Blood Pressure During Siesta: Effect on 24-H Ambulatory Blood Pressure Proﬁles Analysis

Blood pressure during siesta: effect on 24-h ambulatory blood pressure proﬁles analysis

GS Stergiou, JS Malakos, AS Zourbaki, AD Achimastos and TD Mountokalakis Hypertension Centre, Third University Department of Medicine, Sotiria Hospital, Athens, Greece

Blood pressure (BP) during siesta declines to levels The magnitude of night time BP drop was greater with similar to those of night time sleep. The objective of the ACT intervals, resulting in a lower percentage of non- study was to assess the effect of siesta on 24-h ambu- dippers (P Ͻ 0.001). Among 49 recordings without a latory BP (ABP) data. Two different approaches were siesta, differences between ACT and ARB BPs were less employed for the definition of day and night periods: (1) pronounced for daytime but not for night time. Differ- actual patient reported day and night intervals (ACT) ences in the magnitude of nocturnal BP drop between with siesta period analysed as a third time period; and ACT and ARB periods, although statistically significant, (2) arbitrary day and night time intervals (ARB) with the did not affect the prevalence of non-dippers. In con- presence of siesta being ignored. A total of 203 24-h clusion, analysis of 24-h BP profiles by using ARB ABP recordings were analysed, with a siesta during instead of ACT day and night intervals results in under- ABP monitoring reported in 154 of them. Mean siesta BP estimation of the nocturnal BP drop and overestimation was very close to ACT night time BP. Among recordings of the proportion of non-dippers. This bias is more with a siesta, ACT daytime BP was higher and night time pronounced in patients who take a siesta during ABP BP lower than the corresponding ARB BPs (P Ͻ 0.001). monitoring.

Keywords: siesta; ambulatory blood pressure; data analysis; non-dippers

Introduction running for 15 ± 2 h (07.00 to 22.00), and night time for 9 ± 2 h (20.00 to 07.00) as reference intervals.16 Non-invasive ambulatory blood pressure (ABP) However, there is no consensus for the definition of monitoring allows BP to be measured repeatedly night time period and a great variety of arbitrary away from the clinic setting. In addition ABP moni- time schedules has been employed in different stud- toring offers the unique opportunity to take intermit- ies.8,9,11–15,17 Narrow time intervals (for example, tent BP measurements during sleep. Studies with 00.00 to 05.00 or 01.00 to 07.00)13,14 have been pro- ABP monitoring have shown that in about 20% of posed by some investigators in order to ignore inter- essential hypertensive patients the normal BP fall mediate hours just before and after falling asleep, during sleeping hours is reduced or absent, allowing whereas wider intervals that include late afternoon division of patients into dippers and non-dippers.1,2 hours have been preferred by others, the 22.00 to The attenuation of nocturnal BP drop has been 06.00 interval being the most commonly used.17,18 shown to be associated with greater echocardio- Analysis of 24-h ABP data based on arbitrary instead graphic left ventricular mass,3–5 higher incidence of of actual sleeping hours results in inclusion of a silent cerebrovascular damage6 and increased risk number of ‘in bed’ readings in the daytime period for future cardiovascular morbid events.1,7–9 It is and ‘out of bed’ readings in the night time per- noteworthy that the Joint National Committee in its 11,17,19,20 fifth report suggest that non-invasive ABP monitor- iod. Thus, calculated average day and night time BP and noctural BP drop may be significantly ing may be useful in selected cases in practice, 11,17,20,21 including the evaluation of nocturnal BP changes.10 affected. This bias is expected to be more In the analysis of 24-h ABP data, fixed day and pronounced in patients with daytime sleep (siesta) night periods are routinely applied while actual since BP during siesta is reported to decline to levels 19,22 patient reported sleeping