Classification of Blood Pressure Levels by Ambulatory Blood Pressure in Hypertension Hypertension Staging Through Ambulatory Blood Pressure Monitoring Masked Hypertension How Can We Use the Results of Ambulatory Blood Pressure Monitoring   in Clinical Practice? Prognostic Value of Ambulatory Blood-Pressure Recordings    in Patients    with  Treated  Hypertension

Classification of Blood Pressure Levels by Ambulatory Blood Pressure in Hypertension

The study was performed at the Hypertension Unit of the Department of Emergency Medicine in Vienna. Overall, 736 patients were included between January 1994 and June 2001. All patients were informed about the study protocol and gave their informed consent before study inclusion. Inclusion criteria were evidence of hypertension defined as blood pressure 140/90 mm Hg evaluated by 3 measurements on 3 consecutive visits according American Heart Association guidelines.⁹ Patients with secondary hypertension were excluded. The presence of previous cardiovascular events did not constitute exclusion in subjects maintaining their normal physical and work activities. At the beginning, the study procedure included a medical visit, which took place in the morning and consisted of a comprehensive medical history, a physical examination, and assessment of antihypertensive drug treatment. Twenty-four-hour ABPM was performed at the time of entrance. The incidence of cardiovascular events during the time of follow-up was recorded. A minimal 6-month follow-up was required for being included in the analysis.

Clinic Blood Pressure Measurements
During the physician’s visit (8 to 11 AM), blood pressure was measured in a quiet environment with a mercury sphygmomanometer with the patient in a sitting position after 5 minutes of rest, following the recommendations of the British Hypertension Society.¹⁰ Systolic and diastolic blood pressure values (Korotkoff phase I and phase V, respectively) represented in each visit the mean of 3 different readings measured at 5-minute intervals. In any patient, sphygmomanometric measurements were obtained by the same medical doctor.

Ambulatory Blood Pressure Monitoring
Ambulatory blood pressure monitoring was performed with oscillometric Meditech ABPM-04 devices, which had previously been validated and recommended for clinical use.^11,12 The monitoring equipment was applied at the end of the medical visit. The cuff was fixed to the nondominant arm, and 3 blood pressure readings were taken concomitantly with sphygmomanometric measurements to ensure that the average of the 2 sets of values did not differ by >5 mm Hg. The device was set to measure blood pressure at 15-minute intervals during the daytime (6 AM to 10 PM) and at 30-minute intervals during the nighttime (10 PM to 6 AM). The patient was sent home with instructions to hold the arm immobile at the time of measurements, to keep a diary of daily activities and quality of night rest, and to return to the hospital 24 hours later. The monitoring was always done on a working day and for treated patients during the normal intake of the usual antihypertensive treatment. The patients had no access to the ambulatory blood pressure values. Each of the 6 ambulatory blood pressure devices available for the study was checked monthly as described previously.^3,4

Data Analysis
In each participant, the blood pressure values obtained by the sphygmomanometer (before and after 24-hour ABPM) were averaged to calculate a single systolic and diastolic clinic blood pressure value. Ambulatory blood pressure data were edited for artifacts according to the following criteria.

Measurements recorded during the ambulatory period were stored on a personal computer and screened for editing of artificial values by applying previously described criteria.¹³ A 24-hour record was rejected for analysis if more than one third of potential daytime and nighttime measurements were absent (daytime minimum, 18; nighttime minimum, 8).¹⁴ The editing criteria¹³ that were considered removed <1.0% of the readings without any effect on the findings. The ambulatory blood pressure values are expressed as 24-hour average systolic and diastolic pressures. Each patient was then classified according to ambulatory blood pressure values (normal stage <132/81 mm Hg; stage I <140/88 mm Hg; stage II <148/94 mm Hg; and stage III >148/94 mm Hg). When systolic and diastolic pressures fall into different categories, the higher category was selected to classify the individual’s blood pressure status.

