Sleep Problems in Patients with Rheumatoid Arthritis

Study design and recruitment

This was an observational, cross-sectional, multicenter study in a population of Belgian patients with established RA. Rheumatology clinics were invited to participate if their treatment protocols included conventional and biologic DMARD. To study a representative sample, academic, as well as nonacademic, RA clinics were invited to participate. Further, participants were categorized according to baseline disease activity level into remission/LDA [28-joint Disease Activity Score (DAS28)-C-reactive protein (CRP) ≤ 3.2] or moderate to high disease activity (DAS28-CRP > 3.2). Next, each site was asked to include a predefined proportion of patients according to the distribution of disease activity levels, as observed in a previous Belgian population-based study — that is, 50.6% patients in remission/LDA, 35.8% with moderate disease activity (3.2 < DAS28-CRP < 5.1), and 13.6% with high disease activity (DAS28-CRP ≥ 5.1)²².

Study population

Participants were recruited during their routine outpatient visits and were eligible if they met the following criteria: aged 18–75 years, a diagnosis of RA according to the revised 1987 criteria of the American College of Rheumatology²³, currently receiving conventional and/or biologic DMARD, fulfilled the center’s predefined disease activity criteria, and able to provide written informed consent. Exclusion criteria were a past history of major depressive disorder, psychiatric illness, or substance abuse; a concurrent diagnosis of fibromyalgia (FM); a lifestyle that placed the patient at serious risk of sleep disturbances (i.e., shift work or night work); traveling through more than 3 time zones during the week before screening or during the study; a body mass index > 35 kg/m²; sleep-related breathing disorders; restless legs syndrome; and/or periodic limb movement disorder.

This study was approved by the Medical Ethics Committee of the UZ KU Leuven.

Sociodemographic and clinical status assessment

Data were collected on the year of RA diagnosis, socioeconomic and demographic status, past and current pharmacological treatment for RA, concurrent medications (including sleep medication), medical history, and comorbidities that were recorded according to the body system affected.

RA disease activity assessment

The level of RA disease activity was assessed according to the DAS28-CRP European League Against Rheumatism response criteria²⁴.

Patient-reported sleep measures assessment

The Athens Insomnia Scale (AIS) is an 8-item questionnaire designed to measure sleep difficulty based on the 10th revision of the International Statistical Classification of Diseases and Related Health Problems criteria for insomnia. The first 4 items assess difficulty with sleep quantity, including sleep induction, nighttime awakenings, early morning awakenings, and total sleep duration. The fifth item relates to overall sleep quality, and the last 3 items refer to the effect of nocturnal sleep disturbance on daytime performance. Each item is scored from 0 (no problem) to 3 (very serious problem), corresponding with the experience of sleep difficulty in each item for at least 3 times a week, during the last month. Total scores range from 0 (absence of any sleep-related problem) to 24 (the most severe degree of insomnia), with a cutoff score of ≥ 6 for a diagnosis of insomnia^25,26.

The Pittsburgh Sleep Quality Index (PSQI) measures sleep quality over the past 4 weeks, using 19 of the overall 24 items. Good sleepers can be distinguished from poor sleepers through the measurement of 7 subscales: subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances (i.e., number of awakenings during the night and number of arousals), use of sleep medication, and daytime dysfunction. Each subscale is rated from 0 to 3, where 3 reflects a more severe sleep complaint. A total PSQI score ≥ 5 is indicative of a poor sleeper²⁷.

The Epworth Sleepiness Scale (ESS) measures excessive daytime sleepiness over the past 2 weeks. Respondents are asked how likely they are to doze in the following situations: sitting and reading, watching television, sitting inactive in a public place (e.g., a meeting), as a passenger in a car for an hour without a break, lying down to rest in the afternoon when circumstances permit, sitting and talking to someone, sitting quietly after a lunch without alcohol, and in a car while stopped for a few minutes in traffic. Each situation is scored from 0 (would never doze) to 3 (high chance of dozing). The commonly applied cutoff score of ≥ 9 was used to indicate excessive daytime sleepiness and reflects “very sleepy and should seek medical advice”²⁸.

