Abstract
BACKGROUND: Obesity is one of the most prominent risk factors for obstructive sleep apnea (OSA). Weight loss decreases the number of shorter respiratory events (<40 s), whereas the number of longer events remains almost unchanged. However, it is unknown how body mass index (BMI) affects individual obstruction event severity within OSA severity categories when age, sex, smoking, daytime sleepiness, snoring, hypertension, heart failure, and sleeping posture are considered. Therefore, we investigated whether individual obstruction event severity varies with BMI within the OSA severity categories when considering the most important confounding factors.
METHODS: Polygraphic recordings of 723 subjects without CPAP treatment and with an apnea-hypopnea index (AHI) of ≥5 events/hour were reanalyzed retrospectively. The effect of BMI on the severities of OSA and individual obstruction events was evaluated in general, within OSA severity categories, and between different BMI groups (ie, BMI < 25; 25 ≤ BMI < 30; 30 ≤ BMI < 35; 35 ≤ BMI < 40; and BMI ≥ 40 kg/m2).
RESULTS: AHI increased in mild and severe (β ≥ 0.10, P < .001) OSA categories, with increasing BMI. However, the proportion of apneas from all respiratory events decreased (β = −0.55, P = .068) with increasing BMI in all the OSA categories. An increase in BMI led to a decrease in the median duration of individual apneas, hypopneas, and desaturations in all OSA categories, whereas desaturation depth increased statistically significantly in the severe category (β = 0.20, P < .001).
CONCLUSIONS: Because BMI is related to the duration of individual obstruction events, its effect on OSA severity is more complex than its effect on AHI would implicate. Therefore, overall severity of OSA may not be the same for non-obese patient and obese patient who have similar AHI. Thus, these patient-specific characteristics of individual breathing cessations should be considered when estimating the severity of disease and risk of related adverse health effects.
Introduction
Obesity is one of the most prominent risk factors for obstructive sleep apnea (OSA); several studies demonstrate a strong association between being overweight and OSA.1,2 In individuals who are obese and individuals who are morbidly obese, the prevalence of OSA has been estimated to be ≥30% and 50–98%, respectively.3,4 Furthermore, it has been estimated that 60–90% of adults with OSA are overweight and that individuals who are obese have over 10-fold probability of developing OSA compared with individuals who are not obese.3 However, the link between obesity and OSA is complex, and increased body mass index (BMI) affects breathing in many ways. For example, pharyngeal size and upper-airway muscle force can be reduced or upper-airway structure can be altered due to an increased amount of adipose tissue in the upper airway or its muscles.3,5 Even though it has been shown that OSA severity increases with weight gain and decreases with weight reduction,6,7 the effect of weight change on the severity of OSA might not be as straightforward as conventional apnea-hypopnea index (AHI), used to estimate the severity of OSA, indicates. Previously, we showed that the level of weight change affects the number but also the duration of individual obstruction events (ie, apneas, hypopneas, and desaturations).7 It has been shown that weight reduction decreases and weight gain increases the number of shorter obstruction events, whereas the number of longer obstruction events tended to remain almost unchanged.7,8
In addition to BMI, age, male sex, and smoking are well-known risk factors for OSA.9–11 As with BMI, OSA has been connected to cardiovascular outcomes (eg, hypertension), and its most common symptoms are excessive daytime sleepiness and snoring.12–14 The number and severity of obstruction events have been shown to differ between sleeping positions; in the supine position, they are increased compared with lateral positions.10,15,16 In addition to the fact that obesity is an independent risk factor for cardiovascular disease,17 the severity of individual obstruction events is also linked to an increased risk of cardiovascular morbidities in patients with moderate-to-severe OSA.18
However, it has not been previously investigated how a change in BMI affects the severity of individual obstruction events within different OSA severity categories (ie, mild, moderate, severe). Therefore, because both BMI and severity of individual obstruction events have been linked to an elevated risk of cardiovascular outcomes, the information on the effect of BMI on the severity of individual obstruction events might enhance understanding of the link between BMI and overall severity of OSA. In the present study, we investigated the effect of BMI on the severity of individual obstruction events adjusted for age, sex, smoking habits, excessive daytime sleepiness, snoring, hypertension, heart failure, and proportion of supine sleep in general, and also within the different OSA severity categories. We hypothesized that an increase in BMI would increase the severity of OSA by increasing the number and the severity of individual obstruction events.