Patient Follow-Up
After the initial evaluation, at intervals a physical examination was performed. A comparison of the incidence of new cardiovascular events, fatal and nonfatal, between blood pressure groups was made during the follow-up. Patients who died of a noncardiovascular cause were considered to have been event-free until death. In subjects with multiple nonfatal events, the analysis included only the first event. Cardiovascular events included myocardial infarction, angina pectoris, coronary revascularization, arrhythmia (eg, atrial fibrillation), stroke, transient ischemic attack, peripheral artery disease, acute left ventricular failure, hypertensive crisis requiring hospitalization, and recurrence of aortic aneurysm.

Statistical Analysis
Data are presented as mean and standard deviation or 95% CI or number and percentage. The 24-hour mean values of ABPM measurements and the mean of 6 clinic blood pressure measurements either before or after 24-hour ABPM from each patient were used for the calculations. Systolic and diastolic blood pressure values were analyzed separately. To assess the association between ABPM and clinic blood pressure values, we used Pearson’s linear correlation. Linear regression analysis was used to quantify the association between clinic blood pressure and ABPM and the difference of the two methods. To assess the influence of the absolute blood pressure level on the difference between ambulatory blood pressure and clinic blood pressure, we plotted the difference between ambulatory blood pressure and clinic blood pressure against clinic blood pressure. According to the visual aspect of the data distribution, we calculated linear regressions of this association. The regression equations were used to calculate the cutoff values of ABPM corresponding to the clinic blood pressure values as defined by the JNC-VI and WHO guidelines.^1,2 We calculated the mean difference (ie, the bias) between ABPM and clinic blood pressure measurements. The limits of agreement were calculated as 1.96xSD of the mean bias. For statistical comparison of ABPM and clinic blood pressure values, the paired t test was used. Additionally, we used the ² test for trend to assess linear association between ambulatory blood pressure stages and proportion of cardiovascular events. Kaplan-Meier estimates were used to assess the probability of cardiovascular events for the different ambulatory blood pressure groups. Differences in probability of cardiovascular events were calculated by use of the log rank test. Data processing was performed with Microsoft Excel 97 for Windows and SPSS 7.5 for Windows. A 2-sided probability value <0.05 was considered statistically significant.
General Data
Overall, 736 (362 male) patients could be enrolled into the study. The average age of the patients was 55±14 years. Initially, 557 (75%; 270 male) of the enrolled patients were treated. The treatment consisted of commonly available drugs including ß-adrenergic-blocking drugs, calcium channel antagonists, ACE inhibitors, -adrenergic-blocking drugs, and thiazide diuretics alone or combined. The average duration of hypertension was 6.4±8.4 years. Laboratory values concerning renal function and serum electrolytes were within the normal range in all patients (serum creatinine, 1.01±0.21 mg/100 mL; blood urea nitrogen, 15.6±4.9 U/L; serum sodium, 140.5±5.8; serum potassium, 4.2±1.6). During follow-up in 442 (60%) patients, treatment was modified. Table 1 demonstrated the average decrease of systolic and diastolic blood pressures in each group during first year of follow-up.

Clinic and Ambulatory Average Blood Pressures
A total of 528 patients (72%) had systolic clinic blood pressure values 140 mm Hg, and 308 patients (42%) had diastolic clinic blood pressure values 90 mm Hg. The mean systolic as well as diastolic blood pressure values were similar before and after the 24-hour ABPM (148±14 versus 149±16 mm Hg; 87±9 versus 86±8 mm Hg). Mean ambulatory blood pressure values were 135±13 mm Hg and 79±10 mm Hg for systolic and diastolic blood pressure, respectively.

Twenty-four-hour ambulatory blood pressure and clinic blood pressure were associated with a correlation coefficient of 0.46 and 0.61 for systolic and diastolic blood pressure, respectively (P<0.0001 for both; Figures 1A and 1B). The linear regression coefficients were 0.405 and 0.307 and the intercepts 75.2 and 25.4 for systolic and diastolic blood pressure, respectively.

Figure 1. A, Relation of systolic clinic blood pressure and 24-hour average systolic ambulatory blood pressure. B, Relation of diastolic clinic blood pressure and 24-hour average diastolic ambulatory blood pressure.

