Skip to main content
Annals of the Rheumatic Diseases logoLink to Annals of the Rheumatic Diseases
. 2006 Aug;65(8):1074–1079. doi: 10.1136/ard.2005.048835

Childhood onset arthritis is associated with an increased risk of fracture: a population based study using the General Practice Research Database

J M Burnham 1,2,3,4, J Shults 1,2,3,4, R Weinstein 1,2,3,4, J D Lewis 1,2,3,4, M B Leonard 1,2,3,4
PMCID: PMC1798264  PMID: 16627541

Abstract

Background

Childhood onset arthritis is associated with low bone mass and strength.

Objective

To determine whether childhood onset arthritis is associated with greater fracture risk.

Methods

In a retrospective cohort study all subjects with onset of arthritis between 1 and 19 years of age in the United Kingdom General Practice Research Database were identified. As controls, all sex and age matched subjects from a practice that included a subject with arthritis were included. Incidence rate ratios (IRRs) for first fracture were generated using Mantel‐Haenszel methods and Poisson regression.

Results

1939 subjects with arthritis (51% female) and 207 072 controls (53% female) were identified. The median age at arthritis diagnosis was 10.9 years. A total of 129 (6.7%) first fractures were noted in subjects with arthritis compared with 6910 (3.3%) in controls over a median follow up of 3.90 and 3.95 years in the subjects with arthritis and controls, respectively. The IRR (95% confidence interval) for first fracture among subjects with arthritis, compared with controls, according to the age at the start of follow up were 1.49 (0.91 to 2.31) for age <10 years, 3.13 (2.21 to 4.33) at 10–15 years, 1.75 (1.18 to 2.51) at 15–20 years, 1.40 (0.91 to 2.08) at 20–45 years, and 3.97 (2.23 to 6.59) at >45 years.

Conclusions

Childhood onset arthritis is associated with a clinically significant increased risk of fracture in children, adolescents and, possibly, adults. Studies are urgently needed to characterise the determinants of structural bone abnormalities in childhood arthritis and devise prevention and treatment strategies.

Keywords: children, arthritis, fracture, epidemiology, osteoporosis


Juvenile idiopathic arthritis (JIA) is the most common rheumatic disease of childhood.1,2 Children with JIA have multiple risk factors for low bone mass, including delayed puberty, malnutrition, weakness, inactivity, inflammation, and glucocorticoid treatment. Throughout childhood and adolescence, bone mineral accretion results in ethnic‐, sex‐, and maturation‐specific increases in bone dimensions and density that are critical to both short and long term skeletal integrity.3,4,5 Subjects with higher peak bone mass after adolescence have a greater protective advantage when the inexorable decline in bone mass associated with increasing age occurs. Thus, the impact of childhood arthritis on bone health may be immediate, resulting in childhood fractures, or delayed, owing to suboptimal peak bone mass attainment or persistent disease activity.

Numerous studies suggest that there are significant bone deficits in JIA.6 Roth et al recently examined children with active JIA using peripheral quantitative computed tomography (pQCT) of the forearm.7 Subjects with polyarticular JIA had low trabecular volumetric density. Cortical bone strength deficits were noted in all JIA subtypes. Bianchi et al have recommended bisphosphonate treatment for prevention of osteoporosis in children with rheumatic diseases, such as JIA, who are receiving long term glucocorticoid treatment.8

Although imaging based studies of children with JIA are critical for characterising skeletal pathology precisely and for determining JIA characteristics associated with osteoporosis, bone density and strength estimates are surrogate measures of fracture risk. Case reports of fractures in JIA have been published.9,10,11,12 Before the use of methotrexate and biological therapies, vertebral fractures were seen in up to 50% of children with JIA.13 Lien et al studied changes in bone mass over 2 years in children with JIA and healthy controls.14 Despite a similar percentage of fractures at baseline, a higher percentage of the JIA group had new fractures during the follow up period (9% of the JIA group v 5% of controls), but this difference was not statistically significant.

In recent years, investigators have used population based methods to examine the epidemiology of fractures during both childhood and adulthood.15,16,17,18,19,20 In particular, the United Kingdom General Practice Research Database (GPRD) has been used to determine the risk of fracture attributable to glucocorticoid use and conditions such as inflammatory bowel disease and epilepsy. The purpose of this study was to determine whether a population based sample of subjects with childhood onset arthritis have an increased risk of fracture as children and as adults, compared with a healthy control group, using the GPRD.

Methods

Data source

We conducted a retrospective cohort study using the GPRD, the largest longitudinal database of anonymised medical records from the primary care setting. Currently, there are 8.9 million patient records, and over 35 million patient years of observational data. Data collected include registration date, date of birth, sex, diagnostic codes for acute and chronic illnesses and injuries, and prescriptions provided by the primary care provider. GPRD contributors are rigorously trained, and data are subject to quality control audits to be considered “up to standard” (UTS).21

Studies have demonstrated the completeness and accuracy of GPRD diagnostic and therapeutic data,22,23,24,25 which have been used to identify chronic conditions and drug exposures associated with increased fracture rates. In a cohort of oral and non‐systemic glucocorticoid users, van Staa et al found 87% and 91% confirmation of vertebral and hip fractures, respectively. No group differences in the proportion of confirmed fractures were seen, suggesting an absence of bias in fracture reporting due to underlying chronic illness.26

This study was approved by the Institutional Review Boards of the Children's Hospital of Philadelphia, the University of Pennsylvania School of Medicine, and the Scientific and Ethical Advisory Board of the GPRD.

