Obstructive sleep apnea (OSA) is a common sleep disorder characterized by recurrent upper airway blockages, leading to reduced oxygen levels and disrupted sleep patterns [1]. Approximately 1 billion individuals aged 30 to 69 worldwide are estimated to be affected by OSA [2]. Risk factors for OSA include male sex, obesity, aging, menopause, and impairments in upper airway anatomy or muscle function [3,4]. The exact cause of OSA is not fully understood, but it is believed to result from upper airway obstruction due to negative pressure collapse during inspiration and gradual narrowing in the retro-palatal area during exhalation [5]. Symptoms commonly exhibited by patients with OSA include snoring, gasping for air during sleep, daytime sleepiness, morning headaches, and cognitive impairment [6].

The clinical presentation and severity of OSA can vary among individuals due to differences in OSA-related parameters, as well as variations in the duration and type of sleep. A significant distinction in OSA lies in the relationship between the supine apnea-hypopnea index (AHI) and the non-supine AHI, categorizing patients into positional and nonpositional OSA groups. Positional dependent OSA was first described by Cartwright, et al. [7] in 1991, identifying patients with a supine AHI more than twice their non-supine AHI as having positional dependent OSA. This categorization, based on the propensity for apneic events to occur in supine versus non-supine sleeping positions, is clinically significant and carries implications for diagnostic, therapeutic, and prognostic considerations [8,9]. Oksenberg, et al. [8] found that OSA patients who experience more severe apneic events in the supine position may benefit from positional therapy, which encourages sleeping in a non-supine position. Additionally, an inverse relationship between body mass index (BMI) and positional dependency suggests that weight loss may be beneficial for individuals with non-positional dependency [9].

Although the differentiation between positional and non-positional OSA is increasingly recognized, comprehensive investigations into the disparities between these groups are imperative. This study aims to identify differences in the clinical features of positional dependent OSA and analyze treatment methods based on positional dependency.

Positional OSA, which entails a 50% or greater reduction in the AHI between supine and non-supine sleep positions, has been widely adopted in many studies [7]. In a South Korean study of 1052 OSA adults, positional OSA was found in 75.6% of cases [14]. Our study, involving 558 OSA patients, also showed a high prevalence, with 76.70% having positional OSA. This finding is consistent with another cohort study of 1224 OSA participants, in which 75% were identified with positional OSA [15].

Prior to data adjustment using PSM, the mean AHI for positional OSA was 28.47±16.56, whereas for non-positional OSA, it was 56.04±29.83. A study conducted by Beyhan Sagmen and Cömert [16] reported similar results, with non-positional OSA exhibiting greater AHI severity, presenting an AHI of 39.99±27.52 compared to positional OSA, which had an AHI of 15.56±10.61 (p<0.001). However, comparing clinical features using raw data would yield statistically inaccurate and biased results due to the large differences in group size and average AHI. Thus, we conducted PSM to select patients with similar AHI values, enhancing the accuracy of our results. This approach allowed us to compare the specific influence of positional and non-positional factors in OSA patients, excluding the impact of OSA severity. After performing PSM to balance the two OSA groups, we observed that patient demographics and clinical characteristics, including BMI and neck/waist/hip circumferences, were similar between the positional and non-positional OSA groups. Using the R program, pre- and post-PSM variables were compared to check for changes in distributions. Except for AHI, which was the variable used for matching the two groups, the distributions of age, BMI, neck, waist, hip circumference, average and minimum SpO2, and ESS were similar (Fig. 2). It is noteworthy that some studies have suggested that patients with positional OSA tend to exhibit lower BMI or smaller neck circumference [15,17]. But none of these studies accurately matched AHI between the two groups.

In our study, participants with non-positional OSA had significantly higher ESS scores compared to those in positional OSA group. Pearson correlation analysis revealed a positive link between ESS scores and AHI values in non-positional OSA, even when controlling for BMI. Temirbekov, et al.’s [18] study also supported this correlation in patients with moderate or severe OSA. This underscores the significance of assessing sleepiness as a potential marker for OSA severity in non-positional cases. In contrast, positional OSA may not demonstrate a significant symptom severity-AHI association. Thus, it’s crucial to inform patients that even with milder symptoms, they could have severe OSA requiring treatment based on AHI severity.

We also found that both OSA groups exhibited similar sleep quality and sleep stage distribution, except for oxygen levels. During sleep, the positional OSA group displayed significantly higher average and minimum SpO2 levels, and significantly lower T90, suggesting better oxygenation throughout the night. This difference may also relate to variations in symptom severity between the two groups. Furthermore, the severity of subjective symptoms could have influenced the utilization of APAP therapy.

