Prediction of Stroke by Home “Morning” Versus “Evening” Blood Pressure Values
The Ohasama Study
Predictive power of self-measured blood pressure at home (home BP) for cardiovascular disease risk has been reported to be higher than casual-screening BP. However, the differential prognostic significance of home BP in the morning (morning BP) and in the evening (evening BP), respectively, has not been elucidated. In the Ohasama study, 1766 subjects (≥40 years) were followed up for an average of 11 years. The predictive power for stroke incidence of evening BP was compared with that of morning BP as continuous variables. The Cox regression model demonstrated that evening BP and morning BP predicted future stroke risk equally. Subjects were also assigned to 1 of 4 categories based on home BP. In this analysis, stroke risk in morning hypertension ([HT] morning BP ≥135/85 mm Hg and evening BP <135/85 mm Hg; relative hazard (RH): 2.66; 95% CI:1.64 to 4.33) and that in sustained HT(morning BP and evening BP ≥135/85 mm Hg; RH: 2.38; 95% CI: 1.65 to 3.45) was significantly higher than that in normotension (morning BP and evening BP <135/85 mm Hg). The risk in morning HT was more remarkable in subjects taking antihypertensive medication (RH: 3.55; 95% CI: 1.70 to 7.38). Although the risk in evening HT (morning BP <135/85 mm Hg and evening BP ≥135/85 mm Hg) was higher than that in normotension, the differences were not significant. In conclusion, morning BP and evening BP provide equally useful information for stroke risk, whereas morning HT, which indicates HT specifically observed in the morning, might be a good predictor of stroke, particularly among individuals using anti-HT medication.
- home blood pressure
- general population
- morning-home blood pressure
- evening-home blood pressure
- Ohasama study
Self-measurement of blood pressure (BP) at home (home BP) by individual patients is highly reproducible and reliable and is acknowledged worldwide as a useful clinical tool.1–4 We have reported previously the strong predictive power of home BP measurements in the morning for cardiovascular disease mortality and stroke incidence.5–7 It is generally agreed that the prognostic power of home BP is higher than casual-screening BP in accordance with these recent studies.5–8
Several guidelines recommended that home BP should be measured both in the morning and in the evening (eg, European Society of Hypertension guidelines based on the German Hypertension League recommendation9,10), and it has been recommended in the 7th Report of the Joint National Committee that the home BP level should be evaluated as the average of all of the BP values measured.1 Because of circadian BP variation and other latent confounding factors, the characteristics of home BP in the morning (morning BP) and that in the evening (evening BP) must be different.11,12 Our previous study showed that there was a substantial difference between morning BP and evening BP.11 Moreover, antihypertensive medications were reported to affect circadian BP variation.13 Although morning BP values have a high predictive power, little is known about the predictive value of evening BP. Bobrie et al8 demonstrated that the home BP, which was averaged from the morning BP and evening BP readings, had good prognostic value among elderly (≥60 years) hypertensive patients. However, the clinical significance of the home BP in the evening versus that in the morning was unclear. The purpose of the present study is to evaluate the clinical significance of evening BP as well as morning BP for prediction of stroke incidence using data derived from the Ohasama study, a long-term cohort study in the northern part of Japan.
The present study is a part of the longitudinal observational study of subjects who have been participating in our home BP measurement project in Ohasama, a rural community in the northern part of Japan, since 1987. The study protocol was approved by the Institutional Review Board of Tohoku University School of Medicine and by the Department of Health of the Ohasama Town Government. Informed consent was obtained from each subject.
The socioeconomic and demographic characteristics of this region and the details of the selection procedure of study populations have been described previously.5–7,14–16 There were 1913 eligible individuals aged ≥40 years who measured their morning BP ≥3 times (3 days). This criterion was based on our previous observation that home BP level obtained for the first 3 days was not significantly different from that obtained for the entire study period.14 For the current analysis, we excluded 56 individuals who did not measure their evening BP ≥3 times. Because 91 individuals had a previous history of stroke, they were excluded from the present analysis to examine the risk of the first onset of stroke. Therefore, the study population consisted of 1766 individuals.
