Sleep Disorders and Dementia Risk in Older Patients with Kidney Failure

Visual Abstract

Abstract

Key Points

• Older patients with kidney failure who are newly diagnosed with sleep disorders are at higher risk of developing any type of dementia, vascular dementia, and other/mixed types of dementia.

• For older patients with kidney failure who are diagnosed with obstructive sleep apnea, positive airway pressure therapy is an intervention that is associated with lower dementia risk.

Background

Community-dwelling older adults with sleep disorders are at higher risk of developing dementia. Greater than 50% of older patients with kidney failure experience sleep disorders, which may explain their high burden of dementia.

Methods

Among 216,158 patients (aged 66 years and older) with kidney failure (United States Renal Data System; 2008–2019), we estimated the risk of dementia (including subtypes) associated with sleep disorders using Cox proportional hazards models with propensity score weighting. We tested whether positive airway pressure (PAP) therapy was associated with reduced dementia risk among patients with obstructive sleep apnea (OSA).

Results

26.3% of patients were diagnosed with sleep disorders; these patients had a higher 5-year unadjusted cumulative incidence for any type of dementia (36.2% versus 32.3%; P < 0.001), vascular dementia (4.4% versus 3.7%; P < 0.001), and other/mixed dementia (29.3% versus 25.8%; P < 0.001). Higher risk of any type of dementia was identified in patients with insomnia (adjusted hazard ratio [aHR], 1.42; 95% confidence interval [CI], 1.34 to 1.51), sleep-related breathing disorders (SRBDs) (aHR, 1.20, 95% CI, 1.17 to 1.23), and other sleep disorders (aHR, 1.24; 95% CI, 1.11 to 1.39). Higher vascular dementia risk was observed in patients with insomnia (aHR, 1.43; 95% CI, 1.19 to 1.73) and SRBDs (aHR, 1.15; 95% CI, 1.07 to 1.24). Patients with SRBDs (aHR, 1.07; 95% CI, 1.00 to 1.15) were at higher risk of Alzheimer disease. Among patients with OSA, PAP therapy was associated with lower risk of any type of dementia (aHR, 0.82; 95% CI, 0.76 to 0.90) and vascular dementia (aHR, 0.65; 95% CI, 0.50 to 0.85).

Conclusions

Older patients with kidney failure and sleep disorders are at a higher risk of dementia. Sleep is an important modifiable factor that should be considered for targeted interventions to mitigate dementia risk in patients with kidney failure. For patients with OSA, PAP therapy is associated with lower dementia risk.

Introduction

Older patients with kidney failure have compromised cognition^1,2 because of comorbidities and dialysis, resulting in a higher risk of dementia and Alzheimer disease (AD).^3–5 Once diagnosed with dementia, older patients with kidney failure are at a two-fold higher risk of mortality and graft failure.^3,5–8

Sleep is a complex biological process that is important for cognitive function, including attention and memory consolidation.^9,10 Community-dwelling older adults with sleep disorders, including insomnia and sleep-related breathing disorders (SRBDs), are at higher odds for developing dementia and AD.^9,11–15 Older women with SRBD have an 85% greater risk of mild cognitive impairment or dementia.¹⁶ The relationships between sleep disorders and cognitive impairment are multifactorial, ranging from circadian disruption, neurodegeneration, vascular changes, to neuroinflammation.^9,17

Sleep disorders are prevalent among patients with kidney failure undergoing dialysis,¹⁸ with >50% reporting sleep disturbances including insomnia and obstructive sleep apnea (OSA).^19–22 Older patients with kidney failure have 1.31 times the risk of OSA compared with those without kidney failure.²³ Patients with kidney failure who developed insomnia have a 31% higher risk of mortality and 21% higher risk of major adverse cardiac and cerebrovascular events.²⁴ Despite the prevalence of sleep disorders in this population, the extent to which sleep disorders and their treatment are associated with dementia are unclear. Given that older patients with kidney failure are already at an elevated risk of developing cognitive impairments because of complications of kidney failure and dialysis,^25,26 the magnitude of association between sleep disorders and dementia and the effect of treatment cannot be generalized from community-dwelling older adults.

