Introduction

Sleep plays a pivotal function in life. The 24-h sleep and wake rhythm is related to certain circadian patterns of immune responses [1, 2]. Sleep is reciprocally related to immune system [3]. Sleep disorders (SDs) may alter immune function, and immune disorders eventually impair sleep quality [4,5,6,7].

Rheumatoid arthritis (RA) is a chronic and systemic inflammatory disease, which causes articular demolition and protean systemic presentation [8, 9]. Most reports have focused on the treatment of RA and improvement for quality of life [10,11,12,13]. Studies investigating the risk factors for developing RA are scant.

Cross-sectional studies have indicated that patients with poor control of RA exhibit a reduction in sleep quality [7, 14]. Inadequate sleep may exacerbate pain severity, fatigue, and depression in patients with RA, potentially limiting their daily activity [15]. Whether sleep problems cause or are an independent risk factor of RA remains unclear in these cross-sectional studies. Therefore, we conducted a longitudinal cohort study to investigate the effect of SDs on the risk of developing RA.

Methods

Data source

This current study involved analyzing data sets in the Taiwan National Health Insurance Research Database (NHIRD). The government initiated the National Health Insurance (NHI) program, which is a single-payer system on March 1, 1995 and covers approximately 99% of the residents in Taiwan [16]. The National Health Insurance Administration (NHIA) managed and maintained the NHIRD. We used the Longitudinal Health Insurance Database (LHID), which contains claims data of 1,000,000 people randomly selected insured people for the period of 1996–2011 in the NHIRD. All study participants were anonymous and deidentified by the NHIA before release of claim data for public research. The Institutional Review Board of China Medical University and Hospital in Taiwan approved this study (CMU-REC-101-012).

Sampled patients

For the study cohorts, patients diagnosed with SDs (N = 67,455) between 1998 and 2002 were divided into two cohorts: those with sleep apnea (SA) syndrome (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 780.51 indicating insomnia with SA, 780.53 indicating hypersomnia with SA, and 780.57 indicating unspecified SA) and non-apnea SDs (ICD-9-CM codes 307.4 comprising non-organic SDs, insomnia, parasomnia, and circadian rhythm SDs, and 780.5 indicating sleep disturbance, except 780.51, 780.53, and 780.57). The index date was defined as the first date of SD diagnosis. We excluded patients with a prior history of RA (ICD-9-CM code 714) (N = 134) or aged < 20 years (N = 1567). We selected four non-SD patients for each SD patient from the LHID randomly by frequency-matched with age (every 5-year interval), sex, and the year of index date to conduct the comparison cohort (N = 64,350). Both cohorts used the same exclusion criteria.

Outcome

We identified patients with RA in the LHID and confirmed the diagnoses from the Registry of Catastrophic Illness Database (RCIPD) between 1998 and 2011 as the study end point. In Taiwan, the NHIA can issue a catastrophic illness card for a patient with RA who fulfills four or more diagnostic criteria according to the American College of Rheumatology criteria for RA diagnosis [17] through a scrutinized peer review process. RA patients with a catastrophic illness card can receive medical therapy without copayment. We followed all the patients till the date of diagnosis with RA, censoring, or the end of 2011.

Variables of interest

We categorized age into ≤ 34, 35–49, 50–64, and ≥ 65 years. The patients’ monthly income based on insurance premiums were classified into < 15,000, 15,000–19,999, and ≥ 20,000 NTD (1 USD equals approximately 30 NTD). The NHIA report beneficiaries residing in urbanization from level 1 to level 7 (levels 1 and 7 were the highest and lowest levels of urbanization, respectively) [18]. Because few people resided in levels 4–7, we combined the least urbanized populations into level 4. The baseline comorbidity included obesity (ICD-9-CM code 278), stroke (ICD-9-CM codes 430–438), coronary artery disease (CAD; ICD-9-CM codes 410–414), anxiety (ICD-9-CM codes 300.00), depression (ICD-9-CM codes 296.2, 296.3, 300.4, 311), diabetes mellitus (ICD-9-CM code 250), and other rheumatic diseases (ICD-9-CM code 710.0, 710.1, 710.2, 710.3, 710.4, 720).