Study population

The exposure of interest was entry of an Oxford Medical Information Systems (OXMIS)/READ diagnostic code consistent with arthritis between the ages of 1 and 19 years. This definition of arthritis was designed to identify subjects of all ages with a history of arthritis during childhood and adolescence and diminish the possibility of capturing infants with congenital or infectious arthropathies. Both incident and prevalent cases were identified, regardless of current age or disease activity.

The incidence of fracture varies by sex, age, race, socioeconomic status, and geographic region.27 Accordingly, control subjects were matched by sex, year of birth, and practice. All sex and age matched subjects from a practice that included a subject with arthritis were included.

Identification of fractures

OXMIS/READ diagnostic codes consistent with fractures18 were identified and classified according to anatomical location. For subjects with arthritis, the start of the follow up period began after the date of diagnosis of arthritis and when the participating practice's data were considered UTS, whichever was later. Control subjects were followed up from their registration date in the practice or the first UTS date, whichever was later. The follow up period was ended if the subject had a first fracture, if the practice was no longer UTS, if a subject left the participating practice or died.

Fractures occurring before the start of the follow up period were noted. For arthritic subjects, these fractures were defined as occurring at least 6 months before the first diagnosis of arthritis to avoid a common clinical situation where new onset arthritis may be misclassified as a fracture.

Identification of covariates and exclusion criteria

We identified codes consistent with conditions and drugs known to have an impact on bone health. Table 1 gives a list of these conditions drug exposures.

Table 1 Conditions and drug exposures excluded in study subjects.

Exclusion criteria in patients with arthritis and controls Exclusion criteria in controls only
Conditions Drugs Conditions Drugs
SLE Behçet's disease Antiepileptics Uveitis Prednisone
MCTD IBD Medroxyprogesterone acetate Diabetes Methylprednisolone
Sjögren's syndrome Sickle cell disease Lithium Hyperthyroidism Methotrexate
Scleroderma Rheumatic fever Cholestyramine Hypothyroidism Etanercept
Vasculitis Rickets Leuprolide Lymphoedema Infliximab
Sarcoidosis Cerebral palsy Antipsychotic drugs Psoriasis Anakinra
Inflammatory neuropathy Cystic fibrosis Leflunomide
Multiple sclerosis Seizures Hydroxychloroquine
Inflammatory myopathy Asthma Sulfasalazine
Non‐inflammatory myopathy Coeliac disease Ciclosporin A
Osteogenesis imperfecta Renal insufficiency Thyroid active drugs
Organ transplant Malignancy

SLE, systemic lupus erythematosus; MCTD, mixed connective tissue disease; IBD, inflammatory bowel disease.

Some conditions and drugs were considered criteria for exclusion in both the subjects with arthritis and controls. For example, subjects with epilepsy, including those treated with antiepileptic drugs have an increased risk of fracture.20 Because epilepsy is not a direct cause or result of arthritis, subjects with epilepsy were excluded from both the arthritis and the control group. In contrast, JIA is associated with uveitis,28 and may require the use of disease modifying antirheumatic drugs (DMARDs) to prevent visual loss. Subjects in the control group with uveitis, who also may require DMARD use, were excluded from the study, but arthritic subjects were not.

Statistical analysis

Analyses were conducted using Stata 8.2 (Stata Corporation, TX). Descriptive analyses included means, standard deviations, median and ranges of continuous variables, and distributions of categorical variables. Differences of means and medians were assessed using Student's t test or the Wilcoxon rank sum test, respectively. Group differences in categorical variables were assessed using the χ2 or Fisher's exact test, where appropriate. Two sided tests of hypotheses were used and a p value <0.05 was considered to be significant.

Fracture incidence is dependent on sex and age in healthy children. Cooper et al described similar fracture rates in boys and girls under the age of 10. Thereafter, fracture rates continue to rise in boys until the peak incidence is reached at about 13–15 years.27 We used Mantel‐Haenszel methods to estimate the incidence rate ratio (IRR) for fracture with 95% confidence intervals (CIs) in the arthritis group, stratified by sex and age group at the start of follow up (<10 years, 10–15 years, 15–20 years, 20–45 years, and >45 years). If the Mantel‐Haenszel test for heterogeneity strongly suggested heterogeneity (p<0.10), then effect modification was deemed to be present. For homogeneity across strata, the Mantel‐Haenszel combined IRR estimate with 95% CIs is presented.

Models were constructed to assess the sensitivity of the fracture outcome according to restricted definitions of the arthritis exposure. Fracture rates were assessed in subjects with at least two arthritis diagnostic entries, in those with arthritis diagnosed only during a UTS period, and in patients with arthritis receiving non‐steroidal anti‐inflammatory drugs (NSAIDs). Multivariable Poisson regression models were used to determine IRR estimates for fracture when adjustment for multiple covariates, such as age category and sex, was required. Poisson regression models were checked for goodness of fit.

