Abstract
Objective. Ankylosing spondylitis (AS) is a common inflammatory rheumatic disease. A higher prevalence of psychiatric comorbidities, including depressive disorder, has been proven in patients with AS. However, a clear temporal causal relationship between AS and psychiatric disorders has not been well established. We performed a nationwide population-based retrospective cohort study to analyze the relationship between AS and the subsequent development of psychiatric disorders, including schizophrenia, bipolar disorder, depressive disorders, anxiety disorders, and sleep disorders.
Methods. We identified subjects who were newly diagnosed with AS between January 1, 2000, and December 31, 2008, in the Taiwan National Health Insurance (NHI) Research Database. A comparison cohort was constructed of patients without AS who were matched according to age and sex. All patients with AS and control patients were observed until diagnosed with psychiatric disorders, or until death or withdrawal from the NHI system, or until December 31, 2009.
Results. The AS cohort consisted of 2331 patients and the comparison cohort consisted of 9324 matched control patients without AS. The adjusted HR for depressive disorders, anxiety disorders, and sleep disorders in subjects with AS were higher than those of the controls during followup (HR 1.718, 95% CI 1.303–2.265; HR 1.848, 95% CI 1.369–2.494; and HR 1.494, 95% CI 1.031–2.162, respectively).
Conclusion. AS might increase the risk of a subsequent newly diagnosed depressive disorder, anxiety disorder, or sleep disorder, but not schizophrenia or bipolar disorder. These observations highlight the need for psychiatric evaluation and intervention for patients with AS.
Ankylosing spondylitis (AS) is a common inflammatory rheumatic disease that mainly involves the axial skeleton, causing frequent pain and stiffness in the lower back and backside, which can lead to functional impairments and decreased quality of life1. A recent report analyzing 36 eligible studies demonstrated that the mean prevalence of AS per 10,000 people was 23.8 in Europe, 16.7 in Asia, 31.9 in North America, 10.2 in Latin America, and 7.4 in Africa2. AS mostly affects young people and about 80% of patients develop the first symptoms when aged younger than 30 years3. Men are affected more often than women, by a ratio of about 2 to 13. In addition to the back pain and stiffness that characterize AS, anterior uveitis, psoriasis, diabetes mellitus (DM), cardiovascular, and cerebrovascular diseases also commonly coexist in subjects with AS4,5,6,7,8. Chronic diseases and pain are usually accompanied by psychological abnormalities. Depression, anxiety, and sleep disorders have been reported in a significant number of patients with AS.
In 1993, Barlow, et al reported that about one-third of patients with AS experience a high level of depressive symptoms and that women report more depression than men9. They also showed that pain was a major determinant of depression for women, but was of lesser importance for men. In a nationwide population-based study of comorbidity profiles among patients with AS in Taiwan, it was found that patients with AS had higher OR for mental illness, including depression and psychoses, than the comparison group7. Comorbid depression can lead to work disability, unemployment10, disturbed sexual relationships11, decreased quality of life12, and fatigue symptoms13 in subjects with AS.
In a Swedish national cohort study with a 31-year followup designed to analyze the subsequent risk of hospitalization for psychiatric disorders in patients with rheumatic diseases, it was found that those with AS and systemic lupus erythematosus (SLE) had a higher risk of subsequent psychiatric disorders than did patients with rheumatoid arthritis (RA)14. Patients with AS were at increased risk of mood disorders (including depressive disorders and bipolar disorders) and anxiety disorders. In addition, it is noteworthy that among the patients with AS, the standardized incidence ratio was highest for anxiety disorders than any other subtype of psychiatric disorder, including mood disorders. Moreover, the risk of developing anxiety disorders among the patients with AS was higher than those with SLE or RA14.
In a study of 80 patients with AS and 52 age- and sex-matched controls, the patients with AS had significantly less favorable scores in the subjective sleep quality and habitual sleep efficiency domains, and the total Pittsburgh Sleep Quality Index score15. Poorer sleep quality is statistically associated with pain, disease activity, depression, decreased quality of life, and increased limits on mobility. The higher prevalence of sleep disturbances in patients with AS has been confirmed by another study of Chinese subjects16.
