Introduction
Assessment of pulmonary function is essential for diagnosing, monitoring and managing respiratory diseases.1 Chronic Obstructive Pulmonary Disease (COPD), one of the most common respiratory conditions, is the third leading cause of death globally.2 Early detection of airflow obstruction through effective case-finding strategies enables timely interventions and may improve long-term outcomes.3,4
Spirometry is the clinical standard for detecting airflow limitation.1,5 Key spirometry parameters include Forced Expiratory Volume in the first second (FEV1) and Forced Vital Capacity (FVC), with the FEV1/FVC ratio being the primary indicator of airway obstruction. COPD is typically defined by a post-bronchodilator FEV1/FVC ratio below 0.70.5–9 Despite its diagnostic value, spirometry remains underused in clinical practice, particularly in primary care, contributing to misdiagnosis and underdiagnosis of COPD.10 This underuse persists despite adequate availability in high-resource settings and it attributed to practical barriers such as time constraints, insufficient training, and competing priorities,10 while in low-resource settings access to equipment and trained personnel remains limited.11–13 In addition, spirometry requires active patient cooperation during forced expiratory maneuvers, which can be challenging for frail older adults or individuals with cognitive impairment, potentially resulting in failed or unreliable measurements.10 In such situations, there is a need for simple, low-burden case-finding methods that can provide an accessible means of detecting airflow obstruction in primary care when spirometry is not performed, not feasible, or yields unreliable results. Importantly, such methods are intended to support clinical decision-making in these specific contexts by identifying patients who may benefit from confirmatory spirometry rather than replacing spirometry or introducing an additional barrier in the diagnostic pathway.
Tidal breathing analysis may offer a simple and minimally demanding case-finding approach, as it can be performed during quiet breathing without forced maneuvers.14,15 Tidal breathing patterns are known to be altered in obstructive pulmonary diseases.16–20 Recent developments enable use of nasal pressure measurements to derive scaled tidal flow-volume curves,21 providing a low-burden repeatable means of monitoring respiratory patterns.21,22 However, while previous studies have demonstrated the potential of such measures to distinguish individuals with COPD from healthy controls,21 their diagnostic performance for detecting spirometry-defined airflow obstruction in heterogeneous, clinically relevant populations remains unclear.
This study aimed to determine the diagnostic performance of nasal-pressure tidal breathing patterns for detecting spirometry-defined airflow obstruction (FEV1/FVC < 0.70) in a heterogeneous population referred for diagnostic spirometry. We hypothesized that these measures can discriminate between individuals with and without airflow obstruction and support identification of patients requiring confirmatory spirometry.
Material and Methods
Study Design and Participants
For this prospective diagnostic accuracy study, all subjects referred for a pulmonary function test at Maastricht UMC+ (MUMC+) between April 2025 and July 2025 were consecutively approached for participation. This population included unselected referrals from primary and secondary care, including individuals with and without respiratory symptoms, and with or without a prior respiratory diagnosis. Participants were eligible if they were ≥ 18 years of age.
Exclusion applied to participants who were unable to maintain nasal breathing, received chronic oxygen therapy delivered via a nasal cannula, or were incapable of providing informed consent for participation in the research. The Medical Ethics Committee of Maastricht UMC+ reviewed the study and determined that it did not fall under the scope of the Medical Research Involving Human Subjects Act (2024–0479). All participants provided written informed consent and the study was conducted in accordance with the Declaration of Helsinki and applicable privacy regulations.
This study was written in accordance with the STARD 2015 guidelines.23
Baseline Characteristics
Participant characteristics were obtained from the medical records at the time of inclusion. Demographic data included age and sex, and anthropometric measurements (height and weight) were used to calculate body mass index (BMI). Clinical data included smoking history, with ≥ 10 pack-years classified as (ex-)smokers and < 10 pack-years as nonsmokers, consistent with previous COPD research.24–26 Prior respiratory diagnoses documented by treating physicians in the medical records were also recorded.
Lung function parameters were obtained from spirometry.
