Introduction

Blood pressure (BP) fluctuates over time. However, a precise assessment of BP variability is only possible with beat-to-beat BP recordings using a specific methodology, such as intra-arterial BP monitoring or continuous finger BP measurements [1, 2]. Increases in BP variability can be observed across various timeframes [3, 4]. Over a short time period, ambulatory BP monitoring (ABPM) using a device that intermittently measures BP at fixed time intervals is widely used and has allowed for the estimation of BP variability, which has been shown to be associated with target organ damage and cardiovascular prognosis [5,6,7,8,9].

Pulse transit time (PTT), which refers to the travel time of the systolic pressure wave between two arterial sites, typically the aortic valve and a peripheral site, has been developed as a novel cuffless form of continuous BP monitoring [10,11,12]. Several previous studies reported that the PTT-estimated BP value was validated by the BP value evaluated by a mercury sphygmomanometer [10, 11]. The BP variability assessed by beat-to-beat PTT-estimated BP may detect BP variability to a greater extent than ABPM in situations of dramatic BP change. However, regarding the methodology of BP evaluation using PTT, there has been no study on whether transient BP elevation and fluctuation assessed by beat-to-beat PTT-estimated BP provides a clinical advantage compared to that assessed by intermittent BP readings at fixed time intervals.

In sleep disorder breathing (SDB), acute transient BP elevation and fluctuation are observed at the end of desaturation episodes during sleep [4, 13,14,15,16,17]. We hypothesized that BP readings assessed by beat-to-beat PTT-estimated BP variability (eBPBTB) would be a more precise measure of BP variability than those assessed by intermittent PTT-estimated BP variability at a fixed time interval (eBPINT) in patients with SDB. The aim of this study was to investigate the difference in BP parameters, including BP variability, between those assessed by eBPBTB and those assessed by eBPINT in patients suspected of SDB from a multicenter study.

Subjects and methods

Subjects

The subjects enrolled in this study were analyzed using a SOMNO touch RESP (Fukuda Denshi Co., LTd., Tokyo, Japan), which recorded nasal airflow, snoring sounds, thoracic and abdominal respiratory effort signals, ECG, oxygen saturation (SpO2) via pulse oximetry, PTT, R-R timing, finger plethysmography and body position. The subjects comprised 330 patients suspected of having SDB at 8 institutes (Supplementary Material) between May 2016 and August 2019. All recordings were analyzed for the oxygen desaturation index (ODI) using DOMINO Light software version 1.5.0 (Somnomedics, Randersacker, Germany). The 3% ODI was calculated as the number of oxygen desaturation events (reduction of 3% from baseline) per hour during the entire recording time. The Institutional Review Board of Jichi Medical University approved the study with a waiver of informed consent.

Demographic information was collected by physicians at each participating institute. Diagnosed hypertension, dyslipidemia and diabetes were defined as a self-reported physician’s diagnosis or current use of the respective treatment medication(s). Body mass index (BMI) was calculated from measured weight and height. Office BP measurements were obtained at local medical centers using validated cuff oscillometric devices.

Determinant of blood pressure by pulse transit time

Several previous studies have reported validation of the determination of BP by PTT [11]. In brief, PTT is defined as the travel time between the R-wave of the ECG and the pulse wave at the site of the finger in plethysmography (Supplementary Fig. 1). Arrival refers to the steepest part of the leading edge of the pulse wave. Pulse wave velocity was determined by PTT, height, and body composition factors [10, 18]. Systolic BP (SBP) and diastolic BP (DBP) values determined by PTT were calculated automatically using DOMINO Light software version 1.5.0 based on a patented algorithm (11/364174 US 2006/0217616 A1, 7374542). Detailed information for the calculation of beat-to-beat PTT-estimated BP is shown in the Supplementary Material. Resting BP, which was measured just before placing SOMNO touch RESP on patients in the supine position, was used for calibration by a manual, cuff-based method. All BP readings estimated by PTT from the start to the end of the SOMNO touch RESP recording were used to calculate average systolic and diastolic eBPBTB values. eBPINT was defined every 30 min during the recording, and average systolic and diastolic eBPINT values were calculated from those values. The maximum values of eBPBTB and eBPINT were defined as the highest values among all BP values of eBPBTB and eBPINT, respectively. The minimum values of eBPBTB and eBPINT were defined as the lowest values among all BP values of eBPBTB and eBPINT, respectively. For BP variability indices, we calculated 1) the standard deviation (SD) of the average eBPBTB and eBPINT and 2) the coefficient of variation (CV) of the average eBPBTB and eBPINT.