We leveraged national registry data to (1) quantify the dementia risk associated with sleep disorders, (2) determine whether these risks differed by sleep disorder subtype, and (3) test whether positive airway pressure (PAP) therapy mitigated dementia risks for patients with OSA.

Methods

Data Source

The retrospective cohort study used data from the United States Renal Data System (USRDS), which maintains data on all patients with kidney failure and kidney transplant recipients in the United States and integrates data from the United Network for Organ Sharing/Organ Procurement and Transplantation Network, Centers for Medicare & Medicaid Services, and Medicare claims data. The Medical Evidence Report forms were used to obtain demographic data as well as kidney failure etiology and the presence of comorbid conditions. United Network for Organ Sharing/Organ Procurement and Transplantation Network records collected in the USRDS were used to ascertain additional patient characteristics. Medicare billing claims were used to determine recipients' sleep disorder and dementia diagnoses.²⁷

Study Population

We identified 218,608 adults aged 66 years and older who initiated dialysis between January 2008 and December 2019. Patients were included if they (1) had Medicare as their primary payer 1 year before dialysis and at dialysis initiation; (2) were not diagnosed with dementia, sleep disorders, or prescribed PAP therapy within 1 year before dialysis; and (3) had the Medical Evidence Report form filled and entered into the USRDS system within 1 year after the first dialysis service. For the analytical cohort (N=216,158), patients with missing covariates of interest were further excluded, resulting in 2450 patients (1.1%) with missing data on comorbid conditions being dropped (Supplemental Figure 1).

This study underwent review by the institutional review board at the New York University Grossman School of Medicine and was granted exemption under s22-00535.

Sleep Disorders

Sleep disorders were ascertained from Medicare claims by identifying International Classification of Diseases and related Health Problem codes (ICD-9/ICD-10) (Supplemental Table 1) within the first 180 days after dialysis initiation.²⁸ Sleep disorders, defined as a time-fixed exposure, consisted of seven subtypes: insomnia, hypersomnia, parasomnia, SRBDs (including OSA), circadian rhythm sleep disorders, sleep-related movement disorders, and drug-induced sleep disorders. Only patients who had ≥1 inpatient claim or ≥2 outpatient claims were defined as having a sleep disorder.²⁹ Among patients with OSA, we further identified those who were prescribed PAP treatment within the first 180 days from dialysis initiation, on the basis of the presence of Healthcare Common Procedure Coding System codes (Supplemental Table 2), indicating ≥1 prescription fill for PAP equipment.²⁸

Incident Dementia

We identified dementia diagnoses (including subtypes: AD, vascular dementia, and other/mixed dementia) using ICD-9/ICD-10 codes from Medicare claims data (Supplemental Table 3; sensitivity and specificity on dementia were reported to be 85% and 89%, respectively).³⁰ Follow-up for dementia started after the first 180 days after dialysis initiation if there were no claims for sleep disorders. We started follow-up for dementia after the ascertainment of first diagnosis of sleep disorder to ensure temporality and address immortal person–time bias.

Statistical Analyses

We described patient characteristics by sleep disorders. The cumulative incidence of dementia was estimated using the Kaplan–Meier method. We quantified the associations between sleep disorders and dementia using Cox proportional hazards models, with the propensity score weighting method (stabilized weights) to adjust for the imbalanced distribution of patient characteristics. The propensities of having sleep disorders were estimated by logistic regression that accounted for age; sex; race; employment status; Medicaid eligibility; cause of kidney failure; body mass index; dialysis type; pre–kidney failure depression; and comorbid conditions including cerebrovascular disease, peripheral vascular disease, hypertension, diabetes, chronic obstructive pulmonary disease, heart failure, coronary artery disease, and tobacco use; factors were selected on the basis of our conceptual model reflecting the potential mechanisms linking sleep with dementia in kidney failure (Supplemental Figure 2). Sleep disorders were further categorized into four subtypes: (1) no sleep disorder, (2) SRBD, (3) insomnia, and (4) sleep-related movement disorder/other sleep disorder, to study the association between dementia and specific subtype, with a similar statistical approach being used, except that the propensity score model for sleep disorder subtypes was a multinomial logistic regression model. Finally, we performed a subanalysis studying whether using PAP was associated with lower dementia risk among patients diagnosed with OSA. The propensities of using PAP were fitted by a logistic regression model adjusting for the same covariates as discussed above. Risk factors associated with dementia were also identified (Supplemental Table 4).

