Abstract
Study Objectives:
We performed a case-control study to investigate the correlation between the apnea-hypopnea index (AHI) and the retinal vascular fractal dimension (FD).
Methods:
We selected 527 individuals who underwent polysomnography during health checkups at the Huadong Sanatorium from January to December 2021 as the study population, of whom 468 were included and 59 were excluded. All participants underwent a detailed health examination, including medical history assessment, physical examination, assessment of lifestyle factors, fundus photography, and laboratory examinations. The retinal vasculature was quantitatively assessed using Singapore I Vessel Assessment (SIVA) software. The relationship between the AHI and the retinal vessel quantitative was examined by multiple linear regression analyses and restricted cubic spline.
Results:
Among the 468 studied individuals, the average age was 51.51 (43–58) years, with 369 (78.85%) men and 99 (21.15%) women. According to the AHI indicator, 355 individuals were diagnosed with obstructive sleep apnea (OSA) syndrome, with an average AHI of 17.00 (9.200–30.130) events/h; 113 individuals were classified as controls, with an average AHI of 2.13 (0.88–3.63) events/h. In multiple linear regression, following varying degrees of adjustment for confounding factors, FD was reduced by 0.013 (P = .012; 95% confidence interval [CI]: −0.024 to −0.003), FD arteriole (FDa) was reduced by 0.013 (P = .019; 95% CI: −0.024 to −0.002), and FD venule (FDv) was reduced by 0.014 (P = .08; 95% CI: −0.024 to −0.004) in the high-AHI group compared with the low-AHI group. All tests for trend P values were < .05. The restricted cubic spline in the overall OSA population and the individuals without diabetes revealed a U-shaped pattern of decreasing, then increasing, FD, FDa, and FDv with a rising AHI. In the OSA individual with diabetes, FD, FDa, and FDv gradually decreased with increasing AHI.
Conclusions:
The FD is associated with AHI in OSA individuals. The link between AHI and FD varied for OSA individuals with and without diabetes.
Citation:
Wang J, Chen T, Qi X, Li Y, Yang X, Meng X. Retinal vascular fractal dimension measurements in patients with obstructive sleep apnea syndrome: a retrospective case-control study. J Clin Sleep Med. 2023;19(3):479–490.
Keywords: obstructive sleep apnea syndrome, retinal, fractal dimension, polysomnographic, fundus photographs
BRIEF SUMMARY
Current Knowledge/Study Rationale: Previous studies on obstructive sleep apnea syndrome and quantitative measures of retinal vascularity have been conducted. However, the results were inconsistent. There were no studies that investigated the relationship between the apnea-hypopnea index (AHI) index and fractal dimension (FD). This study focused on the relationship between the AHI index and FD.
Study Impact: The FD is associated with the AHI index. The link between the AHI index and FD varied for individuals with obstructive sleep apnea with and without diabetes.
INTRODUCTION
Obstructive sleep apnea (OSA) syndrome (OSAS) refers to the recurrent partial or complete collapse of the upper airway during sleep, causing reduced or absent airflow that lasts at least 10 seconds, associated with cortical arousal or reduced oxygen saturation.1 There are approximately 936 million adults between the ages of 30 and 69 years with mild to severe OSAS worldwide, and the largest number of individuals affected are in China, the United States, Brazil, and India.2 Hypertension, obesity, diabetes, and hyperlipidemia, which rank among the top 10 risk factors for early death and disability-adjusted life-years, are strongly associated with OSAS.3 Severe OSAS doubles the risk of stroke, an acute neurological event of focal cerebral ischemia, affecting especially young to middle-aged patients.4
OSAS is also associated with many ocular diseases, such as glaucoma, nonarteritic anterior ischemic optic neuropathy, and retinal vein occlusion.5 The primary drivers causing ocular complications in OSAS are intermittent hypoxia, sympathetic nervous system activation, oxidative stress, and the adverse effects of endothelin-1.6
The retinal microvasculature is the only part of the human circulatory system that is clearly visible and can be observed directly without invasive methods.7 The retina is the organ with the highest weight of oxygen-consuming units of all tissues in the human body, and any alteration in the circulation can lead to dysfunction and tissue damage.8 As a result, the retina can usually exhibit changes secondary to hypoxic diseases such as OSAS in the early stages of the disease course due to its high oxygen-demand requirements.6 Furthermore, retinal vascular abnormalities can be used as a predictive marker for the onset and progression of cardiocerebrovascular diseases.9
Previous studies have investigated retinal vascular alterations in patients with OSAS.9,10 However, these studies used specialized analysis software to focus on the central retinal arteriolar equivalent (CRAE), central retinal venular equivalent (CRVE), and arteriolar-to-venular diameter ratio (AVR) metrics of the retina, without involving a comprehensive assessment of retinal microcirculation patterns. The retinal vascular fractal dimension (FD) is a global measure of the complex retinal vascular branching pattern and density.11 The FD has been used to evaluate some ocular diseases and predict the outcomes of cardiovascular and cerebrovascular diseases.12,13
OSAS can cause alterations in cerebral hemodynamics and the retinal microcirculation.14,15 We hypothesized that patients with OSAS may exhibit alterations in the retinal vascular FD. The purpose of our study was to quantitatively examine the correlation between the severity of OSAS and retinal vascular FD.
METHODS
Participants
The retrospective study was conducted at the Huadong Sanatorium, which is a tertiary hospital specializing in health care and health checkups in China. We selected individuals who underwent health screening at our hospital from January to December 2021 as the study population.
The inclusion criteria were as follows: (1) adults aged > 18 years, (2) individuals who volunteered to participate in the study, (3) individuals who had undergone both polysomnographic monitoring and fundoscopic examination, and (4) individuals who could cooperate in obtaining satisfactory fundus photographs.
The exclusion criteria were as follows: (1) uncontrolled hypertension (systolic blood pressure > 140 mm Hg or diastolic blood pressure > 90 mm Hg), acute myocardial infarction within 6 months, acute stroke within 6 months, inflammatory or infectious diseases, sustained arrhythmia (eg, atrial fibrillation), pulmonary arterial hypertension, malignant tumors, or peripheral vascular disease; (2) history of severe ocular diseases (eg, corneal disorders, glaucoma, macular degeneration, or ocular trauma); (3) fundus photographs of poor quality preventing quantitative assessment with the software; and (4) missing data.
