Prehypertension and Insulin Resistance in a Nationally Representative Adult Population

Abstract

Prehypertension is associated with increased risk of cardiovascular disease and progression to hypertension. Insulin resistance (IR) is also related to cardiovascular risk. It is unknown whether individuals with prehypertension also have higher IR. The purpose of this study was to examine the association between prehypertension and IR. The National Health and Nutrition Examination Survey 1999–2002 was used to determine odds of IR by fasting insulin level >12.2 μU/mL or homeostasis model assessment (HOMA) ≥2.6 among nondiabetic adults aged 20 to 80 years across blood pressure categories. Compared with normotensives, odds of IR were over 60% higher for prehypertensive individuals by both IR measures: fasting insulin (odds ratio [OR], 1.67; 95% confidence interval [CI], 1.12–2.48) and HOMA (OR, 1.67; 95% CI, 1.22–2.30). After stratifying by sex, IR was associated with prehypertension in only men for both IR measures. In conclusion, prehypertension is associated with higher IR in men, which may confer additional cardiovascular disease risk.

In 2003, the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC) issued its seventh report on blood pressure and treatment guidelines. In addition to the various hypertension categories, JNC introduced a new blood pressure classification for prehypertension. ¹ This new category combined normal and high normal categories from previous JNC reports and is defined as a systolic blood pressure (SBP) of 120 to 139 mm Hg or a diastolic blood pressure (DBP) of 80 to 90 mm Hg. One prevalence study in 2005 ² estimated that 27.8 million women and 41.9 million men in the United States have prehypertension. Cardiovascular risks associated with hypertension are well established, ³ , ⁴ but those associated with prehypertension are relatively unknown given that it is a fairly new classification. Evidence suggests that persons with prehypertension have greater overall cardiovascular disease (CVD) as well as more risk factors, including elevated cholesterol and obesity, compared with normotensive patients. ⁵ , ⁶

Insulin resistance (IR) has become an area of research interest as a marker of early CVD. IR relates to impairment of the physiologic response of tissues to insulin leading to disruption of various metabolic processes. Patients with IR are more likely to progress to various types of CVD including hypertension, coronary heart disease, and type 2 diabetes. ⁷ , ⁸ Likewise, individuals with prehypertension are also more likely to progress to hypertension. ⁹ Of interest, Greenlund and colleagues ¹⁰ showed that unlike patients with hypertension, those with prehypertension did not have a significantly higher incidence of type 2 diabetes than did normotensive patients. It is unknown, however, whether individuals with prehypertension have increased IR. A prospective study in Spain showed a link between metabolic syndrome and surrogate markers of IR such as triglyceride to high‐density lipoprotein ratio, ¹¹ but no study has attempted to investigate whether established markers of IR including fasting insulin level and homeostatic model assessment (HOMA) are independently associated with prehypertension.

Given that much still remains unknown about the cardiovascular risks associated with prehypertension and the implications of the diagnosis, further characterization and risk stratification of these patients are needed. Thus, the purpose of this study is to evaluate the association between IR as measured by fasting insulin level and HOMA with normotensive, prehypertensive, and hypertensive individuals in a nationally representative adult population.

METHODS

Survey Description

We analyzed data from the 1999–2002 National Health and Nutrition Examination Survey (NHANES). The NHANES 1999–2002 is a product of the National Center for Health Statistics, and it consists of detailed household interviews and physical and laboratory examinations. It is a continuous annual survey that involves participants from a nationally representative sample of noninstitutionalized residents of the United States. Certain groups such as African Americans, Mexican Americans, and older persons are oversampled to ensure adequate numbers for subgroup analyses. Samples are weighted so that they are representative of the US population. Unequal probabilities of selection due to sample design and oversampling were taken into account when calculating sample weights, which were then matched to known population control totals to be representative of the US population. Descriptions of the NHANES design and sampling methods are available from the National Center for Health Statistics. ¹² , ¹³

