How to best monitor blood pressure at home? Assessing numbers and individual patients

In March 2009, an article entitled 'Home blood-pressure monitoring: US and European consensus' appeared in the Lancet1, underlying the deep agreement between two guidelines documents on home blood pressure (BP) monitoring (HBPM) published almost simultaneously at the two sides of the Atlantic, one by the European Society of Hypertension 2 and another by the American Heart Association 3. On the background of the increasingly wide use of self-BPM by hypertensive patients at home in several countries, these publications aim at optimizing and boosting the application of this method worldwide. Both the European and the American guidelines for HBPM recommend the use of upper arm cuff automated electronic (oscillometric) devices if validated using an established protocol 2,3. Devices equipped with automated memory to store measurements should be preferred, as well as those with a software able to provide whole period averages according to the specific recommendations for optimal HBPM schedule provided in an identical fashion by both the European and the American guidelines 2,3. According to these recommendations, before each doctor's visit a minimum of 3 and preferably 7 days of HBPM should be performed with duplicate morning and evening measurements per day, and the average of all readings should be calculated after discarding readings of the first day. Different sized cuffs should be available to fit the arm circumference of all potential users. The issue of the optimal HBPM schedule needed to provide a reliable picture of the BP levels prevailing at home has been extensively investigated and discussed 4–17. There are two approaches to investigate this issue. One is the 'statistical approach' that assesses the reproducibility of average HBP values, their stability (standard deviation) and their relationship with ambulatory BP 4,5. The second is the 'clinical approach', which assesses the relationship of HBP values with the consequences of hypertension (target organ damage and cardiovascular events) 4,5. It might be argued that the clinical approach would be the only one needed because it reflects the precision of risk prediction by HBP measurements. Indeed, BP is only a surrogate (intermediate) endpoint, and its measurement is only useful because of its ability to predict the risk of future cardiovascular events. However, HBPM is being used by practising physicians for long-term treatment decisions in individual patients. For this reason, the statistical approach is also necessary to complement the clinical one in order to ensure a reliable assessment, in terms of stability and reproducibility, of an individual's 'usual' average HBP values. This is exemplified by the fact that, although in large outcome trials the first single HBP reading is strongly related with cardiovascular events (clinical approach) 16,17, in the individual patient, the first reading has very little diagnostic or prognostic value and repeated measurements are needed on several occasions to provide a reliable and reproducible estimate of the 'true' HBP characterizing a given individual (statistical approach). Several studies 7–15 have used a statistical approach to investigate what an optimal HBPM schedule might be, whereas only two studies 16,17 have used prognostic data to address this issue on the basis of cardiovascular events' risk. Interestingly, all the above-mentioned studies, when defining the ideal number of HBP measurements to obtain, agree that 12–14 HBP readings is the minimum number required 4,5,16,17. However, all investigators also agreed that averaging 24–28 readings gives an additional, although small, benefit and should therefore be preferred 4,5,16,17. Furthermore, studies 4,5,16,17 using either the statistical or the clinical approach showed the first HBPM day to be the least reliable. A recent analysis of the Didima study 17 that combined the statistical with the clinical approach in the same dataset showed a striking similarity in the conclusion derived from both approaches (Fig. 1). By applying the criteria of stabilization of mean HBP, its variability (SD) and hazard ratios of cardiovascular events, this study showed that by averaging more readings, there is a progressive decline in average HBP and in its SD together with an increase in the hazard ratio of cardiovascular events. Most of these benefits were achieved on the second day (eight readings), and little additional benefit was obtained on the third day (12 readings) (Fig. 1) 17. Thus, the first day clearly gave higher HBP values, which were more unstable (higher SD) and had lower prognostic ability. Again, in this study, little difference was found between morning and evening HBP values with no prognostic superiority of morning readings 18. These findings thus validate and fully support the European Society of Hypertension 2 and the American Heart Association 3 recommendations for the optimal HBPM schedule.Fig. 1In this issue of the Journal of Hypertension, Johansson et al.19 offer new data on the same issue; that is, on the assessment of the optimal HBPM schedule. This study is unique in that