Measurements – Reference Standard: Spirometry
Spirometry was conducted as part of routine clinical care during routine visits. The spirometry was performed according to the American Thoracic Society/European Respiratory Society (ATS/ERS) guidelines.1,9 A minimum of three acceptable flow-volume curves were produced. The flow-volume curve with the best expiration was selected for analysis in this study.
The parameters FEV1 and FVC were taken as the maximum values among the acceptable curves.1 Spirometry measurements were performed by trained personnel and all maneuvers were reviewed for technical acceptability according to ATS/ERS criteria.9 A pre-specified cut-off of post-bronchodilator FEV1/FVC < 0.70 was used to define airway obstruction,5,9 with post-bronchodilator measurements performed following administration of Airomir (4x 100µg; total 400µg). In this study, spirometry was used as the reference standard, whereas the nasal pressure-tidal curve method was considered the index test. Assessors of the reference standard were blinded to index test results.
Measurements – Index Test: Nasal Pressure-Tidal Curves
Nasal pressure derived tidal breathing curves were measured at rest using a prototype device developed by AMES BV (the Netherlands). Both tests were performed at a single time point, with spirometry conducted first followed by the nasal pressure-tidal curve measurement after completion of the routine spirometry visit. The device consists of a pressure sensor (SDP810-500Pa, Sensirion, Switzerland) connected to a nasal cannula (Salter Labs nasal), enabling the recording of nasal pressure signals.
Participants were instructed to sit in a comfortable position and breathe naturally through their noses at their own pace, without any specific breathing technique being required. It was emphasized that they should avoid breathing through their mouth and focus on nasal breathing for a duration of two minutes. The recorded nasal pressure signals were processed using a previously described algorithm to derive scaled tidal flow-volume curves from the nasal pressure signal.22 A detailed description of the signal processing and curve-generation procedure is provided in the data processing section.
The resulting curves were subsequently analyzed to estimate the FEV1/FVC ratio. The estimated FEV1/FVC ratio was compared with the spirometry-derived FEV1/FVC ratio used as the reference measure. A technical issue with the index test was predefined as the inability to obtain a valid, interpretable measurement from the nasal pressure-tidal curve method, resulting in no valid index test result.
These participants were excluded from analysis. All tidal breathing data were analyzed blinded to clinical information (age, sex, diagnosis).
Data Processing
The flow-volume shape reconstruction from the pressure recordings was performed using the method of Hebbink and Hagmeijer,22 whereas estimation of the FEV1/FVC ratio followed the method of Hebbink et al.21 The overall procedure is summarized briefly below, with a schematic overview of the complete processing workflow provided in Supplementary figure S1.
The pressure difference Δp generated by airflow over the nasal cannula is measured at the device-end of the cannula tube. The flow rate Q is assumed to be related to the pressure difference through a power-law relationship: Q=a (Δp)b, where the constants a and b differ between inspiration and expiration. Once the two a values and two b values are defined, respiratory volume is obtained by integrating the reconstructed flow over time, resulting in a tidal flow-volume curve.
The two prefactors a depend on the individual nasal cannula-nostril configuration and therefore vary between individual measurements. Assuming that, over ten consecutive breathing cycles, the inspired amount of air is equal to the expired amount of air yields the ratio of the two prefactors. This ratio is sufficient to reconstruct the shape of the flow-volume curve without determining its absolute scale.
The two powers b were derived from experiments in 3D printed airways and depend on the degree of turbulence of the nasal flow. For inspiration b=0.75, and for expiration b=0.55. By measuring both tidal and forced flow-volume curves in a group of subjects, a statistical relation was established between the shape of the tidal flow-volume curve and the FEV1/FVC ratio derived from the forced flow-volume curve.
In this study, this relation is referred to as the spirometric relation. To facilitate a convenient mathematical description of this relation, the shape of the tidal flow-volume curve is accurately captured by 15 coefficients. The spirometric relation is therefore formulated as a statistical relation between these 15 coefficients and the spirometry-derived FEV1/FVC ratio.
Once established, this relation can be used to estimate the FEV1/FVC ratio of an individual subject from the shape of their tidal flow-volume curve, resulting in an estimated FEV1/FVC ratio.22
Model Development and Evaluation Strategy
To construct the spirometric relation, the available participants were randomly divided into a model group and a control group. The two groups formed complementary parts of the total cohort, such that each participant was included in only one group. The model group was used to establish the spirometric relation and the control group was used to evaluate its validity.