The mean bias between ABPM and clinic blood pressure was -13.3 mm Hg (95% CI, -14.3 to -12.2; 1.96xSD limits of agreement, 15.7 to -42.3) and -7.3 mm Hg (95% CI, -7.9 to -6.6; 1.96xSD limits of agreement. 9.8 to -24.3) for systolic and diastolic blood pressure (P>0.0001 for both), respectively. The 95% CI at the cut-point of the regression line (ambulatory blood pressure, 132/82 mm Hg; clinic blood pressure, 140/90 mm Hg) was 116 to 148 mm Hg for systolic blood pressure and 70 to 94 mm Hg for diastolic blood pressure.

We found a linear relation between clinic blood pressure and the difference between the methods for systolic and diastolic blood pressure (Figures 2A and 2B). Using the above regression equations, we calculated the ambulatory blood pressure cutoff values corresponding to the recent recommended guidelines for clinic blood pressure, which are presented in Table 2. According to these calculations, stage 1 hypertension is defined from 132 to 140 mm Hg systolic and from 82 to 87 mm Hg diastolic ambulatory blood pressure and stage 2 hypertension from 140/88 to 148/94 mm Hg, respectively.

Figure 2. A, Relation between systolic clinic blood pressure and the difference between both methods. B, Relation between diastolic clinic blood pressure and the difference between both methods.

Frequency Distribution of Age and Average Blood Pressure Values in Different Ages
The distribution of age in the different blood pressure groups are demonstrated in Figure 3. No significant difference in the distribution of age are noted. In patients <65 years of age, the upper normal limit of ambulatory blood pressure was 132 mm Hg and 82 mm Hg for systolic and diastolic blood pressure, respectively. The cutoff values in patients >65 years of age were 132 mm Hg for systolic and 81 mm Hg for diastolic ambulatory blood pressure.

Frequency of Cardiovascular Events
The mean time of observation was 52 months, ranging from 6 to 96 months (median, 48 months). Overall, 82 (11.1%) patients had nonfatal clinical cardiovascular events and 9 (1.2%) patients died of cardiovascular causes. Death was caused in 4 patients by acute myocardial infarction, in 3 patients by cerebral infarction, and in 2 patients by cerebral hemorrhage. In 26 patients, causes for nonfatal clinical cardiovascular events were coronary heart disease, myocardial infarction, angina pectoris, and atrial fibrillation; in 15 patients, cerebrovascular disease, stroke, or transient ischemic attack; in 11 patients peripheral artery disease; in 28 patients, acute left ventricular failure and hypertensive crisis requiring hospitalization; and in 2 patients, recurrence of aortic aneurysm.

Ambulatory Blood Pressure Stages and Cardiovascular Events
According to the ABPM values 260 (35%), patients were assigned in normal (<132/81 mm Hg), 216 (29%) patients in stage I (<140/88 mm Hg), 131 (18%) patients in stage II (<148/94 mm Hg), and 129 (18%) patients were assigned in stage III (>148/94 mm Hg). The distribution of the nonfatal and fatal clinical cardiovascular events are demonstrated in Table 3. We found a linear association of increasing ABPM value and the number of cardiovascular events (P<0.006) (Figure 4). The Kaplan-Meier plot demonstrating the survival probability of the different ambulatory blood pressure groups is presented in Figure 5. We found a statistical trend toward a difference in survival probability between the ambulatory blood pressure groups within a mean observational period of 52 months (P=0.07).

Figure 4. Association of different ABPM stages and incidence of cardiovascular events.

Figure 5. Probability of cardiovascular events in different blood pressure groups (Kaplan-Meier curves).

Comparison of Clinic and Ambulatory Blood Pressure
Our study of 736 participants provides new information on the relation between clinic and ambulatory blood pressure obtained from a moderate to severe hypertensive population. First, the mean difference between ambulatory blood pressure and blood pressure assessed by a doctor in the clinic environment was significant at all blood pressure levels. Second, the mean difference between ambulatory blood pressure and clinic blood pressure increases with increasing blood pressure values. Whereas the mean difference of systolic blood pressure between both methods was 7 mm at the level of 135 mm Hg, this difference increases to 32 mm Hg at a level of 180 mm Hg. A similar pattern of increasing disparity between both methods was observed for diastolic blood pressure. Our findings extend previous findings that ambulatory blood pressure is significantly lower than clinic blood pressure even in patients above the normal values of 140/90 mm Hg. These results are in line with the data provided by the PAMELA study, which also reported an increasing difference between both methods, dependent on the actual clinic blood pressure in a normotensive population.^3,4

Because of the increasing difference between clinic blood pressure and ambulatory blood pressure, a direct conversion of ABPM results into clinical stages would be an incorrect procedure that may result in a lower staging of patients with severely elevated ambulatory blood pressure.

