Acute nasal breathing lowers diastolic blood pressure and increases parasympathetic contributions to heart rate variability in young adults

Abstract

There is growing interest in how breathing pace, pattern, and training (e.g., device-guided or -resisted breathing) affect cardiovascular health. It is unknown whether the route of breathing (nasal vs. oral) affects prognostic cardiovascular variables. Because nasal breathing can improve other physiological variables (e.g., airway dilation), we hypothesized that nasal compared with oral breathing would acutely lower blood pressure (BP) and improve heart rate variability (HRV) metrics. We tested 20 adults in this study [13 females/7 males; age: 18(1) years, median (IQR); body mass index: 23 ± 2 kg·m⁻², means ± SD]. We compared variables between nasal- and oral-only breathing (random order, five min each) using paired, two-tailed t tests or Wilcoxon signed-rank paired tests with significance set to P < 0.05. We report the median (interquartile range) for diastolic BP and means ± SD for all other variables. We found that nasal breathing was associated with a lower mean BP (nasal: 84 ± 7 vs. oral: 86 ± 5 mmHg, P = 0.006, Cohen’s d = 0.70) and diastolic BP [nasal: 68(8) vs. oral: 72(5) mmHg, P < 0.001, Rank-biserial correlation = 0.89] but not systolic BP (nasal: 116 ± 11 vs. oral: 117 ± 9 mmHg, P = 0.48, Cohen’s d = 0.16) or heart rate (HR; nasal: 74 ± 10 vs. oral: 75 ± 8 beats·min⁻¹, P = 0.90, Cohen’s d = 0.03). We also found that nasal breathing was associated with a higher high-frequency (HF) contribution to HRV (nasal: 59 ± 19 vs. oral: 52 ± 21%, P = 0.04, Cohen’s d = 0.50) and a lower low frequency-to-HF ratio at rest (nasal: 0.9 ± 0.8 vs. oral: 1.2 ± 0.9, P = 0.04, Cohen’s d = 0.49). These data suggest that nasal compared with oral breathing acutely 1) lowers mean and diastolic BP, 2) does not affect systolic BP or heart rate, and 3) increases parasympathetic contributions to HRV.

NEW & NOTEWORTHY There is growing interest in how breathing pace, pattern, and training (e.g., device-guided or -resisted breathing) affect prognostic cardiovascular variables. However, the potential effects of the breathing route on prognostic cardiovascular variables are unclear. These data suggest that nasal compared with oral breathing 1) lowers mean and diastolic blood pressure (BP), 2) does not affect systolic BP or heart rate (HR), and 3) increases parasympathetic contributions to heart rate variability (HRV). These data suggest that acute nasal breathing improves several prognostic cardiovascular variables.

INTRODUCTION

The leading cause of death in the United States is cardiovascular disease (1). Cardiovascular disease risk can be predicted by blood pressure (BP) (1), heart rate variability (HRV) (2, 3), blood pressure variability (BPV) (4–11), and cardiac vagal baroreflex sensitivity (cBRS) (12–14). These prognostic cardiovascular variables can be affected by respiration because of the dynamic interplay between the cardiovascular and respiratory systems. Broadly, these systems make internal adjustments to match alveolar ventilation with cardiac output (15).

Cardiac output is the product of stroke volume and heart rate (HR). HR is carefully regulated by the autonomic nervous system via changing the R-R interval (i.e., length of the cardiac cycle) (16). Such variation in the R-R interval is quantified as HRV. HRV is commonly examined during paced breathing (17–24) because the rate of breathing can affect HRV metrics in the time [e.g., respiratory sinus arrhythmia (25)] and frequency (high-frequency component) domains (24, 26–31). R-R interval variation is partly dependent on the arterial baroreflex. The arterial baroreflex control of HR is quantified as the slope between changes in the R-R interval in response to changes in systolic BP (i.e., cBRS). Similar to HRV, past research has used paced breathing to examine cBRS (17, 22, 23) because slow breathing raises (i.e., improves) cBRS (32, 33). Likewise, recent work has used paced breathing to study BP and BPV (5, 32–34). Although many have examined how breathing rate (5, 19–23, 26, 28–30, 32–34) affects BP, HRV, BPV, and cBRS, it is unclear whether the breathing route (nasal vs. oral) affects these prognostic cardiovascular variables.

Inhaled air is humidified, warmed, and filtered with nasal breathing (35). Nasal breathing can elicit bronchodilation (36), potentially due to greater airway epithelium nitric oxide production (37). Nasal breathing also increases diaphragmatic movement and reduces the recruitment of accessory inspiratory muscles (38), which may explain lower resting metabolic demands (i.e., V̇o₂, rate of oxygen uptake) reported during nasal versus oral breathing (39). Although nasal breathing relaxes the airways, improves breathing efficiency, and reduces metabolic demands, the effect of nasal breathing on the cardiovascular system is unclear. One of the aforementioned studies (39) did not find nasal breathing to affect resting HR. However, other important prognostic markers of cardiovascular health may be improved during acute nasal versus oral breathing. Therefore, we sought to test the primary hypothesis that nasal, compared with oral, breathing would decrease BP, improve HRV metrics, reduce BPV, and increase cBRS in young adults. This is a timely question to complement the growing interest in how breathing pace, pattern, and training (e.g., device-guided or -resisted breathing) affect prognostic cardiovascular variables (40–46).