For each participant in the control group, the coefficients derived from the tidal flow-volume curve were compared with the coefficient values predicted by the spirometric relation across the range of possible FEV1/FVC ratio values. The estimated FEV1/FVC ratio was defined as the FEV1/FVC ratio yielding the minimum sum of squared differences between the observed and model-predicted coefficients. The estimated FEV1/FVC ratios were then compared with the corresponding spirometry-derived FEV1/FVC ratios.
Agreement between the estimated and spirometry-derived FEV1/FVC ratios was assessed, and the ability of the method to identify airflow obstruction, defined as a spirometry-derived FEV1/FVC < 0.70, was evaluated.5 The subset-based analysis was performed to assess the internal consistency and robustness of the spirometric relation using an independent subset of participants. Subsequently, a secondary analysis was performed using the full cohort to establish the spirometric relation to estimate the FEV1/FVC ratio for all participants.
Sample Size Calculation
Sample size was estimated for a paired diagnostic accuracy study comparing the nasal pressure-tidal curve method with spirometry as the reference standard. Conservative a priori assumptions were applied (expected positive test proportions of 90% for the reference standard and 80% for the index test, with a minimum clinically relevant difference of 10%), with 80% power and a 5% type I error. Depending on assumptions regarding the probability of disagreement between tests, required sample sizes ranged from 78 to 164 participants after continuity correction.
To ensure a conservative estimate, the maximum sample size was adopted, resulting in 164 participants undergoing both the reference and index test.27
Statistics
Data were presented as median (interquartile range), mean (standard deviation), counts, and percentages, as appropriate. Normality was assessed using histograms and the Shapiro–Wilk test. Continuous variables were compared using the independent samples t-test or the Mann–Whitney U-test, and paired comparisons using the Wilcoxon signed-rank test.
Categorical variables were compared with the chi-square test. Diagnostic accuracy of nasal pressure-tidal curves for detecting obstructive ventilatory impairments was evaluated using an FEV1/FVC < 0.70 as the reference standard. To determine the diagnostic performance of the nasal pressure-tidal curves, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and overall accuracy were calculated with two-by-two contingency tables with spirometry as reference.
Receiver operating characteristic (ROC) analysis was used to assess alternative cut-off values for the nasal pressure-tidal curve method across multiple spirometry-defined thresholds. Optimal cut-offs were determined using the Youden index. Agreement between the estimated FEV1/FVC ratio and the spirometry-derived FEV1/FVC ratio was assessed using a Two-way Mixed-Effects Model Intraclass Correlation Coefficient (ICC (3,1)) and interpreted as poor (<0.50), moderate (0.50–0.75), good (0.75–0.90), or excellent (>0.90).28 Agreement was further evaluated with a Bland-Altman plot.
All statistical tests were conducted in SPSS v28.0.1 (SPSS Inc, Chicago, IL, USA), with a two-sided significance level of 0.05.
Results
A total of 201 participants were measured between April and July 2025. Of these, nine participants were excluded due to technical issues (n=5), withdrawn consent (n=1), inability to breathe through the nose for 2 minutes (n=2), no data or diagnosis available due to screening for transplantation (n=1). The remaining 192 participants were enrolled and completed both spirometry and nasal pressure-tidal curve measurements.
Baseline demographics and clinical characteristics are summarized in Table 1. 61.5% of the total population was female. The median age was 66 (56.0–74.0) years, median BMI was 26.9 (24.4–30.8) kg/m2 and median FEV1%predicted was 84.0 (60.0–101.8)%.
Of these 192 participants, 94 participants (49.0%) had spirometry-defined airflow obstruction and 98 participants (51.0%) had non-obstructive spirometry (See Supplementary Table S1 for a detailed breakdown of diagnoses). Participants in the obstructive group were older, with a median age of 68.0 (61.0–76.0) years, and demonstrated a lower median FEV1%predicted (64.93 (44.75–82.25)%) and spirometry-derived FEV1/FVC ratio (0.55 (0.40–0.64)) compared to the non-obstructive group. The estimated FEV1/FVC ratio derived from the nasal pressure-tidal curve method was lowest in the obstructive group (0.52 (0.39–0.80)), reflecting the pattern observed with conventional spirometry.