hours are used only similar to those of night time sleep. occasionally.3,4,8,9,11–15 This approach has been The main objective of the present study was to encouraged by the Scientific Committee of the 1990 compare the effect of siesta on ABP data (mean day- International Conference on Indirect ABP Monitor- time, night time and nocturnal BP drop) analysed by ing that recommended the use of daytime period using either arbitrary day and night time intervals or actual patient reported sleeping hours. Secondary objectives were to assess the incidence of siesta in Correspondence: George S Stergiou, Hypertension Centre, Third patients undergoing 24-h ABP monitoring in a ter- University Department of Medicine, Sotiria Hospital, 152 Meso- tiary Hypertension Clinic, and to investigate the gion Ave, Athens 11527, Greece Received 16 August 1996; revised 4 November 1996; accepted 11 magnitude of the BP decline during siesta compared November 1996 to noctural BP drop. We compared ABP monitoring Siesta and analysis of ambulatory BP profiles GS Stergiou et al 126 data analysed by using actual patient reported sleep- ‘out of bed’ periods, and with siesta period, if ing hours, or arbitrary day and night intervals in present, analysed as a third time period; and (2) arbi- patients that had a siesta during ABP monitoring trary (ARB) periods by using arbitrary limits with and in patients that slept only at night. daytime as 06.00 to 22.00 and night time as 22.00 to 06.00, with the presence of siesta being ignored. Patients were classified as non-dippers if noctur- Materials and methods nal BP fall was less than 10% of the daytime value Material for SBP and/or DBP.3 Data are expressed as mean ± s.d. except where indicated otherwise. Paired t-test All 24-h ABP recordings performed over a 2-year was used for the comparison of BPs calculated by period (April 1993 to July 1995) in patients referred using different day and night times, and with Bon- to the Hypertension Clinic were analysed retrospec- ferroni’s correction for multiple comparisons tively. Recordings with less than 30 successful day- applied where appropriate. Differences in classi- time BP measurements (07.00 to 22.30) and/or less fication of non-dippers was assessed by using ␹2 test than 15 night time measurements (23.30 to 07.00) 15 with Yates correction. The relation between BP drop were excluded from analysis. Average siesta BP during siesta and night was assessed by using Pear- was calculated when at least three BP measurements son correlations. Bland-Altman approach24 was were recorded during patient reported daytime employed in order to investigate the degree of simi- sleep. larity between ACT and ARB nocturnal BP drop. Analysis was performed using the Minitab Statisti- Measurements cal Software, 8th PC Version (Minitab Inc, State Col- lege, USA). A probability value P Ͻ 0.05 was con- ABP was measured by using non-invasive portable sidered statistically significant. oscillometric devices SpaceLabs 90207 (SpaceLabs Inc, Redmond, WA, USA). The recorders were pro- grammed to measure BP at 20 min intervals for 24 h Results and were always applied on a routine working day. ABP data To ensure good function of the device, two to three succeeding readings were taken in the clinic by A total of 264 consecutive 24-h ABP recordings were manual activation. Patients were instructed to carry drawn from the computer. ABP monitoring was performed in the context of clinical trials (about two- out their usual daily activities, but to stay still with 10 the forearm extended during each reading. They thirds of cases) or on clinical indication. Eighteen were asked to keep a brief diary specifying the time recordings were discarded because of inadequate BP when they went to bed and were out of bed during data. A diary with patients sleeping hours was avail- daytime and night time sleep. able in 203 cases. Only night time sleep was Clinic BP measured at a routine clinic visit within reported in 49 cases (24%) and both daytime and 2 weeks preceding the 24-h