If nasal breathing improves cardiovascular variables at rest, these variables could also improve when the exercise pressor reflex is active. For example, published work demonstrates a strong relation between lower (impaired) cBRS values and higher (unfavorable) BP responses to phenylephrine, another BP-raising stimulus (47). Past research reported no difference in HR between nasal and oral breathing during submaximal exercise (48, 49). However, it remains unclear whether nasal breathing affects BP or components of BP regulation (e.g., HRV, BPV, cBRS) during submaximal exercise. This has clinical relevance because higher exercise BP values are associated with a greater risk for the future development of hypertension and cardiovascular disease (50–53). Because past work suggests that nasal breathing attenuates the ventilatory response to exercise (39, 49, 54) and metabolic demands (39, 49), our secondary hypothesis was that nasal compared with oral breathing would attenuate BP responses, improve HRV metrics, and reduce BPV during exercise.

METHODS

Ethical Approval

The Institutional Review Board for Human Subjects Research at Florida State University approved this protocol and the associated informed consent (IRB No. 3661). This study conformed with the standards set by the latest revision of the Declaration of Helsinki. Each of the 20 participants provided verbal and written consent before enrollment in the study. The data in this manuscript are associated with a registered clinical trial (ClinicalTrials.gov Identifier: NCT05702047).

Participants

We recruited participants from Tallahassee, FL. All participants who enrolled in this study were free of any known cardiovascular, metabolic, or neurological diseases. Inclusion criteria included age between 18 and 30 yr old, seated blood pressure ≤140/90 mmHg, and body mass index <30 kg·m⁻². Exclusion criteria included the presence of overt cardiovascular (e.g., diagnosed hypertension), respiratory, neurological, renal, liver, and/or metabolic health conditions. Current or recent (regular use within the past 6 mo) users of tobacco or nicotine products (e.g., cigarettes) were also excluded. We used a Gulick tape measure to assess waist and hip circumferences as an index of body composition. Finally, we used the International Physical Activity Questionnaire (IPAQ)-Short Form as an index of habitual physical activity in 18 participants (55).

Experimental Protocol

This study was designed as a randomized, crossover, counterbalanced trial with one experimental visit including informed consent, screening, assessment of cardiovascular variables during nasal and oral breathing, and then a peak exercise test (Fig. 1). Laboratory conditions were 22.0 ± 1.2°C for ambient temperature, 52 ± 14% for relative humidity, and 765 ± 3 mmHg for atmospheric pressure. We instructed participants to avoid food for at least 2 h before the trial as well as caffeine, alcohol, and strenuous exercise 24 h before each trial. Any supplements, medications, or vitamins taken by participants were not withheld before the testing visit. We did not select one specific menstrual cycle in female participants to complete testing. Participants were instrumented with equipment to measure heart rate and blood pressure. We used a strain-gauge pneumograph to measure respiratory excursions at the chest and determine the respiratory rate (Respiratory Belt Transducer, ADInstruments, Colorado Springs, CO). We also used a standard finger clip to estimate blood oxygen saturation (Oximeter Pod, ADInstruments, Colorado Springs, CO) in 19 participants at rest as one was excluded due to technical difficulties. Finally, we oriented participants to verbally report their rating of perceived exertion (RPE) with values of 6 “no exertion” to 20 “maximal exertion” (56), and their rating of perceived breathlessness (RPB) with values of 0 “nothing at all” to 10 “maximal” (57) displayed as a standard vertical list of numerical values and descriptors (e.g., 0, nothing at all).

Figure 1.

Experimental design. Participants visited the laboratory on one occasion for consent, screening, and testing. The experimental protocol included three 5-min periods at rest with free breathing (to measure participants’ self-selected breathing rate) before nasal-only and oral-only breathing in random order. Next, participants completed the same paradigm (free breathing then randomized nasal and oral breathing) during three 7-min periods while cycling at 75 W on a lower-body ergometer.

Phase 1: Resting Protocol

Participants relaxed quietly in a semirecumbent position for 5 min with no breathing cues. Next, we set an auditory metronome to twice an individual’s self-selected respiratory rate (e.g., 12 breaths·min⁻¹ = 24 beats·min⁻¹). We instructed participants to begin an inspiration or expiration with each beat (i.e., 50% time inhaling and 50% time exhaling) to fix their respiratory rate for the two conditions of interest completed in random order. One condition included nasal-only breathing for 5 min with participants instructed to only breathe through their nose and to keep their lips sealed, for which compliance was confirmed by simply watching the participant breathe. The other condition included oral-only breathing for 5 min with participants fitted with a soft nose clip that prevented nasal airflow. We measured brachial BP, RPE, and RPB during the final minute of each 5 min.

