Jump to
Abstract
Previous assessments of the prevalence of resistant hypertension (RH) in uncontrolled blood pressure (BP) have ranged from 3% to 30%. Using real-world data, our aim was to estimate the prevalence of RH in patients belonging to the Maccabi Healthcare Services, a 2-million-member health organization in Israel. From 2010 to 2011, all hypertensive patients with ≥2 recorded BP measurements during a minimum period of 6 months were identified. Patients were considered uncontrolled if their most recent BP during the study period and their mean systolic BP or diastolic BP during a preceding period of ≥6months were systolic BP ≥140 mm Hg or diastolic BP ≥90 mm Hg, or systolic BP ≥130 mm Hg or diastolic BP ≥80 mm Hg in chronic kidney disease or diabetes mellitus. Uncontrolled patients taking diuretics and ≥2 antihypertensive therapy classes at their maximal recommended dose were regarded as resistant hypertensives. A total of 172 432 patients were eligible for the study. Uncontrolled BP was found in 35.9% (n=65 710). Overall, 2.2% of the uncontrolled patients (n=1487) were resistant hypertensives. Patients with RH were characterized by a significantly (P<0.01) older age, higher body mass index, and multicomorbidity (including dyslipidemia, diabetes mellitus, and impaired renal function) compared with patients with controlled hypertension receiving equivalent treatment. The results of this large population-based study indicate a substantially lower prevalence of RH than previously reported. Most patients with uncontrolled BP took less than the maximal recommended antihypertensive treatment.
Introduction
Arterial hypertension is associated with an increased risk of cardiovascular morbidity and mortality.1,2 Blood pressure (BP) reduction has been shown to reduce fatal and nonfatal outcomes in hypertensive patients.3,4
Despite the improvement in diagnostic options and therapeutic interventions, observational studies continue to show persistently low rates of BP control in the general population of hypertensive patients.5,6 In some of these patients, hypertension is resistant to treatment and persists despite the use of a combination of drugs. Multiple previous cross-sectional studies have consistently indicated an increased frequency of cardiovascular complications in patients with resistant hypertension (RH) compared with subjects without RH.7–10
Daugherty et al11 in a longitudinal retrospective study found a 47% higher risk for cardiovascular complications in patients with RH compared with patients with non-RH. Studies reporting on RH have shown prevalence ranging between 3% and 38%.7–9,12–14 However, these studies differ in the definition of RH (eg, number of drug classes included, time interval with uncontrolled BP), including or ignoring pharmacological treatment dosage and adherence to treatment, data source (eg, self-reported, physician-reported, or computerized database), method of BP assessment (eg, office or ambulatory BP monitoring), and use of a single or repeated assessments of BP. The current retrospective population-based study aimed at assessing the prevalence of RH in the hypertensive population of the Maccabi Healthcare Services (MHS) using its comprehensive computerized database, including drug dispense data.
Methods
Study Population
All data were obtained from the computerized databases of MHS. The study population included all patients aged >18 years who had entered the MHS’s hypertension registry before January 1, 2010. Inclusion was based on ≥2 physician’s diagnoses or hospital records and ≥2 BP measurements of systolic BP (SBP) ≥140 or diastolic BP (DBP) ≥ 90 mm Hg. For cases with abnormal measurements but no diagnoses, 4 documented measurements were required where ≥50% of the measurements were SBP >160 mm Hg or DBP >90 mm Hg. Also included were patients with ≥6 dispensed medications for hypertension. Patients were eligible for the study if they had had ≥2 BP measurements during a period of ≥6 months between January 1, 2010, and December 31, 2011. The mean (SD) number of BP measurements per patient was 3.9 (5.1). Patients who left MHS or died before the end of the 6-month follow-up were excluded from the study.
