The need for repeated blood pressure measurements
Blood pressure in humans is characterized by continuous and pronounced variations over time due to the complex interplay of several different mechanisms . This has led to the recommendation that blood pressure should be repeatedly measured, before taking any diagnostic or therapeutic decision in hypertensive patients. The recent 2003 European Society of Cardiology (ESC) and European Society of Hypertension (ESH) Guidelines for hypertension management , emphasized that this recommendation should apply to all available blood pressure monitoring techniques, including ambulatory blood pressure monitoring (ABPM) over 24 h, conventional blood pressure measurements in the physician's office and self blood pressure monitoring at the patient's home (HBP). This recommendation has been consistently supported by the evidence that the availability of a large number of blood pressure readings carries important advantages in the clinical management of hypertensive patients, at least when making use of ABPM  and when considering blood pressure measurements obtained in the physician's office or in the clinic [5,6].
The importance of obtaining a high number of blood pressure readings should in theory apply also to HBP monitoring (HBPM). This approach is being increasingly used not only in clinical research, but also in daily practice as a supplementary source of information for the diagnosis of sustained and white-coat hypertension and for the long-term follow-up of blood pressure control by treatment [6,7]. Indeed, the possibility to easily obtain repeated blood pressure measurements over time is regarded as one of the most important advantages offered by this technique. However, at variance from clinic and ambulatory blood pressure measurements, no precise and generally accepted recommendation is presently available on what an optimal schedule for HBPM might be. In other words, even if some indications on this issue have been provided by the recent guidelines for blood pressure measurement issued by the European Society of Hypertension Working Group on Blood Pressure Monitoring , there is no general consensus on how many measurements are needed, and on how many days, to obtain the best assessment of the actual HBP levels in a given subject. This problem is exemplified by the large discrepancy in the frequency of HBP readings found between different studies carried out on this issue, which makes a proper comparison of their results very difficult. While some investigators have focused on HBP measurements obtained on a single day [8–10], others have made use of HBP readings obtained over 3–4 days [11–14], over 1–2 weeks [15–21] or even over 1 month . Not only the number of HBPM days, but also the number of HBP readings per monitoring day showed a marked between-studies discrepancy, ranging from a single morning reading per day , to one morning and one evening reading [9,10,15,16], to duplicate [14,20,21] or triplicate morning and evening readings [11–13,17,18], up to five sequential readings per day .
Studies trying to provide an answer to the question of what the most suitable schedule for HBPM might be have made use of two different methods, namely a statistical and a clinical approach. According to the statistical approach, criteria for defining the best frequency of HBP measurements should be based on (i) the reproducibility of HBP values obtained; (ii) their stability over time (in the same or in subsequent visits); and (iii) their relation with average ABPM values, considered as the gold standard references [4,6]. Conversely, the clinical approach, which was adopted in the study by Ohkubo et al. published in this issue of the Journal of Hypertension , is based on the assessment of the power of average HBP values, derived from different HBPM schedules, to predict the occurrence of cardiovascular events. The results offered by these two approaches will be briefly discussed.
How to define the optimal HBPM schedule: a statistical approach
The reproducibility of HBP has been reported to directly depend on the number of HBP measurements to be averaged , as previously shown for ABPM [23–26]. Chatellier et al.  showed that the maximal reduction in the standard deviation of the mean difference between the average values of two HBPM sessions is obtained when each average value is derived from 30 HBP measurements (three measurements per day for 10 consecutive days), although 80% of this maximal reduction was already obtained by averaging 15 measurements of the first 5 days. On the other hand, Imai et al. , although confirming that the reproducibility of HBP is at least in part dependent on the number of measurements, provided evidence that no further improvement is obtained by increasing the number of HBP readings above five . This conclusion is in agreement with the results of the SMART study, which also showed that, after only six HBP measurements, a small decrease in the standard deviation of the mean difference in average HBP values between two sessions is achieved . The first study to specifically address the issue of the minimum number of HBP measurements necessary to provide reliable and reproducible mean values included 189 hypertensive subjects, in whom clinical, home and ambulatory blood pressure were measured . The study endpoints were the reliability and the reproducibility of HBP. Criteria for reliability were the stabilization of mean HBP over time, its variability (standard deviation of average value within a session) and its relationship with the average 24-h ABP values, taken as the blood pressure gold standard. The reproducibility of HBP was quantified using test–retest correlations and the standard deviation of mean differences between average HBP values of different days. The average HBP of the initial monitoring day provided higher and more unstable values (higher standard deviation) than the values obtained over the following days, although no difference was observed among days 2–6 . When measurements of the initial day were excluded, the average HBP value of the whole monitoring period was significantly reduced. However, at least two monitoring days were needed for the reproducibility of HBP to be superior to that of clinic blood pressure . These findings were confirmed by another study by the same group in which 133 untreated hypertensives were included. The reproducibility of HBP in this case also improved by discarding the initial monitoring day, whereas only a minimal improvement was achieved by averaging measurements taken over more than 2 days (average of days 2–3) . On the basis of these data, it was concluded that, after excluding measurements obtained on the initial day, the average of duplicate morning and evening HBP measurements of 2 days is the minimum requirement to ensure a reliable estimate of the average level of blood pressure at home [20,21].
