Abstract
Objective. To identify the risk of hip and vertebral fractures in patients with rheumatic disorders (RD) and inflammatory bowel diseases (IBD).
Methods. This population-based case-control study assessed the fracture risk of patients with rheumatoid arthritis, juvenile idiopathic arthritis (JIA), ankylosing spondylitis (AS), systemic lupus erythematosus, polymyositis/dermatomyositis (PM/DM), systemic sclerosis (SSc), Crohn’s disease, and ulcerative colitis (UC). The study cohort comprised 53,108 patients with fracture (66% women) and 370,602 age-matched and sex-matched controls. Conditional logistic regression analysis was performed and results were expressed as OR with corresponding 95% CI.
Results. There was a statistically significant increased fracture risk for all RD and for IBD compared with controls. The magnitude of fracture risk was higher for patients with RD (OR 3, 95% CI 2.9–3.2) than for those with IBD (OR 1.6, 1.4–1.8). The OR in RD ranged from 2.6 (1.3–4.9) for SSc to 4 (3.4–4.6) for AS. The largest increased fracture risk for vertebral fractures was seen in AS (OR 7.1, 6–8.4) and for hip fractures in JIA (OR 4.1, 2.4–6.9).
Conclusion. Our results highlight the existence of an increased fracture risk from a variety of underlying causes in patients with RD and IBD. In many inflammatory diseases, implementation of fracture prevention strategies may be beneficial.
Chronic inflammatory diseases, such as rheumatic disorders (RD) and inflammatory bowel diseases (IBD), have cost implications for healthcare provision, employers, and insurers. Some of this expense is due to comorbidity, including bone fractures1,2,3,4,5,6,7,8,9,10,11. The multifactorial etiology of fractures includes inflammation in RD, causing low bone mass with periarticular osteoporosis, as well as osteomalacia from malnutrition and malabsorption. The catabolic effect of systemic glucocorticosteroid therapy in these diseases, as well as an increased risk of falling, may also contribute to fracture risk.
This large case-control study examined hip and vertebral fracture risk across a range of inflammatory diseases including RD and IBD.
MATERIALS AND METHODS
Source of data
The Swedish National Hospital Discharge Register (SNHDR), which collects information on all inpatient care in Sweden (including diagnoses using the International Classification of Diseases, ICD), provided the data for this study12. Data linkage used the personal identity number issued to all Swedish residents.
Study design
We identified all individuals registered in the SNHDR with ICD-9 and ICD-10 codes of hip fractures (820*, S720*–S722*) or vertebral fractures (805*, 806*, S120*–S127*, S129*, S220*, S221*, S320*–S322*). The study period was 1987 to 2004.
Each patient in the fracture cohort was matched with 7 controls by sex, age, and residential area (using the Total Population Register). We selected both populations (cases and controls) equally, i.e. for all cases it was the first hip or spine fracture. The controls did not have a hip or spine fracture at the time of the matching process or before.
We investigated any hospital admissions due to RD or IBD in cases and controls prior to the case’s admission for fracture using these codes: rheumatoid arthritis (RA; 714A, 714B, 714C, 714W, M05*, M060–M063, M068, M069), juvenile idiopathic arthritis (JIA; 714D, M080), ankylosing spondylitis (AS; 720A, M45*), systemic lupus erythematosus (SLE; 710A, M32*), polymyositis/dermatomyositis (PM/DM; 710D, 710E, M33*), systemic sclerosis (SSc; 710B, M34*), Crohn’s disease (CD; 555*, K50*), and ulcerative colitis (UC; 556*, K51*). Rheumatic disorders and IBD could have been the reason for admission or a comorbidity.
The local Ethics Committee of Stockholm North approved the study (DN 2005/1332-31/2).
Statistical analysis
Median values and interquartile ranges (IQR) were used descriptively. Conditional logistic regression analysis based on risk-set (patients with fracture and controls matched for sex, age, and residential area) was performed. The results were expressed as OR with corresponding 95% CI.
