Bidirectional association between Parkinson’s disease and obstructive sleep apnea: a cohort study

Abstract

Study Objectives:

Chronic intermittent hypoxia due to obstructive sleep apnea (OSA) causes oxidative stress, which may contribute to the pathophysiology of Parkinson’s disease (PD). However, the bidirectional relationship between PD and OSA has not been satisfactorily established. The objective of this study was to try to estimate whether there is a bidirectional relationship between PD and OSA through a retrospective cohort study in the South Korean population.

Methods:

This study used data from the Korean National Health Information Database of the National Health Insurance Service, which contains data from 3.5 million individuals evenly distributed. In study 1, patients with OSA were matched in a 1:2 ratio with non-OSA controls. In study 2, patients with PD were matched in a 1:2 ratio with non-PD controls. A stratified Cox proportional hazards model was used to calculate hazard ratios.

Results:

In study 1, which included 6,396 patients with OSA and 12,792 non-OSA controls, the incidence of PD per 10,000 person-years was 11.59 in the OSA group and 8.46 in the non-OSA group. The OSA group demonstrated a 1.54-fold higher incidence of PD than the non-OSA group (95% confidence interval, 1.14–2.07; P < .05). In study 2, which included 3,427 patients with PD and 6,854 non-PD controls, the incidence of OSA per 10,000 person-years was 14.97 in the PD group and 7.72 in the non-PD group. The PD group demonstrated a 1.92-fold higher incidence of OSA than the non-PD group (95% confidence interval, 1.32–2.78; P < .05).

Conclusions:

This study supports a possible bidirectional relationship between PD and OSA.

Citation:

Jeon S-H, Hwang YS, Oh S-Y, et al. Bidirectional association between Parkinson’s disease and obstructive sleep apnea: a cohort study. J Clin Sleep Med. 2023;19(9):1615–1623.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Obstructive sleep apnea (OSA) causes oxidative stress, which may contribute to the pathophysiology of Parkinson’s disease (PD); however, the quantitative bidirectional relationship between PD and OSA has not been satisfactorily established in various ethnic populations.

Study Impact: Among the OSA and propensity score–matched non-OSA groups, the overall hazard ratio of OSA for PD was 1.54 (95% confidence, 1.14–2.1). Similarly, the overall HR of PD for OSA was 1.92 (95% confidence interval,1.32–2.78). A bidirectional association was observed between PD and OSA, which suggests a possible shared mechanism between PD and OSA.

INTRODUCTION

Parkinson’s disease (PD) is the second most common neurodegenerative disease and is a multisystem disorder.¹ Cardinal features of PD are motor symptoms of resting tremor, bradykinesia, and rigidity, but these are accompanied by a wide range of nonmotor features such as hyposmia, gastrointestinal dysfunction, psychiatric problems, cognitive impairment, and sleep problems.^2,3 Among nonmotor symptoms, sleep problems occur in 74–98% of patients with PD, and sleep fragmentation is frequently observed in patients with sleep problems. Sleep fragmentation is a cause of excessive daytime sleepiness (EDS) and fatigue, is more common in PD than in age-matched populations, and worsens with disease progression.^2,4–6

Obstructive sleep apnea (OSA) is a treatable sleep disorder frequently observed in the general population, and arousal from sleep with oxygen desaturation occurs because of complete or partial obstruction of the upper airway during sleep.⁶ Correlations between OSA and the onset of cognitive dysfunction, metabolic disease, coronary artery disease, arrhythmia, and stroke have been successfully established.^7,8 Snoring is a major risk factor for OSA, and is also observed in 70% of patients with PD. Intermittent cerebral hypoxia caused by oxygen desaturation causes oxidative stress and inflammation, leading to alpha-synuclein aggregation and dopaminergic cell death.^9,10

The aforementioned studies suggest the possibility that physiological changes accompanying OSA may elevate the risk of PD. To date, there have been several studies supporting this relationship despite inconsistent results.^11–14 However, there have been very few studies in large populations, and these studies are concentrated in just one country, Taiwan. Therefore, studies in various ethnic populations are required to establish the relationship between OSA and PD. In addition, although PD and OSA share the same manifestation of snoring, there have only been a few studies on the incident risk of OSA in patients with PD, each based on a small number of patients.¹⁵ Therefore, the incidence of OSA in patients with PD has not been satisfactorily established.