The number of HBP readings required for a reliable estimate of the true level of HBP in a given subject was also investigated in a study on stress management training, carried out in 43 hypertensive patients . Test–retest correlation coefficients and the standard deviation of mean differences between the averages of repeated measurements, as well as other more complex statistical models, were applied to HBP readings obtained before and 8 weeks after treatment. The results showed that the average of single evening readings taken over three consecutive days was required to provide a reliable estimate of HBP levels. Conversely, another study using single afternoon HBP measurements in 74 elderly subjects did not find any significant difference among group average sitting HBP values when considering readings obtained only on the first day, over 3 days and over 10 days . However, the HBP values of the initial day tended to be higher and had the highest within group standard deviation. The authors concluded that at least 3 days of HBP measurements may be needed to obtain stable HBP values . Brook  further addressed the effect of different HBP measurement schedules on the accuracy of the resulting HBP averages and assessed the optimal frequency of HBP measurements to be recommended, by reviewing 12 studies all based on HBP measurements but characterized by considerably variable HBP monitoring schedules. The accuracy of HBP average values, as determined by their agreement with average ABP values, was not related to any of the HBPM schedule parameters considered, including total number of measurements, number of measurements per session, number of sessions per day, and total duration of monitoring. The author also concluded that, on a statistical basis, most of the benefits derived from HBPM can be achieved by obtaining only a few HBP measurements through a very simple monitoring schedule  (i.e. by considering as few as two HBP readings obtained on a single day).
The results of all these studies appear therefore to suggest that the usefulness of HBPM in the management of hypertensive patients  depends not only on the statistical advantages associated with the availability of repeated measurements (the reliability of average HBP values increasing with the increase in the number of measurements available) [18,23–25], but also on the possibility offered by even a few HBP measurements to obtain information on blood pressure levels away from the clinic setting, in the protected environment of the patient's home.
A debated issue in this respect is the actual value of the data provided on the first day of HBPM. Although HBP is devoid of a white-coat effect [6,30,31], higher and unstable values are usually obtained on the first HBPM day [12,20,21]. Moreover, HBP tends to decline with repeated measurements, both in the same session  and over consecutive days [13,20,21], similarly to what was observed for clinic blood pressure measurements . Thus, the choice of focusing on a single HBPM day, and on a single measurement per session, may not be the most appropriate one. Indeed, it was suggested that the initial HBPM day should be regarded as a training day, and that the corresponding HBP values should be discarded [12,20,21]. This suggestion is further supported by a recent study comparing the effects of two antihypertensive drugs on HBP, with data being obtained in repeated sessions over a long observation time. The average HBP values of the initial day were also unreliable in this study too, and were unable to identify any between-drug difference in efficacy for diastolic blood pressure . Furthermore, the drawback of the initial day of HBP measurement was invariably observed on the first day of any of three HBPM sessions (each including 5-day HBPM), repeated at 5-week intervals during the study . The suggestion to exclude HBP values obtained on the first HBPM day from data analysis has been implemented in the recent recommendations published by the European Society of Hypertension Working Group on Blood Pressure Monitoring .
