A relatively common occurrence of spontaneous subarachnoid hemorrhage (SAH) in patients with systemic lupus erythematosus (SLE) has been noted; however, the subsequent studies were conflicting. This nationwide population-based study aimed to evaluate the risk of SAH in patients with SLE.
We identified 16,967 SLE patients from the Taiwan National Health Insurance (NHI) database between 2000 and 2006, and compared the incidence rate of SAH with 16,967 randomly selected age- and sex-matched non-SLE subjects. A Cox multivariable proportional hazards model was used to evaluate the risk factors of SAH in the SLE cohort.
The SLE cohort had a higher risk of SAH, with an incidence rate ratio of 4.84 (P < 0.001). Despite a younger age, the mortality rate after SAH was significantly higher in the SLE cohort compared to all of the non-SLE SAH patients identified from the 1 million NHI beneficiaries (60.0% versus 38.9%; P = 0.007). Age (hazard ratio [HR] 1.03, 95% confidence interval [95% CI] 1.01–1.05), platelet transfusion (HR 2.75, 95% CI 1.46–5.17), red blood cell transfusion (HR 7.11, 95% CI 2.81–17.97), and a mean daily steroid dose >10 mg of prednisolone or equivalent (HR 4.36, 95% CI 2.19–8.68) were independent risk factors for the new onset of SAH.
This study demonstrated that SAH is a rare but associated complication of SLE with a high mortality rate. Other than age, higher mean daily steroid use and a history of platelet or red blood cell transfusion were associated with the occurrence of SAH in patients with SLE.
Systemic lupus erythematosus (SLE) is a systemic autoimmune disease affecting mainly women of childbearing age. The clinical manifestations spare no particular organ, but vary between individuals (1). Neuropsychiatric events, one of the major complications of SLE, occur in 30–40% of patients and warrant aggressive management (2). Among various neuropsychiatric manifestations, cerebrovascular diseases account for considerable morbidity and mortality (3). In contrast to ischemic stroke, where the increased risk and associated mechanisms in SLE patients are well studied, subarachnoid hemorrhage (SAH) is much less well understood (3–8). Spontaneous SAH due to an aneurysm rupture usually occurs between ages 40 and 60 years, with a slightly higher incidence in women. The major clinical symptom is the sudden onset of a severe headache, i.e., a “thunderclap headache,” in 97% of patients. The mortality and morbidity rate of SAH is high (9, 10). Due to the extremely low incidence of SAH, studying the risk of SAH in patients with SLE is difficult and the results are easily influenced by selection bias. Two studies reported a relatively common occurrence of SAH in SLE patients (3, 5); several case reports echoed these and implied that lupus-related intracranial vasculitis and aneurysm formation might be the causes (11–15). However, the only case–control study that has been undertaken unexpectedly reported a lower risk of SAH in patients with SLE (4).
The National Health Insurance (NHI) data set in Taiwan includes comprehensive medical care data of a nationwide population and was released for research purposes in 2000. The NHI data set has been successfully used to analyze the risk of stroke in different diseases, including SLE (16). Furthermore, all patients with SLE are registered in the NHI database of catastrophic illness. Therefore, this nationwide database is a good tool to analyze rare complications of SLE without a selection bias. We designed this study to explore the risk of SAH in patients with SLE and to determine related risk factors.
Significance & Innovations
Studies investigating the risk of subarachnoid hemorrhage (SAH) in patients with systemic lupus erythematosus (SLE) are limited and conflicting, mainly due to the extremely low incidence rate of SAH. Analyzing the nationwide SLE cohort and randomly selected age- and sex-matched non-SLE controls adapted from the Taiwan National Health Insurance database provided the chance to evaluate the risk of such rare complications without selection bias.
Our results proved that SAH is a rare but associated complication of SLE with a high mortality rate. Other than older age, a higher mean daily steroid dose and a history of platelet or red blood cell transfusion were risk factors for the occurrence of SAH in SLE patients.
SUBJECTS AND METHODS
Taiwan began the NHI program in 1995. This mandatory universal health insurance program provides comprehensive health care, including outpatient, inpatient, emergency, dental, and traditional Chinese medicine services, for all citizens in Taiwan, with a coverage rate of ∼98% (17). This comprehensive health care data set includes enrollment files, a registry of patients with catastrophic illness, and details of inpatient and ambulatory care orders, including all of the examinations and prescribed medications. For research purposes, the National Health Research Institute publicly released the NHI database. All information that may potentially identify any individual patient has been managed. The NHI data set consists of deidentified secondary data for research purposes; therefore, the study was exempted from a full review by the Institutional Review Board of Taipei Veterans General Hospital.
