The effectiveness of sleep breathing impairment index in assessing obstructive sleep apnea severity

Free AccessScientific InvestigationsThe effectiveness of sleep breathing impairment index in assessing obstructive sleep apnea severity Lu Dai, MD, Wenhao Cao, MD, Jinmei Luo, MD, Rong Huang, MD, Yi Xiao, MD Lu Dai, MD Department of Respiratory and Critical Care Medicine, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China , Wenhao Cao, MD Department of Respiratory and Critical Care Medicine, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China , Jinmei Luo, MD Department of Respiratory and Critical Care Medicine, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China , Rong Huang, MD Department of Respiratory and Critical Care Medicine, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China , Yi Xiao, MD Address correspondence to: Yi Xiao, MD, No. 1 Shuaifuyuan, Dongcheng District, Beijing, China 100730; Email: E-mail Address: [email protected] Department of Respiratory and Critical Care Medicine, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China Published Online:February 1, 2023https://doi.org/10.5664/jcsm.10302SectionsAbstractEpubPDF ShareShare onFacebookTwitterLinkedInRedditEmail ToolsAdd to favoritesDownload CitationsTrack Citations AboutABSTRACTStudy Objectives:Using the apnea-hypopnea index (AHI) and the sleep breathing impairment index (SBII) to assess the severity of obstructive sleep apnea (OSA) to study how effective SBII is in assessing the severity and cardiovascular disease (CVD) prognosis.Methods:This study comprised a total of 147 patients with diagnosed OSA. The AHI and SBII were calculated from the polysomnography. Patients were enrolled in the cluster analysis using 20 symptoms and the SBII. The prognostic indicator was determined as the moderate-to-high Framingham 10-year CVD risk.Results:Cluster analysis revealed 3 separate groups: cluster 1 (n = 45, 30.61%) had the lowest symptoms complaints yet the highest PSQI score; cluster 2 (n = 70, 47.62%) had considerably increased symptom complaints but the lowest Epworth Sleepiness Scale score, intermediate PSG indices, a higher low arousal threshold possibility, and a lower SBII quantile; cluster 3 (n = 32, 21.77%) had the largest percentage of smokers, a predominant symptom of restless sleep, severe PSG characteristics, a lower low arousal threshold likelihood, a greater SBII quantile and a higher Framingham CVD risk. There were no differences in severity indicated by AHI between groups. Higher SBII rather than AHI is associated with an increased 10-year CVD risk.Conclusions:SBII provides higher sensitivity when evaluating OSA severity and better predictive capabilities for CVD outcomes. SBII may be a more effective substitute for AHI in the future.Citation:Dai L, Cao W, Luo J, Huang R, Xiao Y. The effectiveness of sleep breathing impairment index in assessing obstructive sleep apnea severity. J Clin Sleep Med. 2023;19(2):267–274.BRIEF SUMMARYCurrent Knowledge/Study Rationale: The sleep breathing impairment index (SBII) is a novel index to assess the severity of obstructive sleep apnea (OSA). However, whether this metric can be a good alternative to the apnea-hypopnea index (AHI) was unclear. This study aimed to investigate its effectiveness in assessing the severity and prognosis of OSA by cluster analysis.Study Impact: In this study, we found that SBII as a novel index provides higher sensitivity when evaluating OSA severity and better predictive capabilities for cardiovascular disease outcomes, which indicated that SBII may be a better substitute for AHI in future evaluations.INTRODUCTIONObstructive sleep apnea (OSA) is a disease that causes intermittent hypoxia, sleep fragmentation, and sympathetic activation.1 It is caused by a repeated collapse in the upper airway during sleep. The gold standard for diagnosing OSA is the apnea-hypopnea index (AHI) generated from polysomnography (PSG). However, based just on the frequency of respiratory episodes, the AHI is unable to adequately assess the severity of OSA.2,3 According to Punjabi,3 AHI merely equalizes apneas and hypopneas in determining event frequency, despite the fact that they are not pathophysiologically