How to define the optimal HBPM schedule: a clinical approach
A clinical approach for the identification of the best frequency of HBP measurements to be recommended, based on the prognostic value of the average HBP data obtained, was employed in the study by Ohkubo et al. . This study, which includes data from the outcome population study carried out in Ohasama (Japan), has the merit of addressing for the first time the issue of how to identify the best possible HBPM schedule by comparing the ability of average HBP values, based on a different number of readings, to predict the risk of developing stroke. Subjects had their clinic blood pressure measured twice at the screening visit, in addition to being asked to monitor their HBP once a day in the morning, every day for 4 weeks. The authors showed that even the first single HBP measurement had a greater prognostic value in relation to stroke risk than duplicate initial clinic blood pressure measurements. The predictive value of average HBP increased progressively, without any threshold, by increasing the number of measurements up to 14, with the highest predictive value being associated with the average of all measurements (n = 25) .
In spite of their interest, these findings should be interpreted with some caution because of the unusual schedule for HBPM used in the Ohasama study. Only single morning measurements were performed, for a period of 4 weeks, whereas in most of the previous studies, from two to six HBP measurements were obtained on each monitoring day. This difference in the number of measurements compared to other studies may be clinically important because of the evidence indicating that HBP values are invariably lower by 1–2 mmHg when going from the first to a second HBP measurement taken in the same session, and this difference remained detectable even after repeated measurement sessions . Moreover, HBP values are also known to be different between morning and afternoon sessions [13,18], which further emphasizes the limitations of a single daily measurement.
Notwithstanding such a discrepancy with the usual approach to HBPM, the results of the study by Ohkubo et al.  are quite relevant against the backgound discussion of whether the advantages of HBP versus clinical blood pressure measurements should mainly be attributed to the larger number of measurements obtained, or to the fact that these measurements are taken away from the artificial setting of a clinic, in the usual daily life environment of each individual. The finding that even a single HBP measurement on the first day offered prognostic information superior to that offered by the average of two clinic blood pressure measurements appears to suggest that the setting where blood pressure is measured may be more important than the number of times blood pressure is taken, at least when only a few readings are considered. This observation is not completely surprising, and is in agreement with the findings of the SAMPLE study (Study on Ambulatory Monitoring of Pressure and Lisinopril Evaluation), which showed treatment- induced changes in HBP (average of only two measurements) to be significantly related to treatment-induced changes in left ventricular mass index, whereas changes in clinic blood pressure (also the average of two measurements) did not . However, the study by Ohkubo et al.  also emphasizes that, besides the effect of the measurement site, the number of HBP measurements is also an important factor when considering repeated HBP sessions, as shown by the progressive increase in the prognostic value of average HBP values by increasing the number of measurements up to 14–25 . Again, this is in agreement with the data of the SAMPLE study in which the relationship between treatment-induced regression of left ventricular hypertrophy and the corresponding blood pressure reduction was closest when considering changes in 24-h average ambulatory blood pressure. The advantages offered by 24 h average blood pressure were likely to depend not only on the inclusion of day and night values, but also on the fact that it was based on the average of approximately 90 measurements .
Another interesting finding of the study by Ohkubo et al.  is that there was no threshold in the number of HBP measurements to be considered in relation to the increasing power of stroke risk prediction, up to 14 measurements obtained over 2 weeks, with a modest additional benefit by averaging all 25 measurements . This is more or less in agreement with previous shorter-term studies that focused on the reproducibility of HBP. Chatellier et al.  suggested that 15 HBP measurements should be the minimun number required, with the possibility of some further improvement up to 30 measurements, whereas Stergiou et al.  proposed a minimum of 12 measurements. Concerning the number of HBPM sessions to be recommended, on the basis of the Ohasama outcome study, HBPM carried out for at least 1 or 2 weeks (7 or 14 measurements) would be required to obtain clinically relevant information .
