Key points are not available for this paper at this time.
Design
Editorial
Answers are grounded in this paper and cite their sources.
While standardizing blood pressure measurement procedures is crucial for epidemiological assessment, initial readings should not be discarded as they capture important alerting reactions and form the basis of historical cardiovascular risk data.
Blood pressure is nothing if not variable [1]. All clinicians know this, many patients are confused by it, and the perennial monograph Clinical Hypertension, devotes an entire chapter to the seemingly simple matter of blood pressure measurement [2]. The substantial within-individual variability in blood pressure presents particular problems for investigators because such variability denotes poor reliability and dubious validity. Taking multiple blood pressure readings certainly improves measurement stability. Indeed, authoritative clinical guidelines [3,4] and recent epidemiological studies [5,6] base the diagnosis of hypertension upon the average of two to three blood pressure readings obtained at each of two assessments. Notably, however, the bulk of the prospective evidence linking elevated blood pressure with cardiovascular sequelae (evidence that forms the basis of all hypertension research and clinical interventions) was generated in studies using only one or two blood pressure readings [7,8]. Because the objective is to estimate ‘usual’ blood pressure and blood pressure fluctuates with time of day, activity and psychological state, much research now utilizes automatic ambulatory blood pressure monitoring. Besides providing many measurements, ambulatory monitoring obtains blood pressure readings in the subject's natural environment. Moreover, accumulating evidence suggests that ambulatory blood pressure is better than office or clinic readings in predicting cardiovascular disease risk [9–11] and benefit from anti-hypertensive treatments [12,13]. This literature has also shown that ambulatory blood pressure is lower than office or clinical blood pressure by 5–15 mmHg [14]. Against this backdrop, in the current issue of the journal, Bovet et al. [15] describe the results of their hypertension screening program in which they obtained repeated measures of blood pressure, taking three blood pressure measurments on four separate visits [15]. The study was conducted in a large number of poor residents living in Dar es Salaam in Tanzania. Blood pressure fell with successive readings, and dropped by an average of 15 mmHg systolic and 8 mmHg diastolic from the first reading on the first visit to the third reading on the fourth visit. Furthermore, the authors observed that hypertension prevalence declines as more readings and more visits are used. A modest decline in blood pressure with repeated measurement is a widely recognized phenomenon, presumably reflecting habituation to the experience of undergoing blood pressure measurement. In this study, the fall may have been exaggerated due to the novelty of blood pressure measurement in this population. Also, some subjects were selected for participation on the basis of blood pressure elevation, such that the fall in blood pressure across successive readings may reflect, in part, regression to the mean. In their discussion, Bovet and colleagues review previous studies that also described the effects on hypertension prevalence estimates of multiple, as opposed to single, visits for blood pressure measurement. In as much as elevated initial blood pressure readings might erroneously label some individuals as hypertensive, the authors suggest that, ‘… prevalence estimates based on only the 3rd reading – or the average of the 2nd and 3rd readings – at the second visit would be more appropriate to reliably estimate the true prevalence of hypertension in the population'. Although reasonable at first glance, such a recommendation raises a number of issues. First, it espouses the use of only one or two blood pressure readings. By not taking into consideration multiple readings across two visits, this suggested approach would put epidemiological methods at odds with current clinical guidelines for hypertension detection. This is not to say that epidemiological studies have always used consistent methods or adhered to clinical approaches. For example, in the National Health and Nutrition Examination Survey in the USA, adults were examined at each of four separate surveys during the early 1960s, early 1970s, late 1970s and late 1980s [5]. Protocols for blood pressure measurement varied from one survey to the next. Additionally, the threshold for defining hypertension changed from 160/95 to 140/90 mmHg during that period. Therefore, when comparing prevalence of hypertension longitudinally, differences in both methodology of obtaining blood pressure, as well as definition of hypertension, need to be taken into account. The dichotomous classification of individuals as either normotensive or hypertensive gives an approximate estimate of the prevalence of disease burden for a particular population. These estimates can have substantial social and political impact on the allocation of healthcare resources. Therefore, any change in assessment method must be carefully justified and, in this case, it is clear that hypertension prevalence figures will decline if more blood pressure readings are employed. Finally, we should be reluctant to discard initial blood pressure readings because, as stated, the essential link between elevated blood pressure and cardiovascular sequelae derives largely from studies measuring blood pressure once or twice at a single visit. Here, we need more empirical evidence showing that subsequent blood pressure readings are at least as powerful as initial readings in predicting cardiovascular events [16]. Furthermore, the initial blood pressure reading captures each individual's blood pressure response during an alerting reaction. Individuals likely differ in the magnitude of this response, and such acute blood pressure responses to changes in environment are of some interest. An exaggerated blood pressure reaction to psychological stress (reliably measured in the laboratory), as well as the degree of variability in blood pressure, may convey independent risk for cardiovascular disease [17,18]. The related phenomenon of white coat hypertension also comes to mind, and patients whose blood pressure rises during clinical examinations may experience increased cardiovascular disease [19,20]. During the last 40 years, we have changed our procedures for both determining the level of blood pressure (types of manometers, numbers of repeated measurements), and for defining hypertension (based on level of blood pressure). Furthermore, the newer automatic instruments now in use may not be comparable to the readings from the mercury manometers. In any case, we agree with the authors’ recommendation to standardize procedures for epidemiological assessment of hypertension across populations, but would not discard initial blood pressure readings. In any case, several cases remain unresolved. Will new blood pressure data, whether taken in the home or in a clinic by the various new devices, be comparable to the old data using different instruments and different methods? Will this new data be comparable to that of older studies in terms of cardiovascular risk assessment? Finally, given that yet other methods and techniques in blood pressure assessment are likely to appear, might we be trying to shoot at a moving target?
No takes yet. Share an insight, caveat, or question.
Muldoon et al. (2003) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: