Measuring blood pressure: who and how?

The measurement of blood pressure (BP) is often described as the commonest procedure carried out by doctors in the care of patients. The community-based BP measurement is usually performed to diagnose or manage hypertension, a condition with at least 30% prevalence in the Western world. It would seem unnecessary, therefore, to question the importance of careful and accurate attention to detail when carrying out this ‘simple’ task and there are many published guidelines that describe in great detail how this should be done [1]. It is a moot point, however, whether most doctors, in their busy practice, are as careful as they could be in making these recordings. Nonetheless, we have known for more than half a century that even a casual measurement of BP predicts cardiovascular outcome in large populations [2]. It is worth reflecting on a number of ‘truisms’ around the measurement of BP. The term hypertension is an arbitrary concept and it is only in recent times that we have come to accept that the risks of a given level of BP are very different in say a 75-year-old man with type 2 diabetes than they are in a 30-year-old woman with no other health problems. The recording of BP therefore, though important, should be seen as part of a package in assessing vascular risk. There exists a marked intrinsic variability of BP; such that even if an observer is as careful as possible in following published guidelines, regarding measurement, the value obtained will not be the same the next time the BP is recorded [3,4]. As a result, the certainty of labelling an individual as truly having hypertension is open to question [5]. It is not uncommon for BP to be much higher in a doctor's office than in the ‘out of office’ setting, the difference being referred to as ‘the white-coat effect’ and there is a large and increasing body of information indicating that the ‘out of office’ BP, whether recorded via ambulatory BP monitoring (ABPM) [6] or self-BP monitoring (SBPM) [7], is a better predictor of outcome than that measured by a doctor in the office (data are now available from many thousand patients). Most would now accept the view that ‘white-coat hypertension’ (elevated office BP with normal ABPM or SBPM) is considerably less risky than sustained hypertension both in and out of the office [8]. This author has always believed (assumed) that the reason for this may be no more complicated than that the average of a large number of BP measurements represents a more reproducible measure than any one or two office recordings. In the accompanying article by Woodiwiss et al.[9] this assumption is challenged. The authors hypothesized that a single trained observer measuring BP under strictly standardized conditions might record a BP that could equally predict the presence of a wide range of markers of target organ damage as a standardized ABPM. The same nurse using a mercury sphygmomanometer measured BP in all 458 patients. All had had the procedure previously described and had rested for 5–10 min, in the sitting position, in a quiet room, away from onlookers. BP was then measured (to the nearest 2 mmHg, diastolic phase V) five times consecutively, 2–3 min apart. All patients then underwent a 24 h BP monitoring procedure (Spacelabs 90207 – arguably the most widely used ABPM in the world) on the same day as the office BP. Measurements were made at 15 min intervals from 0600 to 2200 h and at 30 min intervals from 2200 to 0600 h. Day/night times were analysed either as the actual time spent in and out of bed or by using the mean sleep/wake cycle for the group; results being weighted for the time interval between measurements. The various surrogate markers of outcome were as follows: left ventricular dimensions and mass index (echocardiography), urinary albumin excretion and elements of pulse wave analysis. Most of the patients were female (64%) and 40% were defined as having hypertension with only a half of these on therapy. Some of the results were as expected. Office BP was significantly higher than day, night or 24 h BP whether the average of all five office readings, the average of three or only the first reading was considered. For the average of three office readings, the mean difference from the daytime ABPM was 8.7/6.8 mmHg. The SBP difference increased to 18 mmHg when office BP was above the group mean, and all differences were broadly the same, however, the daytime ABPM was calculated. Such differences are in line with many other similar comparisons [10]. The main results from the study lay in the relationships (correlations) between the various BP measurements and the different target organ markers. The bottom line was that the office systolic BP (less so for diastolic BP) was as good as the ABPM in predicting abnormality in most areas but, even more dramatically, this seemed to apply not only when an average of all five office readings was used but also for an average of three and even, simply, the first reading itself. Despite the plethora of earlier publications to the contrary, these authors have shown that under the specific conditions of this study a single office BP gives all the BP information that is needed for predicting outcome and therefore, presumably, for managing individual patients with hypertension. Is this plausible? There can be little doubt that nurses take more care in the measurement of BP than doctors and the white-coat effect is strikingly less apparent [11]. Some have even gone so far as to suggest that doctors should stop measuring BP entirely [11] – a view with which this author has a lot of sympathy but it is still hard to understand how the results in this study could have occurred. The authors themselves are at great pains to emphasize that their study is unique in that a single trained observer was responsible for every BP measurement. The patients were carefully prepared beforehand and rested for a long time prior to BP measurement. It is also of note that most were women and most did not have raised BP. All of this leaves us with the inevitable question as to whether these results can be extrapolated to men and to all patients with hypertension (the authors do indicate that the results were similar for untreated individuals and those with higher BP levels, although no data are presented). I am not sure whether it would have impacted on the outcome but we are not given information on those with white coat or masked hypertension (it is likely that a significant number of those with office normotension will have fallen into this latter category). What, if any, are the implications of this study for the routine practice of the management of hypertension? It is clearly not practicable (even in a single centre) to have all BPs measured by the same individual in perpetuity; it is certainly not possible to make every patient rest for 15 min and then measure five readings over a further 15 min. As is often the case, the rigors of a clinical trial are unlikely to be replicated in the ‘real world’, although even within the confines of an outcome trial, patients having repeated office readings will move from categories that would be labelled hypertensive to unequivocal normotension (and vice versa) despite only receiving placebo therapy [12]. We cannot afford, therefore, to let one study (even if there is support from other smaller trials) overly influence our routine practice. In their introduction section the authors acknowledge the wealth of data that indicate the primacy of out of office BP in predicting risk but at the same time they also emphasized the impracticability of this technology (ABPM) for the routine management of patients. If office BP is so fallible in routine clinical care, what is the answer? Mercury manometers will soon disappear due to concerns regarding the environment, aneroid devices are not recommended due to the difficulties of maintaining accuracy and there can be little doubt that the future lies in the use of increasingly inexpensive electronic devices (largely measuring BP oscillometrically) [13]. At the most basic level such devices remove the potential for error introduced by the observer and they represent a remarkably cost-effective way of obtaining large numbers of BP readings [14]. As indicated before, the large variability of BP makes it difficult to interpret readings taken at different times given that the natural variation can be greater than the anticipated effect of a drug. For the best office BP measurement, the coefficient of variability of repeated measures is of the order of 6% (unpublished data from this department on an analysis of the placebo arm of the MRC treatment trial of mild hypertension [15]) and for routine hospital clinics this can be doubled [16]. To allow confidence in practice, we may need to average 15–20 readings and this can only be realistically achieved by patients themselves using a home monitoring device [14]. A timely meta-analysis along with an accompanying editorial has recently emphasized the pros and cons of such an approach [17,18] but, whatever the views of the medical profession may be, the patient population is already voting with between 60 and 85% of patients with hypertension owning monitors worldwide [19]. We should embrace this wave of enthusiasm and channel it appropriately rather than remain with our outdated current Luddite practices!