The study was conducted in strict compliance with the principles of the Declaration of Helsinki, which was granted by the Medical Ethics Committee of Huadong Sanatorium (grant number: ECHS2022-04). This study pre-anonymized private personal information in advance, maintained strict confidentiality during statistical analyses, and used it solely for scientific purposes. As a result, the requirement for informed patient consent was waived.
Examination
Routine health checkups
All participants underwent a detailed health examination, including medical history assessment, physical examination, assessment of lifestyle factors, and laboratory tests. Data recorded for all participants included age and sex, smoking, drinking history, body mass index (BMI), systolic pressure, diastolic pressure, triglycerides, cholesterol, high-density lipoprotein, low-density lipoprotein, fasting plasma glucose, glycated hemoglobin, and uric acid.
Polysomnographic monitoring
Each individual underwent 8 hours of polysomnography (PSG) (Alice PDX; Philips Respironics, Murrysville, PA, USA) during the night in a quiet environment. Two sleep specialists evaluated the PSG. Obstructive apnea is defined by a nearly complete cessation of airflow (> 90%) for longer than 10 seconds while sleeping, whereas hypopnea is defined by a reduction in airflow of over 30% with a concurrent decrease in oxyhemoglobin saturation of at least 3% or arousal from sleep.16 The number of episodes of apnea and hypopnea per hour during sleep is referred to as the apnea-hypopnea index (AHI).16 Based on the AHI results, all participants were divided into a control group (AHI < 5 events/h), a mild OSAS group (AHI ≥ 5 events/h but AHI < 15 events/h), a moderate OSAS group (AHI ≥ 15 events/h but AHI < 30 events/h), and a severe OSAS group (AHI ≥ 30 events/h).17 According to the tertile of AHI, these individuals were divided into 3 groups: low-, intermediate-, and high-AHI groups.
Acquisition of fundus photographs and measurements of quantitative fundus vascular indicators
All participants underwent fundus photography with a nonmydriatic method. Fundus photographs were acquired using a no-dilatation fundus camera (NW400; Topcon Corporation, Tokyo, Japan) centered on the optic disc with a 45° field of view. Fundus photographs were analyzed quantitatively in the right eye by 2 trained ophthalmologists using the Singapore I Vessel Assessment (SIVA) software (version 4.0; School of Computing, National University of Singapore, Singapore). If the quality of the fundus photograph of the right eye was not satisfactory, the left one was used for evaluation. Due to errors in automatic software measurements and interference from inadequate photograph quality, most photographs were corrected manually and inspected. To guarantee consistent and reproducible results, the quantitative analysis of fundus photographs was conducted strictly according to the manual instruction for retinal angiometry. As shown in Figure 1, the area from 0.5 to 1 papillary diameter (PD) from the edge of the optic disc is zone B, whereas the area from 1 to 2 PD from the edge of the optic disc is zone C.18 The morphological indicators of retinal vessels in the region were measured by the software, including CRAE, CRVE, AVR, FD, FD arteriole (FDa), and FD venule (FDv). The CRAE, CRVE, and AVR values were obtained from simultaneous measurements in the B and C zones, which were denoted as CRAE-B, CRAE-C, CRVE-C, CRVE-B, AVR-B, and AVR-C, respectively. Retinal fundus photographs and skeletonized images of patients with different severity of OSAS are shown in Figure 2.
Figure 1. Image showing the operating interface of SIVA software (version 4.0; School of Computing, National University of Singapore, Singapore) for retinal vascular FD measurement.
The area from 0.5 to 1 PD from the edge of the optic disc shown in the figure is zone B, while the area from 1 to 2 PD from the edge of the optic disc is zone C. FD = fractal dimension, PD = papillary diameter, SIVA = Singapore I Vessel Assessment.
Figure 2. (A–H) Images showing retinal fundus photographs and skeletonized images in the control, mild, moderate, and severe OSAS groups.
In the skeletonized images of the retinal vasculature, the vascular network of the retinal arteries is highlighted in red and the vascular network of the retinal veins in blue. OSAS = obstructive sleep apnea syndrome.
Statistical analyses
All of the statistical analyses were performed using R statistical software (R Core Team, Vienna, Austria), with a test level of α = .05. Continuous variables are expressed as means (interquartile range), and groups were compared using the variance test. The count data are expressed as n (%), and the chi-square test was used to compare the 2 groups. A linear trend test was performed to identify trends. Bivariate associations were assessed by one-way analysis of variance and linear regression. Those independent variables found to be statistically significant in the bivariate analysis (α = 0.05) were added to the stepwise multivariate regression model to determine the correlation between FD, CRAE, and OSAS severity. The variance inflation factor procedure was used to test the collinearity of all variables, and the variance inflation factor for all covariates was less than 3.33 (mean [standard deviation]: 1.688 [0.645]).
The final model was evaluated for multicollinearity by examining the variance inflation factor. Three models were fitted with different levels of adjustment for confounding factors. Confounding factors were identified by comparing regression coefficients with and without the inclusion of potential confounders in the model. The covariates were identified as potential confounders by introducing covariates in the basic model and eliminating covariates with an effect of greater than 10% on the regression coefficient of the outcome variable, or with a P value of less than .1 after including the covariates in the regression equation, from the full model.19 With regard to treatment of missing values, using multiple interpolation, we created 5 complete datasets, analyzed each imputed dataset separately, and then averaged the resulting estimates. Sensitivity analyses were performed on both the raw and imputed data to assess the impact of the missing data, as shown in Table S1 (249.6KB, pdf) in the supplemental material. Restricted cubic spline plots were used to explore the shape of the association between the AHI index and FD, fitting an restricted cubic spline function with 3 knots (10th, 50th, and 90th centiles).
RESULTS
Characteristics of the sample
A total of 527 individuals received PSG examinations. Of these participants, 31 individuals were excluded for systemic disease reasons: 23 cases of uncontrolled hypertension, 3 cases of acute myocardial infarction, 2 cases of cerebrovascular disease, 1 case of cardiac arrhythmia, 1 case of pulmonary arterial hypertension, and 1 case of a malignant tumor. Then, 28 cases were excluded for ocular disease reasons (history of severe ocular trauma in 1 case, glaucoma in 3 cases, macular degeneration in 8 cases, corneal diseases in 1 case, and unclear fundus photographs in 15 cases). The flowchart for the exclusion of study participants in this study is shown in Figure 3.
