Trends in Prevalence, Awareness, Treatment, and Control of Hypertension in the United States, 1988-2000 FREE

Hypertension is a major risk factor for cardiovascular disease morbidity and mortality.^1- 3 The National Health and Nutrition Examination Survey (NHANES), conducted by the National Center for Health Statistics, provides periodic information on the health of the US population. Based on earlier analyses of NHANES data, with the possible exception of black men aged 50 years or older, hypertension prevalence has decreased between 1960 and 1991.⁴ In addition, during this time, the entire population-based blood pressure (BP) distribution shifted downward.⁵

In apparent contrast with these favorable trends, according to the Behavioral Risk Factor Surveillance System (BRFSS), self-reported hypertension was 24.9% in 1999 compared with 22.9% in 1991.⁶ This increase may represent an increase in awareness, but because the BRFSS does not include BP measurements, the current hypertension prevalence rate in the community is not known.

Despite the overwhelming evidence that hypertension control is associated with a significant reduction in cardiovascular events,^7- 9 less than 25% of the US population with hypertension had their BP controlled in 1988-1991.¹⁰ The goal of the US Department of Health and Human Services that 50% of Americans with hypertension have their BP controlled by the year 2000 was not met, and this goal has been reestablished to be achieved by 2010.^11- 12 Recent trends in hypertension control in the US population are not known.

We analyzed NHANES data to determine trends in the prevalence, awareness, treatment, and control of hypertension in the United States between 1988 and 2000.

NHANES is a stratified multistage probability sample of the civilian noninstitutionalized US population. The third NHANES was conducted in 2 phases between 1988 and 1994; the first phase in 1988-1991 and the second phase in 1991-1994.¹³ The most recent NHANES was conducted in 1999 and 2000.¹² Although each of the surveys has an overall similar design, the 1999-2000 survey has a smaller sample size and is the first part of the continuous NHANES.¹⁴ This may lead to a higher SE for variables measured in the 1999-2000 survey.¹⁵ All participants provided informed consent and the data were approved by the Centers for Disease Control and Prevention Institutional Review Board as ensuring confidentiality.¹⁴ For this analysis, we included data on all participants aged 18 years or older. This includes age, sex, and self-assigned race/ethnicity (non-Hispanic whites, non-Hispanic blacks, Mexican Americans, and other racial groups, including non–Mexican American Hispanic whites and blacks). In addition, body mass index (BMI; calculated as weight in kilograms divided by the square of height in meters) and BP measurements and hypertension and diabetes mellitus history were included.

A certified operator performed BP measurements using a mercury sphygmomanometer and a standardized procedure in each survey during the home interview and during the physical examination at home or at a mobile examination center. A cuff size appropriate for the participant's arm circumference was used. Quality control was ensured by operator recertification, procedural checklists, and data review.^16- 17 In the 1999-2000 phase, 9.7% had fewer than 3 BP readings compared with 0.5% in the 1991-1994 phase and 0.4% in the 1988-1991 phase (P<.01). A small (<1%) and similar number of participants in all 3 phases reported that they had never had a BP measurement. Those interviewed but not examined were not included in the analysis (n = 1076 in 1988-1991, n = 790 in 1991-1994, and n = 472 in 1999-2000).

