Obstructive sleep apnea‐related intermittent hypoxaemia is associated with impaired vigilance

Obstructive sleep apnea (OSA)‐related intermittent hypoxaemia is a potential risk factor for different OSA comorbidities, for example cardiovascular disease. However, conflicting results are found as to whether intermittent hypoxaemia is associated with impaired vigilance. Therefore, we aimed to investigate how desaturation characteristics differ between the non‐impaired vigilance and impaired vigilance patient groups formed based on psychomotor vigilance task (PVT) performance and compared with traditional OSA severity parameters. The study population comprised 863 patients with suspected OSA who underwent a PVT test before polysomnography. The conventional OSA parameters, for example, the apnea‐hypopnea index, oxygen desaturation index, and arousal index were computed. Furthermore, the median desaturation area, fall area, recovery area, and desaturation depth were computed with the pre‐event baseline reference and with reference to the 100% oxygen saturation level. Patients were grouped into best‐ and worst‐performing quartiles based on the number of lapses in PVT (Q1: PVT lapses <5 and Q4: PVT lapses >36). The association between parameters and impaired vigilance was evaluated by cumulative distribution functions (CDFs) and binomial logistic regression. Based on the CDFs, patients in Q4 had larger desaturation areas, recovery areas, and deeper desaturations when these were referenced to 100% saturation compared with Q1. The odds ratio (OR) of the median desaturation area (OR = 1.56), recovery area (OR = 1.71), and depth (OR = 1.65) were significantly elevated in Q4 in regression models. However, conventional OSA parameters were not significantly associated with impaired vigilance (ORs: 0.79–1.09). Considering desaturation parameters with a 100% SpO2 reference in the diagnosis of OSA could provide additional information on the severity of OSA and related daytime vigilance impairment.