Although the above-mentioned research has provided insight into associations between AS and comorbid psychiatric disorders, most of those studies were cross-sectional and lacked a longitudinal perspective. In addition, psychiatric disorders in these studies were often evaluated using rating scales, such as the Mini International Neuropsychiatric Interview, the Beck Depression Inventory, or the Hamilton Depression Rating Scale, rather than a diagnosis by a psychiatrist. Further, the small sample sizes of most of these studies prevent generalization.
To date, there has been a lack of national data, with very few longitudinal studies regarding the risk of psychiatric disorders in patients with AS. We hypothesized that patients with AS may have a higher risk of developing psychiatric disorders, and therefore conducted a nationwide population-based retrospective cohort study to investigate a possible link between these 2 illnesses.
MATERIALS AND METHODS
Data sources
The National Health Insurance (NHI) program is a mandatory health insurance program in Taiwan that was instituted in 1995 and offers comprehensive medical care coverage, including outpatient, inpatient, emergency, and traditional Chinese medicine to all the residents, with a coverage rate of up to 98%17. The NHI Research Database (NHIRD) contains comprehensive information regarding clinical visits, including prescription details and diagnostic codes based on the A-code and the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM). The A-code is a disease classification system developed especially for filing medical claims, which was mainly used for ambulatory care before 2000 in Taiwan. The A-code is a much simplified version of the ICD-9-CM codes. To maintain consistency between different claims records and to truly reflect the distribution of different diseases, the NHI program has switched to the ICD-9-CM codes since 2000. In our study, the A-code was mainly used to exclude patients who were diagnosed with AS between 1996 and 2000, and also to exclude patients who were diagnosed with psychiatric disorders before they were diagnosed with AS. The NHIRD is managed by the National Health Research Institutes (NHRI), and is confidentially maintained according to the directives of the Bureau of NHI. The data source for our study was the Longitudinal Health Insurance Database 2005 (LHID 2005), a dataset released by the NHRI containing all the original claims data for 1 million randomly selected beneficiaries in the 2005 Registry of Beneficiaries.
Ethics statement
The Institutional Review Board of Taipei Veterans General Hospital approved this study. Written consent was not obtained from the study patients because the NHI dataset consists of deidentified secondary data for research purposes and the Institutional Review Board of Taipei Veterans General Hospital has issued a formal written waiver of the need for consent.
Study population
Using data extracted from the LHID 2005, we conducted a retrospective cohort study of patients who were newly diagnosed with AS (A-code: A433, ICD-9-CM code: 720) between January 1, 2000, and December 31, 2008. To ensure diagnostic validity and patient homogeneity in the study group, we selected only patients who were diagnosed with AS by an orthopedist or rheumatologist and had at least 2 consensus AS diagnoses7. We excluded patients who were diagnosed with AS between January 1, 1996, and December 31, 1999. We also excluded patients who were diagnosed with psychiatric disorders (A-codes: A210–A219, ICD-9-CM codes: 290–319) before they were diagnosed with AS. For each patient with AS included in the final cohort, 4 age- and sex-matched control patients without psychiatric disorders were randomly selected from the LHID 2005 dataset. All patients with AS and control patients were observed until a psychiatrist diagnosed schizophrenia (ICD-9-CM code: 295), depressive disorder (ICD-9-CM codes: 296.2, 296.3, 300.4, and 311), bipolar disorder (ICD-9-CM codes: 296.0, 296.1, 296.4, 296.5, 296.6, 296.7, 296.8, 296.80, and 296.89), anxiety disorder (ICD-9-CM codes: 300.0, 300.2, 300.3, 308.3, and 309.81), or sleep disorder (ICD-9-CM codes: 780.5, 307.4, excluding 780.51, 780.53, 780.57), or until death or withdrawal from the NHI system, or until December 31, 2009. The primary clinical outcomes assessed were psychiatrist-diagnosed schizophrenia, depressive disorder, bipolar disorder, anxiety disorder, or sleep disorder.