All patients were followed from the 180 days after dialysis initiation until dementia diagnosis, censoring at the earliest of death, transplant, end of Medicare coverage, or end of the study (December 31, 2019). The proportional hazards assumption was verified by visually inspecting the log–log plots. All analyses were performed using doubly robust estimation in which the covariates included in the propensity score models were again adjusted for in the outcome models. To ensure that all relevant and standard factors were appropriately included in the models, we based our selection on the existing literature on cognitive decline among community-dwelling older adults²⁸ and added factors specific to patients with kidney failure. The Wald test with robust variance was used to evaluate statistical significance.

Sensitivity Analyses

We performed three sensitivity analyses: (1) We limited patients to those who initiated dialysis between 2016 and 2019 to test for any changes in capturing sleep disorder diagnoses after the transition to ICD-10 from ICD-9; (2) we used cause-specific hazards models to estimate dementia risk associated with sleep disorder, treating death as a competing event; and (3) missing indicators were created for each covariate with missing values to accompany the covariate in all analyses.

Results

Characteristics of Older Patients with Kidney Failure

Among 218,608 older patients with kidney failure, the median (interquartile range [IQR]) age at dialysis initiation was 75 (70–80) years, and 57,445 (26.3%) had a first-time sleep disorder diagnosis within the first 180 days after dialysis initiation. Among patients with sleep disorders, 86.3% had SRBD with OSA accounting for 32.8% and 9.5% for insomnia. Sleep disorders were diagnosed in 28.2% of non-Hispanic White, 23.2% of non-Hispanic Black, 22.4% of Hispanic, and 19.8% of Asian patients with kidney failure (Supplemental Table 5). Patients with sleep disorders were more likely to be female (45.7% versus 43.4%); eligible for Medicaid (19.2% versus 18.3%); and have comorbidities, such as obesity (38.8% versus 30.0%) and chronic obstructive pulmonary disease (18.0% versus 13.5%), compared with those without sleep disorders (Table 1).

Table 1.

Characteristics of older (aged ≥66 years) patients with kidney failure by first-time diagnosis of sleep disorder in 2008–2019 (N=218,608) within 180 days after the first kidney failure service

Characteristics	No Sleep Disorder	With Sleep Disorder
	(n=161,163)	(n=57,445)
Age at the first kidney failure service, median (IQR)	75 (70–81)	74 (70–80)
Female sex, %	43.4	45.7
Race/ethnicity, %
Non-Hispanic Asian	4.6	3.2
Non-Hispanic Black	19.2	16.3
Hispanic	9.4	7.6
Non-Hispanic White	64.8	71.4
Other	2	1.4
Employment status, %
Employed	2.7	2.1
Unemployed	12.3	11.9
Retired	85	86
Medicaid coverage, %	18.1	18.9
Cause of kidney failure, %
Diabetes mellitus	43.1	44.2
Hypertension	36	34.3
GN	5.6	5.2
Others	15.3	16.2
BMI kg/m 2 , %
<18.5	3.4	3.1
18.5–24.9	34.4	28.9
25–29.9	32.2	29.2
≥30	30	38.8
Comorbidities, %
Obesity	30	38.8
Cerebrovascular disease	9.4	10.4
Peripheral vascular disease	13.4	15.5
Tobacco use	4	4.3
Alcohol dependence	0.7	0.8
Hypertension	89.2	88.1
Heart failure	31.6	41.8
Coronary artery disease	22.3	25.7
COPD	9.2	16.4
Functional impairmenta	13.5	17.9
Diabetes	53.7	56.7
Depression	3.2	4.5
Dialysis modality at initiation, %
Hemodialysis	90.9	94.2
Peritoneal dialysis	9.1	5.8

BMI, body mass index; COPD, chronic obstructive pulmonary disease; IQR, interquartile range.

