Nighttime Blood Pressure Phenotype and Cardiovascular Prognosis: Practitioner-Based Nationwide JAMP Study

The practitioner-based, nationwide, multicenter, prospective, observational study (University Hospital Medical Information Network trials registration: UMIN000020377) was conducted in Japan. Recruitment occurred between 2008 and 2017, with completion of follow-up by December 31, 2022. This analysis includes data from all patients registered between 2008 and 2017 with follow-up information to the end of March 2019. The protocol was approved by the Institutional Review Board of Jichi Medical School, Tochigi, Japan, and the study was conducted according to Declaration of Helsinki principles. All participants provided written informed consent. The data, analytic methods, and study materials will not be made available to other researchers for purposes of reproducing the results or replicating the procedure.

The JAMP study was designed to investigate the prognostic effect of ABPM parameters in general practice. Participants were recruited by general practitioners from 36 prefectures around Japan using the following inclusion and exclusion criteria. Patients with at least 1 of the following cardiovascular risk factors were included: diabetes or glucose intolerance; dyslipidemia; hypertension; current smoking; renal disease; atrial fibrillation; metabolic syndrome; chronic obstructive pulmonary disease;¹⁷ or sleep apnea syndrome¹⁸ (see Methods in the Data Supplement for full definitions and exclusion criteria).

The primary end point was total cardiovascular events, including atherosclerotic CVD (ASCVD; fatal and nonfatal stroke, fatal and nonfatal coronary artery disease [CAD]) and HF (see Methods in the Data Supplement for full event definitions). Patients were followed up annually (clinic visit or by telephone) to determine vital status and the occurrence of cardiovascular events over the preceding year. All end point events were adjudicated by an independent end point committee who were unaware of the patient’s ambulatory BP profile and clinical characteristics. If events occurred on 2 or more occasions, the first occurrence was included in the analysis. CAD events in patients with HF were classified as CAD.

Patient history data were obtained from medical records. Office BP was measured using an upper arm cuff and standard mercury sphygmomanometer or validated oscillometric device. Ambulatory BP was determined using a validated device (TM-2425/2431; A&D Co, Saitama, Japan)¹⁹ programmed to take readings every 30 minutes and repeat a reading if SBP (60–280 mm Hg), diastolic BP (40–160 mm Hg), pulse pressure (10–150 mm Hg), or pulse rate (30–200 beats/min) fell outside predefined acceptable ranges. Individuals were instructed to rest or sleep during nighttime and maintain usual daytime activities. Daily activities and sleep and wake times were recorded in a diary.

After wearing monitors for at least 24 hours, participants returned to the clinic, and device data were downloaded. Nighttime readings were defined as those taken from the self-reported time of falling asleep to waking time; all other readings were defined as daytime values. For analysis, ambulatory recordings had to include at least 6 daytime and 3 nighttime readings.²⁰ Patients were classified into 4 groups based on the reduction in SBP while asleep versus awake: extreme dippers (≥20%); dippers (10%–<20%); nondippers (0%–<10%); and risers (any increase).¹⁹

Data are presented as mean values with SD, and percentages. Demographic variables and clinical characteristics were compared between the nocturnal BP dipping pattern groups using a t test for the slope in linear regression models (continuous variables) or the Cochran-Armitage test (trends in categorical variables). The cumulative incidence of cardiovascular events in nocturnal dipping pattern subgroups was visualized using Kaplan-Meier curves, adjusted for covariates (age, sex, body mass index, smoking, alcohol intake, diabetes, dyslipidemia, CVD history, antihypertensive drug use, bedtime antihypertensive dosing), and office and 24-hour SBP.