ABP monitoring was night time sleep were reported in 154 cases (76%). used for analysis. Clinic BP was measured with stan- There was no significant difference between groups dard mercury sphygmomanometers (bladder size regarding patient characteristics, duration of clinic 15 × 35 cm, Korotkoff phase V for diastolic BP (DBP) attendance, antihypertensive treatment status and by three physicians that satisfied accuracy criteria clinic or 24-h ABP (Table 1). ± ± for observer agreement recommended by the British A total of 78.1 6.5 (mean s.d.) readings were 23 obtained with 24-h ABP monitoring. A fraction, Hypertension Society. Following Hypertension ± Clinic protocol all patients had triplicate measure- 11.8 5.2%, of the readings that satisfied at least one ments of sitting BP after 5 min rest and one measure- of the editing criteria were considered erroneous ment in the standing position, with at least 1 min and were discarded. In the majority of time points between recordings. Mean value of second and third sitting BP measurement was recorded and used for Table 1 Patient characteristics, duration of clinic attendance and analysis. clinic and ABP

No diary No siesta Siesta Analysis No. 43 49 154 BP measurements flagged by the software of the ± ± ± monitors as technically erroneous were excluded, as Age (years) 53.5 13.9 52.1 10.6 54.4 10.8 Ͻ Sex (M/F) 0.87 0.78 1.1 were measurements with systolic BP (SBP) 70 BMI (g/m2) 26.9 ± 3.4 28.1 ± 3.6 28.5 ± 3.4 mm Hg or Ͼ260 mm Hg or with DBP Ͻ40 mm Hg or Treated (%) 71.7 69.4 64.3 Ͼ 150 mm Hg. Early readings taken less than 20 min Clinical visits 4.3 ± 2.4 4.7 ± 3.3 4.3 ± 3.2 after the monitor was attached to the patient were Attendance (months) 6.9 ± 8.3 6.6 ± 9.2 5.7 ± 8.0 also excluded as these were taken in the clinic. Clinic BP Average 24-h, daytime, siesta and night time ABP Sitting SBP 144.0 ± 21.1 143.8 ± 16.4 144.9 ± 15.5 were calculated for each monitoring. Two different Sitting DBP 90.8 ± 11.2 91.8 ± 9.5 92.5 ± 10.0 approaches were used for the definition of daytime Ambulatory BP and night time periods: (1) actual (ACT) day and 24-h SBP 132.8 ± 17.6 131.5 ± 13.0 132.6 ± 12.9 night time periods by using individual patient 24-h DBP 82.8 ± 10.6 83.8 ± 12.9 83.6 ± 10.0 reported sleeping hours designated as ‘in bed’ and Siesta and analysis of ambulatory BP profiles GS Stergiou et al 127 where erroneous recordings were obtained, a suc- are presented in Figure 2. The percentage of non- cessful recording was obtained on automatically dippers (defined as subjects with less than 10% SBP repeated measurement 3 min later (57%). However, and/or DBP nocturnal drop) was greatly reduced 3.6 ± 3.9% of time points were not represented in when ACT night time was used (for SBP criterion the 24-h ABP profile since both initial and repeated 70 and 101 non-dippers for ACT and ARB periods recordings were considered erroneous. A total of respectively, P Ͻ 0.01, and for DBP criterion 30 and 43.6 ± 6.2 successful readings were obtained during 59, P Ͻ 0.01) (Figure 3). daytime, 21.4 ± 4.2 during night time and 5.0 ± 1.7 Average siesta BP drop was 8.7 ± 7.8/13.9 ± 10.5% recordings during daytime sleep. The average dur- (SBP/DBP drop compared to ACT daytime values) ation of siesta was 1.9 ± 0.7 h whereas the average and was closely related to ACT night time BP drop duration of night time sleep was 7.1 ± 1.3 h in (coefficient of correlation r 0.38/0.36 for SBP/DBP, patients that had a siesta and 7.4 ± 1.4 h in those that P Ͻ 0.0001). When the definition for dippers/ slept only during the night. non-dippers state was applied on siesta BP data, night time dipping pattern was in agreement with siesta dip in 61% of cases for SBP and in 68% of Patients with siesta cases for DBP (Table 4). Furthermore, there was no Among 154 patients that had a siesta during ABP indication of bimodality in the distribution of siesta monitoring, 99 (64.3%) were on antihypertensive BP dip or ACT night time BP dip (Figure 4). There drug treatment, 39 (25.3%) were untreated