comparable. Furthermore, AHI fails to eliminate the temporal distribution of events since the impacts of events clustered and scattered throughout the night are distinct. Moreover, AHI is insufficient for determining illness complexity since it cannot accurately reflect the severity of desaturation and the length of events (including respiratory events and desaturation events).As a result, numerous novel criteria that can assess OSA more thoroughly have recently emerged and have demonstrated their benefits in several trials.4–6 Among these, the sleep breathing impairment index (SBII), which takes into account both respiratory events and hypoxia, has been proven to be beneficial in predicting 10-year cardiovascular disease (CVD).7 It is necessary to determine whether this metric can be a good alternative to AHI in assessing OSA severity and its prognosis.Individual differences exist from clinical and physiological mechanisms to future prognosis,8–10 as evidenced by earlier research indicating that OSA is a heterogeneous disease.11 Several researchers used cluster analysis to categorize patients with OSA with comparable clinical complaints into the same clinical phenotypes and discovered differences between each phenotype,12–15 proving the usefulness of cluster analysis. The purpose of this study was to use cluster analysis to investigate the predictability of SBII in determining OSA severity and CVD prognosis.METHODSParticipantsThis was a cross-sectional study. Patients who visited the sleep center of Peking Union Medical College Hospital for suspected OSA from February 2020 to January 2022 were enrolled. The exclusion criteria were as follows: (1) having received OSA treatment, such as continuous positive airway pressure (CPAP), prior to enrollment; (2) having total sleep time (TST) < 4 hours; (3) unable to finish questionnaires; (4) incomplete polysomnographic data; and (5) age < 18 years old. The study was approved by the ethics committees of Peking Union Medical College Hospital (K1474) and was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants.Data collectionDemographic data of each participant were collected, including age, sex, and body mass index (BMI). Smoking status, drinking consumption, and medical conditions were collected via questionnaires. All participants filled out the Pittsburgh Sleep Quality Index (PSQI) and Epworth Sleepiness Scale (ESS) questionnaires; thus, several clinical symptoms were obtained from them. The Framingham CVD risk was determined using a sex-specific multivariable risk factor algorithm that took into account age, total cholesterol, high-density lipoprotein cholesterol, systolic blood pressure, medication for hypertension, smoking, and diabetes status.16 For better analysis, a risk that was predicted to be ">30%" was ultimately determined to be 30%. Risks of moderate to high CVD were defined as those more than 6%.Polysomnographic recordingAll participants underwent whole-night PSG (Embla N7000; Natus Medical Incorporated, Orlando, FL, USA) from 11 pm to 6 am in the sleep center. Data were recorded and analyzed by skilled sleep laboratory technicians following the standard criteria recommended by the American Academy of Sleep Medicine.17 AHI was defined as the number of apnea and hypopnea events per hour. Oxygen desaturation index (ODI) was defined as the number of desaturations per hour. Total sleep time (TST), the lowest oxygen saturation by pulse oximetry (LSpO2), the percentage of time spent with SpO2 below 90% (T90), sleep efficiency, the percentage of each sleep stage, and the maximum time for each respiratory event were also collected. The low arousal threshold (LAT) was calculated through the method proposed by Edwards et al18: AHI < 30 events/h, LSpO2 > 82.5%, and the percentage of hypopneas > 58.3%. Each criterion earns 1 score, a score of 2 or above predicts the existence of LAT.SBII measurementSBII is a novel index combining the degree and duration of each event-related desaturation, and the frequency of events, which was more comprehensive when describing the severity of OSA than AHI, which only describes the frequency. The event-associated desaturation was defined when its start was in a 100-second window from a