Cox proportional hazards models were used to calculate hazard ratio (HR) and 95% CI values for the risk of cardiovascular events associated with a 20-mm Hg increase in SBP, and a 10% increase in nocturnal SBP dipping (after adjustment for the above covariates); the dipper group was used as the reference in dipping status models. Additional adjustments included office SBP (model 1); office SBP and 24-hour SBP (model 2); office SBP and daytime SBP (model 3); and office SBP and nighttime SBP (model 4). The proportionality assumption for the Cox analyses was confirmed graphically, and comparison of the discriminative ability of each model was conducted using Harrell’s C-statistics (with 95% CI values calculated by bootstrapping),²¹ net reclassification improvement, and integrated discrimination improvement.²² Heat maps were used to visually represent the relative associations between nocturnal SBP dipping status and nighttime SBP and the occurrence of cardiovascular events.

All statistical analyses were performed with SAS system, v9.4 (SAS Institute, Cary, NC). Two-sided P values <0.05 were defined as statistically significant.

A total of 130 doctors from 116 institutions (72 primary practices, 40 hospital-based outpatient clinics, 3 specialized university hospitals, and 1 national center hospital) recruited patients to the study. Of the 6772 participants enrolled, 6359 were included in the final analysis; 6288/6772 participants had 20 daytime and 7 nighttime ABPM recordings. Information about patient withdrawals is provided in the Methods in the Data Supplement (Figure I in the Data Supplement). Almost half of all participants were men (47.6%), 76.7% were taking antihypertensive medication, and mean (±SD) age was 68.6±11.7 years (range, 21–96 years, 67.5% age >65 years) (Table 1). Mean (±SD) follow-up duration was 4.5±2.4 years, during which there were 306 total CVD events: 218 ASCVD events including 119 strokes and 99 CAD events, and 88 episodes of HF (Table I in the Data Supplement).

Adjusted cumulative incidence rates of CVD events by baseline nocturnal SBP dipping status are shown in Figure 1A. The riser versus dipper pattern was associated with a significantly increased risk of total CVD (P=0.024) and HF (P=0.004), independent of office and 24-hour SBP. Adjustment for nighttime rather than 24-hour SBP showed that the riser versus dipper pattern was significantly associated with the occurrence of HF (P=0.027) but not CAD (Figure II in the Data Supplement).

Increasing nighttime, but not office, SBP was also significantly associated with CVD event risk. A 20-mm Hg increase in nighttime SBP was significantly associated with a 21% to 36% increase in CVD event risk, whereas daytime SBP was significantly associated with a 22% increase in stroke risk (Table 2). The riser pattern was also significantly associated with higher total CVD event risk compared with the dipper pattern (58% increase; P=0.007) (Table 2). The risk associated with a riser pattern was greatest for HF (HR, 2.53; P=0.002) and remained significant after adjustment for office SBP and 24-hour SBP (HR, 2.45; P=0.004), for office SBP and daytime SBP (HR, 2.54; P=0.002), and for office SBP and nighttime SBP (HR, 2.23; P=0.027) (Table 2). The risk associated with the riser pattern also persisted after the addition of morning SBP surge to model 1 (total CVD: HR, 1.48, P=0.034; HF: HR, 2.53, P=0.005).

Heat maps for the 5-year risk of CVD showed that higher nighttime SBP was associated with a greater relative risk for CVD events than dipping status (Figure 1B). The risk of both ASCVD and stroke increased as nighttime SBP decreased in extreme dippers but increased in parallel with increasing nighttime SBP in the other dipping status groups (Figure 1B). The risk of CAD and HF was highest in individuals with a riser pattern and higher nighttime SBP (Figure 1B). There was a significantly higher risk of stroke in patients with well-controlled 24-hour SBP and an extreme dipper pattern (HR, 2.30 [95% CI, 1.22–4.35]; P=0.010), and a significantly higher risk of HF in those with well-controlled SBP and a riser pattern (HR, 3.78 [95% CI, 1.61–8.89]; P=0.002) (Figure 2).