Figure 3. Flowchart of the study participant exclusion criteria applied in this study.
OSAS = obstructive sleep apnea syndrome, PSG = polysomnography.
A total of 468 individuals were included, of whom 369 (78.85%) were male and 99 (21.15%) were female. According to the AHI indicator, 355 individuals were diagnosed with OSAS, with an average AHI of 17.00 (9.200–30.130) events/h and 113 were healthy individuals without OSAS with an average AHI of 2.13 (0.88–3.63) events/h. The baseline characteristics of this study are presented in Table 1.
Table 1.
Demographic data and clinical characteristics.
| AHI tertile | Low | Middle | High | P | P for trend |
|---|---|---|---|---|---|
| Age, median (IQR), y | 47 (39–56) | 52 (46–60) | 54 (47–60) | <.001 | <.001 |
| Sex, n (%) | <.001 | <.001 | |||
| Male | 105 (68.182%) | 129 (81.646%) | 135 (86.538%) | ||
| Female | 49 (31.818%) | 29 (18.354%) | 21 (13.462%) | ||
| Smoking status, n (%) | .012 | .002 | |||
| Nonsmoking | 110 (71.429%) | 99 (62.658%) | 86 (55.128%) | ||
| Smoking | 44 (28.571%) | 59 (37.342%) | 70 (44.872%) | ||
| Drinking, n (%) | .040 | .013 | |||
| Nondrinking | 96 (62.338%) | 85 (53.797%) | 75 (48.077%) | ||
| Drinking | 58 (37.662%) | 73 (46.203%) | 81 (51.923%) | ||
| BMI, mean (SD), kg/m2 | 24.642 (3.128) | 25.599 (2.902) | 27.189 (2.869) | <.001 | <.001 |
| Systolic pressure, median (IQR), mmHg | 121 (110–130) | 121 (114–128) | 124 (114–132) | .106 | .180 |
| Diastolic pressure, median (IQR), mmHg | 72 (66–80) | 73 (67–80) | 76 (70–83) | .029 | .037 |
| Triglyceride, median (IQR), mmol/L | 1.465 (0.978–2.188) | 1.340 (0.925–1.985) | 1.710 (1.220–2.130) | .002 | .029 |
| Cholesterol, median (IQR), mmol/L | 4.840 (4.100–5.475) | 4.720 (4.225–5.270) | 4.790 (4.360–5.620) | .297 | .377 |
| HDL, median (IQR), mmol/L | 1.195 (1.000–1.410) | 1.230 (1.020–1.425) | 1.160 (1.027–1.365) | .259 | .09 |
| LDL, mean (SD), mmol/L | 3.105 (0.883) | 2.987 (0.774) | 3.130 (0.777) | .254 | .25 |
| FPG, median (IQR), mmol/L | 5.205 (4.888–5.645) | 5.220 (4.885–5.735) | 5.450 (5.090–5.960) | .001 | .077 |
| HbA1c median (IQR), % | 5.700 (5.400–5.900) | 5.700 (5.400–6.000) | 5.800 (5.600–6.100) | .019 | .078 |
| Uric acid, mean (SD), µmol/L | 354.383 (88.292) | 361.334 (87.386) | 383.405 (76.774) | .008 | .005 |
| Hypertension (n), % | 31 (20.130%) | 41 (25.949%) | 55 (35.256%) | .010 | .002 |
| Diabetes (n), % | 37 (24.026%) | 39 (24.684%) | 45 (28.846%) | .574 | .273 |
| Hyperlipidemia (n), % | 46 (29.870%) | 59 (37.342%) | 67 (42.949%) | .057 | .068 |
| TS90, median (IQR), s | 1.000 (0–6.000) | 5.000 (1.000–23.000) | 38.500 (6.925–88.250) | <.001 | <.001 |
| AHI, median (IQR) | 3.340 (1.160–5.000) | 11.775 (8.648–15.363) | 31.450 (25.500–41.750) | <.001 | <.001 |
| TAmax, median (IQR), s | 28.000 (17.000–49.000) | 53.250 (35.250–84.375) | 60.000 (46.000–88.625) | <.001 | <.001 |
| The longest time of hypopnea, median (IQR), s | 49.000 (23.250–58.000) | 61.000 (54.125–70.000) | 62.000 (56.000–75.000) | <.001 | <.001 |
| MinSaO2, median (IQR), % | 87.000 (83.500–90.000) | 84.000 (79.000–87.000) | 77.000 (71.000–81.000) | <.001 | <.001 |
| Mean oxygen saturation, median (IQR), % | 94.500 (93.000–95.000) | 94.000 (93.000–95.000) | 93.000 (92.000–94.000) | <.001 | <.001 |
| CRAE-B, median (IQR) | 175.940 (162.693–189.421) | 174.093 (163.399–188.684) | 170.173 (157.028–183.788) | .063 | .015 |
| CRVE-B, median (IQR) | 244.734 (227.724–263.076) | 245.730 (227.068–263.555) | 242.147 (220.460–259.874) | .164 | .016 |
| AVR-B, median (IQR) | 0.719 (0.677–0.755) | 0.725 (0.679–0.768) | 0.708 (0.674–0.760) | .399 | .774 |
| CRAE-C, median (IQR) | 171.417 (158.502–183.569) | 169.265 (156.548–185.286) | 163.883 (148.689–178.178) | .008 | .001 |
| CRVE-C, median (IQR) | 228.673 (214.227–245.812) | 229.195 (213.576–244.399) | 226.892 (213.111–242.207) | .506 | .079 |
| AVR-C, median (IQR) | 0.739 (0.700–0.784) | 0.748 (0.693–0.806) | 0.719 (0.677–0.778) | .004 | .082 |
| FD, median (IQR) | 1.311 (1.284–1.328) | 1.300 (1.270–1.327) | 1.292 (1.262–1.317) | <.001 | <.001 |
| FDa, median (IQR) | 1.172 (1.135–1.192) | 1.158 (1.125–1.181) | 1.147 (1.112–1.174) | <.001 | <.001 |
| FDv, median (IQR) | 1.150 (1.132–1.170) | 1.141 (1.114–1.170) | 1.134 (1.107–1.159) | <.001 | <.001 |
A, B, C = zones A, B, and C, respectively, AHI = apnea-hypopnea index, AVR, arteriole-to-venular diameter ratio, BMI = body mass index, CRAE = central retinal arteriolar equivalent, CRVE = central retinal venular equivalent, FD = fractal dimension, FDa = fractal dimension arteriole, FDv = fractal dimension venule, 20. HbA1c = glycated haemoglobin, HDL = high-density lipoprotein, IQR = interquartile range, LDL = low-density lipoprotein, SD = standard deviation, TS90:the percentage of time spent in sleep below 90% oxygen saturation. TAmax, the longest time of apnea. MinSaO2, Minimum oxygen saturation.