Patients with certain severe diseases, including SLE, were enrolled in the registry of catastrophic illnesses through the NHI program. The certification of catastrophic illness exempts these patients from related medical payments. We conducted a retrospective cohort study from January 1, 2000 to December 31, 2006. Patients with SLE catastrophic illness certification (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] code 710.0) in the registry for catastrophic illnesses were enrolled. Those who had SAH prior to the diagnosis of SLE on or before January 1, 2000, who developed SAH within 30 days of SLE diagnosis, or who were followed up for <30 days were excluded from the study. Comorbidities and medications data of the enrolled patients were collected for analyses. Medication usage was defined as any prescription for ≥28 days before the end point, with the exception that corticosteroids were defined as an equivalent dose of prednisolone >10 mg daily.
Subjects without SLE were used to populate a matched control cohort and were randomly selected from 1 million NHI beneficiaries of a total population of 21,400,826 enrollees throughout Taiwan in 2000. Each SLE patient was age and sex matched to 1 non-SLE subject within the same observational period. Those who had SAH before January 1, 2000 or who were followed up for <30 days were excluded from the study. Otherwise, we also identified all new SAH events from the 1 million NHI beneficiaries without SLE in the study period in order to compare the outcomes to the SLE cohort with SAH.
Potential risk factors for SAH in SLE patients.
In addition to the traditional risk factors for SAH, we also identified potential risk factors that were relevant to patients with SLE and might be associated with an increased risk of SAH, including 1) a bleeding tendency possibly due to lupus hematologic involvement, 2) aneurysm formation resulting from vasculitis or a higher risk of atherosclerosis in lupus patients, and 3) the other unidentified factors in the SLE cohort such as immunosuppressants. Therefore, we included immunosuppressants, platelet or red blood cell transfusion therapy, major bleeding history (upper gastrointestinal bleeding), comorbidities related to atherosclerosis, and “surrogates” for lupus activity. The comorbidities were identified by the ICD-9-CM codes (see Supplementary Table 1, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.21846/abstract) in this study. On account of lacking the clinical information needed to calculate lupus disease activity, we used a higher mean daily steroid dose and aggressive treatment for SLE as surrogates in this study, including cyclophosphamide pulse therapy and plasmapheresis. In this study, a higher steroid dosage was defined as >10 mg of prednisolone or equivalent per day based on the cutoff value of steroid usage in the upper one-third of patients with SLE.
The main dependent variable was SAH incidence. SAH events were identified according to the discharge diagnosis (ICD-9-CM code 430). In order to ensure diagnostic homogeneity, those without brain images (e.g., computed tomography [CT] or magnetic resonance imaging [MRI]) during or within 1 month before the index hospitalization were excluded. The 2 cohorts were followed until the development of SAH, death, or the end of the study in 2006. Cumulative incidence rates were analyzed using the Kaplan-Meier analysis and a Cox proportional hazards model was employed for the risk factors. Of all of the factors, antiplatelet agents and warfarin were used as time-dependent covariates in the Cox regression model to evaluate their bleeding complication during usage. The extraction and computation of data were performed using the Perl programming language, version 5.12.2. Microsoft SQL Server 2008 was used for data linkage, processing, and sampling. All statistical analyses were performed using SPSS statistical software for Windows, version 17.0. A P value less than 0.05 was considered to be statistically significant.
SAH in the general population.
Of the 1 million NHI beneficiaries, a total of 642 patients developed SAH within the study period, with a slightly higher incidence for women (n = 342 [53.3%]). The mean age at onset was 56.9 years. The mortality rate at the index admission was 38.9% (250 of 642 patients). All of these results were consistent with prior epidemiologic data published by the Department of Health, Executive Yuan, Taiwan in 1993 (18).
Risk of SAH in the SLE and matched cohorts.
During the study period, a total of 17,404 SLE patients were identified. We excluded 359 patients with a followup duration of <30 days and 78 patients with a history of traumatic SAH or SAH prior to the enrollment date. In total, 16,967 SLE patients were enrolled. An equal number of age- and sex-matched non-SLE controls with the same exclusion criteria were randomly selected and studied (Table 1). The mean age of the patients was 35.5 years and the majority (89.5%) was female. The mean followup period for the SLE and matched cohorts was 5.20 and 5.37 years, respectively. Despite the significantly shorter duration of followup, the SLE cohort had significantly more events (45 versus 9; P < 0.001) and a higher incidence rate of SAH (49.4 versus 10.2 per 100,000 person-years) than the control cohort, with an incidence rate ratio (IRR) of 4.84 (P < 0.001). Figure 1 shows the Kaplan-Meier analysis, which revealed a significantly higher cumulative incidence of SAH in the SLE patients compared to the matched control (log rank test P < 0.001). The age at SAH was significantly lower in the SLE group (44.5 years versus 57.7 years; P = 0.015). Of the 45 SAH events in the SLE cohort, 27 (60.0%) of the patients died at the index hospitalization. We compared the mortality rate of the 45 SLE patients with SAH with the 642 SAH patients from the non-SLE general population. Despite the younger age, the mortality rate after SAH was significantly higher in the SLE cohort (60.0% versus 38.9%; Fisher's exact test P = 0.007).
