Periodic leg movements and cognitive function in middle-aged and elderly adults with mediating role of REM duration and PLMS-related arousal

Abstract

The relationship between periodic leg movements of sleep (PLMS) and cognitive function is controversial. This study aims to explore: (1) the effects of PLMS on cognitive function in middle-aged and elderly patients; and (2) the mediating role of rapid eye movement (REM) sleep duration and PLMS-related arousal between PLMS and cognitive performance. This retrospective study was conducted based on inpatient medical records. Middle-aged and elderly patients hospitalized in the Department of Neurology at the Affiliated Hospital of Yangzhou University from September 2023 to August 2024 were selected. Participants were those who underwent polysomnography (PSG) due to sleep-related complaints, such as daytime sleepiness, nocturnal snoring, insomnia, or frequent nighttime awakenings. The Montreal Cognitive Assessment (MoCA) was used to evaluate the cognitive function of patients. Participants were stratified into two groups based on MoCA scores (MoCA ≥ 26 and MoCA < 26) to investigate the differential effects of PLMS on cognitive function across distinct cognitive profiles. Multivariate linear regression models were used to evaluate the association between PLMS (periodic limb movement index, PLMI) and cognitive performance, adjusted for covariates including age, gender, education, smoking, total sleep time, and apnea–hypopnea index. Mediation analyses were performed to test whether REM sleep duration and PLMS-related arousal index (PLMAI) mediated the PLMS-cognition relationship. A total of 818 patients were included. Regression analysis showed that PLMI (β = − 0.033, 95% CI − 0.039 to − 0.026, p < 0.001) was negatively correlated with cognitive function. After adjusting for REM sleep duration and PLMAI, PLMI was still significantly associated with cognitive function (β = − 0.026, 95% CI − 0.034 to − 0.018, p < 0.001). PLMI was negatively correlated with REM sleep duration (β = − 0.157, 95% CI − 0.258 to − 0.056, p = 0.002) and positively correlated with PLMS-related arousal (β = 0.036, 95% CI 0.029–0.043, p < 0.001). The results of mediation analysis showed that REM sleep duration and PLMS-related arousal partially mediated the association between PLMS and cognitive function (indirect effect estimate = − 0.003, − 0.007; direct effect estimate = − 0.033). PLMS is associated with poorer cognitive function in middle-aged and elderly people, and REM sleep duration and PLMS-related arousal play a partial mediating role. Clinicians can utilize periodic limb movements during sleep (PLMS) as a potential early warning indicator of cognitive decline, enabling timely interventions to slow its progression. Additionally, improving REM sleep duration and suppressing PLMS-related arousals through pharmacological or non-pharmacological treatments may serve as critical strategies for preserving cognitive function.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-04660-7.

Introduction

Cognitive impairment has a high incidence in the elderly. By 2021, the number of people with cognitive impairment worldwide had reached 56.9 million, which poses a serious threat to the health of the elderly and has become a public health problem that cannot be ignored¹. Therefore, it is important to explore the factors and potential mechanisms that affect cognitive function. Periodic leg movements of sleep (PLMS) is a sleep-related unconscious phenomenon characterized by periodic, repetitive, stereotypical limb movements, often manifested as toe and ankle dorsiflexion, and occasionally hip and knee flexion². PLMS is usually found during polysomnography, and patients often have no complaints³. However, studies have shown that PLMS is related to a variety of diseases, such as cardiovascular disease⁴, sleep and dysfunxsction⁵, neurodegenerative diseases,⁶ and even increased mortality in patients with heart failure and end-stage renal disease^7,8. The pathophysiological mechanism needs to be further studied.

There have been few and controversial studies on the relationship between PLMS and cognitive function, Yue Leng et al. studied 2636 community-dwelling elderly men aged 65 years or older without dementia and showed that a higher frequency of PLMS was associated with greater cognitive decline, especially in executive function⁹. However, additional cross-sectional studies have shown no association between PLMS severity and cognitive function¹⁰. This inconsistency creates a pressing clinical dilemma: if PLMS is indeed a modifiable risk factor for cognitive impairment, current diagnostic and therapeutic protocols may fail to address a preventable contributor to neurodegeneration in aging populations. PLMS is associated with decreased REM sleep duration and increased sleep fragmentation¹¹. Reduced REM sleep duration and increased sleep fragmentation is associated with worse cognitive function^12,13, but no study has systematically tested whether these sleep disruptions mediate the PLMS-cognition relationship. Without such evidence, clinicians cannot prioritize interventions targeting REM continuity or arousal reduction in PLMS patients with cognitive complaints. Therefore, we aimed to (1) examine the correlation between PLMS and cognitive function in middle-aged and older Chinese patients; and (2) explore the mediating effect of REM sleep duration and sleep arousal on the relationship between PLMS and cognitive ability, to provide help for clinical practice. To address the potential heterogeneity in cognitive vulnerability, we categorized participants using the MoCA cutoff of 26, a validated threshold distinguishing normal cognition (MoCA ≥ 26) from cognitive impairment (MoCA < 26)¹⁴. This stratification allows for a targeted analysis of how PLMS impacts individuals with preserved versus compromised cognitive function.