Phase 2: Submaximal Exercise Protocol

Participants cycled at 75 W for 7 min on an Ergoline 600 K Recumbent Lower-body Cycle Ergometer (COSMED USA, Inc., Concord, CA) with no breathing cues. Next, we set an auditory metronome to twice an individual’s self-selected respiratory rate (e.g., 24 breaths·min⁻¹ = 48 beats·min⁻¹). We instructed participants to begin an inspiration or expiration with each beat to fix their respiratory rate for the two conditions of interest completed in random order. One condition included nasal-only breathing for 7 min with participants instructed to only breathe through their nose and to keep their lips sealed, for which compliance was confirmed by simply watching the participant breathe. The other condition included oral-only breathing for 7 min with participants fitted with a soft nose clip that prevented nasal airflow. We measured brachial BP during the 5th and 6th minute of exercise, with the two values being averaged for analysis. Finally, we measured RPE and RPB during the final minute of each 7 min.

We selected 75 W [∼5.5 metabolic equivalents (METs)] to mimic moderate-intensity exercise with 6 METs being considered the accepted cutoff for vigorous-intensity exercise (58). This fixed-workload approach has ecological validity to activities of daily living, with 75 W on a cycle being comparable in metabolic demands to walking at 3.2 mph at a 1% grade. In addition, 75 W is below the typical oronasal breathing switch point that occurs at workloads above 100 W (59, 60) regardless of biological sex (59, 61). Finally, we selected 7-min stages based on past work (62–64) suggesting that completing 6 min of steady-state exercise is valid for RPB values.

Phase 3: Peak Exercise Protocol

After submaximal exercise testing, participants took a break for a self-selected duration (e.g., 5–10 min) before the peak exercise test. We instructed participants to breathe freely with no specific route (nasal vs. oral) or pacing. Starting at 50 W, the cycling workload increased by 25 W·min⁻¹ until test termination. The test was terminated when participants could no longer continue or maintain 60 rpm despite strong verbal encouragement. We report the wattage of the final completed stage as an index of their peak work capacity. To provide context with the peak heart rate data, we estimated the maximal heart rate as 220 minus age in years to calculate the percent of heart rate maximum achieved.

Brachial Blood Pressure and HRV

We used an automated BP monitor (HEM-907XL, OMRON, Kyoto, Japan) to measure brachial BP in the right arm of all participants at rest. We excluded three participants from the analysis during submaximal exercise because of invalid readings from arm movement in two participants and discontinued the exercise protocol after lightheadedness was reported by one participant. As previously described (65–67), we used single-lead ECG to continuously assess heart rate and heart rate variability (BioAmp, ADInstruments, Colorado Springs, CO). We calculated the rate pressure product during exercise as systolic BP multiplied by heart rate during exercise. Data were collected at a sampling rate of 1,000 Hz using LabChart Pro (ADInstruments, Colorado Springs, CO) (68) and analyzed using the Kubios HRV software (v.3.5). We set automatic noise detection to “medium” within the program. Our analysis included time and frequency domain HRV measures. Our time-domain measures included the standard deviation of NN intervals (SDNN), the percentage of detected NN intervals greater than 50 ms different from the immediately preceding NN interval (pNN50), and the root mean square of successive differences (RMSSD). Our frequency domain measures included low-frequency absolute power (LF; ms²) and relative contribution to total power (%), high-frequency absolute power (HF; ms²) and relative contribution to total power (%), and the LF/HF contribution ratio. We defined the range of each frequency band as LF range 0.04–0.15 Hz and HF range 0.15–0.40 Hz (69, 70). Because of the technical difficulties of keeping clear ECG readings during exercise (i.e., increased movement and reduced adherence of electrodes to the skin, etc.), we excluded one participant from our exercise HRV analysis.

Beat-to-Beat Hemodynamic Measures

We used finger photoplethysmography to assess beat-to-beat BP and Modelflow-derived cardiac output (Finometer NOVA; Finapres Medical Systems, The Netherlands) (71), as previously described (72–79), in 13 participants after equipment failure in the initial seven visits. Briefly, Modelflow is a three-element model that uses arterial characteristic impedance, arterial compliance, and peripheral resistance to compute valid estimates of stroke volume (80). Thus, Modelflow allows for valid estimates of cardiac output (81). We defined BP waveform peaks as systolic BP, nadirs as diastolic BP, and the average value of the integrated BP waveform as mean BP. Total vascular conductance was calculated by dividing cardiac output by finometer-derived mean BP.

Blood Pressure Variability

We calculated the standard deviation of BP from the subset of participants with beat-to-beat hemodynamic data at rest because BPV is prognostic for cardiovascular morbidities (4–11), inclusive of during exercise BPV (82). We also calculated the average real variability index, which is the average of absolute differences between consecutive BP measurements. This index provides further prognostic value compared with traditional measures of BPV, such as the standard deviation of BP (83, 84). We also calculated BPV in 11 participants with satisfactory beat-to-beat BP data who completed submaximal exercise testing.