Smoking history was significantly associated with airflow obstruction (χ2=10.69, p=0.001). Among (ex-) smokers, 59% (66/112) had obstructive spirometry, compared to 35% (28/80) among non-smokers. In the total population, the estimated FEV1/FVC ratio derived from nasal pressure-tidal curve measurements and the spirometry-derived FEV1/FVC ratio were comparable (0.69 vs 0.70, p=0.571), although variability in the estimated FEV1/FVC ratio was consistently higher.
Median estimated FEV1/FVC ratio were also closely aligned with spirometry-derived FEV1/FVC ratios in participants with obstruction (0.52 vs 0.55) and without obstruction (0.77 vs 0.78), with no significant differences in any group (p=0.571).Table 1Participants Demographics and Baseline Characteristics for the Total Population and Separately for Participants with and without Spirometry-Defined Airflow ObstructionParticipant CharacteristicsTotalPopulation(n=192)Obstructive(FEV1/FVC < 0.70)(n=94, 49.0%)Non-Obstructive(FEV1/FVC ≥ 0.70)(n=98, 51.0%)p-valueSex, n (%) Male74 (38.5%)43 (45.7%)31 (31.6%)0.054 Female118 (61.5%)51 (54.3%)67 (68.4%)Age (years)66 (56.0–74.0)68 (61.0–76.0)63 (46.8–72.0)<0.001BMI (kg/m2)26.9 (24.4–30.8)26.5 (24.0–29.8)27.6 (24.5–31.5)0.082GOLD, n (%) (COPD only, n=48) GOLD I5 (10.4%)5 (10.4%)N/A GOLD II20 (41.7%)20 (41.7%)N/A GOLD III GOLD IV15 (31.3%)8 (16.7%)15 (31.3%)8 (16.7%)N/AN/ASmoke history, n (%) (ex-)smokers (≥ 10 pack-years)112 (58.3%)66 (70.2%)46 (46.9%)0.001 Non-smokers (< 10 pack-years)80 (41.7%)28 (29.8%)52 (53.1%)FEV1 (L)2.22 (1.5–2.9)1.79 (1.10–2.38)2.67 (2.18–3.33)<0.001Predicted (%)84.00 (60.0–101.8)64.93 (44.75–82.25)97.50 (84.0–109.0)<0.001Z-score−1.05 (−2.50–0.14)−2.03 (−3.30- −1.03)−0.14 (−1.06–0.63)<0.001FVC (L)3.38 ± 1.073.31 ± 1.103.45 ± 1.040.353Predicted (%)93.0 (82.0–110.5)90.0 (78.0–109.0)96.5 (86.0–111.25)0.252Z-score−0.40 (−1.12–0.73)−0.66 (−1.46–0.55)−0.22 (−1.02–0.79)0.334Spirometry-derived FEV1/FVC ratio0.70 (0.56–0.78)0.55 (0.40–0.64)0.78 (0.75–0.84)<0.001Z-score−1.07 (−2.47- −0.08)−2.48 (−3.72- −1.73)−0.14 (−0.50–0.55)<0.001Estimated FEV1/FVC-ratio (Nasal pressure-tidal curve method)0.69 (0.48–0.86)0.52 (0.39–0.80)0.77 (0.58–0.89)<0.001Notes: Variables are presented as n (%), mean (±SD), and median (IQR) with corresponding p values. Columns represent the total population, participants with obstructive spirometry, participants with non-obstructive spirometry, and the p-value for comparison between obstructive and non-obstructive groups. GOLD classification is reported only for participants with COPD within the obstructive group (n=48).
Other obstructive diagnoses are not included in this classification.Abbreviations: BMI; body mass index, GOLD; Global Initiative for Chronic Obstructive Lung Disease, FEV1; forced expiratory volume in 1 s, FVC; forced vital capacity, FEV1/FVC ratio; ratio of volume of air exhaled in the first second (FEV1) compared to the total volume of air exhaled (FVC).