Methods

Subjects

This was a hospital medical records-based retrospective study. Middle-aged and elderly patients hospitalized in the Department of Neurology at the Affiliated Hospital of Yangzhou University from September 2023 to August 2024 were selected. Participants were recruited based on self-reported sleep-related complaints, including but not limited to: Daytime sleepiness, nocturnal snoring, insomnia symptoms, frequent nocturnal awakenings, discomfort in the legs, etc. A list of symptoms and their frequencies is provided in Supplementary Table S1. PSG is used to confirm sleep disorders. Final diagnoses were established by board-certified sleep physicians according to ICSD-3¹⁵, integrating PSG data, clinical interviews, and exclusion of confounding conditions. Inclusion criteria: (1) middle-aged and elderly people of both sexes aged 45–74 years; (2) Complete polysomnography (PSG), sleep-related scales, and Montreal Cognitive Assessment Scale (MoCA); (3) Complete general clinical data. Exclusion criteria: (1) existing cognitive impairment caused by related diseases, such as Alzheimer’s disease, frontotemporal dementia, malignant tumor, epilepsy, alcoholic encephalopathy, psychosis, metabolic encephalopathy, hypothyroidism, etc.; (2) Taking or taking cognition-related medications (e.g., cholinesterase inhibitors, ergonovines, neurotrophins, etc.) or medications that affect the sleep–wake cycle or prolong REM sleep (e.g., antidepressants, benzodiazepines, or antipsychotics, etc.) in the last 6 months; (3) patients with Restless leg syndrome (RLS), periodic limb movement disorder (PLMD) and Rapid eye movement sleep behavior disorder (RBD). PLMD was defined as PLMS index (PLMI) ≥ 15 events/h with clinical symptoms¹⁵. Patients with REM sleep behavior disorder (RBD) were excluded because they are closely related to α-synucleinopathies (such as Parkinson’s disease) and have different pathophysiological mechanisms from PLMS^16,17, which may confound PLMS-specific effects; (4) complicated with severe heart, liver, kidney, lung and other organ failure diseases; (5) patients treated with continuous positive airway pressure ventilation (CPAP). Participants with obstructive sleep apnea (OSA) were included to reflect the real-world prevalence of comorbid sleep disorders in middle-aged and elderly populations, ensuring external validity. To minimize confounding effects, patients undergoing CPAP therapy were excluded, as CPAP alters sleep architecture and may obscure the natural association between PLMS and cognition.

The study was approved by the Ethics Committee of Yangzhou University (Ethics 2023-YKL09), and all patients provided written informed consent.

Research methods

General data characteristics

The general data of the patients were collected, including age, gender, years of education, body mass index (BMI), smoking history, drinking history, and past medical history (hypertension, diabetes, cerebral infarction, coronary heart disease, etc.).

Scale assessment and outcome measures

ESS

The Epworth Sleepiness Scale assesses daytime sleepiness through self-reported likelihood of dozing in eight daily situations. Each item is rated from 0 (no chance of dozing) to 3 (high chance). Total scores range from 0 to 24, with higher scores indicating greater daytime sleepiness (≥ 10 suggests pathological sleepiness)¹⁸.

PSQI

The Pittsburgh Sleep Quality Index (PSQI) scale was used to assess subjective sleep quality over the past month. It included 19 items divided into 7 components (e.g., sleep latency, sleep duration). Each component is scored 0–3, with a global score range of 0–21. Higher scores indicate poorer sleep quality (≥ 5 suggests significant sleep impairment)¹⁹.

HADS

The Hospital Anxiety and Depression Scale (HADS) was used to screen for anxiety (HADS-A) and depression (HADS-D) symptoms in the past week. Each subscale contains 7 items rated from 0 (no symptoms) to 3 (severe symptoms). Subscale scores range from 0 to 21, with scores ≥ 8 indicating clinically significant anxiety/depression²⁰.