Through a nationwide cohort study in South Korea, this study aimed to examine the possible bidirectional association between PD and OSA and discusses their relationship based on shared pathophysiology.

METHODS

Data source

We used data from the National Health Information Database of the National Health Insurance Service (NHIS-NHID) (NHIS 2021-1-689). The NHIS-NHID cohort includes data from 3.5 million individuals evenly distributed through random stratified sampling of the data from the 50 million Korean nationals and includes data for age, sex, residential area, and economic status. In addition, the NHIS-NHID cohort includes hospitalization date, diagnostic code, treatment history during hospitalization, medication history, and death information. The Ethics Committee of Jeonbuk National University Hospital (2021–current) approved the use of these data. The requirement for written informed consent was waived by the Institutional Review Board.

Patient selection

The patient groups were defined as follows: OSA was based on the International Classification of Diseases, 10th Revision (ICD-10), and the participants with OSA were assigned the morbidity code G473 at least once.¹⁶ Participants with PD were assigned code G20 at least once. Study 1 contained data from patients diagnosed with OSA in the 4-year period between January 1, 2008, and December 31, 2012, while study 2 contained data from patients diagnosed with PD in the same period. With data up to December 31, 2019, the follow-up period ranged from 7 to 11 years.

Study setting

Study 1

The experimental cohort was defined as those diagnosed with OSA between January 1, 2008, and December 31, 2012. The moderator variable was surgery for OSA. OSA surgery was defined as patients receiving the following treatment codes during the follow-up period: Q2195, Q2196, Q2197, O1001, O1002, O1003, Q2280, Q2281, Q2300, and Q2310. The definition of target disease was patients diagnosed with PD between January 1, 2008, and December 31, 2012. To avoid selection bias, data from 6,396 patients with OSA were matched in a 1:2 ratio with 12,792 non-OSA control participants, and the hazard ratio (HR) for PD in patients with OSA was analyzed. The adjusted HR was obtained by considering all of the other variables and the unadjusted HR was calculated without considering other variables.

The exclusion criteria were as follows: (1) patients diagnosed with OSA during the period 2013–2019 and (2) patients with records for both OSA and PD in whom PD was diagnosed earlier or at the same time as diagnosis of OSA. The starting point for the OSA group was the date of diagnosis of OSA, and the end point was set as the date of first diagnosis of PD if there was a record of diagnosis of PD or December 31, 2019, if there was no record of PD diagnosis.

Study 2

The criteria used for study 2 were the reverse of those used in study 1. The experimental cohort was defined as patients diagnosed with PD between January 1, 2008, and December 31, 2012. Target disease was defined as patients diagnosed with OSA between January 1, 2008, and December 31, 2012. Data from 3,427 patients with PD were matched in a 1:2 ratio with 6,854 non-PD control participants, and the HR for OSA in patients with PD was analyzed. The adjusted HR was obtained by considering all of the other variables and the unadjusted HR was calculated without considering other variables.

The exclusion criteria were as follows: (1) patients diagnosed with PD during 2013–2019 and (2) patients with records for both OSA and PD in whom OSA was diagnosed earlier or at the same time as diagnosis of PD. The starting point for the PD group was the date of PD diagnosis, and the end point was set as the date of first diagnosis of OSA if there was a record of OSA diagnosis or December 31, 2019, if there was no record of OSA diagnosis.

