Dose–response relationship between obstructive sleep apnoea
severity and C-reactive protein levels: data from the European
Sleep Apnoea Database
Ludger Grote 1,2, Haralampos Gouveris 3, Lea Lethuillier4, Johan Verbraecken 5, Ozen K. Basoglu 6,
Sophia Schiza7, Ondrej Ludka8, Silke Ryan 9, Pavol Joppa 10, Francesco Fanfulla11,
Stefan Mihaicuta 12, Tarja Saaresranta13, Pawel Sliwinski 14, Jan Hedner1,2, Jean Louis Pepin4,
Sebastien Bailly 4 and the ESADA Study Group15
1Centre for Sleep and Wake Disorders, Institute of Medicine, Sahlgrenska Academy, Gothenburg University, Gothenburg, Sweden.
2Pulmonary Department, Sahlgrenska University Hospital, Gothenburg, Sweden. 3Sleep Medicine and Neurostimulation Centre,
Department of Otorhinolaryngology, University Medical Centre, Mainz, Germany. 4Grenoble Alpes University, HP2 Laboratory, INSERM
U1300 and Grenoble Alpes University Hospital, Grenoble, France. 5Multidisciplinary Sleep Disorders Centre, Antwerp University
Hospital and University of Antwerp, Antwerp, Belgium. 6Ege University, Department of Respiratory Medicine, Faculty of Medicine,
Izmir, Turkey. 7University of Crete, Sleep Disorders Unit, Department of Respiratory Medicine, Heraklion, Greece. 8Department of
General Internal Medicine, University Hospital Brno and Faculty of Medicine, Masaryk University, Brno, Czech Republic. 9Department
of Respiratory Medicine, St Vincent’s University Hospital, School of Medicine, University College Dublin, Dublin, Ireland. 10Department
of Respiratory Medicine and Tuberculosis, Faculty of Medicine, P.J. Safarik University, Kosice, Slovakia. 11Unità Operativa diMedicina
del Sonno, Istituto Clinici Scientifici Maugeri di Pavia e Montescano IRCCS, Pavia, Italy. 12Centre for Research and Innovation in
Precision Medicine of Respiratory Diseases, Department of Pulmonology, University of Medicine and Pharmacy, Timisoara, Romania.
13Turku University Hospital, Division of Medicine, Department of Pulmonary Diseases, and Sleep Research Centre, Department of
Pulmonary Diseases and Clinical Allergology, University of Turku, Turku, Finland. 14Institute of Tuberculosis and Lung Diseases, 2nd
Department of Respiratory Medicine, Warsaw, Poland.
Corresponding author: Ludger Grote (ludger.grote@lungall.gu.se)
Shareable abstract (@ERSpublications)
Obstructive sleep apnoea is characterised by elevated systemic inflammation. Analysis performed
in >18 000 patients across Europe established a causal relationship between OSA severity measures
and CRP blood levels independent of confounders. https://bit.ly/44TK60A
Cite this article as: Grote L, Gouveris H, Lethuillier L, et al. Dose–response relationship between
obstructive sleep apnoea severity and C-reactive protein levels: data from the European Sleep Apnoea
Database. ERJ Open Res 2026; 12: 00707-2025 [DOI: 10.1183/23120541.00707-2025].
Abstract
Introduction Obstructive sleep apnoea (OSA) characterised by intermittent hypoxia promotes systemic
inflammation. This study evaluated the association between OSA severity and circulating C-reactive protein
(CRP) levels as marker of systemic inflammation in a pan-European patient cohort.
Methods This cross-sectional analysis of the multicentre European Sleep Apnoea Database (ESADA)
cohort used inverse probability weighted regression adjustment for multiple covariates within a linear
mixed-effects model (LMEM) to test the independent association between OSA severity and CRP levels.
Covariates included anthropometrics and comorbidities. Study centre and year of analysis accounted for
methodological variability in CRP analysis.
Results 18 445 subjects (71% male, median age 53 years (interquartile range 44–62), median apnoea–
hypopnoea index (AHI) 22.1 events per h (9–44.9)) were included. CRP (median 3.0 mg·L−1 (1.2–5.1))
increased in a dose–response fashion across OSA severity categories (2.0 (1.0–4.0) for AHI <5 events
per h; 2.5 (1.0–5.0) for AHI 5–<15 events per h); 2.9 (1.2–5.0) for AHI 15–<30 events per h; and
3.7 mg·L−1 (1.8–6.4) for AHI ⩾30 events per h; p<0.001, respectively). In the final LMEM model, AHI
remained an independent predictor of CRP concentration (p<0.001). Other significant predictors of CRP
were age and female sex. Obesity (body mass index ⩾35 kg·m−2) had, among other comorbidities, the
strongest independent effect on CRP levels with 2.7 mg·L−1 (95% CI 2.45–2.90).
Conclusions Our results showed a consistent and robust dose–response relationship between OSA severity
and systemic inflammation independent of usual confounders. The combination of OSA and obesity
Copyright ©The authors 2026
This version is distributed under
the terms of the Creative
Commons Attribution Non-
Commercial Licence 4.0. For
commercial reproduction rights
and permissions contact
permissions@ersnet.org
Received: 24 May 2025
Accepted: 13 July 2025
https://doi.org/10.1183/23120541.00707-2025 ERJ Open Res 2026; 12: 00707-2025
ERJ OPEN RESEARCH
ORIGINAL RESEARCH ARTICLE
L. GROTE ET AL.
 on January 26, 2026 at Turku University Library. Please see licensing information on first page for reuse rights.https://publications.ersnet.orgDownloaded from 
amplified the association. Future studies should address whether elevated CRP could serve as a prognostic
marker for subsequent cardiovascular events in OSA.
Introduction
Obstructive sleep apnoea (OSA) is a clinical syndrome characterised by repeated complete or partial
obstruction of the upper airway during sleep. OSA alters quality of life and leads to an increase in
cardiometabolic morbidity and overall mortality [1–4].
There is vast evidence that intermittent hypoxia, the hallmark feature of OSA, leads to systemic low-grade
inflammation which is likely a key mediator of cardiometabolic diseases [5–7]. C-reactive protein (CRP) is
recognised as the prototypic circulating marker of inflammation [8]. The CRP concentration, unlike other
markers of inflammation, is not influenced by circadian factors. High-sensitivity CRP (hsCRP) has been
identified as a superior biomarker if only a single blood test should be singled out to characterise
inflammation. Importantly, elevated CRP is a strong predictor of future cardiovascular (CV) events [8].
However, conflicting data have been reported for the link between OSA and CRP. While some studies
demonstrated higher levels in OSA patients compared with non-OSA patients [9–11] others found only a
weak correlation after controlling for the presence of obesity [12–14]. In two meta-analyses, serum CRP/
hs-CRP levels were higher in OSA patients compared with control subjects [15, 16]. Several additional
factors such as age, sex, obesity or comorbidities can influence baseline CRP serum levels [17]. Therefore,
it remains uncertain whether OSA per se is associated with increased CRP levels. Indeed, a recent
American Heart Association statement on OSA did not identify a role for cardiovascular risk stratification
by means of CRP levels in patients with OSA [18].
Thus, the aim of this analysis was to evaluate the role of CRP as a marker of systemic inflammatory
burden in OSA in our large cohort controlling for numerous confounding factors by using a reference
inverse of probability treatment weighting (IPTW) analysis. We hypothesised that OSA increased CRP
independent of confounders. The European Sleep Apnoea Database (ESADA) cohort is suitable to address
potential limitations due to rigorous classification of comorbidities and risk factors, and the adjustment for
potential genetic and lifestyle factors. In addition, the cohort size provided statistical power to apply an
advanced statistical model of causal inference methods on observational data.