MoCA

The Montreal Cognitive Assessment (MoCA) was used to evaluate the cognitive function of the patients. Covering the following seven cognitive subdomains: (1) Visuospatial and executive function: including tasks such as cube copying and clock drawing; (2) Naming: testing language naming ability through the recognition of images of common objects (e.g., lion, rhinoceros); (3) Attention: including digit span tests (forward and backward), digit "1" clicking tasks, and calculation ability; (4) Language: measuring language expression ability through the repetition of complex sentences and vocabulary fluency tasks (e.g., listing animal names within one minute); (5) Abstract thinking: requiring explanations of word similarities; (6) Delayed recall: after completing other tasks, asking subjects to recall five previously presented words to assess short-term memory ability; (7) Orientation: testing orientation to time (date, season) and place (city, hospital name).The total score was 30 points, and usually ≥ 26 was defined as normal cognitive function. An additional 1 point was added for participants with ≤ 12 years of education to adjust for educational bias²¹.

Cognitive function was assessed with MoCA scale scores as continuous variables to capture subtle cognitive changes. Although a MoCA score < 26 generally indicates mild cognitive impairment (MCI), the study did not use formal MCI diagnostic criteria, which avoided diagnostic misclassification while improving sensitivity to subtle associations.

These tools were selected for their validated sensitivity to sleep disturbances in aging populations, and their established relevance to cognitive outcomes, enabling control of confounding effects from subjective sleep complaints and mood disorders.

Polysomnography

The included patients had been monitored by SOMNOmedics V6 polysomnography equipment. Signal acquisition encompasses electroencephalogram (F3-O2 derivation), electrooculogram, electromyogram, and electrocardiogram. Respiratory event data are obtained via nasal airflow monitoring, thermistor-based airflow detection, snore microphone recording, thoracoabdominal respiratory effort belts, and pulse oximetry. Participants maintained habitual sleep schedules. Bedtime and wake time were self-reported and synchronized with PSG recording periods.

The following data were collected: total sleep time (TST), sleep efficiency (SE), Sleep latency, wake time after sleep onset (WASO), proportion and duration of sleep stages (N1, N2, N3, REM), apnoea‒hypopnea index (AHI), arousal index (Arl), PLMS-related arousal index (PLMAI), periodic limb movement index (PLMI).

According to the American Academy of Sleep Medicine Sleep Staging and Related Events Interpretation Rule, version 2.6²², sleep staging and related events were analyzed manually by three professional polysomnography technicians, and the decision was made by the three technicians after the different results of the interpretations were agreed upon. Apnea was defined as ≥ 90% airflow reduction for ≥ 10 s. Hypopnea was defined as ≥ 30% airflow reduction with ≥ 3% SpO₂ desaturation or arousal. AHI was calculated as events/hour. Periodic limb movement index (PLMI) was defined as the number of periodic leg movements per hour, with each movement lasting 0.5–1.0 s and occurring at intervals of 5.0–90.0 s. The occurrence of ≥ 4 was considered as the occurrence of a single periodic limb movement, and ≥ 5/h was considered abnormal. The periodic leg movement arousal index of sleep (PLMAI) was the number of awakenings or microarousals per hour due to PLMS.

In this study, the AASM rule was used for PLMS assessment, which was chosen for its wide clinical applicability²². However, it should be noted that previous studies have shown that the AASM rules may be sensitive to non-periodic leg movements, which may overestimate PLMS to a certain extent²³.

Statistical methods

SPSS 26.0 (IBM Corp., Armonk, USA) software was used to process the data. All data first underwent normality and homogeneity of variance tests. Normally distributed numerical data are expressed as mean ± standard deviation (SD), while skewed numerical data are presented as median and interquartile range (IQR). Independent samples t-test and ANOVA were employed for analyzing baseline patient characteristics. According to the MoCA score, the patients were divided into two groups (MoCA ≥ 26 group and MoCA < 26 group). Independent sample t-test or rank sum test was used to compare the scale and PSG results of the two groups. Four multivariate linear regression models were used to assess the direct effects of sleep periodic limb movements and REM sleep duration, PLMS arousal index on cognitive function. Model 1 was a multiple linear regression analysis between MoCA scale scores and PLMI. Model 2 was a multiple linear regression analysis between MoCA scale scores and PLMI after adjustment for REM sleep duration and PLMS arousal index. Models 3 and 4 were multiple linear regression analyses of the relationship between REM sleep duration, PLMS arousal index, and PLMI, respectively. All four models were adjusted for relevant covariates, including gender, age, education, smoking, total sleep time and AHI. To explore age-related differences and differences according to diagnostic criteria, we performed main regression analyses within each subgroup. Bonferroni correction was used for baseline characteristics (e.g., sex and education), main regression analyses (models 1–4 in Table 3), and subgroup analyses stratified by age and diagnostic criteria. In order to explore the internal mechanism of the influence of PLMS on cognitive function, the mediating effect was tested by using Model 4 in PROCESS(version 4.1)²⁴, a plug-in compiled by Hayes in SPSS, and the test level was α = 0.05. All statistical tests were two-sided and showed odds ratio (OR) with 95% confidence interval (CI), p < 0.05 for statistical significance.

