Muscle relaxant use and the associated risk of Sjögren’s syndrome and dry eye disease (including keratoconjunctivitis sicca) in patients with ankylosing a population-based cohort study

Authors: Weijie Wang, Jingyang Huang, Shujun Xia, Lin Cai, James Cheng-Chung Wei

Abstract

Background:

Muscle relaxants are widely used for the treatment of ankylosing spondylitis (AS) and inevitably have many adverse reactions. Individuals with AS frequently present with ocular symptoms. However, the association of muscle relaxants with the increase of dry eye or Sjogren’s syndrome (SS) in AS patients remains unclear.

Objectives:

To investigate the association between muscle relaxant use and the risks of Sjögren’s syndrome and dry eye disease (including keratoconjunctivitis sicca) in patients with AS.

Design:

A population-based retrospective cohort study.

Methods:

This population-based retrospective cohort study identified patients suffering from AS between 2007 and 2017 (N = 26,806,963) in the National Health Insurance Research Database (NHIRD) of Taiwan. The demographic data of two study groups, namely muscle relaxant users and non-users, were collected in the present study. The inverse probability of treatment weighting (IPTW) was used to balance the clinical confounders. Furthermore, we used Kaplan–Meier analyses and Cox proportional hazard regression to analyze the association between cumulative muscle relaxant use and SS or dry eye risk in AS patients.

Results:

After exclusion, 68,970 muscle relaxant users and 31,863 non-users were included in our study. The risks of composite outcome (hazard ratios (HR) 1.37 (1.31–1.43), aHR1.18 (1.13–1.24), and IPTW HR 1.17 (1.13–1.22)), Sjogren’s syndrome (HR 1.63 (1.47–1.79), aHR 1.44 (1.30–1.59), and IPTW HR 1.45 (1.32–1.59)), Dry eye disease (including keratoconjunctivitis sicca) (HR 1.32 (1.26–1.39), aHR 1.12 (1.06–1.17), and IPTW HR 1.11 (1.06–1.16)) were all significantly higher among muscle relaxant users (p < 0.0001). Moreover, higher doses of muscle relaxant use increased higher risks of composite outcome (25< medication possession ratio (MPR), aHR = 1.77 (1.60–1.96)) and dry eye disease (25< MPR, aHR = 1.74 (1.56–1.94)) (p < 0.0001 for trend). However, there was higher risk of Sjogren’s syndrome among the patients who received 5< MPR ⩽10 (aHR = 1.70 (1.49–1.93)). Muscle relaxant users who received ethers chemically related to antihistamines and oral muscle relaxant users had higher risks of composite outcome, Sjogren’s syndrome, and dry eye disease.

Conclusion:

This study demonstrates that cumulative muscle relaxant exposure might increase the risk of dry eye or even SS at a dose-dependent manner.

Introduction

Sicca syndrome, also known as Sjögren’s syndrome (SS), is an autoimmune disease of exocrine glands that may occur as primary or in association with other connective tissue diseases, known as secondary SS. Secondary SS appears more frequently in association with rheumatoid arthritis (RA), systemic lupus erythematosus (SLE), and progressive systemic sclerosis.^ 1 ^ Dry eye disease is one of the most prevalent complications of Sjögren’s syndrome, affecting most patients.^ 2 ^ Ankylosing spondylitis (AS) is a chronic systemic inflammatory rheumatic disease involving the axial skeleton characterized by lower back pain.^ 3 ^ It was suggested that the concomitant occurrence of SS and AS might be more likely seen in women.^ 4 ^ However, in a previous study, an estimated 10% of 70 AS patients suffered from SS. 80% of the study population were men. Of these seven patients, one was anti-Ro-positive, and two were anti-La-positive. It seems that sex and antibody positivity may be associated with secondary SS among AS patients.^ 5 ^ It is regretful that there were only a few wide population-based cohort studies about the coexistence of SS and AS and the reasons for the associated risk.