Results

Study population

A total of 2792 (53.9% female) subjects with a history of arthritis and 303 711 (53.3% female) age, sex, and practice matched controls were identified. After the application of exclusion criteria (table 1), the study population included 1939 (51.4% female) subjects with arthritis and 207 072 (53.1% female) controls. Table 2 shows the most common diagnostic codes entered.

Table 2 Frequency of all arthritis diagnostic codes entered in 1939 subjects with arthritis.

Arthritis codes No Frequency Exposed to DMARD*
(%) (%)
Arthritis 1284 39.09 7.1
Synovitis 731 22.25 0.6
Juvenile arthritis 348 10.59 14.5
Juvenile rheumatoid arthritis 305 9.28 9.8
Ankylosing spondylitis 253 7.70 9.3
Polyarthritis 187 5.69 4.4
Psoriatic arthropathy 117 3.56 12.2
Other† 60 1.83 6.3
Total 3285‡ 100

*Subjects were categorised according to their last diagnosis of arthritis for the assessment of DMARD exposure; †codes in the “Other” category include: rheumatoid arthritis (RA) monitoring, rheumatoid disease, RA increased activity, spondylarthrosis, synovial cyst, and Reiter's syndrome; ‡note that there were multiple diagnostic code entries for individual subjects with arthritis.

Table 3 details characteristics of the study population. The arthritic patients were slightly younger at the start of follow up (median age 17.3 v 19.7 years, p = 0.005), but the duration of follow up between groups was similar. Fractures before the start of the follow up period were significantly more common in the subjects with arthritis than in controls (7.1% v 3.2%, p<0.001). For the arthritic subjects, these fractures occurred either before or after arthritis onset, because both incident and prevalent subjects with arthritis are represented in the study population. The higher proportion of subjects in the arthritis group with documented NSAID prescriptions (54.0% v 12.7%, p<0.001) is consistent with arthritis treatment. A small percentage of subjects with arthritis received prescriptions for DMARDs (5.7%) and glucocorticoids (4.9%). There were significant differences in DMARD exposure according to arthritis group (p<0.001), with DMARD use being most prevalent in those with juvenile arthritis and psoriatic arthropathy codes recorded.

Table 3 Characteristics of the study population and concurrent exposures.

Characteristics Arthritis group Control group p Value
(n = 1939) (n = 207 072)
Female (%) 51.4 53.1 0.15
Age at 1st recorded diagnosis, mean (SD) 10.9 (5.2)
Age at start of follow up 17.3 (1–96) 19.7 (0–104) 0.005
Follow up duration (years)* 3.90 (0.003–14.0) 3.95 (0.003–14.0) >0.2
Fracture before follow up (%)† 7.1 3.2 <0.001
DMARD treatment (%)‡ 5.7
Glucocorticoid treatment (%) 4.9
NSAID treatment (%) 54.0 12.7 <0.001

Medians and ranges provided, unless otherwise noted.

*Follow up duration was calculated as follows: (date that the subject has first fracture, dies, leaves the eligible practice, or that the practice is no longer up to standard (UTS) − date of start of follow up)/365.25; †assessed as fractures during UTS or non‐UTS periods; ‡DMARDs include methotrexate, ciclosporin, sulfasalazine, and tumour necrosis factor α inhibitors.

Incidence and determinants of fracture in patients with arthritis

The patients with arthritis and control subjects experienced 129 (6.7%) and 6910 (3.3%) first fractures during the follow up period, respectively (p<0.001). In female and male subjects, effect modification between the diagnosis of arthritis and the age category at the start of follow up on fracture risk was noted (p = 0.05 and p = 0.04, respectively; table 4). However, we did not observe effect modification by sex within the stratified age categories. The IRRs (95% CIs) for first fracture among female and male subjects with arthritis, compared with controls, according to age at the start of follow up, were 1.49 (0.91 to 2.31) for age <10 years, 3.13 (2.21 to 4.33) at 10–15 years, 1.75 (1.18 to 2.51) at 15–20 years, 1.40 (0.91 to 2.08) at 20–45 years, and 3.97 (2.23 to 6.59) at >45 years. There were no differences in these results when the subjects with “Other” arthritis codes were excluded from the analysis. In those initiating follow up at <20 years of age, the median disease duration at the time of first fracture was 3.6 (0.2–18.4) years.

Table 4 Incident rate ratio* for first fracture in the subjects with arthritis compared with healthy controls.

Age category (years) Female subjects Male subjects Combined
<10 1.94 (0.96 to 3.48) 1.17 (0.54 to 2.24) 1.49 (0.91 to 2.31)
10–15 3.71 (1.96 to 6.42) 2.79 (1.80 to 4.15) 3.13 (2.21 to 4.33)
15–20 1.37 (0.50 to 3.02) 1.89 (1.21 to 2.84) 1.75 (1.18 to 2.51)
20–45 1.57 (0.63 to 3.26) 1.25 (0.74 to 1.98) 1.40 (0.91 to 2.08)
>45 4.25 (2.07 to 7.89) 3.15 (1.00 to 7.57) 3.97 (2.23 to 6.59)
Test for heterogeneity p = 0.05 p = 0.04 p<0.001

*IRR (95% confidence interval); †when models were stratified by sex and age category at the start of follow up, we did not identify effect modification between the diagnosis of arthritis and sex on the risk of fracture in any age category.