Diagnostic Accuracy of Nasal Pressure-Tidal Curve Method Using a Predefined FEV1/FVC Cut-off
Figure 1 illustrates the relationship between the estimated FEV1/FVC ratio derived from the nasal pressure-tidal curve method and the spirometry-derived FEV1/FVC ratio, separately for participants with and without spirometry-defined airflow obstruction. Misclassification occurred in both directions. The nasal pressure-tidal curve method classified 36.7% of participants with non-obstructive values (FEV1/FVC ≥ 0.70) as obstructive, and failed to identify obstruction in a subset of patients with spirometry-defined airflow limitation (FEV1/FVC < 0.70) (n=32).
In a limited number of cases, the estimated FEV1/FVC ratio exceeded the physiological range (≥ 1.0), reflecting limitations of the estimation algorithm. Figure 1Scatterplot of the spirometry-derived FEV1/FVC ratio (x-axis) vs the estimated FEV1/FVC ratio from the nasal pressure-tidal curve method (y-axis), separately for participants with and without spirometry-defined airflow obstruction. Vertical and horizontal dashed lines indicate the predefined diagnostic cut-off of 0.70.
The plot visualizes the distribution of measurements and highlights concordant and discordant classifications in relation to airflow obstruction, illustrating overlap and potential misclassification. True positives (TP=62), false negatives (FN=32), true negatives (TN=62), false positives (FP=36).A scatter plot features two marker types with a legend: Non-obstructive and Obstructive. The horizontal axis is labeled ′FEV1/FVC ratio Spirometry′ with ticks at 0.20, 0.40, 0.60, 0.80 and 1.00.
The vertical axis is labeled ′FEV1/FVC ratio Nasal pressure tidal curve method′ with ticks at 0.00, 0.20, 0.40, 0.60, 0.80, 1.00 and 1.20. Dashed lines at 0.70 on both axes divide the plot into four quadrants: FN 32 (upper left), TN 62 (upper right), TP 62 (lower left) and FP 36 (lower right). Obstructive markers are mainly left of 0.70 horizontally, ranging from 0.20 to 0.70 and vertically from 0.10 to 1.15, with some above 1.00.
Non-obstructive markers are mainly right of 0.70 horizontally, ranging from 0.70 to 1.00 and vertically from 0.30 to 1.20, clustering around 0.70 to 0.90.A scatter plot comparing the FEV1/FVC ratio measured by spirometry with the FEV1/FVC ratio measured using the nasal-pressure tidal curve method.
At the standard spirometry cut-off of FEV1/FVC < 0.70, the nasal pressure-tidal curve method yielded 36 false-positive and 32 false-negative classifications (Table 2), resulting in an overall accuracy of 64.6%, with a sensitivity of 66.0% and a specificity of 63.3% (Table 3).Table 22×2 Contingency Table Comparing the Estimated FEV1/FVC Ratio Derived from the Nasal Pressure-Tidal Curve Method (Index Test) with the Spirometry-Derived FEV1/FVC Ratio (Reference Standard), with Airflow Obstruction Defined as FEV1/FVC < 0.70Reference Standard: Non-Obstructive(Spirometry-Derived FEV1/FVC ≥ 0.70)Reference Standard: Obstructive(Spirometry-Derived FEV1/FVC < 0.70)TotalIndex test: Non-obstructive(estimated FEV1/FVC ≥ 0.70)62 (true negatives)32 (false negatives)94Index test: Obstructive(estimated FEV1/FVC < 0.70)36 (false positives)62 (true positives)98Total9894192Abbreviations: FEV1, forced expiratory volume in the first second; FVC, forced vital capacity; FEV1/FVC, ratio of forced expiratory volume in the first second to forced vital capacity.Table 3Diagnostic Performance of the Nasal Pressure-Tidal Curve Method for Detecting Airflow Obstruction (Reference Standard: FEV1/FVC < 0.70), Shown for an Identical Cut-off (0.70) and the Youden-Optimized Cut-off (0.56)MeasureCutoff0.700.56Value (95% CI)Value (95% CI)Sensitivity66.0% (55.5–75.4)55.3% (44.7–65.6)Specificity63.3% (52.9–72.8)79.6% (70.3–87.1)Positive Predictive Value (PPV)63.3% (56.1–69.9)72.2% (62.3–80.0)Negative Predictive Value (NPV)66.0% (58.5–72.7)65.0% (59.2–70.4)Accuracy64.6% (57.4–71.3)67.7% (60.6–74.3)
At the standard spirometry cut-off of 0.70, the Youden-optimized nasal cut-off was 0.56 (Youden index = 0.35). Application of this optimized threshold increased overall accuracy to 67.7% and specificity to 79.6%, while sensitivity decreased to 55.3% (Table 3). Details of the 2×2 contingency table are provided in Supplementary Table S2.