QUICK LOOK
Current Knowledge
Obesity is strongly associated with the severity of obstructive sleep apnea (OSA) when the severity is estimated based on the apnea-hypopnea index (ie, number of obstruction events). Even though the severity estimation of OSA is improved by considering also the duration and morphology of individual obstruction events, the effect of the body mass index on the severity of individual obstruction events has not been previously investigated within the different OSA severity categories.
What This Paper Contributes To Our Knowledge
An increased body mass index increased the frequency of apneas and hypopneas but decreased their duration. The effect of the body mass index on OSA severity may not be as straightforward as its effect on apnea-hypopnea index would suggest. Therefore, the severity of OSA may differ between patients with a different body mass index but similar apnea-hypopnea index.
Methods
During the years 1992–2003, polygraphic recordings of 2,057 patients with clinically suspected OSA were conducted by using custom-made 4-channel ambulatory devices without electroencephalography registration.19,20 The devices recorded air flow with a thermistor, abdominal respiratory movements, blood oxygen saturation, and sleeping position. Furthermore, all the recordings were manually reanalyzed during the years 2012–2015 by using RemLogic software (Embla, Thornton, Massachusetts) in accordance with standard respiratory rules provided by the American Academy of Sleep Medicine.21 Based on the 2007 recommendations, and because electroencephalography was not recorded, hypopneas were scored using 4% threshold for desaturations (rule 4A).21 A favorable statement on the data collection was given by The Research Ethics Committee of the Hospital District of Northern Savo, Kuopio, Finland (decision 127/2004 and 24/2013).
Within the whole population, 568 male and 155 female subjects with an AHI of ≥ 5 events/hour and not treated with CPAP. These 723 subjects were included in further analysis. All subjects were classified into mild OSA (ie, 5 ≤ AHI < 15 events/hour), moderate OSA (ie, 15 ≤ AHI < 30 events/hour), or severe OSA (ie, AHI ≥ 30 events/hour) categories based on conventional AHI. In addition to AHI, the values of oxygen desaturation index (ODI), apnea index, hypopnea index, obstruction severity parameter,20 and proportion of apneas from all breathing cessation events were calculated from the reanalyzed polygraphic recordings. The obstruction severity parameter is defined as the sum of the products of respiratory event duration and related desaturation event area divided by index time (ie, sleeping time).20 Durations of individual apneas and hypopneas, and durations, depths, and areas of individual desaturations were calculated by using custom-made MATLAB functions (Mathworks, Natick, Massachusetts).20 These calculations were based on manually scored events, as illustrated in Figure 1. Basic anthropometric data and information on smoking history, heart failure, hypertension, snoring, and excessive daytime sleepiness were collected from subjects' medical records in Kuopio University Hospital. Before this information was used in analyses, variables were categorized. Smoking was categorized into 3 classes: “yes,” “no” or “quit”; and heart failure, hypertension, snoring, and excessive daytime sleepiness were categorized into 2 classes: “yes” or “no.” If a subject did not have a history of heart failure, hypertension, snoring, or suffered from excessive daytime sleepiness before the polygraphic recording, then these variables were categorized as “no” and, otherwise, the classification was set to be “yes.” Next, the subjects were divided into different BMI groups (ie, Normal: BMI < 25; Overweight: 25 ≤ BMI < 30; Obese: 30 ≤ BMI < 35; Severe Obesity: 35 ≤ BMI < 40; and Morbid Obesity: BMI ≥ 40 kg/m2) and the calculated polygraphic parameters and severity of individual obstruction events were compared between the groups.
