Article Text

Extended report
Increased risk of osteoporotic fractures in patients with systemic sclerosis: a nationwide population-based study
  1. Chien-Chih Lai1,2,
  2. Shu-Hung Wang2,3,
  3. Wei-Sheng Chen1,2,
  4. Chia-Jen Liu4,5,
  5. Tzeng-Ji Chen2,6,
  6. Pui-Ching Lee7,
  7. Yu-Sheng Chang2,3
  1. 1Division of Allergy, Immunology, and Rheumatology, Department of Medicine, Taipei Veterans General Hospital, Taipei City, Taiwan
  2. 2National Yang-Ming University, Taipei, Taiwan
  3. 3Division of Allergy, Immunology, and Rheumatology, Department of Internal Medicine, Shuang Ho Hospital, Taipei Medical University, New Taipei City, Taiwan
  4. 4Division of Hematology and Oncology, Department of Medicine, Taipei Veterans General Hospital, Taipei City, Taiwan
  5. 5Institute of Public Health & School of Medicine, National Yang-Ming University, Taipei, Taiwan
  6. 6Department of Family Medicine, Taipei Veterans General Hospital, Taipei, Taiwan
  7. 7Department of Medicine, Taipei Veterans General Hospital, Taipei, Taiwan
  1. Correspondence to Dr YS Chang, Division of Allergy, Immunology, and Rheumatology, Department of Internal Medicine, Shuang Ho Hospital, Taipei Medical University, No. 291, Zhongzheng Road, Zhonghe District, New Taipei City 23561, Taiwan; risea65{at}


Objectives To identify the incidence rate (IR) and risk factors of osteoporotic fractures (OFs) among systemic sclerosis (SSc) patients.

Methods A cohort study was conducted using the Taiwan National Health Insurance database. Patients with SSc and respective age- and gender-matched controls without SSc were enrolled. The primary endpoint was the first occurrence of OF. The Cox proportional hazard model was used to investigate the risk factor of OFs in the SSc cohort.

Results Among 1712 SSc patients (77.8% female, mean age 50.3 years) with a median follow-up of 5.2 years, 54 patients developed vertebral fractures, 17 patients developed hip fractures, and 7 patients developed radius fractures (IR: 6.99, 2.18 and 0.90 per 1000 person-years, respectively). Compared with the controls, the incidence rate ratios (IRRs) (95% CIs) among SSc patients were 1.78 (1.30 to 2.39, p<0.001) for vertebral fractures and 1.89 (1.05 to 3.22, p=0.026) for hip fractures. The IRRs for overall OFs were 1.74 (1.32 to 2.27, p<0.001) for women and 1.06 (0.33 to 2.66, p=0.856) for men. The SSc patients experienced hip fractures at a younger age (67.2 vs 75.2 years, p=0.005), and had a higher 1-year mortality rate (13% vs 3%, p=0.006) of vertebral fractures than did the controls. Multivariable Cox regression analyses indicated that older age, being female, using daily prednisolone equivalent to >7.5 mg, and bowel dysmotility treated with intravenous metoclopramide are associated with OF.

Conclusions SSc patients had a high IR of vertebral and hip fractures, especially those who were female, older, used a high dose of corticosteroid or experienced bowel dysmotility.

  • Systemic Sclerosis
  • Osteoporosis
  • Epidemiology

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Systemic sclerosis (SSc) is a chronic debilitating systemic autoimmune disease with a reported prevalence of 56.3–341 cases per million population.1–3 SSc patients are characterised by scleroderma and have a high risk of certain complications, including digital ischaemic ulcers, interstitial lung disease (ILD), pulmonary arterial hypertension (PAH), renal crisis and bowel dysmotility.4

Osteoporosis is a systemic skeletal disease characterised by decreased bone mineral density (BMD), disrupted microarchitecture of the skeleton and subsequent fracture susceptibility.5 Because of increased secular trends in the incidence of osteoporotic hip fractures occurring in numerous Western and Asian populations,6 the medical care costs caused by osteoporotic fractures (OFs) have increased.7 ,8 Approximately half of all patients with hip fractures lose their physical independence and require institutionalisation, and 20–25% expire after 1 year of living with a hip fracture.9 ,10 Patients with autoimmune disease may have a high risk of osteoporosis and fracture.11 Emerging evidence has indicated that SSc patients are associated with low BMD12–19 and osteoporosis.13–16 However, the risk of OF in SSc patients has not been extensively investigated.