Each survey used similar questions for hypertension history. We considered a participant to be hypertensive if the average systolic BP (SBP) was 140 mm Hg or higher or the average diastolic BP (DBP) was 90 mm Hg or higher or if the participant was currently receiving antihypertensive medications. The same criteria were used for both diabetic and nondiabetic participants. Participants were considered to be aware of their hypertension if they answered "yes" to the interview question "Have you ever been told by a doctor or health professional that you had hypertension, also called high blood pressure?" Participants were considered to be treated if they answered "yes" to the interview question "Because of your high blood pressure/hypertension are you now taking prescribed medicine?" Participants were considered to have their hypertension controlled if they were receiving antihypertensive medications and their average SBP was less than 140 mm Hg and average DBP was less than 90 mm Hg. A separate control variable was defined for diabetic participants. Participants were considered diabetic if they answered "yes" to at least 1 of these interview questions: "Have you ever been told by a doctor that you have diabetes or sugar diabetes?" or "Are you now taking insulin?" or "Are you now taking diabetes pills to lower blood sugar?" Diabetic participants were considered to have controlled hypertension if they were receiving antihypertensive medication and their mean BP was below the recommended target. Because of the change in recommendations between 1988 and 2000 to a lower target BP in diabetic individuals, we defined control in those with diabetes based on 2 criteria for BP: a mean SBP of less than 140 mm Hg and a mean DBP of less than 90 mm Hg, and a mean SBP of less than 130 mm Hg and a mean DBP of less than 85 mm Hg.¹⁸

Each survey oversampled African Americans, Mexican Americans, and individuals aged 60 years or older.^14,19 To adjust for oversampling, special sampling weights were included in our analysis. These weights also adjusted for nonresponse bias and poststratification population totals.²⁰ To obtain estimates of the sampling errors, a Taylor linear approximation method and jackknife 1 procedures were used in each of the 3 surveys.^19,21- 22 In consideration of the complex sample design, SAS version 8.02 (SURVEYMEANS, SURVEYREG)²³ and WesVar version 4.2 (Westat Inc, Rockville, Md) software were used. For statistical analyses of differences over time, data from the 1999-2000 survey were compared with 1988-1991 data rather than the entire NHANES III survey to provide more comparable sample sizes. The test for a linear trend over the 3 phases of the NHANES survey is the same as for a comparison of the first and last surveys. In accordance with the Joint Policy on Variance Estimation, estimates with a coefficient of variation larger than 0.30 or a sample size smaller than the recommended size for the design effect or the estimated proportion were considered unreliable.^15,24 Hypothesis testing was not performed on unreliable estimates. Age adjustment and population estimates were calculated using the standard 2000 US population estimates, with a population breakdown by 10-year increments.²⁵ The following age groups were used in this analysis in accordance with the 1999-2000 survey's analytic guidelines: 18-39 years, 40-59 years, and 60 years or older.¹⁵ Tests for trend were performed using the t test for comparisons of weighted means (to adjust for sampling weights) between the 1988-1991 and 1999-2000 data. We used a conservative approach for statistical significance (α = .01) to help account for multiple comparisons. Multiple regression analysis was performed with hypertension as the outcome variable and age, sex, race/ethnicity, and BMI as predictors to calculate the independent association between the predictors and the outcome variables. This analysis was performed separately for each of the 3 phases. Results were nonstandardized. We used SAS SURVEYREG and WesVar to account for survey design and sampling weights. The increase in hypertension prevalence attributable to BMI was determined by running the regression model and performing analysis of covariance to determine the independent associations between hypertension and BMI.

Table 1 provides the sample sizes by sex, race/ethnicity, and age. Average age of the participants, sex, racial/ethnic distribution, and mean SBP and DBP were similar in each of the 3 phases (Table 2.) Participants in 1999-2000 had a higher BMI (P<.001) and a higher prevalence of diabetes than in 1988-1991 (P<.001).

In 1999-2000, the age-adjusted hypertension prevalence was 28.7% (Table 3), an absolute increase of 3.7% compared with 1988-1991. Non-Hispanic blacks had the highest prevalence of hypertension (P = .04 for 1988-1991, P = .001 for 1991-1994, and P = .01 for 1999-2000) and Mexican Americans had the lowest rates (P<.001 for all 3) in all 3 surveys, and rates for all racial/ethnic groups tended to increase between 1988 and 2000. Hypertension prevalence increased to a greater extent in individuals aged 60 years or older than in younger individuals. Non-Hispanic black women had the greatest increase in hypertension prevalence and non-Hispanic white men had the smallest (7.2% of non-Hispanic black women vs 1.0% of non-Hispanic white men). Hypertension prevalence in all age, sex, and race/ethnicity groups is shown in Figure 1. Based on the 1999-2000 survey, among both men and women, hypertension prevalence in those aged 40 to 59 years and those aged 60 years or older was higher in non-Hispanic blacks than in non-Hispanic whites and Mexican Americans.

