Diagnostic certainty in hypertension

The relationship between blood pressure and the risk of future cardiovascular events is continuous [1]: the higher the blood pressure, the greater the chance of adverse events. But what exactly do we mean by blood pressure in this context? Blood pressure changes rapidly depending on a multitude of factors, including the level of blood pressure measured [2], as well as the examiner, situation, posture, emotion, habituation and climate, to name but a few [3–6]. Having taken one or two readings in the office on a single occasion, what does that portend for the next occasion? In this issue of the journal, Chiu et al. [7] provide novel and potentially useful answers to this question faced by physicians, their patients and epidemiologists alike. Our knowledge of the adverse effects of hypertension is largely based on studies where blood pressure has been taken by either nurses or physicians in an ‘office’ setting in the seated position, often repeatedly and on different days [8–12]. Consequently, current guidelines for the diagnosis and classification of hypertension are structured along the same lines, requiring confirmation on a different day [12]. This is a luxury commonly lacking in epidemiological studies where, for cost and practicality reasons, subjects can only be assessed on a single visit. It also does not solve the more common problem of understanding the meaning of a first elevated blood pressure reading taken in the office and further communicating this meaning to our patients in the light of uncertainty. A doctor's visit, an investigation or, more generally, the unknown presumably cause a heightened state of anxiety, which in turns raises blood pressure [6]. Figure 1 demonstrates this well-known effect. During ambulatory blood pressure monitoring, the first few readings are usually the highest, with a small but measurable rise upon the return of the device to the office. Thus, situational habituation does occur, but does not disappear completely and, even when this is the case, it does so to an individually varying degree. ‘White-coat’ hypertension, where blood pressure is only elevated in the office setting, but is normal during ambulatory conditions, confers a level of risk between consistently normotensive and consistently hypertensive subjects [13]. In placebo-controlled trials, blood pressure is known to decrease even in the placebo arm, despite an expectation that it should increase with time due to ageing [14]. Again, this is presumably a combination of situational habituation and regression to the mean from a pre-defined study entry cut-off value. In the recently published vascular substudy of the Australian National Blood Pressure Study 2, blood pressure taken by dedicated study nurses in general practice was higher than later during vascular assessment [15]. The diastolic blood pressure difference was lower in those individuals who had an adverse event during 4 years of follow-up than those who remained event-free (4 ± 11 versus 6 ± 10 mmHg, P = 0.003). This provides further support to the notion that the risk for subjects with excessively elevated office blood pressure (e.g. as in ‘white-coat’ hypertension) is somewhere in between.Fig. 1Hypertension acts independently from, but in concert with, other risk factors to cause target organ damage and subsequent mortality and morbidity. The JNC 7 guidelines identify a group of subjects who are at high risk of developing hypertension, so-called pre-hypertension [12]. The JNC 7 recommendations suggest that, in this group, drug therapy should only be initiated in subjects who also suffer from diabetes and/or kidney disease. The 2003 ESH/ESC guidelines take a differently structured approach to treatment and future surveillance. In particular, in the pre-hypertensive group (‘high–normal’ in ESH/ESC), the frequency of future surveillance is contingent on the number of risk factors and associated clinical conditions [16]. Chiu et al. [7] clarify this message and provide further scientific rationale for this approach. If only two measurements of blood pressure on a single occasion are available, what is the likelihood of this patient subsequently developing hypertension? Chiu et al. [7] demonstrate that, having taken exactly the same blood pressure reading, the presence of other coexisting risk factors makes a subsequent diagnosis of hypertension (transition) more or less likely. As an example, for a subject found to be in the pre-hypertensive range of blood pressure, being male or having a family history of hypertension increased the likelihood that he would be found to be hypertensive on the next occasion. At the same time, it decreased the likelihood of becoming normotensive [7]. These two factors are reasonably intuitive. Given the recruitment of a predominantly normotensive and pre-hypertensive population, the power of the study is greater in relation to the normal to pre-hypertensive transition and vice versa. A most interesting and significant new finding is the detrimental effect of current betel nut consumption, which had the single biggest effect of all the dichotomous variables evaluated in the study. This needs to be confirmed by a more targeted study. Chiu et al. [7] further identified factors that increase occasion-to-occasion variability rather than being directional, such as low high-density lipoprotein cholesterol, which increases the likelihood of transition in both directions. This is of relevance because blood pressure variability (albeit only at night) has been identified as an independent risk factor [17]. The results obtained by Chiu et al. [7] may further suggest a novel approach to the ‘white-coat’ effect, concentrating more on associated risk factors that increase the pre- and post-test likelihood of underlying true hypertension rather than the conventional approach that relies on out-of-office measurements, or alternatively evaluation for target-organ damage [3–5]. A single visit measurement, as performed in the study by Chiu et al. [7], even when taking the lower of two readings, would overestimate prevalence of hypertension if it was compared with the lower reading of more than two readings obtained on multiple visits [2]. Hence, when evaluating the study by Chiu et al. [7], comparing it with NHANES II is a logical choice. Both studies used a fairly similar methodology for blood pressure measurement with two readings obtained in the seated position. Table 1 compares prevalence by age and gender from NHANES II [18] and the study by Chiu et al. [7]. Despite being separated by location, race and time, the similarity in prevalence rates is striking. Chiu et al. [7] only report the proportion of currently hypertensive patients who are also receiving treatment, and similarly with respect to awareness. It would be hoped that effective treatment to targets achieves at least pre-hypertension or even normotension in an individual who was initially hypertensive. Hence, to compare whether the level of awareness and treatment was similar in both studies would require knowledge of how many subjects in the normotensive and pre-hypertensive groups are either aware of or are treated for hypertension. Such data has not been reported in the paper by Chiu et al. [7]. Notwithstanding its limited relevance, hypertensive subjects in NHANES II reported treatment at a similar percentage when compared with those in Chiu et al. [7,18,19]. It is hoped that, at some stage, Chiu et al. [7] will make their data publicly available (as the NHANES data already are) so that such questions can be answered with respect to their important and very large data set.Table 1: Prevalence in the percentage of pre-hypertension and stage I and II hypertensionaNHANES II was conducted almost three decades ago. Subsequently, blood pressure control has improved substantially in the USA, with each subsequent survey showing increased rates of awareness, and most importantly treatment, with a substantial progressive reduction in age-adjusted average population blood pressure [20,21], particularly in the white population [22]. In their discussion, Chiu et al. [7] compare their data with the NHANES 1999–2000 data [23], which is presented in Table 1 using comparable methodology (i.e. using the lower of the first two blood pressure measurements only). Even then, it would appear that the prevalence described by Chiu et al. [7] is more comparable with NHANES II than it is with the data for 1999–2000, at least in the non-Hispanic white population. This similarity with NHANES II, rather than the more contemporary data, would suggest that the cohort studied by Chiu et al. [7] is lagging many decades behind the USA. It further suggests a considerable potentially preventable future burden of cardiovascular disease in Taiwan with all its personal, social and economic implications.