Several studies have revealed a possible association between SSc and OF.11–13 However, these were relatively small-scale, single-centre studies using cross-sectional or case–control designs. Furthermore, the incidence rate (IR), characteristics and subtype differences of OF in the SSc cohort are unknown, and the risk factors have not been comprehensively studied. Therefore, we conducted this cohort study using a nationwide population-based dataset to investigate this major issue.

Patients and methods

Data sources

The National Health Insurance (NHI) programme, a universal health insurance system, was established in 1995 in Taiwan. Taiwanese citizens, predominantly of Han Chinese and aboriginal ancestry, are required by law to join the programme. The NHI programme provides comprehensive medical services including outpatient, inpatient, emergency and dental services, and also provides traditional Chinese medicine. By the end of 2006, the NHI programme provided a coverage rate of more than 99%.20 The NHI Research Database (NHIRD) data—which include NHI enrolment files; claim data for management, examinations and drug prescriptions; and the diagnoses for all admitted patients and outpatients—were released for research purposes. The International Classification of Diseases, ninth revision, Clinical Modification codes (ICD-9-CM codes; 2001 revision) were used to perform disease coding in the system. This study was approved by the Institutional Review Board at Taipei Veterans General Hospital.

Study cohorts

In the NHI system, SSc is defined as a catastrophic illness. Patients who fulfil the 1980 American College of Rheumatology (ACR) classification criteria for SSc21 are enrolled in the Registry of Catastrophic Illnesses after being verified by a review committee. We conducted a cohort study by analysing the SSc catastrophic illness database. In the study period between 1 January 2000 and 31 December 2006, all patients with SSc catastrophic illness certification (ICD-9-CM: 710.1) were identified.

Patients who (1) experienced an OF before enrolment, (2) were aged less than 18 years, or (3) had pathological fractures (ICD-9-CM: 733.14 and 733.15) or malignancies were excluded.

The NHI dataset containing 1 million NHI beneficiaries, who were randomly selected from a total of 21 400 826 enrolees throughout Taiwan in 2000, was adopted for the non-SSc control cohort. The same exclusion criteria were applied to the matched cohort, and six age- and gender-matched non-SSc controls were randomly selected from the database for each SSc patient.

Primary outcomes and subgroup analysis

The hip fractures, distal radius fractures documented in the inpatient database, or thoracic spine and lumbar spine vertebral fractures documented in the inpatient and outpatient dataset which occurred in every patient during the study period following relevant radiographic examinations were identified as OF after excluding patients accompanied by codes indicating traumatic co-injuries (see online supplementary table S1 for relevant ICD-9-CM codes). In the individual analysis of the incidence and risk of vertebral fracture, hip fracture or distal radius fracture, the first occurrence of each type of fracture was identified as the endpoint. Regarding the analysis of overall OF, the endpoint was designated as the first occurrence of either type of fracture. Follow-up procedures were performed until the occurrence of the endpoint, permanent disenrollment from the NHI, or the end of the study. Deaths that occurred within 1 year following the index OF event were also recorded.

To analyse the clinical relevance of OF, we further stratified the data for various fracture sites according to age and gender. Because menopause status was not available in the NHI database, subgroup analysis was conducted assuming that patients aged ≥50 years have an increased fracture risk, because the mean age of menopause in Taiwan is approximately 50 years.22

Vertebral fracture risk analysis with an adjusted control cohort

In contrast to patients with hip and distal radius fractures, not all those with vertebral fractures seek medical attention. The lower medical utilisation in the non-SSc cohort might lead to underdiagnosis of vertebral fracture and observation bias. Thus, an adjusted control cohort was randomly selected from subjects receiving at least one plain x-ray examination during the study period in the 1 million NHI beneficiaries at a 6:1 ratio to the SSc cohort; that is, six age- and gender-matched non-SSc subjects for each SSc patient with the same exclusion criteria. Vertebral fractures were identified and compared to the SSc cohort.

Risk factors of fractures in SSc patients

Age, gender, pre-existing co-morbidities, overlap with other autoimmune diseases, SSc-related complications, and medication use were recorded. Co-morbidities were ascertained based on the ICD-9-CM codes included in the inpatient and outpatient datasets before enrolment. Overlap syndrome was defined as having any of the following diseases: rheumatoid arthritis, systemic lupus erythematosus, Sjögren syndrome and inflammatory myositis. SSc duration and SSc-related complications, including ILD, PAH, malnutrition and ileus, were also recorded.