Table 3.

Regression analysis.

	Model 1: MoCA			Model 2: MoCA			Model 3: REM sleep duration			Model 4: PLMAI
	β	p	95% CI	β	p	95% CI	β	p	95% CI	β	p	95% CI
REM sleep duration	–	–	–	0.006	0.021	(0.001, 0.012)	–	–	–	0.005	0.008	(− 0.001, 0.010)
PLMAI	–	–	–	− 0.165	< 0.001	(− 0.241, − 0.090)	0.869	0.080	(− 0.103, 1.841)	–	–	–
PLMI	− 0.033	< 0.001	(− 0.039, − 0.026)	− 0.026	< 0.001	(− 0.034, − 0.018)	− 0.157	0.002	(− 0.258, − 0.056)	0.036	< 0.001	(0.029, 0.043)
Gender	0.715	0.006	(0.205, 1.226)	0.845	0.002	(0.319, 1.370)	4.999	0.149	(− 1.791, 11.788)	0.369	0.144	(− 0.126, 0.865)
Age	− 0.230	< 0.001	(− 0.281, − 0.179)	− 0.232	< 0.001	(− 0.284, − 0.181)	0.021	0.951	(− 0.647, 0.689)	− 0.002	0.927	(− 0.051, 0.046)
Education	0.710	0.009	(0.180, 1.239)	0.793	0.004	(0.257, 1.329)	− 1.653	0.640	(− 8.591, 5.285)	0.237	0.358	(− 0.269, 0.743)
Smoking	− 0.170	0.528	(− 0.699, 0.358)	− 0.207	0.451	(− 0.746, 0.332)	− 6.412	0.071	(− 13.369, 0.544)	− 0.142	0.584	(− 0.367, 0.651)
TST	0.001	0.530	(− 0.001, 0.003)	− 0.001	0.547	(− 0.003, 0.002)	0.208	< 0.001	(0.182, 0.234)	− 0.002	0.127	(− 0.004, 0.000)
AHI	− 0.013	0.033	(− 0.024, − 0.001)	− 0.010	0.085	(− 0.022, 0.001)	− 0.190	0.016	(− 0.343, − 0.036)	0.008	0.148	(− 0.003, 0.020)
R2	0.284			0.296			0.287			0.147
Adjusted R2	0.277			0.287			0.280			0.138

Model 1: multiple linear regression analysis between PLMS and cognitive score; Model 2: multiple linear regression analysis between PLMS and cognitive score adjusted by PLMAI; Model 3: multiple linear regression analysis between PLMS and REM duration; Model 4: multiple linear regression analysis between PLMS and PLMAI; Adjusted by gender, age, education, smoking total sleep time and AHI.

Based on the anticipated effect size (β = − 0.032) and the number of covariates, a power analysis was performed using G*Power 3.1. The results indicated that a minimum of 720 samples were required to achieve 80% statistical power (α = 0.05). This study included 818 patients, meeting the sample size requirement. For the mediation analysis, the Bootstrap method (5000 resamples) was employed to validate the stability of the indirect effects.

Results

Comparison of cognitive function at baseline data

A total of 893 patients were collected, and 22 patients with RLS, 17 patients with PLMD, and 36 patients with RBD were excluded. Finally, 818 patients who met the inclusion and exclusion criteria were enrolled, including 497 males (60.8%) and 321 females (39.2%). Patients with younger age, male gender, longer education, and no smoking habit had better cognitive function (p < 0.001; p = 0.005; p < 0.001; p = 0.040, Table 1). There was no significant difference in cognitive function among patients with different BMI, alcohol consumption, history of hypertension, diabetes, coronary heart disease, and cerebral infarction (all p > 0.05).

Table 1.

Baseline demographic and cognitive characteristics.

