Sleep Matters: Exploring the Link Between Sleep Disturbances and Fatigue in Rheumatoid Arthritis

Abstract

Objective

Fatigue is a prevalent and debilitating symptom for patients with rheumatoid arthritis (RA). Although patients and rheumatologists often attribute fatigue to inflammation, other factors such as sleep disturbances are frequently overlooked. This study aims to explore the relationship between subjective (self‐reported) and objective (actigraphy based) sleep parameters and self‐reported fatigue in patients with RA.

Methods

This cross‐sectional analysis included data from 48 adult patients with RA from a single academic rheumatology practice. Sleep data were obtained daily over 14 days with actigraphy (objective) and the Karolinska Sleep Diary (subjective). Fatigue was assessed using the Patient‐Reported Outcome Measurement Information System (PROMIS) fatigue computerized adaptive test. Spearman's correlations and linear regression analyses were used to examine associations between sleep parameters and fatigue, adjusting for swollen joint count, pain intensity, and symptoms of depression.

Results

Subjective sleep parameters showed significant correlations with PROMIS fatigue. Longer total sleep time (ρ = −0.4, P < 0.01), higher sleep efficiency (ρ = −0.42, P < 0.01), and better sleep quality (ρ = −0.5, P < 0.01) were associated with lower levels of fatigue. Objective actigraphy‐based sleep parameters were not significantly associated with PROMIS fatigue. Separate linear regression models demonstrated that each subjective sleep parameter remained significantly associated with fatigue after adjusting for covariates.

Conclusion

Self‐reported poor sleep duration, efficiency, and quality were significantly associated with fatigue in patients with RA, whereas objective actigraphy‐based sleep parameters were not, supporting the integration of self‐reported assessment of sleep disturbances into RA treatment plans to improve patient outcomes.

INTRODUCTION

Over 50% of patients with rheumatoid arthritis (RA) report fatigue, often on a daily basis. ¹ Despite fatigue being one of the most commonly reported and psychologically distressing symptoms experienced by patients with RA, it is rarely directly targeted for treatment. ² Patients with RA and rheumatologists often attribute fatigue to disease activity and inflammation and rely on standard‐of‐care therapy with disease‐modifying antirheumatic drugs (DMARDs) to manage fatigue. However, several studies have reported that the association between fatigue and systemic markers of inflammation, such as erythrocyte sedimentation rate and C‐reactive protein, are modest at best, ³ and although treatment with DMARDs improves fatigue, the magnitude of effect is small. ⁴

SIGNIFICANCE & INNOVATIONS.

• Use of daily sleep diaries: Most studies of rheumatoid arthritis (RA) have used retrospective assessments of sleep quality over the past week or month. By employing daily sleep diaries over a 14‐day period, this research provides a more nuanced assessment of the relationship between specific self‐reported sleep parameters and fatigue.

• Independent impact of sleep on fatigue: This study demonstrates that self‐reported measures of poor sleep (short duration, low efficiency, and poor quality) are independently associated with fatigue in patients with RA, even after adjusting for inflammation, pain, and symptoms of depression. This finding challenges the traditional view that fatigue in RA is primarily driven by inflammatory disease activity and pain.

• Targeting sleep in RA management: Our results highlight the importance of routinely assessing and addressing self‐reported sleep disturbances as part of comprehensive RA treatment plans, as interventions aimed at improving self‐reported sleep could help alleviate fatigue and enhance overall patient well‐being.

• Foundation for future research: This study showcases the need for future longitudinal research to establish causal relationships between sleep disturbances and fatigue in RA and develop effective sleep‐focused interventions for managing fatigue in patients with RA.

This focus on RA‐related factors has often contributed to the oversight of other important contributors, such as sleep disturbances. Dar et al found that patients with RA predominantly attributed fatigue to joint inflammation and pain, with minimal consideration given to sleep quality. ⁵ Similarly, da Silva et al reported that patients and rheumatologists often prioritize rheumatic disease management over addressing sleep‐related issues. ⁶

This study explores the relationship between objective (actigraphy) and self‐reported sleep parameters and fatigue in patients with RA. We aim to identify modifiable sleep‐related factors that could become specific targets for interventions aimed at mitigating fatigue. We hypothesize that impairments in specific sleep parameters (eg, low sleep duration, poor sleep efficiency, impaired wake after sleep onset [WASO], and poor sleep quality) are significantly associated with higher fatigue, even when controlling for RA‐related factors such as inflammation and pain.