Methods
Subjects
This was a cross-sectional analysis of the multicentre ESADA cohort. Data were collected as previously
described [19]. Briefly, patients with suspected OSA, aged 18–80 years, were recruited from March 2007
to December 2022. Patients needed to provide oral and/or written informed consent to be included in the
database. The ESADA database has been reviewed by the local independent ethics review boards at each
participating European sleep centre.
Anthropometrics, medical history and concomitant medication were captured at baseline and subjects
underwent sleep apnoea testing via respiratory polygraphy or a polysomnography technique according to
local practice. Respiratory events and sleep were classified according to the American Academy of Sleep
Medicine criteria with 4% desaturation criteria for hypopnoeas [20]. The apnoea–hypopnoea index (AHI)
was used to classify patients into four severity groups for OSA: no OSA (AHI<5), mild OSA
(5⩽AHI<15), moderate OSA (15⩽AHI<30) and severe OSA (AHI⩾30) [2]. In sensitivity analysis, OSA
severity was also classified for measures of nocturnal hypoxia, including oxygen desaturation index (ODI)
tertiles and tertiles of time below 90% oxygen saturation.
CRP
CRP was obtained by a venous blood sample taken in conjunction with the clinical evaluation at the sleep
centre and the analysis in plasma or serum was performed according to local routine designed to provide
either a low-sensitivity (ls) or predominantly high-sensitivity (hs) value. All patients who had a
measurement of CRP at baseline were included. CRP values equal to exactly 0 mg·L−1 were considered
missing (technical failure). Patients with a high inflammatory state as indicated by an elevated CRP
>20 mg·L−1 [21], those taking corticosteroids (ATC H02) or patients with any form of known cancer were
excluded from the analysis.
Data analysis
Data were expressed as number and percentage for qualitative variables and median and interquartile range
(IQR) for quantitative variables. Comparisons between groups were performed using the Pearson’s
https://doi.org/10.1183/23120541.00707-2025 2
ERJ OPEN RESEARCH ORIGINAL RESEARCH ARTICLE | L. GROTE ET AL.
 on January 26, 2026 at Turku University Library. Please see licensing information on first page for reuse rights.https://publications.ersnet.orgDownloaded from 
chi-squared test for categorical variables. Due to the non-normal and asymmetric distribution of several
quantitative variables (as assessed by visual inspection and Shapiro–Wilk tests), nonparametric Kruskal–
Wallis tests were used to compare groups for continuous variables. Given the large sample size, these tests
provide robust inference while avoiding assumptions about data distribution and variance homogeneity.
Due to the low rate of missing values (<2% for each variable), single imputation using the median or
mode was applied. Given the limited extent of missingness and the descriptive role of the imputed
variables, we considered this approach sufficient to preserve the structure of the dataset without
introducing model complexity such as multiple imputation [22]. A comparison of distributions before and
after imputation is provided in supplementary table S1.
To assess the “average treatment effect” of OSA on CRP values, we used an inverse probability of
treatment weighting (IPTW) regression adjusted (RA) approach for multiple exposure. For this, a directed
acyclic graph (DAG) was drawn to represent the relations between the variables (figure 1) [23]. The
relations between variables in the DAG were selected based on expert knowledge. Both pure risk factors
and confounders had to be included to compute the weights. Instrumental variables, which are variables
only related to the exposure and not related to the outcome, were not included as they will increase the
variance without reducing the bias [24]. A correctly specified DAG can then provide a rational choice of
confounding variables to adjust for in the computation of weights, and therefore increases the credibility of
the conditional exchangeability [25].
More details of the statistical model are provided in the supplementary material. To summarise, first an
ordinal regression was performed to assess individual weights by computing, for each individual, their own
probability of belonging to their OSA group. The ordinal model accounts for the fact that the transition
from no OSA to mild OSA may not be equivalent, in clinical impact or distribution, to the transition from
moderate to severe OSA. This framework allows us to preserve the ordinal nature of the exposure while
avoiding arbitrary assumptions about equal spacing between categories. The standardised mean difference
was used to assess balance for covariates before and after weighting (supplementary figure S1). To account
for variability in CRP measurements across European sleep centres and over time (e.g. differences in
methodology or high- versus low-sensitivity assays, timing of blood sampling), random effects were
included for both European “sleep centre” and “year” of CRP measurement. These random effects capture
unobserved heterogeneity related to site-specific practices and temporal trends, and account for the
Age
Sex
ESS
Comorbidities
SmokingBMI
Environmental
factors
OSA
severity
CRP
Outcome
Exposure
Measured confounders
Unmeasured confounders
FIGURE 1 Directed acyclic graph for the causal relation between obstructive sleep apnoea (OSA) severity and
C-reactive protein (CRP) levels. Dotted arrow represents the causal relation under investigation. Solid arrows
represent the known relations. ESS: Epworth sleepiness scale; BMI: body mass index.
https://doi.org/10.1183/23120541.00707-2025 3
ERJ OPEN RESEARCH ORIGINAL RESEARCH ARTICLE | L. GROTE ET AL.
 on January 26, 2026 at Turku University Library. Please see licensing information on first page for reuse rights.https://publications.ersnet.orgDownloaded from 
clustering of patients within centres and calendar years. This approach improves the generalisability of the
results beyond the observed units by assuming that the included European sleep centres and years represent
samples from larger populations. [26]. Ordinal logistic regression coefficient and weight distribution are
presented in supplementary tables S2 and S3. Second, to assess the impact of OSA on CRP, a weighted
mixed linear model was performed, adjusted on all the covariates used in the ordinal logistic regression
(IPTW-RA). In this model, the interaction between sex and AHI severity group has been tested, but was not
significant, and therefore was not included. Demographic and anthropometric data, Epworth sleepiness scale
(ESS), cardiometabolic comorbidities and COPD, were included in the analysis as potential confounders.
To avoid model-based standard errors we used bootstrap confidence interval as recommended [27]. The
stabilised weights and the estimates of the model were computed for each of the 1000 bootstrap samples.
The SD of the estimate across those samples was considered the SE of the estimate in the original one.
Finally, the residuals of the model were investigated to verify the assumptions of linear models. Also, the
significance of the random effects was assessed by removing the random-effect term and computing a
likelihood ratio test between the real model and the model without random effects.
Using the same method, sensitivity analyses were performed for men and women, as well as for ODI or
time <90% as markers of OSA severity instead of the AHI.
Statistical analyses were performed using R software (v.4.2.0, R Foundation for Statistical Computing,
Vienna, Austria) and a p-value threshold of 0.05 was considered significant.