Summary
Obstructive sleep apnea (OSA)-related intermittent hypoxaemia is a potential risk factor for different OSA comorbidities, for example cardiovascular disease. However, conflicting results are found as to whether intermittent hypoxaemia is associated with impaired vigilance. Therefore, we aimed to investigate how desaturation characteristics differ between the non-impaired vigilance and impaired vigilance patient groups formed based on psychomotor vigilance task (PVT) performance and compared with traditional OSA severity parameters. The study population comprised 863 patients with suspected OSA who underwent a PVT test before polysomnography. The conventional OSA parameters, for example, the apnea-hypopnea index, oxygen desaturation index, and arousal index were computed. Furthermore, the median desaturation area, fall area, recovery area, and desaturation depth were computed with the preevent baseline reference and with reference to the 100% oxygen saturation level.
Patients were grouped into best-and worst-performing quartiles based on the number of lapses in PVT (Q1: PVT lapses <5 and Q4: PVT lapses >36). The association between parameters and impaired vigilance was evaluated by cumulative distribution functions (CDFs) and binomial logistic regression. Based on the CDFs, patients in Q4 had larger desaturation areas, recovery areas, and deeper desaturations when these were referenced to 100% saturation compared with Q1. The odds ratio (OR) of the median desaturation area (OR = 1.56), recovery area (OR = 1.71), and depth (OR = 1.65) were significantly elevated in Q4 in regression models. However, conventional OSA parameters were not significantly associated with impaired vigilance (ORs: 0.79-1.09). Considering desaturation parameters with a 100% SpO 2 reference in the diagnosis of OSA could provide additional information on the severity of OSA and related daytime vigilance impairment.  (Benjafield et al., 2019). In OSA, relaxation of the upper airway muscles during sleep causes repetitive breathing obstructions that lead to episodes of hypoxaemia (i.e. oxygen desaturations). In addition, breathing obstructions often lead to sleep fragmentation, mood swings, and excessive daytime sleepiness (Hossain et al., 2002;Mitler et al., 1988). Moreover, inadequate sleep decreases productivity, work performance, and daytime vigilance. Additionally, patients with severe obstructive sleep apnea have a higher probability of increased morbidity and risk for motor vehicle accidents (Barbe et al., 1998;Brookings et al., 1996;Körber et al., 2015;Reinerman-Jones et al., 2016).
Polysomnography (PSG) is the primary method utilised in the diagnosis of obstructive sleep apnea (Mulgrew et al., 2007). The apnea-hypopnea index (AHI), oxygen-desaturation index (ODI), and arousal index (ArI), which can be defined from the PSG, are predominantly used as a clinical index for OSA severity (Berry et al., 2012).
However, several studies suggest that these parameters might not be the most optimal for the assessment of the severity of OSA Leppänen et al., 2016;Muraja-Murro et al., 2014;Pevernagie et al., 2020). In contrast, previous studies show a significant association between the desaturation severity (calculated based on desaturation areas) and OSA severity (Azarbarzin et al., 2019;Gami et al., 2007;Kulkas et al., 2017;Leppänen et al., 2017;Muraja-Murro et al., 2013;Trzepizur et al., 2022). The desaturation area describes the entire area from the onset of an oxygen desaturation to the point when resaturation again reaches the baseline and the recovery area describes only the resaturation segment where the oxygen saturation level reaches the baseline from the nadir (Figure 1).
Vigilance is defined as the ability to remain attentive and to sustain alertness to a particular task over a period of time (Parasuraman et al., 1987). The psychomotor vigilance task (PVT) is a 10 min test which is regarded as a good indicator of fatigue and cognitive impairment (Lee et al., 2010). The psychomotor vigilance task measures the reaction time (RT) to visual stimuli. If the reaction time exceeds 500 milliseconds that is considered a lapse or a microsleep. Although PVT is commonly used in sleep deprivation studies, its use in patients with obstructive sleep apnea is not as widespread. Studies that have examined PVT performance in OSA show a poor correlation between PVT performance and conventional OSA severity parameters (Batool-Anwar et al., 2014;Kainulainen et al., 2019;Kainulainen, Duce, et al., 2020;Kim et al., 2007;Lee et al., 2010;Li et al., 2017;Sforza et al., 2004;Tanno et al., 2017). Intermittent hypoxaemia appears to have the strongest association with PVT performances in patients with obstructive sleep apnea (Kainulainen, Duce, et al., 2020;Sforza et al., 2004;Tanno et al., 2017).
Indeed, recent studies have shown that desaturation parameters are better at describing OSA severity than conventional OSA parameters (Kainulainen et al., 2019;Kulkas, Tiihonen, Eskola, et al., 2013;. In addition, oxygen saturation characteristics from the pre-event baseline described that severe desaturations have a stronger association with impaired vigilance in patients with obstructive sleep apnea (Kainulainen, Duce, et al., 2020). Furthermore, the oxygen saturation characteristics derived from the 100% reference level have shown cardiovascular consequences in patients with obstructive sleep apnea (de Chazal et al., 2021). However, referencing the 100% SpO 2 level is not commonly considered when evaluating desaturation events and may also provide further F I G U R E 1 Visualisation of the desaturation parameters computed in this study. SpO 2 , blood oxygen saturation; DesaturationArea100, desaturation area with 100% reference; FallArea100, fall area with 100% reference; RecoveryArea100, recovery area with 100% reference; Depth100, desaturation depth with 100% reference insights into OSA severity characteristics. Additionally, the role of recovery areas of oxygen desaturations with respect to daytime symptoms has not been studied thoroughly. These parameters could provide novel and additional information for the diagnosis of obstructive sleep apnea.
Hence, we aimed to investigate our hypothesis of whether a more detailed characterisation of desaturation events from the pre-event baseline and the 100% reference level provides new insight into the causes of impaired vigilance in patients with obstructive sleep apnea. Desaturations were initially scored using the 3% criterion from the onset of the oxygen desaturation to the point where the saturation values returned to the normal baseline that is, the pre-event baseline reference ( Figure 1). As SpO 2 values below 50% are clinically unreliable, they were considered artifacts (Chan et al., 2013). The Ethical Committee of Princess Alexandra Hospital has approved the use of data related to this retrospective study (HREC/16/QPAH/021 and LNR/2019/QMS/54313).
The psychomotor vigilance tasks were conducted for all patients in the evening before the PSGs. PVTs were performed with the Psychology Experiment Building Language program on an ASUS (Taipei, Taiwan) transformer pad using a 10 min protocol (Mueller & Piper, 2014). Patients were instructed to use their thumb or index finger to respond to visual stimuli (n = 120) appearing every 2 to 10 s as fast as possible. From the recorded reaction times, we calculated the number of lapses, that is, the number of responses with over 500 ms reaction times. Patients who had missing background information or incomplete PSG or PVT data (n = 49) were excluded from the study.
Thus, a total of 863 patients were included in the analyses.

| Parameters
Next, all SpO 2 signals, scored apnea, hypopnea, arousal, and desaturation event information (e.g., event start time, durations, end time), and sleep stages were imported to MATLAB (R2020b; MathWorks; Natick, MA, USA) for further analyses. The total sleep time was calculated based on the scored sleep stages. To describe the severity of obstructive sleep apnea, we calculated the conventional parameters used for OSA diagnostics (i.e. AHI, ODI, and ArI). Furthermore, we calculated several SpO 2 -based parameters including average SpO 2 (Avg. SpO 2 ), time spent with SpO 2 under 90% (t90), and median desaturation characteristicsdesaturation area, fall area, recovery area, and desaturation depth (Table 1). Fall areas were calculated from the onset of an oxygen desaturation to nadir and recovery areas from the nadir to the end of that desaturation (i.e. to the point where oxygen saturation level reached the baseline). Desaturation areas were defined as a sum of the fall and recovery areas. In addition, the median desaturation area (DesaturationArea100), fall area (FallArea100), recovery area (RecoveryArea100), and desaturation depth (Depth100) were similarly calculated from the 100% SpO 2 reference level ( Figure 1).