Statistical analysis

Comparison of the demographics and comorbidities between the SD (SA and non-apnea SD) and comparison cohorts used the chi-squared test for binomial variables and the Student’s t test for continuous variables. The follow-up time (person-years) estimated incidence densities according to demographic status, monthly income, urbanization level, and comorbidity for each cohort. Cox proportional hazard regression models compared the risk of developing RA-associated SD in the SD and comparison cohorts presenting hazard ratios (HRs) and 95% confidence intervals (CIs). In addition, adjusted HRs were determined after we controlled for age, sex, monthly income (NTD), urbanization level, and the comorbidities such as obesity, stroke, CAD, anxiety, depression, diabetes mellitus, and other rheumatic diseases. The Cox models were applied to compare the HRs and 95% CIs of RA for the SA cohort with those of the non-apnea SD cohort. We conducted a survival analysis to estimate the cumulative incidence of RA in the SD and non-SD cohorts. The SAS statistical package for Windows (version 9.3, SAS Institute Inc., Cary, NC, USA) was used for analyses. A statistical significance was set by two-tailed P value of < 0.05.

Results

Comparison of demographics and comorbidities in both cohorts

We identified 65,754 newly diagnosed patients with SD (1404 SA patients and 64,350 non-apnea SD patients) and 65,753 non-SD controls in this study. The patients in both the SD and comparison cohorts were aged ≤ 49 years (50.1%), predominantly women (62.8%), lived in urbanized areas (56.1 vs 58.8%), and had income levels between 15,000 and 19,999 NTD (50.2 vs 48.4%). The SD cohort had a significantly higher proportion of comorbidities, namely, obesity, stroke, CAD, anxiety, depression, diabetes mellitus, and other rheumatic diseases, compared with the comparison cohort (all P < 0.001). The patients in the SA cohort were predominantly men (62.5%), whereas those in the non-apnea SD cohort were mostly women (63.4%) (Table 1).

Table 1 Comparison of demographics and comorbidity between the sleep disorders and comparison cohorts
Full size table

Comparison of incidence densities and risks of rheumatoid arthritis between the patients with and without sleep disorders according to demographic characteristics and comorbidities

The SD cohort showed higher overall incidence densities of RA than did the comparison cohort (3.31 vs 2.15 per 10,000 person-years, respectively). After we controlled for covariates, a 1.49-fold adjusted HR of RA was present in the SD cohort compared with the non-SD cohort (95% CI = 1.19–1.85). The patients aged younger than 49 years in the SD cohort had a 2.27-fold increased risk of RA compared with those in the counterparts (95% CI = 1.57–3.28). The incidence densities of RA were higher for the women than those for the men in both cohorts. The adjusted HR of RA for the SD cohort to comparison cohort was significantly higher in both the women (adjusted HR = 1.44, 95% CI = 1.13–1.82) and men (adjusted HR = 1.87, 95% CI = 1.03–3.40). For the patients with SDs, a monthly income < 15,000 NTD was associated with a significantly increased risk of RA relative to that of the counterparts (adjusted HR = 2.17, 95% CI = 1.29–3.67). The risk of RA for the SD cohort relative to the comparison cohort was significantly higher for those living in regions with the highest and second highest urbanization levels (adjusted HR = 1.94, 95% CI = 1.28–2.92 and adjusted HR = 1.59, 95% CI = 1.04–2.44, respectively). The relative risk of RA for the SD-to-comparison cohort was significantly higher for those who did not have comorbidity (adjusted HR = 1.60, 95% CI = 1.24–2.05) (Table 2).