Physicians and well-trained public health nurses conducted health education classes to inform the subjects on how to measure and record home BP. After their ability to measure home BP was verified, subjects were asked to measure their own BPs once in the morning, in the sitting position within 1 hour after awaking, and after ≥2 minutes of rest and to record the measurements for 4 weeks. When individuals were taking antihypertensive medications, morning BP was measured before taking medications. Subjects were also asked to measure their BPs once in the evening just before going to bed for 4 weeks. We allowed subjects to measure their own BP ≥2 times on each occasion; however, the first measurement value from each occasion was the value that was used for analysis to exclude subjects’ selection biases. All of the subjects were instructed to hold their cuff-covered arms at heart level during home BP measurements. These procedures were described in detail in our previous report14 and followed the Japanese guidelines for self-monitoring of BP at home.3 Home BP was measured using the HEM 401C (Omron Healthcare Co Ltd), a semiautomatic device based on the cuff-oscillometric principle, which generates a digital display of both systolic and diastolic BP.17
Casual BP was measured among 1661 of the 1766 study subjects. Subjects were seated and at rest for ≥2 minutes, then casual BP was consecutively measured 2 times by nurses or technicians. A semiautomatic BP measuring device (USM700F; Ueda Electronic Work Co, Ltd) based on the microphone method was used.14 Casual BP of each subject was the average of 2 consecutive casual BP readings taken at the beginning of the study.
The devices met the criteria set by the Association for the Advancement of Medical Instrumentation.18 We used a standard arm cuff for both casual and home BP measurements, because the arm circumference of subjects was ≤34 cm.
Classification of Groups Based on Home BP
The morning BP values and the evening BP values were averaged separately in individuals, eg, the morning BP value in an individual who measured his/her BP for 20 days was the average of these 20 measurements. The combined BP values were the average of morning BP and evening BP.
In the present analysis, we set the criteria of hypertension (HT) based on home BP as ≥135/85 mm Hg according to recent guidelines.1,2,19 When a systolic or diastolic BP was in a different category, the higher category was applied. As shown in Figure 1, all of the subjects were assigned to 1 of 4 categories based on their own home BP: normotension ([NT] both morning BP and evening BP <135/85 mm Hg); morning HT (morning BP ≥135/85 mm Hg and evening BP <135/85 mm Hg); evening HT (morning BP <135/85 mm Hg and evening BP ≥135/85 mm Hg); and sustained HT (both morning BP and evening BP ≥135/85 mm Hg).
Follow-Up and Risk Assessment
We accumulated follow-up data from 1987 until December 31, 2001. The subjects’ residence status in Ohasama was confirmed by registration cards. These cards are accurate and reliable, because they are used for pensions and social security benefits in Japan.
The incidence and past history of stroke and transient ischemic attack (TIA) were investigated through the Stroke Registration System of Iwate Prefecture, death certificates, receipt of National Health Insurance, and questionnaires sent to each household at the time of home BP measurement. The information was then confirmed by checking the medical records of Ohasama hospital where >90% of the subjects had their regular health checkups. We used computed tomography scans and MRI reports to determine the clinical definition of stroke. For 3% of stroke cases, death certificates were the only source of information. The analysis included only the first event in those who had multiple nonfatal events. The diagnostic criteria of stroke, TIA, and their subtypes were based on the system for the Classification of Cerebrovascular Disease III by the National Institute of Neurological Disorders and Stroke.20
Other information for individuals such as height, weight, smoking status, use of antihypertensive medication at baseline, or history of heart disease, hypercholesterolemia, or diabetes mellitus was obtained from questionnaires sent to each household at the time of home BP measurements, from records of annual health checkups, and from medical records at Ohasama Hospital. Subjects using lipid-lowering drugs or those with serum cholesterol levels of ≥5.68 mmol/L (220 mg/dL) were considered to have hypercholesterolemia. Subjects with a fasting glucose level of ≥7 mmol/L (126 mg/dL), nonfasting glucose level of ≥11.11 mmol/L (200 mg/dL), or those using insulin or oral antihyperglycemic drugs were defined as having diabetes mellitus. A past history of cardiovascular disease included a history of myocardial infarction, angina pectoris, atrial fibrillation, and cardiac failure.