Characteristics	Frequency	Proportion (%)	MoCA Score( ‾x ± s)	t/F	p	Effect size
Age
45–55	210	25.7	25.36 ± 3.44	94.49	< 0.001	η2 = 0.188
55–65	286	34.4	22.52 ± 3.07
65–74	322	38.7	21.06 ± 3.96
Gender
Male	497	60.8	22.17 ± 4.34	− 2.81	0.005	Cohen’s d = 0.52
Female	321	39.2	19.39 ± 6.69
BMI
< 24	286	35.0	22.80 ± 3.92	0.674	0.500
≥ 24	532	65.0	22.60 ± 3.93
Education
Primary school and below	312	38.8	21.34 ± 3.93	51.32	< 0.001	η2 = 0.099
Junior high school	306	36.8	22.69 ± 3.21
High school and above	200	24.4	24.73 ± 4.02
Smoking
No	543	66.4	22.87 ± 3.83	2.05	0.040	Cohen’s d = 0.15
Yes	275	33.6	22.28 ± 4.06
Smoking
No	601	73.5	22.64 ± 3.95	− 0.37	0.711
Yes	217	26.5	22.76 ± 3.85
Hypertension
No	383	46.8	22.60 ± 4.11	− 0.48	0.631
Yes	435	53.2	22.73 ± 3.74
Diabetes
No	564	68.9	22.61 ± 3.90	− 0.65	0.518
Yes	254	31.1	22.80 ± 3.97
Cerebral infarction
No	544	66.5	22.63 ± 3.85	− 0.38	0.704
Yes	274	33.5	22.74 ± 4.06
Coronary heart disease
No	713	87.2	22.73 ± 3.90	1.13	0.258
Yes	105	12.8	22.27 ± 4.04

BMI, Body mass index. Older age, lower education, and female gender were associated with poorer cognitive scores, highlighting the need for targeted cognitive monitoring in these higher-risk subgroups.

Comparison of scale and PSG results

As shown in Table 2, patients with MoCA score ≥ 26 had longer total sleep time, longer proportion and duration of REM sleep, less proportion of N1 sleep, and lower PLMS arousal index and PLMI (p = 0.013; p = 0.011; p = 0.001; p = 0.035; p < 0.001; p < 0.001). There were no significant differences in ESS, HADS, PSQI scale scores, SE, WASO, other sleep proportion and duration, AHI, and arousal index between the two groups (all p > 0.05).

Table 2.

Comparison of scale scores and PSG results.

	MoCA ≥ 26 (n = 283)	MoCA < 26 (n = 535)	t/z	p	Effect size
ESS	6 (3,12)	7 (3, 12)	− 0.60	0.548
HADS(A)	1 (0,4)	1 (0, 4)	− 0.54	0.590
HADS(D)	1 (0, 4)	1 (0, 4)	− 0.59	0.554
PSQI	9.08 ± 4.35	9.04 ± 4.45	0.11	0.911
TST (min)	400.81 ± 117.08	378.43 ± 123.77	2.50	0.013	Cohen’s d = 0.18
S-efficiency (%)	78.90 ± 16.29	76.53 ± 24.00	1.49	0.138
WASO (min)	60.20 (33.95, 127.50)	73.00 (38.25, 138.55)	− 1.92	0.054
Sleep latency (min)	7.70 (3.70, 15.85)	8.50 (3.30, 17.50)	− 0.55	0.584
REM (%TST)	18.10 (13.00, 25.93)	17.20 (10.40, 24.10)	− 2.55	0.011	Cliff’s delta = 0.13
N1% (%TST)	4.80 (2.40, 10.63)	5.50 (2.80, 11.53)	− 2.11	0.035
N2% (%TST)	58.70 (47.48, 67.60)	59.8 (48.10, 69.10)	− 1.34	0.182
N3% (%TST)	12.55 (5.68, 19.83)	11.50 (3.80, 21.00)	− 1.00	0.317
REM duration (min)	74.80 (48.00, 106.25)	64.00 (34.00, 97.50)	− 3.26	0.001	Cliff’s delta = 0.15
N1 duration (min)	20.00 (9.65, 38.00)	19.50 (10.00, 38.50)	− 0.86	0.393
N2 duration (min)	230.65 ± 88.69	220.55 ± 94.38	1.48	0.131
N3 duration (min)	46.00 (20.38, 81.00)	40.50 (13.80, 80.00)	− 1.35	0.177
AHI	15.40 (6.20, 38.35)	19.30 (9.00, 35.50)	− 1.55	0.122
Arl	29.05 (17.17, 40.00)	28.20 (17.65, 37.90)	− 0.79	0.428
PLMAI	0.60 (0.20,1.90)	2.20 (0.68, 5.00)	− 8.89	< 0.001	Cliff’s delta = 0.39
PLMI	3.45 (1.1,8.53)	18.25 (7.15, 38.58)	− 12.08	< 0.001	Cliff’s delta = 0.52

ESS: Epworth Sleepiness Scale; PSQI: Pittsburgh Sleep Quality Index scale; HADS-A: Hospital Anxiety and Depression Scale—Anxiety subscale; HADS-D: Hospital Anxiety and Depression Scale—Depression subscale; TST, Total sleep slips; WASO, wake time after sleep onset; REM, rapid eye movement; AHI, apnoea‒hypopnea index; Arl, arousal index; PLMAI, PLMS-related arousal index; PLMI, periodic limb movement index. Older age, lower education, and female gender were associated with poorer cognitive scores, highlighting the need for targeted cognitive monitoring in these higher-risk subgroups.