Subsample Population

The subpopulation analyzed included adult men and women aged 20 to 80 years in NHANES 1999–2002. Participants with a fasting insulin and fasting glucose levels were selected. Because we were interested in looking at IR as a prediabetic state and as a risk factor for later disease, we excluded any patients with a history of diabetes because they are known to already have IR. We defined diabetes as either ever having been told by a doctor that the participant has diabetes or sugar diabetes or currently taking medications for diabetes. IR was defined by the established marker of fasting insulin level >12.2 μU/mL, which was established by correlation with the hyperinsulinemic euglycemic clamp method ¹⁴ and is considered the gold standard for determining IR. We also used the homeostasis model assessment for IR (HOMA‐IR). HOMA‐IR is calculated by multiplying fasting serum insulin level (μU/mL) and fasting plasma glucose level (mg/dL) and then dividing by 405. ¹⁵ Standardized cutoff values for HOMA‐IR have been determined at the 75th percentile in only one study. ¹⁶ However, this level of ≥2.6 corresponds to a fasting insulin level representative of IR established by the hyperinsulinemic euglycemic clamp method.

Blood Pressure Variables

Blood pressure was measured during the NHANES examination section by trained technicians. Details of blood pressure‐measuring protocol can be found on the National Center for Health Statistics Web site. ¹³ Four measurements of SBP and 4 measurements of DBP were taken, and then an average of each was reported to the participants. The reported average was used in this analysis. For this analysis, 3 blood pressure categories were designated for each participant: hypertension, prehypertension, and normotension. Persons assigned to the hypertension category were defined as those who reported having ever been told by a doctor that they have hypertension or high blood pressure, those on medications for blood pressure (by stating that they are taking medications or on an antihypertensive in the medication inventory part of the exam), and those with a SBP ≥140 mm Hg or DBP ≥90 mm Hg. Participants were designated as having prehypertension if criteria for hypertension were not met and they had SBP of 120–139 mm Hg or DBP of 80–89 mm Hg. Finally, those with SBP <120 mm Hg and DBP <80 mm Hg who did not meet criteria for hypertension or prehypertension were considered to be normotensive.

Covariates

Several potential confounding variables were assessed. For race/ethnicity, participants were categorized into white, black, Mexican American, other Hispanic, and other, including multiracial, based on patient self‐report. Regular vs sedentary physical activity was defined as moderate or vigorous activity in the last 30 days, also based on participant self‐report. Family history of early CVD in an individual's parents was also included, defined as having a mother or father with myocardial infarction or angina before age 50. Smoking status was categorized as current smoker or current nonsmoker.

Age in years, body mass index (BMI; kg/m²), and average daily carbohydrate (g) intake (g) were each categorized for univariate analysis and included as continuous variables in regression modeling. BMI was calculated in NHANES from the measured weight and height of the participants. Dietary information was derived from a 24‐hour dietary recall obtained using a multiple‐pass method.

Data Analyses

Descriptive statistics and regression modeling of the sample population were conducted using SUDAAN 9.0.1 software (Research Triangle Institute, Research Triangle Park, NC) to account for the complex sampling methods of the NHANES data. Sample size and weighted population estimates are given for the total sample and for each blood pressure category. Simple logistic regression was performed for each IR factor (fasting insulin and HOMA) and compared across each blood pressure category. Multivariable logistic regression models predicting odds of IR were also performed controlling for sex, race, exercise level, smoking status, and family history of early CVD as categoric variables and age, BMI, and carbohydrate intake as continuous variables. Results were also stratified by sex. Results for all regression models are presented as odds ratios (OR) with 95% confidence intervals (CI), with normotension as the referent category.

RESULTS

Characteristics of the sample are presented in Table I. Briefly, there are 3203 unweighted individuals in the subpopulation representing almost 164 million people. Of those, 39.8% were normotensive, 25.8% prehypertensive and 34.4% hypertensive. Approximately 52% were women and 48% were men. The distribution of IR among normotensive individuals was 18.9% by fasting insulin level and 23.6% by HOMA and for prehypertensive individuals, 32.6% and 40.5%, respectively. IR in hypertensive individuals was 43.7% by fasting insulin level and 54.9% by HOMA. Mean values for fasting insulin and HOMA‐IR levels by sex and blood pressure categories are presented in Table II.

Table I.