METHODS

We conducted a cross‐sectional analysis of 48 adult patients with RA enrolled in the Sleep, Pain, and Autonomic Function in RA (SPAN‐RA) study. ⁷ All participants were from a single academic rheumatology practice, ≥18 years old, and met the American College of Rheumatology/EULAR 2010 criteria for RA. ⁸ Because the original SPAN‐RA study was designed to assess heart rate variability, the enrollment criteria excluded patients with diagnoses of cardiac arrhythmias and obstructive sleep apnea (OSA), as well as anyone routinely taking as well as anyone routinely taking beta blockers, central acting pain medications, opioids, and sedatives. Ethics approval was obtained from the Institutional Review Board of Northwestern University. All participants provided written informed consent.

Actigraphy‐based sleep measures

Actigraphy data were recorded using the Actiwatch Spectrum (Phillips Respironics) and worn on participants’ nondominant wrists for 14 days. Data were sampled at 30‐second intervals and processed using the manufacturer's proprietary software (Actiware, version 6.0). Main resting intervals were manually scored. Any 24‐hour recording period including ≥4 hours of nonwear was excluded. Sleep parameters included sleep duration, sleep efficiency, and WASO. Sleep duration, measured as the total time spent asleep in the main rest interval, was calculated in minutes. Sleep efficiency measured the percentage of time spent asleep while in bed and was calculated as 100% × total sleep time over time in bed. WASO, defined as the time, in minutes, spent awake after the onset of sleep, was calculated as total time of the intervals scored as wake between sleep onset and sleep offset in the sleep interval. Sleep measures were averaged across all valid days in the recording period.

Self‐reported sleep measures

Self‐reported measures of sleep were collected using the Karolinska Sleep Diary. ⁹ Participants were instructed to complete this diary every morning for 14 days. Subjective sleep parameters included duration (subjective sleep length in 10‐minute intervals), efficiency (derived from sleep length/time spent in bed), WASO (minutes spent awake after falling asleep), and quality. Sleep quality was derived from a summated index (score range 4–20, higher scores indicating better sleep quality) constructed from four items in the Karolinska Sleep Diary log (sleep quality, restless sleep, difficulties falling asleep, and premature [final] awakening). ⁹ All measures were averaged over the total number of sleep diary days.

Other measures

All individuals completed the Patient‐Reported Outcome Measurement Information System (PROMIS) fatigue and depression computerized adaptive tests. PROMIS T scores are standardized metrics derived from item response theory–based assessments. T scores have a mean of 50 (representing the general population) and an SD of 10. Higher T scores indicate more of the concept being measured. For example, PROMIS fatigue T scores above 55 reflect moderate fatigue, and scores above 60 indicate severe fatigue. ¹⁰ Pain intensity was assessed via a pain intensity numerical rating scale ranging from 0 to 10. Swollen joint count was determined by a single trained assessor using a standardized 28‐joint count.

Statistical analysis

Descriptive statistics were computed to summarize the demographic and clinical characteristics of the cohort. Spearman's correlations (rho) were performed to determine the strength of correlations between each wrist actigraphy‐ and diary‐based sleep parameter and the outcome of fatigue. Only sleep parameters that demonstrated significant correlations with fatigue were included in subsequent linear regression analyses.

Separate linear regression models were conducted to assess the associations between different sleep parameters and fatigue, with the PROMIS fatigue T score as the outcome. For the purposes of analysis, sleep duration was expressed in 10‐minute increments to provide a more interpretable measure of its association with fatigue. Both unadjusted and adjusted models were constructed, with the latter controlling for potential confounders of depression, swollen joint count, and pain intensity.

All data analyses were performed using R (version 4.2.2). The strength and direction of associations were determined using regression coefficients (beta) with 95% confidence intervals (CIs). An alpha level was set at P < 0.05 for all analyses. With a sample size of 48, we have 80% power to detect a medium effect size of Cohen's f² = 0.19 using a multiple regression model with four independent variables at an α level of 0.05. Additionally, we have adequate power (greater than 80%) to detect correlations of 0.42 or greater using a correlation test at an α level of 0.05.