Surveillance bias is possible when the exposure disease is a medical condition that may result in a higher probability of detection of outcome in the exposed individuals, so secondary analyses were performed: (1) only the primary diagnoses of the exposure diseases in the SNHDR were included; (2) only matched controls who at least once had been admitted to a hospital were included; and (3) all hospital admissions 1 year prior to fracture were excluded. The risk of fractures for 2 control diseases, hypertension (401*, I10) and nephrolithiasis (592*, N200–N209), were also studied.
Statistical analyses were done using SPSS 15 (SPSS Inc., Chicago, IL, USA).
RESULTS
Study population
The fracture cases comprised 53,108 patients (35,174 women; 66%) with a hip or a vertebral fracture, who were compared with 370,602 controls. The median age of the case group was 71 years (IQR 61–78). Among cases, 47,282 (89%) had a hip fracture at a median age of 72 years (IQR 64–79). Vertebral fracture was found in 6216 (12%) cases at a median age of 41 years (IQR 25–58).
We identified 3884 patients diagnosed with RD. Of those, 3379 (87%) had RA, 34 JIA (1%), 265 AS (7%), 136 SLE (3%), 36 PM/DM (1%), and 34 SSc (1%). Among 1238 patients with IBD, 644 (52%) had CD and 594 (48%) had UC.
Risk of fracture
Table 1 illustrates fracture risk for the exposure diseases. The magnitude of fracture risk was higher for patients with RD (OR 3, 95% CI 2.9–3.2) than for IBD (OR 1.6, 95% CI 1.4–1.8), compared with controls. There was a statistically significant increase in risk of fracture for all RD, with OR ranging from 2.6 (95% CI 1.3–4.9) for SSc to 4 (95% CI 3.4–4.6) for AS. The largest increased risk for vertebral fractures was seen in AS (OR 7.1, 95% CI 6–8.4) and for hip fractures in JIA (OR 4.1, 95% CI 2.4–6.9). Both CD (OR 1.8, 95% 1.5–2.1) and UC (OR 1.4, 95% CI 1.1–1.7) had an increased fracture risk compared with controls.
Fracture risk in rheumatic disorders and inflammatory bowel diseases.
For RD, we divided the study period into 3 time periods and found a decrease in fracture risk for later years [1987–1994: OR 4 (95% CI 3.7–4.3), 1995–1999: OR 2.9 (95% CI 2.6–3.2), and 2000–2004: OR 1.8 (95% CI 1.6–2.1)].
Analyses to tackle surveillance bias did not notably alter the overall results. We identified 3503 patients with nephrolithiasis and there was no associated fracture risk (OR 1.1, 95% CI 0.9–1.2, p = nonsignificant). However, 25,078 patients who had been diagnosed with hypertension had an increased risk of fracture (OR 2.1, 95% CI 2.1–2.2, p < 0.001).
DISCUSSION
Hip and vertebral fracture risk of patients with RD is great, with OR between 2.5 and 7.1 compared with controls. Similarly, patients with IBD have a statistically significant increase in fracture risk, of lower magnitude than for RD.
Hip and vertebrae are typical fracture sites in patients with low bone mass due to osteoporosis or osteomalacia. Hip fractures represent serious medical incidents with high economic relevance, often requiring cost-intensive surgical treatments, hospitalization, and longterm rehabilitation. We selected these fracture types as indicators for the assessment of the overall risk of a fracture in the general population.
The results are consistent with other studies showing that patients with RA and JIA have an increased fracture risk2,3,4,5,13,14,15. Patients with RA showed a similar risk of hip fractures compared with a Finnish study4. However, our fracture risk was more prominent than in earlier reports from the UK2 and the United States3. The risk for vertebral fractures was similar to a report from Spector, et al5.
AS is associated with bone remodeling, especially bone loss in the vertebral bodies and the hip due to axial and articular inflammation16. The rigidity of the spine may contribute to an elevated vertebral vulnerability in patients with AS. We found a considerable increase for vertebral and hip fractures in this subtype of RD. This is consistent with Cooper, et al, who reported estimates of similar magnitude for clinical vertebral fractures in patients diagnosed with AS6. Although other studies showed an overall risk for vertebral fractures, increased hip fracture risk has not always been indicated7.