Statistical analysis
The incidence of newly diagnosed schizophrenia, depressive disorder, bipolar disorder, anxiety disorder, or sleep disorder in the patients with AS and control patients was calculated, and independent Student t tests and chi-square tests were conducted to examine differences in characteristics between the patients with AS and control patients. To investigate potential surveillance bias, subgroups were stratified according to the duration since AS diagnosis. A Cox proportional hazards regression model was constructed to calculate the HR for schizophrenia, depressive disorders, bipolar disorders, anxiety disorders, and sleep disorders of the AS cohort and control cohort. Further, we performed a Cox proportional hazards regression model to identify variables that predicted psychiatric disorders in the patients with AS. Control variables were included as covariates in the univariate model: age; sex; common comorbidities including hypertension, DM, dyslipidemia, coronary artery disease, congestive heart failure, chronic pulmonary disease, and malignancy; urbanization; and monthly income. Factors that demonstrated a moderately significant statistical relationship in the univariate analysis (p < 0.1) were entered by forward selection in a multivariate Cox proportional hazards regression model18. In addition, we tested the proportionality of hazards using Scaled Schoenfeld Residuals.
The SAS statistical software for Windows, Version 9.3 (SAS Institute) was used for data extraction, computation, linkage, processing, and sampling. All other statistical analyses were performed using the SPSS statistical software for Windows, Version 20 (IBM). A p value of < 0.05 was considered statistically significant.
RESULTS
Our study sample consisted of 2331 patients with AS (64.9% men) and 9324 control patients without AS. Comparisons of the demographic and clinical variables between the patients with AS and control patients are presented in Table 1. The median age of the patients was 36.50 years (interquartile range 27.25–48.18 yrs) and the median followup durations for the AS and control cohorts were 5.99 and 5.95 years, respectively. A higher percentage of patients with AS was observed in the 20–39 years age group.
Characteristics of patients with ankylosing spondylitis and control subjects. Values are n (%) or median (interquartile range) unless otherwise specified.
During the followup period, 190 (8.15%) of the patients with AS and 467 (5.01%) of the control patients were diagnosed with psychiatric disorders (p < 0.001). The most common psychiatric disorders occurring subsequently in patients diagnosed with AS were depressive disorder in 73 patients (3.1%), anxiety disorder in 64 patients (2.7%), and sleep disorder in 39 patients (1.7%). Overall, significantly higher incidences of depressive disorder (p < 0.001) and anxiety disorder (p < 0.001) were observed in the patients with AS than in the control patients.
In addition, a Cox proportional hazards regression analysis was conducted to calculate the HR of the newly diagnosed psychiatric disorders in patients with AS compared with matched controls (Table 2). The results indicated that patients with AS exhibited a markedly higher risk of developing a subsequent depressive disorder (adjusted HR 1.718, 95% CI 1.303–2.265), anxiety disorder (adjusted HR 1.848, 95% CI 1.369–2.494), or sleep disorder (adjusted HR 1.494, 95% CI 1.031–2.162) after adjusting for age, sex, comorbidities, urbanization, and monthly income.
HR of time until psychiatric disorders between patients with ankylosing spondylitis and control subjects during a 10-year followup period.
Further, a subanalysis based on the duration of followup revealed that the risk of a newly diagnosed depressive disorder became significantly elevated not only within the first year, but also beyond the fifth year following an AS diagnosis. On the other hand, the risk of a newly diagnosed anxiety disorder only increased beyond the first year following an AS diagnosis. The results of the subanalysis are summarized in Table 3.
Number of newly diagnosed depressive, anxiety, and sleep disorders between patients with AS and control subjects, stratified by followup duration.
Finally, we performed a Cox proportional hazards regression model to identify variables that predicted psychiatric disorders in the patients with AS. The results showed that age, sex, and comorbidities were not significant prognostic factors for psychiatric disorders in the patients with AS (data not shown).
The results of testing the proportionality of hazards showed that there was no evidence that the proportional hazards assumption had been violated in each of the Cox regression models in our study.
DISCUSSION
To our knowledge, this is the first large nationwide population-based cohort study to compare the HR for psychiatric disorders in patients with and without AS. The main finding of our study is that Cox proportional hazards analysis, after adjusting for age, sex, comorbidities, urbanization, and monthly income, yielded an adjusted HR for depressive disorders that was 1.718 times greater for patients with AS than for the comparison cohort (Table 2). We also found that subjects with AS were more likely to develop anxiety disorders and sleep disorders than controls during the followup. Further, our analysis showed that hypertension, DM, dyslipidemia, coronary artery disease, and chronic pulmonary disease were more prevalent in patients with AS than in patients without AS (Table 1), which is consistent with the results of previous studies and strengthens the reliability of our findings4,5,6,7,8.