Comparison of Our Data With Previously Published Data
As our study population consisted of a high percentage of hypertensive individuals, our mean systolic and diastolic blood pressure was significantly higher compared with three other population studies. Whereas our mean ambulatory blood pressure was 135/79 mm Hg, the mean values of the previous studies were 118/74, 119/71, and 119/70 mm Hg, respectively.^3–5,15 Similar differences were observed concerning the clinic blood pressure. Despite these differences between our mean blood pressure values and the previously reported results, the normal value of the 24-hour ABPM assessed in our study population was similar to the PAMELA study and the results published by Staessen et al.^3–5 Our upper limit of normality for 24-hour ambulatory blood pressure is 132/81 mm Hg, which is similar to the data of a Belgian population (129/80 mm Hg),⁵ of the PAMELA study (128/82 mm Hg),^3,4 and of an international database (133/82 mm Hg).¹⁶ In contrast to these previous studies, our study population consists of normotensive as well as hypertensive subjects of all stages.

Definition of Corresponding Stages Between Ambulatory and Clinic Blood Pressure
Because >70% of our patients had a systolic value >140 mm Hg and >40% had a diastolic value >90 mm Hg, corresponding stages between clinic and ambulatory blood pressure over a wide range of blood pressure values above normality were assessed. The definition of corresponding stages between both methods provides some clinically relevant advantages:

First, patients classified as severely hypertensive by 24-hour ambulatory blood pressure measurement (>148/94 mm Hg) will no longer be classified as mild or moderately hypertensive, according to clinic blood pressure stages. Second, ambulatory blood pressure values can be used for treatment decision according to the recently published guidelines because the cutoff values obtained in this study correspond with the cutoff values recommended for the clinic blood pressure.

Prognostic Value of the Newly Defined Stages of Ambulatory Blood Pressure
The calculation of corresponding stages between ABPM and clinic blood pressure without the evaluation of the prognostic value is only of limited clinical relevance. Our data clearly demonstrate a significant association between the frequency of cardiovascular events and height of initial ambulatory blood pressure. Patients belonging to stage III assessed by ambulatory blood pressure had the highest frequency of cardiovascular events. The difference between patients with different stages of hypertension remained unchanged over the observation period of 5 years, as demonstrated by Kaplan-Meier curves. The risk for a patient belonging to the highest blood pressure group to have a cardiovascular event showed a trend to be elevated compared with patients with an ambulatory blood pressure below the upper normal limit.

These findings are in line with recently published data,^17–20 which demonstrated a correlation between ABPM and the extent of left ventricular hypertrophy or microalbuminuria. Both conditions are associated with an increased risk for cardiovascular events.^21,22 A former published study has demonstrated the prognostic value of ABPM.²³ The patients belonging to the highest tertile of 24-hour ambulatory pressure had the highest rate of cardiovascular events. However, in this former study, the stages for ambulatory blood pressure did not correspond to clinic blood pressure stages. Therefore, it seems rather difficult to use these cutoffs in daily clinical practice and to estimate risk for an individual patient.

Influence of Age
It must be mentioned that age is an independent prognostic parameter for subsequent cardiovascular events in all stages of hypertension. However, the distribution of age was similar in all four groups, indicating a similar effect of age on prognosis in each group. Moreover, average blood pressure as well as the upper normal limit do not differ between patients >65 years of age and younger people. Our data are in line with the report of O’Brien et al,¹⁴ who demonstrated similar blood pressure values in patients 50 to 79 years of age. We therefore conclude that age contributes equally to the frequency of cardiovascular events in all four groups.