Immunosuppressive drugs for treating SSc, including oral D-penicillamine and intravenous cyclophosphamide that were prescribed within the follow-up period, were extracted for analysis. Prokinetic drugs, including oral metoclopramide, mosapride, domperidone and intravenous metoclopramide, were identified as markers of potential SSc-related bowel dysmotility and malabsorption, which is associated with bone loss in female SSc patients.15 Medications for treating osteoporosis were excluded from the analysis to prevent potential susceptibility bias and misinterpretation, as described previously.23 To analyse the effect of steroids on OF, cumulative doses and the mean daily dose of oral corticosteroids (prednisolone or its equivalent) were identified. To prevent immortal time bias, only the mean daily dose of oral corticosteroids was included in the analysis. We also recorded the use of proton pump inhibitors and intravenous methylprednisolone during the follow-up period to evaluate its effect on fractures.

Statistical analysis

The IRs of major outcomes were compared between the SSc and control cohorts. Fracture-free survival was assessed using Kaplan–Meier analysis. The risk factors for fractures in SSc patients were evaluated using a multivariable Cox proportional hazard model. p Values <0.05 were considered to be statistically significant. Extraction and computation of data were performed using the Perl programming language (V.5.12.2). An SQL Server 2005 (Microsoft, Redmond, Washington, USA) was used for data linkage, processing and sampling. All statistical analyses were performed using SPSS V.19.


Clinical characteristics of the study population

From 2000 to 2006, 1832 patients with SSc catastrophic illness certification were identified. After excluding 43 patients having OF before enrolment, 64 patients aged <18 years, one patient with a pathological fracture and 12 patients with malignancies, 1712 adult SSc patients were recruited, with 10 272 age- and gender-matched non-SSc controls. The mean age of the study cohorts was 50.3 years, and 50.1% of the population was aged ≥50 years (table 1). The female to male ratio was 3.5:1. Although the median follow-up years was significantly lower in the SSc group than in the control group (5.2 vs 7.0 years, p<0.001), similar rates of overall OF were calculated (4.3% vs 3.7%, respectively). The SSc patients were significantly younger when they exhibited overall OF than were those in the control group (66.8 vs 69.7 years, p=0.049), especially those with hip fractures (67.2 vs 75.2 years, p=0.005). The 1-year mortality rates after the index fracture were higher among SSc patients than among controls, and the 1-year mortality rates after overall and vertebral fractures were statistically significant (table 1).

Table 1

Characteristics of patients with SSc versus age- and sex-matched cohorts

Incidence of OF

The SSc cohort had a significantly higher IR of overall OF (incidence rate ratio (IRR) 1.69, 95% CI 1.30 to 2.18, p<0.001), vertebral fracture (IRR 1.78, 95% CI 1.30 to 2.39, p<0.001) and hip fracture (IRR 1.89, 95% CI 1.05 to 3.22, p=0.026) than did the controls, but no difference in distal radius fracture was observed (table 2), which is consistent with the result of the Kaplan–Meier analysis (figure 1). When compared with the adjusted control cohort, the SSc cohort still had a significantly higher risk of vertebral fracture (IRR 1.75, 95% CI 1.28 to 2.35, p<0.001). The higher incidence of OF in the SSc cohort was primarily caused by the increased risk of vertebral and hip fracture in female patients (table 1), because the fracture incidence of male patients was similar to that of the controls (IRR 1.06, 95% CI 0.33 to 2.66, p=0.856). In the subgroup analyses, although the incidence of hip and vertebral fractures substantially increased in female SSc patients over 50 years of age, those who were under 50 years had a significantly higher incidence than did those in the matched cohort (vertebral fracture IRR 3.43, 95% CI 1.65 to 6.75, p<0.001). Regarding the IR, 1-year mortality rate, and further subgroup analyses of the radius fracture, no statistical differences were observed between the SSc and control groups.

Table 2

IRs and IRRs of OF: overall and subgroup analysis

Figure 1

Kaplan–Meier curves of the fracture-free survival of (A) overall fractures, (B) vertebral fractures, (C) hip fractures, and (D) distal radius fractures in systemic sclerosis (SSc) versus the matched cohorts.