RESULTS

Baseline participant characteristics

The study cohort comprised 48 patients, all with complete sleep and fatigue data (Table 1). Participants were predominantly female (95.8%), White (52.1%), seropositive (70.8%), and using DMARDs (89.6%). Actigraphy‐derived sleep characteristics were reported in a previous publication. ⁷

Table 1.

Baseline characteristics of study participants, N = 48^*

	Value
Demographics
Age, y	55.0 (41.5–64.8)
BMI	26.6 (22.4–32.8)
Female, %	95.8
Race and ethnicity, %
White	52.1
Asian	6.3
Black or African American	12.5
Hispanic or Latino	25.0
Other	4.2
Clinical factors
Clinical Disease Activity Index	5.0 (2.0–12.0)
Swollen joint count (0–28)	0.0 (0.0–2.0)
Tender joint count (0–28)	1.0 (0.0–4.2)
Patient global (patient global assessment of disease activity) (0–10)	2.0 (1.0–4.0)
Assessor global (assessor global assessment of disease activity) (0–10)	1.0 (0.0–2.0)
Pain intensity NRS (0–10)	3.0 (1.0–5.0)
PROMIS depression	47.2 (39.6–51.8)
PROMIS fatigue	48.6 (42.5–57.1)
Steroid use, %	12.5
RA duration, y	9.8 (4.0–16.2)
Seropositivity, %	70.8
Any DMARD use, %	89.6
Conventional DMARD, %	62.5
Biologics and targeted synthetic DMARD, %	54.2
Sleep diary measures
Sleep duration, min	427.4 (402.1–454.1)
Sleep efficiency, %	89.9 (86.0–94.0)
Wake after sleep onset, min	11.9 (6.5–30.1)
Sleep quality index	14.1 (12.4–16.3)

^*Values are median (Q1–Q3), unless noted otherwise. BMI, body mass index; DMARD, disease‐modifying antirheumatic drug; NRS, numeric rating scale; PROMIS, Patient‐Reported Outcome Measurement Information System; RA, rheumatoid arthritis.

Correlations between sleep parameters and fatigue

Self‐reported sleep measures

Participants who reported longer total sleep time and higher sleep efficiency had significantly lower fatigue (ρ = −0.4, P < 0.01, and ρ = −0.42, P < 0.01, respectively; Figure 1B and D). Similarly, a better sleep quality index score was associated with lower fatigue (ρ = −0.5, P < 0.01; Figure 1C). Remaining self‐reported sleep measures, including WASO, onset latency, number of awakenings, and other self‐reported sleep log measures, were not significantly associated with fatigue.

Figure 1.

Scatter plots illustrating the relationship between self‐reported sleep parameters and PROMIS fatigue. Each plot shows the correlation between PROMIS fatigue and (A) WASO, (B) sleep efficiency, (C) sleep quality index, and (D) sleep duration, based on Spearman's correlation coefficients. PROMIS, Patient‐Reported Outcome Measurement Information System; WASO, wake after sleep onset.

Actigraphy‐based sleep measures

Actigraphy‐based measures of sleep duration (ρ = −0.08, P = 0.608), sleep efficiency (ρ = 0.00, P = 0.986), WASO (ρ = 0.01, P = 0.921), sleep fragmentation (ρ = 0.09, P = 0.564), number of awakenings (ρ = 0.1, P = 0.491), and sleep onset latency (ρ = 0.01, P = 0.921) were not significantly associated with PROMIS fatigue. Because the actigraphy‐based measures were not significantly associated with PROMIS fatigue, they were not included in subsequent analyses.

Linear regression analysis of fatigue on sleep diary measures

Shorter self‐reported sleep duration (β = −0.66, 95% CI −1.2 to −0.09), lower sleep efficiency (β = −0.67, 95% CI −1.10 to −0.25), and worse sleep quality (β = −1.50, 95% CI −2.40 to −0.62) were significantly associated with worse fatigue in unadjusted models. The association between WASO and fatigue was not statistically significant (β = 0.06, 95% CI −0.09 to 0.21). After adjustment for symptoms of depression, swollen joint count, and pain intensity, significant associations between self‐reported sleep duration (β = −0.50, 95% CI −0.94 to −0.07), sleep efficiency (β = −0.41, 95% CI −0.76 to −0.05), sleep quality (β = −0.81, 95% CI −1.60 to −0.05), and fatigue persisted, although these associations were attenuated (Table 2). The association between WASO and fatigue was not statistically significant after covariate adjustment (β = 0.02, 95% CI −0.10 to 0.13).