Limitations
Some limitations of the study must be emphasized. First, ABPM is influenced by the diurnal rhythm, which may contribute to the discrepancy between clinic and ABPM. The reduction of blood pressure during the night may contribute to the lower level of ambulatory blood pressure compared with clinic blood pressure values, which were assessed in the morning. However, even home blood pressure measurements taken at different times of the day remained higher than ABPM.³ We therefore assume that the diurnal rhythm of blood pressure is only a minor contributor to the discrepancy between clinic and ambulatory blood pressure. Second, most of the patients included in this study were receiving active treatment. Drug treatment influences the course of blood pressure, dependent on the pharmacodynamic and pharmacokinetic properties of the drug. Antihypertensive drugs with a marked peak-to-through ratio may contribute to lower mean ambulatory blood pressure values compared with clinic blood pressure values obtained in the morning. However, this discrepancy between both methods was also observed in normotensive as well as nontreated hypertensive patients. Third, the modification of antihypertensive drug treatment after the initial evaluation has influenced the frequency of subsequent cardiovascular events. A point of criticism may be that the initial ambulatory blood pressure may have only limited prognostic value. However, by analyzing the event rate over time, a significant disparity between the groups remained despite the most pronounced decrease of clinic blood pressure in stage III patients. We therefore assume that initial ambulatory blood pressure remained an independent prognostic parameter. Our data are in line with the results of the MRFIT study, which also demonstrated a prognostic impact of the initially measured clinic blood pressure on cardiovascular events despite subsequent therapeutic interventions.²⁴

In conclusion, ambulatory blood pressure is significantly lower than clinic blood pressure, even in patients with moderate and severe hypertension. The disparity between both methods increased with increasing clinic blood pressure values. Nevertheless, with ABPM, different stages of hypertension according the recent guidelines of clinic blood pressure could be identified.

Perspectives
The staging of hypertension by ABPM may facilitate the use of this method in daily clinical practice, as the 24-hour ambulatory blood pressure values can now be used not only to confirm the diagnosis of hypertension but to assess the severity and prognostic value of hypertensive disease.

Andreas Bur; Harald Herkner; Marianne Vlcek; Christian Woisetschläger; Ulla Derhaschnig; Michael M. Hirschl

 Hypertension. 2002;40:817 

Hypertension Staging Through Ambulatory Blood Pressure Monitoring

This issue of Hypertension includes an article by Bur et al¹ that focuses on the comparison between clinic and ambulatory blood pressure (ABP) values in patients with moderate to severe hypertension. The primary goal of the Bur study was to obtain a classification of hypertensive patients, based on the ABP values corresponding to the clinic blood pressure (BP) values that have been used to stage hypertension by the World Health Organization–International Society of Hypertension (WHO-ISH) and the Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure (JNC) VI guidelines.^2,3 An additional goal was to evaluate whether this ABP-based classification has prognostic value, as shown for the prognostic value of the clinic BP staging. Both clinic BP and ABP were measured in 736 hypertensive patients (557 of whom were under treatment) at the time of their first admission to the local Hypertension Unit. All patients then entered a follow-up period with an average duration of 52 months (range, 6 to 96 months), during which only clinic BP was obtained. During the observation time, 82 patients had nonfatal cardiovascular events and 9 patients died of cardiovascular causes.

The article adds interesting information to the existing database on the clinical value of ABP. In particular, it contributes to the available knowledge on the prognostic importance of ABP as well as on its relation to clinic BP in the context of treating patients in a hypertension center.⁴

Stratifying patients into different risk categories on the basis of ABP values requires studies that (1) establish in populations or in large groups of hypertensive patients the relation of cardiovascular morbidity and mortality with the different 24-hour ABP values selected^5,6 and (2) evaluate how prognosis of patients is modified when ABP is reduced by treatment, leading to a change in the ABP-based staging. This information is only partly available, however, because the association between the incidence of cardiovascular disease and ABP has been examined in only a few studies of suitable size.⁷ Furthermore, the few intervention studies addressing the prognostic value of treatment-induced changes in ABP have been undermined by problems such as a low number of BP measurements during ABP monitoring, use of surrogate end points rather than morbid or fatal events, uncontrolled experimental designs, small numbers of patients (and thus insufficient statistical power to back the study conclusions), and lack of ABP measurements during antihypertensive treatment.⁶ The available evidence clearly indicates that the upper limits of normal 24-hour average ABP are markedly <140/90 mm Hg. However, because of these problems, it does not provide any classification of hypertensive patients on the basis of ABP levels that might parallel the staging of hypertension, based on clinic BP, proposed by WHO-ISH and JNC VI guidelines.^2,3

