Troubles respiratoires du sommeil au-delà de la somnolence diurne : dépistage par le STOP-BANG dans un centre cardiologique de référence d’Afrique de l’Ouest.

BC BOKA¹, F SALL², LD FLAINDE¹, N MOBIO³, J MIDAGO¹, F TRAORɹ, Y NK N’GORAN¹, KE KRAMOH¹.

 

RESUME

 

Introduction. L’hypersomnolence diurne représente un phénotype fréquemment observé dans les troubles respiratoires du sommeil, mais la plainte subjective de somnolence et la sévérité des TRS correspondent à des axes cliniques en grande partie dissociables. S’en remettre uniquement à la somnolence pour indiquer des explorations objectives expose dès lors au risque d’ignorer des présentations cliniquement significatives. Nous avons caractérisé les TRS en stratifiant selon le statut de somnolence et évalué la valeur discriminante du STOP-BANG pour détecter les formes sévères chez des adultes non somnolents.

Méthodes. Dans cette étude rétrospective transversale, nous avons revu les dossiers d’adultes adressés pour une polygraphie ventilatoire nocturne à l’Institut de Cardiologie d’Abidjan, en Côte d’Ivoire, entre janvier 2023 et décembre 2025. L’analyse principale a retenu les participants âgés de 18 ans ou plus disposant à la fois d’une polygraphie interprétable pour la sévérité et d’un score valide à l’échelle de somnolence d’Epworth (ESS). La somnolence diurne excessive était définie par un score ESS > 10 ; les participants ayant un score ESS ≤ 10 étaient considérés comme non somnolents. Les performances du STOP-BANG ont été évaluées chez les adultes non somnolents ayant un score valide compris entre 0 et 8.

Résultats. Parmi les 538 adultes inclus dans l'analyse principale, 497 (92,4 %) présentaient un TRS, 407 (75,7 %) une forme modérée à sévère et 259 (48,1 %) une forme sévère. Au total, 353/538 (65,6 %) étaient non somnolents. Une forme sévère était observée chez 160/353 (45,3 %) adultes non somnolents et 99/185 (53,5 %) adultes somnolents (RR 1,18 ; IC à 95 % 0,99–1,41 ; p = 0,084). Les adultes non somnolents représentaient 65,6 % de la cohorte, et 61,8 % des formes sévères étaient observées dans ce groupe. La corrélation entre l’ESS et la sévérité polygraphique était faible (ρ = 0,142 ; p = 0,001). Chez les 349 adultes non somnolents ayant un score STOP-BANG valide, les aires sous la courbe ROC (AUC) étaient de 0,816 pour la présence d’un TRS, de 0,736 pour les formes modérées à sévères et de 0,749 pour les formes sévères. Au seuil STOP-BANG ≥ 3, la sensibilité pour les formes sévères était de 91,8 % (IC à 95 % : 86,3–95,5) et la valeur prédictive négative de 86,0 % (IC à 95 % : 77,3–92,3).

Conclusion. Dans cette cohorte sélectionnée en milieu cardiologique, la somnolence diurne subjective apportait peu d’éléments pour distinguer les différents niveaux de sévérité polygraphique des TRS. L’ESS ne devrait donc pas être utilisée seule pour décider de la réalisation d’une exploration objective du sommeil. Chez les adultes non somnolents, le STOP-BANG présentait une sensibilité élevée pour les formes sévères et pourrait aider à orienter les explorations du sommeil lorsque les ressources diagnostiques sont limitées.

 

 

MOTS CLES 

Apnée obstructive du sommeil ; troubles respiratoires du sommeil ; somnolence ; STOP-BANG ; maladies cardiovasculaires ; Côte d’Ivoire.

 

 

SUMMARY

 

  

Introduction. Excessive daytime sleepiness is frequently observed in sleep-disordered breathing (SDB), yet how sleepy a person feels does not always match the underlying severity of SDB. Using sleepiness alone to guide who receives objective testing can therefore miss clinically significant cases. We analyzed SDB patterns by daytime sleepiness status and evaluated whether the STOP-BANG questionnaire can detect severe SDB in adults who do not report excessive sleepiness.

Methods. In this retrospective cross-sectional study, we reviewed records from adults referred for overnight respiratory polygraphy at the Institut de Cardiologie d'Abidjan, Côte d'Ivoire, between January 2023 and December 2025. The primary analysis included participants aged 18 years or older with both an interpretable polygraphic severity classification and a valid Epworth Sleepiness Scale (ESS) score. Excessive daytime sleepiness was defined as ESS > 10, while those with ESS ≤ 10 were considered non-sleepy. STOP-BANG performance was evaluated among non-sleepy adults with valid scores from 0 to 8.