Table 2 Comparison of incidence densities and hazard ratios of rheumatoid arthritis between the patients with and without sleep disorders according to demographic characteristics and comorbidity
Full size table

To compare incidence and risks of rheumatoid arthritis among the SA, non-apnea SD, and non-SD cohorts

The incidence of RA was the highest in the patients with SA, followed by those with non-apnea SD and the comparison cohort (4.11, 3.29, and 2.15 per 10,000 person-years, respectively). Compared with the comparison cohort, the adjusted HR of RA was 2.56 (95% CI = 1.11–5.91) and 1.47 (95% CI = 1.18–1.84) for the apnea SD and non-apnea SD cohorts, respectively. The adjusted HRs of RA for the SA and non-apnea SD cohorts were significantly increased for those aged ≤ 49 years compared with the counterparts. The incidence densities of RA among the three cohorts were higher in the women than in the men. The sex-specific adjusted HR for the SA cohort relative to the comparison cohort was 3.06-fold in the women (95% CI = 1.23–7.61). The risk of RA was higher in the patients with non-apnea SDs than that in the patients without SDs for both the women and men. Regarding monthly income stratification, the risk of RA was significantly higher in the patients with non-apnea SDs than in the patients without SDs for monthly incomes of < 15,000 (adjusted HR = 2.21, 95% CI = 1.31–3.73) and 15,000–19,999 NTD (adjusted HR = 1.39, 95% CI = 1.02–1.89).

The non-apnea SD patients living in the regions with the highest and second highest urbanization levels were related to increased risks of RA (adjusted HR = 1.93, 95% CI = 1.27–2.92 and adjusted HR = 1.58, 95% CI = 1.03–2.43, respectively). In the non-comorbid subgroup, the risk of RA was significantly higher for the non-apnea SD cohort (adjusted HR = 1.60, 95% CI = 1.24–2.05) compared with the comparison cohort (Table 3).

Table 3 Comparisons of incidence densities and hazard ratio of rheumatoid arthritis in the sleep apnea, non-apnea sleep disorders, and comparison cohorts
Full size table

Incidence rates and risks of rheumatoid arthritis in patients with diverse types of non-apnea sleep disorders

We further categorized non-apnea SDs into insomnia, sleep disturbance, and other SDs. Compared with people without SDs, patients with insomnia and sleep disturbance exhibited a substantially elevated incidence of RA with a significantly increased risk of RA development (adjusted HR = 1.47, 95% CI = 1.12–1.94 for insomnia and adjusted HR = 1.55, 95% CI = 1.20–2.00 for sleep disturbance) (Table 4).

Table 4 Incidence rates and hazard ratios of rheumatoid arthritis in patients with diverse types of non-apnea sleep disorders
Full size table

Comparison of cumulative incidence of rheumatoid arthritis between the SD and non-SD cohorts

The RA cumulative incidence curve illustrated an increased risk of developing RA in the SD cohort compared with that in the non-SD cohort (P < 0.001, log-rank test) (Fig. 1).

Fig. 1
figure 1

Kaplan-Meier method determined cumulative incidence of rheumatoid arthritis between sleep disorder cohorts and comparisons without sleep disorder

Full size image

Discussion

This longitudinal cohort study revealed that the patients with SDs exhibited a 1.49-fold greater risk compared with the general population after adjustment for covariates through 1,300,420 follow-up person-years. The incidence density of RA was the highest for the patients with SA, followed by those with non-apnea SD and the comparison cohort (4.11, 3.29, and 2.15 per 10,000 person-years, respectively). After we controlled for covariates, the risk of developing RA was 2.56 (95% CI = 1.11–5.91) for the SA cohort and 1.47 (95% CI = 1.18–1.84) for the non-apnea SD cohort when compared with the non-SD cohort.

