Cross-national differences in the use of 7 antihypertensive drug classes and combination drug therapy among treated hypertensive patients. ACE indicates angiotensin-converting enzyme; ARB, angiotensin II receptor blocker; and CCB, calcium channel blocker.
Cross-national differences in hypertension control (defined as a latest systolic blood pressure level of <140 mm Hg and a diastolic blood pressure level of <90 mm Hg) and medication increase for those with inadequately controlled hypertension.
Wang YR, Alexander GC, Stafford RS. Outpatient Hypertension Treatment, Treatment Intensification, and Control in Western Europe and the United States. Arch Intern Med. 2007;167(2):141-147. doi:10.1001/archinte.167.2.141
Hypertension guidelines in the United States tend to have more aggressive treatment recommendations than those in European countries.
To explore international differences in hypertension treatment, treatment intensification, and hypertension control in western Europe and the United States, we conducted cross-sectional analyses of the nationally representative CardioMonitor 2004 survey, which included 21 053 hypertensive patients visiting 291 cardiologists and 1284 primary care physicians in 5 western European countries and the United States. The main outcome measures were latest systolic and diastolic blood pressure (BP) levels, hypertension control (latest BP level, <140/90 mm Hg), and medication increase (dose escalation or an addition to or switch of drug therapy) for inadequately controlled hypertension.
At least 92% of patients in each country received antihypertensive drug treatment. The initial pretreatment BP levels were lowest and the use of combination drug therapy (≥2 antihypertensive drug classes) was highest in the United States. Multivariate analyses controlling for age, sex, current smoking, and physician specialty indicated that, compared with US patients, European patients had higher latest systolic BP levels (by 5.3-10.2 mm Hg across countries examined) and diastolic BP levels (by 1.9-5.3 mm Hg), a smaller likelihood of hypertension control (odds ratios, 0.27-0.50), and a smaller likelihood of medication increase for inadequately controlled hypertension (odds ratios, 0.29-0.65) (all P<.001). In addition, controlling for initial pretreatment BP level attenuated the differences in latest systolic and diastolic BP levels and the likelihood of hypertension control.
Lower treatment thresholds and more intensive treatment contribute to better hypertension control in the United States compared with the western European countries studied.
Hypertension is a highly prevalent disease and a strong risk factor for cardiovascular disease in industrialized countries.1- 4 Numerous studies suggest its widespread underdiagnosis and undertreatment in Europe5- 7 and the United States.8 Guidelines in these countries differ regarding the blood pressure (BP) thresholds at which to initiate drug treatment and the definitions of high risk, with the guidelines in the United States being more aggressive than others.9- 17 According to national surveys conducted in the 1990s, the rates of hypertension treatment and control were lower in 5 European countries and Canada compared with the United States.18,19 These findings were consistent with the more aggressive hypertension treatment guidelines in the United States, although the use of aggregate-level data by age group and sex precluded controlling for important patient-level differences, such as pretreatment BP level and cardiovascular comorbidities, across countries.
We hypothesized that the more aggressive hypertension treatment guidelines in the United States are associated with lower drug treatment thresholds and more intensive drug treatment, both of which contribute to better hypertension control. To test this hypothesis, we used original patient-level data from a geographically representative survey of outpatient visits during which a diagnosis of hypertension was made in 5 western European countries and the United States. We analyzed cross-national differences in initial pretreatment BP level, comorbidities, and antihypertensive drug treatment and investigated how these factors explained differences in latest BP levels and hypertension control. For patients with inadequately controlled hypertension, we further analyzed the cross-national differences in medication increase (dose escalation or an addition to or switching of drug therapy) to learn more about the intensity of drug treatment.