The statistical significance of differences in demographic data between different OSA severity categories was evaluated by using the Mann-Whitney U test for continuous variables and the chi-square test for categorical variables. The effect of BMI on OSA parameters (ie, AHI, ODI, apnea index, hypopnea index, and obstruction severity parameter) and the severity of individual obstruction events within different OSA severity categories were evaluated by using a general linear model univariate analysis adjusted for age, sex, smoking habits, snoring, excessive daytime sleepiness, heart failure, hypertension, and the proportion of supine sleep. The general linear model was the statistical method of choice because it can be used in situations when several dependent variables are correlated and values of predictor variables are not linearly independent. Furthermore, it is normally used, unlike the multiple regression analysis, in cases when predictor variables include both continuous and categorical variables. The statistical significance of differences in anthropometric data and background information between the BMI groups was evaluated by using the Mann-Whitney U test and the chi-square test. Post hoc tests (Tukey's honestly significant difference test for equal variances and the Games-Howell test for unequal variances) were used to investigate the statistical significance of the differences in polygraphic data between the BMI groups. One-way analysis of variance ie, ANOVA (for equal variances) and the Welch test (for unequal variances) were used to evaluate the statistical significance of the trend of variation in polygraphic parameter values between the BMI groups. SPSS version 20.0 software (SPSS, Chicago, Illinois) was used for statistical analyses, and P < .05 was set to be the limit of statistical significance.
Results
Based on the AHI, 329, 140, and 99 male, and 102, 34, and 19 female subjects were categorized as having mild, moderate, and severe OSA, respectively (Table 1). The sex distribution did not differ statistically significantly between any of the severity categories (P = .16) (Table 1). In mild, moderate, and severe OSA groups, the median ages were 51.2, 51.2, and 50.7 y, and the median BMIs were 28.8, 30.1, and 33.3 kg/m2, respectively (Table 1). Age was not statistically significantly different (P = .63) between the OSA severity categories, but BMI increased statistically significantly towards more-severe OSA categories (mild-moderate: P = .001 and moderate-severe: P = .002) (Table 1). The incidence of heart failure, hypertension, snoring, and excessive daytime sleepiness as well as information on smoking status, the median values of AHI, ODI, apnea index, hypopnea index, obstruction severity parameter, and the severity of individual obstruction events are presented in Table 1.
AHI, ODI, apnea index, hypopnea index, obstruction severity parameter, and the severity of individual obstruction events were strongly associated with BMI in the entire cohort (Table 2). In more-detailed analysis, values of AHI, ODI, and hypopnea index increased statistically significantly (β = 0.10, P = .002) with increasing BMI in all the OSA categories, except in the moderate OSA category, in which an increase in AHI and ODI did not reach the limit of statistical significance (β = 0.047, P = .20) (Table 2). In contrast, an increase in BMI decreased the apnea index (β = −0.056, P = .02) in mild and moderate OSA categories and the proportion of apneas (β = −0.545, P = .068) in all OSA categories (Table 2). Median durations of individual apneas decreased in all OSA categories (mild, β = −0.245, P = .003; moderate, β = −0.334, P < .001; severe, β = −0.065, P = .42) with increasing BMI as did the median durations of individual hypopneas (mild, β = −0.273, P < .001; moderate, β = −0.110, P = .21; severe, β = −0.318, P < .001) (Table 2).
The same was also observed with the median durations of desaturations (mild, β = −0.236, P = .004; moderate, β = −0.115, P = .18; severe, β = −0.240, P = .005) (Table 2). Furthermore, areas of individual desaturation events decreased (β = −0.737, P = .007) in the mild OSA category with increasing BMI, whereas the depth of desaturations did not change statistically significantly (β = 0.000, P = .97) (Table 2). In contrast, the depths and areas of individual desaturation events increased (β = 0.201, P = .040) in the severe OSA category with increasing BMI (Table 2). Despite the changes in the severity of individual obstruction events, the obstruction severity parameter increased with increasing BMI in a statistically significant manner only in the severe OSA category (β = 2.809, P = .004) (Table 2).
The subjects with BMI ≥ 25 kg/m2 tended to sleep less in the supine position and had a higher rate of heart failure and hypertension than subjects of normal weight (BMI < 25 kg/m2) (see the supplementary materials at http://www.rcjournal.com). No statistically significant difference was seen between the BMI groups in smoking history, snoring, and excessive daytime sleepiness. The trends for AHI, ODI, obstruction severity parameter, and hypopnea index were found to be statistically significant (P < .001) as a function of BMI: the values of these parameters increased with increasing BMI at the group level. Furthermore, with increasing BMI, the decreasing trends for the median durations of individual hypopneas and desaturations and the decreasing trend for areas of individual desaturations were found to be statistically significant (P < .001). In contrast, the depth of the desaturations increased statistically significantly (increasing trend P < .001) with increasing BMI. In addition, the median durations of individual apneas showed decreasing trend with increasing BMI (at the group level), although this decrease did not reach statistical significance (P = .08) (see the supplementary materials at http://www.rcjournal.com).