it is the only one that used indices of target organ damage (microalbuminuria and echocardiographic left ventricular mass index). In line with all the previous reports, this study confirmed that by averaging more measurements, the reliability of HBP is improved 19. Most of this improvement was achieved on the fourth day (16 readings), and there was marginal further benefit by averaging 28 readings (7 days). Again, in this study, morning HBP measurements offered no advantage compared with evening measurements 18. The only important difference in this analysis compared with the previous reports is that the authors concluded that there is no need to discard HBP measurements of the first day 19. This conclusion by Johansson et al.19 about the first HBPM day has been based on the lack of a clear difference in the strength of the association of average HBP with ambulatory BP and target organ damage when measurements of days 1–3 versus days 2–3 or those of days 1–7 versus 2–7 were used 19. However, the study data clearly showed that the largest decline in average HBP and the largest improvement in the strength of the relationship with ambulatory BP, microalbuminuria and left ventricular mass index were achieved between day 1 and day 2 (table 2 and fig. 1 in 19). It is only because data for cumulative rather than individual days are reported in the study that the drawback of the first day did not become so evident. Moreover, although in this study the correlation coefficients of different average HBP values with ambulatory BP and target organ damage had a relatively narrow range (table 2 in 19), it is important to notice that the change in the coefficient values was the same for HBP at day 1 versus day 2 (cumulative days 1–2) as it was for day 2 versus days 2–7 (change in correlation coefficient value by 2–3 points for both systolic and diastolic HBP). Another issue to mention in the study by Johansson et al.19 is that it included a random population sample and a group of newly diagnosed hypertensive patients with office BP more than 180/100 mmHg, thus excluding groups of individuals that might carry diagnostic problems. Indeed, individuals with BP levels within the range of stage I hypertension are those who are most often misclassified and might benefit the most from out-of-office BP monitoring. Whether measurements of the first day should be discarded is obviously dependent on the number of measurements obtained. If 7 or more days' measurements are averaged, then the inclusion or exclusion of the first day is expected to have only a negligible effect on calculated average HBP. However, in real-life conditions, when patients might take only few measurements, the impact of the first day is expected to be larger. Notwithstanding these differences, for the sake of clarity it does not seems to be appropriate to give a different recommendation to the practising physicians regarding the exclusion or not of the first day on the basis of the number of HBP measurements obtained by individual patients. Although Johansson et al.19 found no statistically significant differences between average HBP of days 1–7 versus 2–7 and also between days 1–3 and 2–3, there is a clinical issue that should be considered. Studies always assess and report data of many patients analysed altogether, whereas physicians assess one patient at a time. The higher and more unstable (larger variation) HBP of the first day that has been consistently shown in all studies does not necessarily demonstrate that all patients on their first monitoring day had a higher HBP by a few mmHg with higher SD, but that some of them had significantly higher and unstable HBP on their first day whether others did not. The physicians should be aware of the 'first-day' HBP reaction and should better have a homogeneous approach in all their patients. The practical recommendation would thus be to always ignore these measurements rather than deciding in each case whether there is a clinically relevant first-day reaction or whether the number of measurements is adequate so as to preclude a significant effect of the first-day measurements on average HBP. Devices with embedded guidelines-based HBPM schedule and automated systems that display an average BP of 3–7 days after discarding the first day are available in the market 20. Such devices may facilitate the implementation of the HBPM schedule in clinical practice as it is today recommended by international consensus 1. In conclusion, for self-monitoring of BP at home, the most important issues to remember are using validated devices and appropriate measurement conditions and obtaining at least 12 measurements (over 3 days) and preferably 28 measurements (over 7 days) 2,3. The first monitoring day provides higher and more unstable HBP readings with lower prognostic ability and should better be discarded, particularly if only the minimum number of the recommended HBP measurement has been obtained. This HBPM schedule should be implemented by all patients before each visit to the doctor's office to adequately complement the conventional measurements taken during consultation and improve the accuracy of BP assessment and hypertension diagnosis 2,3.