The catabolic effect of systemic glucocorticosteroid therapy in SLE and glomerulonephritis with renal impairment, a common involvement in SLE, could have adverse effects on bone mineralization. Patients with SLE had a combined risk for hip and vertebral fractures with an OR of 2.9. An earlier report showed a higher estimate of fracture risk in SLE8. In contrast with our study, the latter report analyzed all fracture types in women.
Yuen, et al, found the same occurrence of fractures in patients with SSc compared with RA9. Although the literature on fracture risk of both PM/DM and SSc is sparse, both disorders have been found to be independently risky for lumbar and femoral osteoporosis17,18. This may explain the increase in fracture risk for those disorders in our study.
Many patients with RD are treated with longterm systemic glucocorticosteroids, which may lead to decreased bone mineral density. Nevertheless, we do not believe that this solely explains the increased fracture risk in RD, since an increased fracture risk has also been apparent in patients without glucocorticosteroids1. Other risk factors such as inflammation and bone quality, including microarchitectural weakness, bone remodeling, bone turnover, and mineralization, may be important as well. This situation is also confounded by an increased incidence of falling in these patients.
Overall, the magnitude of fracture risk for patients with IBD was lower than in RD compared with controls. Different effects on bone metabolism and varying disease severity of the various conditions may explain the lower risk. In agreement with our results, 2 other studies reported a similar relative risk of fractures in patients with IBD10,11. We confirm the findings of van Staa, et al, who showed a higher fracture risk in patients with CD compared with those with UC10.
We found a decrease in fracture risk of patients with RD during the study period. It is difficult to determine the reasons for the decreasing risk due to changes in medicinal therapy over time. This is because temporal variations in admissions policies will also have changed the characteristics of the patient population admitted with inflammatory diseases. Patients who experienced fractures during any given time period will have had their RD diagnosed in a variety of different time periods and so their treatment history will be somewhat heterogeneous.
Our study is limited by the lack of information on other explanatory factors such as patient characteristics, lifestyle factors, and medication use, all data not included in the SNHDR. However, validation studies of other medical diagnoses indicate that the accuracy of the SNHDR is close to 90%12. By having very large samples in both the fracture and control group, we believe that possible bias through false data entry should be equal in both cohorts. Still, our results may have been biased by an inappropriate classification of the RD and IBD studied. The finding of a positive association between the exposure diseases and fracture risk, which is in line with our primary hypothesis, however, adds to the validity of our data.
Because we are using inpatient diagnoses, our study may identify a subset of the population with these diagnoses who have a more severe disease or possible other sources of adversity that increase both the risk of hospital admission and fracture. It is possible that among patients with less severe diseases, who received only outpatient care, the association with fracture will be lower. Still, our sample in this study represents the most complete inpatient population. While we are less certain about admissions for IBD, we know that in patients with RA, about 50% are admitted to a hospital at some time, as estimated in 2007 (personal communication, J. Askling, Department of Medicine, Karolinska University Hospital, Karolinska Institutet, Stockholm, Sweden). It is likely that those admitted will have more severe disease and possibly be more likely to have comorbid conditions that could also be relevant to fracture risk.
Nonvertebral fractures such as hip fractures are usually easy to diagnose, as patients have severe pain and are unable to walk. By contrast, diagnosing vertebral fractures can be more complicated. Therefore our study population represents more precisely a cohort of patients with clinical vertebral fractures, because morphometric fractures of the spine are often missed.
As expected, there was no association between fractures and nephrolithiasis. However, patients with hypertension also had a higher risk of fractures, which has been described13. The authors speculated that the mechanism behind the elevated risk of fractures in hypertension may be an increased loss of calcium in the urine, leading to a negative calcium balance19.
Our results highlight an increased fracture risk in patients with RD and IBD. Therefore, fracture prevention strategies may be of benefit and warrant further research addressing the important issue of fractures in chronic inflammatory diseases.
Footnotes
-
Supported by grants from the Karolinska Institutet, the Stockholm County Council, the Swedish Orthopaedic Association, and the Capio Research Foundation.
- Accepted for publication June 14, 2010.