Although the current official guidelines generally do not explicitly recommend muscle relaxants as a primary or standard therapy for the treatment of axial spondyloarthritis, which includes ankylosing spondylitis.^6,7^ Muscle relaxants, frequently used to treat musculoskeletal conditions, act as a useful adjunct to steroids or non-steroidal anti-inflammatory drugs (NSAIDs) in the treatment of ankylosing spondylitis.^8 –10^ A meta-analysis of muscle relaxants demonstrated that muscle relaxant users were linked to a doubled risk of central nervous system disturbance compared to non-users.^ 11 ^ The risk could increase with longer exposure to muscle relaxants. Furthermore, in a longitudinal cohort trial involving patients with diabetic renal disease, users had a 26% greater chance of developing incident frailty than never-users. The probability further increased if the users were administered higher doses or longer durations of muscle relaxants.^ 12 ^ It has also been confirmed that medications like benzodiazepines, tricyclic antidepressants, and antihypertensive agents may induce sicca syndrome.^ 13 ^

Dry eye disease (DED) is a heterogeneous condition, often drug-induced, and a common ocular manifestation of SS. Furthermore, patients with undiagnosed SS often presented with symptoms of dry eye.^ 14 ^ However, dry eye symptoms are a known adverse effect of muscle relaxants, especially those with anticholinergic actions. Anticholinergic effects can reduce aqueous and mucous secretions from lacrimal grands and goblet cells and can therefore cause or aggravate sicca symptoms.^ 15 ^ This adverse effect is also mentioned in the labels/SmPCs of several authorized muscle relaxants (e.g., baclofen, cyclobenzaprine, orphenadrine, tizanidine).^ 16 ^ On the other hand, some other muscle relaxants have not been reported to cause dry eyes. Therefore, in this present study, the association between muscle relaxants and sicca syndrome among AS patients was explored via a population-based cohort with a population of more than 20 million and a follow-up of 11 years.

Methods

Data sources

The data were obtained from the National Health Insurance Research Database (NHIRD) of Taiwan, maintained by the National Research Agency. To date, more than 99% of Taiwanese citizens are registered in this database. NHIRD contains information on the history of outpatients, inpatients, medical treatments given to inpatients, as well as further drug prescriptions for outpatients. The NHIRD used in this study is held by the Taiwan Ministry of Health and Welfare and under controlled access. The NHIRD approved our application to access these data. The reporting of this study conforms to the STROBE statement.^ 17 ^

Study participants

As illustrated in Figure 1, the NHIRD enrolled a total of 26,806,963 patients between 2007 and 2017. Diseases were defined according to the International Classification of Diseases, Ninth Revision (ICD-9) and Tenth Edition (ICD-10) codes. The participants of the present study were patients diagnosed with AS (ICD-9 code 720.0; ICD-10 code M45) between Jan 2009 and Sep 2017 (n = 115,216). Moreover, we imposed a stringent criterion for recognizing AS by requiring at least two outpatient visits or one hospital admission within 1 year.^18,19^ The index date was defined as +3 months after being newly diagnosed with ankylosing spondylitis. Patients were categorized into two groups according to muscle relaxant use during the baseline period, defined as the 15 months preceding the index date. The 15-month baseline period comprised the 12 months prior to the initial AS diagnosis and the subsequent 3 months after diagnosis, thereby capturing pre-existing medication use as well as early treatment patterns following AS diagnosis, while ensuring temporal separation between exposure assessment and outcome ascertainment. Non-users were defined as patients without any prescription records of muscle relaxants during the baseline period (n = 357,77). Users were defined as patients with at least one prescription for muscle relaxants during the baseline period (n = 79,439).

The exclusion criteria were as (1) Missing demographic non-users (n = 247) and muscle relaxant users (n = 722); (2) Any study before index non-user (n = 3610) and relaxants user (n = 9576); (3) Dead before index non-user (n = 57) and relaxants user (n = 171). Finally, 31,863 non-users and 68,970 muscle relaxant users were included in the study group.