A multiple regression analysis was carried out to assess the independent association of hypertension prevalence with demographics and BMI, considered simultaneously (Table 4). In each of the 3 surveys, hypertension was associated with increasing age (P<.001), non-Hispanic black race/ethnicity (P<.001), and higher BMI (P<.001). Hypertension was not associated with sex (P = .37) in the regression analysis. In all 3 surveys, demographic characteristics and BMI accounted for 30% to 31% (age accounted for 26%, non-Hispanic black race/ethnicity for 0.4%, and BMI for 5%) of the hypertension prevalence.

To address the possibility that the increase in the age-adjusted hypertension prevalence could be secondary to the increase in BMI from 1988-1991 to 1999-2000, we calculated the contribution of this increase in BMI to the increase in hypertension prevalence. Adjusting the 1988-1991 hypertension prevalence for the differences between 1988-1991 and 1999-2000 covariates using analysis of covariance, 2.0% of the 3.6% increase of hypertension prevalence was attributable to an increase in BMI.

Because the recommended goal for hypertension control has been changed to less than 130/85 mm Hg in those with diabetes,¹⁸ we considered the possibility that the increase in prevalence in the 1999-2000 phase may be due to initiation of antihypertensive therapy at lower BP levels, in accord with this new recommendation. A separate analysis of age-adjusted prevalence excluding those with diabetes demonstrated an increase in hypertension prevalence between 1988 and 2000, but this increase was no longer significant (24.3 [SE, 2.0] in 1988-1991 vs 27.8 [SE, 1.9] in 1999-2000; P = .10).

In 1999-2000, 68.9% of all hypertensive participants were aware of their hypertension, and 58.4% were receiving BP medication (Table 5). Hypertension was controlled in 53.1% of those taking medication, and among all hypertensive participants, BP was controlled in only 31.0%. Awareness remained unchanged from 1988 to 2000, although treatment and overall control increased by 6.0% (P = .007) and 6.4% (P = .004), respectively. Among participants with diabetes, control rates declined but not significantly, possibly because of the small number of hypertensive individuals with diabetes. Based on the BP criterion of less than 140/90 mm Hg, 46.9% of all patients with hypertension and diabetes had their hypertension controlled in 1999-2000 compared with 53.1% in 1988-1991. Control rates were lower for participants with diabetes, based on the BP criterion of less than 130/85 mm Hg.

We then evaluated the separate effects of sex, race/ethnicity, and age on hypertension awareness, treatment, and control (Table 6). Between 1988 and 2000, hypertension awareness rates remained stable in both men and women. Hypertension treatment and control rates increased in men (P<.001 for both) but did not change significantly in women. In the 1999-2000 survey, hypertension awareness was similar between men and women (P = .12), whereas treatment rates were higher, although nonsignificantly (P = .03), and control rates in those being treated were lower (P = .006) in women than in men.

Hypertension awareness rates were unchanged and treatment rates improved in all 3 racial/ethnic groups between 1988 and 2000, but the increase was only significant in non-Hispanic whites (non-Hispanic whites, P = .009; non-Hispanic blacks, P = .02; and Mexican Americans, P = .05). Hypertension control rates improved in non-Hispanic whites (P = .01 for treated hypertension and P = .002 for all hypertension); non-Hispanic white men demonstrated the highest increase in control rates (P<.001) (Table 6). Control rates did not change significantly for Mexican Americans between 1988 and 2000. In the 1999-2000 survey, hypertension awareness (P = .005), treatment (P<.001), and overall control rates (P<.001) were lower in Mexican Americans than in non-Hispanic whites and non-Hispanic blacks. Furthermore, rates of hypertension control among those being treated were lower in non-Hispanic blacks than in non-Hispanic whites (P = .02) (Table 5).