Sensitivity Analysis of Alternative Cut-off Values
Diagnostic performance of the nasal pressure-tidal curve method was assessed across alternative spirometry cut-offs (FEV1/FVC 0.6 to 0.3) using both identical and Youden-optimized nasal pressure-tidal curve thresholds (see Supplementary Table S3 for optimized thresholds and Youden indices).
Diagnostic performance improved progressively with lower spirometry thresholds. When identical nasal pressure-tidal curve cut-offs were applied, accuracy increased from 70.8% (95% CI: 63.9–77.2) at a cut-off of FEV1/FVC < 0.60 to 96.4% (95% CI: 92.6–98.5) at a cut-off of < 0.30 (Figure 2A). Application of Youden-optimized nasal pressure-tidal curve cut-offs improved diagnostic accuracy at the higher spirometry thresholds (FEV1/FVC < 0.70–0.50), whereas at lower thresholds (FEV1/FVC < 0.40–0.30) accuracy was comparable to, or lower than that obtained with identical cut-offs.
The highest accuracy was observed at a threshold of 0.30 for both methods (96.4% (95% CI: 92.6–98.5)). Figure 2Accuracy (A), sensitivity (B) and specificity (C) of the nasal pressure–tidal curve method across spirometry thresholds (FEV1/FVC 0.7–0.3). The x-axis shows spirometry thresholds; Youden-optimized nasal cut-offs corresponding to each spirometry cut-off ranged from 0.56 to 0.43 (Supplementary Table S3).
Error bars represent 95% confidence intervals.Image A displays a line graph of Accuracy (%) vs Spirometry cut-off (FEV1/FVC ratio). The x-axis spans 0.7 to 0.3 and the y-axis ranges from 0 to 100%. Two series are shown: Identical cut-off and Youden optimized cut-off, both with error bars.
Identical cut-off values: 65 at 0.7, 71 at 0.6, 78 at 0.5, 89 at 0.4, 96 at 0.3. Youden optimized values: 67 at 0.7, 81 at 0.6, 84 at 0.5, 85 at 0.4, 87 at 0.3. Image B shows Sensitivity (%) vs Spirometry cut-off.
Identical cut-off values: 66 at 0.7, 73 at 0.6, 72 at 0.5, 68 at 0.4, 67 at 0.3. Youden optimized values: 55 at 0.7, 60 at 0.6, 69 at 0.5, 86 at 0.4, 100 at 0.3, with a long error bar at 0.3. Image C illustrates Specificity (%) vs Spirometry cut-off.
Identical cut-off values: 63 at 0.7, 70 at 0.6, 80 at 0.5, 91 at 0.4, 98 at 0.3. Youden optimized values: 80 at 0.7, 90 at 0.6, 88 at 0.5, 85 at 0.4, 86 at 0.3, all with error bars.Three line graphs showing accuracy, sensitivity and specificity across spirometry cut off thresholds.
Sensitivity increased with decreasing spirometry thresholds for Youden-optimized cut-offs, surpassing that of identical cut-offs at lower thresholds (Figure 2B), while sensitivity remained stable for identical cut-offs. Specificity was higher for Youden-optimized cut-offs at higher thresholds, whereas identical cut-offs performed better at lower thresholds (Figure 2C). Sensitivity estimates showed wide 95% confidence intervals due to the small number of obstructive cases (n=9), whereas specificity estimates were more precise.
Detailed diagnostic metrics for each threshold combination are presented in Supplementary Table S4.