respiratory event's beginning and its area includes both desaturation and recovery parts. Therefore, the desaturation area can be regarded as a triangle with the duration as its base and the degree as its height. SBII was calculated by a customized and automated program managed by python, with the sum of the products of the desaturation area and the related events' duration, and then divided by the TST. The unit of SBII is (%min2)/hour. Rapid eye movement (REM)-SBII and Supine-SBII could also be obtained accordingly.Cluster analysisA total of 21 variables were included in the cluster analysis in the OSA population, including 20 symptoms and 1 OSA severity variable, SBII. Twenty symptoms are listed as followed: snoring, sleepy, difficulty maintaining sleep, difficulty falling asleep, waking up in the middle of the night or early morning, nocturia, wake up suddenly and cannot breathe, nightmares while sleeping, waking up at night with a headache, restless in sleep, perspire heavily at night, nasal congestion at night, been told they stop breathing during sleep, physically tired after waking up, dry mouth in the morning, headache in the morning, hypomnesia, concentration drops, unresponsiveness, and short of breath during the day. To better make comparisons among clusters, SBII severity was defined from 1 to 4 by its quantiles, while AHI severity was defined from 1 to 4 by mild (AHI 5–15 events/h), moderate (AHI 15–30 events/h), severe (AHI 30–60 events/h), and very severe (AHI > 60 events/h) OSA.Statistical analysisCategorical data are presented as numbers (percentage). Continuous variables are shown as mean ± standard deviation or median (interquartile range, 25–75%) depending on whether the data were normally distributed or not. Two-step cluster in SPSS (IBM Corporation, Armonk, NY, USA) was used to cluster participants into groups based on symptoms and SBII. Differences among clusters were examined via chi-square, analysis of variance (ANOVA), or Kruskal–Wallis test as appropriate. The correlation between SBII and moderate-to-high Framingham CVD risks was discovered using logistic regression analysis. SPSS version 26.0 was used for data analysis. A 2-sided P value < .05 was considered statistically significant.RESULTSBaseline characteristics of enrolled patientsA total of 147 participants with diagnosed OSA were recruited for the study, with 129 men and 18 women. The baseline characteristics are presented in Table 1. The median age of all participants was 43 years, and the median BMI was 27.43 kg/m2. Participants had a median AHI of 27.3 events/h, and 48.30% of participants had a prevalence of moderate-to-severe OSA. The incidences of hypertension, diabetes mellitus, smoking, and drinking were 31.97%, 8.84%, 43.54%, and 80.95%, respectively.Table 1 Baseline characteristics of all participants.VariablesValuesAge, y43 (36–51)Male, n (%)129 (87.76%)Female, n (%)18 (12.24%)BMI, kg/m227.43 (25.00–29.41)HTN, n (%)47 (31.97%)DM, n (%)13 (8.84%)Smoking, n (%)64 (43.54%)Drinking, n (%)119 (80.95%)SBII, (%min2)/h23.65 (7.42–74.75)REM-SBII, (%min2)/h6.74 (1.51–24.64)Supine-SBII, (%min2)/h19.02 (5.19–58.66)AHI, events/h27.30 (12.90–52.70)AHI 5–15 events/h, n (%)44 (29.93%)AHI 15–30 events/h, n (%)32 (21.77%)AHI 30–60 events/h, n (%)40 (27.21%)AHI > 60 events/h, n (%)31 (21.09%)ODI, events/h19.90 (8.90–47.80)LSpO2, %84 (75–88)T90, %0.69 (0.06–4.29)ESS11.0 (7.0–16.0)PSQI7.0 (6.0–9.0)Data are presented as median (interquartile range, 25–75%) unless otherwise indicated; n = 147. AHI = apnea-hypopnea index, BMI = body mass index, DM = diabetes mellitus, ESS = Epworth Sleepiness Scale, HTN = hypertension, LSpO2 = lowest oxygen saturation by pulse oximetry, ODI = oxygen desaturation index, OSA = obstructive sleep apnea, PSQI = Pittsburgh Sleep Quality Index, REM = rapid eye movement, SBII = sleep breathing impairment index, T90, percent of time spent with SpO2 below 90%.Cluster analysisBasic characteristics and comorbiditiesA total of 147 participants met the criteria of OSA diagnosis; thus, the cluster analysis was conducted in these participants. Three distinct clusters were identified based on symptom experiences and SBII: cluster 1 (n = 45, 30.61%), cluster 2 (n = 70, 47.62%), and