Table 1. Baseline characteristics of patients with SLE and the age- and sex-matched cohort*
SLE (n = 16,967)
Controls (n = 16,967)
Values are the number (percentage) unless otherwise indicated. SLE = systemic lupus erythematosus; SAH = subarachnoid hemorrhage.
Age, mean ± SD years
35.5 ± 15.0
35.5 ± 15.0
30 and <50
Followup, mean ± SD years
5.20 ± 2.2
5.37 ± 2.2
SAH events, no.
SAH incidence (per 100,000 person-years)
Mortality after SAH
Risk factors of SAH in SLE patients.
A Cox univariate proportional hazards analysis found an increased risk of SAH in SLE patients with one of the following characteristics: older age, hypertension, use of statins, use of clopidogrel, use of aspirin, use of cyclophosphamide, history of platelet transfusion, history of red blood cell transfusion, history of plasmapheresis, and/or a mean daily prednisolone dose >10 mg (Table 2). Of these, only age (hazard ratio [HR] 1.03, 95% confidence interval [95% CI] 1.01–1.05; P = 0.002), platelet transfusion (HR 2.75, 95% CI 1.46–5.17; P = 0.002), red blood cell transfusion (HR 7.11, 95% CI 2.81–17.97; P < 0.001), and a mean daily prednisolone dose >10 mg (HR 4.36, 95% CI 2.19–8.68; P < 0.001) were independent risk factors for the onset of SAH by using a Cox multivariable proportional hazards analysis.
Table 2. Cox regression analyses: variables related to subarachnoid hemorrhage in patients with systemic lupus erythematosus*
HR (95% CI)
HR (95% CI)
All factors with P < 0.05 in univariate analyses were selected for Cox multivariable forward stepwise analysis. HR = hazard ratio; 95% CI = 95% confidence interval; NS = not significant.
The steroid dose was the mean daily prednisolone equivalent.
This nationwide population-based study revealed a significantly higher risk of SAH in 16,967 Taiwanese SLE patients, with an IRR of 4.84, compared to age- and sex-matched non-SLE controls. After SAH events, the mortality rate (60.0%) was much higher in SLE patients than in the non-SLE general population (38.9%). The major risk factors of SAH in patients with SLE included older age, higher mean daily steroid use, and previous transfusion of red blood cells or platelets.
In the NHI system, patients with the catastrophic illness certification for SLE can be exempted from related medical expenses. The verification, however, requires fulfillment of the 1997 American College of Rheumatology (ACR) classification criteria for SLE (19), as supported by medical records and examination reports. These features make the SLE diagnosis exhaustive and reliable in the nationwide population and suitable for studying rare complications of SLE without a selection bias. On the other hand, the diagnosis of SAH in our study was made after the image studies. The symptoms and complications of SAH are too severe to be missed, especially in Taiwan, where the availability and utilization of image studies, such as CT and MRI, are very high (20). However, since we could not access the results of the imaging and the chart records, the exact validity of the diagnosis cannot be provided. Furthermore, we confirmed the epidemiologic data of image-confirmed SAH in the 1 million beneficiaries and found the results to be reliable, regardless of the aspects of incidence, female to male ratio, age, and mortality rate, compared to the data published by the Department of Health, Executive Yuan, Taiwan in 1993 (18).
Previous studies reported conflicting results (3–5). Two Japanese studies reported a higher occurrence of SAH in patients with SLE (3, 5). However, these 2 studies were at a disadvantage because of a lack of a control group, a small sample size, and a single center. Other than the inevitable selection bias of the single center–based data, there was a chronological bias worthy to be mentioned. These 2 studies enrolled patients from 1972–1989 and from 1978–1998, respectively. The ACR classification criteria for SLE were revised twice (in 1982 and 1997) (19, 21); thus, there might be more heterogeneity in the SLE diagnosis. Most important of all, during the long study period, there were great advances in the treatment of SLE. In contrast, one US study reported opposite findings. This large-scale, national, population-based study found a significantly lower risk of SAH in patients with SLE by analyzing hospitalization data in the US (4). Other than ethnic differences, a selection bias could not be excluded because this study calculated the odds ratios by comparing the incidence of SAH among subjects who were hospitalized from 2001–2002, rather than from the general population. Furthermore, the discrepancy of the admission criteria between SLE and non-SLE patients might lead to differing or opposing results.