Agreement Between Reference Standard and Index Test
The overall ICC for the total population was 0.51 (95% CI: 0.40–0.61), indicating a poor to moderate reliability. Subgroup analyses showed poor agreement in both the obstructive (ICC 0.45, 95% CI: 0.27–0.60) and non-obstructive group (ICC 0.18, 95% CI: −0.02–0.36) (Table 4). A Bland-Altman plot (Figure 3) demonstrated a small mean difference of −0.016, but wide 95% limits of agreement ranging from −0.43 to 0.40, indicating substantial variability and limited agreement between methods.Table 4Reliability of Measurements Expressed as Intraclass Correlation Coefficients (ICCs) for the Total Population and Separately for Participants with and without Spirometry-Defined Airflow ObstructionParticipant GroupNICC (95% CIs)InterpretationTotal population1920.51 (0.40–0.61)ModerateObstructive (FEV1/FVC < 0.70)940.45 (0.27–0.60)PoorNon-obstructive (FEV1/FVC ≥ 0.70)980.18 (−0.02–0.36)PoorAbbreviations: FEV1, forced expiratory volume in the first second; FVC, forced vital capacity; FEV1/FVC, ratio of forced expiratory volume in the first second to forced vital capacity.Figure 3Bland-Altman plot comparing the estimated FEV1/FVC ratio derived from the nasal pressure-tidal curve method with the spirometry-derived FEV1/FVC ratio.
Each point represents an individual measurement, separately for participants with and without spirometry-defined airflow obstruction. The solid horizontal line indicates the mean difference (bias) of −0.016, reflecting systematic deviation between methods. The dashed lines represent the 95% limits of agreement, ranging from −0.43 to 0.40.Bland-Altman scatter plot: the mean FEV1/FVC ratio ranges from 0.00 to 1.20 and the difference ranges from -0.50 to 0.50.
Discussion
In this study, the diagnostic performance of the nasal pressure-tidal curve method was evaluated against spirometry, with the aim of distinguishing spirometry-defined airflow obstruction from non-obstructive spirometry. At the standard spirometry cut-off (FEV1/FVC < 0.70), the nasal pressure-tidal curve method demonstrated moderate diagnostic performance (sensitivity 66.0%, specificity 63.3%, accuracy 64.6%). Misclassification occurred in both directions, reflecting the inherent physiological overlap between tidal breathing patterns in individuals with and without airflow obstruction.
These findings suggest that the method can distinguish obstructive from non-obstructive ventilatory impairments to a limited extent, but also highlights substantial variability at the individual patient level.
ROC analysis showed that optimizing cut-offs resulted in only modest improvements in diagnostic performance at the standard threshold, highlighting the inherent trade-off between sensitivity and specificity. Diagnostic accuracy increased at lower spirometry thresholds, primarily due to improved detection of more severe airflow obstruction, where physiological differences are more pronounced. In contrast, mild airflow obstruction is more difficult to detect using tidal breathing measurements, as resting breathing patterns may remain relatively preserved.
Optimization of nasal pressure-tidal curve cut-offs appears to provide the greatest benefit at higher spirometry thresholds, where physiological overlap is more substantial. For severe obstruction, group separation is already pronounced and optimization offers limited additional gain. As primary care screening aims to detect early disease, performance at FEV1/FVC < 0.70 remains most clinically relevant.
Together, these findings suggest that the nasal pressure-tidal curve method demonstrates insufficient discriminatory ability at the conventional threshold and performs increasingly well in the presence of more severe airflow limitation.
Previous work by Hebbink et al provided important context for interpreting these findings.22 In-vitro studies demonstrated that nasal pressure-derived tidal curves can be reconstructed with high technical accuracy, showing near-perfect agreement with reference flow-volume curves.22 However, subsequent analyses revealed substantial inter-individual variability in tidal breathing patterns, limiting the ability to reliably infer airway obstruction from tidal breathing alone.21 The present study confirms these observations in a clinical setting. Despite small mean differences, Bland-Altman analysis showed wide limits of agreement, and ICC values indicated poor to moderate agreement, suggesting substantial variability at the individual level rather than systematic measurement bias. The moderate sensitivity and specificity observed likely reflect the inherent physiological overlap between tidal breathing patterns between individuals with and without airflow obstruction.