cluster 3 (n = 32, 21.77%). Demographic characteristics and comorbidities are summarized in Table 2. The table shows that cluster 2 had the lowest ESS score and cluster 1 had the highest PSQI score, with significance noted separately when compared to other clusters. Participants in cluster 3 had the highest percentage of smokers.Table 2 Demographic characteristics and comorbidities of participants by cluster.Cluster 1Cluster 2Cluster 3PSubjects, n (%)45 (30.61%)70 (47.62%)32 (21.77%)Male, n (%)38 (84.44%)61 (87.14%)30 (93.75%).515Age, y43.98 ± 10.0344.31 ± 11.8346.03 ± 10.47.695BMI, kg/m227.01 ± 4.7828.42 ± 5.4128.77 ± 5.85.272ESS13 (11–16)a9 (5–13.25)a,b12.5 (6.5–18.75)b.001PSQI8.67 ± 2.79a,c6.93 ± 2.26a6.78 ± 2.31c<.001HTN, n (%)12 (26.67%)24 (34.29%)11 (34.38%).657T2DM, n (%)2 (4.44%)9 (12.86%)2 (6.25%).284Smoking, n (%)17 (37.78%)c26 (37.14%)b21 (65.63%)b,c.017Drinking, n (%)38 (84.44%)55 (78.57%)26 (81.25%).735CVD, n (%)1 (2.22%)6 (8.57%)6 (18.75%).038Hyperlipemia, n (%)5 (11.11%)14 (20.00%)9 (28.13%).166COPD, n (%)1 (2.22%)0 (0.00%)0 (0.00%).524Rhinitis, n (%)12 (26.67%)21 (30.00%)8 (25.00%).852Data are presented as mean ± standard deviation or median (interquartile range, 25–75%) unless otherwise indicated. aP < .05 between cluster 1 and cluster 2; bP < .05 between cluster 2 and cluster 3; cP < .05 between cluster 1 and cluster 3. BMI = body mass index, COPD = chronic obstructive pulmonary disease, CVD = cardiovascular disease, ESS = Epworth Sleepiness Scale, HTN = hypertension, PSQI = Pittsburgh Sleep Quality Index, T2DM = type 2 diabetes mellitus.Symptom experiences and PSG characteristicsThe frequencies of each symptom of patients who enrolled in cluster analysis are detailed in Table 3 and Figure 1. According to the symptoms, cluster 1 can be defined as the "minimally symptomatic group," as members in this cluster hold the lowest probability of the majority of symptoms than the other 2 clusters. Cluster 2 was the "disturbed sleep group," with a significantly higher probability of experiencing several symptoms, including difficulty maintaining sleep (95.71%), waking up suddenly and not able to breathe (71.43%), with headaches in the morning (80.00%), and feeling short of breath during the day (78.57%). Other symptoms were also prominent compared with cluster 1, such as difficulty falling asleep (64.29%), nocturia (31.43%), nightmares while sleeping (78.57%), perspiring heavily at night (54.29%), nasal congestion at night (51.43%), been told they stop breathing during sleep (51.43%), physically tired after waking up (31.43%), and with decreases in concentration (27.14%), and unresponsiveness (58.57%). Cluster 3 had predominantly symptoms of being restless in sleep (100.00%) but the lowest probabilities in waking up suddenly and being not able to breathe (0.00%) and being told they stop breathing during sleep (3.13%).Table 3 Symptoms of participants by clusters.Cluster 1Cluster 2Cluster 3PSnoring4.447.140.00.368Sleepy66.6775.7165.63.449Difficulty maintaining sleep51.11a95.71a,b31.25b<.001Difficulty falling asleep22.22a,c64.29a59.38c<.001Waking up in the middle of the night or early morning22.2240.0034.38.141Nocturia11.11a31.43a15.63.023Wake up suddenly and cannot breathe17.78a,c71.43a,b0.00b,c<.001Nightmares while sleeping24.44a,c78.57a71.88c<.001Waking up at night with a headache33.33a,c81.43a62.50c<.001Restless in sleep8.89a,c77.14a,b100.00b,c<.001Perspire heavily at night22.22a,c54.29a56.25c.001Nasal congestion at night26.67a51.43a31.25.017Been told stop breathing during sleep33.33c51.43b3.13b,c<.001Physically tired after waking up0.00ac31.43a21.88c<.001Dry mouth in the morning4.4411.436.25.391Headache in the morning24.44a80.00a,b46.88b<.001Hypomnesia6.6720.006.25.056Concentration drops0.00a,c27.14a18.75c.001Unresponsiveness24.44a58.57a46.88.002Short of breath during the day20.00a78.57a,b31.25b<.001Data are presented as %. aP < .05 between cluster 1 and cluster 2; bP < .05 between cluster 2 and cluster 3; cP < .05 between cluster 1 and cluster 3.Figure 1: Probability of having a symptom within each cluster.Download FigureTable 4 lists the PSG characteristics of each cluster. The results indicate that cluster 3 has the highest level of SBII