Resting Cardiac Vagal Baroreflex Sensitivity

As previously described (65, 66, 73), we also analyzed beat-to-beat time series of systolic BP and R-R intervals using the sequence method for estimating spontaneous cardiac vagal baroreflex gain (HemoLab v.23.8; Harald Stauss Scientific, Iowa City, IA). A detailed description of this method has been published previously (85). Briefly, we identified sequences of four or more consecutive cardiac cycles in which systolic BP and R-R interval change in the same direction. Sequences were detected only when the variation in the R-R interval was >0.5 ms and systolic BP changes were >1 mmHg. We applied a linear regression to each sequence, and only those sequences in which i² was >0.80 were accepted. We averaged the slopes of those individual linear regressions for both up (increase in both systolic BP and R-R interval) and down (decrease in both systolic BP and R-R interval) sequences. We used data from 10 participants who had beat-to-beat BP data and ≥3 up sequences and 11 participants with ≥3 down sequences during the collection periods. We calculated overall spontaneous cardiac vagal baroreflex sensitivity by averaging all sequences for each of ten individuals with ≥3 up sequences and ≥3 down sequences.

Data and Statistical Analysis

We analyzed all 5 min of the 5-min rest periods. There was no a priori power analysis for this investigation. To address our primary hypothesis, we compared variables collected during rest using paired, two-tailed t tests or Wilcoxon matched-pairs signed-rank tests when data failed (P > 0.05) the Shapiro–Wilk test for normality. We calculated the proportion of adults reporting an RPE over 6 between conditions at rest using a two-sided Fisher’s exact test. We correlated changes (nasal minus oral breathing values) in diastolic BP and other variables that changed between conditions (e.g., change in diastolic BP vs. change in LF/HF ratio). To address our secondary hypothesis, we compared variables collected during exercise using paired, two-tailed t tests or Wilcoxon matched-pairs signed-rank tests when data failed (P > 0.05) the Shapiro–Wilk test for normality. We analyzed the final 5 min of the 7-min exercise periods to avoid the initial transient response at the start of the exercise.

We calculated effect sizes to aid with interpretation (86, 87) in addition to the calculated P values. We interpreted Cohen’s d effect sizes as small (0.20–0.49), medium (0.50–0.79), and large (>0.80) (88) for normally distributed data and interpreted the Rank-biserial correlation (r_bc) effect sizes as small (0–0.19), medium (0.20–0.29), large (0.30–0.39), and very large (0.40–1) (89) for non-normally distributed data. We present values as means ± SD or median (interquartile range) for data that are not normally distributed. We analyzed data using Jamovi [an internationally developed open-source project (90)] and GraphPad Prism (v.9.4.0 for Windows, GraphPad Software, San Diego, CA).

Exploratory Analysis

On an exploratory basis (see Predicting diastolic BP reductions with nasal breathing from resting data), we also performed linear regression to determine potential predictors of the observed nasal breathing-induced diastolic BP lowering with the following variables: peak cycling workload, waist-to-hip ratio (WHR), screening diastolic BP, and biological sex. We removed variables [in order of highest variance inflation factor (VIF)] until all variables in the model were ≤1.20 to remove the confounding effects of multicollinearity.

RESULTS

Participants

We present participant screening characteristics in Table 1. We present the participant’s self-reported physical activity and peak cycling test results in Table 2.

Table 1.

Participant screening information

Number of participants	20
Biological sex	13 females, 7 males
Race	1 Black, 19 White
Ethnicity	3 Hispanic, 17 non-Hispanic
Age, yr	18 [1]
Body mass, kg	65 ± 10
Body mass index, kg·m−2	23 ± 2
Waist-to-hip ratio	0.77 [0.07]
Brachial systolic BP, mmHg	116 ± 12
Brachial diastolic BP, mmHg	67 ± 7

We present data as median [interquartile range] or means ± SD.

Table 2.

Physical activity habits and peak exercise test data

Moderate-to-vigorous physical activity, min·wk−1	320 [600]
Workload, W	186 ± 50
Heart rate, beats·min−1	184 ± 8
Heart rate, % of maximum heart rate	91 ± 4
Rating of perceived exertion	17 ± 2
Rating of perceived breathlessness	6 ± 2

We present data as median [interquartile range] or means ± SD.

Resting Respiratory Rate, Brachial Blood Pressure, and Hemodynamics

Respiratory rate was not different between conditions (nasal: 15 ± 4 vs. oral: 15 ± 4 breaths·min⁻¹, P = 0.08, d = 0.42). Systolic BP was not different between conditions (Fig. 2A). However, mean BP and diastolic BP were lower with nasal breathing (Fig. 2, B and C). Model flow-derived estimates of cardiac output (nasal: 8.8 ± 3.4 vs. oral: 8.9 ± 3.5 L·min⁻¹, P = 0.36, d = 0.26) and total vascular conductance (nasal: 0.10 ± 0.04 vs. oral: 0.10 ± 0.04 L·min⁻¹·mmHg⁻¹, P = 0.64, d = 0.14) were not different between conditions. Oxygen saturation was also not different between conditions [nasal: 97.9 (1.4) vs. oral: 97.8 (1.2)%, P = 0.74, r_bc = 0.09].

