics that were unobserved and uncorrelated with the variables included in our regression model (eg, occupation). Our results suggest a plausible causal pathway based on the relationship between awareness and medication adherence. In addition, lack of awareness of hypertension may be associated with other health behaviors, such as diet.

Fourth, we could not identify patients with uncontrolled hypertension. Patients who had hypertension but did not self-report it may have believed that their response was correct if their hypertension was controlled. If that were the case, we would expect patients in this group to have fewer CVD-related events. However, our results suggest that patients who did not report high blood pressure were either not aware of their hypertension or mistakenly believed their blood pressure was under control, because these patients were more likely to have a CVD-related event than patients who self-reported high blood pressure.

Finally, although our study measured CVD-related events during a calendar year, the time period was relatively short. The difference in outcomes based on awareness of hypertension would likely grow larger during a longer follow-up period.

Conclusions

Many patients in private health plans who were unaware that they had hypertension but were filling prescriptions for antihypertensive medications were more likely to have a CVD-related event and incurred higher medical costs than patients who were unaware that they had hypertension but were filling prescriptions for antihypertensive medications. Our results suggest that interventions that improve patients’ awareness of their hypertension may improve antihypertensive medication use and reduce CVD-related events.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors declared the following financial support with respect to the research, authorship, and/or publication of this article: This research was supported by the Centers for Disease Control and Prevention (CDC). The findings and conclusions in this article are those of the authors and do not represent the official position of CDC.

ORCID iD

Madeleine M. Baker-Goering, PhD

References

Mozaffarian

Roger

et al. Heart disease and stroke statistics—2013 update: a report from the American Heart Association [published correction appears in Circulation. 2013;127(33):e841]. Circulation. 2013;127(1):e6–e245. doi:10.1161/CIR.0b013e31828124ad

Valderrama

Gillespie

King

George

Hong

Gregg

. Vital signs: awareness and treatment of uncontrolled hypertension among adults—United States, 2003-2010. MMWR Morb Mortal Wkly Rep. 2012;61(35):703–709.

Alexander

Gordon

Davis

Chen

. Patient knowledge and awareness of hypertension is suboptimal: results from a large health maintenance organization. J Clin Hypertens (Greenwich). 2003;5(4):254–260.

Safeer

Cooke

Keenan

. The impact of health literacy on cardiovascular disease. Vasc Health Risk Manag. 2006;2(4):457–464.

Gazmararian

Baker

Williams

et al. Health literacy among Medicare enrollees in a managed care organization. JAMA. 1999;281(6):545–551. doi:10.1001/jama.281.6.545

Joyner-Grantham

Mount

McCorkle

Simmons

Ferrario

Cline

. Self-reported influences of hopelessness, health literacy, lifestyle action, and patient inertia on blood pressure control in a hypertensive emergency department population. Am J Med Sci. 2009;338(5):368–372. doi:10.1097/MAJ.0b013e3181b473 dc

Williams

Baker

Parker

Nurss

. Relationship of functional health literacy to patients’ knowledge of their chronic disease: a study of patients with hypertension and diabetes. Arch Intern Med. 1998;158(2):166–172.

Cooper

Roter

Carson

et al. A randomized trial to improve patient-centered care and hypertension control in underserved primary care patients. J Gen Intern Med. 2011;26(11):1297–1304. doi:10.1007/s11606-011-1794-6

Howard

Jacobson

Kripalani

. Doctor talk: physicians’ use of clear verbal communication. J Health Commun. 2013;18(8):991–1001. doi:10.1080/10810730.2012.757398

10.

Chowdhury

Khan

Heydon

et al. Adherence to cardiovascular therapy: a meta-analysis of prevalence and clinical consequences. Eur Heart J. 2013;34(38):2940–2948. doi:1093/eurheartj/eht295

11.

Bosworth

Granger

Mendys

et al. Medication adherence: a call for action. Am Heart J. 2011;162(3):412–424. doi:10.1016/j.ahj.2011.06.007

12.

Hyre

Krousel-Wood

Muntner

Kawasaki

DeSalvo

. Prevalence and predictors of poor antihypertensive medication adherence in an urban health clinic setting. J Clin Hypertens (Greenwich). 2007;9(3):179–186.

13.

Naderi

Bestwick

Wald

. Adherence to drugs that prevent cardiovascular disease: meta-analysis on 376,162 patients. Am J Med. 2012;125(9):882–887. doi:10.1016/j.amjmed.2011.12.013

14.