We used the 2004 data from CardioMonitor, an ongoing survey of physician visits by ambulatory patients with cardiovascular diseases in selected countries. The survey has been conducted since the 1980s by Taylor Nelson Sofres (TNS) Healthcare, an international market research company based in London, England.20 CardioMonitor data have been validated against other control data and were found to be consistently reliable. These data have been used previously to assess hypertension treatment and control in Europe and the United States.10,21,22
In CardioMonitor, a nationally representative sample of physicians in a country are recruited at random by nationally distributed interviewers. Whether a physician fulfills the sampling quota is determined via a screening questionnaire, which includes a requirement that the physician sees a minimum of 15 patients per week who have a cardiovascular disorder (hypertension, coronary artery disease, heart failure, arrhythmia, arteriosclerosis, cerebral ischemia, peripheral vascular disease, or hyperlipidemia). The quotas were designed to be representative in terms of physician type, region, sex, and age or years of experience. After telephone confirmation of a physician's eligibility and willingness to participate, a field-worker delivers 15 copies of a 2-page diary to the physician's office and explains the tasks to the physician in person. The diaries are then left with the physician to record the visit of each patient with a cardiovascular disorder of interest who is seen in 1 week or the visits of the first 15 cardiovascular patients seen, whichever occurs sooner. One week later, the field-worker returns to the physician's office to collect the diaries. Extensive quality control and editing checks are made before the data are processed and entered into the database.
The 2-page diary contains mostly closed questions, except for open-ended questions about drug therapy and other diagnoses. Survey variables include physician specialty; patient age, sex, and BP levels (including initial pretreatment BP level, if known); presence of specific cardiovascular and other diagnoses (eg, diabetes mellitus); and up to 5 cardiovascular drugs (including over-the-counter ones) currently being taken by the patient. No information is collected on when the initial pretreatment BP level was measured.
Our study data set included 5 western Europeans countries—France, Germany, Italy, Spain, and the United Kingdom—and the United States (Table 1). We identified all visits of adult patients (≥18 years old) with a primary or secondary diagnosis of hypertension, excluding the 2% of hypertensive patients with age, sex, or the latest BP level missing. These patients visited a total of 291 cardiologists, 990 family practice or general practice physicians, and 294 internists in the study countries. We examined patient characteristics by country, including age, sex, current smoking, physician specialty (cardiology, family practice or general practice, or internal medicine), initial pretreatment BP level (if reported), and comorbid conditions. Next, we examined cross-national differences in the use of 7 antihypertensive drug classes (thiazides, other diuretics, β-blockers, calcium channel blockers, angiotensin-converting enzyme inhibitors, angiotensin II receptor blockers, and others) and combination drug therapy (use of ≥2 antihypertensive drug classes). For a combination drug with multiple ingredients (eg, the combination of an angiotensin II receptor blocker and a thiazide), each ingredient was treated as a separate drug.
Four outcome measures were compared across the study countries, using the United States as the comparison (or base) country. The first 2 measures were latest systolic and diastolic BP levels. The third measure was hypertension control, defined as the latest BP level being below 140/90 mm Hg. For patients with inadequately controlled hypertension (ie, those whose latest systolic BP level was ≥140 mm Hg or whose latest diastolic BP level was ≥90 mm Hg), we defined the fourth measure, medication increase for inadequately controlled hypertension, as a dose escalation or an addition to or switch of drug therapy during the patient visit.
We used linear regression analyses to estimate cross-national differences in latest systolic and diastolic BP levels and logistic regression analyses to estimate cross-national differences in hypertension control and medication increase for inadequately controlled hypertension. To assess the individual contributions of initial pretreatment BP level, comorbid conditions, and concurrent drug treatment, we adopted a sequential, 4-step approach, adding more explanatory variables in each step. These 4 steps represented, in order, demographics-adjusted, comorbidity-adjusted, treatment-adjusted, and initial or latest BP–adjusted models.
In the demographics-adjusted model, the explanatory variables included age, sex, current smoking, whether the visit was to a cardiologist, and 5 country indicator variables (vs the United States). In the comorbidity-adjusted model, we added 8 indicator variables representing the presence of 7 cardiovascular comorbidities and diabetes mellitus. In the treatment-adjusted model, we added 7 indicator variables representing the concurrent use of 7 antihypertensive drug classes. In the initial or latest BP–adjusted model, we added the initial pretreatment systolic and diastolic BP levels (for the first 3 outcome measures, ie, latest systolic and diastolic BP levels and hypertension control) or the latest systolic and diastolic BP levels (for the fourth outcome measure, ie, medication increase for inadequately controlled hypertension). Patient visits without initial pretreatment BP levels reported were therefore excluded.