Hypertension awareness did not change significantly from 1988 to 2000 in any age group (Table 7). Treatment (P = .01) and control (P = .006 for treated and P<.001 for all) improved significantly in those aged 40 to 59 years between 1988 and 2000, whereas only overall control rates improved in those aged 60 years or older (P = .02). Despite similar hypertension awareness and treatment rates, control rates were lower during each of the 3 surveys in participants aged 60 years or older than in those aged 40 to 59 years (P = .03 for 1988-1991 and P<.001 for 1999-2000) (Table 7). Hypertension awareness, treatment, and control rates tended to be lowest in participants aged 18 to 39 years in each of the 3 surveys. However, because of the low hypertension prevalence and the high relative SE in this age group, these observations must be interpreted cautiously.

Figure 2 depicts overall hypertension control rates by age group (excluding those ≤39 years) and by race/ethnicity for men and women in the 1999-2000 survey. For both sexes and both age groups, Mexican Americans had the lowest rates of hypertension control. In those aged 40 to 59 years, non-Hispanic black men and women had lower rates of hypertension control than non-Hispanic whites. However, in those aged 60 years or older, hypertension control rates were similar in non-Hispanic white and non-Hispanic black men, whereas the rate of hypertension control was lower in non-Hispanic white than in non-Hispanic black women aged 60 years or older.

This analysis indicates that almost 29% of the adult US population, an estimated 58.4 million individuals, had hypertension in 1999-2000. This is an overall increase from 1988-1991 and demonstrates a reversal of the previously reported trend of an overall decline in hypertension prevalence between 1960 and 1991 using the NHANES survey.⁴ It also shows that women, non-Hispanic blacks, and the older population have the highest rates of hypertension. In 1999-2000, almost 30% of all hypertensive individuals were unaware of their illness, 42% were not being treated, and, at the time that their BP was measured, 69% did not have their hypertension controlled. Women, older participants, and Mexican Americans tended to have the lowest rates of control. In addition, in 1999-2000, nearly 75% of all patients with diabetes and hypertension did not have their hypertension controlled to the BP target of less than 130/85 mm Hg recommended in 1997.¹⁸

The observed increase in hypertension prevalence during the last 10 years is in accordance with results from the BRFSS.⁶ However, as expected, the prevalence based on NHANES data is higher because BP measurement was not performed in the BRFSS survey. Our data suggest that the increased prevalence is not explained by an increased awareness of hypertension. Potentially, the increased prevalence may be related to an increase in BMI as well as the aging of the US population. Based on an analysis of NHANES data, Flegal et al²⁶ recently reported an increase in prevalence of obesity during this same period. Our analysis indicates that BMI was associated with higher hypertension prevalence after adjusting for age, sex, and race/ethnicity and contributed to 2%—more than half—of the 3.6% increase in hypertension prevalence. Therefore, the increased prevalence of hypertension between 1988-1991 and 1999-2000 appears to be partially related to but not solely secondary to the increase in BMI.

Although awareness rates remained unchanged, treatment and control rates have each increased by approximately 6% between 1988 and 2000. In the latest survey, treatment and control rates of 58% and 31% are comparable with rates observed in the Framingham data between 1990 and 1995, in which 60.7% of hypertensive patients were treated and 29% had their hypertension controlled.²⁷ Furthermore, our analysis shows that, compared with younger individuals with hypertension, older individuals with hypertension have a lower rate of control despite being equally likely to be treated. Although previous studies have also shown that hypertension is less likely to be controlled in older patients,^27- 30 this analysis demonstrates that since 1988 the control rate in the elderly population has minimally increased and remains lower than in those aged 40 to 59 years. We also observed that Mexican Americans had lower rates of awareness, treatment, and control than non-Hispanic whites and non-Hispanic blacks. This pattern persisted between 1988 and 2000.