Results. Among 538 adults in the primary analysis, 497 (92.4%) had SDB, 407 (75.7%) had moderate-to-severe SDB, and 259 (48.1%) had severe SDB. Overall, 353/538 (65.6%) were non-sleepy. Severe SDB occurred in 160/353 (45.3%) non-sleepy adults and 99/185 (53.5%) sleepy adults (RR 1.18, 95% CI 0.99–1.41; p = 0.084). Non-sleepy adults accounted for 65.6% of the cohort, and 61.8% of severe SDB cases occurred in this group. ESS correlated only weakly with polygraphic severity (Spearman ρ = 0.142; p = 0.001). Among 349 non-sleepy adults with valid STOP-BANG scores, the areas under the receiver operating characteristic curves (AUCs) were 0.816 for any SDB, 0.736 for moderate-to-severe SDB, and 0.749 for severe SDB. At STOP-BANG ≥ 3, sensitivity for severe SDB was 91.8% (95% CI 86.3–95.5) and negative predictive value (NPV) was 86.0% (95% CI 77.3–92.3).

Conclusion. Subjective daytime sleepiness provided limited discriminatory information about polygraphic SDB severity in this selected cardiovascular referral cohort. ESS should therefore not be used alone to decide who should undergo objective sleep evaluation. STOP-BANG showed high sensitivity for severe SDB among non-sleepy adults and may help guide further sleep testing where diagnostic resources are limited.

 

 

KEY WORDS 

 

Sleep Apnea, Obstructive; sleep-disordered breathing; sleepiness; STOP-BANG; cardiovascular disease; Côte d’Ivoire

 

1. Université Félix Houphouët-Boigny /Institut de Cardiologie d'Abidjan, Abidjan, Côte d'Ivoire

2. Université Alassane Ouattara, Institut de Cardiologie de Bouaké, Bouaké, Côte d'Ivoire

3. Université Félix Houphouët-Boigny / service de Pneumo-phtisiologie, CHU de Cocody, Abidjan, Côte d'Ivoire

 

 

Adresse pour correspondance

BOKA BC

  This email address is being protected from spambots. You need JavaScript enabled to view it.

 

 

 

 

 

INTRODUCTION

 

 

 

Obstructive sleep apnea (OSA) is the most common form of sleep-disordered breathing (SDB) and affects a large number of people worldwide, making it an important public health concern [1].

 

Repeated upper-airway obstruction can produce intermittent hypoxemia, sleep fragmentation, autonomic activation, oxidative stress, inflammation, and large fluctuations in intrathoracic pressure. These mechanisms provide biologically plausible links with hypertension, atrial fibrillation, heart failure, coronary disease, and stroke [2,3].

 

Clinical suspicion of sleep apnea remains strongly influenced by symptoms, particularly excessive daytime sleepiness. The Epworth Sleepiness Scale (ESS), developed by Johns, quantifies an individual's general propensity to doze in common daytime situations [4]. It is therefore a measure of subjective sleepiness rather than a direct measure of respiratory-event frequency or severity, and it does not reliably reflect the severity of OSA [5].

 

Several major cardiovascular trials have shown that clinically significant OSA can occur even in patients who report little or no excessive daytime sleepiness, including Barbé et al., RICCADSA, SAVE, and ISAACC [6–9].

STOP-BANG assesses eight clinical features associated with OSA risk: snoring, tiredness, observed apnea, high blood pressure, body mass index, age, neck circumference, and sex [10,11]. Previous studies have reported higher sensitivity for STOP-BANG than for the ESS in detecting clinically significant sleep-disordered breathing [12], and its usefulness as a screening tool has also been described in sub-Saharan African populations [13]. 

 

 

 

This is especially relevant in regions where access to sleep testing is limited. Comparable polygraphy studies from West Africa remain scarce. In Burkina Faso, a sleep-clinic study found OSA to be common in patients evaluated with respiratory polygraphy [14]. A low ESS score may seem reassuring, even when other clinical features still suggest OSA. Relying on sleepiness alone may therefore overlook patients who need objective sleep testing.

 

We therefore examined SDB according to daytime sleepiness status among adults evaluated at the Sleep Exploration Unit of the Institut de Cardiologie d'Abidjan. The primary objective was to quantify any, moderate-to-severe, and severe SDB among adults with and without excessive daytime sleepiness. The secondary objective was to assess the screening performance of STOP-BANG for severe SDB among non-sleepy adults. We hypothesized that ESS would show limited discrimination of polygraphic severity and that STOP-BANG would show useful screening performance among adults without excessive daytime sleepiness.

 

 

 

 

 

METHODS

 

 

 

Study design and setting

 

 

 

We conducted a retrospective cross-sectional study at the Sleep Exploration Unit of the Institut de Cardiologie d'Abidjan (ICA), Côte d'Ivoire. The ICA is a tertiary national cardiovascular referral center with a unit dedicated to the investigation of sleep-related breathing disorders. The study period extended from 1 January 2023 through 31 December 2025. We reported the study in line with the STROBE statement for observational research [15]. For the STOP-BANG diagnostic-accuracy analysis, we also followed STARD 2015 recommendations [16]. Completed STROBE and STARD checklists are provided as supplementary files.

 

 

 

Study population

 

 

 

The registry included 766 records evaluated during the study period. Of these, 740 had an interpretable polygraphic severity classification and 544 also had a valid ESS score between 0 and 24. The primary analysis included adults aged 18 years or older with both an interpretable polygraphy result and a valid ESS score. Five patients younger than 18 years and one patient with no recorded age were excluded, leaving 538 adults for analysis. No sample-size calculation was performed in advance because the study included all eligible routine-care records available in the registry during the predefined study period. Precision was therefore assessed through the 95% confidence intervals reported for the main effect estimates and diagnostic-accuracy measures.