Table 2.

Linear regression analysis of PROMIS fatigue on sleep diary measures, N = 48^*

Characteristic	Unadjusted			Adjusted
	β	95% CI	P value	β	95% CI	P value
Main predictor: sleep duration
Sleep duration, 10‐min intervals	−0.66	−1.20 to −0.09	0.024	−0.50	−0.94 to −0.07	0.025
Swollen joint count	–	–	–	0.72	−0.33 to 1.8	0.2
Pain intensity numeric rating scale	–	–	–	1.0	0.27 to 1.8	0.008
PROMIS depression T score	–	–	–	0.42	0.18 to 0.66	<0.001
Main predictor: sleep efficiency
Sleep efficiency	−0.67	−1.1 to −0.25	0.002	−0.41	−0.76 to −0.05	0.028
Swollen joint count	–	–	–	0.33	−0.78 to 1.4	0.5
Pain intensity numeric rating scale	–	–	–	0.93	0.18 to 1.7	0.016
PROMIS depression T score	–	–	–	0.43	0.19 to 0.67	<0.001
Main predictor: WASO
WASO, min	0.06	−0.09 to 0.21	0.4	0.02	−0.10 to 0.13	0.7
Swollen joint count	–	–	–	0.70	−0.43 to 1.8	0.2
Pain intensity numeric rating scale	–	–	–	1.0	0.22 to 1.8	0.013
PROMIS depression T score	–	–	–	0.45	0.20 to 0.71	<0.001
Main predictor: sleep quality
Sleep quality index	−1.5	−2.4 to −0.62	0.001	−0.81	−1.6 to −0.05	0.038
Swollen joint count	–	–	–	0.72	−0.35 to 1.8	0.2
Pain intensity numeric rating scale	–	–	–	0.72	−0.06 to 1.5	0.071
PROMIS depression T score	–	–	–	0.43	0.19 to 0.67	<0.001

^*CI, confidence interval; PROMIS, Patient‐Reported Outcome Measurement Information System; WASO, wake after sleep onset.

DISCUSSION

In this study of patients with established RA, longer self‐reported sleep duration, higher self‐reported sleep efficiency, and better self‐reported sleep quality were associated with lower levels of self‐reported fatigue. These relationships persisted after adjusting for joint inflammation, pain, and depressive symptoms. These findings suggest that self‐reported poor sleep contributes independently to the experience of fatigue among patients with RA, challenging the commonly held view that fatigue is mainly driven by inflammatory disease activity or pain in this population. ²

Our study builds on previous studies examining the association between sleep quality and fatigue. One of the largest studies (N = 158) was a cross‐sectional analysis by Katz et al, which showed that poor sleep quality (assessed by the Pittsburgh Sleep Quality Index), inactivity, depression, and obesity were all associated with fatigue. ¹¹ Our study extends these results by providing information on specific self‐reported sleep parameters (eg, sleep duration and sleep efficiency) through daily sleep diaries. This more nuanced view of self‐reported sleep disturbances enabled us to identify sleep duration and sleep efficiency (in addition to the more nebulous concept of sleep quality) as specific targets for interventions to improve fatigue.

Our study is consistent with a mobile health study performed in 254 patients with self‐reported RA recruited from the National Rheumatoid Arthritis Society, a patient organization in the United Kingdom. ¹² In this study, McBeth et al reported significant associations between both subjective and objective sleep parameters and health‐related quality of life. Although the authors did not specifically report associations between sleep parameters and fatigue, the association between sleep and health‐related quality of life diminished when fatigue was included in the models, implicating fatigue as a potential mediating factor. Furthermore, the importance of sleep duration was highlighted by an interventional in‐laboratory study, in which 27 patients with RA underwent partial night sleep deprivation. The morning following partial night sleep deprivation, patients with RA reported significant increases in fatigue and other related symptoms, such as pain and self‐reported assessments of disease activity. ¹³ Together, these findings suggest that interventions aimed at increasing total sleep duration—and possibly enhancing sleep efficiency—may be effective for improving fatigue and other health‐related outcomes.