We estimated the linear and logistic regression analyses using the GENMOD procedure in SAS statistical software, version 8.1 (SAS Institute Inc, Cary, NC), with the REPEATED statement to adjust for multiple patients (up to 15) seen by the same physicians in a country. The SAS GENMOD procedure derives the generalized estimating equation estimates using the maximum likelihood method.
We carried out several sensitivity analyses to test the robustness of our findings. First, rather than the standard definition of hypertension control, we used an alternative definition: the latest BP level below 130/80 mm Hg for patients with diabetes and below 140/90 mm Hg for all others. This alternative definition of hypertension control is consistent with hypertension guidelines that were effective in the study countries in 2004.14,15 We then examined hypertension control and medication increase for inadequately controlled hypertension using this alternative definition. We did not define a different goal BP level for patients with coronary artery disease or chronic heart failure because of potentially different diagnostic criteria across countries, as indicated by significant cross-national variations in these comorbidities. In addition, renal disease was not systematically reported in CardioMonitor. Second, we repeated the multivariate regressions on the 4 outcome measures using 2 separate patient subgroups, ie, those with an initial pretreatment BP level of 160/100 mm Hg or higher and those visiting primary care (noncardiologist) physicians, to check whether our findings remained true for late-stage hypertension (stage 2 in JNC 7 [the seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure]14) and in the primary care setting. Finally, we repeated the multivariate regression analyses using only noncardiologist visits, to check whether our findings remained true in the primary care setting.
Our study sample consisted of 21 053 hypertensive patients in the 5 western European countries and the United States (Table 2). The mean patient age was 65 years, 53% of patients were male, and 23% of them were diagnosed as having diabetes mellitus. Patients in the United States were similar to those in western European countries in the distributions of age, sex, current smoking, and cardiologist visits. Reported comorbidities varied considerably across countries, with Germany leading in the reporting of 7 comorbidities and the United States leading in the reporting of hyperlipidemia and being the second in the reporting of 5 others. At least 92% of patients in each country received antihypertensive drug treatment. Initial pretreatment BP level was reported for 61% to 80% of patients; it was lowest in the United States (mean, 161/94 mm Hg vs 167-173/96-99 mm Hg across European countries), as was the latest BP level (mean, 134/79 mm Hg vs 139-144/80-84 mm Hg across European countries). Alternatively, only 65% of US patients had an initial pretreatment BP level of 160/100 mm Hg or higher (vs 81%-90% of European patients). The 10th and 25th percentiles of initial pretreatment BP level were 140/80 and 150/90 mm Hg, respectively, in the United States.
The use of thiazides was quite similar across countries (29%-31%) (Figure 1). In contrast, the use of other antihypertensive drug classes varied considerably by country, especially for β-blockers (range, 20%-49%), angiotensin-converting enzyme inhibitors (range, 27%-52%), and angiotensin II receptor blockers (range, 18%-36%). Finally, the use of combination drug therapy (use of ≥2 antihypertensive drug classes) was highest in the United States (64% vs 44%-59% across European countries).
The rate of hypertension control (latest BP level, <140/90 mm Hg) was highest in the United States (63% vs 31%-46% of patients across European countries) (Figure 2). In addition, of the 11 969 patients with inadequately controlled hypertension, the United States had the highest percentage of patients receiving any medication increase during the visit (38% vs 15%-28% across European countries). These patterns did not change under the alternative definition of hypertension control (latest BP level, <130/80 mm Hg for patients with diabetes and <140/90 mm Hg for all others). Hypertension control was achieved by 53% of patients in the United States and 27% to 40% of patients across European countries; a medication increase for inadequately controlled hypertension was achieved in 32% of patient visits in the United States compared with 14% to 26% across European countries.