Results
Population
A total of 18 445 patients with a median age of 53 years (IQR 44–62), 71% male sex and median body
mass index (BMI) of 30.5 kg·m−2 (26–35) from 29 European sleep centres with a median of 271 (59–769)
patients enrolled per centre were included in this analysis (figure 2 and table 1). Cardiovascular
comorbidities were prevalent with 42% for pre-existing systemic hypertension, 15% for diabetes mellitus,
8% for ischaemic heart disease and 6.5% for COPD. With increasing OSA severity, defined by AHI
severity, subjects were older, more obese, more likely to be male and showed a higher burden of
Patients in ESADA cohort
n=36 308
Excluded (n=17 442; 48%):
  No value of CRP (n=16 667)
  CRP value >20 mg·L–1 (n=775)
Excluded (n=421; 2%):
  No value of AHI (n=188)
  Taking ATC H02 (n=191)
  Cancer (n=42)
Patients eligible for baseline analysis
n=18 445
Patients with correct CRP value
n=18 866
No OSA:
AHI 0–<5
n=2884 (15.6%)
Mild OSA:
AHI 5–<15
n=3947 (21.4%)
Moderate OSA:
AHI 15–<30
n=4284 (23.2%)
Severe OSA:
AHI ≥30
n=7330 (39.7%)
FIGURE 2 Flowchart depicting the selection of patients from the European Sleep Apnoea Database (ESADA)
and categorisation of the population based on the severity of obstructive sleep apnoea (OSA), according to the
apnoea–hypopnoea index (AHI) metric. CRP: C-reactive protein; ATC: anatomical therapeutic chemical (in this
case, systemic corticosteroids).
https://doi.org/10.1183/23120541.00707-2025 4
ERJ OPEN RESEARCH ORIGINAL RESEARCH ARTICLE | L. GROTE ET AL.
 on January 26, 2026 at Turku University Library. Please see licensing information on first page for reuse rights.https://publications.ersnet.orgDownloaded from 
comorbidities. Anthropometrics and comorbidities were comparable without clinically significant
differences in ESADA subjects with and without (n=16 667) available CRP data (supplementary table S4).
CRP
The median value of CRP in the whole cohort was 3.0 mg·L−1 (1.2–5.1) and values increased in parallel
with an increase in OSA severity defined by AHI. The median (IQR) for CRP values were 2 mg·L−1
(1.0–4.0) for no-OSA group, 2.5 mg·L−1 (1.0–5.0) for mild OSA, 2.9 mg·L−1 (1.2–5.0) for moderate OSA
and 3.7 mg·L−1 (1.8–6.4) for severe OSA (p<0.0001) (figure 3). In all AHI categories, males had
significantly lower values compared with females across categories.
In multivariable analysis controlling for obesity, anthropometrics and comorbidities, patients with moderate
OSA and patients with severe OSA had a median CRP value augmented compared with no OSA patients
(estimate of 0.28 (0.07–0.49); p=0.002; and 0.58 (0.37–0.80); p<0.001, respectively) (table 2). Males had
significantly lower values than females with a mean CRP difference of 0.77 (−0.87–−0.61) (p<0.001).
Regarding comorbidities, having diabetes mellitus (p<0.001), a high BMI (p<0.001), left ventricular
hypertrophy (p<0.001), cardiac failure (p<0.001), COPD (p<0.001), neurological disease (p=0.017) and/or
inflammatory disease (p<0.001) was significantly associated with increased CRP values. However, there
was no significant association between the ESS score, systemic hypertension, transient ischaemic attack
(TIA) or stroke, ischaemic heart disease and psychiatric disease and serum CRP concentration (table 2
and figure 4).
Sensitivity analyses for CRP
The association between OSA severity and CRP levels, separated for male and female patients, is
presented in table 3. In males, the presence of both, moderate or severe OSA, significantly increased the
CRP levels compared with no OSA (estimate of 0.33 (0.11–0.55); p=0.003; and 0.71 (0.50–0.92);
p<0.001, respectively). However, in females there was a significant difference only when comparing those
with severe OSA with those with no OSA (estimate of 0.49 (0.19–0.79); p=0.001). Overall, the value of
those estimates was higher for males than females. In the analysis of CRP in relation to menopausal status,
TABLE 1 Clinical data: x-axis: AHI severity classes and statistics before weighting
Characteristic All population
n=18 445
No OSA
n=2884
Mild OSA
n=3947
Moderate OSA
n=4284
Severe OSA
n=7330
p-value#
C-reactive protein, mg·L−1 3.00 (1.20–5.10) 2.0 (1.0–4.0) 2.5 (1.0–5.0) 2.9 (1.2–5.0) 3.8 (1.8–6.4) <0.001
Age, years 53 (44–62) 46 (36–55) 52 (43–61) 55 (46–63) 55 (46–63) <0.001
ESS score 9 (6.0–13.0) 9.0 (5.0–13.0) 9.0 (5.0–13.0) 9.0 (5.0–13.0) 10.0 (6.0–14.0) <0.001
BMI, kg·m−2 30.5 (27.0–34.9) 27.2 (24.5–30.7) 29.0 (26.0–32.5) 30.1 (27.1–34.0) 33.1 (29.5–37.4) <0.001
BMI categories <0.001
<25 kg·m−2 2360 (13) 859 (30) 686 (17) 491 (11) 334 (4.6)
25–<30 kg·m−2 6067 (33) 1175 (41) 1573 (40) 1594 (37) 1715 (23)
30–<35 kg·m−2 5367 (29) 560 (19) 1116 (28) 1306 (30) 2511 (34)
⩾35 kg·m−2 4524 (25) 290 (10) 572 (14) 893 (21) 2770 (38)
Current smoking 4618 (25) 721 (25) 954 (24) 991 (23) 1952 (27) <0.001
Sex, male 13 036 (71) 1583 (55) 2574 (65) 3041 (71) 5838 (80) <0.001
Diabetes mellitus 2665 (15) 191 (6.6) 423 (11) 606 (14) 1445 (20) <0.001
Left ventricular hypertrophy 183 (1.0) 13 (0.5) 31 (0.8) 39 (0.9) 100 (1.4) <0.001
Systemic hypertension 7678 (42) 617 (21) 1429 (36) 1823 (43) 3809 (52) <0.001
Ischaemic heart disease 1525 (8.4) 103 (3.6) 308 (7.8) 359 (8.4) 755 (10) <0.001
TIA or stroke 417 (2.3) 42 (1.5) 95 (2.4) 107 (2.5) 173 (2.4) 0.017
Status post-myocardial infarction 393 (2.2) 31 (1.1) 82 (2.1) 99 (2.3) 181 (2.5) <0.001
Cardiac failure 720 (4.0) 57 (2.0) 109 (2.8) 160 (3.7) 394 (5.4) <0.001
Other CV comorbidities 2173 (16) 316 (11) 699 (18) 525 (12) 633 (8.6) <0.001
COPD 1183 (6.5) 123 (4.3) 211 (5.3) 259 (6.0) 590 (8.0) <0.001
Neurological disease 1089 (6.0) 193 (6.7) 254 (6.4) 270 (6.3) 372 (5.1) 0.001
Psychiatric disease 1993 (11) 405 (14) 451 (11) 447 (10) 690 (9.4) <0.001
Inflammatory disease 669 (3.7) 122 (4.2) 165 (4.2) 150 (3.5) 232 (3.2) 0.011
Data are presented as median (IQR) or n (%). OSA: obstructive sleep apnoea; ESS: Epworth sleepiness scale; BMI: body mass index; TIA: transient
ischaemic attack; CV: cardiovascular. “Other CV comorbidities” denote information derived from free text not captured by the categories above.
#: a Kruskal–Wallis test was performed for quantitative variables and a Pearson’s chi-squared test was performed for qualitative variables.
https://doi.org/10.1183/23120541.00707-2025 5
ERJ OPEN RESEARCH ORIGINAL RESEARCH ARTICLE | L. GROTE ET AL.
 on January 26, 2026 at Turku University Library. Please see licensing information on first page for reuse rights.https://publications.ersnet.orgDownloaded from 
we could not identify any significant difference in CRP values (3.20 mg·L−1 (1.22–6.70) for women under
45 years (n=1176) compared with 3.30 mg·L−1 (1.50–6.00) in women older than 55 years (n=2873);
p=0.80.
A diagnosis of insomnia was reported in 4.2% of the population (n=762), and CRP values tended to be
slightly lower in patients who reported insomnia (2.90 mg·L−1 (1.00–5.00)) compared with patients
without reported insomnia of 3.00 mg·L−1 (1.20–5.00) (n=17 358, 96%); p=0.10.
