Effect of arterial blood bicarbonate (HCO₃⁻) concentration on the accuracy of STOP-Bang questionnaire screening for obstructive sleep apnea

This prospective study was approved by the institutional ethics committee of the Third Affiliated Hospital of Anhui Medical University, and informed consent was obtained from all patients to include their data in this study. Patients who underwent PSG examination at the Sleep Center of the Third Affiliated Hospital of Anhui Medical University (First People's Hospital of Hefei) from March 2019 to October 2020 were screened for study inclusion. Inclusion criteria were: (1) age > 18 years old; (2) complete autonomous behavior and cognitive ability; (3) arterial blood gas analysis and PSG monitoring performed during hospitalization; and (4) ability to answer the questionnaire completely and accurately. Exclusion criteria were: (1) associated diseases such as chronic obstructive pulmonary disease, acute attacks of asthma, interstitial pulmonary disease, laryngospasm, vocal cord paralysis, tracheal foreign bodies, anemia, and electrolyte disorders that may affect blood gas analysis; (2) other underlying diseases that may affect HCO₃⁻ concentration, such as liver, kidney, lung, and heart dysfunction; (3) use of a ventilator in the past three months; (4) other common sleep respiratory disorders besides OSA; (5) pregnancy, lactation, puerperium, and mental disorders (including a history of depression or anxiety); (6) use of sedatives or antipsychotic treatment over the past three months; (7) patients with an abnormal EEG and (8) if patients are using diuretics.

All subjects were monitored using an Alice6 PSG instrument (Philips Respironics, USA) and analyzed manually by sleep technicians to confirm or exclude OSA. PSG was performed with 16 channels, including: electrooculography, electroencephalography, electrocardiography, electromyography (submental and bilateral tibial), airflow measurements with both oronasalthermal sensors and nasal air pressure transducers, transtracheal sounds via microphone, rib cage and abdominal movement by inductance plethysmography using thoracoabdominal belts, and continuous pulse oximetry. Studies were scored using the 2012 American Academy of Sleep Medicine (AASM) [7] scoring guidelines with hypopneas scored as a 30% drop in the nasal pressure from baseline for at least 10 s and associated with either arousal or drop in oxygen saturation by 3%. The apnea hypopnea index (AHI), lowest blood oxygen saturation (LSaO₂) and mean blood oxygen saturation (MSaO₂) were recorded. OSA was diagnosed according to the Chinese Guidelines for Primary Diagnosis and Treatment of Adult Obstructive Sleep Apnea (2018). These include (1) clinical symptoms of any one or more of the following: daytime sleepiness, non-recovery of energy after waking, fatigue, or insomnia; waking up because of breathlessness, poor breathing, or suffocation at night; habitual snoring and breathing interruption; and hypertension, coronary heart disease, stroke, heart failure, atrial fibrillation, type 2 diabetes, mood disorders, or cognitive impairment; (2) PSG monitoring: AHI ≥ 5 times/h, mainly obstructive events; (3) none of the above symptoms and PSG monitoring: AHI ≥ 15 times/h, mainly obstructive events. Adult OSA can be diagnosed if criteria (1) and (2) are met or only criterion (3) is met. Grade of disease: AHI: 5–15 times/h is mild, > 15–30 times/h is moderate, and > 30 times/h is severe. The percentage of total sleep time when blood oxygen is less than 90% (CT90%), duration of apnea hypopnea in total sleep time (AHT%), mean apnea-hypopnea duration (MAD), and duration of apnea hypopnea per hour (HAD) were calculated based on the reported data. PSG is the gold standard for the diagnosis of OSA.

Two milliliters of radial artery blood were collected with a special syringe for blood gas analysis (American Westmed). Samples were obtained from study subjects following PSG monitoring and when they were awake and seated in a quiet room. The sample was mixed and sealed, and blood gas analysis was performed within 30 min to detect HCO₃⁻. The process of specimen extraction strictly followed the requirements of aseptic operation.

The ESS asks individuals to grade their sleepiness during several routine activities including sitting and reading, watching TV, sitting in a public place, a long ride (more than 1 h), talking with people, resting after dinner (not drinking), driving, and reposing at rest in the afternoon. Each condition is divided into four grades: never (0), rarely (1), sometimes (2) and often (3). The subjects will score according to their own conditions, and the researchers will calculate the total score. An ESS score ≥ 9 was associated with high risk of OSA, and an ESS score < 9 was associated with a low risk of OSA (shown in Additional file 1: Fig. 1).

The questionnaire comprised eight characteristics: S, snoring; T, tiredness; O, observed apnea; P, high blood pressure; B, body mass index > 35 kg/m²; A, older than 50 years; N, neck circumference > 40 cm; and G, male sex. Patients answered the first four questions (STOP questions), and the staff in the sleep room measured the height, weight, blood pressure, and neck circumference of the patients. Then, the respondents answered the last four questions (BANG questions). The answer was yes (1 points) or no (0 points), and the total score was calculated. A score ≥ 3 was defined as a high risk of OSA, while a score < 3 was defined as a low risk of OSA (shown in Additional file 2: Fig. 2).

The Berlin questionnaire consists of ten questions plus information on height and weight arranged in three categories: snoring and cessation of breathing (category 1; five questions); symptoms of excessive daytime sleepiness (category 2; four questions); and BMI and hypertension (category 3; one question and height and weight information). Some of the questions had to be answered by the patient's family members or co-residents to ensure the accuracy of the answers. Positive scores in 2 or more categories suggest that the respondent is at high risk for OSA (shown in Additional file 3: Fig. 3).