Distribution of Characteristics of Sample Population by Blood Pressure Category

	Total	Normotension	Prehypertension	Hypertension
Sample size, No.	3203	1216	757	1230
Population represented, No.	163,942,181	65,258,000	42,286,148	56,398,033
Sex
Female	51.6	60.4 (1.5)	38.3 (2.4)	51.4 (1.6)
Male	48.4	39.6 (1.5)	61.7 (2.4)	48.6 (1.6)
Age, y
20–44	54.6	75.0 (1.9)	56.9 (2.7)	29.3 (1.9)
45–64	33.7	22.9 (1.7)	35.8 (2.6)	44.6 (2.2)
65–80	11.8	2.2 (0.5)	7.4 (0.8)	26.1 (1.5)
Ethnicity
White	73.5	71.9 (2.3)	71.9 (2.7)	76.6 (2.1)
Black	10.0	9.1 (1.3)	9.6 (1.5)	11.3 (1.5)
Mexican American	7.3	9.9 (1.3)	7.1 (1.2)	4.5 (0.7)
Other Hispanic	6.3	6.7 (1.6)	7.0 (2.3)	5.3 (1.9)
Other	2.9	2.3 (0.6)	4.4 (1.4)	2.4 (0.7)
Current smoker
Yes	24.9	28.0 (1.8)	25.9 (2.0)	20.7 (1.6)
No	75.1	72.0 (1.8)	74.1 (2.0)	79.3 (1.6)
Moderate to vigorous exercise in past month
Yes	63.9	67.6 (1.9)	65.3 (2.1)	58.6 (2.3)
No	36.1	32.4 (1.9)	34.7 (2.1)	41.4 (2.3)
BMI (kg/m2)
<30	71.5	81.8 (1.4)	73.7 (2.0)	58.0 (1.5)
≥30	28.5	18.2 (1.4)	26.3 (2.0)	42.0 (1.5)
Carbohydrate intake, d (g)
<200	29.7	27.3 (1.1)	27.0 (2.6)	34.4 (1.8)
≥200	70.3	72.7 (1.1)	73.0 (2.6)	65.6 (1.8)
Family history of CVD
Yes	7.7	6.3 (1.0)	8.5 (1.5)	8.7 (0.9)
No	92.3	93.7 (1.0)	91.5 (1.5)	91.3 (0.9)
Values are expressed as percentage or percentage (standard error). Abbreviations: BMI, body mass index; CVD, cardiovascular disease.

Table II.

Mean Fasting Insulin and HOMA‐IR Levels by Sex and Blood Pressure Category

	Total	Normotension	Prehypertension	Hypertension
Fasting insulin (µIU/mL)	12.08	9.41	12.16	14.54
Men
Women	10.88	9.64	10.67	12.68
HOMA‐IR
Men	3.08	2.28	3.03	3.90
Women	2.64	2.20	2.58	3.27
Abbreviation: HOMA‐IR, homeostasis model assessment‐insulin resistance.

In an unadjusted analysis, individuals with prehypertension or hypertension were significantly more likely to have IR as measured by both fasting insulin and HOMA, compared with normotensive persons (Table III). After adjustment for age, sex, race, smoking status, BMI, carbohydrate intake, exercise level, and family history of early CVD, IR remained significantly associated with both prehypertension and hypertension. The odds of IR as assessed by fasting insulin and HOMA levels in those with prehypertension were 1.67 times the odds for those with normotension, while the odds of IR in hypertensive individuals was more than 2 times that of normotensives (Table III). Once stratified by sex, IR was associated with prehypertension in only men (Table IV) for both fasting insulin (OR, 2.12; 95% CI, 1.47–3.07) and HOMA (OR, 1.79; 95% CI, 1.31–2.45)

Table III.

Odds Ratios and 95% Confidence Intervals for the Association Between Blood Pressure Category and Measures of Insulin Resistance

	Normotension	Prehypertension	Hypertension
		OR (95% Cl)
Fasting insulin (µIU/mL)
>12.2a	1.00	2.08 (1.52–2.85)	3.34 (2.47–4.52)
>12.2b	1.00	1.67 (1.12–2.48)	2.12 (1.42–3.18)
HOMA
≥2.6a	1.00	2.20 (1.73–2.81)	3.93 (3.07–5.03)
≥2.6b	1.00	1.67 (1.22–2.30)	2.33 (1.68–3.22)
aUnadjusted analysis with simple logistic regression. bMultivariable logistic regression analysis controlling for age, sex, race, smoking status, body mass index, carbohydrate intake, exercise level, and family history of early cardiovascular disease. Abbreviations: OR, odds ratio; CI, confidence interval; HOMA, homeostasis model assessment.