The association between hypoxic markers of OSA and CRP levels showed consistently a significant dose–
response relationship. The second and third tertiles of the ODI4 were associated with elevated CRP levels
by 0.68 mg·L−1 (0.55–0.81) for ODI 5.8–<40 events per h and 1.8 mg·L−1 (1.7–2.0) for ODI ⩾40 events
per h; both p<0.001, n=18 031, respectively (supplementary figures S2 and S3 and supplementary
table S5). Comparably, the second and third tertile of T90 were associated with marked increased levels of
10
15
20
p<0.01
5
0
C
R
P
 (
m
g
·L
–
1
)
No OSA Mild OSA Moderate OSA Severe OSA All
p<0.01 p<0.01 p<0.01 p<0.01
Women Men All patients
FIGURE 3 Boxplot of C-reactive protein (CRP) values according to sex and severity of obstructive sleep apnoea
(OSA). Box-and-whisker plots represent CRP levels (mg·L−1) across apnoea–hypopnoea index (AHI)-based OSA
severity categories (no OSA, mild, moderate, severe) and for the overall population (“All”). Boxes indicate the
interquartile range (IQR), with the horizontal line representing the median. Whiskers extend to the most
extreme data points within 1.5×IQR from the lower and upper quartiles. Values beyond this range are displayed
as individual points (outliers). Comparisons across OSA severity categories were assessed using Kruskal–Wallis
tests; brackets indicate statistically significant differences (p<0.01). CRP values are plotted on a linear scale to
preserve interpretability across the full range of observations.
https://doi.org/10.1183/23120541.00707-2025 6
ERJ OPEN RESEARCH ORIGINAL RESEARCH ARTICLE | L. GROTE ET AL.
 on January 26, 2026 at Turku University Library. Please see licensing information on first page for reuse rights.https://publications.ersnet.orgDownloaded from 
CRP by 0.57 mg·L−1 (0.41–0.73) for T90 0.8–<51.40 min and 1.9 mg·L−1 (1.8; 2.1) for T90⩾51.4 min;
both p<0.001, n=12 691, respectively (supplementary figures S4 and S5 and supplementary table S6). The
results show that the highest tertiles of these two measures of hypoxic burden had a numerically stronger
influence on CRP levels than the highest AHI category (1.8 and 1.9 mg·L−1 versus 0.58 mg·L−1).
Discussion
Our data derived from the large ESADA cohort demonstrate a robust dose–response relationship between
several OSA severity measures and CRP levels, independent of obesity and other frequent cardiometabolic
and pulmonary comorbidities. Furthermore, OSA tended to associate with elevated CRP concentration
more extensively in men than in women. This unique and large study supports the recognition of OSA as a
proinflammatory disease, in particular when OSA is associated with profound hypoxic burden.
A meta-analysis, including 50 studies in adults, showed significantly elevated CRP levels in OSA patients
compared with controls [16]. Interestingly, another meta-analysis including 96 studies in adults, concluded
different relationships depending on the type of CRP assessment. hsCRP was significantly higher
according to OSA severity compared with controls, whereas ls-CRP was not [28]. Our results, exceeding
the number of patients analysed in previous meta-analyses, are consistent with the reported findings of the
first meta-analysis. The collection and analysis of many confounders allowed the use of the IPTW-RA
method to estimate a causal effect on observational data. Even though IPTW-RA is a well-established
method to estimate the causal effect of a binary exposure, it is a novel approach to apply an extension of
this method for an ordinal exposure [29]. A doubly robust estimator such as IPTW-RA is a well-known
causal inference method for observational data, allowing us to account for measured confounders and
reducing any mis-specification in the model [30]. Indeed, the robust dose–response relationship between
OSA severity measures (event frequency and hypoxic load) and CRP levels established in our dataset
further strengthens the causality between OSA and systemic inflammation.
TABLE 2 Summary of the results of univariable and multivariable weighted linear regression
Characteristic Univariate analysis Multivariable analysis
β 95% CI p-value β 95% CI p-value
OSA severity <0.001 <0.001
No OSA Ref. Ref. Ref. Ref.
Mild OSA 0.22 0.04–0.39 0.015 0.10 −0.10–0.31 0.23
Moderate OSA 0.55 0.37–0.73 <0.001 0.28 0.07–0.49 0.002
Severe OSA 1.2 1.1–1.4 <0.001 0.60 0.39–0.80 <0.001
Age 0.01 0.00–0.01 0.005 −0.01 −0.01–0.00 <0.001
ESS score 0.03 0.02–0.04 <0.001 0.00 −0.01–0.01 0.84
BMI <0.001 <0.001
<25 kg·m−2 Ref. Ref. Ref. Ref.
25–<30 kg·m−2 0.51 0.35–0.68 <0.001 0.53 0.34–0.72 <0.001
30–<35 kg·m−2 1.3 1.2–1.5 <0.001 1.2 0.97–1.36 <0.001
⩾35 kg·m−2 3.1 2.9–3.2 <0.001 2.7 2.45–2.90 <0.001
Smoking 0.30 0.18–0.42 <0.001 0.30 0.16–0.44 <0.001
Sex male 0.77 −0.89– −0.66 <0.001 −0.75 −0.89– −0.61 <0.001
Diabetes mellitus 1.2 1.1–1.4 <0.001 0.55 0.35–0.76 <0.001
Left ventricular hypertrophy 2.0 1.5–2.5 <0.001 1.3 0.61–1.97 <0.001
Systemic hypertension 0.69 0.58–0.79 <0.001 0.09 −0.03–0.21 0.11
Ischaemic heart disease 0.50 0.31–0.69 <0.001 −0.02 −0.25–0.21 0.84
TIA or stroke 0.35 0.00–0.70 0.049 0.14 −0.22–0.49 0.43
Status post-myocardial infarction 0.32 −0.04–0.68 0.082 NA NA NA
Cardiac failure 1.0 0.74–1.3 <0.001 0.48 0.11–0.84 <0.001
Other CV comorbidities 0.10 −0.27–0.06 0.21 NA NA NA
COPD 1.2 0.98–1.4 <0.001 0.82 0.53–1.11 <0.001
Neurological disease 0.32 0.09–0.55 0.006 0.26 0.03–0.50 0.017
Psychiatric disease 0.33 0.16–0.50 <0.001 0.05 −0.13–0.23 0.53
Inflammatory disease 0.63 0.36–0.91 <0.001 0.59 0.26–0.92 <0.001
CI: confidence interval; OSA: obstructive sleep apnoea; NA: not applicable to variables that were not significant
in unvariable analyses and not considered for multivariable analysis; ESS: Epworth sleepiness scale; BMI: body
mass index; TIA: transient ischaemic attack; CV: cardiovascular. Ref.: reference category. “Other CV
comorbidities” are derived from free text and not captured by the categories above.
https://doi.org/10.1183/23120541.00707-2025 7
ERJ OPEN RESEARCH ORIGINAL RESEARCH ARTICLE | L. GROTE ET AL.
 on January 26, 2026 at Turku University Library. Please see licensing information on first page for reuse rights.https://publications.ersnet.orgDownloaded from 
The present study also showed that females had significantly higher CRP levels compared with males, in
agreement with other studies [31, 32]. The mean difference in the ESADA population was 0.77 mg·L−1,
whereas sex differences of 1.5 mg·L−1 [32] and 1.13 mg·L−1 [31] have been reported elsewhere. In the
latter study, the mean difference changed to 0.52 mg·L−1 after excluding females using oestrogens and
individuals with CRP>10 mg·L−1. Thus, the value found in our study has a consistent order of magnitude.
