We aimed this study to test the hypothesis that short-term blood pressure (BP) variability and abnormal patterns of diurnal BP variation, evaluated by ambulatory BP (ABP), predicts risk of incident cardiovascular disease (CVD) in patients with type 2 diabetes (T2DM).
ABP monitoring (ABPM) was performed in 300 patients with uncomplicated T2DM without known CVD and without BP medications, who were followed for 54 ± 20 months. The relationships of different measures of BP variability, the presence of abnormal patterns of diurnal BP variation (nondipper, riser, or morning BP surge) and the standard deviations of awake and asleep ABP were determined. Cox proportional hazards models were used to estimate hazard ratios (HRs) and their 95% confidence intervals (CIs) before and after controlling for various covariates.
The mean age was 67.8 ± 9.6 years, 48% were male, 253 (84%) had a diagnosis of hypertension, and the mean of the standard deviations of awake systolic BP/diastolic BP (SBP/DBP) were 18 ± 6/11 ± 4 mm Hg, and those of sleep SBP/DBP were 13 ± 5/9 ± 3 mm Hg. During follow-up, there were 29 cardiovascular events. In multivariable analyses, the standard deviations of sleep SBP (HR = 1.08; 95% CI, 1.01–1.16, P < 0.05) and sleep DBP (HR = 1.13; 1.04–1.23, P < 0.01) were independently associated with incident CVD. Neither the nondipper and riser patterns nor the morning BP surge were associated with incident CVD events independently of clinic and 24-h BP levels.
Abnormal diurnal BP variation was not a predictor of CVD in patients with T2DM. Night time BP variability was an independent predictor of future incidence of CVD, suggesting that this measure could reflect pathophysiology of T2DM.
American Journal of Hypertension (2009). doi:10.1038/ajh.2008.294
There is accumulating evidence that ambulatory blood pressure (ABP) predicts cardiovascular disease (CVD) better than clinic blood pressure (BP), however, the use of ABP monitoring (ABPM) is not recommended in the current guidelines of American Diabetes Association.1 In addition to BP levels,2 BP variability can also be evaluated by ABP. Variability of BP, assessed by the s.d. of ABP readings, has been reported to be associated with an adverse cardiovascular prognosis in hypertensive patients.3–6 In these reports, there is conflicting evidence as to which types of BP variability measures best predict future cardiovascular events. BP variability during the daytime,3 night time,5,6 and both daytime and night time4 have all been reported to be associated with a higher incidence of CVD or CVD mortality. However, it has also been reported that standard deviations of 24-h, daytime, and night time BP were not associated with CVD when adjusted by 24-h average BP.7 Another measure of BP variability is the diurnal change of ABP, evaluated either as the dipping pattern, such as nondipping, reverse dipping (i.e., rising), or as the morning BP surge,8 which have also been found to predict future cardiovascular events in subjects with hypertension7,9,10 and type 2 diabetes (T2DM).11,12
There have been no studies comparing the associations of these different measures of BP variability with incident CVD in patients with T2DM. We performed this study to investigate the prognostic significance of these different measures of BP variability in an exclusively diabetic population, and thereby examine the generalizability of previous reports to diabetic patients.
This prospective study was performed in a sample of 300 asymptomatic T2DM patients who were seen in clinics at nine participating institutions in Japan: three clinics, two hospitals, and one outpatient clinic of a university hospital (the Jichi Medical School ABPM Study Wave 1); and one clinic and two hospitals in the Karatsu-Nishiarita Study.13–15 The subjects in this study are recruited from the same population who met the criteria of diabetes mellitus in our recent publication.2
Subjects and definitions. During the period of recruitment, 1990–1998 for the Jichi Medical School ABPM Study Wave 1 sample and 1996–2002 for the Karatsu-Nishiarita Study, subjects were enrolled consecutively while being treated or evaluated for hypertension in the clinic. The present study is restricted to those who had T2DM. At least two clinic BP readings were taken on each of two separate occasions after at least 5 min of rest in the sitting position, which were taken both before and after being fitted with an ABPM in subjects who stopped medication for ABPM. Hypertension was diagnosed when the clinic systolic BP (SBP) was >140 and/or diastolic BP (DBP) was >90 mm Hg on at least two occasions according to current guidelines,16 or by a previous diagnosis of hypertension with current antihypertensive medication use. Subjects took no antihypertensive medications for a minimum of 7 days before the ABPM and >95% took no medications during the 14 days preceding the ABPM study. T2DM was diagnosed according to the guidelines of the American Diabetes Association17 or a previous diagnosis and currently taking antidiabetic medication. We excluded patients with type 1 or secondary diabetes, renal dysfunction (serum creatinine >1.9 mg/dl), active hepatitis (type B, type C, alcoholic), or liver scirrosis, ischemic heart disease or other cardiac diseases, congestive heart failure, arrhythmias (including atrial fibrillation), stroke (including transient ischemic attacks), or other major concomitant non-CVDs. Left ventricular hypertrophy was not excluded, and noncardiac exclusion criteria were screened by routine clinical practice procedure, such as blood test, X-ray, abdominal echo, or gastric fiberscope. No shift workers existed in this population. Body mass index was calculated as weight/height2 (kg/m2). Smoking was defined as current smoking status. This study was approved by the Institutional Review Board of each participating hospital or clinic. All subjects were ambulatory and gave informed consent for the study.