 

Analyses involving ESS were based on complete cases. We did not impute missing ESS values because ESS itself was used to define the sleepiness groups. To see whether missing ESS data might have affected the study population, we compared the 544 records with a valid ESS score with the 196 patients who had interpretable polygraphy but no valid ESS score, using age, sex, body mass index, and polygraphic severity (Supplementary Table S1). This comparison was performed before the age restriction because its purpose was to assess ESS missingness across all interpretable polygraphies. During database quality control, 31 potential duplicate pairs (62 records) were identified because some patient names were highly similar (similarity range 0.90–0.98). As these pairs could not be confirmed as true duplicate examinations, we kept them in the primary analysis. We then repeated the analysis after excluding all records involved in a suspected duplicate pair to check whether this affected the results.

 

 

 

 

 

Respiratory polygraphy and disease classification

 

 

 

Overnight cardiorespiratory polygraphy was carried out as part of routine care using three portable systems available in the unit: Nox T3 (Nox Medical, Reykjavik, Iceland), BMC H2 Polypro (BMC Medical, Beijing, China), and Somté (Compumedics, Abbotsford, Australia). Overnight respiratory polygraphy included nasal airflow, thoracic and abdominal respiratory movements, oxygen saturation, heart rate, and body position. Recordings were interpreted as part of routine care by physicians trained in sleep medicine.

 

For this retrospective study, we used the original clinical reports and did not perform centralized rescoring. Disease severity and the predominant respiratory pattern were taken directly from these reports. The registry did not consistently identify the physician who reviewed each recording or indicate whether automated scoring had been used before physician review. Interobserver agreement could therefore not be evaluated.

 

As electroencephalography was not recorded, the apnea-hypopnea index (AHI) was expressed per hour of recording rather than per hour of confirmed sleep. Respiratory events were scored using AASM-based criteria, with hypopneas defined by an oxygen desaturation threshold of at least 3% [17]. The registry did not consistently document the exact version of the AASM scoring manual applied to each historical examination.

 

AHI was classified as normal at < 5 events/h, mild at 5–14.9 events/h, moderate at 15–29.9 events/h, and severe at ≥ 30 events/h. Three outcomes were considered in the main analysis: any SDB, moderate-to-severe SDB, and severe SDB. Because obstructive, central, and mixed breathing patterns were all represented in the registry, we used the broader term sleep-disordered breathing (SDB) for these analyses.

 

When available in the clinical record, the respiratory pattern was classified as obstructive, central, or mixed. Since STOP-BANG was developed specifically for screening obstructive sleep apnea, we also performed a sensitivity analysis limited to patients with a documented obstructive pattern. Patients with normal polygraphy were retained as controls.

 

 

 

Epworth Sleepiness Scale

 

 

 

The ESS contains eight items scored from 0 to 3, giving a total score from 0 to 24 [4,5]. Excessive daytime sleepiness was defined using the widely used threshold ESS > 10 (ESS ≥ 11); patients with ESS ≤ 10 were classified as non-sleepy. ESS was interpreted as a measure of subjective sleep propensity rather than as a diagnostic test for SDB.

 

 

 

STOP-BANG

 

 

 

STOP-BANG consists of eight dichotomous items: snoring, tiredness, observed apnea, high blood pressure, body mass index > 35 kg/m², age > 50 years, neck circumference > 40 cm, and male sex [10,11]. Each positive item added one point to the STOP-BANG score, giving a total score from 0 to 8. In routine care, the total score was recorded on clinical questionnaires or forms and later entered into the final study database. We used this recorded total for the analysis rather than reconstructing the score from the individual items. Because the Tiredness item overlaps with daytime sleepiness, a seven-item version excluding Tiredness would have been useful as a sensitivity analysis in the non-sleepy subgroup. This could not be implemented because the registry did not consistently retain responses for all eight items, and documentation of the original neck circumference measurement and the operational criteria used for the STOP-BANG hypertension item was incomplete. The separate registry variable used to describe documented hypertension in the cohort was not used to reconstruct the STOP-BANG hypertension item. As a result, the seven-item score could not be reconstructed with sufficient reliability. We prespecified a threshold of 3 or greater as the primary high-sensitivity cut point to minimize the likelihood of missing severe SDB [11]. During database checks, three STOP-BANG values were corrected after verification against the source records. One value had been entered outside the valid 0–8 range. After correction, no out-of-range STOP-BANG values remained; records without a usable score were retained as missing and excluded only from STOP-BANG analyses.

 

 

 

Statistical analysis

 

 

 

Continuous variables were described using mean ± standard deviation when their distributions were approximately normal; otherwise, median and interquartile range (IQR) were used. Categorical variables are presented as counts and percentages. For each of the three prespecified SDB outcomes, we compared sleepy and non-sleepy adults using Fisher’s exact test. We used the same test throughout so that the comparisons did not depend on large-sample approximations. Because these outcomes were common, we reported relative risks (RRs) with 95% confidence intervals (CIs).