Concerning sex differences in inflammation due to OSA, GAINES et al. [33] found that males with OSA
had a greater proinflammatory profile. This finding is supported by our results demonstrating an
independent effect size of severe OSA on median CRP levels, which was 0.71 mg·L−1 in males and
0.49 mg·L−1 in women; this difference is apparently not influenced by menopausal status. The more
pronounced proinflammatory effect of OSA in males may be a result of a higher degree of hypoxia, a
longer history of OSA exposure and/or a higher frequency of respiratory events often reported in male
compared with female OSA patients. However, more epidemiological and mechanistic studies are needed
to better understand these findings.
The sensitivity analysis suggests that hypoxic burden, reflected as intermittent hypoxia in the ODI4 or as
sustained hypoxia in the T90 measure, has a very strong influence on CRP levels. This finding is in line
BMI >35 kg·m–2
Left ventricular hypertrophy
BMI 30–35 kg·m–2
COPD
Severe OSA
Inflammatory disease
Diabetes
BMI 25–30 kg·m–2 
Cardiac failure
Smoking
Moderate OSA
Neurological disease
TIA or stroke
Mild OSA
Hypertension
Psychiatric disease
ESS
Age
Ischaemic heart disease
Sex male
Estimate
0–1 1 2 3
FIGURE 4 Plot of the coefficient estimates of the weighted linear regression explaining C-reactive protein
values. BMI: body mass index; TIA: transient ischaemic attack; ESS: Epworth sleepiness scale; OSA: obstructive
sleep apnoea. Mild OSA is defined as AHI 5–<15 events per h; moderate OSA as AHI 15–<30 events per h; severe
OSA as AHI ⩾30 events per h.
TABLE 3 Subgroup analysis, summary of the weighted linear regression for males and females
Males; n=13 036 Females; n=5409
Estimate (95% CI) p-value Estimate (95% CI) p-value
OSA severity <0.001 <0.001
No OSA Ref. Ref.
Mild OSA 0.21 (−0.01–0.43) 0.057 −0.18 (−0.47–0.10) 0.21
Moderate OSA 0.33 (0.11–0.55) 0.003 0.24 (−0.06–0.54) 0.12
Severe OSA 0.71 (0.50–0.92) <0.001 0.49 (0.19–0.79) 0.001
CI: confidence interval; OSA: obstructive sleep apnoea; Ref.: reference category. The model was adjusted on the
entire set of covariables but only the coefficients of the variable of interest (i.e. OSA severity) are displayed on
the table.
https://doi.org/10.1183/23120541.00707-2025 8
ERJ OPEN RESEARCH ORIGINAL RESEARCH ARTICLE | L. GROTE ET AL.
 on January 26, 2026 at Turku University Library. Please see licensing information on first page for reuse rights.https://publications.ersnet.orgDownloaded from 
with the large body of evidence linking intermittent hypoxia to inflammatory processes [3–5, 12]. The
consistency and robustness of this finding further strengthen the overall study results suggesting that OSA
and its associated hypoxic burden has a significant and independent contribution to low-degree
inflammation.
In our study, morbid obesity was the strongest predictor of CRP levels supporting the importance of
elevated BMI as a risk factor for low-grade inflammation and overall CV risk in the OSA population [34].
In this context, the therapy of comorbid obesity is critical to reduce inflammation, in particular in younger
individuals. Indeed, two recent randomised trials demonstrated that both lifestyle interventions [35] and
pharmacological therapy [36] targeting obesity in OSA cause a significant reduction in inflammatory
markers such as CRP.
Strengths and limitations of this study
This study has several strengths. First, it is by far the largest study analysing the link between CRP as a
biomarker of systemic inflammation and OSA, with a total of 18 445 patients analysed. Second, the studied
population is representative of the European population, and not only of one country or region, which
increases the generalisability of the results. In addition, the ESADA cohort has an extended reporting of
comorbidities based on detailed data from the patient’s medical record, which is important for the
confounder assessment in this study [19]. Third, appropriate and advanced causal inference methods to
estimate the causal effect of OSA on CRP levels have been applied in this study for the first time. Fourth,
multiple sensitivity analyses all show highly conclusive results linking different measures of OSA severity,
in particular hypoxia during sleep, to increased inflammation, both in male and female patients.
Altogether, our study contributes robust data and significant novelty to the question of whether untreated
OSA causes systemic low-grade inflammation.
However, several study limitations need to be recognised. Considerable variability in CRP measurements
might have been introduced due to methodological differences between the different centres as well as over
the study period over almost one and a half decade. Indeed, there is no annotation in the database on the
specific use of the analysis kit and the accreditation of the analysis method in the different hospital clinical
laboratories. There is also variability in the measurement of the AHI value in the ESADA [20], as patients
were diagnosed with either home sleep apnoea testing (polygraphy or polysomnography) and the manual
analysis was performed by multiple scorers over time. The type of sleep test used could not be included in
the model because it would have violated the positivity assumption, as some centres only use one type of
sleep test procedure. We have overcome this study limitation by the use of “centre” and “year” as random
effects in our analysis, implying that scoring and assessment methods were stable over at least 1 year. The
robustness of our statistical models despite a rather highly variable CRP analysis methodology may argue
for an even stronger true association between OSA and CRP. Another aspect is the fact that a majority of
patients in our study are White, a fact that could limit the generalisability of the results to other ethnic
groups, as significant race differences exist in the population distribution of CRP [32, 37]. The exact
menopausal status in women included in our study has not been captured in the database. However, we
applied the separation of age groups <45 years and >55 years as an established method to study the
systematic effect of menopausal status. Evidence suggests that OSA severity is significantly more
associated with oxidative stress in elderly females when compared with elderly males [38]. However, our
sensitivity analysis suggests no significant difference in CRP levels in pre- and post-menopausal women
with OSA. Finally, this is an observational study and not a randomised controlled trial (RCT). CRP data
were available in 54% of the patients and clinical data were comparable with the remaining individuals
without CRP data, suggesting a significant bias to be unlikely. Some unmeasured or even unknown
confounders could still influence the results and hence violate the conditional exchangeability assumption
for the use of IPTW-RA. However, RCTs have the limitation of inclusion and exclusion of certain patient
groups reducing the generalisability of the findings. The use of appropriate causal inference statistical
methods appears then to be a favourable solution for such a significant pre-selection problem, even if some
limitations remain.
Study implications and future research questions
Our data implicates the need for further future studies on this topic. An interventional study to address the
impact of continuous positive airway pressure (CPAP) therapy on CRP concentration during at least
6 months would shed further light on associations and causality. CPAP is known to substantially reduce
intermittent hypoxia and sympathetic activation in OSA. Indeed, a recent meta-analysis including
randomised trials suggest that CPAP treatment causes beneficial effects on systemic inflammation and CRP
levels, but the overall level of evidence was still considered to be limited [39]. There is also evidence
suggesting that the ODI may be a better predictor than AHI to assess OSA severity [7, 40]. It would be
https://doi.org/10.1183/23120541.00707-2025 9
ERJ OPEN RESEARCH ORIGINAL RESEARCH ARTICLE | L. GROTE ET AL.
 on January 26, 2026 at Turku University Library. Please see licensing information on first page for reuse rights.https://publications.ersnet.orgDownloaded from 
also interesting to test the impact of OSA severity metrics such as hypoxic burden [41] or arousal
burden [42] on CRP as a marker of low-degree systemic inflammation. Finally, the important question of
whether CRP can serve as a potential risk marker for future CV events in OSA, alone or combined with
the above-mentioned markers of OSA severity, needs to be addressed in prospective outcome studies
within the ESADA and other large-scale cohorts.