Regression analysis

To exclude confounding factors, gender, age, years of education, smoking, total sleep time and AHI were used as covariates for regression analysis, as shown in Table 3. The results of Model 1 showed increased PLMI was significantly associated with decreased MoCA score (β = − 0.033, 95% CI − 0.039 to − 0.026, p < 0.001). To address potential confounding by low baseline cognition, we repeated the primary analysis after excluding participants with MoCA scores < 26 (n = 535). The association between PLMI and cognitive function remained significant (β =  − 0.026, 95% CI − 0.031 to − 0.020, p < 0.001). After adjusting for REM sleep duration and PLMAI in Model 2, PLMI (β = − 0.026, 95% CI − 0.034 to − 0.018, p < 0.001) was still significantly associated with cognitive function. Model 3 showed a negative association between PLMI (β = − 0.157, 95% CI − 0.258 to − 0.056, p = 0.002) and REM sleep duration. Model 4 showed a positive correlation between PLMI (β = 0.036, 95% CI 0.029–0.043, p < 0.001) and PLMAI.

Stratified analysis of disease diagnosis showed that PLMI was significantly correlated with MoCA scores in different diseases. (Supplementary Table S2) Age-stratified analyses demonstrated a stronger association between PLMI and cognitive decline in the 65–74 years old group (β = − 0.035, 95% CI − 0.045 to − 0.002, *p* < 0.001) compared to the other two groups (Supplementary Table S3).

Mediation effect test

The path coefficients of REM sleep duration and PLMAI between PLMS and cognitive function are shown in Fig. 1. Table 4 shows that the total effect value of PLMS and cognitive function was c = − 0.041, and the direct effect value was c′ = − 0.033. The mediating effect value of REM sleep duration and PLMAI was β1 = − 0.002 and β2 = − 0.007, accounting for 4.1% and 15.8% of the total effect, respectively. The lower and upper limits of the bootstrap 95% confidence intervals for effect sizes did not include 0. Both direct and mediating effects were statistically significant, indicating that PLMS not only had a direct effect on cognitive function but also had a mediating effect on cognitive function by reducing REM sleep duration and increasing PLMS-related arousal (increased PLMAI). Clinically, these percentages (4.1% and 15.8%) suggest that even modest improvements in REM continuity and reductions in PLMS-related arousal could mitigate approximately 20% of PLMS-driven cognitive decline—a meaningful target for intervention in vulnerable patients.

Fig. 1.

Mediation model and regression coefficients of REM sleep duration and PLMAI between PLMS and cognitive function. Note: *p < 0.05, **p < 0.01, ***p < 0.001; a represents the effect size of PLMS on REM sleep duration, b represents the effect size of REM sleep duration on cognitive function, a' represents the effect size of PLMS on PLMS-related arousal, b' represents the effect size of PLMS-related arousal on cognitive function, c represents the total effect size of PLMS on cognitive function, c' represents the direct effect size of PLMS on cognitive function.

Table 4.

Mediation effect test.

		Effect value	SE	LLCI	ULI	Proportion(%)
Total effect value	–	− 0.041	0.004	− 0.048	− 0.033
Direct effect value	–	− 0.033	0.004	− 0.041	− 0.025
Indirect effect value	REM sleep duration	− 0.002	0.001	− 0003	− 0.001	0.041
	PLMAI	− 0.007	0.002	− 0.011	− 0.003	0.158

Discussion

The main objectives of this study were to analyze the potential association between PLMS and cognitive function in middle and old age and to explore the mediating role of REM sleep duration and PLMS-related arousal in the relationship between PLMS and cognitive function. Our results showed that heavier PLMS was significantly associated with worse cognitive function. Moreover, it leads to poor cognitive function in middle and old age by reducing REM sleep duration and increasing PLMS-related arousal.