ABPM. Noninvasive ABPM was performed on a weekday with an automatic system (either ABPM-630 (Omron Colin, Tokyo, Japan), TM2421, or TM2425 (A&D, Tokyo, Japan)) which recorded BP, using the oscillometric method and pulse rate (PR) every 30 min for 24 h. These devices have been previously validated.18,19 Awake and sleep time were defined based on patients' written diaries recorded during ABPM. Mean awake and sleep levels of SBP and DBP were calculated and the nocturnal BP fall (%) was calculated as (awake SBP – sleep SBP)/awake SBP. Nocturnal BP fall was classified as follows: dipper, if the fall was >10%; nondipper, if it was >0% but <10%; and riser, if it was <0%;9,10,13 and these three categories were used as measures of diurnal BP variation. Morning BP was defined as the mean BP during the 2 h after waking, and the morning BP surge as the morning BP minus the mean BP during the 1 h that included the lowest BP at night.8 Short-term BP variability was estimated by s.d. of the daytime and night time SBP and DBP.
Follow-up and events. The subjects' medical records were reviewed once a year after ABPM for the purpose of identifying incident CVD. The 99 participants from Jichi Medical School ABPM Study Wave 1 were followed from 1996 to 1998 for up to 5.7 years or until they moved, changed their telephone number, or died; the 201 participants from the Karatsu-Nishiarita Study were similarly followed from March 2004 to October 2006 for up to 9.7 years. Participants who developed a malignant disease and/or died from noncardiovascular causes were censored as of the time such events took place. The median follow-up period (5th and 95th percentiles) was 54.6 months (21 and 91 months). When subjects did not visit the clinics, we interviewed them by telephone. We defined three outcomes: stroke, fatal or nonfatal myocardial infarction, and sudden cardiac death. Strokes and cardiac events were diagnosed by the physician caring for the patient at the time of the event, and independent neurologists or cardiologists reviewed the cases and confirmed the diagnosis by referrals or medical records. Stroke was diagnosed on the basis of sudden onset of a neurological deficit that persisted for >24 h in the absence of any other disease process that could explain the symptoms. Stroke events included ischemic stroke (cerebral infarction and cerebral embolism), hemorrhagic stroke (cerebral hemorrhage and subarachnoid hemorrhage), and undefined types of stroke. We excluded transient ischemic attacks, in which the neurological deficit cleared completely in <24 h.10 Myocardial infarction was diagnosed based on the AHA criterion of “definite” myocardial infarction.20 Angina and congestive heart failure were not treated as end points.
Statistical analyses. All statistical analyses were carried out using SPSS/Windows, version 13.0 (SPSS, Chicago, IL). The data are expressed as the mean (±s.d.) or percentage. Adjusted hazard ratios (HRs) with 95% confidence intervals (CIs) were based on multivariable Cox regression analysis in which age (years), sex (male = 1, female = 0), body mass index (kg/m2), and either clinic SBP/DBP or 24-h mean ambulatory SBP/DBP were controlled a priori, whereas current smoking status (yes or no), antihypertensive medication (yes or no), total cholesterol, and serum creatinine (mg/dl) were treated as potential additional covariates and controlled only if they significantly predicted the outcome. The predictive utility of each BP variability measures (nondipper pattern and riser pattern, yes or no, morning SBP/DBP surge, standard deviations of awake, sleep SBP/DBP) was analyzed controlling for age, current smoking, and reference BP levels. We performed the Kolmogorov supremum test, as implemented in the PROC PHREG procedure of the SAS package for each of the Cox regression models reported in our article. For every model the result of the Kolmogorov-type supremum test was not significant (all Ps > 0.20). Therefore, the “proportional hazards assumption” was not violated. The null hypothesis was rejected when two-tailed P < 0.05.
The baseline characteristics are shown in Table 1. The mean age was 67.8 ± 9.6 years; there were 145 men and 155 women; 57% of subjects were taking antihypertensive medication.