Risk factors of OF in patients with SSc

In the univariate analyses, older age; being female; co-morbidities including hypertension, diabetes, coronary artery disease, heart failure, chronic obstructive pulmonary disease, asthma and malnutrition; medication with intravenous methylprednisolone, intravenous metoclopramide, oral prokinetic agents or oral proton pump inhibitors; and a mean daily dose of oral prednisolone equivalent to >7.5 mg were associated with an increased risk of OF in SSc patients (table 3). Of these factors, age, being female, a mean daily dose of oral prednisolone equivalent to >7.5 mg and the use of intravenous metoclopramide were designated as independent risk factors in the multivariable Cox regression model.

Table 3

Risk factor analysis of OF in patients with SSc*


Based on a thorough review of relevant research, this is the first cohort study that focused on the risk of OF using the largest SSc cohort to date. This nationwide population-based study reported a high risk of OF in 1712 adult Taiwanese SSc patients with an IRR of 1.69, which was primarily due to the increased vertebral (IRR 1.76) and hip (IRR 2.30) fracture risk in female patients. In addition, those younger than 50 years developed an increased risk of OF, and SSc patients were significantly younger when hip fracture occurred than were those in the control cohort. Older age, being female, medication with a daily dose of oral prednisolone equivalent to >7.5 mg and bowel dysmotility treated with intravenous metoclopramide were independent risk factors of OF in SSc patients.

The validity of this study is supported by the unbiased patient selection and a median follow-up duration of more than 5 years. In the NHI system, patients with a certification of catastrophic illness can be exempted from copayment for related medical care. The verification process requires the precise fulfilment of the 1980 ACR classification criteria for SSc. These factors ensure that the diagnosis of SSc is reliable. Thus, using this nationwide SSc cohort and a relatively long follow-up period ensure that this study is suitable for analysing rare complications and performing further subgroup analysis without selection bias. Furthermore, in light of observation bias we compared the risk of vertebral fractures between the SSc and adjusted control cohorts. The SSc cohort still had a higher risk of vertebral fracture compared with the adjusted control cohort, which appeared to have higher healthcare utilisation, but might be unhealthier than and less representative of the general non-SSc cohort.

The study by Yuen et al did not reveal a higher rate of OF in SSc patients compared with patients with non-inflammatory musculoskeletal diseases.24 However, we believe that these results were limited by the small sample size because the rates of osteoporosis and BMD in the SSc cohort were similar to those of rheumatoid arthritis. A case–control study by Weiss et al indicated an increased risk of hip fracture in SSc patients compared with healthy controls11; this was further supported by the cross-sectional study of Avouac et al, which reported a significantly higher prevalence of OF in 71 SSc patients compared with 227 healthy controls.13 Several results of the present study are consistent with these results. This study established a time relationship by excluding patients with OF before being diagnosed with SSc, and demonstrated that a significantly higher incidence, rather than association or prevalence, of OF occurred in the SSc cohort. Moreover, this study used a large SSc cohort to perform subgroup analyses. Compared with the control group, the risks of osteoporosis-related vertebral and hip fractures were not significantly high in male SSc patients, but were significantly high in female SSc patients, even those below the age of 50. In addition, the risk of distal radius fracture was not increased in the SSc patients.

The conclusion that male patients in the SSc and control cohorts exhibited similar IRs of OF is not irrefutable because mortality was a strong competitive factor of OF in the SSc cohort. In the study period, the SSc patients had both a significantly higher mortality rate and a younger mortality age than did those in the control cohort (total SSc vs control: 62.2 vs 69.7 years, p<0.001; male: 64.3 vs 70.1 years, p=0.001; female: 61.2 vs 69.4 years, p<0.001). Despite the competitive factor, the female SSc cohort still maintained a higher OF rate than did the control cohort. In addition, factors such as small peak bone mass,25 rapid bone loss after menopause and a significant decrease in trabecular number26 that render females susceptible to osteoporosis, resulted in an independent risk factor to develop OF in the SSc cohort.