Definition of Outcome
An index BP measurement was defined as the last measurement available during the study period (2010–2011). Patients were defined as having uncontrolled hypertension if both the index measurement and the mean of all measurements performed during a period of ≥6 months before the index BP measurement were above the goal. The goal was defined as SBP <140 or DBP <90 mm Hg in patients without diabetes mellitus (DM) or chronic kidney disease (CKD; estimated glomerular filtration rate [eGFR] <60 mL/min) and SBP <130 or DBP <80 mm Hg in diabetic patients or those with CKD15 (Figure S1 in the online-only Data Supplement). The BP was measured in the outpatient clinics in the sitting position by either the primary care physician or the nurse and was recorded in the computerized patient’s file. To assess RH, medication dispense data during the last 30 days, within the 6-month period (Figure S1), were searched. Patients were considered RH if ≥25 days of the 30-day period were covered by medications from ≥3 classes, including a diuretic, all at maximal dosage. The latter was regarded as maximal treatment intensity. Antihypertensive drug classes included β-blockers, angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, diuretics, calcium antagonists, aldosterone antagonists, α-blockers, and direct vasodilators. Patients taking a combination of antihypertensive pills were counted as taking separate classes for each drug (ie, thiazide diuretic and angiotensin-converting enzyme inhibitor). Full dosage level was defined as the active ingredient of all drugs available in the MHS (Table S1). Sensitivity of the assessment of RH prevalence was tested by applying the same definition of RH to a 60-day treatment period before the last BP measurement. For this time period, RH was established if ≥50 days were covered by treatment.
Data Extraction and Definition
Background data extracted for the study included age, sex, smoking status (coded as ever, never, and missing), body mass index, country of birth, and region in Israel where health services are given. Data also included socioeconomic status level, ranging between 1 (lowest) and 20 (highest), according to the poverty index of the member’s enumeration area as defined by the 1995 national census.16
Clinical data included date of hypertension registry entry as well as dates of entry into the following MHS’ registries: DM,17 subregistries of the cardiovascular disease registry,18 congestive heart disease, ischemic heart disease, peripheral vascular disease, atrial fibrillation, stroke, transient ischemic attack, CKD (based on eGFR <60 mL/min or report of dialysis), and chronic obstructive pulmonary disease.
History of additional diseases was determined by the International Classification of Diseases, Ninth Revision (ICD-9) codes and relevant internal MHS codes for diagnoses reported in the computerized medical record, that is, if diagnosis was reported by a physician with a relevant specialization or at least twice by a general or family physician. Diagnoses (ICD-9 codes) include sleep apnea (ICD-327.2 and ICD-780.51), dyslipidemia (ICD-272 and its subhierarchy), and asthma (ICD-493 and its subhierarchy).
Data as to clinical tests and test results were retrieved for the following tests: serum creatinine, glucose, sodium, potassium, lipid profile, urine microalbumin, plasma renin activity, blood aldosterone in the recumbent and upright position, 24-hour urine aldosterone excretion, 24-hour urine norepinephrine excretion, renal ultrasound, Doppler of the renal arteries, and 24-hour ambulatory BP monitoring (24H ABPM).
We calculated the mean test value of tests performed within the 6-month index period and ≤6 months prior for serum creatinine, glucose, sodium, potassium, urine microalbumin, and ≤12 months prior for lipids.
Drug dispense data included brand name, active ingredients, dose per tablet, date of purchase, number of packets/tablets, number of treatment days of antihypertensive drugs (Table S1), as well as BP-elevating drugs (Table S2).
Finally, the medical records of 57 random cases of RH were accessed directly and searched for BP measurements and prescribed drugs. In 7 cases, inconsistencies were found between the medical record and the central database (6 were prescribed <3 drug classes and 1 had an unreported BP measurement in the central database). Therefore, the positive predictive value of RH in this study was 87.7% (95% confidence interval, 76.3–94.9%).
Statistical Analysis
Proportions were computed for prevalence measures. Descriptive statistics were presented with count and percentage for categorical variables and SD for continuous variables. Histograms and scatter plots were used to describe distribution of SBP and DBP in uncontrolled patients with less than maximal treatment. χ2 and Mann–Whitney tests compared categorical data and continuous variables between groups, respectively. Multivariate logistic regression models were used for assessing the associations between background factors and uncontrolled versus controlled hypertension groups, both at maximal treatment. Variables maintained in the multivariate model were those with a statistically significant association or confounders of associations of other variables, including age, sex, body mass index, DM, CKD, stroke or transient ischemic attack, potassium levels, low-density lipoprotein cholesterol levels, and region of the MHS clinic. Odds ratios and 95% confidence intervals are presented.