During the follow-up of 53.8 ± 20.6 months, 29 cardiovascular events occurred. The incidence of CVD was 2.2/100 person-years. In univariable analyses, in addition to clinic and ABP levels, s.d. of awake DBP, and standard deviations of sleep SBP and DBP, and a riser pattern were all significantly associated with incident CVD, but morning BP surge was not (Figure 1). Regarding PRs, the HR (95% CI) of awake PR was 1.03 (0.99–1.07), P = 0.108, and sleep PR, HR = 1.05 (1.01–1.09), P = 0.024, but 24-h PR, HR = 1.03 (0.99–1.07), P = 0.18.
Univariable Cox regression analysis predicting cardiovascular events. Plots (bars) are shown as hazard ratio (95% CI) by 10 mm Hg increase of each blood pressure and 5 mm Hg increase of each s.d. of blood pressure values. CI, confidence interval; DBP, diastolic blood pressure; MS, morning surge; SBP, systolic blood pressure. *P < 0.05, **P < 0.01.
In our recent publication, the higher the ABP levels, the more CVD events occurred in diabetes.2 Based on the univariable analyses, we used clinic or ABP as a covariate for adjustment together with age, and current smoking, both of which were selected from a preliminary Cox regression analysis including essential and significant variables (age, sex, body mass index, smoking, and serum creatinine). In multivariable Cox regression analyses that controlled for age, current smoking, and clinic SBP/DBP, neither the nondipper or riser patterns nor the morning BP surge were associated with CVD (Figure 2). The standard deviations of awake SBP/DBP were not associated with CVD, however, the standard deviations of sleep SBP/DBP were significantly associated with incident CVD (Figure 2). When the same analyses were repeated using 24-h SBP/DBP in place of clinic SBP/DBP as a covariate (Figure 3), the association of the standard deviations of sleep SBP/DBP with CVD events continued to be significant, but none of the other measures of BP variability were significantly associated with incident CVD. The predictive value of sleep BP variability remained significant when the corresponding sleep BP levels were used instead of 24-h SBP/DBP as a covariate. When the nondipper/riser patterns were added in the same models based on our recent publication,2 the sleep BP variability also remained significant (data not shown). When we added sleep PR in multivariable models, the predictability of sleep SBP variability disappeared, but that of sleep DBP remained significant. There is a moderate, but not large, degree of collinearity (VIF <3 in all cases), our analyses showed that sleep BP variability has unique predictive utility whereas clinic BP/ABP level has predictive ability by itself, but not unique predictive ability if BP variability is also present. We have plotted the CVD incidence rate by quartile of BP variability and by 24 h BP level as shown in Figure 4. Consistent with our reported finding of the independent effect of DBP variability on CVD incidence rate (Figure 3), there is a fairly clear dose–response pattern when comparing the two extreme quartiles against the middle 50% (the P value for a linear dose–response pattern is 0.04).
Cardiovascular disease (CVD) incidence rates by quartiles of (a) sleep blood pressure (BP) variability and (b) 24-h BP levels. Sex and age (four categories) were adjusted using the direct method.
The predictive utility of each blood pressure (BP) measures (nondipper/riser pattern, yes or no, morning BP surge, and awake/sleep BP variability measures) adjusted by 24-h BP levels. Plots (bars) are shown as multivariable hazard ratio (95% confidence interval) by 10 mm Hg increase of each systolic BP/diastolic BP (SBP/DBP) and 5 mm Hg increase of each s.d. of SBP/DBP values controlling for age, current smoking, and 24-h SBP/DBP (each) as significant potential covariates. MS, morning surge. *P < 0.05, **P < 0.01.
The predictive utility of each blood pressure (BP) measures (nondipper/riser pattern, yes or no, morning BP surge, and awake/sleep BP variability measures). Plots (bars) are shown as multivariable hazard ratio (95% CI) by 10 mm Hg increase of each systolic BP/diastolic BP (SBP/DBP) and 5 mm Hg increase of each s.d. of SBP/DBP values controlling for age, current smoking, and clinic SBP/DBP (each) as significant potential covariates. MS, morning surge. *P < 0.05, **P < 0.01.
In T2DM patients with or without hypertension, night time BP variability was a strong predictor for CVD events independent of other covariates, including 24-h BP level and an abnormal circadian BP rhythm. This study is the first study showing the impact of BP variability on cardiovascular prognosis in type 2 diabetic patients. Although one abnormal diurnal BP variation, the riser pattern, was a significant predictor of incident CVD events,2 it was not independent of clinic or 24-h SBP. Neither the nondipper pattern nor the morning surge of BP was a predictor of incident CVD events.