Statistical analysis

The control groups were obtained by propensity score (PS) matching from the experimental cohort, considering other variables using a greedy-nearest-neighbor algorithm. Confirmation that PS matching was appropriate was judged quantitatively using standardized mean differences (SMDs) and qualitatively using numerical values.^17–23 All independent variables were divided into two or three categories as follows:

1. Age < 50 and age ≥ 50 years

2. Female or male sex

3. Residential area—metropolitan or rural areas

4. From the 10th quintile for economic status, the lower 30% were of low economic status, the higher 30% were of high economic status and the rest were of average economic status

5. Obesity (body mass index > 25 kg/m²) or normal

6. History or no history of hypertension (HTN)

7. History or no history of diabetes mellitus (DM)

8. History or no history of chronic kidney disease (CKD)

The definitions of histories 6–8 are provided in Table S1 ^{(202.3KB, pdf)} in the supplemental material. Sensitivity analysis was performed according to (1) before and after PS matching; (2) matching ratio of 1:1, 1:2, or 1:3; (3) matching variables (minimally matching [sex + age] or fully matching; and (4) the recruitment period of patients with OSA or PD and controls (3, 4, or 5 years).^17–23 The robustness of the HR was evaluated according to matching variables.

Data analysis was conducted between November and December 2021. Two data analysts (M.G.K., S.W.Y.) independently conducted data analysis. The HR in the Cox proportional hazards model was calculated considering the time variable (between the end point and start point). When considering the relationship between one independent variable and the dependent variable, the adjusted HR was obtained by considering all other variables, while the unadjusted HR was calculated without considering other variables. The cumulative HR was obtained through Kaplan–Meier survival analysis, and the R 3.5.3 statistical program (R Foundation for Statistical Computing, Vienna, Austria) was used to analyze the results.

RESULTS

Study 1

The number of participants in the OSA group was 6,396 and the number in the PS-matched non-OSA control group was 12,792. These groups were well distributed with regard to sex, age, economic status, residential area, DM, HTN, and CKD (Table 1). All of the SMDs were less than 0.1, indicating that they were well distributed.

Table 1.

Demographics of OSA and non-OSA groups: study 1.

Variable	Control (non-OSA) Group (n = 12,792)	Study (OSA) Group (n = 6,396)	SMD
Sex			0.024
Male	10,553	5,218
Female	2,239	1,178
Age			0.022
Young < 50 years	8,825	4,478
Old ≥ 50 years	3,967	1,918
Obesity			0.026
No	5,507	2,670
Yes	7,285	3,726
Residential area			0.017
Rural	6,238	3,172
Metropolitan	6,554	3,224
Economic status			0.057
Low	1,322	768
Middle	6,563	3,284
High	4,907	2,344
HTN			0.070
No	9,700	4,655
Yes	3,092	1,741
DM			0.100
No	11,747	5,683
Yes	1,045	713
CKD			0.059
No	12,241	6,038
Yes	551	358
OSA surgery			0.805
No	12,792	4,830
Yes	0	1,566

CKD = chronic kidney disease, DM = diabetes mellitus, HTN = hypertension, OSA = obstructive sleep apnea, SMD = standardized mean difference.

The incidence of PD per 10,000 person-years was 11.59 in the OSA group and 8.46 in the non-OSA control group. The adjusted HR for PD in patients with OSA was 1.54 (95% confidence interval [CI]: 1.14–2.07) (Figure 1A, Figure 2A, Table 2). The difference in unadjusted HR and adjusted HR was not significant. Surgery for OSA significantly lowered the incidence of PD (HR: 0.47; 95% CI: 0.21–0.99) (Figure 1B, Figure 2A, Table 2). In addition to OSA and OSA surgery, age ≥ 50 years, HTN, and CKD history were significant factors in a multivariate model considering 11 variables. The adjusted HRs for these three factors were 6.19 (95% CI: 4.17–9.18), 2.18 (95% CI: 1.56–3.04), and 1.68 (95% CI: 1.11–2.55), respectively (Figure 2A, Table 2).