severity, thus defined as the "sleep breathing impairment group," but no significant differences were found among the 3 clusters in AHI severity. Other PSG characteristics show that cluster 2 has the shortest maximum time for obstruction apnea, total apnea, and total respiratory events. Differences were manifested between cluster 2 and cluster 3, such that the latter group has a higher ODI (40.17 events/h vs 26.53 events/h), longer maximum time for mixed apnea (33.33 vs 20.68 seconds), lower LAT score (0.72 vs 1.47), and a lower percentage of having LAT (25.00% vs 55.71%).Table 4 PSG characteristics of participants by clusters.Cluster 1Cluster 2Cluster 3PSBII quantiles2.58 ± 1.102.23 ± 1.12b2.97 ± 1.00b.007AHI severity2.40 ± 1.142.40 ± 1.122.31 ± 1.15.890ODI, /h28.87 ± 26.0226.53 ± 25.02b40.17 ± 25.95b.043LSpO2, %79.53 ± 10.5182.33 ± 9.2377.66 ± 10.05.066T90, %6.38 ± 14.565.02 ± 12.886.65 ± 11.94.793Sleep efficiency, %0.87 ± 0.130.86 ± 0.140.88 ± 0.13.778Stage 1, %TST0.08 ± 0.050.08 ± 0.070.08 ± 0.07.769Stage 2, %TST0.48 ± 0.130.49 ± 0.120.52 ± 0.15.417Stage 3, %TST0.27 ± 0.140.26 ± 0.130.23 ± 0.16.506Stage REM, %TST0.17 ± 0.060.17 ± 0.060.16 ± 0.06.704Maximum time for obstructive apnea, s58.7 ± 29.81a42.35 ± 26.6a,b64.2 ± 23.88b<.001Maximum time for central apnea, s5.93 ± 10.828.11 ± 9.85.93 ± 8.37.408Maximum time for mixed apnea, s26.35 ± 26.0620.68 ± 22.65b33.33 ± 24.79b.049Maximum time for hypopnea, s52.24 ± 23.5343.66 ± 18.6550.1 ± 24.23.091Maximum time for apnea, s58.76 ± 29.77a43.31 ± 26.54a,b64.86 ± 22.82b<.001Maximum time for respiratory events, s68.05 ± 26.61a56.23 ± 21.6a,b71.02 ± 22.09b.003LAT, score1.22 ± 0.971.47 ± 1.03b0.72 ± 0.85b.002Having LAT, n (%)46.67%55.71%b25.00%b.015Data are presented as mean ± standard deviation unless otherwise indicated. aP < .05 between cluster 1 and cluster 2; bP < .05 between cluster 2 and cluster 3. AHI = apnea-hypopnea index, LAT = low arousal threshold, LSpO2 = lowest values of peripheral blood oxygen saturation, ODI = oxygen desaturation index, PSG = polysomnography, REM = rapid eye movement, SBII = sleep breathing impairment index, T90 = percent of time spent at peripheral blood oxygen saturation beneath 90%, TST = total sleep time.The association between SBII and the Framingham CVD riskA total of 64 patients with OSA with relevant data were analyzed to determine the Framingham CVD risks after gathering laboratory data and removing patients with pre-existing CVD. First, we treated SBII as a continuous variable, and logistic regression revealed an elevated proportion of moderate-to-high Framingham CVD risk with an unadjusted odds ratio (OR) of 1.02 (95% confidence interval [CI], 1.00–1.03). The SBII was then divided into quantiles for subsequent regression. According to Table 5, individuals with SBII in the fourth quantile had a 13-fold higher probability of having moderate-to-high Framingham CVD risk (OR, 13.33; 95% CI, 1.42–124.88; P = .023). This association persisted after adjusting for sex (OR, 13.80; 95% CI, 1.33–143.22; P = .028) and appears to persist after adjusting for both sex and BMI (OR, 10.09; 95% CI, 0.91–111.75; P = .059). Similar correlations between the AHI or T90 and the Framingham CVD risk, however, could not be found. The differences in CVD risks among the 3 clusters acquired above were looked at in light of the substantial link between SBII quantiles and CVD risk, and the results are shown in Table 6. Similar to their SBII trends, cluster 3 exhibited a much higher CVD risk than cluster 2. Cluster 3 also had the highest proportion of moderate-to-high CVD risks, although this was not statistically significant.Table 5 The association between SBII and moderate-to-high Framingham CVD risk.SBII QuantileOR (95% CI)Model 0Model 1Model 2First111Second3.56 (0.73, 17.32)3.99 (0.72, 22.20)4.66 (0.80, 26.99)Third1.14 (0.29, 4.51)0.77 (0.18, 3.28)0.63 (0.14, 2.83)Fourth13.33 (1.42, 124.88)*13.80 (1.33, 143.22)*10.09 (0.91, 111.75)*P < .05. Model 1: adjusted for sex as a covariate. Model 2: adjusted for sex and BMI as covariates. BMI = body mass index, CI = confidence interval, CVD = cardiovascular disease, OR = odds ratio, SBII = sleep breathing impairment index.Table 6 The Framingham CVD risks of participants