Results
Uncontrolled Hypertension
As of January 1, 2010, 230 750 patients were listed in the hypertension registry of the MHS. Of these, 58 318 (25%) were excluded from the study because of <6 months follow-up and missing BP measurements; 172 432 (75%) were eligible for the study (Figure 1). Those excluded were 3.5 years younger on average, more likely to be men, and were less likely to have comorbidities than those included (Table S3). Prevalence of uncontrolled hypertension in the eligible population was 35.9% (n=65 710; Figure 2). Diabetic patients and those with a CKD had a 3-fold prevalence of uncontrolled hypertension than patients without these conditions (Table S4). When the new BP goals (SBP <140 or DBP <90 mm Hg)19 were applied to diabetic patients and those with CKD, prevalence of uncontrolled hypertension was only 21.7% (n=37 451).
Reasons for exclusion from the study. BP indicates blood pressure; and MHS, Maccabi Healthcare Services.
Flow diagram of treatment patterns in the eligible population by status of blood pressure (BP) control.
The prevalence was similar in those with CKD and slightly higher in those with DM or both DM and CKD than in those without any of them (22.8% and 26.9% versus 20.5%, respectively; P<0.001; Table S4). Among those without DM or CKD, 86.4% of the uncontrolled patients had SBP ranging from 140 to 159 mm Hg; 58.4% had DBP ranging from 90 to 99 mm Hg (Figure S2). Among diabetic or CKD uncontrolled patients, 47.7% had SBP ranging from 130 to 139 mm Hg; 31.1% had SBP ranging from 140 to 149 mm Hg; 86.6% had DBP ranging from 80 to 89 mm Hg. Most patients without DM and CKD had uncontrolled systolic and diastolic BP (68.3%), and only a few had isolated diastolic uncontrolled hypertension (5.7%; Figure S2). Among those with DM or CKD, the largest percentage had uncontrolled SBP (48.5%; Figure S2).
Approximately 98% (n=64 223) of the uncontrolled patients did not fulfill the definition of maximal treatment during the 30 days before the last BP measurement. Stratification of these 64 233 uncontrolled patients by treatment pattern showed that 20% purchased 3 classes of drugs including a diuretic, but the dosages were less than maximal, or the number of days covered during the 30 days before the last BP measurement was <25 (Figure 2). Almost 50% of the uncontrolled patients purchased drugs from only 1 or 2 drug classes and another 20% had no coverage of purchased drugs during the 30-day period.
Resistant Hypertension
The number of patients who fulfilled the definition of RH reached 1487, comprising 2.24% of the uncontrolled group and 0.86% of the entire eligible population (Figure 2). Sensitivity analysis was performed by considering antihypertension treatment during the 60 days before the last BP measurement, instead of 30 days, resulting in a prevalence of 3.24% (n=2131) of RH in the uncontrolled group or 1.24% of the entire eligible population.
When the new hypertension goals19 were used for diabetic patients and those with CKD, only 785 patients were still defined as RH. By applying this definition, prevalence of RH was 1.19% of the uncontrolled population and only 0.45% of the entire eligible population.
Characteristics of patients with RH were compared with 1343 controlled patients who fulfilled the same definition of maximal treatment (Figure 2). Patients with RH were older, had a higher mean body mass index, and more comorbidities including DM, CKD, dyslipidemia, peripheral vascular disease, and stroke or transient ischemic attack compared with patients with controlled hypertension. Patients with RH also had lower levels of eGFR and high-density lipoprotein cholesterol and higher levels of glucose, potassium, and low-density lipoprotein cholesterol (Table 1). No difference was found between RH and controlled patients with regard to smoking status, socioeconomic status, country of birth, and region of health services. Patients with RH used, in addition to diuretics, more antihypertensive agents than the controlled patients (Table 1). All patients used, in addition to diuretics, blockers of the renin–angiotensin system; >70% used calcium antagonists. Patients with RH used more β-blockers than controlled patients. Very few patients used aldosterone antagonists. In the multivariate model, RH was associated with age, morbid obesity (body mass index ≥30), DM, CKD, potassium levels >5 mmol/dL, and increased low-density lipoprotein cholesterol levels (Table 2).