There were statistically significant (P < .05) differences in baseline characteristics (age, sex, smoking, history of alcohol consumption, BMI, diastolic pressure, triglycerides, fasting plasma glucose, glycated hemoglobin, uric acid, percentage of hypertension) between the low-, intermediate-, and high-AHI groups.
The trend test revealed that, as the AHI index rose, age, the proportion of smokers, the proportion of drinkers, the proportion of males, BMI, triglycerides, the proportion of hypertension, diastolic pressure, triglyceride, and uric acid all increased (P value for trend test < .05). PSG indicators for different grades of the AHI showed a trend toward increasing the percentage of time spent in sleep below 90% oxygen saturation (TS90), AHI, the longest time of apnea (TAmax), and maximal hypoventilation time (P value for trend test < .001) and a trend toward decreasing MinSaO2 and mean oxygen saturation (P value for trend test < .001) with increasing AHI.
Quantitative fundus vascular indicator measurements in different groups
The results indicated that the FDs in the low-, intermediate-, and high-AHI groups were 1.311 (1.284–1.328), 1.300 (1.270–1.327), and 1.292 (1.262–1.317), with a P value of < .0001 for the trend test. These results indicate a gradual decrease in FD as the AHI increases. The analysis of quantitative fundus vascular indicators for various degrees of AHI levels that CRAE-B, CRAE-C, AVR, FD, FDa, and FDv all exhibited a decreasing trend as the AHI increased (P value for trend test < .001). By contrast, CRVE-B, CRVE-C, AVR-B, and AVR-C showed no significant trend (see Table 1).
Correlation analysis of PSG indexes and quantitative fundus vascular indicators
The AHI was negatively correlated with CRAE-B (r = –.110, P = .019), CRAE-C (r = –.149, P = .001), AVR-C (r = –.106, P = .026), FD (r = –.173, P < .001), FDa (r = –.175, P < .0001), and FDv (r = –.135, P = .004); TS90 was negatively correlated with CRAE-C (r = –.134, P = .006), FD (r = –.142, P = .003), and FDv (r = –.118, P = .013). Mean oxygen saturation and CRAE-C were positively correlated (r = .098, P = .036) (see Table 2).
Table 2.
Linear relationship between PSG indicators and quantitative fundus vascular indicators.
| CRAE-B | CRVE-B | AVR-B | CRAE-C | CRVE-C | AVR-C | FD | FDa | FDv | |
|---|---|---|---|---|---|---|---|---|---|
| TS90 | |||||||||
| r | −.083 | −.052 | −.046 | −.134 | −.085 | −.078 | −.142 | −.126 | −.118 |
| P | .085 | .264 | .346 | .006b | .067 | .127 | .003b | .008b | .013a |
| AHI | |||||||||
| r | −.110 | −.063 | −.049 | −.149 | −.062 | −.106 | −.173 | −.175 | −.135 |
| P | .019a | .176 | .304 | .001b | .185 | .026a | <.001b | <.001b | .004b |
| TAmax | |||||||||
| r | −.021 | −.05 | .038 | −.011 | −.057 | .057 | −.051 | −.037 | −.063 |
| P | .652 | .285 | .415 | .805 | .217 | .217 | .271 | .429 | .174 |
| Longest time of hypopnea | |||||||||
| r | −.074 | −.069 | −.007 | −.051 | −.06 | .007 | −.072 | −.071 | −.091 |
| P | .108 | .134 | .887 | .274 | .194 | .887 | .118 | .126 | .05 |
| Mean oxygen saturation | |||||||||
| r | .036 | .019 | .02 | .098 | .051 | .07 | .085 | .079 | .057 |
| P | .447 | .685 | .673 | .036a | .273 | .132 | .071 | .094 | .22 |
| MinSaO2 | |||||||||
| r | .069 | .027 | .055 | .061 | .034 | .041 | .091 | .081 | .086 |
| P | .141 | .57 | .243 | .19 | .46 | .383 | .051 | .082 | .068 |
aP < .05. bP < .01. B, C = zones B, and C, AHI = apnea-hypopnea index, AVR = arteriole-to-venular diameter ratio diameter ratio, CRAE = central retinal arteriolar equivalent, CRVE = central retinal venular equivalent, FD = fractal dimension, FDa = fractal dimension arteriole, FDv = fractal dimension venule, MinSaO2 = Minimum oxygen saturation. PSG = polysomnography, TS90 = the percentage of time spent in sleep below 90% oxygen saturation,TAmax = the longest time of apnea.
Multiple linear regression models for OSAS severity classification and quantitative fundus vascular indicators
After varying degrees of adjustment for confounding factors, FD was reduced by 0.003 (P = .59; 95% confidence interval [CI]: −0.012, 0.007), FDa by 0.004 (P = .514; 95% CI: −0.013, 0.007), and FDv by 0.003 (P = .489; 95% CI: −0.013, 0.006) in the intermediate-AHI group compared with the low-AHI group. FD was reduced by 0.013 (P = .012; 95% CI: −0.024, −0.003), FDa by 0.013 (P = .019; 95% CI: −0.024, −0.002), and FDv by 0.014 (P = .08; 95% CI: −0.024, −0.004) in the high-AHI group compared with the low-AHI group. All tests for trend P values were < .05. The multiple linear regression model is shown in Table 3. Thus, decreased FD, FDa, and FDv were all associated with increasing the AHI. However, after adjusting for all of the confounding factors, the changes in CRAE-B, CRAE-C, and AVR-C were not statistically significant as the AHI increased (see Table 4).