Outcome and covariates

The primary outcome was a composite outcome comprising newly diagnosed Sjögren’s syndrome and DED, including keratoconjunctivitis sicca. Sjögren’s syndrome was identified using ICD-9-CM code 710.2 or ICD-10-CM code M35.0. DED, including keratoconjunctivitis sicca, was defined by ICD-9-CM codes 375.15 (tear film insufficiency) and 370.33 (keratoconjunctivitis sicca), as well as ICD-10-CM codes H04.12x and H16.22x.^20,21^ The comorbidities analyzed in this study were illustrated in Supplemental Table 1.

Muscle relaxant exposure

Muscle relaxants were classified according to the Anatomical Therapeutic Chemical (ATC) classification system. To focus on centrally acting muscle relaxants relevant to chronic musculoskeletal conditions, ATC level 4 classifications within centrally acting agents (ATC code M03B) were selected a priori to examine potential heterogeneity in the risk of Sjögren’s syndrome or DED across different pharmacological mechanisms. The following subgroups were carbamic acid esters (M03BA), oxazole/thiazine derivatives (M03BB), ethers chemically related to antihistamines (M03BC), and other centrally acting agents (M03BX). These subgroup analyses aimed to assess whether agents with anticholinergic properties or distinct central nervous system effects were associated with differential risks of Sjögren’s syndrome or DED. In addition, drug exposure was further stratified by route of administration, categorized as oral or parenteral formulations, to assess whether differences in systemic exposure patterns influenced the observed associations.

Statistical analysis

Baseline covariate balance between exposure groups was assessed using absolute standardized differences (ASDs), with values <0.1 indicating adequate balance. For analyses incorporating inverse probability of treatment weighting (IPTW), ASDs were calculated using weighted samples.

Incidence density rates of Sjögren’s syndrome or DED were calculated as events per 10,000 person-months. Crude incidence rates and corresponding 95% confidence intervals were estimated using Poisson regression models with a log link function and an offset for person-time at risk.

Time-to-event analyses were performed using Cox proportional hazards regression models, which were used to estimate adjusted hazard ratios (aHRs) and IPTW-weighted hazard ratios, along with their 95% confidence intervals. IPTW was implemented using stabilized inverse probability weights to account for baseline confounding.

Kaplan–Meier curves were constructed to illustrate cumulative incidence over follow-up, and group differences were assessed using the log-rank test. Statistical significance was defined as a two-sided p value <0.05. All analyses were performed using SAS software, version 9.4 (SAS Institute Inc., Cary, NC, USA).

The medication possession ratio (MPR) was calculated over a 15-month baseline window (comprising the 12 months preceding and the 3 months following the diagnosis of AS) to serve as a proxy for baseline exposure intensity to muscle relaxants. This metric was defined as the total days’ supply of muscle relaxants dispensed divided by the total duration of the baseline window. To further evaluate the temporal relationship between muscle relaxant exposure and study outcomes, we conducted landmark and time-varying Cox proportional hazards analyses. Exposure status was updated using cumulative MPR assessed over successive landmark periods, including baseline to 1 year, 1–3 years, 3–5 years, and 5–8 years after the index date. For each landmark analysis, only patients who were event-free and uncensored at the beginning of the respective period were included. These models were adjusted for baseline demographic characteristics as well as time-varying comorbidities and concomitant medication.