In multivariable Poisson regression models in the subjects with arthritis accounting for age at the start of follow up and sex of the subject, there was a suggestion of a protective effect of DMARD treatment on fracture incidence (IRR 0.34; 95% CI 0.10 to 1.0; p = 0.063), but this effect was not significant. No significant associations between glucocorticoid (IRR 0.78; 95% CI 0.34 to 1.77; p>0.2) or NSAID treatment (IRR 0.81; 95% CI 0.57 to 1.17; p>0.2) were seen in similar multivariable Poisson regression models.

Analyses were performed to determine whether alterations in the diagnostic criteria for arthritis would impact fracture risk estimates (table 5). In subjects with arthritis diagnosed during UTS periods, we did not identify effect modification by age at the start of follow up. Effect modification by age at the start of follow up was present when considering subjects with two diagnostic codes consistent with arthritis at least 6 weeks apart (64% female), and subjects with arthritis who received NSAID treatment (58% female).

Table 5 Sensitivity of fracture incidence rate ratio* estimates to more restrictive assessments of arthritis diagnoses.

Age category (years) During UTS† Two diagnoses‡ NSAIDs prescribed
(n = 844) (n = 381) (n = 1047)
<10 1.84 (1.09 to 2.93) 0.98 (0.20 to 2.86) 1.34 (0.58 to 2.65)
10–15 2.86 (1.79 to 4.36) 2.29 (0.99 to 4.54) 2.54 (1.55 to 3.94)
15–20 1.86 (1.06 to 3.02) 1.32 (0.43 to 3.10) 1.66 (1.02 to 2.55)
20–45 1.03 (0.28 to 2.64) 1.40 (0.77 to 2.36)
>45 5.56 (1.79 to 13.17) 3.60 (1.63 to 6.95)
Test for heterogeneity p>0.2 p = 0.04 p = 0.09
Combined estimate§ 2.15 (1.65 to 2.80)

*IRR (95% confidence interval); †subjects ⩾20 years at the start of follow up were not analysed because of the small number of subjects with arthritis (n = 10, no fractures); ‡more than 6 weeks apart; §the Mantel‐Haenszel combined estimate is provided when there is homogeneity across age category strata.

Assessment of fracture by skeletal location

Table 6 shows fracture sites in the arthritis and control groups. The anatomical distribution of fractures was consistent with a prior report using the GPRD,27 with significant differences in the proportions of fracture between the subjects with arthritis and controls at the forearm and wrist (p<0.001), humerus and elbow (p = 0.003), femur (p = 0.005), lower leg (p<0.001), foot (p = 0.007), and “other” (p<0.001) categories.

Table 6 Incidence of first fracture by anatomical location.

Fracture sites Arthritis group Control group p Value
No (%) No (%)
Vertebral 1 (0.05) 68 (0.03) >0.2
Forearm* and wrist 36 (1.86) 1686 (0.81) <0.001
Humerus and elbow 11 (0.57) 435 (0.21) 0.003
Femur 4 (0.21) 72 (0.03) 0.005
Lower leg and ankle† 18 (0.93) 778 (0.38) <0.001
Hand/finger 11 (0.57) 1378 (0.67) >0.2
Foot 14 (0.72) 706 (0.34) 0.007
Scapula/clavicle 7 (0.36) 397 (0.19) 0.09
Pelvis 0 (–) 32 (0.02) >0.2
Rib 2 (0.10) 180 (0.09) >0.2
Skull‡ 9 (0.46) 524 (0.25) 0.07
Other§ 16 (0.83) 654 (0.32) <0.001
Total 129 (6.65) 6910 (3.34) <0.001

*Radius and ulna; †tibia, fibula, ankle; ‡includes fractures of facial bones and nose; §includes recorded fractures that were not specific to individual bones, such as “arm” or “leg” fractures.

Significance of fractures before the follow up interval

Fractures before the start of follow up occurred in 137 patients with arthritis and 6686 controls. Among those subjects, 12 (8.6%) patients with arthritis and 536 (6.5%) controls had a fracture during the follow up interval (p>0.2). In a multivariable Poisson regression model adjusting for arthritis diagnosis, sex, and age category at the start of follow up, a fracture before the start of follow up was a significant risk factor for future fracture (IRR 1.86; 95% CI 1.70 to 2.03; p<0.001). A similar model that included a prior fracture by arthritis interaction term showed no significant interaction (p>0.2), indicating that the added risk of fracture (86%) attributable to prior fractures is similar between subjects with arthritis and healthy controls.