Table 3.
Multiple linear regression model with FD and OSAS severity classification.
| Model 1 | Model 2 | Model 3 | ||||
|---|---|---|---|---|---|---|
| Group | B (95% CI) | P | B (95% CI) | P | B (95% CI) | P |
| FDa | ||||||
| Low | – | – | – | |||
| Middle | −0.003 (−0.012, 0.005) | .435 | −0.004 (−0.013, 0.004) | .317 | −0.003 (−0.012, 0.007) | .59 |
| High | −0.011 (−0.020, −0.002) | .013 | −0.013 (−0.022, −0.004) | .006 | −0.013 (−0.024, −0.003) | .012 |
| P for trend | .012 | .005 | .013 | |||
| FDab | ||||||
| Low | – | – | – | |||
| Middle | −0.004 (−0.014, 0.005) | .353 | −0.005 (−0.015, 0.004) | .254 | −0.003 (−0.013, 0.007) | .514 |
| High | −0.012 (−0.021, −0.002) | .014 | −0.013 (−0.023, −0.004) | .007 | −0.013 (−0.024, −0.002) | .019 |
| P for trend | .013 | .007 | .021 | |||
| FDvb | ||||||
| Low | – | – | – | |||
| Middle | −0.004 (−0.012, 0.005) | .403 | −0.005 (−0.013, 0.004) | .286 | −0.003 (−0.013, 0.006) | .489 |
| High | −0.010 (−0.018, −0.001) | .034 | −0.011 (−0.020, −0.003) | .012 | −0.014 (−0.024, −0.004) | .008 |
| P for trend | .033 | .008 | .012 | |||
aModel 1: adjusted for age and sex; model 2: adjusted for age, sex, smoking, hypertension, diabetes, and hyperlipidemia; model 3: adjusted for age, sex, smoking, hypertension, diabetes, hyperlipidemia, BMI, triglycerides, uric acid, FBG, and HbA1c. bModel 1: adjusted for age and sex; model 2: adjusted for age, sex, smoking, hypertension, diabetes, and hyperlipidemia; model 3: adjusted for age, sex, smoking, hypertension, diabetes, hyperlipidemia, BMI, FBG, and HbA1c. BMI = body mass index, CI = confidence interval, FBG = fasting blood glucose, FD = fractal dimension, FDa = fractal dimension arteriole, FDv = fractal dimension venule, HbA1c = glycated hemoglobin, OSAS = obstructive sleep apnea syndrome.
Table 4.
Multiple linear regression models for OSAS severity and CRAE-B, CRAE-C and AVR-C.
| Model 1 | Model 2 | Model 3 | ||||
|---|---|---|---|---|---|---|
| OSAS | B (95% CI) | P | B (95% CI) | P | B (95% CI) | P |
| CRAE-Ba | ||||||
| Low | – | – | – | |||
| Middle | 2.435 (−2.474, 7.344) | .332 | 2.583 (−2.354, 7.520) | .332 | 5.516 (0.272, 10.761) | .040 |
| High | −2.277 (−7.287, 2.732) | .373 | −1.640 (−6.735, 3.455) | .528 | −0.918 (−6.644, 4.808) | .753 |
| P for trend | .348 | .499 | .786 | |||
| CRAE-Ca | ||||||
| Low | – | – | – | |||
| Middle | 3.252 (−1.402, 7.906) | .172 | 3.554 (−1.125, 8.233) | .137 | 5.282 (0.361, 10.203) | .036 |
| High | −3.072 (−7.821, 1.677) | .206 | −2.353 (−7.182, 2.476) | .34 | −2.648 (−8.021, 2.725) | .335 |
| P for trend | .183 | .309 | .365 | |||
| AVR-Cb | ||||||
| Low | – | – | – | |||
| Middle | 0.015 (−0.002, 0.033) | .093 | 0.015 (−0.002, 0.033) | .091 | 0.016 (−0.003, 0.035) | .103 |
| High | −0.015 (−0.033, 0.003) | .099 | −0.013 (−0.031, 0.005) | .16 | −0.008 (−0.029, 0.013) | .448 |
| P for trend | .085 | .140 | .472 | |||
Multiple linear regression models showed no statistically significant changes in CRAE-B, CRAE-C, and AVR-C as the severity of OSAS increased after adjusting for the confounding factors. aModel 1: adjusted for age and sex; model 2: adjusted for age, sex, smoking, drinking, hypertension, diabetes, and hyperlipidemia; model 3: adjusted for age, sex, smoking, drinking, hypertension, diabetes, hyperlipidemia, systolic pressure, diastolic pressure, BMI, triglycerides, HDL, uric acid and HbA1c. bModel 1: adjusted for age and sex; model 2: adjusted for age, sex, smoking, drinking, hypertension, diabetes, and hyperlipidemia; model 3: adjusted for age, sex, smoking, drinking, hypertension, diabetes, hyperlipidemia, BMI, systolic pressure, diastolic pressure, triglycerides, HDL, uric acid, FPG, and HbA1c. B, C = zone B, and C. AVR = arteriole-to-venular diameter ratio, BMI = body mass index, CI = confidence interval, CRAE = central retinal arteriolar equivalent, CRVE = central retinal venular equivalent, FPG = fasting plasma glucose, HbA1c = glycated hemoglobin, HDL = high-density lipoprotein, OSAS = obstructive sleep apnea syndrome.
Dose–response relationships between the AHI and the FD using Restricted cubic spline functions
The results showed that FD, FDa, and FDv decreased and then increased with increasing AHI scores in a U-shaped trend in all individuals and individuals without diabetes mellitus (Figure 4). The AHI corresponding to the lowest point of the curve alteration is depicted by the dotted lines in Figure 4. However, FD, FDa, and FDv gradually decreased with increasing AHI scores in diabetic individuals.
Figure 4. Images showing the dose–response relationships between AHI and the FD according to restricted cubic spline functions using 3 knots in all individuals and individuals with DM and without DM.