Discussion
In this study, the effect of BMI on the AHI, ODI, apnea index, hypopnea index, obstruction severity parameter, and on the severity of individual obstruction events was investigated within different OSA severity categories and between different BMI groups. In general, AHI, ODI, and hypopnea index were found to increase with increasing BMI within the OSA severity categories and between the different BMI groups. In contrast, the apnea index; apnea proportion; and individual apnea, hypopnea, and desaturation durations tended to decrease with increasing BMI. Therefore, the effect of the change in BMI on the overall severity of OSA might be more complex than seems based on the AHI only. The results imply that hypopneas are more-dynamic events than apneas because the number of hypopneas increased with increasing BMI. In contrast, apneas may be seen to represent static events with more-severe pathophysiologic effects because apneas have been shown to result in deeper desaturations compared with hypopneas.22 Therefore, because individual obstruction event severity and the proportion of apneas from all events are modulated by the BMI, the AHI (which weights apneas and hypopneas equally) may not be the most sophisticated parameter to assess the overall severity of OSA. We suggest that severity estimation of OSA should be based on more-detailed analysis of characteristics of individual obstruction events (eg, adjusted-AHI parameter19) and that apneas and hypopneas should be weighted in a different manner.
Previously, it has been shown that the severity of OSA (based on AHI) is modulated by weight change.6,7 The present study is in line with this because it was found that AHI, hypopnea index, and ODI increased with increasing BMI. In contrast, apnea index and the proportion of apneas decreased with increasing BMI, especially in mild and moderate OSA categories and between normal and overweight BMI groups. These findings are in line with a previous study in which very obese (BMI ≥ 45 kg/m2) OSA patients were compared with OSA patients having a BMI <35 kg/m2.23 It has been reported that the minimum intraluminal airway pressure that is needed to keep the upper airway open is one of the most important factors in the pathogenesis of apneas.24 In contrast, the occurrence of hypopneas is more dependent on upstream resistance,24 which may indicate different mechanisms behind the pathogenesis of apneas and hypopneas.23
In addition, increased genioglossal fatigability has been reported in non-obese OSA patients, whereas, in obese OSA patients, the performance of the genioglossus muscle did not differ from that of obese controls not having OSA.25 Therefore, it can be speculated that, if obese patients with OSA have better performance of genioglossus muscle compared with patients with OSA and who are not obese, this could lead to a higher occurrence of hypopneas and reduce the number of apneas. Also, increase in hypopnea index and decrease in apnea index along with increasing BMI might be due to increased amount of adipose tissue in the upper airway but also increased muscle mass.23 Because the increased amount of adipose tissue in the upper airway leads to reduction in upper-airway size, it could be assumed that the upper airway is more prone for partial occlusions (ie, hypopneas).
An increase in BMI has been reported to increase the depth of desaturations, independent of age, sex, sleeping position, smoking habits, baseline oxygen saturation, and respiratory event duration.22 In the present study, the desaturation depth did not change statistically significantly with increasing BMI in the mild and moderate OSA categories, albeit increasing trend towards higher BMI was observed in the BMI group level. This might be because the general linear model used in this study was not adjusted for the duration of respiratory events and the durations of apneas and hypopneas were found to decrease with increasing BMI. It has been shown that longer respiratory events cause greater decrease in oxygen saturation and that apneas lead to deeper desaturations compared to hypopneas.22 However, in line with Sato et al,26 we found that the depth and area of desaturation events increased statistically significantly with increasing BMI only in the severe OSA category. This can also be seen in the obstruction severity parameter, which increased statistically significantly with increasing BMI only in subjects with severe OSA due to greater hypoxemia. Because the subjects in the severe OSA category were the most obese, this finding was probably caused by restrictions on pulmonary function.27
Median durations of apneas, hypopneas, and desaturations tended to decrease with increasing BMI, whereas no statistically significant change was observed in the depth of desaturations in mild and moderate OSA categories. This may indicate that, with a higher BMI, short respiratory events are sufficient to cause a rapid decrease in oxygen saturation. Furthermore, because median duration of desaturations decreased at the same time, it could be assumed that the physiologic response for decreased oxygen saturation might be more rapid in patients with a higher BMI. Previously, it has been shown that the severity of respiratory events is modulated by the level of weight change — weight loss mainly decreases and weight gain mainly increases the number of shorter respiratory events.7 If the number of shorter events increases proportionally more than that of longer events while BMI increases, this might be the reason why median durations of obstruction events decreased in the present study.