The risk of the first stroke or TIA onset was examined using the Cox proportional hazards model adjusted for possible confounding factors: age, sex, body mass index ≥25 kg/m2, current or ex-smoking, diabetes mellitus, hypercholesterolemia, and past history of cardiovascular disease. The dependent variable was the number of days from the initial home BP measurement to the date of stroke, TIA, or censoring. Participants who died from causes other than fatal stroke or who were lost to follow-up were treated as censored.
The estimated relative hazard (RH) and the 95% CI of variables were derived from the coefficient and SEM determined by the Cox proportional hazards model. The RH is expressed as the RH for each 10-mm Hg (systolic) or 5-mm Hg (diastolic) increase in BP. Comparison of corresponding regression coefficients and log likelihoods in the Cox model were used to compare the strength of each morning BP, evening BP, and combined BP. In the categorical analysis, the RH is expressed relative to the reference group (RH=1). Separate models were used for BP classification after verification of the assumption of proportionality for the Cox proportional hazards model.21 We also assessed the interaction between antihypertensive medication and BP values or BP groups by adding an interaction variable to the Cox model.21 All of the data were shown as mean (SD) unless otherwise stated. Student t test, Fisher’s exact test, and ANOVA were used for appropriate analysis. A P<0.05 (2-sided test) was accepted as indicative of statistical significance. All of the statistical calculations were conducted using the SAS system (version 9.1, SAS Institute Inc).
The characteristics of subjects classified by 4 BP categories and by use of antihypertensive medications are presented in Table 1. The mean (SD) age was 60.1 (11.0) years, whereas the average age in the sustained HT group was significantly higher than that in the other groups. The ratio of men to women was 40:60. The mean numbers of measurements for morning BP and evening BP of all of the subjects were 23.0 (7.0) and 23.6 (7.1), and the mean systolic/diastolic morning BP and evening BP values of all of the subjects were 125.0 (15.0)/75.0 (10.0) mm Hg and 123.0 (14.5)/73.2 (9.5) mm Hg, respectively. Of the 1766 study subjects, 504 (29%) were treated with antihypertensive medication at baseline, 394 (22%) were current or ex-smokers, 15 (1%) had a history of heart disease, 212 (12%) had a history of diabetes mellitus, and 204 (12%) had a history of hypercholesterolemia. Among 387 treated subjects, 72.9% were prescribed calcium channel blockers, 30.0% diuretics, 27.4% β-blockers, 10.9% angiotensin-converting enzyme inhibitors, 1.6% α-blockers, and 6.7% others.
The subjects were followed up for a median of 10.6 (interquartile range: 8.9 to 13.9) years to a maximum of 13.9 years. Thirty-four subjects (1.9%) had moved away and could not be followed up, and 262 deaths (14.8%) were identified from the residents’ registration cards. We observed 156 incident cases of first stroke or TIA among the 1766 individuals: 106 (68%) cerebral infarctions, 31 (20%) intracerebral hemorrhages, 12 (8%) subarachnoid hemorrhages, 4 (3%) TIAs, and 3 (2%) unknown causes. Among untreated subjects, there were 78 (6.2%) stroke or TIA cases, whereas there were 78 (15.5%) cases among subjects taking antihypertensive medication.