(A) The FD first decreases and then increases slightly as the AHI increases from 0 to 30.39. (B) In individuals without DM, the FD first decreases and then increases slightly as the AHI increases from 0 to 25.84. (C) As the AHI increased, the FD decreased gradually in individuals with DM. (D) The FDa first decreases and then increases slightly as the AHI increases from 0 to 29.21. (E) In individuals without DM, the FDa first decreases and then increases slightly as the AHI increases from 0 to 25.84. (F) As the AHI increased, the FDa decreased gradually in individuals with DM. (G) The FDv first decreases and then increases slightly as the AHI increases from 0 to 30.98. (H) In individuals without DM, the FDv first decreases and then increases slightly as the AHI increases from 0 to 26.12. (I) As the AHI increased, the FDv gradually reduced in individuals with DM. AHI = apnea-hypopnea index, CI = confidence interval, DM = diabetes mellitus, FD = fractal dimension, FDa = fractal dimension arteriole, FDv = fractal dimension venule. The red solid line indicates the beta regression coefficient, red shaded areas indicate confidence intervals.
DISCUSSION
Our findings confirmed the hypothesis that patients with OSAS may have microvascular alterations in the ocular fundus. To the best of our knowledge, this is the first study to examine retinal vascular FD alterations in patients with OSAS. The present investigation revealed a significant correlation between OSAS and fundus FD. As the AHI increased, FD, FDa, and FDv decreased and then increased in a U-like trend in patients with OSA. In the diabetic subgroup study, the trend was the U-shaped in the patients with OSA without diabetes, whereas the FD, FDa, and FDv gradually decreased with increasing AHI in the patients with OSA with diabetes.
Previous findings regarding the correlation between OSAS and retinal vascular alterations have been inconsistent.9,10 Wang et al9 assessed retinal CRAE, CRVE, and AVR in patients with OSAS. They found that CRAE and CRVE were not significantly different in subgroups of various severity, while AVR increased with increasing AHI. However, Tong et al10 found that AVR and CRAE were reduced as AHI increased, indicating that worsening OSAS is associated with retinal arteriolar narrowing. Our findings are not consistent with those of previous studies. Multiple linear regression models indicated no significant change in CRAE-B, CRAE-C, or AVR-C with increasing OSAS severity. Furthermore, no significant change in the trend for CRVE-B, CRVE-C, AVR-B, and AVR-C was observed in the different OSAS severity groups. Previous studies have only examined the diameters of retinal arteries and venules within a circular area of 0.5 to 1 PD centered on the optic disc.10 However, our data were taken over a 2-PD circular area centered on the optic disc, reflecting blood circulation in the retina more accurately. Moreover, compared with the existing literature, our study benefited from a larger sample size.
FD is a common metric used by ophthalmologists to examine the vascular condition of a patient’s fundus. It is an overall measure of retinal vascular complexity that reflects the density of blood vessels within the retinal region of interest. Human retinal vascular networks are fractal, and FD can be used as both a metric for quantifying the retinal microvasculature and a useful screening tool.20 A larger FD suggests a denser vascular network with more complicated retinal vascular branching, whereas a reduced FD indicates a sparser and more distributed retinal vascular network.11 Low FD was associated with a significantly increased risk of accidental mortality, hypertension, renal failure, type 2 diabetes, congestive heart failure, sleep apnea, and anemia.21 The decreased FD may be a marker of microvascular alterations produced by elevated blood pressure.22 Reduced FD implies the risk of microvascular complications such as diabetic retinopathy, neuropathy, and nephropathy in patients with type 1 diabetes.23 Long-term cohort studies have shown that reduced FD is independently associated with mortality from coronary heart disease and stroke.11,24 Among the disorders related to reduced FD, hypertension, diabetes, and stroke all have a significant association with OSAS.3,4 Thus, patients with OSAS who have not yet experienced complications but have a lower FD are expected to be at a higher risk of developing severe complications in the future.
By stratifying individuals based on diabetes, we investigated the dose–response relationship between AHI and FD alterations in diabetic and nondiabetic patients. In the nondiabetic population, retinal FD showed a trend of decreasing and then increasing with increasing AHI, which may be associated with impaired retinal microcirculation due to hypoxia. In the diabetic population, retinal FD decreased gradually as the AHI increased, which could be attributable to aberrant retinal microcirculation in diabetic individuals themselves. Some studies have shown that retinal FD is reduced in diabetic patients compared with healthy individuals, which results in a different basal FD in patients with diabetes than in patients without diabetes.25,26 Additionally, several investigations have indicated that retinal hypoxia gradually increases from nondiabetic to nonproliferative to proliferative diabetic retinopathy, while retinal FD tends to grow first and then decline.27–29 It differs from the situation in patients with OSA without diabetes, where FD first decreases and then increases with increasing AHI. We therefore thought that the dose–response relationship between AHI and FD for patients with OSA with and without diabetes was different.
Recently, retinal vascular changes in patients with OSAS have garnered interest, and various techniques have been used to explore the alterations in their retinal microcirculation. Using optical coherence tomography angiography, Ava et al30 found decreased vascular density in the superficial and deep capillary plexuses and an enlarged foveal avascular zone in the deep capillary plexus in patients with OSAS. By contrast, Moyal et al31 observed that OSAS did not affect vascular density in the retina’s superficial or deep capillary plexus layers with optical coherence tomography angiography. Shiba et al32 observed the optic nerve head microcirculation in patients with OSAS using laser speckle flowgraphy and discovered that the lowest blood oxygen saturation (SpO2) had a substantial impact on retinal microcirculation in female patients. Tong et al,10 using a dynamic vascular analyzer, reported a strong correlation between higher AHI and a reduced amplitude of retinal vascular pulsations in individuals with OSAS. The mechanism of damage to the retinal microvasculature associated with OSAS remains unclear. It may involve multiple mechanisms. Intermittent hypoxia, intrathoracic pressure fluctuations, and arousal are the primary acute physiological effects of OSAS.33 These pathophysiological processes can disrupt vascular endothelial function, which refers to the activity of endothelial cells in sensing and regulating blood flow.34 In a study that used real-time measurement of critical endothelial factors, OSAS caused microcirculatory endothelial dysfunction by increasing endothelial nitric oxide synthase and decreasing the availability of nitric oxide in oxidative stress.35 A recent ex vivo tissue study in patients with OSAS also showed that OSAS causes microcirculatory endothelial dysfunction.36 Intermittent hypoxia stimulates the sympathetic nervous system, causing blood pressure to rise,1 and hypertension may play a more critical role in vascular damage.9 Intermittent hypoxia can elevate intracranial pressure and increase venous pressure in the optic nerve head due to upper airway obstruction, mechanically limiting the retinal circulation and reducing spontaneous retinal venous pulsation.10
This study also has some limitations. First, the present study was retrospective, with the limitations inherent in a retrospective analysis. Thus, a prospective study is required in the future. Second, OSAS has traditionally been regarded as a primarily male-related disorder. However, the clinical presentation of female patients with OSAS differs from that of males, and female patients with OSAS are frequently underdiagnosed and undertreated compared with males.37 This study included fewer female patients, which may have introduced some bias. Third, no OSAS symptom–related questionnaires were given to participants in this study (eg, the Epworth Sleepiness Scale).38 As a result, our data did not include a link between symptoms and FD in patients with OSAS.