Discussion

As far as we know, our study is the first to examine population based fracture risk in a large cohort of subjects with childhood arthritis. We demonstrated a significantly increased fracture risk in subjects with arthritis that was most pronounced during the adolescent years, when fractures are common occurrences, and after the age of 45 years, when bone mass begins to decline. The IRR for first fracture among children aged <10 years and adults between 20 and 45 years of age at the start of follow up suggested an increased fracture risk of 49% and 40%, respectively, but this did not achieve statistical significance. Fractures were more common in subjects with arthritis at clinically significant sites, such as the humerus, forearm, femur, and lower leg. The fracture risk estimates were robust in sensitivity analyses designed to maximise the specificity of the arthritis group. The sex distribution of subjects with arthritis in the sensitivity analyses was consistent with that seen in the tertiary care setting.29

The increased risks for fracture observed in this study are comparable to those seen in other high risk groups.16,17,18,20,30,31,32,33 For example, van Staa et al found that patients with inflammatory bowel disease have a 59% increase in hip fracture risk.16 In addition, because of the high rates of fractures among children in general, even modest relative risks translate into large risk differences.

About one third of children have a fracture. Using the GPRD, Cooper et al found that fractures are more common in boys, particularly around the time of puberty.27 Peak fracture incidence occurred at 13–15 years in boys and 10–12 years in girls. Their estimates of forearm fracture incidence were similar to those in a study by Khosla et al in Minnesota,15 which attests to the generalisability of our findings to populations outside the United Kingdom. Our study is strengthened by the use of the same fracture codes as those used by Cooper et al (van Staa TP, personal communication).

Throughout growth, increases in bone density and dimensions result in increased bone strength. The risk of fracture seems to be dependent on these structural determinants of bone integrity. Goulding and colleagues demonstrated that children with forearm fractures had lower dual x ray absorptiometry (DXA) derived bone mass than controls.34,35 In a pQCT study examining girls with forearm fractures, Skaggs et al found that despite similar cortical and trabecular density, the cross sectional area of the radius was 8% lower in the fracture group, consistent with diminished bone strength.36

Recent studies in children with JIA have demonstrated variable bone deficits,6 possibly related to differences among arthritis groups, suboptimal control populations, and challenges in the analysis and interpretation of DXA and pQCT results. In general, studies have been limited by the confounding effect of short stature on DXA estimates of bone health. pQCT derived trabecular volumetric density is thought to be less susceptible to confounding by skeletal size, and the precise measurement of cortical dimensions allows for an assessment of bone strength. Roth et al examined children with active JIA and found low trabecular volumetric density in subjects with polyarticular disease, and both smaller periosteal dimensions and thinner cortices in all JIA subtypes.7 These deficits occurred in association with low muscle mass, consistent with a bone disorder secondary to sarcopenia.37 The authors suggested that effective arthritis treatment and physical activity promotion may be the optimal methods to enhance bone health in JIA.

Coordination and sedentariness also influence fracture susceptibility.38,39,40 In JIA, flexion contractures, muscle weakness, and musculoskeletal pain may contribute to an abnormal gait, poor balance, and increased risk of injury.41,42 Even in the absence of active arthritis, gait may be abnormal and peak impact during jumping may be increased and imbalanced.43 Fusion or abnormal range of motion of the wrist in arthritis may unfavourably alter forces on an outstretched arm during a fall, amplifying the risk of fracture. The extent to which lower physical activity contributes to the risk of fracture in children with arthritis is unknown.44

There are limitations to our study that should be mentioned. Firstly, we chose a broad definition of childhood arthritis to identify our study population. The most common diagnostic codes were “arthritis” and “synovitis,” which potentially reflect a lack of familiarity among primary care providers with the classification of JIA.45 Also, our study group may include children with brief episodes of reactive arthritis. However, the lack of specificity in the diagnostic codes used to identify subjects with arthritis would probably reduce the observed fracture risk. Secondly, adults who communicated a history of arthritis during childhood might not be representative because those with severe or persistent disease may have been more likely to report the diagnosis. This potential for recall bias does not negate the high fracture risk in adults reporting childhood onset arthritis, but the finding may not be generalisable to those with a history of mild or inactive disease. Additionally, these subjects were likely to have received chronic glucocorticoid treatment, placing them at increased risk for fracture.46

We were disappointed by the paucity of detail in the records documenting DMARD use. Possibly, DMARDs were more commonly prescribed by rheumatologists and not entered into the GPRD medication record, although this does not appear to be the case in adults with RA.47 A survey recently revealed that 82% of family practitioners were more confident about managing arthritis in adults than in children.45 The sparse documentation of DMARD use precluded a meaningful analysis of fracture incidence in subjects with arthritis exposed to drugs potentially associated with fracture risk. There was a suggestion that methotrexate use may be protective against fracture, but this effect was not statistically significant.

Three possibilities may explain our the observed lack of association between DMARD use and fractures. Firstly, DMARD and glucocorticoid use may be protective, because improved disease control may protect against poor bone accrual or frank bone loss. Recent studies in adults with RA support this hypothesis.48,49 Haugeberg and colleagues recently reported that hand bone loss in RA was associated with mean C reactive protein and rheumatoid factor status, but not with age, Health Assessment Questionnaire score, or glucocorticoid use.48 In a randomised clinical trial, Haugeberg et al found that glucocorticoid use was protective against bone loss in the hand.49 In the glucocorticoid treated group, the relation between C reactive protein levels and bone loss was reduced. Secondly, some subjects with a history of severe but inactive disease may be represented in the database. These subjects were likely to remain at high risk for fracture owing to a history of poor bone accrual or low peak bone mass. Thirdly, differential misclassification of DMARD or glucocorticoid treated subjects as non‐treated would potentially bias results toward the null hypothesis if these subjects were, in fact, at higher risk for fracture.