Conclusion
Our study showed a dose–response relationship between OSA severity and CRP levels as a biomarker of
systemic inflammation in a large prospective cohort using causal inference statistical methods, suggesting a
strong relationship between OSA and low-grade inflammation as a potential risk factor for cardiovascular
disease. We identified a sex difference, with women having higher overall CRP levels, but men showing a
more robust proinflammatory profile related to conventionally determined OSA severity. Obesity and
cardiopulmonary comorbidities further increased CRP blood levels in this patient group. Further research is
warranted to explore the potential relevance of measuring CRP as prognostic indicator for CV events in
OSA.
Acknowledgements: The ESADA network was supported by the European Union COST action B26 (2005–2009). In
addition, the European Respiratory Society (ERS) has funded ESADA as a Clinical Research Collaboration (2015–
ongoing). The ResMed Foundation and the Philips Respironics Foundation have provided unrestricted seeding
grants for establishment of the database in 2007 and 2011. The ESADA has a scientific collaboration with Bayer AG
(2018–2022). 24 ESADA centres participate in the EU Horizon 2020-funded Sleep Revolution project (965417).
Provenance: Submitted article, peer reviewed.
The ESADA study group: Alexandroupolis, Greece: P. Steiropoulos, Sleep Unit, Department of Pneumonology,
Democritus University of Thrace, Alexandroupolis, Greece. Antwerp, Belgium: J. Verbraecken, Multidisciplinary
Sleep Disorders Centre, Antwerp University Hospital and University of Antwerp, Antwerp, Belgium; E. Petiet,
Multidisciplinary Sleep Disorders Centre, Antwerp University Hospital and University of Antwerp, Antwerp, Belgium.
Athens, Greece: G. Trakada, Pulmonary Medicine, National and Kapodistrian University of Athens, Athens, Greece.
Berlin, Germany: I. Fietze, Schlafmedizinisches Zentrum, Charité – Universitätsmedizin Berlin, Berlin, Germany;
T. Penzel, Schlafmedizinisches Zentrum, Charité – Universitätsmedizin Berlin, Germany. Brno, Czech Republic:
O. Ludka, Department of Cardiology, University Hospital Brno and International Clinical Research Center, St Ann’s
University Hospital, Brno, Czech Republic. Crete, Greece: I. Bouloukaki, Sleep Disorders Unit, Department of
Respiratory Medicine, Medical School, University of Crete, Greece; S. Schiza, Sleep Disorders Unit, Department
of Respiratory Medicine, Medical School, University of Crete, Greece. Dublin, Ireland: W.T. McNicholas,
Department of Respiratory Medicine, St Vincent’s University Hospital, Dublin, Ireland; S. Ryan, Pulmonary and
Sleep Disorders Unit, St Vincent’s University Hospital, Dublin, Ireland. Edinburgh, UK: R.L. Riha, Department of
Sleep Medicine, Royal Infirmary Edinburgh, UK. Førde, Norway: J.A. Kvamme, Sleep Laboratory, ENT Department,
Førde Central Hospital, Førde, Norway. Gothenburg, Sweden: L. Grote, Sleep Disorders Center, Pulmonary
Department, Sahlgrenska University Hospital, and Center of Sleep and Wake Disorders, Sahlgrenska Academy,
Gothenburg University, Göteborg, Sweden; J. Hedner, Sleep Disorders Center, Pulmonary Department, Sahlgrenska
University Hospital, and Center of Sleep and Wake Disorders, Sahlgrenska Academy, Gothenburg University,
Göteborg, Sweden; D. Zou, Sleep Disorders Center, Pulmonary Department, Sahlgrenska University Hospital, and
Center of Sleep and Wake Disorders, Sahlgrenska Academy, Gothenburg University, Göteborg, Sweden. Ghent,
Belgium: K. Hertegonne, Department of Respiratory Medicine, Ghent University Hospital, and Department of
Internal Medicine and Pediatrics, Faculty of Medicine and Health Sciences, Ghent University, Ghent, Belgium;
D. Pevernagie, Department of Respiratory Medicine, Ghent University Hospital, and Department of Internal
Medicine and Pediatrics, Faculty of Medicine and Health Sciences, Ghent University, Ghent, Belgium. Grenoble,
France: S. Bailly, Université Grenoble Alpes, INSERM HP2 (U1300) and Grenoble University Hospital, Grenoble,
France; J.L. Pépin, Université Grenoble Alpes, INSERM HP2 (U1300) and Grenoble University Hospital, Grenoble,
France; R. Tamisier, Université Grenoble Alpes, INSERM HP2 (U1300) and Grenoble University Hospital, Grenoble,
France. Hamburg, Germany: H. Hein, Sleep Disorders Center, St Adolf Stift, Reinbeck, Germany. Izmir, Turkey:
O.K. Basoglu, Department of Respiratory Medicine, Ege University, Izmir, Turkey; M.S. Tasbakan, Department of
Respiratory Medicine, Ege University, Izmir, Turkey. Klecany, Czech Republic: J. Buskova, Department of Sleep
Medicine, National Institute of Mental Health, Klecany, Czech Republic. Kosice, Slovakia: P. Joppa, Department of
Respiratory Medicine and Tuberculosis, Faculty of Medicine, P.J. Safarik University and L. Pasteur University
Hospital, Kosice, Slovakia. Lisbon, Portugal: R. Staats, Department of Respiratory Medicine, Hospital de Santa
Maria, Lisbon, Portugal. Leuven, Belgium: D. Testelmans, Sleep Disorders Centre, University Hospital Gasthuisberg,
Leuven, Belgium; A. Kalkanis, Sleep Disorders Centre, University Hospital Gasthuisberg, Leuven, Belgium. Mainz,
Germany: H. Gouveris, ENT department at Mainz University Hospital, Mainz, Germany; K. Ludwig, ENT department
https://doi.org/10.1183/23120541.00707-2025 10
ERJ OPEN RESEARCH ORIGINAL RESEARCH ARTICLE | L. GROTE ET AL.
 on January 26, 2026 at Turku University Library. Please see licensing information on first page for reuse rights.https://publications.ersnet.orgDownloaded from 
at Mainz University Hospital, Mainz, Germany. Milan, Italy: C. Lombardi, IstitutoAuxologicoItaliano, IRCCS,
Department of Cardiovascular, Neural and Metabolic Sciences, St Luke Hospital, and Department of Medicine and
Surgery, University of Milano-Bicocca, Milan, Italy. G. Parati, IstitutoAuxologicoItaliano, IRCCS, Department of
Cardiovascular, Neural and Metabolic Sciences, St Luke Hospital, and Department of Medicine and Surgery;
University of Milano-Bicocca, Milan, Italy. Palermo, Italy: M.R. Bonsignore, PROMISE Dept, University of Palermo,
Palermo, Italy. Pavia, Italy: F. Fanfulla, Unità Operativa di Medicina del Sonno,Istituto Scientifico di Pavia and
Montescano IRCCS, Pavia, Italy. Porto, Portugal: M. Drummond, Pulmonology Department Hospital São João,
Medicine Faculty of Porto University, Porto, Portugal; M. van Zeller, Pulmonology Department Hospital São
João, Medicine Faculty of Porto University, Porto, Portugal. Solingen, Germany: S. Matthes, Respiratory Research
Institute, Pulmonary Clinic, Solingen, Germany; W. Randerath, Sleep Disorders Centre, Pulmonary Clinic, Solingen,
Germany; M. Treml, Respiratory Research Institute, Pulmonary Clinic, Solingen, Germany. Split, Croatia: Z. Dogas,
Sleep Medicine Center, Department of Neuroscience, University of Split School of Medicine, Split, Croatia;
R. Pecotic, Sleep Medicine Center, Department of Neuroscience, University of Split School of Medicine, Split,
Croatia. Thessaloniki, Greece: A. Pataka, Respiratory Failure Unit, G. Papanikolaou Hospital, Aristotle University of
Thessaloniki, Greece. Timisoara, Rumania: S. Mihaicuta, Center for Research and Innovation in Precision Medicine
and Pharmacy, “Victor Babes” University of Medicine and Pharmacy, Timisoara, Romania. Turku, Finland:
U. Anttalainen, Division of Medicine, Department of Pulmonary Diseases, Turku University Hospital and Sleep
Research Centre, Department of Pulmonary Diseases and Clinical Allergology, University of Turku, Finland;
T. Saaresranta, Division of Medicine, Department of Pulmonary Diseases, Turku University Hospital and Sleep
Research Centre, Department of Pulmonary Diseases and Clinical Allergology, University of Turku, Finland. Warsaw,
Poland: P. Sliwinski, 2nd Department of Respiratory Medicine, Institute of Tuberculosis and Lung Diseases,
Warsaw, Poland.