 

We also calculated the proportion of adults with each SDB outcome who had an ESS score ≤10, together with Wilson 95% confidence intervals. The relationship between ESS score and polygraphic severity was assessed using Spearman correlation. To see whether missing ESS data might have introduced selection bias, we compared patients with and without a valid ESS score by age, sex, BMI, and polygraphic severity. Welch’s t-test was used for age and BMI, and chi-square tests for sex and polygraphic severity.

 

Among non-sleepy adults, STOP-BANG performance was evaluated using receiver operating characteristic (ROC) curves. Areas under the curve (AUCs) were reported with 95% confidence intervals obtained from 5,000 percentile bootstrap resamples. At the ≥ 3 threshold, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were reported with exact 95% Clopper-Pearson confidence intervals. Positive and negative likelihood ratios (LR+ and LR−) were also reported with 95% confidence intervals calculated on the logarithmic scale. For severe SDB, we also examined how diagnostic performance changed across thresholds from ≥ 2 to ≥ 6 (Supplementary Table S2).

 

Sensitivity analyses included one analysis restricted to patients with a documented obstructive phenotype and another excluding all records involved in suspected duplicate pairs. We also performed an exploratory complete-case modified Poisson regression with robust variance (n = 531) to estimate adjusted relative risks comparing ESS > 10 with ESS ≤ 10, while accounting for age, sex, and BMI. No formal adjustment for multiple testing was applied across the three related SDB outcomes, so p-values were interpreted together with effect estimates and confidence intervals. All tests were two-sided, with p < 0.05 considered statistically significant. Analyses were performed in Python 3.13.5 using SciPy 1.17.0, scikit-learn 1.8.0, statsmodels 0.14.6, and NumPy 2.3.5.

 

 

 

Ethical considerations

 

 

 

This retrospective study was conducted in accordance with the principles of the Declaration of Helsinki. The Medical and Scientific Directorate of the Institut de Cardiologie d'Abidjan approved the study and authorized the secondary use of routinely collected clinical data in anonymized form. No additional approval from a research ethics committee was obtained. All records were irreversibly anonymized before any analyses were conducted. The study involved no ancillary examinations or therapeutic interventions, no patient contact, and no research-specific collection of biological specimens.

 

 

 

 

 

 

 

RESULTS

 

Study population

 

 

 

The registry included 766 patients, of whom 740 had a polygraphic study with an interpretable severity classification. Among these, 544 had a valid ESS score. Five patients younger than 18 years and one patient with no recorded age were excluded from the adult primary analysis, leaving 538 adults (Figure 1). Of the 353 non-sleepy adults, four did not have a usable STOP-BANG score, leaving 349 for the STOP-BANG analyses.

 

In the adult primary cohort, mean age was 52.3 ± 12.0 years (range 19–92 years). Sex was available in 536 records, of which 327 (61.0%) were male. BMI was available in 533 records and averaged 31.1 ± 6.8 kg/m². Mean ESS was 9.3 ± 4.9. Hypertension was documented in 301 adults (55.9%), and heart failure in 36 (6.7%). Other conditions of interest—including diabetes, atrial fibrillation, coronary disease, and stroke—were not recorded consistently enough in the final study database to be reported reliably. Exact AHI values were available for 197 adults. In this subgroup, the mean AHI was 36.8 ± 24.2 events/h, and the median was 34 events/h (IQR 17–53). (Table 1)

 

Denominators vary because of missing values. Hypertension and heart failure were reported only when they were documented in the final study database. In the historical records, a missing entry did not always allow us to distinguish between true absence of the condition and missing information.

 

Among the 740 interpretable polygraphy studies, 544 had a valid ESS score and 196 (26.5%) did not. The two groups were similar in age (51.9 ± 12.6 vs 52.2 ± 12.7 years; p = 0.800), BMI (31.0 ± 6.9 vs 31.4 ± 7.1 kg/m²; p = 0.441), and proportion of men (60.9% vs 63.8%; p = 0.476). Polygraphic severity was also similar between groups (overall p = 0.654), with severe SDB present in 47.6% of records with a valid ESS and 44.4% of those without one. Further details are provided in Supplementary Table S1.

 

Among the 497 adults with SDB, the dominant respiratory pattern was documented in 489. Most cases were obstructive (419, 85.7%), while 34 (7.0%) were central and 36 (7.4%) were mixed (Figure 1). 

 

The source registry included 766 records, of which 740 had interpretable polygraphic severity and 544 had a valid ESS score. After excluding five patients younger than 18 years and one patient with no recorded age, 538 adults remained in the primary analysis. Among the 353 non-sleepy adults, 349 had a valid STOP-BANG score; four were excluded from STOP-BANG analyses because no usable score was available.