The European Society of Hypertension Working Group on Blood Pressure Monitoring guidelines recommend duplicate morning and evening HBP measurements to be taken daily for 7 days, and that the measurements taken on the first day should be discarded . Thus, in terms of number of HBP measurements, the European Society of Hypertension Working Group recommends that a total of 24 measurements should be averaged . The evidence from the available few studies based on a statistical approach and the data provided by the outcome study by Ohkubo et al.  appear to suggest that taking fewer measurements offers similar benefits to taking several. However, when deciding on the minimum schedule for HBPM to be applied in individual patients in clinical practice, it is important to consider that the conclusions reached for groups of subjects may not invariably apply to an individual patient . Although the reproducibility of HBP, even when based on a limited number of measurements as suggested by the statistical approach, appears to be excellent in groups of subjects, it probably is still inadequate for decision making in the individual patient . Given that HPBM is easily accepted by patients and has a relatively low cost, it appears thus reasonable to favour the recommendation to average more readings than the statistically reliable minimum number. This is also the final conclusion of the paper by Ohkubo et al. , which thus is in agreement with the suggestions put forward by the European Society of Hypertension Working Group on Blood Pressure Monitoring .
This more cautious approach, based on the average of 14–25 HBP readings, is expected to provide not only more reliable and more reproducible measures of HBP in the individual patient, but also HBP data characterized by a stronger prognostic power in predicting the rate of future cerebrovascular events.
However, it should be emphasized that this conclusion applies to the initial evaluation of subjects recruited from the general population of Japan, where stroke is highly prevalent. What the optimal number of HBP measurements carrying prognostic information might be in Western populations, when considering both cerebrovascular and coronary artery disease, still remains to be assessed. Additional information is also needed to determine the optimal number of HBP readings to be used in the long-term follow-up of patients already diagnosed as hypertensives, as well as in the regular assessment of blood pressure coverage by antihypertensive treatment. These yet unresolved issues need to be assessed in future ad hoc outcome studies, similar to the one performed by Ohkubo et al. .
1. Mancia G, Parati G, Di Rienzo M, Zanchetti A. Pathophysiology of hypertension. In: Zanchetti A, Mancia G (editors): Handbook of hypertension
, vol. 17. Amsterdam: Elsevier Science BV; 1997, pp. 117–169.
2. Mancia G, Ferrari A, Gregorini L, Parati G, Pomodossi G, Grassi G, et al
. Blood pressure and heart rate variabilities in normotensive and hypertensive human beings. Circ Res
3. Guidelines Committee of the 2003 European Society of Hypertension–European Society of Cardiology guidelines for the management of arterial hypertension. J Hypertens
4. Mancia G, Di Rienzo M, Parati G. Ambulatory blood pressure monitoring: use in hypertension research and clinical practice. Hypertension
5. Fagard RH, Staessen JA, Thijs L. Prediction of cardiac structure and function by repeated clinic and ambulatory blood pressure. Hypertension
6. O'Brien E, Asmar R, Beilin L, Imai Y, Mallion JM, Mancia G, et al
., on behalf of the European Society of Hypertension Working Group on Blood Pressure Monitoring. European Society of Hypertension recommendations for conventional, ambulatory and home blood pressure measurement. J Hypertens
7. Parati G, de Leeuw P, Illyes M, Julius S, Kuwajima I, Mallion JM, et al
. Blood pressure measurement in research. Blood Press Monit
8. Staessen J, Fagard R, Lijen P, Thijs L, Van Hulle S, Vyncke G, et al
. Ambulatory blood pressure and blood pressure measured at home: progress report on a population study. J Cardiovasc Pharm
1994; 23 (suppl 5)
9. Mancia G, Zanchetti A, Agabiti-Rosei E, Benemio G, De Cesaris R, Fogari R, et al