Table 2 displays adjusted RH of occurrence of stroke or TIA incidences with a BP per increase of 10 mm Hg (systolic) and 5 mm Hg (diastolic). Among 1766 study subjects, the RHs for total stroke risk based on systolic morning BP, systolic evening BP, diastolic morning BP, and diastolic evening BP increased linearly with the elevation of BP values (all P<0.0001). Log likelihood tests indicated that the predictive power of diastolic combined BP was marginally superior to diastolic morning BP (P=0.047), whereas there were no significant differences between diastolic combined BP and evening BP and among systolic combined BP and each morning BP or evening BP (all P>0.05). In a subgroup of subjects treated with antihypertensive medication, morning BP and evening BP, respectively, predicted the risk of all of the subtypes of stroke except ischemic stroke based on systolic evening BP. No significant increase of hemorrhagic stroke risk was observed in the morning BP and in the evening BP among untreated subjects. There was no significant interaction between the use of antihypertensive medication and each BP value for stroke risk (all P>0.05). Among 1661 subjects who measured casual BP in addition to home BPs, adjustment of casual BP had little impact on the results; casual BP did not predict any subtype of stroke events independent of home BPs (Table 3; all P>0.1 for casual BP).
The risks of first stroke or TIA of the 4 categories based on morning BP and evening BP are shown in Figure 2. Among all of the subjects (Figure 2a), the risk of stroke in morning HT (RH: 2.66; 95% CI: 1.64 to 4.33) and that in sustained HT (RH: 2.38; 95% CI: 1.65 to 3.45) were significantly higher than the risk in NT. As shown in Figure 2b, the risk of stroke in morning HT was more remarkable when subjects were taking antihypertensive medications at baseline (RH: 3.55; 95% CI: 1.70 to 7.38). However, the interaction between use of antihypertensive medication and home BP for stroke risk was not statistically significant (P=0.2 among subjects with NT and those with morning HT). Although the risk in evening HT subjects tended to be higher than that in the NT subjects, there were no significant differences (all P>0.05). There were no significant differences among morning HT, evening HT, and sustained HT when these categories were compared directly in the Cox model (all P>0.2).
We also compared (1) morning HT group (morning HT and sustained HT) with the remaining 2 groups and (2) evening HT group (evening HT and sustained HT) with the remaining 2 groups, respectively. The stroke or TIA risk in the evening HT group was significantly higher than that in the remaining 2 (NT and morning HT) groups (Table 4). Furthermore, individuals in the morning HT group also had a significant stroke or TIA risk when compared with those in the remaining 2 (NT and evening HT) groups (Table 4). The risk in the morning HT group was more pronounced when subjects were taking antihypertensive medications at baseline (Table 4). There were no significant interactions between antihypertensive medication and risk classifications (all P>0.1).
The results of the ischemic type of stroke were essentially similar to those of total stroke; the risk of ischemic stroke in morning HT (RH: 3.00; 95% CI: 1.69 to 5.34) and the risk in sustained HT (RH: 2.55; 95% CI: 1.64 to 3.97) was significantly higher than the risk in NT, whereas the risk in evening HT (RH: 1.87; 95% CI: 0.78 to 4.48) was not significantly different from that in NT. The risk of hemorrhagic type of stroke was not assessed because of the insufficient number of events.
This is the first study to demonstrate the difference in the predictive power of home BP measured at different periods of time, that is, morning and evening. Our findings were based on a comprehensive follow-up system in the Ohasama cohort and on accurate diagnoses of stroke and subtypes on the basis of computed tomography/MRI.6 Our previous studies have shown high reproducibility, reliability, and predictive power of home BP in the morning.5–7,11,14–17,22 In the Self-Measurement of Blood Pressure at Home in the Elderly: Assessment and Follow-Up (SHEAF) study, another prospective study to evaluate the prognostic significance of home BP, elderly hypertensive subjects with antihypertensive medication were asked to measure their BPs at home; the result indicated that the average of morning and evening BP values was a better predictor for cardiovascular disease than the average of casual-screening BP.8 In the present study, we demonstrated that the predictive power of morning BP, as well as evening BP, was essentially similar to that of combined morning and evening BP; although adding diastolic combined BP values marginally improved the goodness of fit in the model based on diastolic morning BP values, further studies are required to clarify the clinical significance. However, it is noteworthy that individuals with morning HT, whose BP values in the evening were relatively lower than those in the morning, had a high risk similar to sustained HT, in particular, among those taking antihypertensive medications.
