Pulse transit time-estimated blood pressure: a comparison of beat-to-beat and intermittent measurement

Blood pressure varies continuously, and accurate quantification of BP variability generally requires continuous monitoring. Pulse transit time (PTT)—the time for a pressure wave to travel from the heart to a peripheral site—enables cuffless continuous BP estimation and has been validated against conventional sphygmomanometry. In sleep-disordered breathing (SDB), transient BP surges occur at the termination of desaturation events, potentially requiring high-resolution measurement to capture extremes and variability. The research question was whether beat-to-beat PTT-estimated BP (eBPBTB) provides different and potentially more informative BP indices, including variability, than intermittent PTT-estimated BP sampled at fixed 30-minute intervals (eBPINT). The study aimed to compare BP parameters between eBPBTB and eBPINT in patients suspected of SDB across multiple centers.

Prior studies have validated PTT-based BP estimation against mercury sphygmomanometer readings and described its use for continuous, cuffless BP monitoring. Ambulatory BP monitoring (ABPM) is widely used and linked to organ damage and outcomes, but it may miss transient surges. Invasive studies have suggested that averages from intermittent sampling at intervals up to 30 minutes can approximate continuous averages, though extremes may be missed. In SDB, desaturation-linked surges and fluctuations in BP are well documented, motivating assessment of how sampling frequency impacts measured minima, maxima, and variability.

Design and population: Multicenter observational analysis of 330 patients suspected of SDB from 8 institutes (May 2016–August 2019). Mean age 66.8 ± 11.9 years; notable comorbidities included hypertension (65.8%), diabetes (33.3%), dyslipidemia (63.6%), atrial fibrillation (35.2%), coronary artery disease (26.4%), stroke (9.1%), and heart failure (45.8). Office BP averaged 128.0 ± 21.6/74.4 ± 16.2 mmHg. The study was IRB-approved with a waiver of informed consent. Measurements: Overnight recordings with SOMNO touch RESP (Fukuda Denshi, Tokyo) included nasal airflow, snoring, thoracoabdominal effort, ECG, SpO2, PTT, R-R timing, finger plethysmography, and body position. ODI (3% desaturation index) was computed with DOMINO Light v1.5.0; 3% ODI defined as number of ≥3% desaturation events per hour. PTT-based BP estimation: PTT defined as time from the ECG R-wave to the steepest upstroke of the finger plethysmography pulse wave. BP was estimated using a patented algorithm (11/364174 US 2006/0217616 A1, 7374542) incorporating PTT, height, and body composition factors. Calibration used resting cuff BP measured supine immediately before device placement. BP indices: eBPBTB comprised all beat-to-beat PTT-derived BP readings across the recording. eBPINT comprised readings sampled every 30 minutes over the same recording. For each method, average SBP/DBP, maximum and minimum SBP/DBP were determined. Variability indices included SD and coefficient of variation (CV) of SBP and DBP. Statistical analysis: Data are mean ± SD or percentages. Paired two-tailed t-tests compared eBPBTB vs eBPINT indices. Bland–Altman plots assessed agreement and regression examined proportional bias. Linear regression tested associations of indices and differences with tertiles of ODI. Significance at two-sided p < 0.05. Analyses used Stata 15.0.

• Cohort: N=330; mean 3% ODI 21.0 ± 15.0/h.
• Comparison of eBPBTB vs eBPINT (Table 2; differences are BTB − INT):
  • Average SBP: 122.2 ± 20.0 vs 122.0 ± 20.1 mmHg; difference +0.21 (p<0.01). Average DBP identical (71.9 ± 13.7 vs 71.9 ± 13.7 mmHg; NS).
  • Maximum SBP: 151.1 ± 26.2 vs 132.1 ± 21.2 mmHg; difference +19.0 (p<0.001). Maximum DBP: 86.5 ± 14.0 vs 78.6 ± 13.8 mmHg; difference +7.8 (p<0.001).
  • Minimum SBP: 106.2 ± 19.1 vs 113.7 ± 19.7 mmHg; difference −7.42 (p<0.001). Minimum DBP: 58.8 ± 15.0 vs 65.2 ± 14.3 mmHg; difference −6.5 (p<0.001).
  • Variability: SD of SBP 5.2 ± 2.1 vs 4.9 ± 2.1 mmHg (difference +0.2; NS in table), SD of DBP 3.7 ± 1.3 vs 3.6 ± 1.5 mmHg (difference +0.1; p<0.05). CV of SBP 4.3 ± 1.8% vs 4.1 ± 1.8% (difference +0.2; p<0.001). CV of DBP 5.4 ± 2.8% vs 5.3 ± 2.9% (difference +0.2; p<0.05).
• Bland–Altman: Close agreement for average SBP and variability measures (SD, CV of SBP); significant disagreement for minimum and maximum SBP/DBP, especially at higher SBP levels.
• Atrial fibrillation subgroup: Disagreement for maximum SBP persisted regardless of AF status; no significant group differences in BTB−INT differences across indices.
• Antihypertensive medication subgroups: Calcium channel blocker users showed larger BTB−INT differences in maximum SBP and minimum DBP; ACE inhibitor, diuretic, or beta-blocker users showed smaller differences in some indices.
• SDB severity (tertiles of 3% ODI): SD and CV of SBP (and similarly DBP) increased across ODI tertiles in both eBPBTB and eBPINT. No interaction between method and ODI tertile; method-related differences were consistent across SDB severity.

The study addressed whether beat-to-beat PTT-estimated BP provides different information compared to 30-minute intermittent PTT sampling in suspected SDB. Findings indicate that while mean nighttime SBP and overall variability indices (SD and CV) are closely comparable between eBPBTB and eBPINT, intermittent sampling misses the extremes: eBPINT substantially underestimates maximum SBP/DBP and overestimates minimum SBP/DBP relative to beat-to-beat measures. This is clinically relevant in SDB, where brief desaturation-linked surges and dips occur; capturing such extremes may affect risk stratification and management. The close agreement in averages and variability aligns with prior invasive evidence that intermittent sampling at ≤30-minute intervals approximates continuous averages, but the present results clarify that extremes are not reliably captured intermittently. The consistency of findings across AF status suggests robustness to rhythm irregularity. Medication subgroup patterns (e.g., larger discrepancies with calcium channel blocker use) likely reflect higher BP levels and variability in those subgroups. The progressive increase in BP variability with higher ODI tertiles, observed similarly with both measurement approaches, underscores the link between SDB severity and BP lability and suggests that either approach can characterize variability trends, though eBPBTB better captures extreme values.

In patients suspected of sleep-disordered breathing, beat-to-beat and 30-minute intermittent PTT-estimated BP provide similar mean BP and variability metrics, but intermittent sampling fails to capture true nocturnal extremes, underestimating peaks and overestimating nadirs. For clinical or research contexts where extremes matter (e.g., nocturnal surges in SDB), beat-to-beat monitoring is preferable, whereas intermittent sampling may suffice for averages and variability indices. Future work should validate these findings against invasive or gold-standard continuous BP, evaluate outcome relevance of missed extremes, optimize intermittent sampling strategies (e.g., event-triggered sampling), and assess generalizability across populations and device algorithms.