Bur et al¹ have made an attempt to cope with at least some of the above problems. The results of their study confirm previous findings that the relation between clinic BP and ABP, although statistically significant, is by no means close. They also confirm the observations made in previous studies that ABP is significantly lower that clinic BP, especially in moderate and severe hypertension,^4–7 because the discrepancy between clinic BP and ABP increases with increasing clinic BP values. Based on these findings, and consistent with previously published recommendations,^2–7 the authors correctly emphasize that the classification of hypertension by WHO-ISH or JNC VI clinic BP criteria should not be directly transferred to ABP data. Finally, and most importantly, the results of Bur et al confirm and extend previous observations on the prognostic value of ABP by showing that staging hypertension by ABP values is indeed related to prognosis of hypertensive patients.

It should be acknowledged that it is difficult to carry out controlled studies on the prognostic value of different ABP levels. This is because, as mentioned above, the study size needs to be large and the observation period to be prolonged in order to obtain a sufficiently high number of events that permit conclusions with high statistical power. In particular, if the purpose is not just to show the prognostic value of ABP (a rather likely finding) but whether ABP is prognostically superior to or adds to the prognostic value of clinic BP, the study size and duration need to be substantially increased. Finally, to assess whether cardiovascular protection depends more on treatment-induced reduction in ABP than in clinic BP (or whether knowledge of ABP reduction by treatment adds to the estimate of protection based on clinic BP reduction), additional requirements need to be fulfilled. These include (1) frequent collection of ABP data during treatment, (2) need to avoid excessive treatment nonhomogeneity, and (3) physicians’ different attitudes and biases on optimal ABP levels to be achieved.

This has not been done in the study by Bur et al.¹ First, the patients included during the 52-month follow-up period had a limited number of cardiovascular events. This problem was made more serious by the fact that the attempt to stage patients on the basis of ABP values required their subdivision into subgroups, which showed only few events and often trivial event differences. Second, ABP was obtained (as it was the case for previous studies) only at the time of the initial evaluation, which was followed in the next 52 months by a modification of antihypertensive treatment, because clinic BP showed on average a substantial reduction. This does not detract from the authors’ conclusion that the ABP values originally obtained at the initial evaluation had a prognostic significance.¹ It does not allow, however, clarification of whether determination of this prognostic significance was modified by the new values subsequently obtained after treatment changes, which is an important issue, given the evidence on the prevailing prognostic importance of BP values achieved by treatment.^8,9 This is especially important because selection of the treatments was unrestricted and the patients were given different drugs or drug combinations, which may have been responsible for differences in cardiovascular protection beyond those accounted for by BP reductions.^10–12 Third, identification in the study by Bur et al of normal ABP values based on the regression between clinic BP and the corresponding clinic BP-ABP difference is open to criticism. This is not because of the use of their specific statistical approach, which in previous population studies has led to valuable data.⁴ It is because the present study involved mainly patients under antihypertensive treatment (75.7%), in whom the differences between clinic BP and ABP may have reflected a differential effect of antihypertensive drugs on the two pressures. This possible explanation is supported by the fact that different drugs have different trough-to-peak ratios and by the evidence that usually the effects of antihypertensive treatment are more pronounced on clinic BP than on ABP.^13,14

However, the authors should be given credit for their finding that in the hypertensive patients in whom clinic BP was in the lowest WHO/ISH or JNC VI stage, 24-hour ABP was similar or only slightly above the normal values calculated for population studies. This means that in this range, we can expect a certain degree of correspondence between these two different BP estimates. This is an important observation, although it is valid only for average group data and not for individual patients.