The associations between self‐reported sleep duration, sleep efficiency, sleep quality, and the outcome of fatigue remained significant, even after adjustment for other hypothesized predictors of fatigue, including joint inflammation, pain, and symptoms of depression. Pain and symptoms of depression were also significantly associated with fatigue, but joint inflammation, measured by swollen joint count, was not. These findings are consistent with other studies, suggesting that despite the focus on inflammation as a cause for fatigue, the experience of fatigue is likely rooted in multiple causes, which may differ from individual to individual. In a population of patients with long‐standing treated RA, inflammation may not be the primary contributor. Routine assessments of self‐reported sleep, pain, and depression should be incorporated in a comprehensive approach to identify and target possible sources of fatigue in patients with RA.

Contrary to the observation that self‐reported measures of sleep were associated with fatigue, actigraphy‐based measures of sleep were not. This discrepancy raises an important question: Is it the actual sleep, as measured objectively, that is most relevant to fatigue, or is it the perception of sleep that matters more? Prior studies suggest that self‐reported sleep quality may be a stronger predictor of fatigue than actigraphy‐derived sleep measures because it captures an individual's experience of restfulness, sleep satisfaction, and nighttime discomfort, which actigraphy cannot fully assess. ¹⁴ Additionally, self‐reported sleep disturbances may reflect broader psychosocial factors, including stress, pain perception, and cognitive biases, that influence fatigue perception. ¹⁵  It is common for self‐reported measures to be more strongly associated with other self‐reported measures than with objectively measured data. ¹⁶ This may partly be due to bias in the way individuals respond to questionnaires. Another potential reason for the discrepancy is that self‐reported and actigraphy‐based measures assess different aspects of sleep. Actigraphy primarily captures movement‐based parameters, whereas self‐reports reflect an individual's perception of sleep quality and disturbances. Both are imperfect assessments, but both can provide valuable information. Our findings highlight the importance of addressing patients’ perceptions of sleep disturbances in clinical management rather than relying solely on objective sleep assessments.

Strengths of this study include the comprehensive assessment of sleep characteristics through daily sleep diary assessments, as well as the consideration of potential confounders. Limitations include the cross‐sectional design, which restricts the ability to infer causality. Additionally, the relatively healthy study population with low disease activity and mild sleep disturbances may not be representative of all patients with RA, potentially limiting the generalizability of these findings. The original SPAN‐RA study enrollment criteria excluded patients with self‐reported OSA. As OSA may not always be clinically diagnosed or captured via actigraphy, its potential impact on sleep quality and fatigue cannot be entirely ruled out. Future research should aim to include a broader range of disease activity levels and sleep disturbances to better understand these relationships. We also plan to evaluate the relationship of physical activity, assessed by actigraphy, and fatigue in this population.

This study highlights the role of self‐reported sleep disturbances in the experience of fatigue among patients with RA. Despite the traditional focus on inflammation and pain, our findings suggest that poor sleep is a significant and independent contributor to fatigue. Notably, our sample was biased toward individuals without certain sleep disorders and not taking specific medications. Associations between sleep disturbances and fatigue may be even more pronounced in patients with existing sleep disturbances. Rheumatologists should recognize the importance of assessing and managing sleep disturbances to improve fatigue and overall well‐being in patients with RA. Additional longitudinal studies are needed to establish causal relationships and develop effective sleep‐focused interventions for this population.

AUTHOR CONTRIBUTIONS

All authors contributed to at least one of the following manuscript preparation roles: conceptualization AND/OR methodology, software, investigation, formal analysis, data curation, visualization, and validation AND drafting or reviewing/editing the final draft. As corresponding author, Dr Lee confirms that all authors have provided the final approval of the version to be published and takes responsibility for the affirmations regarding article submission (eg, not under consideration by another journal), the integrity of the data presented, and the statements regarding compliance with institutional review board/Declaration of Helsinki requirements.

Supporting information

Disclosure form.