 

 

 

Daytime sleepiness and SDB severity

 

 

 

Using ESS > 10 to define excessive daytime sleepiness, 353/538 adults (65.6%) were non-sleepy and 185/538 (34.4%) were excessively sleepy. SDB was present in 319/353 (90.4%) non-sleepy adults, moderate-to-severe SDB in 255/353 (72.2%), and severe SDB in 160/353 (45.3%). (Table 2, Figure 2)

 

Relative risks (RRs) compare ESS > 10 with ESS ≤ 10 for the three prespecified binary outcomes. P-values were obtained using Fisher’s exact test, and 95% confidence intervals for RRs were calculated using the logarithmic method.

 

 

 

Sleepiness status among adults with SDB

 

 

 

Among 497 adults with SDB, 319 (64.2%; 95% CI 59.9–68.3) had ESS ≤ 10. Among 407 adults with moderate-to-severe SDB, 255 (62.7%; 95% CI 57.9–67.2) had ESS ≤ 10. Among the 259 adults with severe SDB, 160 (61.8%; 95% CI 55.7–67.5) were non-sleepy. This finding should be interpreted in the context of the overall cohort, where 65.6% of adults had an ESS score ≤10. (Table 3, Figure 3)

 

ESS ≤ 10 was present in 319 of 497 adults (64.2%) with any SDB, 255 of 407 adults (62.7%) with moderate-to-severe SDB, and 160 of 259 adults (61.8%) with severe SDB.

 

 

 

Relationship between ESS and polygraphic severity

 

 

 

ESS showed a weak positive correlation with the four-level polygraphic severity classification (Spearman ρ = 0.142; p = 0.001). Subjective daytime sleepiness and SDB severity were therefore only weakly aligned at the individual level.

 

 

 

STOP-BANG among non-sleepy adults

 

 

 

Among the 353 non-sleepy adults, 349 had a valid STOP-BANG score from 0 to 8. The median STOP-BANG score was 3 (IQR 2–4). Scores ranged from 0 to 8, with 1, 12, 80, 96, 81, 46, 19, 11, and 3 patients at each successive score, respectively. STOP-BANG AUCs were 0.816 (95% CI 0.745–0.877) for any SDB, 0.736 (0.677–0.791) for moderate-to-severe SDB, and 0.749 (0.696–0.797) for severe SDB. At STOP-BANG ≥ 3, sensitivity for severe SDB was 91.8% (95% CI 86.3–95.5); 145 of 158 severe cases were screen-positive, while 13 had STOP-BANG < 3. As the STOP-BANG threshold increased from ≥ 2 to ≥ 6, sensitivity for severe SDB fell from 98.7% to 15.2%, while specificity rose from 5.8% to 95.3% (Supplementary Table S2). (Table 4)

 

AUC 95% CIs were estimated using percentile bootstrap. Exact Clopper-Pearson 95% CIs were used for sensitivity, specificity, PPV, and NPV, while 95% CIs for LR+ and LR− were calculated on the logarithmic scale.

 

Abbreviations: AUC, area under the curve; CI, confidence interval; PPV, positive predictive value; NPV, negative predictive value; LR+, positive likelihood ratio; LR−, negative likelihood ratio.

 

(Figure 4) The AUC was 0.816 for any SDB, 0.736 for moderate-to-severe SDB, and 0.749 for severe SDB. The markers show the corresponding operating points at a STOP-BANG threshold of ≥ 3.

 

 

 

Sensitivity analyses

 

 

 

In the sensitivity analysis restricted to obstructive SDB, normal polygraphy studies were retained as controls, and SDB cases were included only when the dominant respiratory pattern was documented as obstructive. This analysis included 460 adults. Among 208 adults with severe obstructive SDB, 123 (59.1%) were non-sleepy. Among non-sleepy adults with a valid STOP-BANG score, the sensitivity of STOP-BANG ≥ 3 for severe obstructive SDB was 91.7%.

 

A conservative sensitivity analysis excluded all 62 records involved in the 31 suspected duplicate pairs. Forty-seven of these records belonged to the adult primary cohort, leaving 491 adults, and the main findings remained essentially unchanged: 148 of 241 adults with severe SDB (61.4%) were non-sleepy, while the sensitivity of STOP-BANG ≥ 3 for severe SDB remained 91.2% among non-sleepy adults with valid scores.

 

In an exploratory complete-case analysis (n = 531), ESS > 10 was associated with any SDB (adjusted RR 1.06; 95% CI 1.01–1.11) and moderate-to-severe SDB (adjusted RR 1.12; 95% CI 1.02–1.23), but not with severe SDB (adjusted RR 1.11; 95% CI 0.93–1.33). In a separate exploratory analysis by sex, 63.9% of men and 68.9% of women were non-sleepy (p = 0.263). Among adults with severe SDB, 59.8% of men and 68.5% of women were non-sleepy (p = 0.203). For context, repeating the main comparisons in all 544 records with a valid ESS score, irrespective of age, led to the same overall interpretation (data not shown).