hyperten- merely seemed to be a gradation from patients sives (clinic BP Ͼ140/90 mm Hg and daytime ABP whose BP changed very little between awake and у135/85 mm Hg) and 16 (10.4%) were white coat siesta or night time sleep and those whose BP hypertensives (clinic BP Ͼ140/90 mm Hg and day- changed substantially. The specificity of the ARB time ABP Ͻ135/85 mm Hg). There were no differ- method to identify non-dippers correctly (ACT per- ences between groups regarding patient character- iods used as standard for classification as a non- istics and duration of clinic attendance whereas dipper) was 63% and 76% for SBP and DBP, clinic BP and ABP (24-h, daytime and night time) respectively. were significantly higher in untreated hypertensives compared to treated patients or white coat hyperten- Patients without siesta sives (P Ͻ 0.001). Mean siesta BP (systolic and diastolic) was on Among 49 patients that slept only at night during average 2 mm Hg higher than ACT night time BP ABP monitoring, 34 (69.4%) were on antihyperten- and 12 mm Hg lower than ACT daytime BP sive drug treatment, 10 (20.4%) were untreated (Table 2). ACT daytime BP was higher than ARB hypertensives and 5 (10.2%) were white coat hyper- daytime BP (systolic and diastolic) whereas ACT tensives. ACT daytime SBP was higher than the cor- night time BP was lower than the corresponding responding ARB BP whereas ACT night time BP ARB BP (Figure 1, Table 3). (SBP and DBP) was lower than the corresponding The magnitude of ACT night time BP dip ARB BP (Figure 1, Table 3). However, when com- (10.3 ± 6.4/16.2 ± 7.1% drop for SBP/DBP compared pared with patients had a siesta, the magnitude of to the corresponding ACT daytime values) was these differences was significantly less for daytime greater than ARB night time dip (SBP drop (P Ͻ 0.01 for SBP and P Ͻ 0.0001 for DBP) but not 7.6 ± 6.0%, discrepancy from ACT dip 2.7 ± 3.7%, for night time. P Ͻ 0.001, 95% CI 2.2, 3.3 and DBP drop The magnitude of ACT night time BP dip 12.3 ± 6.5%, discrepancy 4.0 ± 4.2%, P Ͻ 0.001, (13.5 ± 7.5/18.7 ± 9.0% drop for SBP/DBP) was 95%CI 3.3, 4.6). Discrepancies for Bland-Altman greater than ARB night time dip (SBP drop technique between ACT and ARB nocturnal BP drop 10.7 ± 6.6, discrepancy from ACT dip 2.7 ± 5.1%, expressed as per cent change from daytime values, P Ͻ 0.001, 95% CI 1.2, 4.2 and DBP 15.8 ± 8.1, discrepancy 3.0 ± 6.9%, P Ͻ 0.01, 95%CI 1.0, 5.0). However, the percentage of non-dippers was not sig- Table 2 Average differences between actual patient reported day- nificantly different when ACT periods were used time, night time and siesta ABP in patients that had a siesta (for SBP criterion 15 and 21 non-dippers for ACT Mean ± 95% CI and ARB periods respectively, and for DBP criterion s.d.d. (bias) 8 and 10) (Figure 3). The specificity of the ARB method to identify non-dippers correctly (ACT per- Day vs siesta iods used as standard) was 82% and 95% for SBP SBP 12.2 ± 11.1*** 13.9, 10.4 and DBP, respectively. DBP 12.5 ± 9.6*** 14.0, 10.9 Night vs siesta SBP −2.2 ± 11.3* −0.4, −4.0 Discussion − ± − − DBP 2.1 9.5** 0.6, 3.6 Previous studies have shown that BP level during Day vs night siesta is similar to night time sleep BP.19,22 In the ± SBP 14.4 9.1*** 12.9, 15.9 present study, although siesta BP (systolic and DBP 14.6 ± 6.7*** 13.5, 15.6 diastolic) was found to be significantly higher than Bias, mean difference between assessments; s.d.d., standard devi- ACT night time BP (Table 2) the difference was too ation of differences; 95% CI, 95% Confidence Intervals; low to be of clinical importance. It should be noted *P Ͻ 0.05; **P Ͻ 0.01; ***P Ͻ 0.001. that siesta and ACT night time periods were defined Siesta and analysis of ambulatory BP profiles GS Stergiou et al 128

Figure 1 Analysis of ABP proﬁles (mean ± s.e.) by using ACT or ARB time schedules. (ACT, actual patient reported day and night time periods; ARB, arbitrary daytime [06.00 to 22.00] and night time [22.00 to 06.00] periods; *signiﬁcant difference P Ͻ 0.05; **P Ͻ 0.01; ***P Ͻ 0.0001).