Casual BP values in this study were defined as the average of only 2 measurements; this might account for the weaker predictive power of casual BP for stroke or TIA risk when compared with both morning BP and evening BP. However, in our previous study, even 1 measurement value of morning BP on the first occasion was superior to the average of 2 casual BP values in terms of stroke or TIA prediction.23 Such association was similarly observed in the present study for both morning BP and evening BP (data not shown). These results suggest that, in addition to the number of measurements, other factors, such as the lack of the white coat effect, may be associated with the superior predictive power of home BP.
We reported previously that morning HT was observed more frequently in treated hypertensive subjects than in untreated hypertensive subjects.11 In 1992, 70% of hypertensive subjects were treated with nifedipine sustained-release tablets, nicardipine sustained-release tablets, or by diltiazem sustained-release tablets in this Ohasama population.13 At that time, β-blockers, diuretics, and angiotensin-converting enzyme inhibitors were prescribed for 30%, 25%, and 10% of hypertensive subjects, respectively.13 Because the duration of action of most antihypertensive drugs used in 1992 was <12 hours for a twice-a-day prescription and <24 hours for a once-a-day prescription,22 it is likely that morning HT was mediated at least in part by insufficient duration of action of the antihypertensive drugs. Moreover, the trough/peak ratios of most antihypertensive drugs are reported to be <50%, although they are long-acting drugs or long-acting formulae.24 This may be one of the reasons why the predictive power of morning HT for stroke or TIA risk was stronger among treated subjects than that among all of the subjects. In the current study, however, we did not assess changes over time in antihypertensive treatment; it remains to be investigated whether changes in antihypertensive treatment during follow-up could modify the risk of stroke or TIA in subjects. It must be noted that, although new long-acting calcium channel blockers and new angiotensin II receptor blockers have been marketed, the control of morning BP is reported to be far from ideal even at the present time.25
Another possibility for the stronger prediction of morning HT in treated subjects compared with that in untreated ones is that treatment, per se, is a sort of marker of disease severity leading to a greater rate of events. Forty percent of the subjects under antihypertensive medication were included in the NT group, resulting in the low RH in the morning HT (Figure 2a). Most of the previous reports demonstrated a worse prognosis for treated hypertensive subjects than untreated hypertensive subjects.26,27 We should be concerned about residual stroke or TIA risks in hypertensive patients.
Several Japanese studies, including ours, demonstrated that evening BP was lower than morning BP.11,12,28 Conversely, 2 European studies reported that evening BP was higher than morning BP.29,30 One explanation for the difference may be that most Japanese people bathe every night. Another reason for the difference might be the time of evening BP measurement. European studies measured evening BP in the early evening (before dinner29 or 6:00 pm to 10:00 pm30), whereas in Japanese studies, measurements were taken in the late evening (before bedtime), based on the Japanese guidelines for home BP measurement.3 Approximately one third of adults in Ohasama drink alcohol in the evening,31 and this might also cause a transient fall in BP in the late evening. Home BP measured in the morning under well-controlled conditions may increase its predictive power, whereas it is likely that the uncontrolled measurement conditions in the evening may result in the insufficient predictive power of evening BP, especially of evening HT, for stroke and TIA incidence. Thus, the prognostic significance of evening home BP for people from other countries still remains to be investigated.