Distal radius fractures, which were the only exception of increased OF risk in the SSc cohort, frequently occur in relatively healthy and active people with excellent neuromuscular functions, and are associated with low BMD, poor visual acuity, a previous history of falls and frequent walking.27 By contrast, performing physical functions with difficulty is associated with a reduced risk of distal forearm fracture.28 The SSc patients experienced more frequent myopathy, arthralgia and tendon friction rub than expected29; such conditions may cause physical disabilities, which may subsequently reduce the occurrence of distal radius fracture in SSc patients.

Using glucocorticoids is a well documented risk factor of osteoporosis, and the risk of fracture increased rapidly when glucocorticoid use persisted for 3–6 months.30 Up to 23–58% of the SSc patients in various studies had been exposed to glucocorticoids.13 ,17 ,19 ,24 We determined that a 4.75-fold fracture risk occurred in SSc patients when the daily dose was >7.5 mg of prednisolone equivalents. However, the claim-based data used in this study lacked indication of steroid use. Thus, evaluating the effect of increased risk caused by SSc inflammatory manifestations indicating steroid use, such as ILD, arthritis or inflammatory myopathy, is difficult.

For certain SSc-related complications, their rarity rendered it unlikely to reach statistical significance in our Cox model. Patients with these complications had a higher mortality rate and a shorter follow-up duration. Taking PAH (n=28) and ILD (n=65) for instance, the small case numbers may partially result from underdiagnosis or underrepresentation. Thus these complications, when presented in a milder condition, may not be coded accordingly in addition to SSc. Similarly, malnutrition (n=30) and ileus (n=74) may also have insufficient representation in the dataset. These two SSc gastrointestinal complications were pertinent to a poorer nutritional status, and in particular shortage of calcium and vitamin D, which may contribute to reduced BMD15 and subsequent OF13 in SSc patients. Thus, we used intravenous metoclopramide as a surrogate for malnutrition, ileus and possibly lower body mass index in our analysis. The prokinetic agent appeared to be an independent risk factor in a multivariable model of OF in SSc patients.

This study presents several clinical implications. Physicians caring for SSc patients should focus on the increased risk of OF, especially those with risk factors such as being female, exhibiting bowel dysmotility and receiving a high dose of steroids. Adequate osteoporosis risk assessment may ensure that preventive measures are promptly taken. We also suggest that physicians make efforts to improve bowel motility and nutritional status of SSc patients, limit the use of steroids and attempt to reduce the fall risk in these patients.

This study has several limitations. Data regarding SSc subtype, severity of skin tightening and joint contracture, BMD, body mass index, the menopause status, autoantibody specificities and radiographic findings such as Genant's scores of vertebral fractures of individual patients were not available in the claim-based dataset. However, the relationship between skin involvement and BMD remained uncertain.17 ,31 Also, previous studies failed to find significant associations between other factors and OF in SSc patients.12 ,13 Several risk factors for OF, such as propensity of falls, family history of OF, calcium and vitamin D intake, alcohol consumption and smoking status, were also unknown. The ethnic background of the patients in this study was predominantly Asian and diagnoses of SSc were not made based on the 2013 criteria,31 which limits the generalisability of the results to other ethnic groups and to earlier or milder SSc patients who only fulfil the new classification criteria.

In conclusion, SSc patients are at high risk of vertebral and hip fractures, and experience hip fractures at a younger age and exhibit a higher 1-year mortality of vertebral fracture than does the general population. Medication with glucocorticoids and the status of gastrointestinal involvement may play a crucial role in the occurrence of OF in SSc patients. Further study is required to elucidate the pathophysiology, treatment and prevention of OF in SSc patients.


We thank the Biostatistics Task Force of Taipei Veteran General Hospital and Ms Chiu-Mei Yeh for assistance with statistical analysis.


Supplementary materials

  • Lay summary

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Handling editor Tore K Kvien

  • CCL and SHW contributed equally.

  • Contributors Study conception and design: C-CL, Y-SC. Acquisition of data: Y-SC, S-HW, C-JL, T-JC. Analysis and interpretation of data: C-CL, S-HW, Y-SC, W-SC, P-CL. All of the authors were involved in drafting the article or revising it critically for valuable intellectual content, and approved the final manuscript.

  • Funding This work was supported in part by grants from the Taipei Veterans General Hospital (V101A-025, V101A-013 and V102A-028), Taiwan.

  • Competing interests None.

  • Ethics approval Institutional Review Board at Taipei Veterans General Hospital.

  • Provenance and peer review Not commissioned; externally peer reviewed.