Characteristics of Patients Treated With ≥3 Different Drug Classes at Maximal Dose Among Patients With Controlled and Uncontrolled BP
OR and 95% CI for Predictors of Uncontrolled BP Among Patients Receiving Maximal Therapy, Mutually Adjusted Using a Multivariable Logistic Regression Model
The proportion of RH patients who performed clinical tests for further investigation of the cause of RH ranged between 2.5% and 5% (Table S5). Only 4.2% of the patients with RH performed 24H ABPM to exclude white coat hypertension. Aldosterone antagonist treatment was used by 8.3% of the RH group versus 6.5% in the controlled BP group.
Discussion
In the present study, we found that BP was uncontrolled in >33% of the hypertensive patients. The rate of uncontrolled hypertension was ≈20% in patients without DM or CKD and ≈60% in patients with DM or CKD. However, when we used the new BP goals for patients with DM or CKD (SBP <140 or DBP <90 mm Hg),19 the rate of uncontrolled BP for the entire population was only 21.7%. Unlike many other studies, we found that most patients with uncontrolled hypertension had SBP levels <160 and DBP <100 mm Hg. Only a small percentage of the uncontrolled patients had severe hypertension.
Most patients with uncontrolled hypertension who did not take the recommended maximal dose of antihypertensive drugs were noncompliant or took <3 antihypertensive drugs. About 9% of the patients did not use diuretics; thus, actual RH was relatively low. In the entire hypertensive population, the rate of RH was <1%. Among those with uncontrolled hypertension, ≈2% had RH.
The rate of RH in our study was lower than reported previously. Most previous reports are based on cross-sectional studies, international registries, claims databases, or large clinical studies.7–9,13,14,19,20 Two large surveys found that 9% to 12% of treated hypertensive patients met the criteria of RH.8,9 In most studies, RH was defined as BP above goal with ≥3 antihypertensive medications or controlled to goal on ≥4 antihypertensive medications including diuretics prescribed at optimal dose.21 This definition is arbitrary and does not state the number of patients with BP who could not be controlled with the current antihypertensive armaments.
In the present study, we used practical strict criteria in defining RH. We tried to identify those who may benefit from further evaluation and more aggressive treatment. Therefore, we defined RH only when hypertension was not at the goal according to the criteria of the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure15 with ≥3 drugs including a diuretic at the maximal recommended dose. Because goal BP recently changed,19 we also calculated the rate of RH according to the new goals. With the new less aggressive goals for patients with DM and CKD, the rate of RH is even lower than the rate found according the old goals.
Our study is unique because it is based on a large registry that also allows verifying patients’ compliance and assessing the evaluation done for excluding pseudo-RH and secondary hypertension. Only patients who had ≥2 BP measurements ≥6-month interval during the study period were included. We applied strict criteria when defining RH, use of 3 drugs including a diuretic at the maximal dose, and 85% compliance. Those excluded because of missing BP measurements were younger and had lesser comorbidities, which may explain why they did not regularly visit their primary physician. It is possible that this group’s rate of uncontrolled BP and RH may even be lower. The rate of RH may have been underestimated because we do not know why patients did not receive the maximal dose and why 9% of the patients did not receive diuretics. It is possible that some patients were unable to tolerate the diuretics or did not tolerate the maximal therapeutic dose of other agents. These patients should practically also be defined as having RH. For those who could receive diuretics and maximal dosages of other drugs, we are presently unaware whether the addition of these agents or increasing the dose of other antihypertensive agents would control the BP. It is possible that some of the uncontrolled patients, not defined as RH in the present analysis, did indeed have RH. On the contrary, the rate of RH may have been overestimated because most patients did not perform a 24H ABPM to exclude white coat hypertension.
It is known that ≈20% of patients with apparent RH have white coat hypertension and should not be defined as RH.22 However, most studies that evaluated the rate of RH did not use 24H ABPM for diagnosis. Goal BP according to 24H ABPM is not well defined. We think that the rate of RH in our study was low because, unlike previous studies, strict criteria were used to define RH. It is noteworthy that when we used the new recommended BP goals for patients with DM and CKD, the rate of RH was even lower. Because the target BP in diabetic patients has changed, we think that henceforth the definition of RH should be the same for all patients.