 

 

 

 

 

 

 

DISCUSSION

 

Principal findings

 

 

 

In this selected cardiovascular referral cohort, the absence of excessive daytime sleepiness did not identify a low-risk group for SDB. Severe SDB was present in 45.3% of non-sleepy adults and 53.5% of sleepy adults, corresponding to an RR of 1.18 (95% CI 0.99–1.41; p = 0.084). The correlation between ESS and polygraphic severity was weak (ρ = 0.142). Taken together, these findings suggest that subjective sleepiness has limited ability to distinguish between levels of SDB severity. Finally, the high overall frequency of SDB (92.4%) reflects the selected population referred to a specialized cardiovascular sleep unit and should not be interpreted as population prevalence.

 

Because 65.6% of adults in the cohort had an ESS score ≤10, the finding that 61.8% of severe SDB cases were non-sleepy mainly reflects the composition of the study population. It should therefore not be interpreted as an independent effect or as evidence of a distinct epidemiological subgroup. Within the non-sleepy subgroup, however, STOP-BANG ≥ 3 identified 91.8% of severe SDB cases.

 

 

 

Subjective sleepiness and polygraphic severity

 

 

 

ESS and STOP-BANG provide different types of clinical information. ESS quantifies subjective daytime sleep propensity [4,5], whereas STOP-BANG combines symptoms and clinical characteristics associated with OSA [10,11]. In the adult cohort, ESS showed only a weak correlation with polygraphic severity (ρ = 0.142), suggesting that greater SDB severity was not consistently associated with greater subjective sleepiness.

 

Silva et al. similarly reported greater sensitivity for STOP-BANG than for ESS in detecting moderate-to-severe and severe SDB in Sleep Heart Health Study data [12]. The measurement properties of ESS also limit its usefulness as a marker of OSA severity [5]. Differences in reported sleepiness may be influenced by arousal propensity, sleep duration, obesity, sex-related differences, comorbid conditions, and cultural or individual perceptions of sleepiness. These factors were not measured in the present study.

 

 

 

Cardiovascular evidence for OSA without excessive sleepiness

 

 

 

Several major cardiovascular trials, including Barbé et al., RICCADSA, SAVE, and ISAACC, have shown that clinically significant OSA can occur in cardiovascular patients with little or no excessive daytime sleepiness [6–9]. The definition of non-sleepiness was not identical across studies: RICCADSA used ESS <10, whereas we defined excessive daytime sleepiness as ESS > 10 and therefore classified patients with ESS ≤ 10 as non-sleepy. This small difference in threshold should be kept in mind when comparing results across studies.

 

 

 

Interpretation of sleepiness status among severe SDB cases

 

 

 

The clinically relevant finding is not the 61.8% proportion by itself, but the limited ability of sleepiness status to discriminate between severe and non-severe SDB. Severe SDB was more frequent among sleepy adults, although the difference was not statistically significant. A low ESS score should therefore not be used alone to exclude the need for objective sleep evaluation when the overall clinical profile suggests a risk of OSA.

 

This does not mean that non-sleepy adults would necessarily go unrecognized in routine practice. Clinicians also consider snoring, witnessed apneas, obesity, hypertension, cardiovascular disease, and other risk features. The data support taking these clinical factors into account rather than relying on subjective sleepiness alone to decide who should undergo further evaluation.

 

 

 

STOP-BANG performance in non-sleepy adults

 

At the prespecified high-sensitivity threshold of ≥ 3, STOP-BANG identified 91.8% of severe SDB cases, with an LR− of 0.20 (95% CI 0.11–0.34). As expected, increasing the threshold improved specificity but markedly reduced sensitivity. This pattern is consistent with previous STOP-BANG meta-analyses [18] and with findings in cardiovascular populations, where a pooled sensitivity of 93.9% for severe OSA has been reported at a threshold of ≥ 3 [19].

 

These findings support the use of STOP-BANG as a screening tool to guide further sleep evaluation rather than as a diagnostic test. The AUC for severe SDB was 0.749, compared with a pooled AUC of 0.52 reported in a meta-analysis of populations at cardiovascular risk [19]. Differences in referral patterns, disease prevalence, and our focus on non-sleepy adults may help explain this difference. The AUC for any SDB was 0.816, but only 33 adults did not have SDB, which limited the precision of this estimate. A similar use of STOP-BANG as a screening tool has been reported in a clinical population from Cameroon [13].

 

 

 

Role of STOP-BANG as a screening tool

 

 

 

At a STOP-BANG threshold of ≥ 3, specificity for severe SDB was 41.9% (95% CI 34.8–49.2), meaning that many adults without severe SDB still screened positive. This is expected when a threshold is chosen to favor sensitivity. As the threshold increased, specificity improved while sensitivity decreased markedly. The most appropriate cut-off therefore depends on the clinical objective: identifying as many cases as possible or reducing the number of patients who need further testing.

 

A positive STOP-BANG result does not confirm the diagnosis. According to guidance from the American Academy of Sleep Medicine, suspected OSA should be confirmed with objective sleep testing rather than questionnaires alone [20]. STOP-BANG is therefore best used to identify patients who should undergo further objective sleep evaluation.

 

 

 

Implications for cardiovascular care

 

 

 

Sleep apnea is common among patients with hypertension, heart failure, arrhythmias, obesity, and vascular disease [3]. In this context, a low ESS score should not provide reassurance when other clinical features indicate substantial OSA risk.