^aFunctional impairment captured from the Centers for Medicare & Medicaid Services 2728 form at dialysis initiation: the need for assistance with daily activities and the inability to ambulate.

Sleep Disorders and Dementia Risk

Any Type of Dementia

There were 42,756 older patients (19.8%) with kidney failure who were diagnosed with any type of dementia during a median follow-up time of 1.8 years (IQR, 0.7–3.4). The 5-year cumulative incidence was higher among older patients with sleep disorders (38.2% versus 33.7%, P < 0.001; Figure 1). Older patients with kidney failure who had sleep disorders were at a 22% higher risk of any type of dementia (adjusted hazard ratio [aHR], 1.22; 95% confidence interval [CI], 1.19 to 1.25). Specifically, patients with insomnia (aHR, 1.42; 95% CI, 1.34 to 1.51), SRBDs (aHR, 1.20; 95% CI, 1.17 to 1.23), and sleep-related movement disorders/other sleep disorders (aHR, 1.24; 95% CI, 1.11 to 1.39) were at a higher risk of any type of dementia (Table 2).

Figure 1.

Cumulative incidence of dementia by sleep disorders among older (aged ≥66 years) patients with kidney failure between 2008-2019. Cumulative incidence of (A) any type of dementia, (B) vascular dementia, and (C) other/mixed dementia after 180 days since dialysis initiation by sleep disorder status among older (aged ≥66 years) patients with kidney failure between 2008 and 2019.

Table 2.

Associations between dementia and sleep disorders among older (aged ≥66 years) patients with kidney failure between 2008 and 2019 (N=216,158)

Outcomes	Sleep Disorder and Subtypes	Unadjusted HR (95% CI)	aHRa (95% CI)
Any type of dementia (n=42,756)	No sleep disorder	Reference	Reference
	Any sleep disorder (N=56,790)	1.19 (1.16 to 1.21)	1.22 (1.19 to 1.25)
	No sleep disorder	Reference	Reference
	Insomnia (N=5422)	1.44 (1.36 to 1.53)	1.42 (1.34 to 1.51)
	SRBD (N=48,960)	1.17 (1.14 to 1.19)	1.20 (1.17 to 1.23)
	Sleep related movement disorders/other sleep disorders (N=2408)	1.14 (1.04 to 1.25)	1.24 (1.11 to 1.39)
Alzheimer's Disease (n=5298)	No sleep disorder	Reference	Reference
	Any sleep disorder	1.03 (0.97 to 1.10)	1.07 (0.99 to 1.14)
	No sleep disorder	Reference	Reference
	Insomnia	1.05 (0.87 to 1.26)	0.99 (0.82 to 1.20)
	SRBD	1.03 (0.97 to 1.11)	1.07 (1.00 to 1.15)
	Sleep-related movement disorders/other sleep disorders	0.92 (0.69 to 1.22)	0.99 (0.71 to 1.39)
Vascular dementia (n=4537)	No sleep disorder	Reference	Reference
	Any sleep disorder	1.17 (1.09 to 1.25)	1.17 (1.09 to 1.25)
	No sleep disorder	Reference	Reference
	Insomnia	1.40 (1.17 to 1.68)	1.43 (1.19 to 1.73)
	SRBD	1.15 (1.07 to 1.24)	1.15 (1.07 to 1.24)
	Sleep-related movement disorders/other sleep disorders	0.97 (0.71 to 1.32)	1.04 (0.71 to 1.51)
Other/mixed dementia (n=32,921)	No sleep disorder	Reference	Reference
	Any sleep disorder	1.21 (1.18 to 1.24)	1.23 (1.20 to 1.27)
	No sleep disorder	Reference	Reference
	Insomnia	1.49 (1.39 to 1.59)	1.47 (1.37 to 1.57)
	SRBD	1.18 (1.15 to 1.21)	1.21 (1.18 to 1.24)
	Sleep-related movement disorders/other sleep disorders	1.19 (1.07 to 1.32)	1.29 (1.14 to 1.47)

aHR, adjusted hazard ratio; CI, confidence interval; HR, hazard ratio; SRBD, sleep-related breathing disorder.