Table 3 Average differences between ACT and ARB ABP in patients recorded ‘in bed’ and ‘out of bed’ rather patients with or without siesta than ‘awake’ and ‘asleep’ periods, and the number of readings taken during siesta was less than a quar- Patients with siesta Patients without siesta ter of night time readings, inclusion in siesta period of BP recordings taken when patients were in bed Mean ± 95% CI Mean ± 95% CI s.d.d. (bias) s.d.d. (bias) but not asleep may have affected mean siesta BP to a greater degree than mean night time BP.19 This Daytime effect may be responsible for the smaller magnitude SBP 2.0 ± 2.3*** 1.6, 2.3 0.9 ± 2.4*† 0.2, 1.6 and the greater variance of BP dip during siesta com- DBP 1.9 ± 1.9*** 1.6, 2.2 0.6 ± 2.1*‡ 0.2, 1.2 pared to that during night time sleep (Table 2). The Night time fact that in 43 out of 246 cases (17%) a diary was SBP −2.0 ± 3.5*** −2.6, −1.5 −2.9 ± 5.7** −4.4, −1.1 DBP −1.9 ± 2.8*** −2.4, −1.5 −2.1 ± 5.0** −3.6, −0.6 not available, is not expected to influence the results, since no differences in patients character- Bias, mean difference between assessments; s.d.d., standard devi- istics, clinic and 24-h ABP and antihypertensive ation of differences; 95% CI, 95% Confidence Intervals; *P for treatment status existed between excluded and stud- discrepancy Ͻ0.05; **P Ͻ 0.01; ***P Ͻ 0.0001; †difference from ied cases (Table 1). Ͻ the corresponding change in patients with siesta P 0.01; In a population study with 24-h ABP monitoring, ‡P Ͻ 0.0001; ACT and ARB as in Figure 1. nocturnal BP fall was normally distributed indicat- ing that the distinction of patients in dippers and as the time when the patients went to bed until the non-dippers is arbitrary.13 A normal distribution of time they were out of bed as they might find dif- BP fall during daytime sleep was evident in the ficulty in reporting the actual time when they fall present study (Figure 4). In addition, BP fall during asleep and woke up. This approach is also sup- daytime sleep was strongly correlated with noctur- ported by a carefully designed study in normoten- nal BP fall. More specifically, dipping pattern was sives and hypertensives, where Khoury et al25 consistent during day and night time sleep in two- showed that BP changes little before and after awak- thirds of the cases (Table 4), a proportion similar to ening if the subjects remain supine. Therefore, since that corresponding to the reproducibility of the noc- Siesta and analysis of ambulatory BP profiles GS Stergiou et al 129

Figure 2 Discrepancies for Bland-Altman technique between ACT and ARB nocturnal BP drop expressed as per cent change from daytime values, in patients that had a siesta. Mean difference in nocturnal BP drop between ACT and ARB methods was 2.7 ± 3.7% (P Ͻ 0.0001, 95% CI 2.2, 3.3) for SBP and 4.0 ± 4.2% (P Ͻ 0.0001, 95%CI 3.3, 4.6) for DBP. (ACT, ARB as in Figure 1).