Other PRO assessment

Additional PRO were measured to explore the overall burden of RA disease and their possible influence on aspects of sleep. All PRO scales included had been validated for use in the Belgian population.

Visual analog scales (VAS) were used to assess fatigue and pain. Patients were asked to rate their experience of fatigue and pain during the last week, each on a VAS of 0–100, with a higher score indicating more severe fatigue or pain²⁹.

Health status was assessed using the Medical Outcomes Study Short Form-36 (SF-36), a generic instrument comprising 8 dimensions: limitations in physical functioning because of health problems, limitations in social functioning because of physical or mental health problems, role limitations caused by physical health problems, role limitations caused by personal or emotional problems, bodily pain, general mental health, vitality, and general health perception^30,31.

The Health Assessment Questionnaire-Disability Index (HAQ-DI) was used to evaluate patients’ functional disability status, measuring the ability to perform daily functional activities in 8 categories: dressing, rising, eating, walking, hygiene, reach, grip, and usual activities³².

The Positive and Negative Affect Schedule measures respondents’ experience of positive and negative emotions during the past 4 weeks. The 10-item positive affect subscale reflects emotions such as interest, determination, enthusiasm, and pride. The 10-item negative affect subscale reflects emotions such as fear, distress, hostility, and shame. Items are scored using a 5-point Likert scale (1 = very slightly or not at all, 5 = extremely)³³.

Coping with pain, defined as patients’ behavioral and cognitive attempts to manage or tolerate pain, was assessed with a Coping Strategies Questionnaire. Two global questions were answered: “having control on pain”, scored between 0 (absence of control) and 6 (total control), and “ability to decrease pain”, scored between 0 (cannot reduce pain) and 6 (can totally reduce pain).

Data analysis

Data were analyzed using statistical software (SAS, version 9.1; SAS Institute Inc.). Descriptive analyses are reported for all sleep quality dimensions, as well as for all PRO. Chi-squared and t tests were used to compare DAS28-CRP and sleep scale scores. A consistency check was also conducted for sleepiness as recorded by sleepiness-related questions of the AIS and PSQI (question 9 “sleepiness during the day”, and question 8 “difficulty to stay awake”, respectively), and each of the 8 questions/variables of the ESS. Pearson’s correlation coefficients were obtained to measure the strength of the association between DAS28-CRP and each sleep quality score. Searching for outliers was performed for all variables and then values were corrected before incorporation into the regression models. Covariates listed in Table 1 were explored in the different models.

The primary analysis model consisted of the dependent variable sleep quality (operationalized as the aggregate PSQI, AIS, or ESS score) and the independent variable disease activity (measured by DAS28-CRP as a continuous or a categorical variable). Analyses were performed on the overall RA sample, as well as by disease activity status: remission/LDA (DAS28-CRP ≤ 3.2) and moderate to high (DAS28-CRP > 3.2). Variables that, in the univariate analysis, showed an association with sleep quality with a p- or F-value of ≤ 0.1 for continuous and categorical covariates, respectively, were entered separately in the multivariate models. The DAS28-CRP covariate was “forced” into multivariate models (i.e., included in the model even if p or F was > 0.1) to determine the best models (i.e., the ones with significant relationships between DAS28-CRP and sleep scores), together with the maximum number of significant additional covariates. Only covariates with at least 5% observations not equal to zero were considered for entry in the multivariate models. The strengths of the associations in logistic models were expressed using OR with 95% CI calculated for all potential factors associated with sleep quality and sleepiness. In the logistic regression models, the odds for success were determined using the following cutoffs: PSQI > 5 (vs ≤ 5), AIS ≥ 6 (median; vs < 6), and ESS ≥ 9 (vs < 9). The fit of each model was also considered for the selection of best association models (Model R² and p value for linear models and likelihood ratio, score, and Wald for logistic models). Collinearity between variables was assessed for the overall RA sample and each sleep questionnaire, and structural equation modeling (SEM) was performed to test the hypothesis about a relationship between the observed covariates with sleep quality scores, and to check for correlations among the study covariates. SEM was run with path analyses to achieve the best models in both significant associations and model fit compared with all other models. LINEQS was used in the models for the direct and indirect effects of the observed covariates on sleep quality scores. The findings with DAS28-CRP as a continuous variable are reported here.