^aModels incorporated the propensity score weighting method and adjusted for patient age, sex, race, Medicaid eligibility, employment status, cause of kidney failure, cerebrovascular disease, peripheral vascular disease, hypertension, diabetes, body mass index, chronic obstructive pulmonary disease, heart failure, coronary artery disease, tobacco use, pre–kidney failure depression, and dialysis type (hemodialysis versus peritoneal dialysis).

AD

During a median follow-up time of 2.1 years (IQR, 0.8–3.8), 5298 older patients (2.5%) with kidney failure were diagnosed with AD. The 5-year cumulative incidences of AD were similar comparing patients with sleep disorders with those without (4.6% versus 4.7%, P = 0.12). Only patients with an SRBD were at a higher risk of AD (aHR, 1.07; 95% CI, 1.00 to 1.15) (Table 2).

Vascular Dementia

During a median follow-up time of 2.0 years (IQR, 0.8–3.8), 4537 older patients (2.1%) with kidney failure were diagnosed with vascular dementia. The 5-year cumulative incidence of vascular dementia was higher among older patients with sleep disorders (4.4% versus 3.7%, P < 0.001; Figure 1). Patients with sleep disorders were at 16% higher risk of developing vascular dementia (aHR, 1.17; 95% CI, 1.09 to 1.25). SRBDs (aHR, 1.15; 95% CI, 1.07 to 1.24) and insomnia (aHR, 1.40; 95% CI, 1.17 to 1.68) were associated with higher risk of vascular dementia (Table 2).

Other/Mixed Dementia

During a median follow-up time of 1.9 years (IQR, 0.8–3.6), 32,921 older patients (15.2%) with kidney failure were diagnosed with other/mixed dementia. The 5-year cumulative incidence of other/mixed dementia was higher among older patients with sleep disorders (30.7% versus 26.5%, P < 0.001; Figure 1). Patients with SRBD had a 1.2-fold higher risk of other/mixed dementia (95% CI, 1.18 to 1.24), and the corresponding estimates for those with insomnia and sleep-related movement disorders/other sleep disorders were 1.47 (95% CI, 1.37 to 1.57) and 1.29 (95% CI, 1.14 to 1.47), respectively (Table 2).

PAP Therapy and Dementia Risk

Among 16,091 older patients with kidney failure diagnosed with OSA, 5082 (31.6%) had ≥1 prescription fill of PAP equipment within 180 days after dialysis initiation. The median (IQR) number of refills was 1 (0–2) within 180 days after dialysis initiation. There were 2830 patients who developed dementia over a median follow-up time of 1.8 years (IQR, 0.7–3.3). PAP therapy was associated with an 18% lower risk of any type of dementia (aHR, 0.82; 95% CI, 0.76 to 0.90), 35% lower risk of vascular dementia (aHR, 0.65; 95% CI, 0.50 to 0.85), and 20% lower risk of other/mixed dementia (aHR, 0.80; 95% CI, 0.73 to 0.80) compared with patients with OSA not on the PAP therapy (Table 3). Every additional refill was associated with a 4% lower risk of any type of dementia (aHR, 0.96; 95% CI, 0.94 to 0.98), 8% lower risk of vascular dementia (aHR, 0.92; 95% CI, 0.85 to 0.99), and 5% lower risk of other/mixed dementia (aHR, 0.95; 95% CI, 0.93 to 0.97) (Table 3).

Table 3.