Stockwell

Madhavan

Cohen

Gibson

Alderman

. The determinants of hypertension awareness, treatment, and control in an insured population. Am J Public Health. 1994;84(11):1768–1774.

15.

Alzaman

Wartak

Friderici

Rothberg

. Effect of patients’ awareness of CVD risk factors on health-related behaviors. South Med J. 2013;106(11):606–609. doi:10.1087/SMJ.0000000000000013

16.

Mosca

Mochari

Christian

et al. National study of women’s awareness, preventive action, and barriers to cardiovascular health. Circulation. 2006;113(4):525–534. doi:10.1161/CIRCULATIONAHA.105.588103

17.

IBM Watson Health. IBM MarketScan Commercial Database 2007-2012. 2014. https://www.ibm.com/us-en/marketplace/marketscan-research-databases. Accessed April 2, 2019.

18.

IBM Watson Health. IBM MarketScan Health Risk Assessment Database 2008-2012. 2014. https://www.ibm.com/downloads/cas/JMPKEDA5. Accessed April 2, 2019.

19.

DeNavas-Walt

Proctor

Smith

. Income, Poverty, and Health Insurance Coverage in the United States: 2012. Current Population Reports, P60-245. Washington, DC: US Government Printing Office; 2013.

20.

Choudhry

Shrank

Levin

et al. Measuring concurrent adherence to multiple related medications. Am J Manag Care. 2009;15(7):457–464.

21.

Centers for Medicare & Medicaid Services. CMS Chronic Conditions Data Warehouse condition categories. https://www.ccwdata.org/documents/10280/19139421/original-ccw-chronic-condition-algorithms.pdf. Accessed December 19, 2018.

22.

Office of the Actuary, Centers for Medicare & Medicaid Services. Table 23 national health expenditures; nominal dollars, real dollars, price indexes and annual percent change, selected calendar years 1980-2012. 2013. https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/NationalHealthExpendData/Downloads/Tables.zip. Accessed November 30, 2018.

23.

Duan

Manning

Jr Morris

Newhouse

. A comparison of alternative models for the demand for medical care. J Bus Econ Stat. 1983;1(2):115–126.

24.

Deb

Trivedi

. The structure of demand for health care: latent class versus two-part models. J Health Econ. 2002;21(4):601–625. doi:10.1016/S0167-6296(02)00008-5

25.

Stata [computer program]. Release 12. College Station, TX: StataCorp LP; 2011.

26.

Edwards

. If my blood pressure is high, do I take it to heart? Behavioral impacts of biomarker collection in the Health and Retirement Study. Demography. 2018;55(2):403–434.

27.

Moy

Bartman

Weir

. Access to hypertensive care: effects of income, insurance, and source of care. Arch Intern Med. 1995;155(14):1497–1502.

28.

Muntner

Chen

Roccella

Streiffer

Whelton

. Factors associated with hypertension control in the general population of the United States. Arch Intern Med. 2002;162(9):1051–1058.

29.

Lurie

Ward

Shapiro

Gallego

Vaghaiwalla

Brook

. Termination of Medi-Cal benefits: a follow-up study one year later. N Engl J Med. 1986;314(19):1266–1268. doi:10.1056/NEJM198605083141934

30.

Fihn

Wicher

. Withdrawing routine outpatient medical services: effects on access and health. J Gen Intern Med. 1988;3(4):356–362.

31.

Berlowitz

Ash

Hickey

et al. Inadequate management of blood pressure in a hypertensive population. N Engl J Med. 1998;339(27):1957–1963. doi:10.1056/NEJM199812313392701

32.

Hyman

Pavlik

. Characteristics of patients with uncontrolled hypertension in the United States [published correction appears in N Engl J Med. 2002;346(7):544]. N Engl J Med. 2001;345(7):479–486. doi:10.1056/NEJMoa10273

33.

Wang

Vasan

. Epidemiology of uncontrolled hypertension in the United States. Circulation. 2005;112(11):1651–1662. doi:10.1161/CIRCULATIONAHA.104.490599

34.

Will

Zhang

Ritchey

Loustalot

. Medication adherence and incident preventable hospitalizations for hypertension. Am J Prev Med. 2016;50(4):489–499. doi:10.1016/j.amepre.2015.08.021