Women had a greater increase in hypertension prevalence than men from 1988 to 2000, and in the 1999-2000 survey, hypertension prevalence was higher in women aged 60 years or older than in any other age or sex group. Other surveys have suggested higher prevalence of hypertension in elderly women. For example, in the National Health Interview Survey, 38% of women 65 to 74 years old reported hypertension compared with 31% of men in the same age group.³¹ In the current analysis, despite the increase in hypertension prevalence in women, awareness, treatment, and control rates remained unchanged from 1988 to 2000.

Lowering BP in patients with diabetes and hypertension is associated with decreases in cardiovascular events and renal failure.^32- 33 Our analysis demonstrates that in 1999-2000, only one quarter of diabetic participants receiving antihypertensive drugs had their hypertension controlled to the recommended BP of less than 130/85 mm Hg. In addition, between 1988 and 2000, the hypertension control rate tended to decline in diabetic participants based on each of the 2 different BP target criteria, although these changes did not reach statistical significance. With the increased prevalence rate of diabetes mellitus observed in this analysis, uncontrolled hypertension with diabetes is an important health problem in the United States.

The low control rates observed in this analysis can be explained in part by lack of awareness or treatment. However, even in those who reported that they were receiving hypertension medications, only 53% had hypertension controlled to the recommended target. In addition, recent analysis of the Framingham Heart Study demonstrated that control was only 33.7% at follow-up in a cohort of participants with hypertension that was treated and uncontrolled at baseline.³⁴ Conceivably, different health insurance coverage may affect different age-related rates of hypertension awareness and control and the effect of age on race/ethnicity and sex differences. However, despite universal health insurance (Medicare) for those aged 65 years or older, hypertension control rates were lowest among individuals aged 60 years or older. Other reports have also noted low hypertension control rates in the elderly,²⁸ in patients treated at a Veterans Administration hospital,³⁵ and in inner-city African Americans,³⁶ despite availability and access to medical care. These observations suggest that improving the health care system to better manage hypertension may improve BP control in the United States. In addition, improving hypertension awareness should result in improved treatment and control rates.

Although measurements obtained at a single point in time may overestimate the prevalence of hypertension, this potential problem was minimized by taking the average of 3 separate measurements obtained under the same standardized conditions in each of the surveys. Other potential sources of bias in the reported estimates of hypertension prevalence include possible BP control by nonpharmacological means (diet and exercise), failure to sample homeless or other unnumerated persons, possible differential response rates, and lack of estimates for smaller racial/ethnic groups. Also, this analysis depends on a national survey that documents self-reported medical history. Although this method has been shown to be accurate in estimating hypertension prevalence,³⁷ limitations exist. Participants who report inaccurately that they are not receiving antihypertensive medications or who cannot recall if their physician has ever told them that they have hypertension are considered nonhypertensive or unaware of their disease. Also, participants who have been told that they have hypertension but are not receiving medications and have BP below the 140/90 mm Hg cut points are not counted as hypertensive. In addition, in this survey, medication use or type was not confirmed, which can lead to misclassification. For example, participants may have been using medications that lower BP for another indication.

Despite progress in treatment and control, hypertension remains an important public health problem, and our analysis of the NHANES data has significant public health implications. The increase in hypertension prevalence in women, non-Hispanic blacks, and older individuals highlights the need for interventions that would target prevention in these groups. Even greater strides will need to be made, as the importance of reducing BP with healthy lifestyle interventions in prehypertensive individuals in the general population (BP of 120-129/80-89 mm Hg) has recently been emphasized, and the target BP goal for antihypertensive therapy in diabetic individuals has been lowered to less than 130/80 mm Hg.³⁸ In addition, improving hypertension treatment and control in women, the older population, Mexican Americans, and diabetic individuals will have a significant impact on overall control rates, which would be translated into improved cardiovascular outcomes. Although control rates have improved since 1988, these rates remain unacceptably low. If the increase in hypertension control rates remains at the current pace, the 50% target for hypertension control by 2010 will not be met. Programs targeting hypertension prevention to achieve the 16% target for hypertension prevalence by 2010 and improving awareness and treatment are of utmost importance for the health of the US population.