Figure 1. Cumulative incidence plots.

(A) Overall cumulative incidence plot for PD in the OSA and matched control (non-OSA) groups. (B) Cumulative incidence plot for PD in the group who had undergone surgery for OSA and the matched control (non-OSA operation) group. (C) Overall cumulative incidence plot for OSA in the PD and matched control (non-PD) groups. OSA = obstructive sleep apnea, PD = Parkinson’s disease.

Figure 2. Forest plots.

(A) Forest plot of the cumulative hazard ratio for each factor for the occurrence of PD. (B) Forest plot of the cumulative hazard ratio for each factor for the occurrence of OSA. CI = confidence interval, CKD = chronic kidney disease, DM = diabetes mellitus, HTN = hypertension, OSA = obstructive sleep apnea, PD = Parkinson’s disease.

Table 2.

Incidence rate, adjusted and unadjusted HRs for each group: study 1.

Variable	Study Group Total	Number of Cases	10,000 PY	HR Adjusted	HR Unadjusted
Total	19,188	192
Group
Non-OSA	12,792	114	8.46	1	1
OSA	6,396	78	11.59	1.54 (1.14–2.07)	1.39 (1.04–1.85)
Sex
Female	3,417	42	11.72	1	1
Male	15,771	150	9.02	0.99 (0.70–1.41)	0.75 (0.54–1.06)
Age
Young < 50 years	13,303	37	2.64	1	1
Old ≥ 50 years	5,885	155	25.13	6.19 (4.17–9.18)	9.59 (6.70–13.72)
Obesity
No	8,177	82	9.54	1	1
Yes	11,011	110	9.47	0.95 (0.71–1.28)	0.98 (0.74–1.30)
Residential area
Rural	9,778	99	9.62	1	1
Metropolitan	9,410	93	9.38	0.95 (0.71–1.26)	0.97 (0.73–1.28)
Economic status
Middle	9,847	76	7.33	1	1
Low	2,090	28	12.78	1.48 (0.96–2.29)	1.75 (1.13–2.69)
High	7,251	88	11.51	1.22 (0.88–1.64)	1.59 (1.17–2.15)
HTN
No	14,355	70	4.36	1	1
Yes	4,833	122	24.03	2.18 (1.56–3.04)	5.22 (3.89–7.00)
DM
No	17,430	150	8.17	1	1
Yes	1,758	42	22.66	0.95 (0.66–1.36)	2.75 (1.95–3.87)
CKD
No	18,279	164	8.52	1	1
Yes	909	28	29.42	1.68 (1.11–2.55)	3.43 (2.30–5.12)
OSA surgery
No	17,662	185	9.97	1	1
Yes	1,566	7	4.25	0.47 (0.21–0.99)	0.42 (0.20–0.90)

CKD = chronic kidney disease, DM = diabetes mellitus, HR = hazard ratio, HTN = hypertension, OSA = obstructive sleep apnea, 10,000 PY = incidence per 10,000 person-years.

Sensitivity analyses also presented similar significant associations of OSA with the occurrence of PD, except for the analysis without PS matching (Figure 2A).

Study 2

The number of participants in the PD group was 3,427 and the number in the PS-matched non-PD control group was 6,854. These groups were well distributed with regard to sex, age, economic status, residential area, DM, HTN, and CKD (Table 3). All of the SMDs were less than 0.1, indicating that they were well distributed.

Table 3.

Demographics of PD and non-PD groups: study 2.