In addition, the first measurement of home BP is reported to be higher than the second or the third home BP values.29 Therefore, both morning BP values and evening BP values might be lower in the present study if the second or the third home BP values were analyzed. However, regression to the mean occurs during long-term measurement of home BP even for the first measurement of home BP.3 Moreover, the reference values for home BP have been derived from studies in the general population that used the average of the first measurement.6,32,33 Further studies are needed to clarify the clinical significance and reference values of the second or the third home BP values compared with the first home BP values.
Morning BP, evening BP, and combined values of the 2, respectively, were a good index for stroke or TIA risk when these parameters were used as continuous variables. Morning HT, which indicates HT specifically observed in the morning, had a high predictive power for stroke or TIA risk, particularly among individuals using antihypertensive medication. We conclude that home BP measurements are applicable to the intervention strategy for primary prevention of cardiovascular disease in individuals whether they measure home BP in the morning or in the evening, whereas morning HT based on home BP might be a good indicator for uncontrolled management of HT.
We are grateful to the staff members of the Iwate Prefectural Stroke Registry and to the public health nurses in Ohasama for their valuable support on this project and on the follow-up survey. We are also grateful to Kazuhiro Suzuki, Ayumi Kurimoto, and Rie Sato for their kind cooperation in the Ohasama study.
Sources of Funding
The Ministries of Education, Culture, Sports, Science and Technology (grants for scientific research: 14657600, 14370217, 15790293, 1654041, and 17790381) and of Health, Labour and Welfare (health science research grants and medical technology evaluation research grants); Grant-in-Aid for Japan Society for the Promotion of Science (JSPS) fellows (16.54041, 18.54042); the Japan Atherosclerosis Prevention Fund; the Uehara Memorial Foundation; and the Takeda Medical Research Foundation supported this work. The study sponsors had no direct role in the study design, data analysis, data interpretation, or writing of this article.
- Received April 3, 2006.
- Revision received April 7, 2006.
- Accepted August 2, 2006.
Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA, Izzo JL, Jones DW, Materson BJ, Oparil S, Wright JT, Roccella EJ. Seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. Hypertension. 2003; 42: 1206–1252.
Stergiou G, Mengden T, Padfield PL, Parati G, O’Brien E. Self monitoring of blood pressure at home. BMJ. 2004; 329: 870–871.
Ohkubo T, Imai Y, Tsuji I, Nagai K, Kato J, Kikuchi N, Nishiyama A, Aihara A, Sekino M, Kikuya M, Ito S, Satoh H, Hisamichi S. Home blood pressure measurement has a stronger predictive power for mortality than does screening blood pressure measurement: a population-based observation in Ohasama, Japan. J Hypertens. 1998; 16: 971–975.
Asayama K, Ohkubo T, Kikuya M, Metoki H, Hoshi H, Hashimoto J, Totsune K, Satoh H, Imai Y. Prediction of stroke by self-measurement of blood pressure at home versus casual screening blood pressure measurement in relation to the JNC-7 classification: the Ohasama study. Stroke. 2004; 35: 2356–2361.
Asayama K, Ohkubo T, Kikuya M, Metoki H, Obara T, Hoshi H, Hashimoto J, Totsune K, Satoh H, Imai Y. Use of 2003 European Society of Hypertension-European Society of Cardiology guidelines for predicting stroke using self-measured blood pressure at home: the Ohasama study. Eur Heart J. 2005; 26: 2026–2031.
O’Brien E, Asmar R, Beilin L, Imai Y, Mallion JM, Mancia G, Mengden T, Myers M, Padfield P, Palatini P, Parati G, Pickering T, Redon J, Staessen J, Stergiou G, Verdecchia P. European Society of Hypertension recommendations for conventional, ambulatory and home blood pressure measurement. J Hypertens. 2003; 21: 821–848.