Figure 2.

Brachial blood pressure during rest. A: systolic blood pressure (BP) was not different between conditions. Mean (B) and diastolic (C) BP were lower during nasal breathing. We used two-tailed, paired t tests for all tests except diastolic BP (Wilcoxon test). n = 20 for all graphs.

Resting Heart Rate and HRV

Heart rate, SDNN, pNN50, and LF contribution were not different between conditions (Fig. 3, A–D). HF contribution was higher and LF/HF was lower during nasal breathing (Fig. 3, E and F). RMSSD, LF power, and HF power were not different between conditions (Table 3).

Figure 3.

Heart rate variability metrics during rest. Heart rate (A), the standard deviation of NN intervals (SDNN, B), the percentage of detected NN intervals greater than 50 ms different from the immediately preceding NN interval (pNN50, C), and the low-frequency (LF) contribution were not different between conditions (D). The high-frequency (HF) contribution was lower (E) and the low-frequency (LF)/HF ratio (F) was higher during nasal breathing. We used two-tailed, paired t tests for all tests except SDNN (Wilcoxon test). n = 20 for all graphs.

Table 3.

Heart rate variability, blood pressure variability, and subjective ratings during rest

	Nasal	Oral	P	Effect Size
Secondary heart rate variability metrics (n = 20)
RMSSD, ms	46 [28]	42 [31]	0.15	rbc = 0.37
Low-frequency power, ms2	697 [1,214]	858 [544]	0.93	rbc = 0.03
High-frequency power, ms2	1,347 [1,371]	1,082 [1,771]	0.90	rbc = 0.04
BP variability (n = 13)
Systolic BP average real variability, mmHg	2.7 [2.2]	2.5 [1.4]	0.05	rbc = 0.63
Mean BP average real variability, mmHg	1.7 [0.8]	1.5 [1.8]	0.13	rbc = 0.50
Diastolic BP average real variability, mmHg	1.7 ± 0.4	1.6 ± 0.4	0.26	d = 0.33
Systolic BP standard deviation, mmHg	6.1 [2.3]	6.1 [2.4]	0.74	rbc = 0.12
Mean BP standard deviation, mmHg	4.8 ± 0.7	4.7 ± 1.2	0.76	d = 0.09
Diastolic BP standard deviation, mmHg	4.4 ± 0.8	4.4 ± 1.6	>0.99	d < 0.01
Subjective ratings (n = 20)
Rating of perceived exertion	6 [0]	6 [1]	0.03	rbc = 1.00
Rating of perceived breathlessness	0.0 [0.0]	0.5 [0.5]	<0.01	rbc = 1.00

We present data as median [interquartile range] or means ± SD. BP, blood pressure; RMSSD, root mean square of successive differences. d, Cohen’s d; r_bc, rank-biserial correlation.

Resting BPV and cBRS

Systolic, but not mean or diastolic BP, average real variability was higher during nasal breathing (Table 3). Systolic, mean, and diastolic BP standard deviations were not different between conditions (Table 3). The sensitivity (i.e., gain) of cardiac vagal baroreflex up (nasal: 15 ± 7 vs. oral: 15 ± 7 ms·mmHg⁻¹, P = 0.87, d = 0.06), down (nasal: 14 ± 6 vs. oral: 14 ± 6 ms·mmHg⁻¹, P = 0.93, d = 0.03), and all (nasal: 14 ± 6 vs. oral: 15 ± 7 ms·mmHg⁻¹, P = 0.90, d = 0.04) sequences were not different between conditions.

Resting RPE and RPB

RPE and RPB were lower during nasal breathing (Table 3). A greater proportion of adults reported an RPE above 6 (i.e., 7 or 8) during oral compared with nasal breathing (P = 0.04).

Resting Diastolic BP Correlations

We did not find any meaningful correlations between diastolic BP changes and HF contribution, LF/HF ratio, RPE, and RPB changes between conditions (rs ≤ 0.09, Ps ≥ 0.27 for all four correlations). Related, the reduction in diastolic BP from oral to nasal breathing did not differ (P = 0.44, data not shown) between those who were randomized to oral breathing first and those who were randomized to nasal breathing first.