Association between dementia and positive airway pressure therapy among patients with kidney failure who were diagnosed with first-time obstructive sleep apnea in 2008–2019 (n=16,091)

Outcomes	PAP Therapy (n=5082)	Unadjusted HR (95% CI)	aHRa (95% CI)
Any type of dementia (n=2830)	No	Reference	Reference
	Yes	0.80 (0.74 to 0.86)	0.82 (0.76 to 0.90)
	No. of refills	0.95 (0.93 to 0.97)	0.96 (0.94 to 0.98)b
Alzheimer's Disease (n=287)	No	Reference	Reference
	Yes	1.17 (0.92 to 1.49)	1.24 (0.97 to 1.58)
	No. of refills	1.02 (0.96 to 1.08)	1.03 (0.97 to 1.09)
Vascular dementia (n=291)	No	Reference	Reference
	Yes	0.66 (0.50 to 0.85)	0.65 (0.50 to 0.85)
	No. of refills	0.91 (0.85 to 0.98)	0.92 (0.85 to 0.99)
Other/mixed dementia (n=2252)	No	Reference	Reference
	Yes	0.78 (0.71 to 0.85)	0.80 (0.73 to 0.88)
	No. of refills	0.94 (0.92 to 0.97)	0.95 (0.93 to 0.97)

AD, Alzheimer’s Disease; aHR, adjusted hazard ratio; CI, confidence interval; HR, hazard ratio; PAP, positive airway pressure.

^aModels incorporated the propensity score weighting method and adjusted for patient age, sex, race, Medicaid eligibility, employment status, cause of kidney failure, cerebrovascular disease, peripheral vascular disease, hypertension, diabetes, body mass index, chronic obstructive pulmonary disease, heart failure, coronary artery disease, tobacco use, pre–kidney failure depression, and dialysis type.

^bExample of the interpretation of the adjusted hazard ratio (95% confidence interval): Every additional refill of the positive airway pressure equipment is associated with 4% of lower risk of any type of dementia (adjusted hazard ratio, 0.96; 95% confidence interval, 0.94 to 0.98).

Sensitivity Analysis

All three sensitivity analyses yield similar inferences (Supplemental Tables 6 and 7). Sleep disorders were associated with a higher risk of any type of dementia (aHR, 1.34; 95% CI, 1.26 to 1.43) and other/mixed dementia (aHR, 1.37; 95% CI, 1.28 to 1.47) after the transition to ICD-10.

Discussion

In this national study of 218,608 older patients with kidney failure, 26% of patients received a first-time diagnosis of a sleep disorder within 180 days of initiating dialysis, and they had significantly greater dementia risk: 22% greater risk of any type of dementia, 17% greater risk of vascular dementia, and 23% greater risk of other/mixed dementia. Older patients with kidney failure diagnosed with SRBD had an greater risk of developing AD, vascular dementia, and other/mixed dementia. For patients with OSA, PAP therapy use was associated with an 18% lower risk of any type of dementia, 35% lower risk of vascular dementia, 20% lower risk of other/mixed dementia.

Most of the older patients with kidney failure with a sleep disorder in our study were found to have SRBD (86.3%), of whom 32.8% had OSA. Our findings are consistent with SRBD prevalence in patients with CKD reported in the literature ranging from 50% to 85%.^31,32 However, the number of patients in our study diagnosed with insomnia and OSA was lower than the prevalence of OSA and insomnia among patients with kidney failure reported in the literature: 48%–56% for OSA^33,34 and 37%–80% for insomnia.^35,36 Sleep disorders were diagnosed in 28.2% of non-Hispanic White individuals and 23.2% of non-Hispanic Black individuals with kidney failure. Several studies and meta-analyses have reported on disparities in sleep disorders and disrupted sleep and the challenges in estimating prevalence because of underdiagnoses, differences in symptoms, and external environmental factors.^37–41 Our findings on the racial/ethnic differences in sleep disorders could be explained by the fact that all patients in our study have kidney failure, which may contribute to the similar burden of sleep disorders between racial/ethnic groups. It is also possible that our findings on the burden of these disorders, in part, reflect difficulties in recognizing and diagnosing sleep disorders among non-Hispanic Black patients captured by ICD-9/10 codes.⁴²