Results

Demographic characteristics and comorbidities

During the study period, 114,296 AS patients were found to be eligible study participants. Among these AS patients, we further selected muscle relaxant users (n = 34,199) and those who had never been prescribed muscle relaxants (n = 80,097). Following the application of the exclusion criteria outlined in Figure 1, 31,863 non-users and 68,970 muscle relaxant users were included in the study groups (Figure 1). Although the number of men in both study groups was slightly higher (non-users: 62.19%, muscle relaxant 59.67%) than the number of women, the age and gender differences between these two groups were negligible. Most AS patients were co-medicated with NSAIDs (non-users: 83.90%, muscle relaxant 97.49%). Although the prevalence of AS has been demonstrated higher in Hen-San (rural area) in Taiwan in a population survey,^ 22 ^ in our cohort, most of the AS patients were in urban areas (non-users: 62.37%, muscle relaxant 59.09%). There were no significant differences between muscle relaxant users and non-users with regard to the urbanization, unit type of insurance, commodity, and co-medication (Table 1).

Risks of new-onset of Sjogren’s syndrome or dry eye among study groups

After a mean of 50 (muscle relaxant users, 54) months of follow-up, the muscle relaxant users exhibited a trend of higher incidence rate compared with the non-users.

The risks of composite outcome exhibited the crude hazard ratios (HR) 1.37 (1.31–1.43), Adjusted HR1.18 (1.13–1.24), and IPTW-weighted HR 1.17 (1.13–1.22), which was statistically significant compared to non-users (p < 0.0001). In addition, the risks of Sjogren’s syndrome exhibited the crude HR 1.63 (1.47–1.79), Adjusted HR 1.44 (1.30–1.59), and IPTW-weighted HR 1.45 (1.32–1.59)with a significant difference (p < 0.0001; Table 2). Furthermore, the risks of DED (including keratoconjunctivitis sicca) exhibited the crude HR 1.32 (1.26–1.39), Adjusted HR 1.12 (1.06–1.17), and IPTW-weighted HR 1.11 (1.06–1.16) among muscle relaxant users(p < 0.0001).The survival curve (Figures 2 and 3) revealed that the cumulative incidences of composite outcome, Sjogren’s syndrome, and DED (including keratoconjunctivitis sicca) were all higher in the muscle relaxants group than in non-users before and after IPTW (p < 0.0001).

Subgroup analysis for risk of composite outcome, Sjogren’s syndrome or dry eye disease

Consequently, we conducted gender, age, urbanization subgroup and with DMARDs treatment, and with Fibromyalgia analyses for the risks of composite outcome, Sjogren’s syndrome and DED among AS patients between the two study groups. Most subgroups demonstrated higher risks of composite outcome, Sjogren’s syndrome and DED among muscle relaxant users except for the urbanization stratification rural subgroup (composite aHR = 0.77 (0.68–0.87), Sjogren’s syndrome aHR = 0.71 (0.56–0.90), DED 0.81 (0.71–0.92). Moreover, most of the subgroups have significant differences except sex subgroup for Sjogren’s syndrome (Supplemental Tables 3–5).

Dose–response for the risk of Sjogren’s syndrome or dry eye

To explore whether greater baseline exposure coverage was associated with elevated risk, we stratified MPR into four prespecified groups (>0–5%, >5–10%, >10–25%, and >25%). Relative to non-users, higher baseline MPR categories were associated with a graded increase in the risk of composite outcome (aHRs: 1.10, 1.29, 1.27, and 1.77, respectively; p for trend <0.0001; Table 3). The MPR represents prescription dispensing coverage rather than actual pharmacological dose or cumulative duration. Thus, these findings indicate an association between baseline exposure intensity and subsequent risk, rather than a causal, time-varying dose–response relationship during the follow-up period.

Sensitivity analysis for different types of muscle relaxants

To focus on centrally acting muscle relaxants relevant to chronic musculoskeletal conditions, the following subgroups were carbamic acid esters, oxazole/thiazine derivatives, ethers chemically related to antihistamines, and other centrally acting agents. Muscle relaxant users who received ethers chemically related to antihistamines had higher risks of composite outcome (aHR = 1.54 (1.40–1.68)), Sjogren’s syndrome (aHR = 2.72 (2.25 3.30)), and DED (aHR = 1.51 (1.37–1.66); Table 4).