The base model yielded a C-statistic value of 0.717 (95% CI, 0.685–0.752) for total CVD, 0.694 (95% CI, 0.662–0.723) for ASCVD, and 0.816 (95% CI, 0.781–0.877) for HF (Table 3). Adding nighttime SBP to the base model resulted in small but statistically significant increments in C-statistic values (P=0.003, P=0.028, and P=0.018, respectively); a similarly small but significant increase in the C-statistic was also seen for HF when dipping status was added to the base model (P=0.037) (Table 3). In addition, increments in net reclassification improvement for total CVD risk after the addition of nighttime SBP and dipping status to the base model were 0.150 (95% CI, 0.026–0.275; P=0.011) and 0.200 (95% CI, 0.059–0.324; P=0.003), respectively, whereas the increment in net reclassification improvement for HF risk when dipping status was added was 0.413 (P<0.001) (Table II in the Data Supplement). Increments in integrated discrimination improvement for the risk of total CVD, ASCVD, and HF when nighttime SBP and dipping status were added to the base model were all statistically significant (Table III in the Data Supplement). However, model performance to predict the risk of total CVD, ASCVD, and HF was not significantly improved by adding daytime SBP when looking at the C-statistic (Table 3) or net reclassification improvement (Table II in the Data Supplement). Adding daytime SBP to the model increased integrated discrimination improvement values for total CVD and ASCVD risk but not HF risk (Table III in the Data Supplement).

Our findings in a population of mostly treated patients with hypertension suggest that each 20-mm Hg increase in nighttime SBP was associated with increased risk of ASCVD (21% increase, P=0.017) and HF (36% increase, P=0.009). Each 20-mm Hg increase in daytime SBP was associated with a significantly increased relative risk of ASCVD (22% increase, P=0.048), but not total CVD (16% increase, P=0.078). The model performance for discrimination of total CVD, ASCVD, and HF risk was slightly but significantly improved by the addition of nighttime BP and dipping status, whereas adding daytime BP slightly improved model performance for ASCVD risk only. Therefore, higher nighttime BP and dipping status appear to be more important contributors to cardiovascular risk than daytime BP, especially for HF.

The clinical impact of elevated nighttime BP versus daytime BP might depend on the stage of hypertension and be affected by antihypertensive drug therapy. In early-stage hypertension where patients have high-normal BP or mild hypertension, daytime BP may be a stronger determinant of left ventricular hypertrophy. Conversely, the prognostic impact of nighttime BP increases in patients with treated hypertension. In the current study, in which 77% of participants were receiving antihypertensive therapy, office SBP was not significantly associated with CVD risk. Physicians often use office BP to guide the initiation and intensification of antihypertensive medication. Thus, during the follow-up period, both office BP and daytime BP might be controlled by antihypertensive medication. However, assessment of only office and daytime BP may leave a significant proportion of medicated patients with undetected nocturnal hypertension, and therefore at high risk for cardiovascular events.

ABPM is the standard approach to determine nighttime BP, but this is not widely implicated in routine clinical practice. Home BP monitoring devices with the ability to monitor nocturnal BP have recently become available^10,23 but are not yet in widespread clinical use. In the J-HOP study (Japan Morning Surge-Home Blood Pressure), we suggested that nocturnal BP measurements using a home device were a significant predictor of ASCVD events, independent of office, morning, and evening home BP.¹¹

In this study, the riser pattern of nighttime BP was significantly associated with the risk of both total CVD events (HR, 1.58; P=0.007) and HF (HR, 2.53; P=0.002). The riser pattern is the extreme end of the continuum of nondipping nocturnal BP and disrupted circadian variation, and has been associated with organ damage and poor cardiovascular prognosis.^12,14 We found that the association between the riser pattern and total CVD events remained significant even after adjusting for 24-hour and daytime SBP, but the association disappeared after adjustment for nighttime SBP. This suggests that higher nighttime BP is more important than the riser patter as a risk factor for total CVD. This is consistent with the recent findings from the International Database of Ambulatory Blood Pressure in relation to Cardiovascular Outcome,¹⁵ suggesting that the risk associated with a riser pattern is at least partly mediated by high nighttime BP. Similarly, the increased HF risk associated with the riser pattern decreased slightly after controlling for nighttime SBP in our study, but the risk remained at least 2 times higher (HR, 2.23; P=0.027) in the riser group, indicating that a riser patter might be more important than nighttime BP as a risk factor for HF. Only 1 previous study has evaluated the link between ambulatory BP and HF.¹³ It showed that a riser pattern was an independent predictor of HF risk, even when cases of myocardial infarction were excluded. The same nonischemic definition of HF events was used in our study.