by cluster.Cluster 1Cluster 2Cluster 3PSubjects, n (%)19 (29.69%)32 (50.00%)13 (20.31%)HDL-c, mg/dL42.36 ± 8.4942.88 ± 6.9139.85 ± 7.76.475TC, mg/dL184.35 ± 36.77a208.40 ± 34.50a183.28 ± 27.49.020Framingham CVD risks, %11.20 (4.70–15.60)8.65 (4.33–13.20)b21.60 (7.90–30.00)b.033Moderate-to-high CVD risks, n (%)12 (63.16%)21 (65.63%)12 (92.31%).152Data are presented as mean ± standard deviation or median (interquartile range, 25–75%) unless otherwise indicated; n = 64. aP < .05 between cluster 1 and cluster 2; bP < .05 between cluster 2 and cluster 3. CVD = cardiovascular disease, HDL-c = high-density lipoprotein cholesterol, TC = total cholesterol.DISCUSSIONThis cross-sectional study explores the effectiveness of SBII in assessing OSA severity and prognosis. After using 20 symptoms and the SBII to perform the cluster analysis, 3 distinct groups were shown and several differences among groups were found, especially the divergence in AHI and SBII. Furthermore, the SBII was found to be associated with 10-year CVD defined by Framingham CVD risk, with the fourth SBII quantile having a significant correlation despite correcting for sex, and this association appears to hold after adjusting for BMI, but no such association was identified with traditional measures. According to these results, SBII may be just as effective as AHI in determining the severity of OSA and its prognosis in the future.We detected 3 separate clusters that were similar to what Ye et al12 discovered in their study, with the exception of cluster 3, which is referred to as the "sleep breathing impairment group." Even if we only use SBII as the analytic variable, as we found before, serving as separate predictors in measuring OSA severity, if the trend of SBII coincides with AHI they should exhibit similar performance among the 3 groups. Surprisingly, the severity of OSA assessed by AHI is comparable in all 3 groups; however, the severity assessed by SBII differs significantly in clusters 2 and 3, with barely the second SBII quantile in cluster 2 and nearly the third SBII quantile in cluster 3.Cluster 3, the "sleep breathing impairment group," had a median symptom manifestation compared with the other 2 groups, but a considerably higher likelihood of feeling restless during sleeping, indicating a poor sleep experience. However, because this group had the lowest chances of waking up suddenly and feeling breathless, and stopping breathing reported by spouses, this indicates that, while the process of a full night's sleep is not ideal or comfortable, these individuals can sustain a generally complete and uninterrupted slumber. This observation is also supported by the LAT index of this group of individuals. Members in this cluster are less likely to experience arousals during sleep, which is reflected in various PSG metrics, as they have the lowest LAT score and the lowest chance of being classified as having LAT. The significantly longer maximum time for obstructive apnea, mixed apnea, total apnea, and total respiratory events, as well as the highest ODI score, all contribute to the fact that these participants' responses to the hypoxia burden caused by respiratory events and respiratory stimuli are not as sensitive, resulting in some stability and integrity in sleep.19 As a result of the prolongation of respiratory and desaturation events, SBII increases. Severe PSG features, on the other hand, may indicate a worse prognosis, as evidenced by the occurrence of CVD in this study. However, the number of cases is insufficient to be considered credible, necessitating a bigger sample and prospective research to confirm. We also discovered a much greater percentage of smokers in this group, which could be linked to a high SBII. Smokers exhibited more sleep apneas20 and lower oxygen saturation overnight21 than nonsmokers, according to a recent study of polysomnographic characteristics. As previously indicated, smoking causes upper airway inflammation and collapse22,23; also, smokers have a higher arousal threshold as a result of nicotine.24 When these putative mechanisms are coupled, smoking has been connected to severe SBII.Cluster 2, the "disturbed sleep group," consists of participants who exhibit higher