 

This is particularly relevant in regions where access to sleep studies is limited. Comparable polygraphy data from West Africa remain scarce. In Burkina Faso, a sleep-clinic study found OSA to be common among patients evaluated with respiratory polygraphy [14]. In such regions, a risk-based approach may help make better use of limited testing resources, while diagnosis still relies on objective polygraphy or polysomnography.

 

 

 

Strengths and limitations

 

 

 

This study has several strengths. It relied on objective polygraphic findings and drew on a relatively large real-world cohort from a region that remains underrepresented in sleep research. Daytime sleepiness and OSA risk were treated as related but distinct clinical dimensions. The main results were very similar when we restricted the analysis to patients with a documented obstructive pattern and when we excluded all records involved in suspected duplicate pairs. We also examined whether missing ESS data could have affected the findings and performed an exploratory analysis adjusted for age, sex, and BMI.

 

Several limitations should be considered. This investigation was a retrospective, single-center analysis restricted to individuals already referred for sleep evaluation. The high observed burden of sleep-disordered breathing reflects the referral-based composition of the cohort and its elevated pretest probability, and should not be taken as an estimate of population prevalence. Of the 740 interpretable polygraphy studies, 196 (26.5%) did not have a valid Epworth Sleepiness Scale (ESS) score and were excluded from the main ESS analysis. Participants with and without a valid ESS score were similar in age, sex, body mass index, and polygraphic severity. However, selection bias related to unmeasured factors cannot be excluded. We also found 31 pairs of records that might have been duplicates. Because we could not confirm that they represented duplicate examinations, we kept them in the main analysis. Excluding all suspected duplicates in a sensitivity analysis did not materially change the results.

 

Respiratory polygraphy was used rather than full polysomnography. Because respiratory events were expressed per hour of recording rather than per hour of confirmed sleep, the AHI may be underestimated when part of the recording is spent awake. Polygraphy also does not provide information on sleep stages or electroencephalographic arousals and therefore cannot identify respiratory effort-related arousals in the same way as polysomnography. The recordings were not re-scored centrally for this retrospective study, and the registry did not consistently record which physician reviewed each examination, so inter-observer agreement could not be assessed. Hypopneas were scored using a ≥ 3% oxygen desaturation threshold, although the exact AASM scoring manual version used during the study period was not consistently documented.

 

The main analysis was restricted to adults because the conventional ESS and STOP-BANG questionnaires were developed for adult populations. The STOP-BANG analysis also has a specific limitation in this setting. The non-sleepy subgroup was defined using ESS ≤ 10, while STOP-BANG includes a Tiredness item that reflects a related symptom. This overlap may have affected the performance of the score in this subgroup. A seven-item score excluding Tiredness would have been a useful sensitivity analysis, but the individual STOP-BANG items were not consistently available in the registry, so this score could not be reconstructed reliably. We did not apply a formal adjustment for multiple testing across the three related SDB outcomes. These outcomes had been defined in advance and followed a clear clinical severity gradient, so we interpreted the results together with the effect estimates and confidence intervals. The multivariable analysis was exploratory, and some residual confounding may remain.

 

Finally, because this was an observational study, we cannot explain why some adults with severe SDB report substantial daytime sleepiness while others do not. We also cannot determine whether using ESS or STOP-BANG to decide who should be tested would lead to better clinical outcomes.

 

 

 

Clinical implications

 

 

 

ESS and STOP-BANG capture related but distinct aspects of sleep-disordered breathing. ESS measures subjective daytime sleepiness, whereas STOP-BANG estimates OSA risk from symptoms and clinical characteristics. In adults with cardiovascular or nocturnal risk factors, an ESS score ≤ 10 should not, on its own, rule out objective sleep evaluation. When the priority is to avoid missing severe SDB, a STOP-BANG score ≥ 3 can help identify patients who warrant further testing. This comes with limited specificity, and sensitivity falls as higher cut-offs are used.

 

 

 

 

 

CONCLUSION 

 

 

 

 

In this West African cardiovascular referral cohort, subjective daytime sleepiness was only weakly associated with polygraphic SDB severity, and severe SDB was not significantly more frequent among adults with excessive sleepiness. The finding that 61.8% of severe SDB cases were non-sleepy should be interpreted in light of the fact that non-sleepy adults made up approximately two-thirds of the cohort.

Among non-sleepy adults, STOP-BANG showed high sensitivity, with a score ≥ 3 identifying 91.8% of severe SDB cases. ESS should therefore be used mainly to assess daytime sleepiness rather than as a stand-alone criterion for deciding who should undergo objective sleep evaluation. A risk-based approach that incorporates clinical factors to guide objective testing may be particularly useful in regions where access to sleep studies is limited.