If the documentation of DMARD and glucocorticoid use in our study was accurate, the cohort presented would represent one with mild and inactive disease, presumably at lower risk of osteoporosis than more severely affected subjects with JIA. Alternatively, infrequent DMARD and glucocorticoid use in active disease would probably contribute to poor bone accrual and heightened fracture risk. Each of these two possibilities may affect the generalisability of our study. Additionally, the lack of anthropometric data in the subject records, as has been noted in a prior paediatric GPRD study,50 precluded an analysis of body mass index as a predictor of fracture risk.

Our study sample includes the period between 1987 and 2002. Therefore, adults and the majority of children did not benefit from biological therapies. Recent improvements in disease control associated with the availability of effective drugs may protect against bone loss, augmenting peak bone mass in afflicted children. However, recent studies demonstrate a continued threat to skeletal integrity, even in the era of biological treatment.7,14

In conclusion, childhood arthritis is associated with a substantially increased risk of fracture that is most marked during adolescence and over the age of 45 years. Although studies of bone health in JIA have been limited by suboptimal imaging techniques and analytical challenges, this study demonstrates a significant clinical problem requiring further study. The immediate goal should be to identify children with JIA at high risk for fracture to target for clinical trials.

Acknowledgements

The Nickolett Foundation for JRA Research provided generous support for data acquisition (JMB and MBL). This project was performed as part a Clinical Investigator Fellowship Award from the American College of Rheumatology Research and Education Foundation (JMB).

Max Herlim, from the Center for Clinical Epidemiology and Biostatistics, provided expert assistance in data acquisition.

Abbreviations

CI - confidence interval

DMARDs - disease modifying antirheumatic drugs

DXA - dual x ray absorptiometry

GPRD - General Practice Research Database

IRR - incidence rate ratio

JIA - juvenile idiopathic arthritis

NSAIDs - non‐steroidal anti‐inflammatory drugs

OXMIS - Oxford Medical Information Systems

pQCT - peripheral quantitative computed tomography

RA - rheumatoid arthritis

UTS - up to standard

Footnotes

Competing interest: Dr Lewis receives research support from Johnson & Johnson, GlaxoSmithKline, and Takeda Pharmaceuticals North America. He also serves as a consultant for Wyeth Pharmaceuticals, Merck & Co, Elan Pharmaceuticals, Synta Pharmaceuticals, Palatin Technologies, and Berlex. None of the other authors have any conflict of interest.