Our central findings that older patients with kidney failure who are diagnosed with sleep disorders are at an greater dementia risk are consistent with the literature. Systematic reviews have shown that individuals in the general population with sleep disorders have a 55% greater risk of developing AD¹³ and those with insomnia have a 53% greater risk of developing dementia.¹² A study from the National Health and Aging Trends Study in older adults showed a 30% higher risk of dementia in older adults who had insomnia.¹⁵ Our study also observed that SRBD was associated with a significantly greater risk of all types of dementia, including AD. This is consistent with a prospective study that showed a higher odds of developing dementia and cognitive impairment associated with SRBD in community-dwelling older women.^9,16 The magnitude of our study's associations may be lower than that observed in the general population because of the challenges of dementia diagnosis in patients with kidney failure because of differing or misattribution of symptoms.²

Sleep disorders and their link with dementia and AD are multifactorial, and the hypothesized mechanism varies by subtype and underlying pathology (Figure 2 for a detailed overview). Both insomnia and sleep-related movement disorders are believed to be associated with an greater dementia risk because of disrupted and poor-quality sleep.^10,43 Sleep disturbance can interfere with normal neuronal functioning, can impair synaptic plasticity, and may contribute to neurodegeneration.⁹ In AD, sleep deprivation may alter amyloid-β accumulation and clearance.¹⁰ Insomnia may affect sleep cycle phases leading to impaired memory formation, decreased acetylcholine levels, and altered amyloid-β accumulation and clearance.⁴⁴ SRBDs are unique in that their association with dementia is believed to be because of hypoxia caused by abnormal respiratory patterns during sleep.¹⁶ OSA characterized by episodic obstruction of the upper airway because of anatomic narrowing leads to intermittent hypoxia and fragmented sleep.^10,45 This chronic intermittent hypoxia may lead to cerebral hypoperfusion, endothelial dysfunction, neuroinflammation,⁴⁶ and cerebral amyloid deposition,⁴⁷ which ultimately contributes to cognitive impairment and dementia.

We found that PAP use in those with an OSA diagnosis was associated with lower risk of dementia, which is consistent with findings from a retrospective study drawing on claims data that demonstrated 22% lower odds of incident diagnoses of AD and 31% lower odds of dementia in older community-dwelling adults with OSA on PAP therapy.²⁸ Patients with kidney failure are already at higher risk of dementia because of vascular risk factors, such as anemia, oxidative stress, chronic inflammation, and uremic toxins.²⁵ These risks are likely worsened by the effect of OSA on cerebrovascular and neurodegenerative pathways, such as cerebral small vessel disease.^46,48 PAP therapy may alleviate the cognitive effects of OSA by reducing inflammation and increasing cerebral blood flow.⁴⁹ Although our results showed that PAP therapy was associated with lower risk of any type of dementia, vascular dementia, and other/mixed dementia risk among those with OSA, it was not significantly associated with lower risk of AD. This may partly be attributed to the differences in pathophysiology between vascular dementia and AD⁵⁰ and how they are affected differently by the vascular effects of SRBD and OSA.⁵¹

A strength of our study includes the generalizability of our findings given that utilization of USRDS national registry data allows us to look at dementia risk in patients with kidney failure in the United States. Some limitations to our study should be noted. The use of ICD codes is limited by the accuracy and completeness of coding of diagnoses and therapies; if the misclassification is differential, it may lead to bias and attenuation of the results. It is more likely that we are underestimating the prevalence of sleep disorders³³ and dementia (which can only be definitively confirmed postmortem) when using claims data, instead of direct clinical evaluation and diagnosis.⁵² Given that dementia⁵³ and sleep disorders⁵⁴ are underdiagnosed and under-reported in the general population, they may be even more challenging to recognize in patients with kidney failure.⁵⁵ Defining the study cohort on the basis of Medicare insurance coverage may also induce selection bias. Owing to data limitations on the duration of PAP therapy use and compliance, our analysis could not estimate a dose response to PAP therapy on dementia risk reduction. We assume any use of PAP therapy once the prescription has been filled, which could greatly underestimate the use of PAP therapy. Given this assumption, it remains even more encouraging that the use of PAP therapy in patients with OSA is associated with significantly lower dementia risk, particularly vascular dementia. Finally, we lacked granular data on individual activity, such as physical activity and activities of daily living. Thus, we were not able to assess the association between physical activity, sleep, and dementia.