Positive airway pressure (PAP) is the primary treatment for OSA, but compliance can be challenging due to issues such as mask discomfort, claustrophobia, pressure intolerance, and lifestyle factors. Compliance with APAP varies across studies; for instance, Dias, et al. [19] reported lower compliance among the positional OSA group. However, in our study, although the non-positional OSA group utilized APAP more frequently (proportion of days using APAP, p=0.009), we observed no significant differences in compliance between the two groups (p=0.063). The increased utilization of APAP in the non-positional OSA group may be associated with higher ESS scores. However, compliance with APAP therapy seems to be affected by factors more complex than just subjective symptoms and positional factors. Therefore, it may be beneficial to consider additional psychological and physiological conditions when encouraging the use of APAP therapy.

We subdivided OSA patients into mild, moderate, and severe categories to analyze factors related to severe OSA. Before PSM, most patients in both groups had severe OSA. We had 15, 17, and 98 cases (11.5%, 13.1%, and 75.4%) for mild, moderate, and severe OSA in the non-positional group respectively, and 100, 162, and 166 cases (23.4%, 37.9%, and 38.8%) in the positional OSA group respectively. However, after PSM matching, both groups had an equal distribution of OSA severity. Logistic regression revealed higher BMI as a common risk factor for severe OSA in both groups. However, the groups differed in other factors. In the positional OSA group, age had a substantial impact on severity, while male sex and retropalatal obstruction were related to severity in the non-positional OSA group. Therefore, these results suggest that there may be differences in OSA pathophysiology between the two groups.

In our study, the non-positional OSA group showed an association with BMI and anatomical narrowing leading to upper airway obstruction such as retropalatal obstruction. Supporting this observation, Lee, et al. [20] noted that changing from supine to lateral sleep positions improved tongue base and larynx obstructions, linking lateral position obstruction to oropharyngeal lateral wall collapsibility. Lan et al.’s [21] multivariate analysis identified oropharyngeal lateral wall collapse as the primary anatomical predictor for non-positional OSA. Notably, 62.5% of these patients had more severe oropharyngeal lateral wall collapse in the lateral position.

In the positional OSA group, age may be associated due to factors like decreased muscle tone in the supine position, leading to increased pharyngeal collapsibility [22]. Several studies suggest that pharyngeal collapsibility in older patients is due to increased pharyngeal fat pads, lengthening of the soft palate, and ultimately narrowing the upper airways. Additionally, a decrease in the negative pressure reflex of the genioglossus muscle, a mechanism that compensates for pharyngeal collapse, has been observed as individuals age [23]. Aging is also associated with an elevated end expiratory lung volume (EELV), which can influence the activity of dilator muscles. Another study by Stanchina, et al. [24] demonstrated that genioglossus muscle activity declines as EELV increases. Furthermore, the aging process is linked to a reduction in the elastic recoil of the lungs, resulting in an overall increase in EELV [25]. As identified in previous studies, multiple factors are related to causing positional and non-positional OSA.

Our study offers several advantages. Firstly, while many studies have compared positional and non-positional OSA groups with significant differences in AHI, our study conducted a comparison between the two groups with similar AHI values using PSM. Additionally, we sought to understand the pathophysiology underlying AHI and OSA severity in each group, going beyond mere clinical differences. Moreover, beyond assessing clinical features, we also evaluated APAP utilization, providing valuable insights for comparing the two groups. However, this study has important limitations to address. Due to the large difference in the number of participants between the two groups (positional OSA: 428 vs. non-positional OSA: 130), the reliability of the analysis results before applying PSM was questioned. To address this issue, we decided to use 1:1 PSM based on AHI values. By matching each participant through 1:1 matching, we were able to precisely adjust the balance between the two groups. However, this adjustment resulted in a significant reduction in sample size post-PSM, from 558 to 180. Additionally, the statistical power of our study is lower compared to utilizing 1:2 matching, which is a notable limitation. Therefore, for future analyses, a larger sample size and the utilization of 1:2 or 1: n matching of AHI scores should be considered when comparing the positional and non-positional OSA groups. In addition, due to the nature of the retrospective study, there was limited information on factors affecting APAP compliance. Thus, further exploration of the various factors affecting APAP compliance is warranted. Further research, fundamentally examining the physiological differences between positional and non-positional AHI, is essential to comprehend the pathophysiological distinctions between the groups.

In conclusion, we found that non-positional OSA patients exhibited more severe symptom scores compared to those with positional OSA. Although a higher proportion of APAP usage was observed in the non-positional OSA group, no significant differences in APAP compliance were found between the two groups. Higher BMI was a common risk factor for severe OSA in both groups. However, age was a significant risk factor in severe positional OSA, while sex and retropalatal obstruction were significant factors in severe non-positional OSA. This suggests that there might be differences in the pathophysiology of non-positional and positional OSA. Further research to elucidate the pathophysiological characteristics of these two groups in more detail would be beneficial.