An additional contributor to this overlap is the inclusion of patients with asthma in the study population. In clinical practice, patients referred for case-finding through spirometry are inherently heterogenous in terms of the underlying (obstructive) respiratory conditions. Obstructive airway diseases share common features while differing in reversibility and temporal variability of airflow limitations.
In asthma, particularly when well controlled or assessed outside exacerbations, tidal breathing patterns may appear relatively normal despite airflow obstruction detected during forced expiratory maneuvers. This may contribute to both false-negative and false-positive classifications. Importantly, this overlap reflects the non-disease specific nature of tidal breathing analysis and suggests that the method is more suited to identifying individuals with potentially abnormal respiratory patterns than to distinguishing between specific obstructive airway diseases.
In this context, the nasal pressure-tidal curve method may fulfil a triage function, albeit with important limitations. Approximately one-third of the patients with spirometry-defined airflow obstruction were not identified, while false-positives occurred among those with non-obstructive spirometry. These findings can be partly explained by fundamental differences in measurement principles: spirometry captures maximal airflow during a forced expiratory maneuver, whereas nasal pressure measurements reflect breathing patterns during tidal respiration.
As these approaches assess different aspects of pulmonary function, limited direct agreement is not unexpected, reinforcing the role of the nasal pressure-tidal curve method as a supportive screening tool rather than a diagnostic replacement.
COPD remains a major global health problem, with many cases undiagnosed.2,10,29 A low-threshold, effort-independent screening tool may be valuable for COPD case-finding in primary care, particularly where spirometry is underused or difficult to perform such as in frail or cognitively impaired patients or resource-limited settings. Importantly, it is not intended to replace spirometry or to serve as a universal pre-screening step in settings where spirometry is readily performed. Instead, its potential value lies in specific clinical scenarios where spirometry is underutilized or difficult to perform.
In such contexts, a positive screening result could support timely referral for confirmatory spirometry, facilitating earlier identification of airflow obstruction in at-risk individuals.
A key strength of this study is the inclusion of an unselected, clinically representative population, comprising referrals from both primary and secondary care. Furthermore, the use of spirometry as reference and the inclusion of both diagnostic accuracy and agreement analysis provides a comprehensive assessment of performance. Several limitations should also be acknowledged.
In a small number of cases, the estimated FEV1/FVC ratios exceeded the physiological range (≥ 1.0), reflecting limitations of the current estimation algorithm rather than true physiological values. These artefacts, likely related to inaccuracies in tidal volume or flow estimation, may have contributed to misclassification near the diagnostic threshold. Additionally, while testing order was standardized, with spirometry performed first followed by tidal measurements, it is unclear whether performing spirometry could have influenced the subsequent tidal recordings.
However, sufficient time was allowed between measurements to minimize this effect. Airflow obstruction was defined by a post-bronchodilator FEV1/FVC < 0.70. While the Lower Limits of Normal (LLN) approach adjusts for age-related decline in FEV1/FVC ratio, the fixed ratio is widely used in clinical practice and guidelines for COPD case-finding.5 Given the focus on primary care screening and the need for a pragmatic, guideline-aligned threshold, the fixed ratio was considered the most appropriate reference standard for this study.
Future studies should investigate whether nasal pressure-tidal curve measurements can detect longitudinal changes in airway function. The growing interest in home-based pulmonary monitoring underscores the need for techniques that are easy to use, minimally effort-dependent, and applicable outside the clinic, a niche that nasal pressure-tidal curve measurements may help to fill.30 The impact of symptom status on diagnostic accuracy should also be evaluated, as symptoms were not incorporated into the present study.
Conclusion
In conclusion, nasal pressure analysis demonstrated moderate diagnostic performance in identifying spirometry-defined airway obstruction. As the method captures general breathing patterns rather than disease-specific pathophysiology, limited agreement with spirometry-defined obstruction is not unexpected. These findings suggest that nasal pressure-derived tidal breathing measurements may have limited potential as a non-invasive tool for detecting airway obstruction, however, further research is required to establish their clinical utility, particularly in primary care settings.