Variable	Control (non-PD) Group (n = 6,854)	Study (PD) Group (n = 3,427)	SMD
Sex			0.014
Male	2,836	1,441
Female	4,018	1,986
Age			0.029
Young < 50 years	665	363
Old ≥ 50 years	6,189	3,064
Obesity			0.004
No	4,531	2,259
Yes	2,323	1,168
Residential area			0.012
Rural	4,087	2,063
Metropolitan	2,767	1,364
Economic status			0.024
Low	1,139	598
Middle	3,259	1,628
High	2,456	1,201
HTN			0.004
No	2,595	1,291
Yes	4,259	2,136
DM			0.034
No	4,623	2,256
Yes	2,231	1,171
CKD			0.050
No	6,023	2,954
Yes	831	473
TH			0.024
No	6,591	3,279
Yes	263	148

CKD = chronic kidney disease, DM = diabetes mellitus, HTN = hypertension, OSA = obstructive sleep apnea, PD = Parkinson’s disease, SMD = standardized mean difference, TH = tonsillar hypertrophy.

The incidence of OSA per 10,000 person-years was 14.97 in the PD group and 7.72 in the non-PD control group. The adjusted HR for OSA in patients with PD was 1.92 (95% CI: 1.32–2.78) (Figure 1C, Figure 2B, Table 4). The difference in unadjusted HR and adjusted HR was not significant. In addition to PD, male sex, age ≥ 50 years, and obesity were significant factors in a multivariate model considering 10 variables. The adjusted HRs for these three factors were 2.19 (95% CI: 1.49–3.20), 0.54 (95% CI: 0.32–0.91), and 1.81 (95% CI: 1.24–2.63), respectively (Figure 2B, Table 4).

Table 4.

Incidence rate, adjusted and unadjusted HR for each group: study 2.

Variable	Study Group Total	Number of Cases	10,000 PY	HR Adjusted	HR Unadjusted
Total	10,281	112
Group
Non-PD	6,854	57	7.72	1	1
PD	3,427	55	14.97	1.92 (1.32–2.78)	1.94 (1.34–2.81)
Sex
Female	6,004	44	6.8	1	1
Male	4,277	68	14.83	2.19 (1.49–3.20)	2.18 (1.49–3.18)
Age
Young < 50 years	1,028	21	19.2	1	1
Old ≥ 50 years	9,253	91	9.13	0.54 (0.32–0.91)	0.48 (0.30–0.77)
Obesity
No	6,790	60	8.2	1	1
Yes	3,491	52	13.9	1.81 (1.24–2.63)	1.69 (1.17–2.45)
Economic status
Middle	4,887	48	9.13	1	1
Low	1,737	16	8.57	0.94 (0.53–1.65)	0.94 (0.53–1.65)
High	3,657	48	12.2	1.39 (0.93–2.08)	1.34 (0.90–1.99)
Residential area
Rural	6,150	62	9.36	1	1
Metropolitan	4,131	50	11.28	1.13 (0.78–1.65)	1.20 (0.83–1.75)
HTN
No	3,886	52	12.44	1	1
Yes	6,395	60	8.72	0.72 (0.48–1.10)	0.70 (0.48–1.02)
DM
No	6,879	76	10.25	1	1
Yes	3,402	36	9.88	1.04 (0.68–1.60)	0.96 (0.65–1.43)
CKD
No	8,977	94	9.72	1	1
Yes	1,304	18	12.96	1.41 (0.84–2.36)	1.33 (0.80–2.20)
TH
No	9,870	107	10.08	1	1
Yes	411	5	11.33	1.04 (0.43–2.57)	1.12 (0.46–2.76)

CKD = chronic kidney disease, DM = diabetes mellitus, HTN = hypertension, OSA = obstructive sleep apnea, PD = Parkinson’s disease, SMD = standardized mean difference, TH = tonsillar hypertrophy, 10,000 PY = incidence per 10,000 person-years.

Sensitivity analyses also presented similar significant associations of PD with the occurrence of OSA, except for the analysis without PS matching (Figure 2B).

DISCUSSION

Our study suggests a possible association between OSA and PD. The main finding was that the incidence of PD was 1.54 times higher in the OSA cohort, after adjusting for possible confounding factors. The fact that the incidence of OSA was 1.92 times higher in the PD cohort was also a difference that has not been reported in other studies. These findings confirmed the bidirectional association between PD and OSA.