In our study, CAD risk was not significantly associated with either higher nighttime SBP or a riser pattern of nocturnal BP. Nevertheless, CAD is an important cause of HF, as shown in a recent study of older patients from Japan.²⁴ In the current study, when acute HF occurred with acute coronary syndrome, it was classified as CAD. Thus, the observed association between nighttime SBP and risk of HF is likely to be independent of a possible overlap between HF and CAD.

Left ventricular hypertrophy is the form of target organ damage most likely to precede nonischemic HF. Other studies have shown that nighttime BP and the nocturnal fall in BP are stronger correlates of left ventricular hypertrophy than daytime or average 24-hour BP.²⁵ In addition, nighttime BP was a risk factor for the development of left ventricular hypertrophy and increased natriuretic peptide levels in the J-HOP study.²⁶ Data from a large general population in Italy were similar, showing that nighttime BP was a reliable independent predictor of the development of left ventricular hypertrophy in subjects with normal left ventricular mass.²⁷ Furthermore, a recent comprehensive 2-dimensional echocardiographic examination including multilayer strain analysis demonstrated an association between the riser pattern and left ventricular mechanical dysfunction in patients with hypertension.²⁸

The association between a riser pattern and HF, independent of 24-hour, daytime and nighttime BP, is indicative of a “beyond BP” pathophysiological mechanism. There are 3 possible mechanisms to explain this. The first is vascular factors, with the riser pattern being associated with advanced vascular disease such as endothelial dysfunction and increased arterial stiffness. The second is circulating volume. Nondipping of nocturnal BP, including a riser pattern, is known to be associated with increased circulating volume, which is predominantly determined by salt sensitivity and salt intake. Decreased salt intake and treatment with diuretics reduces nighttime BP to a greater extent than daytime BP, causing a shift from a nondipper to a dipper pattern.^29,30 The third is sympathetic nerve activity. The riser pattern is characterized by high muscle sympathetic nerve traffic,³¹ and recent trial data showed that renal denervation significantly reduced 24-hour BP, including nighttime BP.³² All these factors increase preload or afterload, or directly impact on the left ventricle, contributing to development of HF.

Our findings suggest that excessive reduction of BP during sleep may also be detrimental. Extreme dipper patients with well-controlled 24-hour SBP showed a significantly increased risk of stroke, and stroke risk increased as nighttime SBP decreased. These associations need to be investigated in prospective studies. Given that the 2 extreme diurnal BP patterns were significantly associated with cardiovascular event risk in patients with well-controlled 24-hour SBP (stroke in extreme dippers and HF for those with a riser pattern), there appears to be a need for pathophysiology-based antihypertensive strategies that can reduce these residual risks.

One of the strengths of this study is the use of consistent technology, methodology, and definitions across centers. In addition, ambulatory BP parameters were defined based on individual awake-sleep behavior, which is the gold standard approach for defining nighttime BP and accurately assessing diurnal BP variation.¹⁹ However, some limitations need to be taken into account when interpreting our findings. First, ambulatory BP data were obtained once at baseline, and we do not have information on the contribution of subsequent changes in ambulatory BP to prognosis. Nevertheless, use of effective antihypertensive therapy is likely to weaken, rather than strengthen, associations between baseline ambulatory BP and cardiovascular outcomes. This analysis focused on systolic, rather than diastolic, BP because of the older age of our study population. Furthermore, study evaluations did not include echocardiography, precluding differentiation between HF with preserved versus reduced ejection fraction. The study population was Japanese, limiting external validity.