Our registry included patients with a definite diagnosis of hypertension or patients without a diagnosis of hypertension but with evidence of ≥4 SBP measurements >140 mm Hg when ≥2 were >160 mm Hg. These strict criteria were chosen to exclude patients with borderline hypertension. It is significant that most patients with RH had BP levels <160/100 mm Hg. It seems that physicians underestimated the cardiovascular risk associated with BP levels ranging from 140 to 160/90 to 100 mm Hg and, therefore, did not refer patients for further evaluations nor added medications.
As in previous studies, we found that RH was more common in patients with DM, renal failure, obese and elderly patients, and patients with comorbidities. We observed a low rate of treatment with aldosterone antagonists because only 8% of patients with RH were treated with aldosterone antagonists. Similar results were described recently by Hanselin et al.20 An addition of aldosterone antagonists to patients with RH, particularly obese and elderly subjects and those with serum potassium levels <4.5 mmol/L, may control BP.21,23,24 Underuse of aldosterone antagonist diuretics may explain RH, and optimization of diuretic therapy should help control BP.
Our study has certain limitations. First, we based our analysis on clinic BP measurements and cannot estimate the accuracy of the measurements. To overcome this drawback, only patients who had >1 BP measurement were included. We used the index measurement and the mean of all measurements performed during the study period to define BP control. Laboratory data were missing for a minority of patients, which did not affect the rate of RH. Data were unavailable as to the side effects of antihypertensive drugs. We were unable to establish the number of patients who did not receive maximal treatment because of the side effects. Our estimation of adherence was based on drug purchase. This calculation is limited because we did not compare the purchase to the prescription. However, because patients with uncontrolled hypertension were taking a higher number of drugs than those with controlled hypertension and because many patients purchased drugs for >25 days per month, we assumed that nonadherence was a major factor in RH.
Perspectives
The main reasons for uncontrolled BP were noncompliance or use of less than the maximal recommended antihypertensive dosage. Therefore, the rate of RH found in this real-world population-based study is relatively low. Most patients with RH have BP levels <160/100 mm Hg. Only a few patients were evaluated to exclude the white coat effect and to identify secondary hypertension. Very few received aldosterone antagonists that could improve BP control. In patients with uncontrolled BP, noncompliance and use of inadequate dose of antihypertensive agents should be excluded. Ambulatory BP monitoring should be used to exclude white coat hypertension. In those with RH, further evaluation should be performed to improve BP control.
Acknowledgments
We thank Phyllis Curchack Kornspan for her editorial services. E. Grossman wrote the first draft of this article. All authors commented on the article. We thank the Morris Kahn Family Foundation for their support of this study.
Sources of Funding
The study was supported by the Morris Kahn Family Foundation.
Disclosures
None.
Footnotes
The online-only Data Supplement is available with this article at http://hyper.ahajournals.org/lookup/suppl/doi:10.1161/HYPERTENSIONAHA.114.03718/-/DC1.
- Received April 10, 2014.
- Revision received April 23, 2014.
- Accepted May 25, 2014.
- © 2014 American Heart Association, Inc.
References
Novelty and Significance
What Is New?
We used real-world data from a large health organization that includes 2 million members. We used a comprehensive computerized database, including drug dispense data. We defined resistant hypertension only when blood pressure (BP) was uncontrolled while patients were taking diuretics and ≥2 antihypertensive therapy classes at their maximal recommended dose, with 85% compliance. We found a substantially lower prevalence of resistant hypertension than previously reported.
What Is Relevant?
Most patients with uncontrolled BP do not receive optimal antihypertensive treatment. Optimizing antihypertensive treatment may improve BP control.
Summary
The rate of uncontrolled BP attributable to true resistant hypertension is low. Most patients with uncontrolled BP do not receive optimal treatment.
This Issue
Jump to
Article Tools
Share this Article
- Prevalence and Factors Associated With Resistant Hypertension in a Large Health Maintenance Organization in IsraelNovelty and Significance