The role of intermittent hypoxia caused by OSA in PD remains uncertain. However, intermittent hypoxemia can cause cerebral hypoperfusion, and the resulting oxidative stress leads to endothelial dysfunction, which is highly associated with systemic inflammation.^24,25 In addition, in animal experiments, OSA induced brain hypoxia and oxidative stress induced an increase in astrocytes and neuronal apoptosis.^26,27 Oxidative stress also results in lipid membrane peroxidation, which disrupts the membrane, activation of glial cells, and the release of proinflammatory cytokines.^28,29 These various mechanisms may contribute to neuroinflammation, which plays an important role in neurodegenerative diseases such as Alzheimer’s disease and PD.^30,31 In particular, sympathetic overactivity during hypoxia in OSA might promote these neuroinflammatory processes via significant surges in proinflammatory cytokines and reactive oxygen species triggered by hyperactive brain renin angiotensin system (RAS) effector peptides, such as angiotensin II.³² In addition to these mechanisms, it has also been confirmed in laboratory studies that oxidative stress induces alpha-synuclein aggregation and dopaminergic cell death.^26,33 These possible mechanisms suggest that OSA may contribute to the development of PD. Moreover, our results indicated a lower risk of PD in patients with OSA who underwent surgery than in those who did not, which supports the hypothesis that OSA is a risk factor for PD.

We also found that the incidence of OSA increased in patients with PD compared with non-PD controls. Some researchers are of the opinion that OSA is less common in PD and does not cause daytime sleepiness. The strongest risk factor for OSA is obesity.³⁴ However, patients with PD had a lower body mass index than healthy controls, and the higher the disease severity, the lower the body mass index.³⁵ Therefore, the risk of OSA in patients with PD might be underestimated.

Few studies have directly compared OSA incidence with PD in the general population; however, there is evidence to support the contribution of brainstem neurodegeneration to the development of OSA. Pharyngeal narrowing due to degeneration of brainstem neurons that act on the pharyngeal dilator muscle plays a very important role in airway patency during sleep in patients with PD.³⁶ In addition, various respiratory disorders, including upper airway obstruction, in patients with PD was noted to improve with dopaminergic medication treatment, and akinesia or impaired central coordination respiratory muscles was suggested as the underlying mechanisms for these respiratory disorders in PD.^37,38 And in one previous study, long-acting levodopa treatment not only improved nocturnal motor function but also reduced the severity of OSA in patients with PD.³⁹ These support the hypothesis that OSA might be one of the nonmotor symptoms of PD, which is consistent with our finding.

Excessive daytime sleepiness (EDS) is a common symptom in patients with OSA, and OSA also has an effect on the occurrence of EDS in patients with PD.³⁰ In addition, EDS correlated with longer duration of PD and disease severity, and cognitive decline was observed more frequently compared with patients without somnolence.^40,41 Therapeutic continuous positive airway pressure (CPAP) reduced daytime sleepiness in patients with PD with OSA by reducing the apnea-hypopnea index, reducing the time for SaO₂ to fall below 90%, decreasing the percentage of N2-stage sleep and increasing the percentage of N3-stage sleep.⁴² In a large community-based study, 15.5% of patients with PD had EDS, which was directly related to the PD stage, and the more severe the EDS, the more frequently cognitive decline was observed.^41,43 These facts suggest that OSA is not only a risk factor for PD but also a symptom of it, and an important factor that affects the clinical course of PD. In this study, we highlighted the reciprocal effect of PD and OSA.