Two final comments deserve to be made. First, evidence is available that the prognostic power of ABP values is a function of their reproducibility,^15,16 which means that averaging data from two ABP recordings may improve it. This methodological issue has not been addressed in the present study, in which initial clinic BP was measured in two different occasions, whereas ABP monitoring was obtained only once. Second, it is not clear in the study by Bur et al whether ABP staging carries an additional prognostic contribution over and above that classically provided by clinic BP. As mentioned above, this is of critical importance because the dilemma we face is not to abandon clinic BP in favor of ABP but to determine whether information on ABP refines the risk assessment made possible by clinic BP to an extent that justifies the additional cost of the ABP monitoring procedure. This could have been addressed by Bur et al, however, by comparing the Kaplan-Meier curves on cardiovascular events based on clinic BP and on ABP or, alternatively, by assessing the prognostic value of ABP after adjusting for the information provided by clinic BP.

Despite these problems, the effort made by Bur et al to provide us with indications on how to classify the severity of hypertension based on ABP data does represent an important step toward the use of ABPM in the prognostic stratification of hypertensive patients, although the data obtained in this study are not robust enough yet to recommend changes in current treatment strategies.

Additional studies are still needed to this aim, including longitudinal investigations on the predictive role of ABP values obtained during a follow-up period in a large sample of patients and intervention studies that might provide evidence on a reduction in cardiovascular risk after treatment-induced reductions in ABP levels. This could be achieved by including serial ABP recordings in clinical trials, either in large longitudinal population risk factor studies, and in large mortality trials addressing the benefits of antihypertensive therapy.

Until the data are available, use of ABP measurements in the routine treatment of hypertensive patients should still be recommended only in selected cases, as a complement to home blood pressure measurements, and to repeatedly and carefully obtained clinic BP readings.

Gianfranco Parati; Giuseppe Mancia

Masked Hypertension

The addition of ambulatory blood pressure monitoring to conventional clinic measurement for defining blood pressure status in clinical practice has added a new complexity to the process, because the separation of normotension and hypertension can be assessed independently by each of the 2 methods. We thus have 4 potential groups of patients who are, first, normotensive by both methods (true normotensives); second, hypertensive by both (true, or sustained, hypertensives); third, hypertensive by clinic measurement and normotensive by ambulatory measurement (white-coat hypertensives); and, fourth, normotensive by clinic measurement and hypertensive by ambulatory measurement. From a clinical point of view, the first 2 groups are easy to deal with, because both methods give the same classification. Of more interest are the groups in which there is disagreement. The third group, usually referred to as white-coat hypertensives, or less frequently, as isolated office hypertensives, have been extensively studied and are generally accepted as being at relatively low risk of cardiovascular morbidity,¹ a view consistent with the concept that ambulatory pressure gives a better prediction of risk than clinic pressure.

Until now, little attention has been given to the fourth group, whose condition has been given the awkward titles of "reverse white-coat hypertension" or "white-coat normotension." If it is true that the ambulatory pressure gives the better classification of risk, it would imply that these people should be regarded as being genuinely hypertensive, as argued below. We also propose that the phenomenon might be called "masked hypertension," on the grounds that the hypertension is not detected by the routine methods. "Undetected ambulatory hypertension" is another possible title. But what evidence is there that this group deserves recognition as a discrete entity, as opposed to being made up of people who happened to have an unusually high ambulatory pressure or a low clinic pressure on that particular occasion? There are potentially several questions that could be asked to decide this issue. First, the phenomenon of masked hypertension would be more credible if it could be shown that it is reproducible on repeat testing. As far as we are aware, this issue has not been examined. Second, patients with masked hypertension should show more extensive target organ damage than true normotensive subjects. Here we are on surer ground. The first study to look at this issue was our publication of 1999,² in which we showed that the masked hypertensive group had left ventricular mass and carotid atherosclerosis that were greater than that of true normotensive subjects and that were similar to true hypertensive subjects. The left ventricular mass index was 73 g/m² in the true normotensive subjects, 86 g/m² in the masked hypertensive subjects, and 90 g/m² in the true hypertensive subjects. Carotid plaque was present in 15% of true normotensives and in 28% of both the masked and true hypertensives. More recently, an analysis of the PAMELA data,³ a population study of 3200 Italians, classified the subjects in the 4 groups that we have described above. Individuals with treated hypertension were excluded from this analysis; 67% were true normotensives, 12% true hypertensives, 12% white-coat hypertensives, and 9% masked hypertensives. The average clinic pressure in the masked hypertensives was 129/84 mm Hg, which, although still within the normal range, was higher than the true normotensives (112/77 mm Hg). The left ventricular mass index was higher in the masked hypertensives (91.2 g/m²) than in the true normotensives (79.4 g/m²) and similar to the true hypertensives (94.2 g/m²). A third issue is whether masked hypertensives are at increased risk of cardiovascular morbidity. This remains to be determined.