In our study, the ratio of men to women for patients with AS is roughly 2 to 1 (1512 vs 819), which is in line with a previous report stating that men are affected twice as often as women3. Because women have a lower prevalence of AS and a higher prevalence of both depressive and anxiety disorders, researchers may falsely evaluate the risk of developing depressive and anxiety disorders by ignoring the sex of patients with AS. Evidence has shown that female sex is associated with depression in AS9 and the effect of sex on the association between anxiety disorders and AS is not clear. In our study, we have investigated the effect of sex on the risk of depressive and anxiety disorders in patients with AS and we found no increased risk of developing depressive and anxiety disorders among female patients with AS (Supplementary Table 1 and Supplementary Table 2 are available from the authors on request).
Several studies have confirmed that depression is the most common comorbidity in patients with AS7,9,14. Our study results indicate that AS might be a risk factor for subsequent depressive disorders. There are several possible explanations for the increased risk of depressive disorders in patients with AS. First, studies have indicated that depressive disorders among patients with AS might be caused by psychological stresses associated with AS. In a study of 200 patients with AS, a significant association was found between AS disease activity and depression severity19. In some cases, it is likely that depression is mediated by a loss of work productivity and the consequent socioeconomic effect of AS20. Second, AS is a chronic inflammatory disease of unknown cause. Studies have indicated that depression is associated with a chronic, low-grade inflammatory response and activation of cell-mediated immunity21. Therefore, depression may be attributable to systemic inflammation and the proinflammatory cytokine milieu of AS. In a study, it was found that infliximab, a tumor necrosis factor-α antagonist drug prescribed to patients with AS, can improve AS symptoms as well as the depression associated with AS22. Third, AS is a complex disease involving multiple risk factors, both genetic and environmental. HLA-B27 is known as a major AS-susceptibility gene and the identification of the genes for IL23R, ERAP1, and the interleukin-1α region in AS pathogenesis has led to a number of replication studies23. Depression and AS may share a common genetic background that involves the inflammatory response. Regardless of the initiating factors, the biological, functional, and socioeconomic consequences of AS can also contribute and perpetuate depressive symptoms in people with AS.
Some studies have assessed the relationship between AS and anxiety24,25. In these 2 studies, anxiety was evaluated by the Hospital Anxiety and Depression Scale, including the depression and anxiety subscales, and both studies showed that anxiety was more frequent than depression among the patients with AS (25% clinical anxiety vs 15% clinical depression and 60% anxiety vs 55.5% depression for each study, respectively). After stratification by followup duration, the result of our study revealed that a longer followup period produced a higher risk of developing subsequent anxiety disorders among the patients with AS (Table 3) and this finding is consistent with the inflammatory theory, which is one of the possible biological mechanisms that could explain the increased risk of anxiety disorders in patients with AS. That is, the peripheral chronic inflammation that causes AS may take a longer time26 to induce a similar level of chronic inflammation in the central nervous system27, which could result in anxiety disorders28,29. To clarify the involvement of depression in the association between AS and subsequent anxiety disorders, we used the Cox regression method to determine whether the association between anxiety disorders and AS is mediated by depression. In our further analysis, only the depressive disorders diagnosed during the same visit or prior to the date of anxiety-disorder diagnoses were included as a new covariate in the Cox regression model. Other control variables, such as age, sex, common comorbidities, urbanization, and monthly income, were still used in the analysis. The results showed that patients with AS still exhibited a higher risk of subsequent anxiety disorders after adjusting for the possible effects of depression (crude HR 1.86, 95% CI 1.39–2.51; adjusted HR 1.73, 95% CI 1.28–2.34).
In our present study, we conducted a subgroup analysis stratified according to the duration between the diagnosis of AS and newly diagnosed psychiatric disorders (Table 3). The results indicated that incident depressive disorders and anxiety disorder were increased not only within the first year, but also beyond the first year of a diagnosis of AS. Patients with AS are likely to exhibit a higher frequency of outpatient visits than those without AS, leading to less chance of psychiatric disorders being underdiagnosed. It is also possible that depressive and anxiety disorders are a stress reaction to the new diagnosis of AS. Our result suggests that this is less likely because the higher risk of these psychiatric disorders was not restricted to the first year after a diagnosis of AS.