Statistical analysis

Data are expressed as means ±standard deviations or percentages. A two-tailed paired t test was used to compare the mean BP indices between eBPBTB and eBPINT. To assess the agreement between eBPBTB and eBPFIX, we performed Bland–Altman analysis. The relationships between eBPBTB and eBPFIX are shown in Bland–Altman plots and were examined by linear regression analysis. Moreover, linear regression analysis was used for the association between eBPBTB and eBPINT parameters, their differences and tertiles of ODI. A two-sided p value <0.05 was accepted as significant. All statistical analyses were performed with Stata ver. 15.0 software (StataCorp, College Station, TX, USA).

Results

Table 1 provides the demographic variables and clinical characteristics of the included patients. The average age was 66.8 ±11.9 years. The proportion of patients with heart failure was relatively high (45.8%). The average ODI was 21.0 ±15.0/h.

Table 1 Patient characteristics (n = 330)
Full size table

Table 2 shows the comparison of each BP index between eBPBTB and eBPINT. Except for diastolic eBPBTB and eBPINT, statistically significant differences in BP indices were found between eBPBTB and eBPINT values. The average systolic eBPBTB, maximum systolic and diastolic eBPBTB, Standard deviation (SD) of systolic and diastolic eBPBTB, and CV of systolic and diastolic eBPBTB were higher than those of eBPINT. Bland–Altman analysis demonstrated a closer agreement between eBPBTB and eBPINT for average SBP, SD of SBP and CV of SBP (i.e., BP variability measures), while there were significant disagreements between both minimum and maximum values of eBPBTB and eBPINT in patients with high SBP (Fig. 1). We performed a stratified analysis according to the presence and absence of atrial fibrillation. In Bland–Altman analysis, there was significant disagreement between the maximum value of systolic eBPBTB and eBPINT irrespective of the presence or absence of atrial fibrillation (Supplementary Figs. 2 and 3). In addition, compared to the difference in eBPBTB and eBPINT between the group with atrial fibrillation and that without, there were no significant differences in any of the BP indices (Supplementary Table 1). Supplementary Tables 2–6 show a comparison of BP indices in eBPBTB and eBPINT according to the use of each antihypertensive drug. The group using calcium channel blockers exhibited a greater difference between eBPBTB and eBPINT in maximum SBP and minimum DBP than the group without calcium channel blockers. In contrast, the group treated with ACE inhibitors, diuretics, or beta-blockers showed a smaller difference between eBPBTB and eBPINT in some BP indices.

Table 2 Comparison of pulse transit time (PTT)-estimated blood pressure parameters between beat-to-beat and intermittent measurement
Full size table
Fig. 1
figure 1

Bland–Altman plots comparing systolic blood pressure parameters between eBPBTB with eBPINT. eBPBTB indicates beat-to-beat PTT-estimated BP, eBPINT intermittent PTT-estimated BP at fixed time intervals

Full size image

Table 3 compares the systolic BP indices of eBPBTB and eBPINT according to the tertiles of 3% ODI. In both eBPBTB and eBPINT, the SD and CV of both systolic eBPBTB and eBPINT incrementally increased according to the tertiles of 3% ODI. However, there was no interaction between BP indices and tertiles of 3% ODI according to the difference in eBPBTB and eBPINT. These associations were observed in the diastolic BP index between eBPBTB and eBPINT according to the tertiles of 3% ODI (Supplementary Table 7).

Table 3 Comparison of pulse transit time (PTT)-estimated systolic blood pressure parameters between beat-to-beat and intermittent measurement according to tertiles of ODI
Full size table

Discussion

This study performed a comparison of PTT-estimated BP indices estimated by beat-to-beat and intermittent measurements in patients with suspected SDB. The results can be summarized as follows. First, although there was a statistically significant difference between average systolic eBPBTB and average systolic eBPINT, the absolute difference was minimal. Second, there were significant disagreements in both minimum and maximum systolic BP between the eBPBTB and eBPINT methods, while there was a close agreement in BP variability between the systolic eBPBTB and systolic eBPINT. Third, BP variability incrementally increased according to the tertiles of 3% ODI in both systolic eBPBTB and systolic eBPINT, and this tendency was essentially the same for systolic eBPBTB and systolic eBPINT.

To detect BP fluctuation over a certain period, continuous BP reading is ideal. However, percutaneous implantation of a monitor in the radial artery is the only established method, and among the numerous approaches that have been proposed as alternatives, ABPM is the closest to percutaneous implantation in the radial artery. BP readings measured by ABPM have been widely used to evaluate the effect of antihypertensive drugs and have been shown to be associated with target organ damage and cardiovascular events [19, 20]. Approximately 40 years ago, di Rienzo et al. compared the data of continuous, 24-h monitoring of each pressure wave and single pressure waves taken at regular intervals of 5, 10, 15, 30, and 60 min using invasive BP measurement [21]. Their results showed almost no difference in average SBP between the continuous measurement of each wave and the measurement of single waves at intervals if the interval was within 30 min. Thus, a previous study showed that the appropriate interval is available to obtain average BP readings comparable to continuous BP readings during a certain period assessed by ABPM.