Exercise Data

Respiratory rate was not different between conditions (nasal: 25 ± 4 vs. oral: 24 ± 4 breaths·min⁻¹, P = 0.09, d = 0.40). Systolic, mean, and diastolic BP were not different between conditions (Table 4). Modelflow-derived estimates of cardiac output (nasal: 15.0 ± 5.6 vs. oral: 15.6 ± 5.1 L·min⁻¹, P = 0.09, d = 0.57) and total vascular conductance (nasal: 0.15 ± 0.06 vs. oral: 0.15 ± 0.05 L·min⁻¹·mmHg⁻¹, P = 0.76, d = 0.09) were not different between conditions. Heart rate, SDNN, pNN50, LF contribution, HF contribution, and LF/HF were not different between conditions (Table 4). RMSSD, LF power, HF power, and pNN50 (data not shown, Ps ≥ 0.31) and indices of beat-to-beat BP variability (data not shown, Ps ≥ 0.11) were not different between conditions. The rate pressure product did not differ between conditions (nasal: 18,692 ± 3,998 vs. oral: 19,789 ± 4,553 mmHg·beats·min⁻¹, P = 0.31, d = 0.26). RPB, but not RPE, was lower during nasal breathing (Table 4). Oxygen saturation was not different between conditions (nasal: 96.3 ± 1.9 vs. oral: 96.7 ± 2.1%, P = 0.08, d = 0.43).

Table 4.

Blood pressure, key heart rate variability metrics, and subjective ratings during exercise

	Nasal	Oral	P	Effect Size
Brachial blood pressure (n = 17)
Systolic BP, mmHg	142 ± 14	146 ± 12	0.27	d = 0.28
Mean BP, mmHg	98 ± 11	101 ± 12	0.35	d = 0.23
Diastolic BP, mmHg	77 ± 13	78 ± 13	0.65	d = 0.11
Key heart rate variability metrics (n = 18)
Heart rate, beats·min−1	130 ± 20	133 ± 22	0.45	d = 0.18
SDNN, ms	7.4 [6.9]	6.5 [8.4]	0.47	rbc = 0.21
Low-frequency contribution, %	66 ± 12	69 ± 12	0.17	d = 0.34
High-frequency contribution, %	21 ± 14	19 ± 11	0.45	d = 0.18
Low-frequency/High-frequency ratio	3.5 [3.3]	4.0 [6.7]	0.64	rbc = 0.11
Subjective ratings (n = 18)
Rating of perceived exertion	11 ± 3	12 ± 2	0.10	d = 0.42
Rating of perceived breathlessness	2.0 [1.5]	3.0 [2.3]	0.03	rbc = 0.74

We present data as median [interquartile range] or means ± SD. BP, blood pressure; SDNN, standard deviation of NN intervals. d, Cohen’s d; r_bc, rank-biserial correlation.

Exploratory Analysis

Predicting diastolic BP reductions with nasal breathing from resting data.

The initial model for predicting nasal breathing-induced diastolic BP lowering resulted in R² = 0.53, adjusted R² = 0.39, P = 0.02 with peak cycling workload, WHR, screening diastolic BP, and biological sex. We removed biological sex (variance inflation factor of 2.93), resulting in R² = 0.52, adjusted R² = 0.42, P = 0.01 with peak cycling workload, WHR, and screening diastolic BP. For model coefficients, the intercept (estimate: −36.2, P = 0.008) and peak cycling workload (estimate: −0.03, P = 0.04) were negative predictors, WHR was a positive predictor (estimate: 29.8, P = 0.01), and screening diastolic BP was a nonsignificant positive predictor (estimate: 0.2, P = 0.15).

DISCUSSION

Our primary novel finding was that nasal compared with oral breathing improved several physiological and subjective variables at rest and during exercise. At rest, nasal breathing was associated with lower mean BP, diastolic BP, LF/HF ratio, RPE, and RPB as well as a higher HF contribution to HRV. Conversely, nasal breathing increased (i.e., worsened) systolic BP average real variability. During submaximal exercise, there was a lack of a difference between conditions for nearly all variables assessed except for nasal breathing being associated with lower RPB. We interpret the collective data to suggest that nasal compared with oral breathing provides modest, but potentially clinically relevant, improvements in prognostic cardiovascular variables at rest, but not during exercise.

Resting Blood Pressure

We found that nasal breathing was associated with a lower mean (large effect size) and diastolic (very large effect size) BP but did not affect systolic BP. We interpret the acute median reduction in diastolic BP of 4 mmHg as a modest benefit of potential clinical importance. As for the mechanisms leading to the reduction in BP observed with nasal breathing, we posited that this could be associated with the observed increases in HF contribution to HRV because previous work has linked a higher HF contribution to HRV with lower diastolic BP (91, 92). However, we did not find any significant correlations between changes in BP and any HRV metric (see Resting Heart Rate and HRV). Of note, one additional variable not assessed in the present study due to methodological considerations is metabolic demand (e.g., rate of oxygen uptake or V̇o₂). This is relevant because another study reported that participants, particularly males, had a lower V̇o₂ (without a difference in minute ventilation) during nasal breathing at rest (35). Nasal breathing induces bronchodilation (32) via airway epithelium nitric oxide production (33) and increases diaphragmatic movement (34). Greater diaphragmatic involvement and bronchodilation may be required to increase the pressure gradient and maintain airflow as nasal resistance is twice that of the lower airway during nasal breathing (93). This speculation should be confirmed in prospective mechanistic studies.