| Statistical analyses
The included 863 patients were divided into quartiles (Q1-Q4) based on the number of lapses in psychomotor vigilance task. Since there are no general guidelines or thresholds on how to define impaired vigilance based on the PVT, and as our goal was to investigate which parameters best explain the impaired vigilance, we only considered the quartiles Q1 and Q4, that is the best and worst-performing patients. The patients in Q1 (n = 255) were considered the nonimpaired vigilance group and the patients in Q4 (n = 223) the impaired vigilance group. In this way, there is a clear division between the groups as it is difficult to define whether patients with average PVT values, that is patients in quartiles Q2 and Q3, have impaired vigilance or not. Demographic and PSG parameters were compared between the Q1 and Q4 using the chi-square test and Mann-Whitney U test. Furthermore, empirical cumulative distribution functions (CDFs) were formed for all parameters and compared between the Q1 and Q4 using the Kolmogorov-Smirnov test.
Two different binomial logistic regression models were utilised to investigate the association between the parameters and impaired T A B L E 1 Definitions of the parameters utilised in this study

| RESULTS
The studied population of 863 patients consisted of 44.6% females (n = 385) and the median age in the whole population was 56.0 years (Table 2). When patients were divided into quartiles based on the number of lapses, the impaired group consisted of 59.7% of females and the non-impaired group of 29.9% (p < 0.001). The impaired group was also older but no differences were found in total sleep time, AHI, ODI, or ArI parameters between the groups. The Avg. SpO 2 value was found to be significantly smaller and t90 higher in the Q4 compared with the Q1 (Table 2, Figure 2). In addition, when the 100% reference level was applied, median desaturations were significantly deeper, and median desaturation and recovery areas significantly larger in the Q4 compared with the Q1.
Cumulative distribution functions were evaluated for all parameters. Similarly, we observed that Avg. SpO 2 , t90, and desaturation parameters with 100% reference were statistically significant (p < 0.05) between Q1 and Q4. Patients in Q4 had larger median desaturation areas, recovery areas, and deeper desaturations when the 100% reference level was considered compared with patients in Q1 (Figure 3).
Unadjusted binomial logistic regression models showed that Avg.
SpO 2 , t90, DesaturationArea100, RecoveryArea100, and Depth100 were significantly associated with impaired vigilance (Table 3). These same parameters were also significantly associated with impaired vigilance in the adjusted models (Table 3).