Table IV.

ORs and 95% CIs for the Association Between Blood Pressure Category and Measures of Insulin Resistance Among Men and Women

	Normotension	Prehypertension	Hypertension
		OR (95% Cl)
Fasting insulin (µIU/mL)
Men >12.2a	1.00	2.12 (1.47–3.07)	2.81 (1.73–4.57)
Women >12.2a	1.00	1.41 (0.71–2.80)	1.70 (1.01–2.88)
HOMA
Men ≥2.6a	1.00	1.79 (1.31–2.45)	2.70 (1.91–3.81)
Women ≥2.6a	1.00	1.56 (0.85–2.86)	1.94 (1.19–3.15)
aMultivariable logistic regression analysis controlling for age, race, smoking status, body mass index, carbohydrate intake, exercise level, and family history of early cardiovascular disease. Abbreviations: OR, odds ratio; CI, confidence interval; HOMA, homeostasis model assessment.

DISCUSSION

Hypertension and diabetes are major public health issues facing the US population. It has been demonstrated that diabetes is associated with hypertension but not with prehypertension. ¹⁰ As expected, our results show that individuals with hypertension who are free of known diabetes have higher IR, but this study also shows that IR is associated with prehypertension, specifically in men. Compared with persons with normal blood pressure, those with prehypertension were more likely to have IR as assessed by fasting insulin and HOMA levels. This is of possible clinical significance because it may indicate that those individuals with prehypertension are at higher risk for developing not only hypertension but also diabetes. The recent Trial of Preventing Hypertension (TROPHY) study ¹⁷ showed that treatment of individuals with prehypertension using candesartan may prevent progression to hypertension. Likewise, further studies are needed to establish whether prehypertension progresses to diabetes at greater rates than normal blood pressure and whether treatment of IR in those with prehypertension prevents progression to diabetes or hypertension.

This study adds to the growing evidence of correlations between IR and CVD risk factors. The findings also provide further characterization of individuals with prehypertension. Demonstrating that blood pressure in the prehypertension range is associated with dysfunctional metabolic states with increased CVD risk, like IR, may be of clinical importance for screening, prevention, and treatment. Finally, the study suggests a need to evaluate whether interventions to control blood pressure at the prehypertension level impact IR and the subsequent disease outcomes related to IR.

This study has some strengths, including the use of a nationally representative data set that allows for population estimates and conclusions generalizable to the US population. The study also further stratifies associated risks with 2 important predisease states. IR in conjunction with prehypertension may confer some additional risk to patients beyond the risks for each individual predisease state. Cohort studies may help to elucidate the answers to these questions.

Some limitations of this study include the cross‐sectional nature of the data, which limits inference beyond association. Thus, we are unable to establish the direction of the association between prehypertension and IR. Further studies are needed to establish IR as a marker of future disease, such as diabetes and CVD, in individuals with prehypertension. The lack of association between IR and prehypertension in women may be due to sample size. Of the 3203 total subjects, 23.6% had prehypertension (757) and of those, fewer than 40% were female. Further studies with larger numbers of women may be useful. Last, the HOMA‐IR level of ≥2.6 used to determine IR is based on a study using the 75th percentile value. We feel reassured that this value is an appropriate measure of IR, because adjusted ORs for HOMA‐IR in those with prehypertension were similar to the odds as measured by fasting insulin level, which is based on the gold standard hyperinsulinemic euglycemic clamp method.

In summary, prehypertension is associated with IR in a nationally representative sample of US men. Continued investigation into links between these 2 important predisease states and future disease outcomes is warranted.

Acknowledgment and disclosure: The authors would like to thank Elizabeth G. Hill, PhD, Medical University of South Carolina, for her statistical support in this study. Supported in part by grants 1D12HP00023 and 1D14HP00161 from the Health Resources and Services Administration.