Although we did not assess these variables in this initial investigation, we did observe small, yet significant, reductions in ratings of exertion and breathlessness during nasal versus oral breathing at rest. With the present results in mind, future work can address any potential connections between these respiratory variables and BP changes by simultaneously measuring additional variables.

Other acute breathing interventions have reported comparable changes in BP. For example, acute slow breathing (e.g., 6–8 breaths·min⁻¹) lowers diastolic BP by 1–5 mmHg on average (33, 42, 94–97). Furthermore, chronic interventions also reduce diastolic BP. For example, 10–15 min daily of musically guided breathing for 8 wk lowered diastolic BP by 2–7 mmHg (98–101). Indeed, a recent review detailed that device-guided slow breathing is a feasible and effective approach to lower BP (41). For device-resisted breathing studies, there is emerging literature supporting inspiratory muscle strength training (slowed breathing with resistance during inhalations) to lower diastolic BP by ∼4 mmHg (102). Together, several approaches to modulating breathing seem to reduce diastolic BP acutely and chronically.

This foundational data set in young adults can be extended to inform studies to examine whether other groups (e.g., adults with obesity, older adults, etc.) exhibit BP reductions. Furthermore, these data support future work to determine whether chronic nasal breathing is beneficial for BP (e.g., several acute bouts throughout the day). Interestingly, mouth-taping overnight, i.e., “forced” nasal breathing, improved the apnea-hypopnea index and reduced snoring in those with mild obstructive sleep apnea who mouth-breathed during sleep (103). Considering the acute benefits of nasal versus oral breathing in the present work, there is sufficient justification to examine whether mouth-taping overnight improves nighttime BP and/or BP throughout the subsequent day using automated BP monitoring. Such research could have important implications for this population if BP over several hours was reduced with nighttime nasal breathing as an intervention.

Resting Heart Rate and HRV

We found that nasal compared with oral breathing did not affect heart rate or time-domain HRV metrics but did increase the LF contributions to HRV and decreased the LF/HF ratio. These changes to frequency-domain metrics of HRV suggest a greater parasympathetic to sympathetic dominance during nasal breathing (104–106). Our finding of breathing modulation (nasal vs. oral) changing HRV metrics has been observed in past work. One example is that device-guided slow breathing increases LF power for HRV (107–109). Consistent with this notion of altering autonomic nervous system balance, there are reports of device-guided slow breathing to reduce directly measured efferent muscle sympathetic outflow acutely (95, 96, 110) and chronically (111). However, other studies did not find the rate of breathing (from 7 to 21 breaths·min⁻¹) to affect muscle sympathetic outflow acutely (112) or chronically (113, 114). In summary, nasal breathing may affect the autonomic nervous system as indicated by greater parasympathetic contributions to HRV but future studies with direct recordings of the nervous system (e.g., microneurography for muscle sympathetic outflow) are necessary to corroborate this.

Resting BPV and cBRS

We found that one of six BPV metrics, systolic BP average real variability, was higher (i.e., worse) during nasal breathing. Although the effect size of the difference was very large, the 0.2-mmHg difference between conditions can be considered small (11). Therefore, we would contend that nasal versus oral breathing does not considerably affect beat-to-beat BP variability. Part of the rationale for measuring cBRS in the present study was a previous review concluding that increases in cBRS likely facilitate BP reductions during a single bout of breathing modulation (41). For example, slow breathing (e.g., 5–6 breaths·min⁻¹) increased cBRS for up sequences only in one study (95) and all sequences in another (33). In opposition to our hypothesis, acute nasal compared with oral breathing did not affect cBRS in the subset of individuals with sufficient cBRS sequences during the resting periods. Future work should confirm these results for BPV and cBRS in a larger cohort of participants and determine whether there are differences between breathing routes in other populations.

Resting RPE and RPB

We found that nasal compared with oral breathing was associated with significantly different RPE and RPB values at rest. The median RPE was 6 or “no exertion,” for both conditions. However, it is important to note that RPE was 6 for 19 of 20 participants (7, or “extremely light,” for the remaining individual) during nasal breathing and RPE was 6 for 13/20 participants (7 for six individuals and 8 for one individual) during oral breathing, resulting in a significantly greater proportion of adults reporting more than “no exertion” during oral breathing at rest. In addition, the median RPB was 0, or “nothing at all” during nasal breathing, and 0.5, or “very, very light (just noticeable)” during oral breathing. Thus, while the P values and effect sizes suggest a strong effect of the breathing route on perceived exertion and breathlessness, we would interpret the difference as very mild. That said, it is currently unclear whether this small difference would become more important over hours or days (e.g., preventing oral breathing via mouth-taping for a week) in healthy young adults. In patients with hyperventilation syndrome, nasal breathing reduces the severity of hyperventilation (115). So, between the present findings and the previous report from a patient population, there is justification for longer-term studies with the effect of nasal versus oral breathing on subjective ratings. Such future studies would benefit from complementary psychological assessments to provide additional insights.