| DISCUSSION
In this study, we carried out a detailed investigation of SpO 2 -based parameters and conventional OSA parameters. The results reveal a strong association between desaturation parameters and impaired vigilance, especially when considering the 100% SpO 2 reference level.
When the 100% reference level was considered, patients with impaired vigilance (Q4) had significantly larger median desaturation and recovery areas, and deeper desaturations compared with nonimpaired patients (Q1). In addition, the average SpO 2 was lower and thus, t90 higher in patients with impaired vigilance. In contrast, the differences in conventional OSA severity parameters (i.e., AHI, ODI, and ArI) between impaired and non-impaired patients were not statistically significant.
AHI, ODI, and ArI consider only the number of events per hour of sleep. Therefore, conventional metrics do not take into account the F I G U R E 2 The morphology of desaturation events. The significant difference between the non-impaired Q1 (n = 225) and impaired groups Q4 (n = 223) was computed with the Mann-Whitney U test and is indicated by an arrow. Notice different scales were used for each parameter.
Avg. SpO 2 , average oxygen saturation; DesaturationArea100, desaturation area with 100% reference; RecoveryArea100, recovery area with 100% reference; Depth100, desaturation depth with 100% reference; t90, time below 90% oxygen saturation level, Q1M, Q1 median; Q4M, Q4 median; No, Number. morphology, duration, or depth of events. In contrast, based on the present results it is evident that the areas and depths of desaturation events provide important diagnostic information in patients with obstructive sleep apnea that is not available through conventional metrics. It is also reported that arousals are not always present with obstruction events (Younes, 2004). Furthermore, studies reported poor intraclass correlation coefficients for arousal scoring and greater variability in the scoring of arousal, making arousal a less reliable index altogether (Drinnan et al., 1998;Kuna et al., 2013). Additionally, in the present study, ArI did not significantly differ between impaired and non-impaired patients, while the severity of desaturation events revealed a difference. This is in agreement with previous studies showing that patients with impaired daytime vigilance had more severe nocturnal hypoxaemia and a lower level of oxygen saturation (Bedard et al., 1991;Kainulainen, Duce, et al., 2020;Tanno et al., 2017). Furthermore, Tanno et al. also reported a similar connection between the severity of intermittent hypoxaemia and psychomotor vigilance task performances in a population with much lower ODI and BMI compared with our study (Tanno et al., 2017). It is well known that obesity is associated with daytime sleepiness, which might be a reason for a higher number of PVT lapses in our studied population (Vgontzas et al., 2006). Although the definition of impaired vigilance has varied between Tanno et al. study and our study, both indicate that the severity of hypoxaemia is strongly associated with impaired vigilance despite the heterogeneous population in both studies. Thus, it is important to provide additional and detailed information on desaturation event characteristics while evaluating the OSA severity and its effects on daytime vigilance.
Another important aspect of our findings is that the 100% SpO 2 reference level plays a significant role while determining the desaturation severity. Throughout our entire study, we observed that impaired vigilance was more strongly associated with desaturation parameters calculated by applying the 100% reference level compared with normal baseline reference (i.e., desaturation from the pre-event baseline reference). This suggests that even if the differences in desaturation parameters are small with normal baseline reference, desaturation parameters determined using the 100% reference may reflect the true level of desaturation severity better. Moreover, it has previously been shown that the desaturation areas calculated from the pre-event baseline are strongly associated with cardiovascular morbidity, daytime sleepiness, and impaired vigilance (Azarbarzin et al., 2019;Kainulainen, Duce, et al., 2020;, while the present study highlights that utilising the 100% SpO 2 reference level could make this association even stronger (de Chazal et al., 2021).
The results from logistic regression analyses also revealed that the severity of intermittent hypoxaemia was significantly elevated in patients with impaired vigilance. The most elevated odds ratios were caused by DesaturationArea100, RecoveryArea100, and Depth100 in both models, indicating a strong correlation between the novel parameters and impaired vigilance ( Table 3). The Avg. SpO 2 and t90 were also significantly associated with impaired vigilance. However, the conventional SpO 2 parameters do not consider the morphology or depth of F I G U R E 3 Empirical cumulative distribution functions of desaturation events. The significant difference between the non-impaired Q1 (n = 225) and impaired groups Q4 (n = 223) was computed with the Kolmogorov-Smirnov test and is indicated by an arrow. Notice different scales were used for each parameter. Avg. SpO 2 , average oxygen saturation; DesaturationArea100, desaturation area with 100% reference; RecoveryArea100, recovery area with 100% reference; Depth100, desaturation depth with 100% reference; t90, time below 90% oxygen saturation level. each desaturation event. Thus, considering desaturation characteristics together with conventional SpO 2 -based parameters can provide more insights related to the severity of OSA-associated intermittent hypoxaemia. In addition, several studies have reported that sleep deprivationrelated impaired vigilance is a potential risk factor for cognitive performance impairment (Doran et al., 2001;Gunzelmann et al., 2009;Odle-Dusseau et al., 2010). Another study revealed that cognitive memory decline in patients with obstructive sleep apnea is more strongly linked with oxygen desaturation compared with AHI or ArI (Hrubos-Strøm et al., 2012). Also, the risk of road accidents is quite high in patients with OSA and impaired vigilance and early recognition of intermittent hypoxaemia can reduce the consequences of obstructive sleep apnea, thus improving the quality everyday life (Barbe et al., 1998;Bedard et al., 1991). Therefore, considering the risk of several factors such as traffic-related accidents together with our present findings and other previous studies suggest that desaturation parameters should be considered to better predict obstructive sleep apnea severity in patients with impaired vigilance (Kainulainen et al., 2019;. Despite the strengths of this study, it has certain limitations. The patients involved in this study underwent the psychomotor vigilance task tests before the polysomnographies. The psychomotor vigilance task tests conducted in the morning after the polysomnography could better describe the quality of sleep, and thus, the polysomnography findings. Furthermore, sleep stages or sleeping positions were not considered while evaluating the desaturation characteristics. As sleep stages and sleeping positions affect the desaturation characteristics, a detailed investigation of such factors could reveal valuable information (Choi et al., 2016). Lastly, we had limited information related to the sleeping habits or medication history of the patients. The sleeping habits prior to polysomnography evaluation, the use of psychotropic drugs, or caffeine can affect the psychomotor vigilance task results.
However, the objective of this study was to investigate SpO 2 signal characteristics of patients with obstructive sleep apnea and impaired vigilance regardless of medication history, thus the lack of information on these can only have slight impact on the present study.

| CONCLUSIONS
This study highlights the importance of considering the level of oxygen saturation and desaturation event characteristics in more detail when assessing the severity of obstructive sleep apnea. When the 100% oxygen saturation level was considered as a reference, the

DATA AVAILABILITY STATEMENT
The data includes medical records and personal information and therefore the data can only be shared within the confinements of the Australian legislation and ethical conventions. Reasonable requests considering data sharing will be individually assessed.