Ethics statement: Patients needed to provide oral and/or written informed consent to be included in the database.
The ESADA database has been reviewed by the local independent Ethics Review Boards at each participating
centre.
Conflict of interest: L. Grote reports grants from Swedish Heart and Lung Foundation, Regional research support
(ALF) and European Union Horizon 2020 grant “Sleep Revolution”; royalties or licenses with Desitin; consulting
fees from Onera; payment or honoraria for lectures, presentations, speakers’ bureaus, manuscript writing or
educational events for ResMed, Lundbeck, Löwenstein, AstraZeneca and ERS courses in 2024; leadership roles with
the national guideline committee for OSA in adults, national quality registry for obstructive sleep apnoea in adults
(SESAR) and ERS Clinical Research Collaboration ESADA; and was an ERS Representative for Assembly 4 in 2020–
2023. H. Gouveris reports grants to their institution (UMC Mainz) from Inspire Medical Systems Inc.; Honoraria
received for lectures from Inspire Medical Systems, Inc.; and participation on Advisory Board for Bioprojet
Deutschland. L. Lethuillier has nothing to disclose. J. Verbraecken reports grants from AirLiquide, Bioprojet,
Desitin, Epilog, Inspire, Medical Systems, Löwenstein Medical, Mediq, Natus Micromed OSG, Nyxoah, Philips,
ProSomnus, ResMed, Sefam, SomnoMed, SOS Oxygène, Tilman, Total Care, Vivisol, Westfalen Medical, Withings,
Zoll Itamar, Bioprojet and Epilog; payment or honoraria for lectures, presentations, speakers’ bureaus, manuscript
writing or educational events for Atos Medical, DEME, Bioprojet, Inspire Medical Systems, Idorsia, SD Worx,
Vlaamse Gemeenschap, Vlerick, Total Care, Azelis and AuroBindo; support for attending meetings and/or travel
from Bioprojet; and is a current member of the ERJ Open Research editorial board. O.K. Basoglu has nothing to
disclose. S. Schiza has nothing to disclose. O. Ludka has nothing to disclose. S. Ryan reports grants from
NovoNordisk and Fitbit; and consulting fees from Irish Rugby Football Union. P. Joppa reports grants from VEGA
1/0393/22 from the Grant Agency of The Ministry of Education of the Slovak Republic. F. Francesco reports
payment or honoraria for lectures, presentations, speakers’ bureaus, manuscript writing or educational events
from GSK, Bioprojet, Idorsia and Sapio Life; and support for attending meetings and/or travel from Vivisol.
S. Mihaicuta has nothing to disclose. T. Saaresranta reports support for the present manuscript from Finnish
Antituberculosis Association Foundation, Tampere Tuberculosis Foundation, The Research Foundation of the
Pulmonary Diseases, Jalmari and Rauha Ahokas Foundation, and Governmental VTR grant number 13542; payment
or honoraria for lectures, presentations, speakers’ bureaus, manuscript writing or educational events from ResMed,
The Finnish Medical Society Duodecim, Chiesi and Boehringer Ingelheim; and a leadership role with Finnish
Current Care Guidelines for Adult Sleep Apnoea. P. Sliwinski has nothing to disclose. J. Hedner reports grants from
Desitin; consulting fees from Somnomed; participation on an Advisory Board for SomnoMed; and a leadership role
with Patient Association, ERS and the Swedish Sleep Apnea Quality Registry (SESAR). J.L. Pépin reports grants
from UGA e-health chair and Sleep health-AI MIAI @ university Grenoble Alpes (ANR-19-P3IA-0003). S. Bailly has
nothing to disclose.
Support statement: L. Grote is supported by the Swedish Heart and Lung Foundation (HLF20240848), the EU
Horizon 2020 grant “Sleep Revolution” (965417), and grants from the Swedish state under the agreement between
the Swedish government and the county councils, the ALF-agreement (ALFGBG1006211). J.L. Pépin, S. Bailly and
https://doi.org/10.1183/23120541.00707-2025 11
ERJ OPEN RESEARCH ORIGINAL RESEARCH ARTICLE | L. GROTE ET AL.
 on January 26, 2026 at Turku University Library. Please see licensing information on first page for reuse rights.https://publications.ersnet.orgDownloaded from 
L. Lethuillier are supported by the French National Research Agency in the framework of the “Investissements
d’avenir” programme (ANR-15-IDEX-02) and the “e-health and integrated care and trajectories medicine and MIAI
artificial intelligence” (ANR-19-P3IA-0003). Chairs of excellence from the Grenoble Alpes University Foundation.
T. Saaresranta has received grants paid to the Turku University Hospital from the Finnish Antituberculosis
Association Foundation, Tampere Tuberculosis Foundation, The Research Foundation of the Pulmonary Diseases,
Jalmari and Rauha Ahokas Foundation, and Governmental VTR grant number 13542. P. Joppa reported VEGA
1/0393/22 grants from the Grant Agency of The Ministry of Education of the Slovak Republic.
References
1 Yaggi HK, Concato J, Kernan WN, et al. Obstructive sleep apnea as a risk factor for stroke and death. N Engl
J Med 2005; 353: 2034–2041.
2 Gottlieb DJ, Yenokyan G, Newman AB, et al. Prospective study of obstructive sleep apnea and incident
coronary heart disease and heart failure: the Sleep Heart Health study. Circulation 2010; 122: 352–360.
3 Javaheri S, Javaheri S, Somers VK, et al. Interactions of obstructive sleep apnea with the pathophysiology of
cardiovascular disease, part 1: JACC state-of-the-art review. J Am Coll Cardiol 2024; 84: 1208–1223.
4 Ryan S, Cummins EP, Farre R, et al. Understanding the pathophysiological mechanisms of cardiometabolic
complications in obstructive sleep apnoea: towards personalised treatment approaches. Eur Respir J 2020;
56: 1902295.
5 Ryan S, Taylor CT, McNicholas WT. Selective activation of inflammatory pathways by intermittent hypoxia in
obstructive sleep apnea syndrome. Circulation 2005; 112: 2660–2667.
6 Lévy P, Kohler M, McNicholas WT, et al. Obstructive sleep apnoea syndrome. Nat Rev Dis Primers 2015; 1:
15015.
7 Bonsignore MR, ESADA Study Group. Adaptive responses to chronic intermittent hypoxia: contributions from
the European Sleep Apnoea Database (ESADA) Cohort. J Physiol 2023; 601: 5467–5480.
8 Willerson JT, Ridker PM. Inflammation as a cardiovascular risk factor. Circulation 2004; 109 21 Suppl 1),
II2–I10.