References

  • 1.Gewanter H L, Roghmann K J, Baum J. The prevalence of juvenile arthritis. Arthritis Rheum 198326599–603. [DOI] [PubMed] [Google Scholar]
  • 2.Manners P J, Diepeveen D A. Prevalence of juvenile chronic arthritis in a population of 12‐year‐old children in urban Australia. Pediatrics 19969884–90. [PubMed] [Google Scholar]
  • 3.Bailey D A, McKay H A, Mirwald R L, Crocker P R, Faulkner R A. A six‐year longitudinal study of the relationship of physical activity to bone mineral accrual in growing children: the university of Saskatchewan bone mineral accrual study. J Bone Miner Res 1999141672–1679. [DOI] [PubMed] [Google Scholar]
  • 4.Gilsanz V, Kovanlikaya A, Costin G, Roe T F, Sayre J, Kaufman F. Differential effect of gender on the sizes of the bones in the axial and appendicular skeletons. J Clin Endocrinol Metab 1997821603–1607. [DOI] [PubMed] [Google Scholar]
  • 5.Gilsanz V, Skaggs D L, Kovanlikaya A, Sayre J, Loro M L, Kaufman F.et al Differential effect of race on the axial and appendicular skeletons of children. J Clin Endocrinol Metab 1998831420–1427. [DOI] [PubMed] [Google Scholar]
  • 6.Burnham J M, Leonard M B. Bone disease in pediatric rheumatologic disorders. Curr Rheumatol Rep 2004670–78. [DOI] [PubMed] [Google Scholar]
  • 7.Roth J, Palm C, Scheunemann I, Ranke M B, Schweizer R, Dannecker G E. Musculoskeletal abnormalities of the forearm in patients with juvenile idiopathic arthritis relate mainly to bone geometry. Arthritis Rheum 2004501277–1285. [DOI] [PubMed] [Google Scholar]
  • 8.Bianchi M L, Cimaz R, Bardare M, Zulian F, Lepore L, Boncompagni A.et al Efficacy and safety of alendronate for the treatment of osteoporosis in diffuse connective tissue diseases in children: a prospective multicenter study. Arthritis Rheum 2000431960–1966. [DOI] [PubMed] [Google Scholar]
  • 9.Maenpaa H, Savolainen A, Lehto M U, Belt E A. Multiple stress fractures in a young girl with chronic idiopathic arthritis. Joint Bone Spine 200168438–442. [DOI] [PubMed] [Google Scholar]
  • 10.Maenpaa H M, Soini I, Lehto M U, Belt E A. Insufficiency fractures in patients with chronic inflammatory joint diseases. Clin Exp Rheumatol 20022077–79. [PubMed] [Google Scholar]
  • 11.Kuemmerle‐Deschner J B, Dammann F, Niethammer D, Dannecker G E. Stress fractures: diagnostic pitfalls in juvenile idiopathic arthritis. Rheumatology (Oxford) 2001401313–1314. [DOI] [PubMed] [Google Scholar]
  • 12.Rudler M, Pouchot J, Paycha F, Gentelle S, Grasland A, Vinceneux P. Low dose methotrexate osteopathy in a patient with polyarticular juvenile idiopathic arthritis. Ann Rheum Dis 200362588–589. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Varonos S, Ansell B M, Reeve J. Vertebral collapse in juvenile chronic arthritis: its relationship with glucocorticoid therapy. Calcif Tissue Int 19874175–78. [DOI] [PubMed] [Google Scholar]
  • 14.Lien G, Selvaag A M, Flato B, Haugen M, Vinje O, Sorskaar D.et al A two‐year prospective controlled study of bone mass and bone turnover in children with early juvenile idiopathic arthritis. Arthritis Rheum 200552833–840. [DOI] [PubMed] [Google Scholar]
  • 15.Khosla S, Melton L J, 3rd, Dekutoski M B, Achenbach S J, Oberg A L, Riggs B L. ncidence of childhood distal forearm fractures over 30 years: a population‐based study. JAMA 20032901479–1485. [DOI] [PubMed] [Google Scholar]
  • 16.van Staa T P, Cooper C, Brusse L S, Leufkens H, Javaid M K, Arden N K. Inflammatory bowel disease and the risk of fracture. Gastroenterology 20031251591–1597. [DOI] [PubMed] [Google Scholar]
  • 17.van Staa T P, Cooper C, Leufkens H G, Bishop N. Children and the risk of fractures caused by oral corticosteroids. J Bone Miner Res 200318913–918. [DOI] [PubMed] [Google Scholar]
  • 18.Van Staa T P, Leufkens H G, Abenhaim L, Zhang B, Cooper C. Use of oral corticosteroids and risk of fractures. J Bone Miner Res 200015993–1000. [DOI] [PubMed] [Google Scholar]
  • 19.van Staa T P, Wegman S, de Vries F, Leufkens B, Cooper C. Use of statins and risk of fractures. JAMA 20012851850–1855. [DOI] [PubMed] [Google Scholar]
  • 20.Souverein P C, Webb D J, Petri H, Weil J, Van Staa T P, Egberts T. Incidence of fractures among epilepsy patients: a population‐based retrospective cohort study in the General Practice Research Database. Epilepsia 200546304–310. [DOI] [PubMed] [Google Scholar]
  • 21.Walley T, Mantgani A. The UK General Practice Research Database. Lancet 19973501097–1099. [DOI] [PubMed] [Google Scholar]
  • 22.Nazareth I, King M, Haines A, Rangel L, Myers S. Accuracy of diagnosis of psychosis on general practice computer system. BMJ 199330732–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Nazareth I, King M, Haines A, Tai S S, Hall G. Care of schizophrenia in general practice. BMJ 1993307910. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Jick S S, Kaye J A, Vasilakis‐Scaramozza C, Garcia Rodriguez L A, Ruigomez A, Meier C R.et al Validity of the general practice research database. Pharmacotherapy 200323686–689. [DOI] [PubMed] [Google Scholar]
  • 25.Jick H, Jick S S, Derby L E. Validation of information recorded on general practitioner based computerised data resource in the United Kingdom. BMJ 1991302766–768. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Van Staa T P, Abenhaim L, Cooper C, Zhang B, Leufkens H G M. The use of a large pharmacoepidemiological database to study exposure to oral corticosteroids and risk of fractures: validation of study population and results. Pharmacoepidemiol Drug Saf 20009359–366. [DOI] [PubMed] [Google Scholar]