 

 

 

 

 

 

 

 

REFERENCES

 

1. Benjafield AV, Ayas NT, Eastwood PR, Heinzer R, Ip MSM, Morrell MJ, et al. Estimation of the global prevalence and burden of obstructive sleep apnoea: a literature-based analysis. Lancet Respir Med. 2019;7(8):687-98. 2. Lévy P, Kohler M, McNicholas WT, Barbé F, McEvoy RD, Somers VK, et al. Obstructive sleep apnoea syndrome. Nat Rev Dis Primers. 2015;1:15015. 3. Yeghiazarians Y, Jneid H, Tietjens JR, Redline S, Brown DL, El-Sherif N, et al. Obstructive sleep apnea and cardiovascular disease: a scientific statement from the American Heart Association. Circulation. 2021;144(3):e56-e67. 4. Johns MW. A new method for measuring daytime sleepiness: the Epworth Sleepiness Scale. Sleep. 1991;14(6):540-5. 5. Kendzerska TB, Smith PM, Brignardello-Petersen R, Leung RS, Tomlinson GA. Evaluation of the measurement properties of the Epworth Sleepiness Scale: a systematic review. Sleep Med Rev. 2014;18(4):321-31. 6. Barbé F, Durán-Cantolla J, Sánchez-de-la-Torre M, Martínez-Alonso M, Carmona C, Barceló A, et al. Effect of continuous positive airway pressure on the incidence of hypertension and cardiovascular events in nonsleepy patients with obstructive sleep apnea: a randomized controlled trial. JAMA. 2012;307(20):2161-8. doi:10.1001/jama.2012.4366. 7. Peker Y, Glantz H, Eulenburg C, Wegscheider K, Herlitz J, Thunström E. Effect of positive airway pressure on cardiovascular outcomes in coronary artery disease patients with nonsleepy obstructive sleep apnea: the RICCADSA randomized controlled trial. Am J Respir Crit Care Med. 2016;194(5):613-20. 8. McEvoy RD, Antic NA, Heeley E, Luo Y, Ou Q, Zhang X, et al. CPAP for prevention of cardiovascular events in obstructive sleep apnea. N Engl J Med. 2016;375(10):919-31. 9. Sánchez-de-la-Torre M, Sánchez-de-la-Torre A, Bertran S, Abad J, Duran-Cantolla J, Cabriada V, et al. Effect of obstructive sleep apnoea and its treatment with continuous positive airway pressure on the prevalence of cardiovascular events in patients with acute coronary syndrome (ISAACC study): a randomised controlled trial. Lancet Respir Med. 2020;8(4):359-67. 10. Chung F, Yegneswaran B, Liao P, Chung SA, Vairavanathan S, Islam S, et al. STOP questionnaire: a tool to screen patients for obstructive sleep apnea. Anesthesiology. 2008;108(5):812-21. 11. Chung F, Abdullah HR, Liao P. STOP-Bang Questionnaire: a practical approach to screen for obstructive sleep apnea. Chest. 2016;149(3):631-8.

 

12. Silva GE, Vana KD, Goodwin JL, Sherrill DL, Quan SF. Identification of patients with sleep disordered breathing: comparing the Four-Variable Screening Tool, STOP, STOP-Bang, and Epworth Sleepiness Scales. J Clin Sleep Med. 2011;7(5):467-72. 13. Massongo M, Balkissou AD, Kenyo CK, Sawa BN, Kanko N, Pefura EW. The STOPBANG score is effective for obstructive sleep apnea syndrome screening and correlates with its features, in a sub-Saharan African population. Pan Afr Med J. 2020;36:93. 14. Ouédraogo AR, Tiendrebeogo A, Boncoungou K, Birba E, Ouédraogo GA, Assao Neino MM, et al. Prevalence of obstructive sleep apnoea in sleep consultations in Burkina Faso: implications for monitoring. Afr J Thorac Crit Care Med. 2020;26(3):74-80. 15. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet. 2007;370(9596):1453-7. 16. Bossuyt PM, Reitsma JB, Bruns DE, Gatsonis CA, Glasziou PP, Irwig L, et al. STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies. BMJ. 2015;351:h5527. 17. Berry RB, Budhiraja R, Gottlieb DJ, Gozal D, Iber C, Kapur VK, et al. Rules for scoring respiratory events in sleep: update of the 2007 AASM Manual for the Scoring of Sleep and Associated Events. Deliberations of the Sleep Apnea Definitions Task Force of the American Academy of Sleep Medicine. J Clin Sleep Med. 2012;8(5):597-619. doi:10.5664/jcsm.2172. 18. Nagappa M, Liao P, Wong J, Auckley D, Ramachandran SK, Memtsoudis S, et al. Validation of the STOP-Bang Questionnaire as a screening tool for obstructive sleep apnea among different populations: a systematic review and meta-analysis. PLoS One. 2015;10(12):e0143697. 19. Hwang M, Zhang K, Nagappa M, Saripella A, Englesakis M, Chung F. Validation of the STOP-Bang questionnaire as a screening tool for obstructive sleep apnoea in patients with cardiovascular risk factors: a systematic review and meta-analysis. BMJ Open Respir Res. 2021;8(1):e000848. 20. Kapur VK, Auckley DH, Chowdhuri S, Kuhlmann DC, Mehra R, Ramar K, et al. Clinical practice guideline for diagnostic testing for adult obstructive sleep apnea: an American Academy of Sleep Medicine clinical practice guideline. J Clin Sleep Med. 2017;13(3):479-504.