Sleep disturbance is often reported by patients with AS, with awakenings attributable to inflammatory pain. In a study of sleep disturbance assessed using the fourth item of the Hamilton Anxiety Scale, sleep disturbance was found in 64.5% of patients with AS30. In our study, patients with AS were found to have a 1.494 times greater risk of developing a sleep disorder than those in the comparison cohort. However, only 1.7% of patients with AS were diagnosed with a sleep disorder. The wide range in the prevalence of sleep disorders in studies of AS is likely because of the different methods used for measuring sleep disorders (rating scale vs ICD-9-CM diagnosis by a psychiatrist).
Evidence from a national study in Sweden has shown that hospitalization because of schizophrenia is not more highly prevalent among patients with AS than the general population14. In a nationwide population-based study of comorbidity profiles among patients with AS in Taiwan, it was found that patients with AS had a higher prevalence of comorbid psychotic disorders than patients without AS7. However, in our nationwide population-based study in Taiwan, no significantly increased risk of subsequent schizophrenia was observed after a diagnosis of AS. The discrepancy between 2 nationwide population-based studies in Taiwan may come from differences in study design (retrospective cohort study vs cross-sectional study) and psychiatric diagnosis [ICD-9-CM codes for schizophrenia (295.x) vs the Elixhauser Comorbidity Index for psychosis (295.x–298.x, 299.1)].
Exploring the association between AS and psychiatric disorders may enhance our current understanding of the mechanisms underlying mental illness, which may be helpful for providing more comprehensive care to patients with AS experiencing comorbid or subsequent psychiatric disorders.
The strengths of our study are the large sample size and the valid diagnosis of AS and psychiatric disorders by specialists in the claims database. However, certain limitations to our findings should be considered. First, the NHIRD does not provide detailed information on patients, such as tobacco use, alcohol consumption, and family history of psychiatric illness. They are all major risk factors for psychiatric disorders. Body mass index data are also not provided. Thus, we were unable to control for these potentially confounding factors. Second, the diagnoses of AS are identified using the ICD-9-CM codes from the database, and its prevalence may be underestimated because only subjects seeking a medical evaluation can be identified; however, this would most likely result in an underestimate of the association between AS and psychiatric disorders. Third, although the data we obtained on AS and the diagnoses of psychiatric disorders were highly reliable, the diagnoses in NHI claims are primarily for administrative billing and do not undergo verification for scientific purposes. Fourth, the classification of anxiety disorders according to the Diagnostic and Statistical Manual of Mental Disorders, 4th edition, text revision was used in our study to define anxiety disorders, and includes generalized anxiety disorder, social anxiety disorder, specific phobia, panic disorder with and without agoraphobia, obsessive compulsive disorder, posttraumatic stress disorder, anxiety secondary to medical condition, acute stress disorder, and substance-induced anxiety disorder. Compared with depressive disorders, the class of anxiety disorders is made up of a more heterogeneous set of specific disorders, and entirely different patterns of association may exist between AS and each of the specific anxiety disorders31. Therefore, the observed association between AS and anxiety disorders in our study may be influenced by a lack of consideration for the competing and canceling effects caused by aggregating a wide range of anxiety disorders. Fifth, prevalence of psychiatric disorders in our study may be underestimated because only subjects seeking a medical evaluation can be identified. Compared with a previous study investigating the prevalence rates of mental disorders of an adult general population sample32, the incidence rates of depressive disorders (2.14%), anxiety disorders (1.74%), and sleep disorders (1.34%) were lower. Therefore, the cases ascertained in our study may represent only a small fraction of all cases of psychiatric disorders. Last, ascertainment bias could not be excluded entirely by our study designs, and findings of our study may be explained by an ascertainment bias.
Our nationwide population-based retrospective cohort study provides further evidence for an increased risk of depressive, anxiety, and sleep disorders in AS subjects. These are treatable conditions and have a great effect on the life quality of patients with AS. Clinicians should be alert to the possibility of depressive, anxiety, and sleep disorders in patients with AS.
Acknowledgment
We thank Emily Ting for English editing.
- Accepted for publication November 24, 2015.
REFERENCES
In this issue