Sensitivity analysis for diverse route of administration of muscle relaxants

Drug exposure was further stratified by route of administration, categorized as oral or parenteral formulations, to assess whether differences in systemic exposure patterns influenced the observed associations. It was indicated that oral muscle relaxant users had higher risks of composite outcome (aHR = 1.18 (1.13–1.24)), Sjogren’s syndrome (aHR = 1.44 (1.30–1.59)), and DED (aHR = 1.12 (1.06–1.17); Table 5).

Landmark and time-varying analysis

In landmark and time-varying Cox regression analyses (Table 6 and Supplemental Table 6), the association between muscle relaxant exposure and study outcomes varied across follow-up periods. Within the first year after the index date, lower MPR categories were not consistently associated with increased risk, and several estimates were below unity, particularly for DED. However, during subsequent follow-up intervals (1–3 years and 3–5 years), higher cumulative MPR categories were generally associated with an increased risk of the composite outcome, Sjögren’s syndrome, and DED. In the longest follow-up period (5–8 years), risk estimates remained elevated in most exposure categories, although confidence intervals widened in some strata, reflecting reduced numbers of patients and events.

Discussion

This is the first study using population-based data to estimate the risk of SS or dry eye in AS patients who take muscle relaxants. Progressive muscle relaxant use was found to be strongly related to an increased risk of SS or dry eye, with the connection becoming stronger with both longer duration of use and larger cumulative doses. Based on our findings, it may be more prudent to limit the use of muscle relaxants in patients with AS, ideally on a short-term basis better within 1 year and at the lowest feasible dose to prevent the risk of SS and DED.

Furthermore, a former population-based association study in the UK demonstrated that age, depression, chronic pain syndrome, the presence of allergy, cataract surgery, rheumatoid arthritis, osteoarthritis, migraine were important risk factors with DED within a female population.^ 23 ^ SS is a common disease with a female to male ratio of at least 1 and an average age at diagnosis of approximately 50 years.^ 24 ^ In our study, female (male vs female aHR 0.380, p < 0.0001) AS patients aged between 30 and 80 have a higher risk of SS or dry eye, among which the age group 50–60 (aHR 1.744, 95% CI 1.567–1.941) and the age group 60–70 (aHR 1.897, 95% CI 1.684–2.137) presented with the most significant difference (p < 0.0001) (Supplemental Table S1). In addition, the use of antihistamines, antidepressants and diuretics were also confirmed to be associated with the risk of dry eye. Benzodiazepines, tricyclic antidepressants, and antihypertensive agents were also frequently associated with dry eye and dry mouth.^ 25 ^ The frequent co-prescription of muscle relaxants with other immune suppressors or system relievers may also increase the incidence of SS. In our study, AS patients co-medicated with hydroxychloroquine (aHR 2.116, 95% CI 1.902–2.353), hypnotic drugs (aHR 1.197, 95% CI 1.126–1.272), and psychotropic drugs (aHR 1.281, 95% CI 1.171–1.403) demonstrated higher risks for SS or dry eye (p < 0.0001) (Supplemental Table S1).

According to a recent large-scale survey in the United States, the rate of prescribing muscle relaxants nearly doubled from 2005 to 2016, especially with high use in older adults. Furthermore, about 67.2% of patients were co-treated with an opioid, which increased the risk of causing severe drug-drug interactions and other adverse effects.^ 26 ^ Muscle relaxants have already been listed as a class of potentially inappropriate medication in the American Geriatrics Society (AGS) Beers Criteria.^ 27 ^