Ishikawa J, Kario K, Hoshide S, Eguchi K, Morinari M, Kaneda R, Umeda Y, Ishikawa S, Kuroda T, Hojo Y, Shimada K. Determinants of exaggerated difference in morning and evening blood pressure measured by self-measured blood pressure monitoring in medicated hypertensive patients: Jichi Morning Hypertension Research (J-MORE) Study. Am J Hypertens. 2005; 18: 958–965.
Chonan K, Hashimoto J, Ohkubo T, Tsuji I, Nagai K, Kikuya M, Hozawa A, Matsubara M, Suzuki M, Fujiwara T, Araki T, Satoh H, Hisamichi S, Imai Y. Insufficient duration of action of antihypertensive drugs mediates high blood pressure in the morning in hypertensive population: the Ohasama study. Clin Exp Hypertens. 2002; 24: 261–275.
Imai Y, Satoh H, Nagai K, Sakuma M, Sakuma H, Minami N, Munakata M, Hashimoto J, Yamagishi T, Watanabe N, Yabe T, Nishiyama A, Nakatsuka H, Koyama H, Abe K. Characteristics of a community-based distribution of home blood pressure in Ohasama in northern Japan. J Hypertens. 1993; 11: 1441–1449.
Nagai K, Imai Y, Tsuji I, Ohkubo T, Sakuma M, Watanabe N, Kato J, Kikuchi N, Nishiyama A, Sekino M, Itoh O, Satoh H, Hisamichi S, Abe K. Prevalence of hypertension and rate of blood pressure control as assessed by home blood pressure measurements in a rural Japanese community, Ohasama. Clin Exp Hypertens. 1996; 18: 713–728.
Tsuji I, Imai Y, Nagai K, Ohkubo T, Watanabe N, Minami N, Itoh O, Bando T, Sakuma M, Fukao A, Satoh H, Hisamichi S, Abe K. Proposal of reference values for home blood pressure measurement: prognostic criteria based on a prospective observation of the general population in Ohasama, Japan. Am J Hypertens. 1997; 10: 409–418.
Imai Y, Abe K, Sasaki S, Minami N, Munakata M, Sakuma H, Hashimoto J, Sekino H, Imai K, Yoshinaga K. Clinical evaluation of semiautomatic and automatic devices for home blood pressure measurement: comparison between cuff-oscillometric and microphone methods. J Hypertens. 1989; 7: 983–990.
Association for the Advancement of Medical Instrumentation. American National Standards for Electronic or Automated Sphygmomanometers. Washington, DC: AAMI Analysis and Review; 1987.
National Institute of Neurological Disorders and Stroke Ad Hoc Committee. Special report from the National Institute of Neurological Disorders and Stroke. Classification of cerebrovascular diseases III. Stroke. 1990; 21: 637–676.
Woodward M. Epidemiology: Study Design and Data Analysis (Texts in Statistical Science), 2nd ed. London, United Kingdom: Chapman and Hall/CRC; 2005.
Ohkubo T, Asayama K, Kikuya M, Metoki H, Hoshi H, Hashimoto J, Totsune K, Satoh H, Imai Y, Ohasama Study. How many times should blood pressure be measured at home for better prediction of stroke risk? Ten-year follow-up results from the Ohasama study. J Hypertens. 2004; 22: 1099–1104.
Zanaad F, Matzinger A, Larche J. Trough/peak ratios of once daily angiotensin converting enzyme inhibitor and calcium antagonist. Am J Hypertens. 1996; 9: 633–643.
Andersson OK, Almgren T, Persson B, Samuelsson O, Hedner T, Wilhelmsen L. Survival in treated hypertension: follow up study after two decades. BMJ. 1998; 317: 167–171.
Nesbitt SD, Amerena JV, Grant E, Jamerson KA, Lu H, Weder A, Julius S. Home blood pressure as a predictor of future blood pressure stability in borderline hypertension. The Tecumseh Study. Am J Hypertens. 1997; 10: 1270–1280.