We acknowledge that the present study had some limitations. Polygraphic recordings were done by using 4-channel devices that did not allow electroencephalography registration. Therefore, hypopneas followed by arousals could not be detected. Furthermore, a 4% threshold level for desaturations was used for scoring hypopneas instead of the 3% threshold level introduced by the American Academy of Sleep Medicine in 2012.28 Using the 3% threshold level for desaturations and also by taking into account hypopneas followed by arousals would have increased the number of the hypopneas and desaturations, and would have affected the average severity of individual events. Thus, the results of the present study need to be further confirmed when considering such events. Also, in the 4-channel recording devices used in this study, respiratory efforts were measured based on abdomen movements only. This may cause inaccuracy in classification of apneas to be obstructive, mixed, or central types. However, in this study, apneas were not separated by their type and, thus, whether obesity has different affects on various types of apneas warrants further investigation.
Furthermore, due to the retrospective design of the present study, information on the hip-to-waist ratio and the neck or waist circumference could not be collected, albeit, for example, waist circumference has been suggested to be a better predictor for OSA than BMI.29 For the same reason, the values of transcutaneous carbon dioxide partial pressure could not be included in the analysis, albeit it is known that alterations in its values are connected to sleep disturbances. In addition, one possible factor that affected the present results was how long the subjects had been obese. This most probably affects the severity of OSA and needs to be investigated in future studies based on follow-up datasets.
Conclusions
The number of hypopneas increased statistically significantly in all OSA severity categories, and the number of apneas decreased statistically significantly in mild and moderate categories with increasing BMI. In addition, the proportion of apneas (from all obstruction events) and median durations of individual hypopnea and desaturation events tended to decrease with increasing BMI in a group level and within all OSA severity categories. No statistically significant BMI-related variations were seen in the depth of desaturation events in mild and moderate OSA categories, whereas, in subjects with severe OSA, the depth of desaturations increased with BMI. Therefore, pathophysiology of OSA might differ between lean subjects and obese subjects, which suggests that the overall severity of OSA may not be equal between lean subjects and obese subjects having a similar AHI. This highlights the importance of considering the individual obstruction event severity while estimating the overall severity of OSA and planning individualized treatment.
Footnotes
- Correspondence: Timo Leppänen, Department of Clinical Neurophysiology, Diagnostic Imaging Center, Kuopio University Hospital, P.O. Box 100, FI-70029 Kuopio, Finland. E-mail: leppanen_timo{at}outlook.com.
Financial support was provided by the Academy of Finland (decision 313697); the Department of Applied Physics, University of Eastern Finland; Kuopio University Hospital, the Research Committee of the Kuopio University Hospital Catchment Area for the State Research Funding (projects 5041754, 5041755, 5041767, and 5041768); Instrumentarium Science Foundation; the Research Foundation of the Pulmonary Diseases; the Respiratory Foundation of Kuopio Region; Seinäjoki Central Hospital (grants 6020 and 6047); the Competitive State Research Financing of Expert Responsibility Area of Tampere University Hospital (grants VTR3221, VTR3228, and EVO2089).
The study was performed at the Department of Clinical Neurophysiology, Diagnostic Imaging Center, Kuopio University Hospital, Kuopio, Finland.
Dr Leppänen presented the research data at the Nordic Congress of Clinical Neurophysiology and Kuopio Epilepsy Symposium at the Kuopio Music Centre, held March 15-17, 2017, in Kuopio, Finland.
The authors have disclosed no conflicts of interest.
Supplementary material related to this paper is available at http://www.rcjournal.com.
- Copyright © 2019 by Daedalus Enterprises