Abnormal diurnal BP rhythm and morning BP surge in T2DM
In our study, neither an abnormal dipping pattern nor the morning BP surge was associated with incident CVD events independent of ABP level. The result with the dipping pattern was consistent with the finding of previous paper showing that ABP level was more important than the dipper/nondipper patterns.2,21 The HRs for risers were about 1.5–2, but our data show that BP levels are more important than the diurnal pattern of BP variation in patients with T2DM.2 The lack of significance of the morning BP surge was not surprising because nearly 50% of our diabetic patients showed a nondipping pattern, and hence a diminished morning surge. Diabetic patients have some degree of neuropathy, which is usually diagnosed at the time of the diagnosis of diabetes. Our results suggest that an abnormal dipping pattern and the morning BP surge in diabetes have no additional impact on CVD over and above the average 24-h BP. A study in diabetic patients has shown that morning BP evaluated at home was associated with micro- and macrovascular complications, but the study design was cross-sectional.22 For the management of BP in diabetic patients, the level of out-of-office BP such as ABP may be the most important measure,2 however, next to ABP levels, the home BP level may have additional impact on future cardiovascular events independent of clinic BP. This finding has important implications for the use of home BP monitoring, where the morning BP can easily be measured.
Night time BP variability in T2DM
We reported that ABP levels were more important than clinic BP level, and confirmed the importance of abnormal circadian BP variation in predicting future CVD events in T2DM.2 In addition to, and independent of these, BP variability during sleep was shown to be a significant predictor of future CVD.
In our study, the standard deviations of sleep SBP and DBP were independent predictors for CVD. However, the s.d. of awake BP was not. This finding is consistent with a report of isolated systolic hypertension showing that increased night time SBP variability was an independent risk factor for stroke.5 However, our findings were different from those of the other reports, in which either awake and sleep BP variability were both associated with cardiovascular outcomes4,6 or none of the s.d. measures were associated with CVD in a general population.7 The daytime BP variability in diabetic patients was reported to be higher than nondiabetic patients.23 The presence of advanced atherosclerosis24 and impaired baroreflex sensitivity, which is associated with diabetic neuropathy,25 can lead to increased variability of BP in diabetic patients.26 The reason why only BP variability during sleep was associated with cardiovascular events is unclear. The significance of sleep SBP variability disappeared when adjusted by sleep PR, which indicates that increased SBP variability can be modulated by impairment of normal parasympathetic activity. In a previous report of diabetic patients, it was reported that parasympathetic nerve activity was relatively low during the night and resulting in nocturnal sympathetic predominance.27 Diminished vagal modulation of the heart is one of the earliest and most prominent features of autonomic neuropathy in diabetic patients,27 which is an independent predictor of CVD28 and CVD mortality.29 The reason why our findings were in agreement with those in isolated systolic hypertension5 is not clear. However, the large fluctuations of awake BP in both diabetes and isolated systolic hypertension could have obscured the predictability of awake BP variability because isolated systolic hypertension in the elderly has sympathetic activation and impaired baroreflex sensitivity,30 both of which are also seen in diabetic patients. The night time BP variability as a significant predictor for CVD could reflect early changes in the autonomic nervous system and could be an additional indicator of adverse cardiovascular events.
Limitations of the study
There are some limitations in this study. First, the small sample size is a major limitation of this study. The sample size for detecting predictability of the standard deviations of awake SBP/DBP and riser/nondipper would be underpowered, which is a limitation to say that sleep BP variability is the only significant variable in predicting cardiovascular events. Because the study patients were recruited from the clinics of general internal medicine or community hospital, the number of diabetic patients for this ABPM study was limited. The small event number is also the limitation, however, the cardiovascular event rates 2.2/100 person-years were similar to the other outcome studies: major macrovascular events were 10.0% in the intense treatment group, and 10.6% in the control group in the ADVANCE trial;31 primary outcome (defined as either nonfatal myocardial infarction, nonfatal stroke, or death from cardiovascular cause) was 2.11% per year in the intensive treatment group, and 2.29% per year in the standard therapy group in the ACCORD study.32 Second, the evaluation of BP variability was made by ABPM. Compared to the intra-arterial method, this method is probably less accurate; however, its noninvasive nature enabled us to study a larger number of patients than previous studies.4,6 Withdrawal of medication for only 1–2 weeks before BP measurements does not exclude the possibility of a persistent, significant antihypertensive effect. Third, information on diabetic neuropathy, which could have influenced the BP variability, were not available in this study. Finally, the data on hemoglobin A1c and urinary albumin were limited to two-thirds of the patients. Because BP variability may not be determined by recent glycemic control, the lack of hemoglobin A1c does not seem to be essential.
In our diabetic population, neither an abnormal dipping pattern nor the morning BP surge was a predictor of CVD events, whereas the night time BP variability appeared to be a strong predictor, independent of ABP level and other traditional risk factors. These results imply that in addition to ABP levels, evaluation of night time BP variability, which could reflect early change of diabetic autonomic neuropathy, is important for the prediction of cardiovascular events.
This work was supported in part by grants-in-aid (1998–1999, 2001–2002, 2004–2005) from the Foundation for the Development of the Community, Tochigi, Japan, and NHLBI (R24 HL76857 and P01 HL 47540).
The authors declared no conflict of interest.