estimates the nocturnal BP drop11 and thereby overestimates the prevalence of non-dippers.17,20,21 The present study indicates that the above effects are influenced by the presence of daytime sleep. Thus, in the analysis of 24-h ABP profiles, ACT daytime BP was underestimated when ARB11,17,20 daytime intervals were used, particularly in patients that had a siesta during ABP monitoring.19 It must be noted that although these differences, particularly for patients that had a siesta, were highly statistically significant, mean bias for daytime was not clinically important, ranging from 1.9 mm Hg (DBP) to 2.0 mm Hg (SBP) in patients with a siesta and from 0.6 mm Hg (DBP) to 0.9 mm Hg (SBP) in those without a siesta. Ninety-five per cent confidence intervals excluded any difference between ACT and ARB day- Figure 3 Comparison of the percentage of non-dippers defined by time values greater than 2.3 mm Hg in patients with using ACT or ARB time schedules (**difference between ACT siesta and 1.6 mm Hg in those without siesta and ARB P Ͻ 0.001; ACT, ARB as in Figure 1). (Table 3). In contrast, ACT night time BP was over- estimated when ARB night time interval was used. Nevertheless, as indicated by 95% confidence inter- Table 4 Dippers and non-dippers during actual patient reported vals, differences between ACT and ARB night time night time and siesta BPs as great as 4 mm Hg could not be excluded, Night Night Total particularly among patients without a siesta time time (Table 3). dippers non- The opposite effects of ACT/ARB intervals on dippers daytime vs night time BPs resulted in underestim- ation of the magnitude of the nocturnal BP drop, and SBP Siesta dippers 47* 23 70 thereby overestimation of the proportion of non-dip- 17,20,21 Siesta non-dippers 37 47* 84 pers, the latter being more pronounced in Total 84 70 154 patients with a siesta during 24-h ABP monitoring.19 DBP However, the small number of patients that slept Siesta dippers 89* 14 103 only at night, precluded any significant difference in Siesta non-dippers 35 16* 51 the classification of non-dippers by using different Total 124 30 154 definitions of night time (Figure 3). On the basis of the present data, we suggest that *Agreement in 94/154 cases (61%) for SBP and in 105/154 cases (68%) for DBP. in patients that had a siesta during ABP monitoring, whenever night time BP and the nocturnal BP drop are of interest, the use of actual patient reported turnal BP drop.13,14,26 These data suggest that the sleeping hours is essential in the analysis of 24-h same factors are responsible for the decline of BP ABP profiles. Currently available ABP equipment during day and night time sleep. software are designed to handle BP data by using Recent studies have shown that analysis of 24-h fixed day/night intervals; thus, the intepretation of ABP profiles by using arbitrary day/night time ABP data by using individual patients reported schedules overestimates night time BP and under- sleeping hours requires additional effort and time. Siesta and analysis of ambulatory BP profiles GS Stergiou et al 130

Figure 4 Distribution of the actual BP drop during night time and siesta.