., for the SAMPLE Study Group. Ambulatory blood pressure is superior to clinic blood pressure in predicting treatment-induced regression of left ventricular hypertrophy. Circulation
10. Mancia G, Sega R, Bravi C, De Vito G, Valagussa F, Cesana G, et al
. Ambulatory blood pressure normality: results from the PAMELA study. J Hypertens
11. Bodrie G, Genes N, Vaur L, Clerson P, Vaisse B, Mallion JN, et al
. Is ‘isolated home’ hypertension as opposed to ‘isolated office’ hypertension a sign of greater cardiovascular risk? Arch Intern Med
12. Chatellier G, Dutrey-Dupagne C, Vaur L, Zannad F, Genes N, Elkik F, et al
. Home self blood pressure measurement in general practice: the SMART study. Am J Hypertens
13. De Gaudemaris R, Chau NP, Mallion JM, for the Groupe de la Mesure, French Society of Hypertension. Home blood pressure: variability, comparison with office reading and proposal for reference values. J Hypertens
14. Stergiou GS, Thomopoulou GC, Skeva II, Mountokalakis TD. Home blood pressure normalcy: the Didima study. Am J Hypertens
15. Mejia AD, Julius S, Jones KA, Schork NJ, Kneisley J. The Tecumseh blood pressure study. Arch Intern Med
16. Mengden T, Binswanger B, Weisser B, Vetter W. An evaluation of self-measured blood pressure in a study 5ith a c3lcium-c2annel a7tagonist versus a beta-blocker. Am J Hypertens
17. Den Hond E, Celis H, Fagard R, Keary L, Leeman M, O'Brien E, et al
., on behalf of the THOP investigators. Self-measured versus ambulatory blood pressure in the diagnosis of hypertension. J Hypertens
18. Chatellier G, Day M, Bodrie G, Menard J. Feasibility study of N-of-1 trials with blood pressure self-monitoring in hypertension. Hypertension
19. Celis H, De Cort P, Fagard R, Thijs L, Staessen JA. For how many days should blood pressure be measured at home in older patients before steady levels are obtained? J Hum Hypertens
20. Stergiou GS, Baibas NM, Gantzarou AP, Skeva II, Kalkana CB, Roussias LG, et al
. Reproducibility of clinic, ambulatory and home blood pressure: implications on the design of trials for the assessment of the efficacy of antihypertensive drugs. Am J Hypertens
21. Stergiou GS, Skeva II, Zourbaki AS, Mountokalakis TD. Self monitoring of blood pressure at home: how many measurements are needed? J Hypertens
22. Ohkubo T, Asayama K, Kikuya M, Metoki H, Harusima H, Hashimoto J, et al
. How many times should blood pressure be measured at home for better prediction of stroke risk? 10-year follow-up results from the Ohasama study. J Hypertens
23. Parati G, Ravogli A, Mutti E, Santucciu C, Omboni S, Mancia G. Ambulatory blood pressure monitoring in the evaluation of antihypertensive drugs. J Hypertens
1994; 12 (suppl 8)
24. Mancia G, Parati G. Ambulatory blood pressure monitoring and organ damage. Hypertension
25. Trazzi S, Mutti E, Frattola A, Imholz BPM, Parati G, Mancia G. Reproducibility of non-invasive and intra-arterial blood pressure monitoring. Implications for studies on antihypertensive treatment. J Hypertens
26. Parati G, Mutti E, Ravogli A, Trazzi S, Villani A, Mancia G. Advantages and disadvantages of non-invasive ambulatory blood pressure monitoring. J Hypertens
1990; 8 (suppl 6)
27. Imai Y, Ohkubo T, Hozawa A, Tsuji I, Matsubara M, Araki T, et al
. Usefulness of home blood pressure measurements in assessing the effect of treatment in a single-blind placebo-controlled open trial. J Hypertens
28. Garcia-Vera MP, Sanz J. How many self-measured blood pressure readings are needed to estimate hypertensive patients’ ‘true’ blood pressure? J Behav Med
29. Brook RD. Home blood pressure: accuracy is independent of monitoring schedules. Am J Hypertens
30. Parati G, Pomidossi G, Casadei V, Mancia G. Lack of alerting reactions and pressor responses to intermittent cuff inflations during non-invasive blood pressure monitoring. Hypertension
31. Stergiou GS, Zourbaki AS, Skeva II, Mountokalakis TD. White coat effect detected using self-monitoring of blood pressure at home: comparison with ambulatory blood pressure. Am J Hypertens
32. Stergiou GS, Efstathiou SP, Skeva II, Baibas NM, Kalkana CB, Mountokalakis TD. Assessment of drug effects on blood pressure and pulse pressure using measurements in the clinic, at home and with ambulatory monitoring. J Hum Hypertens
33. Prisant LM. Ambulatory blood pressure monitoring: test reproducibility and its implications. Blood Press Monit