Our research indicates that pronounced PLMS correlates with diminished cognitive function. This result is similar to that of a prospective study with a large sample size⁹. The study, which looked at 2636 older men without dementia over three visits, found that heavier PLMS was associated with more substantial cognitive decline, particularly in executive function. We focused on the overall cognitive function of patients and did not conduct subdomain studies. The mechanism may be related to the dysfunction of the dopamine system. One study, which assessed striatal presynaptic and postsynaptic dopaminergic status in patients with RLS and PLMS by SPECT imaging, found reduced levels of D2 receptor binding in the striatum of patients with RLS-PLMS²⁵. In addition, nocturnal urinary excretion of dopamine and its metabolite HAV was reduced in subjects with PLMS compared to those without PLMS, demonstrating the reduced activity of the dopamine system in patients with PLMS²⁶. Dopamine deficiency leads to inhibition of prefrontal cortex function, which in turn affects cognitive function, especially in working memory, planning, and attention²⁷. The relationship between PLMS and cognitive function remains controversial, with emerging evidence suggesting potential disorder-specific. For instance, in patients with isolated REM sleep behavior disorder (iRBD), higher PLMS indices were paradoxically associated with better executive function, despite poorer sleep continuity²⁸. Such discrepancies may arise from distinct pathophysiological mechanisms across sleep disorders.

Our study found that PLMS-related arousal leads to worse cognitive function. PLMS is usually accompanied by arouses or microarousals, with EEG frequency activation, resulting in sleep fragmentation^29,30. One study compared polysomnography results of simple stroke patients without cognitive impairment and vascular cognitive impairment-no dementia (VCIND) patients, and found that VCIND patients had varying degrees of sleep abnormalities. These sleep disturbances include clinically diagnosed sleep abnormalities on PSG (e.g., decreased sleep efficiency (SE), increased awakenings, and increased periodic leg movements (PLMS) and self-reported sleep disturbances (e.g., Higher PSQI scores), which may be related to cognitive dysfunction³¹. Frequent awakenings during the night, resulting in decreased sleep efficiency. Another study showed that poor sleep quality can lead to an increase in abnormal levels of beta-amyloid and tau in the cerebrospinal fluid, which can affect cognitive function³².

Data from this study indicate that reduced REM sleep leads to poor cognitive function. In a study of REM sleep deprivation (REMD) in rats via a treadmill with automatic REM sleep detection, an 85% reduction in REM sleep in rats led to a significant reduction in cell proliferation in the dentate gyrus of the rat hippocampus³³. Granule cells within the hippocampal dentate gyrus are important for the maintenance of learning and memory functions²³. Our results support the above experiment that reduced REM sleep leads to poor cognitive function. The mechanism may be as follows: PLMS contributes to frequent nocturnal awakenings, requiring re-initiation of sleep from the N1 stage. This process fragments sleep architecture, reduces total sleep time, and diminishes REM sleep duration, ultimately impairing cognitive function.

In this study, we found that older patients had poor cognitive function. And the older the age, the stronger the association between PLMS and cognitive function. First, with age, the brain volume gradually shrinks, especially the prefrontal cortex and hippocampus volume shrinks more obviously, and these areas are closely related to memory, learning, and cognitive function^34,35. Second, estrogen has a direct effect on brain regions controlling cognitive function. Estrogen can regulate synaptic plasticity in the hippocampus³⁶, and can also be involved in the regulation of health, cognitive function, and memory processes as a neuromodulator³⁷. This could also explain the poorer cognitive function of middle-aged and older women compared to men. Cognitive function is affected by the postmenopausal decline in estrogen levels in middle-aged and older women, especially in executive functions³⁸. Further studies are warranted to explore the age-specific effects of PLMS.

Numerous previous studies have demonstrated the negative effects of smoking on cognitive functioning. Ott et al. conducted a prospective study with a large sample and found that nonsmokers the Mini-Mental State Examination (MMSE) mean score declined by 0.03 points per year, whereas both ex-smokers and current smokers had higher declines in mean MMSE scores than nonsmokers³⁹. This is consistent with our findings that smokers scored significantly lower on cognitive function-related scales than non-smokers. Harmful substances in tobacco lead to the production of reactive oxygen species and oxidative stress, resulting in damage to nerve cells and affecting the processing and transmission of information, thus affecting cognitive function⁴⁰. Long-term smokers showed poorer general intelligence, visuospatial learning and memory, and fine motor dexterity⁴¹.

Our findings demonstrate that patients with > 12 years of education had higher MoCA scale scores than those with ≤ 12 years. Similar to the results of this study, a large-sample study of more than 6000 community residents over a 14-year period found that education was strongly associated with cognitive function but not with the rate of cognitive decline⁴². Some findings suggest that education level can predict cognitive function and decline⁴³. A study of 14,883 subjects found that education was a significant predictor of cognitive decline in patients with baseline MMSE > 23, whereas education was not a significant predictor of cognitive decline in those with baseline MMSE ≤ 23⁴⁴. The relationship between educational level and cognitive functioning needs to be further investigated.