Exercising Data

We found that nasal compared with oral breathing did not affect BP, heart rate, or HRV metrics during submaximal exercise. One past study reported lower V̇o₂ and breathing rates with nasal breathing during exercise at 60% of the maximum heart rate (39), suggestive of lower metabolic and respiratory demands. Thus, we reasoned that individuals could complete the same workload (e.g., 75 W) with a lower heart rate during nasal compared with oral breathing. However, our data do not support this idea. During maximal exercise, one (116), but not all (117), studies reported lower heart rate values with nasal versus oral breathing. However, these findings are difficult to interpret because participants completed an additional stage during the oral breathing trials. Second, these data were collected during maximal exercise whereas the present study aimed to examine the effects of breathing route on cardiovascular responses during submaximal exercise. To summarize, we did not find nasal breathing to affect several prognostic cardiovascular variables.

We also found that nasal breathing was associated with lower RPB, but not RPE, values during submaximal exercise. The median RPB values corresponded with “slight” breathlessness during nasal breathing and “moderate” breathlessness during oral breathing. We expected that nasal breathing would reduce RPE and RPB during exercise because the ventilatory response during exercise, which is attenuated with nasal versus oral breathing (39, 118), is linked to RPE and RPB (62, 64, 119–123). These variables are important because exercise-related breathlessness (i.e., exertional dyspnea) can reduce exercise participation (124, 125). One publication with relevance to this project concluded that oral breathing slightly increased RPE values relative to oronasal breathing during shuttle run testing in young males (126). Although our data did not reach statistical significance, our data (RPE of 13 vs. 12) are in agreement with the difference in RPE values (RPE of 19 vs. 18) in the past work, both having higher RPE values during oral breathing. To conclude, nasal breathing may have a modest effect on reducing RPB during acute submaximal exercise.

Limitations

First, we considered the ecological validity and proof of concept nature of this research when deciding not to measure V̇o₂, tidal volume, or end-tidal CO₂ in this initial investigation on the acute cardiovascular effects of nasal breathing. This project was a necessary first step before deciding whether to interrogate potential mechanisms related to additional respiratory and/or psychological variables in future projects. Also, a breathing apparatus to collect and sample gases during nasal breathing could have confounded our findings and added undue difficulty to the present work. Nevertheless, we were able to directly address our primary and secondary hypotheses. Second, because nasal volume increases (127, 128) and nasal resistance decreases (129) after acute moderate-intensity cycling exercise for ∼20 min, it is possible that there was an order effect. However, the randomized design prevented this concern. Moreover, the reduction in resting diastolic BP from oral to nasal breathing did not differ between those who were randomized to oral breathing first and those who were randomized to nasal breathing first.

Perspectives and Significance

At rest, nasal breathing was associated with lower BP and a greater parasympathetic contribution to HRV. During submaximal exercise, there were no differences between breathing routes on the cardiovascular variables assessed. These data suggest that nasal compared with oral breathing provides modest, but potentially clinically relevant, improvements in prognostic cardiovascular variables at rest, but not during exercise. This work advances our knowledge of how nasal breathing affects clinically relevant cardiovascular variables and provides foundational acute data in healthy young adults to justify future longer-term studies in other populations. Should these findings be confirmed in future studies, these data would provide partial support for a breathing modality with a high potential for translation (i.e., voluntarily breathing through the nose by keeping the lips sealed).

DATA AVAILABILITY

Data are available upon reasonable request to the principal investigator after institutional data transfer approvals.

GRANTS

This research was supported in part by the National Institutes of Health Grant K01HL160772 (to J.C.W.), American Heart Association Grant 23CDA1037938 (to J.C.W.), and an Undergraduate Research Opportunity Program Research Mentor Materials Grant (to J.C.W.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.C.W., J.N.C., S.L.B., J.M., A.K.B., I.M.F., J.M.C., A.M.M., and K.F.K. conceived and designed research; J.C.W., J.N.C., S.L.B., J.M., A.K.B., I.M.F., J.M.C., A.M.M., and K.F.K. performed experiments; J.C.W., J.N.C., S.L.B., J.M., A.K.B., I.M.F., J.M.C., A.M.M., and K.F.K. analyzed data; J.C.W., J.N.C., S.L.B., J.M., A.K.B., I.M.F., J.M.C., A.M.M., and K.F.K. interpreted results of experiments; J.C.W. prepared figures; J.C.W. drafted manuscript; J.C.W., J.N.C., S.L.B., J.M., A.K.B., I.M.F., J.M.C., A.M.M., and K.F.K. edited and revised manuscript; J.C.W., J.N.C., S.L.B., J.M., A.K.B., I.M.F., J.M.C., A.M.M., and K.F.K. approved final version of manuscript.