  • 27.Cooper C, Dennison E M, Leufkens H G, Bishop N, van Staa T P. Epidemiology of childhood fractures in Britain: a study using the general practice research database. J Bone Miner Res 2004191976–1981. [DOI] [PubMed] [Google Scholar]
  • 28.Foster C S. Diagnosis and treatment of juvenile idiopathic arthritis‐associated uveitis. Curr Opin Ophthalmol 200314395–398. [DOI] [PubMed] [Google Scholar]
  • 29.Bowyer S, Roettcher P. Pediatric rheumatology clinic populations in the United States: results of a 3 year survey. Pediatric Rheumatology Database Research Group. J Rheumatol 1996231968–1974. [PubMed] [Google Scholar]
  • 30.Bernstein C N, Blanchard J F, Leslie W, Wajda A, Yu B N. The incidence of fracture among patients with inflammatory bowel disease. A population‐based cohort study. Ann Intern Med 2000133795–799. [DOI] [PubMed] [Google Scholar]
  • 31.Rovner A J, Zemel B S, Leonard M B, Schall J I, Stallings V A. Mild to moderate cystic fibrosis is not associated with increased fracture risk in children and adolescents. J Pediatr 2005147327–331. [DOI] [PubMed] [Google Scholar]
  • 32.Hogler W, Wehl G, van Staa T, Meister B, Klein‐Franke A, Kropshofer G. Incidence of skeletal complications during treatment of childhood acute lymphoblastic leukemia: Comparison of fracture risk with the general practice research database. Pediatr Blood Cancer. 2005 Nov 29; [Epub ahead of print]. [DOI] [PubMed]
  • 33.van der Sluis I M, van den Heuvel‐Eibrink M M, Hahlen K, Krenning E P, de Muinck Keizer‐Schrama S M. Altered bone mineral density and body composition, and increased fracture risk in childhood acute lymphoblastic leukemia. J Pediatr 2002141204–210. [DOI] [PubMed] [Google Scholar]
  • 34.Goulding A, Jones I E, Taylor R W, Manning P J, Williams S M. More broken bones: a 4‐year double cohort study of young girls with and without distal forearm fractures. J Bone Miner Res 2000152011–2018. [DOI] [PubMed] [Google Scholar]
  • 35.Goulding A, Jones I E, Taylor R W, Williams S M, Manning P J. Bone mineral density and body composition in boys with distal forearm fractures: a dual‐energy x‐ray absorptiometry study. J Pediatr 2001139509–515. [DOI] [PubMed] [Google Scholar]
  • 36.Skaggs D L, Loro M L, Pitukcheewanont P, Tolo V, Gilsanz V. Increased body weight and decreased radial cross‐sectional dimensions in girls with forearm fractures. J Bone Miner Res 2001161337–1342. [DOI] [PubMed] [Google Scholar]
  • 37.Schoenau E, Neu C M, Beck B, Manz F, Rauch F. Bone mineral content per muscle cross‐sectional area as an index of the functional muscle‐bone unit. J Bone Miner Res 2002171095–1101. [DOI] [PubMed] [Google Scholar]
  • 38.Ma D, Jones G. Television, computer, and video viewing; physical activity; and upper limb fracture risk in children: a population‐based case control study. J Bone Miner Res 2003181970–1977. [DOI] [PubMed] [Google Scholar]
  • 39.Ma D, Morley R, Jones G. Risk‐taking, coordination and upper limb fractures in children: a population based case‐control study. Osteoporos Int 200415633–638. [DOI] [PubMed] [Google Scholar]
  • 40.Goulding A, Jones I E, Taylor R W, Piggot J M, Taylor D. Dynamic and static tests of balance and postural sway in boys: effects of previous wrist bone fractures and high adiposity. Gait Posture 200317136–141. [DOI] [PubMed] [Google Scholar]
  • 41.Brostrom E, Hagelberg S, Haglund‐Akerlind Y. Effect of joint injections in children with juvenile idiopathic arthritis: evaluation by 3D‐gait analysis. Acta Paediatr 200493906–910. [PubMed] [Google Scholar]
  • 42.Brostrom E, Nordlund M M, Cresswell A G. Plantar‐ and dorsiflexor strength in prepubertal girls with juvenile idiopathic arthritis. Arch Phys Med Rehabil 2004851224–1230. [DOI] [PubMed] [Google Scholar]
  • 43.Myer G D, Brunner H I, Melson P G, Paterno M V, Ford K R, Hewett T E. Specialized neuromuscular training to improve neuromuscular function and biomechanics in a patient with quiescent juvenile rheumatoid arthritis. Phys Ther 200585791–802. [PubMed] [Google Scholar]
  • 44.Henderson C J, Lovell D J, Specker B L, Campaigne B N. Physical activity in children with juvenile rheumatoid arthritis: quantification and evaluation. Arthritis Care Res 19958114–119. [DOI] [PubMed] [Google Scholar]
  • 45.Freed G L, Jee S, Stein L, Spera L, Clark S J. Comparing the self‐reported referral and management preferences of pediatricians and family physicians for children with juvenile rheumatoid arthritis. J Rheumatol 2003302700–2704. [PubMed] [Google Scholar]
  • 46.van Staa T P, Leufkens H G, Abenhaim L, Zhang B, Cooper C. Oral corticosteroids and fracture risk: relationship to daily and cumulative doses. Rheumatology (Oxford) 2000391383–1389. [DOI] [PubMed] [Google Scholar]
  • 47.Edwards C J, Arden N K, Fisher D.et al The changing use of disease‐modifying anti‐rheumatic drugs in individuals with rheumatoid arthritis from the United Kingdom General Practice Research Database. Rheumatology (Oxford) 2005441394–1398. [DOI] [PubMed] [Google Scholar]
  • 48.Haugeberg G, Green M J, Quinn M A.et al Hand bone loss in early undifferentiated arthritis. Evaluating BMD loss before the development of rheumatoid arthritis. Ann Rheum Dis 200665736–740. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Haugeberg G, Strand A, Kvien T K, Kirwan J R. Reduced loss of hand bone density with prednisolone in early rheumatoid arthritis: results from a randomized placebo‐controlled trial. Arch Intern Med 20051651293–1297. [DOI] [PubMed] [Google Scholar]
  • 50.Schlienger R G, Jick S S, Meier C R. Inhaled corticosteroids and the risk of fractures in children and adolescents. Pediatrics 2004114469–473. [DOI] [PubMed] [Google Scholar]

Articles from Annals of the Rheumatic Diseases are provided here courtesy of BMJ Publishing Group

RESOURCES