9 Yokoe T, Minoguchi K, Matsuo H, et al. Elevated levels of C-reactive protein and interleukin-6 in patients with
obstructive sleep apnea syndrome are decreased by nasal continuous positive airway pressure. Circulation
2003; 107: 1129–1134.
10 Drager LF, Lopes HF, Maki-Nunes C, et al. The impact of obstructive sleep apnea on metabolic and
inflammatory markers in consecutive patients with metabolic syndrome. PLoS ONE 2010; 5: e12065.
11 Shah A, Mukherjee S, McArdle N, et al. Circulating C-reactive protein levels in patients with suspected
obstructive sleep apnea. J Clin Sleep Med 2022; 18: 993–1001.
12 Ryan S, Nolan GM, Hannigan E, et al. Cardiovascular risk markers in obstructive sleep apnoea syndrome and
correlation with obesity. Thorax 2007; 62: 509–514.
13 Sharma SK, Mishra HK, Sharma H, et al. Obesity, and not obstructive sleep apnea, is responsible for
increased serum hs-CRP levels in patients with sleep-disordered breathing in Delhi. Sleep Med 2008; 9:
149–156.
14 Korkmaz M, Korkmaz H, Küçüker F, et al. Evaluation of the association of sleep apnea-related systemic
inflammation with CRP, ESR, and neutrophil-to-lymphocyte ratio. Med Sci Monit 2015; 21: 477–481.
15 Li K, Wei P, Qin Y, et al. Is C-reactive protein a marker of obstructive sleep apnea? A meta-analysis. Medicine
(Baltimore) 2017; 96: e6850.
16 Nadeem R, Molnar J, Madbouly EM, et al. Serum inflammatory markers in obstructive sleep apnea:
a meta-analysis. J Clin Sleep Med 2013; 9: 1003–1012.
17 Sproston NR, Ashworth JJ. Role of C-reactive protein at sites of inflammation and infection. Front Immunol
2018; 9: 754.
18 Yeghiazarians Y, Jneid H, Tietjens JR, et al. Obstructive sleep apnea and cardiovascular disease: a scientific
statement from the American Heart Association. Circulation 2021; 144: e56–e67.
19 Hedner J, Grote L, Bonsignore M, et al. The European Sleep Apnoea Database (ESADA): report from 22
European sleep laboratories. Eur Respir J 2011; 38: 635–642.
20 Escourrou P, Grote L, Penzel T, et al. The diagnostic method has a strong influence on classification of
obstructive sleep apnea. J Sleep Res 2015; 24: 730–738.
21 Mac Giollabhui N, Ellman LM, Coe CL, et al. To exclude or not to exclude: Considerations and
recommendations for C-reactive protein values higher than 10 mg/L. Brain Behav Immun 2020; 87: 898–900.
22 Sterne JA, White IR, Carlin JB, et al. Multiple imputation for missing data in epidemiological and clinical
research: potential and pitfalls. BMJ 2009; 338: b2393.
23 Etminan M, Collins GS, Mansournia MA. Using causal diagrams to improve the design and interpretation of
medical research. Chest 2020; 158: S21–S28.
24 Brookhart MA, Schneeweiss S, Rothman KJ, et al. Variable selection for propensity score models. Am
J Epidemiol 2006; 163: 1149–1156.
https://doi.org/10.1183/23120541.00707-2025 12
ERJ OPEN RESEARCH ORIGINAL RESEARCH ARTICLE | L. GROTE ET AL.
 on January 26, 2026 at Turku University Library. Please see licensing information on first page for reuse rights.https://publications.ersnet.orgDownloaded from 
25 Rodrigues D, Kreif N, Lawrence-Jones A, et al. Reflection on modern methods: constructing directed acyclic
graphs (DAGs) with domain experts for health services research. Int J Epidemiol 2022; 51: 1339–1348.
26 Chu R, Thabane L, Ma J, et al. Comparing methods to estimate treatment effects on a continuous outcome
in multicentre randomized controlled trials: a simulation study. BMC Med Res Methodol 2011; 11: 21.
27 Austin PC. Variance estimation when using inverse probability of treatment weighting (IPTW) with survival
analysis. Stat Med 2016; 35: 5642–5655.
28 Imani MM, Sadeghi M, Farokhzadeh F, et al. Evaluation of blood levels of C-reactive protein marker in
obstructive sleep apnea: a systematic review, meta-analysis and meta-regression. Life (Basel) 2021; 11: 362.
29 Yoshida K, Solomon DH, Haneuse S, et al. Multinomial extension of propensity score trimming methods: a
simulation study. Am J Epidemiol 2019; 188: 609–616.
30 Bettega F, Leyrat C, Tamisier R, et al. Application of inverse-probability-of-treatment weighting to estimate
the effect of daytime sleepiness in patients with obstructive sleep apnea. Ann Am Thorac Soc 2022; 19:
1570–1580.
31 Lakoski SG, Cushman M, Criqui M, et al. Gender and C-reactive protein: data from the Multiethnic Study of
Atherosclerosis (MESA) cohort. Am Heart J 2006; 152: 593–598.
32 Khera A, McGuire DK, Murphy SA, et al. Race and gender differences in C-reactive protein levels. J Am
CollCardiol 2005; 46: 464–469.
33 Gaines J, Vgontzas AN, Fernandez-Mendoza J, et al. Gender differences in the association of sleep apnea and
inflammation. Brain Behav Immun 2015; 47: 211–217.
34 Ryan S. Adipose tissue inflammation by intermittent hypoxia: mechanistic link between obstructive sleep
apnoea and metabolic dysfunction. J Physiol 2017; 595: 2423–2430.
35 Chirinos JA, Gurubhagavatula I, Teff K, et al. CPAP, weight loss, or both for obstructive sleep apnea. N Engl
J Med 2014; 370: 2265–2275.
36 Malhotra A, Grunstein RR, Fietze I, et al. Tirzepatide for the treatment of obstructive sleep apnea and obesity.
N Engl J Med 2024; 391: 1193–1205.
37 Wener MH, Daum PR, McQuillan GM. The influence of age, sex, and race on the upper reference limit of
serum C-reactive protein concentration. J Rheumatol 2000; 27: 2351–2359.
38 Ahiawodzi PD, Kerber RA, Taylor KC, et al. Sleep-disordered breathing is associated with higher
carboxymethyllysine level in elderly women but not elderly men in the cardiovascular health study.
Biomarkers 2017; 22: 361–366.
39 Zhu Q, Luo Q, Wang Z, et al. Effects of continuous positive airway pressure therapy on inflammatory markers
in patients with obstructive sleep apnea: a meta-analysis of randomized controlled trials. Sleep Breath 2025;
29: 182.
40 Thunström E, Glantz H, Fu M, et al. Increased inflammatory activity in nonobese patients with coronary
artery disease and obstructive sleep apnea. Sleep 2015; 38: 463–471.
41 Azarbarzin A, Sands SA, Stone KL, et al. The hypoxic burden of sleep apnoea predicts cardiovascular
disease-related mortality: the Osteoporotic Fractures in Men Study and the Sleep Heart Health Study.
Eur Heart J 2019; 40: 1149–1157.
42 Malatantis-Ewert S, Bahr K, Ding H, et al. A novel quantitative arousal-associated EEG-metric to predict
severity of respiratory distress in obstructive sleep apnea patients. Front Physiol 2022; 13: 885270.
https://doi.org/10.1183/23120541.00707-2025 13
ERJ OPEN RESEARCH ORIGINAL RESEARCH ARTICLE | L. GROTE ET AL.
 on January 26, 2026 at Turku University Library. Please see licensing information on first page for reuse rights.https://publications.ersnet.orgDownloaded from