Multivariate linear regression analyses confirmed that the latest systolic and diastolic BP levels were higher in the 5 western European countries compared with the United States (Table 3). In the demographics-adjusted regression, the differences between European countries and the United States were 5.3 to 10.2 mm Hg for systolic BP level and 1.9 to 5.3 mm Hg for diastolic BP level (all P<.001). These cross-national differences remained significant (all P<.01) after further controlling for comorbidities and concurrent drug treatment. In addition, controlling for initial pretreatment BP level narrowed the differences to 2.1 to 6.8 mm Hg for systolic BP and –0.1 to 3.5 mm Hg for diastolic BP (all P<.01, except for a nonstatistically significant difference between France and the United States with regard to diastolic BP level). Among patients who had reported an initial pretreatment BP level (71% of all patients and 61% of US patients), initial pretreatment BP level accounted for more cross-national differences in latest BP level than did comorbidities and concurrent drug treatment combined (data not shown).
Compared with the United States, hypertension control was less likely in the 5 European countries according to the demographics-adjusted regression (odds ratios [ORs], 0.27-0.50; all P<.001) (Table 4). Similar to the differences in the latest systolic and diastolic BP levels, the differences in hypertension control between European countries and the United States remained significant (all P<.001) after controlling for comorbidities, concurrent drug treatment, and initial pretreatment BP level, and initial pretreatment BP level accounted for more cross-national differences than did comorbidities and concurrent drug treatment combined.
Of the 11 969 patients with inadequately controlled hypertension, those in European countries were less likely to receive a medication increase compared with those in the United States (ORs, 0.29-0.65; all P<.001) (Table 4). The cross-national differences in treatment intensification remained significant after controlling for comorbidities, concurrent drug treatment, and the latest BP level (all P<.001).
Compared with patients seen by primary care physicians (general practice, family practice, or internal medicine), those seen by cardiologists had a higher systolic BP level (2.0-3.1 mm Hg; all P<.01), higher diastolic BP level (0.7-1.4 mm Hg; all P<.05), and a smaller likelihood of hypertension control (ORs, 0.79-0.89; all P<.10) in multivariate regression analyses. These findings likely reflect patient selection (sicker, more complicated, or more difficult-to-treat patients are referred to cardiologists) rather than real differences in practice style. Indeed, among patients with inadequately controlled hypertension, those seen by cardiologists were more likely to receive a medication increase than were patients seen by primary care physicians (ORs, 3.0-3.4; all P<.001).
Sensitivity analyses using the alternative definition of hypertension control (latest BP level, <130/80 mm Hg for patients with diabetes and <140/90 mm Hg for all others), focusing on patients with late-stage hypertension (initial pretreatment BP level, ≥160/100 mm Hg), or focusing on primary care (noncardiologist) visits led to similar results.
Using a nationally representative survey of hypertensive outpatient visits in 2004, we found that, compared with the 5 western European countries studied, the latest BP levels were lower and hypertension control was better in the United States. These findings are consistent with previous comparisons between the United States and European countries that used population-level data from the World Health Organization MONICA (Multinational Monitoring of Trends and Determinants in Cardiovascular Disease) Project and national surveys conducted in the 1980s and 1990s.1,2,18,19,23 The availability of comparable, patient-level data from CardioMonitor allowed us to go beyond previous aggregate-level analyses and to assess how initial pretreatment BP level, comorbidities, and drug treatment contributed to cross-national differences in latest BP levels and hypertension control. We showed that a significant portion of the cross-national differences was explained by the lower initial pretreatment BP levels in the United States. Furthermore, for patients with inadequately controlled hypertension, we defined treatment intensification or medication increase and showed that the likelihood of receiving any medication increase was significantly greater in the United States compared with European countries. Together our findings suggest that better hypertension control in the United States may be explained by lower treatment thresholds and more intensive treatment, both of which are consistent with the more aggressive treatment guidelines in the United States compared with the other countries examined.9,12,14- 17 Despite substantial differences in cross-national patterns of antihypertensive drug selection (with the exception of the recommended, first-line thiazides), these variations and their resultant cost differences did not appear to affect BP control. This suggests that BP monitoring and regular follow-up care may be more important than drug selection per se in improving hypertension control.