In conclusion, older patients with kidney failure with sleep disorders are at higher risk of being diagnosed with any type of dementia, specifically vascular dementia and other/mixed types of dementia. As the prevalence and effect of dementia and AD continue to grow, the need for effective interventions to prevent or slow their progression becomes more critical, given their association with greater mortality and disability.^3,5,6 PAP therapy is one possible intervention to decrease dementia risks and preserve cognitive function,⁵⁶ in addition to its benefits of improving cardiovascular outcomes and quality of life.^17,57 Sleep disorders should be taken into consideration by providers when counseling on dementia risks and providing comprehensive management to limit cognitive decline.²⁵ Sleep is an important modifiable factor that can be targeted for interventions to mitigate dementia risk. For older patients with kidney failure who are diagnosed with OSA, PAP therapy is an intervention that is associated with lower dementia risk.

Supplementary Material

cjasn-19-1301-s001.pdf ^{(1.4MB, pdf)}

cjasn-19-1301-s002.pdf ^{(443.5KB, pdf)}

Disclosures

Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/CJN/B950.

Funding

This work was supported by grant numbers K02AG076883 (PI: M.A. McAdams-DeMarco), R01AG055781 (PI: M.A. McAdams-DeMarco), R01AG077888 (PI: M.A. McAdams-DeMarco), F32AG082486 (PI: J.J. Long), and R01AG076834 (PI: A. Mathur) from the National Institute on Aging (NIA); R01DK114074 (PI: M.A. McAdams-DeMarco) and R01DK120518 (PI: M.A. McAdams-DeMarco) from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK); and grant number K24AI144954 (PI: D.L. Segev) from the National Institute of Allergy and Infectious Diseases (NIAID).

Author Contributions

Conceptualization: Mara A. McAdams-DeMarco.

Data curation: Yusi Chen, Mara A. McAdams-DeMarco.

Formal analysis: Yusi Chen, Mara A. McAdams-DeMarco.

Funding acquisition: Mara A. McAdams-DeMarco, Dorry L. Segev.

Investigation: Mara A. McAdams-DeMarco.

Methodology: Mara A. McAdams-DeMarco.

Supervision: Mara A. McAdams-DeMarco, Dorry L. Segev.

Writing – original draft: Sunjae Bae, Yusi Chen, Nadia M. Chu, Byoungjun Kim, Yiting Li, Jane J. Long, Aarti Mathur, Mara A. McAdams-DeMarco, Babak J. Orandi, Dorry L. Segev.

Writing – review & editing: Sunjae Bae, Yusi Chen, Nadia M. Chu, Byoungjun Kim, Yiting Li, Jane J. Long, Aarti Mathur, Mara A. McAdams-DeMarco, Babak J. Orandi, Dorry L. Segev.

Data Sharing Statement

Data cannot be shared. The datasets used and/or analyzed during the current study are available from the United States Renal Data System. Per the Data Use Agreement (DUA) between the authors and the United States Renal Data System, the deposition of data into publicly available repositories is not allowed.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/CJN/B949.

Supplemental Table 1. ICD-9 and ICD-10 codes for diagnosed sleep disorders using claims data.

Supplemental Table 2. HCPCS codes for positive airway pressure (PAP) therapy using claims data.

Supplemental Table 3. ICD-9 and ICD-10 codes for diagnosed dementia using claims data.

Supplemental Table 4. Factors associated with dementia.

Supplemental Table 5. Sleep disorders among older (aged ≥66 years) patients with kidney failure between 2008 and 2019 (N=216,158), by race/ethnicity.

Supplemental Table 6. Associations between dementia and sleep disorders among older (aged ≥66 years) patients with kidney failure between 2008 and 2019 (N=216,158). Cause-specific hazards models were used.

Supplemental Table 7. Associations between dementia and sleep disorders among older (aged ≥66 years) patients with kidney failure between 2008 and 2019 (N=218,608). Missing indicator method was used.

Supplemental Figure 1. Cohort derivation.

Supplemental Figure 2. Potential mechanisms of associations between sleep disorders and dementia.