We also found several differences in patients with SAH between the SLE patients and the matched cohort or the general population. Patients with SLE developed SAH at a mean age of 44.5 years, which is much younger than the matched cohort (57.7 years) or all of the non-SLE SAH patients identified from the 1 million NHI beneficiaries (56.9 years). The SLE patients also experienced a higher mortality rate, despite their younger age. These features were also reported by a Japanese study (6). This discrepancy might imply a different pathogenesis of SAH in patients with SLE. In fact, the causes of a higher risk of SAH in SLE patients are not fully known. In one Japanese study, intracranial aneurysms were found in 38 (76%) of 50 SLE patients with SAH. The higher rates of a saccular-shaped aneurysm and its location in the posterior circulation were noted in SLE subjects compared to the general populations in Japan, which suggested the influence of SLE on the mechanism of aneurysm formation. Several case reports demonstrated intracranial vasculitis in SLE patients with SAH, which might result in arterial dissection, aneurysm formation, or even rupture (11–13).
Although in our claim-based data set certain related information was not available, we found that those receiving a higher mean daily steroid dose were at a higher risk of SAH. In fact, patients with SLE who had higher daily steroid use also had significantly higher rates of admission (21.8 days/year versus 8.7 days/year; P < 0.001), cyclophosphamide pulse therapy (38.8% versus 15.0%; P < 0.001), and plasmapheresis treatment (4.8% versus 1.2%; P < 0.001). We suggest that higher mean daily steroid usage might be a surrogate of lupus disease activity. Those with higher SLE activity might experience lupus-related complications, such as angiopathy due to premature atherosclerosis and transmural vasculitis, and the latter was found to be the cause of atypical cerebral aneurysm formation in SLE patients (11–15). Furthermore, we also found that those who underwent a transfusion of platelets or red blood cells had a higher risk of SAH. We could not jump to the conclusion that SLE patients with hematologic manifestations have a higher risk. The higher risk associated with platelet transfusion might be due to the increased bleeding tendency that results from lupus-related thrombocytopenia. Since the indications of red blood cell transfusion were lacking, the causes of the higher risk, perhaps recurrent bleeding or lupus-related anemia, warrant further studies. The risk factors of SAH in the general population, including antiplatelet therapy, statin use, and hypertension, were also demonstrated in our univariate Cox analysis, but were not statistically significant in the multivariable model after the abovementioned factors were adjusted for.
There are clinical implications of this study. SAH is a rare complication of SLE patients with a high mortality. The extremely low incidence rate invalidates a routine screen. However, in taking care of patients with lupus, one should aggressively control lupus activity and other well-established risk factors of SAH, including hypertension, bleeding tendency, and cigarette smoking (22). Furthermore, in SLE patients with severe headache, despite a relatively young age, SAH should be suspected as an important differential diagnosis, especially in those with higher lupus activity or a history of recurrent platelet or red blood cell transfusion.
Our study has certain limitations. First, our study was based on a claim-based data set. Although many features make it suitable to analyze rare complications of SLE, such as SAH, certain information was lacking, including the clinical manifestations and laboratory or image examination results. Therefore, we cannot make deeper inferences into the pathogenesis of SAH in patients with SLE. Second, our diagnosis of SAH was image based, which may have underestimated the incidence, since some severe cases may have died before image studies were completed, or even before their arrival to the hospital. However, the high density of the hospitals and the availability and utilization of image studies in Taiwan might lessen this reason of underdiagnosis.
In conclusion, our study is the first to demonstrate a higher risk of SAH in SLE patients using a nationwide population data set. Elderly SLE patients and those with a higher mean daily steroid dose and a history of platelet or red blood cell transfusion had a higher risk of developing SAH.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Shuu-Jiun Wang had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Chang, Liu, Wei-Sheng Chen, Lai, Shuu-Jiun Wang.
Acquisition of data. Chang, Liu, Tzeng-Ji Chen, Tzeng, Shuu-Jiun Wang.
Analysis and interpretation of data. Chang, Liu, Shu-Hung Wang, Tsai, Shuu-Jiun Wang.
The authors thank Ms Yeh Chiu-Mei for providing statistical consultation and assistance.