ABP equipment manufacturers should improve and Treatment of High Blood Pressure: The fifth report application software to allow for individualised of the Joint National Committee on the detection, analysis of 24-h ABP profiles. evaluation and treatment of high blood pressure (JNC V). Arch Intern Med 1993; 153: 154–183. 11 Peixoto Filho AJ, Mansor GA, White WB. Effects of Acknowledgement actual vs arbitrary awake and sleep times on analyses of 24-h blood pressure. Am J Hypertens 1995; 8: This article is based on data presented at the 16th 676–680. Scientific Meeting of the International Society of 12 Stewart MJ, McGinley IM, Padfield PL. Analysis of Hypertension in Glasgow, UK, June 1996. ambulatory blood pressure profiles – when is night time? (Abstract). J Hypertens 1991; 9: 1094. 13 Staessen J et al. The diurnal blood pressure profile. A References population study. Am J Hypertens 1992; 5: 386–392. 1 O’Brien E, Sheridan J, O’Malley K. Dippers and non- 14 Palatini P et al. Factors affecting ambulatory blood dippers (letter). Lancet 1988; ii: 397. pressure reproducibility. Results of the HARVEST 2 Pickering TG. The clinical significance of diurnal trial. Hypertension 1994; 23 (Suppl 2): 211–216. blood pressure variations. Dippers and non-dippers. 15 Chanudet X, Phong Chau N, Larroque P. Short-term Circulation 1990; 81: 700–702. representatives of daytime and night-time ambulatory 3 Verdecchia P et al. Circadian blood pressure changes blood pressures. J Hypertens 1992; 10: 595–600. and left ventricular hypertrophy in essential hyperten- 16 The Scientific Committee Consensus document on sion. Circulation 1990; 81: 528–536. non-invasive blood pressure monitoring. J Hypertens 4 Verdecchia P et al. Sex, cardiac hypertrophy and diur- 1990; 8 (Suppl 6): S135–S140. nal blood pressure variations in essential hyperten- 17 Gatzka CD and Schmieder RE. Improved classification sion. J Hypertens 1992; 10: 683–692. of dippers by individualized analysis of ambulatory 5 Fagard R, Staessen JA, Thijs L. The relationships blood pressure profiles. Am J Hypertens 1995; 8: between left ventricular mass and daytime and night 666–671. time blood pressures: a meta-analysis of comparative 18 Schillaci G et al. Non-invasive ambulatory BP moni- studies. J Hypertens 1995; 13: 823–829. toring during the night: randomised comparison of dif- 6 Shimada K, Kawamoto A, Matsubayashi K, Ozawa T. ferent reading intervals. J Hum Hypertens 1994; 8: Silent cerebrovascular disease in the elderly. Corre- 23–27. lation with ambulatory pressure. Hypertension 1990; 19 Bursztyn M, Mekler J, Wachtel N, Ben-Ishay D. Siesta 16: 692–699. and ambulatory blood pressure monitoring, compar- 7 Kobrin I, Oigman W, Kumar A. Diurnal variation of ability of the afternoon nap and night sleep. Am J Hyp- blood pressure in elderly patients with essential ertens 1994; 7: 217–221. hypertension. J Am Geriatr Soc 1984; 32: 896–899. 20 Rosansky SJ, Menachery SJ, Wagner CM, Jackson K. 8 Verdecchia P et al. Blunted nocturnal fall in blood The effect of sleep intervals on analysis of 24-h ambu- pressure in hypertensive women with future cardio- latory blood pressure data. Am J Hypertens 1995; 8: vascular morbid events. Circulation 1993; 88: 986–992. 672–675. 9 Verdecchia P et al. Ambulatory blood pressure. An 21 Pickering TG. How should the diurnal changes of independent predictor of prognosis in essential hyper- blood pressure be expressed? Am J Hypertens 1995; 8: tension. Hypertension 1994; 24: 793–801. 681–682. 10 Joint National Committee on Detection, Evaluation 22 Mulcahy D et al. Heart rate and blood pressure conse- Siesta and analysis of ambulatory BP profiles GS Stergiou et al 131 quences of an afternoon siesta. Am J Cardiol 1993; 71: assessing agreement between two methods of clinical 611–614. measurement. Lancet 1986; i: 307–310. 23 O’Brien E et al. The British Hypertension Study Proco- 25 Khoury AF, Sunderajan P, Kaplan NM. The early tol for the evaluation of automated an semi-automated morning rise in blood pressure is related mainly to blood pressure measuring devices with special refer- ambulation. Am J Hyperten 1992; 5: 339–344. ence to ambulatory systems. J Hypertens 1990; 8: 26 James GD et al. The reproducibility of average ambu- 607–619. latory, home, and clinic pressures. Hypertension 1988; 24 Bland JM, Altman DG. Statistical methods for 11: 545–549.