All patients received PSG monitoring, blood gas analysis, and completed the ESS, SBQ, and Berlin questionnaires. According to AHI and symptoms, the patients were divided into the non-OSA group or the OSA group, and the latter group was further divided into mild (5 times/h ≤ AHI ≤ 15 times/h), moderate (15 times/h < AHI ≤ 30 times/h), and severe (AHI > 30 times/h). Patients were divided into two groups according to the SBQ: the high-risk OSA group (STOP-Bang questionnaire score ≥ 3 points) and low-risk OSA group (STOP-Bang questionnaire score < 3 points). Patients were divided into two groups according to ESS: ESS ≥ 9 was classified as high-risk for OSA, and ESS < 9 was classified as low-risk for OSA. According to the Berlin questionnaire, the patients were divided into two groups: positive for the high-risk OSA group and negative for the low-risk OSA group. A correlation analysis between HCO₃⁻ concentration and AHI, LSAO2, MSAO2, CT90%, AHT%, MAD, and HAD was carried out to judge the correlation degree, and the diagnostic value of HCO₃⁻ concentration on OSA was analyzed and evaluated. The sensitivity and specificity of the three kinds of questionnaires for screening OSA were compared. Then, the diagnostic value of the STOP-Bang questionnaire score alone and SBQ combined with the HCO₃⁻ concentration for OSA was further compared.

SPSS19.0 statistical software was used for data analysis. Patient information was analyzed by descriptive statistics. The normal distribution data is represented by \(\stackrel{-}{X\hspace{0.17em}}\)± s, and the non-normal distribution data is represented by the median (M) and the interquartile range (P75–P25). Correlation analysis was conducted between the HCO₃⁻ concentration and AHI, CT90%, AHT%, MAD, and HAD, and the correlation coefficients were calculated. Then, calculation was done for the degree of sensitivity, specific diagnostic test evaluation of the four tables, and sensitivity to the specific degrees of comparison between different methods using the matching χ2 test (with P < 0.05 for the difference being statistically significant). The optimal truncation value of the HCO₃⁻ concentration for the diagnosis of OSA was analyzed by Youden index. A ROC curve was used to compare the scores of SBQ alone and the diagnostic value of SBQ combined with HCO₃⁻ concentration for OSA.

The general morphological data of 144 patients with suspected diagnosis of OSA are shown in Table 1. According to the PSG monitoring results, the 144 patients with suspected diagnosis were divided as follows: 24 in the non-OSA group and 120 in the OSA group, including 17 mild cases, 12 moderate cases, and 91 severe cases.

The HCO₃⁻ concentration was positively correlated with AHI (r = 0.537, P < 0.001) (Fig. 1A), negatively correlated with LSaO₂ (r =  − 0.251, P = 0.004) (Fig. 1B), negatively correlated with MSaO₂ (r =  − 0.394, P < 0.001) (Fig. 1C), positively correlated with CT90 (r = 0.437, P < 0.001) (Fig. 2A), positively correlated with AHT (r = 0.433, P < 0.001) (Fig. 2B), positively correlated with HAD (r = 0.433, P < 0.001) (Fig. 2C), and had no correlation with MAD.

When the HCO₃⁻ concentration was applied to predict OSA, the corresponding area under the ROC curve (AUC) was 0.80 (Fig. 3A), sensitivity was 80.1% and specificity was 75%. After calculating the corresponding Yoden index, the HCO₃⁻ concentration of 24.6 mmol/L had the best cutoff value for screening (Yoden index = 0.558). The ratio with an HCO₃⁻ concentration ≥ 24.6 mmol/L in the non-OSA group was significantly lower than that in the OSA group (25.0% vs. 80.8%, P < 0.001). The ratio with an HCO₃⁻ concentration ≥ 24.6 mmol/L in the severe OSA group was significantly higher than that in the mild OSA group (87.9% vs. 52.9%, P = 0.001), while there was no statistically significant difference in the ratio between the severe OSA group and the moderate OSA group (87.9% vs. 66.7%, P = 0.127).

Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of the three questionnaires for predicting OSA are shown in Table 2.

The sensitivity of the SBQ questionnaire compared with the ESS questionnaire (97.5% VS 81.7%, P < 0.001) was statistically significant. The sensitivity of the SBQ questionnaire compared with the Berlin questionnaire (97.5% vs. 79.2%, P < 0.001) was also statistically significant. There was no statistically significant difference in sensitivity between the ESS and Berlin questionnaires (81.7% vs. 79.2%, P = 0.701).

There was no statistical significance in the specificity of the three scales (25%, 37.5%, 37.5%).

The sensitivity and specificity of all OSA, moderate and severe OSA, and severe OSA were calculated, with an SBQ score ≥ 3 as the screening cutoff value, as shown in Table 3. A combined SBQ score ≥ 3 points and HCO₃⁻ concentration ≥ 24.6 mmol/L showed increased specificity and decreased sensitivity (Table 3).

When the SBQ was applied alone to predict OSA, the corresponding AUC of the OSA group was 0.613 (P > 0.05). When a combined SBQ score ≥ 3 and the HCO₃⁻ concentration ≥ 24.6 mmol/L, the corresponding AUC in the OSA group increased to 0.771 (P < 0.01) (Fig. 3B).