There is a controversy between total sleep time and cognitive function. The results of this study found longer total sleep duration in the group with better cognitive function. However, the results of a large-sample cross-sectional study of sleep duration and cognitive function in Spanish older adults showed that longer sleep duration was associated with poorer cognitive function⁴⁵. To understand the relationship between sleep duration and cognitive function, a study analyzing two nationally representative aging cohorts found that individuals with extreme sleep durations (≤ 4 or ≥ 10 h per night) experienced faster rates of overall cognitive decline and that there was an inverted U-shaped association between sleep duration and overall cognitive decline⁴⁶. The exact mechanism remains unclear and may be due to the increased β-amyloid load in the human brain as a result of short sleep⁴⁷. Too much sleep may lead to increased activity in inflammatory pathways, which in turn affects cognitive function^48,49.

Although the cross-sectional design of this study revealed the association and mediating pathways between PLMS and cognitive function, the causal direction requires further validation through longitudinal studies. Future research could track the trajectories of PLMS severity, REM sleep dynamics, and cognitive decline in the same cohort—for example, by incorporating repeated measures at intervals of 3–5 years—to clarify whether PLMS drives cognitive deterioration or both phenomena share common underlying pathological mechanisms.

The study population comprised middle-aged and elderly inpatients who underwent polysomnography (PSG) evaluation due to sleep-related complaints, encompassing common symptoms such as daytime sleepiness and nocturnal snoring. Despite clinical heterogeneity (e.g., comorbidities like hypertension and diabetes), this characteristic reflects the diversity of real-world patients with sleep disorders. The findings may be more applicable to community or clinical populations with similar complaints, but caution should be exercised when extrapolating to asymptomatic older adults or specific subgroups.

Our research indicates that focusing on mechanisms related to PLMS, especially the preservation of REM sleep and the reduction of arousal, may help alleviate cognitive decline in individuals who are middle-aged and elderly. Clinicians may consider dopamine agonists to suppress PLMS severity, combined with melatonin to enhance REM continuity. Routine PSG screening in high-risk patients could guide these interventions, with monitoring of cognitive outcomes to assess efficacy. Addressing these modifiable pathways may offset 20% of PLMS-driven cognitive impairment, underscoring the clinical value of integrated sleep-focused care.

Limitations

Our study has some limitations. First, the exclusive use of hospitalized patients may have led to selection bias: our cohort may have represented individuals with more severe sleep complaints or coexisting conditions rather than the general population, which limits its generalizability to adults in the community. Future studies should include community population samples, combine biomarkers and prospective design to distinguish the effects of PLMS and neurodegeneration and improve the generalizability of the results. Second, our patient had only one polysomnography and nocturnal differences cannot be ruled out. Third, the genetic status of the individual was not considered in this study.Fourth, although formal criteria for MCI were not used, sensitivity analyses confirmed that stable PLMS were associated with cognition, so future multi-model studies are warranted to assess the impact on cognitive function. Finally, The evaluation of PLMS in this study was based on the AASM rules, but some studies have pointed out that the 2016 WASM rules can evaluate PLMS more effectively and accurately. Future studies can use double-standard comparative analysis to further verify the robustness of the conclusions²³.

Conclusion

PLMS may affect cognitive function in middle-aged and elderly people, and REM sleep duration and PLMS-related arousal play a partial mediating role. These findings underscore the importance of recognizing PLMS as a potential modifiable risk factor for cognitive impairment.In the future, we need to determine causal relationships through longitudinal studies and assess whether targeted interventions (such as suppressing PLMS or enhancing REM continuity) can alleviate cognitive decline. Incorporating PLMS screening into routine cognitive assessments may provide new avenues for early intervention in high-risk populations.

Author contributions

SS and LX: Conceptualization, Data collection and organization, Writing manuscripts, Writing Review. YZ and WY: Data collection and prepared Table 1–2. TY and GX: Data collection and prepared Table3-4,Fig. 1. CC and TT: Supervision, Funding acquisition, Writing Review. All authors have read and approved the manuscript.

Funding

This study was funded by the Yangzhou Natural Science Foundation-Youth Fund Project (YZ2017111).

Data availability

The datasets used and analysed during the current study available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

The study was approved by the Medical Ethics Committee of Yangzhou University Hospital (Ethics No. 2023-YKL09). Informed consent was obtained from all the patients. All methods were performed in accordance with the relevant guidelines and regulations.

Consent for publication

Not applicable.