CONCLUSIONS
In conclusion, patients with OSAS can exhibit retinal microvascular alterations. There is a correlation between the FD and AHI in patients with OSA. The relationship between the AHI score and FD differed for patients with OSA who had diabetes compared with those who did not have diabetes. When performing funduscopic examinations that reveal alterations in the FD of the patient, OSA factors should be considered.
DISCLOSURE STATEMENT
All authors have seen and approved the manuscript. Work for this study was performed at Huadong Sanatorium. This study was funded by the Clinical Special Program of Shanghai Municipal Health Commission (grant number: 20194Y0437) and the Shanghai Municipal “Rising Stars in Medicine” Young Medical Talents Training Grant Program, China. The authors report no conflicts of interest.
ACKNOWLEDGMENTS
The authors thank all individuals who volunteered to participate in the present study. The authors thank Dr. Lijuan Yang and Dr. Ying Feng from the sleep and respiratory team of Huadong Sanatorium for their collaborative support. The authors thank the Huadong Sanatorium Information Center for its assistance with the data. They acknowledge the Computer Center at the National University of Singapore for supporting the SIVA program.
ABBREVIATIONS
- AHI
apnea-hypopnea index
- AVR
arteriole-to-venular diameter ratio
- BMI
body mass index
- CRAE
central retinal arteriolar equivalent
- CRVE
central retinal venular equivalent
- FD
fractal dimension
- FDa
fractal dimension arteriole
- FDv
fractal dimension venule
- OSA
obstructive sleep apnea
- OSAS
obstructive sleep apnea syndrome
- PD
papillary diameter
- PSG
polysomnography
REFERENCES
- 1. Gottlieb DJ , Punjabi NM . Diagnosis and management of obstructive sleep apnea: a review . JAMA. 2020. ; 323 ( 14 ): 1389 – 1400 . [DOI] [PubMed] [Google Scholar]
- 2. Benjafield AV , Ayas NT , Eastwood PR , et al . Estimation of the global prevalence and burden of obstructive sleep apnoea: a literature-based analysis . Lancet Respir Med. 2019. ; 7 ( 8 ): 687 – 698 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Stanaway JD , Afshin A , Gakidou E , et al. GBD 2017 Risk Factor Collaborators . Global, regional, and national comparative risk assessment of 84 behavioural, environmental and occupational, and metabolic risks or clusters of risks for 195 countries and territories, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017 . Lancet. 2018. ; 392 ( 10159 ): 1923 – 1994 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Bassetti CLA , Randerath W , Vignatelli L , et al . EAN/ERS/ESO/ESRS statement on the impact of sleep disorders on risk and outcome of stroke . Eur Respir J. 2020. ; 55 ( 4 ): 1901104 . [DOI] [PubMed] [Google Scholar]
- 5. Mentek M , Aptel F , Godin-Ribuot D , Tamisier R , Pepin JL , Chiquet C . Diseases of the retina and the optic nerve associated with obstructive sleep apnea . Sleep Med Rev. 2018. ; 38 : 113 – 130 . [DOI] [PubMed] [Google Scholar]
- 6. D’Souza H , Kapoor KG . Retinal vascular manifestations of obstructive sleep apnea . Curr Opin Ophthalmol. 2020. ; 31 ( 6 ): 508 – 513 . [DOI] [PubMed] [Google Scholar]
- 7. Sasongko MB , Wong TY , Donaghue KC , et al . Retinal arteriolar tortuosity is associated with retinopathy and early kidney dysfunction in type 1 diabetes . Am J Ophthalmol. 2012. ; 153 ( 1 ): 176 – 83.e1 . [DOI] [PubMed] [Google Scholar]
- 8. Masters BR . Fractal analysis of the vascular tree in the human retina . Annu Rev Biomed Eng. 2004. ; 6 ( 1 ): 427 – 452 . [DOI] [PubMed] [Google Scholar]
- 9. Wang XY , Wang S , Liu X , Ding X , Li M , Han DM . Retinal vascular morphological changes in patients with extremely severe obstructive sleep apnea syndrome . Chin Med J (Engl). 2017. ; 130 ( 7 ): 805 – 810 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Tong JY , Golzan M , Georgevsky D , et al . Quantitative retinal vascular changes in obstructive sleep apnea . Am J Ophthalmol. 2017. ; 182 : 72 – 80 . [DOI] [PubMed] [Google Scholar]
- 11. Liew G , Gopinath B , White AJ , Burlutsky G , Yin Wong T , Mitchell P . Retinal vasculature fractal and stroke mortality . Stroke. 2021. ; 52 ( 4 ): 1276 – 1282 . [DOI] [PubMed] [Google Scholar]
- 12. Huang F , Dashtbozorg B , Zhang J , et al . Reliability of using retinal vascular fractal dimension as a biomarker in the diabetic retinopathy detection . J Ophthalmol. 2016. ; 2016 : 6259047 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Rim TH , Teo AWJ , Yang HHS , Cheung CY , Wong TY . Retinal vascular signs and cerebrovascular diseases . J Neuroophthalmol. 2020. ; 40 ( 1 ): 44 – 59 . [DOI] [PubMed] [Google Scholar]
- 14. Yang D , Rundek T , Patel SR , et al . Cerebral hemodynamics in sleep apnea and actigraphy-determined sleep duration in a sample of the Hispanic Community Health Study/Study of Latinos . J Clin Sleep Med. 2019. ; 15 ( 1 ): 15 – 21 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Cai Y , Sun GS , Zhao L , Han F , Zhao MW , Shi X . Quantitative evaluation of retinal microvascular circulation in patients with obstructive sleep apnea-hypopnea using optical coherence tomography angiography . Int Ophthalmol. 2020. ; 40 ( 12 ): 3309 – 3321 . [DOI] [PubMed] [Google Scholar]