In addition, in our study, HTN and CKD history were significant risk factors for PD, and male sex and obesity were significant risk factors for OSA. Older age was significantly associated with both OSA and PD risk. These results are somewhat different from the currently accepted risk factors for PD and OSA. Systemic diseases such as CKD, DM, and HTN are often reported as risk factors for PD^44,45; however, as yet, they are not commonly accepted risk factors. It should be noted that this study has a PS-matched design to analyze the relationship between PD and OSA. Because data for these risk factors, which included CKD, DM, and HTN, were collected and controlled as variables for adjusting for PS matching in this study, the HRs of these variables cannot completely represent the actual effects on PD and OSA. Moreover, CKD and HTN were significantly associated with the occurrence of PD, but DM was not, and these systemic diseases did not have a significant effect on the occurrence of OSA, even though they are possibly acceptable risk factors for OSA. These results suggest that data for the variables used for PS matching must be carefully interpreted.^46,47

Another possible reason for these unexpected results is from the introduction of a bias caused by the inadvertent inclusion of vascular parkinsonism in this study based on our analysis of the diagnostic code. Interestingly, the results related to sex in our study have some validity. In OSA, male sex is a known risk factor,⁴⁸ and this is consistent with our results and actual epidemiology in South Korea. The risk of PD is generally increased 1.5-fold in male individuals⁴⁹; however, in South Korea, both incidence and prevalence of PD are higher in female individuals with population-specific epidemiology.⁵⁰

This is the first nationwide cohort study in which a bidirectional association between PD and OSA has been observed. However, this study has several limitations. First, the diagnosis of PD was investigated based on the diagnostic code, so it was difficult to accurately distinguish between Parkinson’s plus syndromes and secondary parkinsonism from primary parkinsonism. However, the diagnosis of PD can be considered relatively accurate because the benefit extension policy was applied, and the diagnosis was made by a neurologist or neurosurgeon. Second, the diagnosis of OSA was investigated based on the diagnostic code. It is impossible to evaluate the severity of sleep apnea using the apnea-hypopnea index; therefore, changes in the incidence of PD based on OSA severity and current treatment could not be identified. Furthermore, while the status for OSA surgery was analyzed in our study, the status of CPAP treatment was unobtainable in our cohort data based on health insurance services because CPAP has only been covered by insurance very recently in South Korea. Third, it was impossible to distinguish whether OSA was a central or obstructive type using only the diagnostic code. Fourth, a patient’s family history, drinking history, dietary pattern, daily exercise, and other health-related conditions act as significant confounding factors not only for OSA but also for PD. However, information regarding these individual factors could not be considered. To correct for these points, social status was adjusted for monthly income, geographic location, etc. Finally, in our study, preferential directionality was found in the HR of PD for the occurrence of OSA compared with the HR of OSA for the occurrence of PD; however, we have not proposed an appropriate hypothesis supporting this result. This should be clarified by further studies.

The incidence of OSA is increased in patients with PD. However, OSA also contributed to an increased risk of PD via various mechanisms, which included oxidative stress and sympathetic overactivity resulting from hypoxia and dopaminergic cell death with alpha-synuclein aggregation. In conclusion, OSA and PD have an adverse effect on each other; therefore, early detection and management of OSA is not only a symptomatic treatment for PD but might also be one of the clues for a disease-modifying strategy for PD.

DISCLOSURE STATEMENT

This study was supported by the project for Industry-Academic Cooperation Based Platform R&D funded by the Korea Ministry of Small and Medium Enterprises (SMEs) and Startups in 2021 (no. S3017921); the Fund of Biomedical Research Institute, Jeonbuk National University Hospital (CUH 2022-0007); an NRF grant funded by the Korean Government (MSIT; no. 2021R1G1A1094681); and by a grant from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (grant number: HI22C1124). The authors report no conflicts of interest.

ABBREVIATIONS

CI

confidence interval

CKD

chronic kidney disease

DM

diabetes mellitus

EDS

excessive daytime sleepiness

HR

hazard ratio

HTN

hypertension

OSA

obstructive sleep apnea

PD

Parkinson’s disease

PS

propensity score

SMD

standardized mean difference