Our study has a number of limitations. First, the CardioMonitor survey is visit-based and likely oversamples sicker patients, who are more likely to visit physicians. In addition, its physician sampling based on quota criteria may not reflect the universe of physicians treating hypertension in a country. Although our rates of hypertension treatment and control are comparable to those of other visit-based studies,24,25 they are more than double those in population-based studies8,18,19 that included undiagnosed and untreated hypertensive patients. Second, the cross-sectional CardioMonitor survey provides only a snapshot of hypertension treatment and control and does not allow us to assess how early diagnosis, early treatment, and intensive treatment over time individually contribute to better hypertension control in the United States. Even though the lower initial pretreatment BP level in the United States was significantly related to better hypertension control, this measure was available for only 71% of all patients (61% of US patients) and its time window of measurement was not collected in CardioMonitor. Whether the lower initial pretreatment BP level reflects early diagnosis, early treatment, or both is an interesting topic for future research. Third, even though it is less likely the case for hypertension than for some other diseases (eg, coronary artery disease or hyperlipidemia), disease diagnostic criteria vary by country and may introduce biases into our study. This may explain some of our observed significant differences in comorbidities across countries. Finally, factors such as health care coverage, financing, and delivery influence disease treatment and control through mechanisms other than those explored herein.26 For example, although health insurance coverage is universal across European countries, regulations on drug pricing and limitations on drug consumption are prevalent. In contrast, drug pricing and consumption are relatively free in the United States, but more than 10% of the US population is uninsured. How health care system factors affect disease treatment and control is an important topic for future research.
Better hypertension control in the United States probably results in an overall lower economic burden of cardiovascular diseases associated with hypertension.3,4,18,19 One potential downside of aggressive case finding and drug treatment initiation is that some patients who do not need drug treatment may be treated, resulting in increased spending for drugs. However, given that the mean BP level before treatment was 161/94 mm Hg (10th percentile, 140/80 mm Hg; 25th percentile, 150/90 mm Hg) in the United States, any increased spending for drugs because of overtreatment among this sample is likely small compared with the future costs of cardiovascular events due to inadequate hypertension control. In our study, even in the United States, nearly half of the hypertensive patients did not achieve their goal BP level set by the JNC 7 guidelines. Thus, better BP control in the United States should not be too quickly praised. Substantial potential for better hypertension control and future cost saving exists in both Europe and the United States.
Correspondence: Y. Richard Wang, MD, PhD, Department of Medicine, Temple University Hospital, 3401 N Broad St, Room 812, Eighth Floor, Parkinson Pavilion, Philadelphia, PA 19140 (email@example.com).
Accepted for Publication: September 15, 2006.
Author Contributions: Dr Wang had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Wang, Alexander, and Stafford. Acquisition of data: Wang. Analysis and interpretation of data: Wang and Alexander. Drafting of the manuscript: Wang and Alexander. Critical revision of the manuscript for important intellectual content: Wang, Alexander, and Stafford. Statistical analysis: Wang, Alexander, and Stafford. Obtained funding: Alexander and Stafford. Administrative, technical, and material support: Wang. Study supervision: Wang, Alexander, and Stafford.
Financial Disclosure: None reported.
Funding/Support: This study was supported by the MacLean Center for Clinical Medical Ethics, a grant 5 K12 AG00488 from the National Institute of Aging (Dr Alexander), and grant RO1-HS11313 from the Agency for Healthcare Research and Quality (Dr Stafford).
Disclaimer: The design of the analysis and interpretation of the results were performed independently by the authors. The views expressed herein are those of the authors and do not necessarily reflect those of the sponsors.
Additional Information: Part of the analysis was completed during Dr Wang's previous employment at AstraZeneca Pharmaceuticals.
Acknowledgment: We thank Philip Chadwell, MBA, DipM, of TNS Healthcare, for technical help with the CardioMonitor database. We also thank an anonymous reviewer and the editors at this journal for helpful comments that substantially improved this article.