frequency of a variety of symptoms, resulting in exceedingly bad sleep integrity. Sleep fragmentation is induced and aggravated by frequent nocturnal and insomnia-related symptoms, as well as repetitive arousals, resulting in the complaint of daytime dysfunction (decreases in concentration and unresponsiveness). We show that the likelihood of these individuals having an LAT is higher (55.71%), which manifests as significantly shorter periods of respiratory events due to the ease of arousals,25 resulting in a relatively low SBII because the periods of events and degrees of desaturations decrease as explained above. Previous research has found that trazodone and eszopiclone can raise the respiratory arousal threshold and lower the AHI in patients with an LAT.26,27 Patients with insomnia may benefit from a combination therapy that includes CPAP and cognitive behavioral therapy.12,28,29 This could lead to individualized treatment recommendations for patients in this cluster. The comparatively high likelihood of tiredness despite a very low ESS score may imply that ESS is unable to capture the complete range of symptoms.29Cluster 1 was the "minimally symptomatic group," with a significantly decreased likelihood of experiencing symptoms. Individuals in this cluster, however, have an intermediate PSG result as well as SBII level, in contrast to the complaints. They also have a considerably higher PSQI score (8.67 ± 2.79) than clusters 2 and 3 (6.93 ± 2.26 and 6.78 ± 2.31, respectively), indicating worse sleep quality. This could be explained by the fact that members in this cluster tend to seek medical help too late due to silent disease progression, resulting in poor sleep quality and severe PSG parameters during their initial visit, leading to an intermediate degree of SBII. Because of the divergence in AHI and SBII, we may underestimate the severity of OSA in these individuals if we simply utilize AHI as a predictor due to their asymptomatic symptoms but concurrent untreated OSA exposure. As Quan et al30 stated and Lee-Iannotti et al31 summarized, since patients with mild OSA had no clinical complaints different from those without OSA, utilizing AHI alone to define the threshold and evaluate the severity of OSA raises the question of whether a patient actually needs therapy at this level. In conjunction with the findings of this paper, the use of SBII may allow clinicians to identify a group of patients who have the same level of AHI, particularly in mild severity, but a different level of SBII, indicating an actual higher impairment caused by OSA and requiring rapid treatment.Numerous previous research has examined the relationship between OSA and CVD and identified CVD as an important and serious complication of OSA.32–35 During the investigations of the relationships, it was discovered that various nocturnal intermittent hypoxia measurements, particularly T90, were connected to CVD. In the study by Kendzerska et al,36 T90 rather than AHI was demonstrated to be a significant predictor of composite cardiovascular outcome. In line with the assertion made by Oldenburg et al,37 Baumert et al38 also demonstrated that T90, a marker of nocturnal hypoxemic burden, was an independent predictor of cardiovascular mortality. The participants in the aforementioned research, however, were either older with multiple comorbidities or patients with heart failure, suggesting that it may be difficult to generalize the findings to the general public. Furthermore, traditional measurements such as with AHI and T90 only examined 1 pathophysiological element of OSA. As an illustration, T90 only records periods of time when SpO2 is below 90%, whereas AHI only counts the frequency of respiratory events. In comparison, SBII takes into account all of these factors. It is a more thorough metric for determining the severity of OSA since it integrates the frequency and duration of respiratory events, as well as the frequency, duration (both the desaturation and recovery periods), and depth of desaturations linked to a corresponding respiratory event. In this study, the divergence of SBII and AHI in 3 clusters demonstrated a higher sensitivity of SBII in assessing the severity of OSA, and the corr