The biological mechanism of RA development among the patients with SD remains unclear. The immunogenetics of RA discloses a central role for aberrant pathways of T cell activation at initiation and progression [19, 20]. CD4+ T cells link with human leucocyte antigen (HLA) class II molecules and certain costimulatory molecules expressed on the surface of antigen-presenting cells. Furthermore, the cooperation of T cells and B cells results in ongoing destruction of the synovial membrane [20]. In addition, numerous proinflammatory cytokines, namely, tumor necrosis factor alpha (TNF-α), interleukin (IL)-1, IL-6, and IL-8, play roles in the pathogenesis of RA [21]. Neuroimmunology indicated that sleep regulation in the central nervous system shows reciprocally related to the immune system [22,23,24]. The natural regulatory T cells follow a sleep-dependent rhythm and sleep can suppress T cell activity [2]. Sleep has been known to enhance immune defense for decades. Sleep impairment may derange immune function, which might perform a role in the risk of infectious diseases and inflammatory disorders [4, 25]. Sleep disturbance impairs T cell regulation and upregulates certain inflammatory cytokines, such as TNF-α, IL-1, and IL-6, as demonstrated in patients with inflammatory bowel disease [26,27,28].

Chronic intermittent hypoxia may activate systemic inflammation in patients with SA [29]. The proinflammatory cytokines, such as TNF-α and IL-6, and an inflammatory marker, namely, C-reactive protein in circulation, are increased in patients with SA [30, 31]. In addition, SA is strongly linked with obesity [32]. Excess adiposity related to impaired immune function has been reported in previous studies [33, 34]. Furthermore, obesity is associated with a condition of low-grade chronic inflammation and disturbed metabolic hormones [35]. Moreover, obesity may influence specific and unspecific immune responses mediated by cell and humoral mechanisms [36, 37]. Furthermore, SA is associated with TNF-α gene polymorphism [38]. Some of the polymorphism forms extend haplotypes with HLA class I and II alleles, which play a role in T cell immune-regulated diseases [39]. These plausible mechanisms may explain why the patients with SA had a considerable risk of RA development in our study.

The prevalence of anxiety and depression were higher in the SD cohort than in the comparison cohort. These findings are compatible with previous reports [40, 41]. The association of SD and anxiety may be associated with adaptive response to stress [42]. Otherwise, the relationship between sleep disturbance and depression is strongly related. SDs may become core symptoms of depression [42].

The incidence density rates of RA were higher for the women than for the men in both cohorts, which are accordant with previous studies [8, 43]. However, the SD cohort still exhibited a substantial risk of developing RA compared with the comparison cohort in both sexes (adjusted HR = 1.87, 95% CI = 1.03–3.40 for the men and adjusted HR = 1.44, 95% CI = 1.13–1.82 for the women). Although the incidence density rates of RA development were higher in the SD cohort than in the comparison cohort, the patients aged 49 years and younger exhibited a considerably higher risk of RA in age-stratified analysis. This result suggested that the effect of SDs on the RA risk was the greatest in younger adults.

The risk of RA for the SD cohort relative to the comparison cohort was significantly higher for those living in regions with the highest and second highest urbanization levels. Numerous studies have indicated that the individuals living in the urban areas are more apt to access medical services than those in the rural areas [44, 45]. For the patients with SDs, a monthly income of < 15,000 NTD was associated with a significantly increased risk of RA relative to that of the patients without SDs. The association of low socioeconomic status and higher risk of RA in this study was consistent with a previous study [46].

The strength of this large sample sized cohort study lies that it provides data on the risk of RA in people with SDs. We could trace the patients throughout the whole study period because the NHI is mandatory in Taiwan and each beneficiary was allocated a personal identification number. Furthermore, RA diagnosis was confirmed in the RCIPD, enhancing a high reliability for our analyses.

However, certain limitations exist when interpreting these findings. First, the NHIRD provides no family history for the patients. Second, patients with SDs may have been underrepresented because patients may rarely visit a clinician to discuss sleep problems [47]. The current study did not elaborate the risk of RA in different types of non-apnea SDs and diverse severity of SA because the result of polysomnography was not available in the NHIRD.

In summary, this longitudinal cohort study of 65,754 SD patients with 662,514 follow-up person-years demonstrated that the patients with SDs had a 49% increased risk of developing RA compared with the comparison cohort. Although the causal relationship between SDs and RA development is modest, a bidirectional relationship exists between SDs and RA. The findings remind the clinicians of being awareness of potential derangement of immune function and an increased RA risk in patients with SDs.