However, this study did reveal the possibility that minimum and maximum BP values evaluated every 30 min are different from beat-to-beat values. Intermittent BP measurement is not able to detect extreme, anomalous BP changes. In clinical practice, detecting the maximum BP value during the nighttime may be important for patients with SDB. Previously, using a noninvasive oscillometric, desaturation-triggered BP measurement device, we found that cases with recurrent stroke and obstructive sleep apnea occasionally registered SBP readings above 200 mmHg even though the average nighttime SBP taken at intermittent intervals was 167 mmHg [22].

In this study, the group using calcium channel blockers exhibited a greater difference between eBPBTB and eBPINT in maximum SBP and minimum DBP than the group that did not use calcium channel blockers. BP variability has been strongly correlated with BP level [23]. In this study, the group that used calcium channel blockers exhibited higher eBPBTB and eBPINT in average SBP and DBP than those that did not use calcium channel blockers. Therefore, the group using calcium channel blockers may exhibit a larger difference between eBPBTB and eBPINT in some BP indices than the group that did not use calcium channel blockers. In contrast, the group treated with ACE inhibitors, diuretics, or beta-blockers showed a smaller difference between eBPBTB and eBPINT in some BP indices. This finding may have been attributable to the patient characteristics in this study, since ~50% of our enrolled patients had prevalent heart failure. These drugs are mainly used in patients with heart failure. Since heart failure involves low cardiac output and autonomic nervous dysfunction, heart failure patients demonstrate a lesser BP response to environmental stimulation.

In this study, BP variability assessed by systolic eBPBTB incrementally increased across the tertiles of ODI. A previous study reported that there is a linear association between systolic eBPBTB and ODI in 242 patients with suspected SDB [24]. Although the results of this study confirmed those of a previous study, there was no interaction of this association with BP variability assessed by systolic BPINT. There have been several previous studies about the association between BP variability assessed by fixed-interval BP measurement using ABPM and the severity of SDB [25, 26]. Steinhorst AP et al. showed that patients with an apnea hypopnea index (AHI) ≥10 had higher nighttime BP variability assessed by ABPM at 20-min intervals than those without, while this association was not found in daytime BP variability [25]. Therefore, fixed-interval BP measurement is useful for detecting nighttime BP variability due to SDB. Interestingly, a previous study reported that when the interval of BP measurement was longer, BP variability evaluated at fixed time intervals was higher than that of beat-to-beat BP measurement [21]. This potential error caused by sampling size may mask the difference between BP variability assessed by systolic eBPBTB and by systolic eBPINT. Regardless of the measurement method, BP variability is expected to increase according to the severity of SDB.

There are important limitations in this study. We acknowledge that there are currently insufficient data to conclude that the PTT-estimated BP value is comparable to the BP value taken by conventional BP measurement techniques, even though the PTT-estimated BP value has been validated with a mercury sphygmomanometer according to the international validation protocol [11]. PTT is calculated with a formula using PWV, which is influenced by many factors, such as age, sex, arterial elastic properties, and respiratory status. In addition, although there is little difference between BP assessed by conventional BP measurement and PTT-estimated BP at rest, there is a large difference of ~20 mmHg between them during exercise [10]. Therefore, in this study, we assessed PTT-estimated BP at night, when the effect of exercise was less pronounced. Furthermore, a previous study indicated that PTT-estimated BP can be used for the assessment of relative BP changes within the same individuals [27]. As a result, international BP guidelines have not yet accepted BP measurement using a wearable device, including the PTT methodology for diagnostic and treatment decisions [28]. The clinical importance of PTT-estimated BP and whether BP indices estimated by PTT are associated with the risk of target organ damage or cardiovascular events more than conventional BP measurements, such as in-office and out-of-office BP measurements, require further research.

In conclusion, this study showed that the PTT-estimated BP difference between beat-to-beat and intermittent measurements was minimal in patients with suspected SDB. In addition, there were disagreements in both minimum and maximum BP between PTT estimated by beat-to-beat monitoring and that estimated by intermittent measurement, while no disagreement was observed in the index of BP variability. The advantage of continuous BP monitoring might be the ability to identify maximum and minimal BPs, the clinical implications of which should be clarified in the future.