- 16. Veasey SC , Rosen IM . Obstructive sleep apnea in adults . N Engl J Med. 2019. ; 380 ( 15 ): 1442 – 1449 . [DOI] [PubMed] [Google Scholar]
- 17. Rajesh S , Wonderling D , Simonds AK ; Guideline Committee . Obstructive sleep apnoea/hypopnoea syndrome and obesity hyperventilation syndrome in over 16s: summary of NICE guidance . BMJ. 2021. ; 375 ( 2360 ): n2360 . [DOI] [PubMed] [Google Scholar]
- 18. Cheung CY , Tay WT , Mitchell P , et al . Quantitative and qualitative retinal microvascular characteristics and blood pressure . J Hypertens. 2011. ; 29 ( 7 ): 1380 – 1391 . [DOI] [PubMed] [Google Scholar]
- 19. Walker E . Applied regression analysis and other multivariable methods . Technometrics. 1989. ; 31 ( 1 ): 117 – 118 . [Google Scholar]
- 20. Ab Hamid F , Che Azemin MZ , Salam A , Aminuddin A , Mohd Daud N , Zahari I . Retinal vasculature fractal dimension measures vessel density . Curr Eye Res. 2016. ; 41 ( 6 ): 823 – 831 . [DOI] [PubMed] [Google Scholar]
- 21. Zekavat SM , Raghu VK , Trinder M , et al . Deep learning of the retina enables phenome- and genome-wide analyses of the microvasculature . Circulation. 2022. ; 145 ( 2 ): 134 – 150 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Liew G , Wang JJ , Cheung N , et al . The retinal vasculature as a fractal: methodology, reliability, and relationship to blood pressure . Ophthalmology. 2008. ; 115 ( 11 ): 1951 – 1956 . [DOI] [PubMed] [Google Scholar]
- 23. Broe R , Rasmussen ML , Frydkjaer-Olsen U , et al . Retinal vascular fractals predict long-term microvascular complications in type 1 diabetes mellitus: the Danish Cohort of Pediatric Diabetes 1987 (DCPD1987) . Diabetologia. 2014. ; 57 ( 10 ): 2215 – 2221 . [DOI] [PubMed] [Google Scholar]
- 24. Liew G , Mitchell P , Rochtchina E , et al . Fractal analysis of retinal microvasculature and coronary heart disease mortality . Eur Heart J. 2011. ; 32 ( 4 ): 422 – 429 . [DOI] [PubMed] [Google Scholar]
- 25. Yau JW , Kawasaki R , Islam FM , et al . Retinal fractal dimension is increased in persons with diabetes but not impaired glucose metabolism: the Australian Diabetes, Obesity and Lifestyle (AusDiab) Study . Diabetologia. 2010. ; 53 ( 9 ): 2042 – 2045 . [DOI] [PubMed] [Google Scholar]
- 26. Aliahmad B , Kumar DK , Sarossy MG , Jain R . Relationship between diabetes and grayscale fractal dimensions of retinal vasculature in the Indian population . BMC Ophthalmol. 2014. ; 14 ( 1 ): 152 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Ţălu Ş , Călugăru DM , Lupaşcu CA . Characterisation of human non-proliferative diabetic retinopathy using the fractal analysis . Int J Ophthalmol. 2015. ; 8 ( 4 ): 770 – 776 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Auvazian SL , Cano J , Leahy S , et al . Relating retinal vascular oxygen saturation and microvasculature morphology at progressive stages of diabetic retinopathy . Transl Vis Sci Technol. 2021. ; 10 ( 6 ): 4 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Felder AE , Wanek J , Blair NP , et al . The effects of diabetic retinopathy stage and light flicker on inner retinal oxygen extraction fraction . Invest Ophthalmol Vis Sci. 2016. ; 57 ( 13 ): 5586 – 5592 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Ava S , Erdem S , Karahan M , et al . Evaluation of the effect of obstructive sleep apnea syndrome on retinal microvascularity by optical coherence tomography angiography . Photodiagnosis Photodyn Ther. 2022. ; 38 : 102761 . [DOI] [PubMed] [Google Scholar]
- 31. Moyal L , Blumen-Ohana E , Blumen M , Blatrix C , Chabolle F , Nordmann JP . Parafoveal and optic disc vessel density in patients with obstructive sleep apnea syndrome: an optical coherence tomography angiography study . Graefes Arch Clin Exp Ophthalmol. 2018. ; 256 ( 7 ): 1235 – 1243 . [DOI] [PubMed] [Google Scholar]
- 32. Shiba T , Takahashi M , Matsumoto T , Hori Y . Gender differences in the influence of obstructive sleep apnea on optic nerve head circulation . Sci Rep. 2019. ; 9 ( 1 ): 18849 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Kohler M , Stradling JR . Mechanisms of vascular damage in obstructive sleep apnea . Nat Rev Cardiol. 2010. ; 7 ( 12 ): 677 – 685 . [DOI] [PubMed] [Google Scholar]
- 34. West SD , Turnbull C . Obstructive sleep apnoea . Eye (Lond). 2018. ; 32 ( 5 ): 889 – 903 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Patt BT , Jarjoura D , Haddad DN , et al . Endothelial dysfunction in the microcirculation of patients with obstructive sleep apnea . Am J Respir Crit Care Med. 2010. ; 182 ( 12 ): 1540 – 1545 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Gavrilin MA , Porter K , Samouilov A , Khayat RN . Pathways of microcirculatory endothelial dysfunction in obstructive sleep apnea: a comprehensive ex vivo evaluation in human tissue . Am J Hypertens. 2022. ; 35 ( 4 ): 347 – 355 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Bonsignore MR , Saaresranta T , Riha RL . Sex differences in obstructive sleep apnoea . Eur Respir Rev. 2019. ; 28 ( 154 ): 190030 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Johns MW . A new method for measuring daytime sleepiness: the Epworth Sleepiness Scale . Sleep. 1991. ; 14 ( 6 ): 540 – 545 . [DOI] [PubMed] [Google Scholar]