Genetic and environmental factors act on the immune system and cause abnormal generation of autoantibody-producing B cells and autoreactive T cells and the production of pro-inflammatory cytokines in autoimmune diseases. There have been hot discussions on the common mechanism among SS, designated as an autoimmune epithelitis, and AS, termed as axial spondyloarthritis. The NOD-like receptor family pyrin domain containing 3 (NLRP3) inflammasome and downstream molecules, including caspase-1, IL-1β, and IL-18, are overactivated in both AS and SS.^ 28 ^ The Th17/Treg balance also plays an essential role between AS and SS.^ 29 ^ The role of mucosal IgAs may also provide a link between these two groups of diseases (with penetration of the agent through the oropharyngeal mucosa in SS and through the intestinal mucosa in AS).^ 30 ^ Moreover, gut microbiota dysbiosis plays a crucial role in autoimmune disease pathogenesis as well as in bowel-related diseases.^ 31 ^ The development of metagenomics enables the study of the common bacteria involved in AS and SS.^ 32 ^ However, whether the muscle relaxants contribute to these mechanisms for the coexistence of the two disease groups still needs further investigation.

Our present study is noteworthy for several reasons. First, it is the first large-sample population-based cohort study to assess the risk of SS or dry eye in AS patients using muscle relaxants. Secondly, the inverse probability of treatment weighting (IPTW) was used in this cohort study for balancing the distribution of clinical confounders.

Limitations

Limitations of our study are inherent to observational studies and studies using administrative data. For instance, detailed information about AS disease activity is lacking. Consequently, it cannot be determined if the observed association is driven by the medication itself or by the underlying severity of the rheumatic disease. Moreover, the observed “dose–response” relationship based on MPR may reflect confounding by severity. Furthermore, the lack of some individual factors such as smoking status, blood pressure, body mass index, patients’ lifestyle, and environmental factors may potentially result in unmeasured confounding. Moreover, patients prescribed muscle relaxants may have more severe diseases or different healthcare-seeking behavior, necessitating more frequent physician visits and closer monitoring than non-users inducing surveillance bias. Furthermore, the anticholinergic side effects of muscle relaxants (e.g., xerostomia) may prompt diagnostic evaluations for sicca that would not otherwise occur in non-users. In addition, patients using over-the-counter medications (e.g., antihistamines), which possess strong anticholinergic properties, could lead to unmeasured confounders. Corneal melt is the most serious side effect of NSAIDs, which almost all AS patients (93%) received NSAID therapy.^ 33 ^ This could cause potential confounding. As the subgroup analysis revealed that patents with DMARDs treatment had a higher risk of SS and dry eye, some DMARDs may have complex drug-drug interactions with the muscle relaxants. A significant limitation of this study is the potential for protopathic bias (reverse causality). The diagnosis of Sjögren’s disease is frequently delayed by several years after symptom onset. Musculoskeletal pain and fatigue are common extra-glandular manifestations that may precede the formal diagnosis of SS. Therefore, the association observed may reflect that patients with undiagnosed, prodromal SS required muscle relaxants for pain management prior to their diagnosis. In addition, our reliance on ICD codes for outcome definition, which were not verified using the American College of Rheumatology (ACR)/EULAR classification criteria, may lead to misclassification, as drug-induced sicca symptoms could potentially be miscoded as Sjögren’s syndrome without objective autoimmune confirmation. Although MPR was used to assess dose–response, dispensing data does not guarantee ingestion and that the fixed baseline window may introduce exposure misclassification due to the intermittent nature of muscle relaxant use. This study was conducted using the NHIRD in Taiwan, a population that is predominantly Han Chinese. Therefore, the results regarding the association between muscle relaxants and SS risk should be extrapolated to non-Asian populations or different healthcare settings with caution. Finally, the weak associations (HRs for Sjögren’s syndrome most <2.0) derived from administrative databases without lab validation should be interpreted with extreme caution regarding causality.

Conclusion

In this population-based cohort study, we demonstrated that cumulative exposure to muscle relaxants might increase the risk of SS or dry eye in patients with AS. Longer durations and higher cumulative doses of muscle relaxants significantly elevated the risk of SS or dry eye in AS patients. Hence, it may be more judicious clinically for AS patients to have a short-term use and the lowest possible dosage of muscle relaxants.

Supplemental Material