What factors might lead to masked hypertension? In principle, there are 2 groups of factors, which are not mutually exclusive. First, the clinic pressure could be relatively low in relation to the ambulatory pressure, or second, there could be factors that selectively raise the ambulatory pressure. With regard to the first possibility, it is generally true that the daytime ambulatory pressure is higher than the clinic pressure in truly normotensive subjects, but in hypertensives the clinic pressure tends to be higher. One reason for this is "regression to the mean," because hypertension status is almost always based on clinic pressure. Many factors could selectively elevate the ambulatory pressure. For example, we showed many years ago that smokers tend to have high daytime ambulatory pressures (when they are likely to be smoking) in comparison with their clinic pressures (when they are not likely to be smoking).⁴ Second, subjects who are more physically active during the day will tend to have higher daytime pressures.

Several population studies have compared clinic and ambulatory blood pressures.^5–7 Some have shown daytime pressures to be a little higher than clinic pressures, whereas others have found the reverse.⁸ One important finding from an Italian population study has been that the ambulatory pressure shows much less increase with age than the clinic pressure.⁷ In a Danish study,⁸ 86% of men 42 years of age had daytime pressures higher than the clinic pressure, whereas this was true of only 51% at the age of 72 years. The white-coat effect (the difference between the clinic and ambulatory pressure) is hence more marked in older people, and because masked hypertension is equivalent to a negative white-coat effect, it is reasonable to suppose that masked hypertension would be less prevalent with increasing age.

A major issue concerns the prevalence of masked hypertension. Although there are no definitive data, the available information is disturbing. In a study of 319 clinically normotensive volunteers, all of whom had 5 clinic measurements and 12-hour daytime ambulatory blood pressure measurements, Selenta et al⁹ found that 23% had masked hypertension, defined as a daytime blood pressure >135/85 mm Hg. Subjects with masked hypertension tended to be male, past smokers, and older, and they had consumed more alcohol. The issue of masked hypertension was also discussed by Belkic et al,¹⁰ who referred to it as occult workplace hypertension. We found that 36 of 267 men (13.5%) in the Cornell Worksite study had masked hypertension, defined as a daytime ambulatory diastolic pressure >85 mm Hg and a clinic pressure <85 mm Hg. Two population-based studies have also described the phenomenon. The first was the Ohasama study,¹¹ conducted in a small Japanese town, which reported that 10.2% of subjects with normal screening blood pressures had ambulatory pressures that were in the "borderline hypertensive" range (>133/78 mm Hg for 24-hour average) and another 3.2% in the definitely hypertensive range (24-hour blood pressure >144/85 mm Hg). The second was the PAMELA study quoted above, which found it in 9% of subjects.³ But even if the prevalence was only 5%, this number applies to the whole adult population, not just the population of hypertensives, so in the case of the United States, this might amount to 10 million people.

It seems clear that masked hypertension should be taken seriously and is a phenomenon worthy of further investigation. If it is accepted that ambulatory blood pressure gives a better prognosis than clinic blood pressure and that the correlation between the two is only moderate, it is logical to propose that there will be a significant number of people who are truly hypertensive but in whom the diagnosis is missed by clinic measurement. But how frequently this phenomenon occurs, and how such individuals should be identified, remains a mystery. Clearly, we cannot argue for screening of the general population, but there are many patients who are referred for suspected hypertension who have normal clinic pressures on repeat testing. Individuals in whom the level of suspicion might be heightened would include those who have a family history of hypertension or other risk factors such as central obesity. Perhaps some of them would benefit from ambulatory monitoring to rule out masked hypertension. Although treatment recommendations may be premature, the finding of masked hypertension in a patient with early signs of target organ damage might act as an incentive to promoting lifestyle changes