Frequency of osteopenia in adolescents with early-onset juvenile idiopathic arthritis: A long-term outcome study of one hundred five patients




To determine the frequency of low bone mineral content (BMC) and low bone mineral density (BMD) as long-term complications in adolescents with early-onset juvenile idiopathic arthritis (JIA), and to identify disease variables, patient characteristics, and biochemical bone markers related to low bone mass.


One hundred five (87%) of 121 adolescent patients with early-onset JIA (ages 13–19 years, 80 girls and 25 boys, mean age at onset of JIA 2.8 years), from a cohort first admitted to the hospital between 1980 and 1985, were assessed after a mean disease duration of 14.2 years. BMC and BMD of the total body, the lumbar spine at L2–L4, and the femoral neck were measured by dual-energy x-ray absorptiometry. Age- and sex-specific reference values from a pooled, healthy reference population were used to calculate Z scores. Low bone mass was defined as a Z score less than −1 SD.


Among the 103 adolescent JIA patients who underwent total-body imaging, 41% had low total-body BMC and 34% had low total-body BMD. Compared with adolescent JIA patients who had normal total-body BMC, those with low BMC had lower mean weight (P < 0.001), height (P < 0.001), lean mass (P < 0.001), and remission rates (P = 0.016), had longer duration of active disease (P = 0.013), had higher numbers of active and mobility-restricted joints (P < 0.001 and P = 0.001, respectively), had more disability (P = 0.011), had higher frequencies of joint erosions (P < 0.001), and had higher erythrocyte sedimentation rates (P = 0.033). In multiple linear regression analyses of total-body BMC, 88% of the variance was explained by the duration of active disease, the number of joints with restricted mobility, the bone area, urinary deoxypyridinoline values, age, weight, and height.


Forty-one percent of the adolescents with early-onset JIA had low bone mass >11 years after disease onset. The development of low total-body BMC was related to the duration of active disease, disease severity, measures of bone resorption, weight, and height.

Osteoporosis is a public health problem worldwide. An important determinant of osteoporosis and future fracture risk is the peak bone mass achieved during the period of skeletal growth. Skeletal maturation in children is dependent on the rate of bone formation exceeding the rate of bone resorption. Peak bone mass is defined as the bone mass present at the end of skeletal maturation, which occurs after age 20 years. If a sound foundation for peak bone mass, which is partly genetically determined, is not established during the second decade of life, the young adult will experience osteopenia and an increased fracture risk (1). Children with a chronic illness such as arthritis are at risk of developing osteopenia that is influenced by (in addition to heredity and hormones) inflammation, medication, nutrition, and physical inactivity.

In several cross-sectional long-term followup studies, reduced bone mineral density (BMD) in adults with juvenile idiopathic arthritis (JIA) has been reported (2–5). Several studies (mainly cross-sectional) have demonstrated reduced BMD or bone mineral content (BMC) in children and adolescents with JIA. These findings were recently summarized in 2 review articles (6, 7).

Pepmueller et al focused on the fact that impairment of bone formation during the critical period of pubertal growth acceleration may not be reparable later in life (8). The extent of low bone mass as a long-term complication in adolescents with onset of JIA as infants or at preschool age is not well documented. In addition, few studies have been published in which disease variables and patient characteristics are related to the development of osteopenia and osteoporosis in these patients.

In order to address these issues, we examined BMD and BMC of the total body, spine, hip, and forearm as measures of long-term outcome in 105 adolescents with early-onset JIA. These adolescents are part of a cohort of patients with JIA, ages 13–31 years, who were first admitted to Rikshospitalet University Hospital between 1980 and 1985 and were followed up for a median of 14.9 years after disease onset.


Patients and controls.

The patients were recruited from a previously described cohort of 400 patients with JIA, all of whom had been admitted for the first time to the Department of Rheumatology, Rikshospitalet University Hospital, between January 1980 and September 1985 (5, 9, 10). All of the adolescent patients with JIA from this cohort who were younger than age 20 years at the time of examination were eligible to participate in the present study. One hundred five (87%) of the 121 adolescent patients (ages 13–19 years) were followed up after a mean ± SD of 14.2 ± 1.4 years of disease duration. Sixteen adolescent patients from the cohort (13%) were not included in the study: 2 patients (2%) had died by the time of followup, 3 patients (2%) could not be found, and 11 patients (9%) chose not to participate. The patients who were not included were comparable with the participants with regard to sex, onset type, age, and duration of symptoms before admission.

All of the patients with JIA from the above-mentioned cohort who were older than age 20 years at followup and whose bone mass had been measured (n = 216), and a group of healthy young Norwegian adults (n = 94) were used as comparison groups for the frequencies of low bone mass (5, 9, 10). The mean height, weight, and body mass index (BMI) in the healthy group were comparable with those described in larger Norwegian population studies (11). Nutritional status, BMD findings, and estimates of acquired peak bone mass in these 2 groups of young adults have been presented previously (5, 12). All of the subjects were white.

Ninety-five of the adolescent patients completed a questionnaire about physical activity, smoking habits, and fracture rates. Individuals matched for age, sex, and county of residence were randomly selected from the national population register to serve as controls for these variables. The patients who completed the questionnaire were comparable with those who did not with regard to age, sex, and onset type.

The Regional Ethics Committee for Medical Research and the Norwegian Radiation Protection Authority approved the study. Informed consent was obtained from all participants.


The adolescent patients with JIA met the criteria for either juvenile rheumatoid arthritis (JRA) (13) or juvenile psoriatic arthritis (14) (n = 96 and n = 9, respectively). None of the patients met the criteria for juvenile ankylosing spondylitis (15), syndrome of seronegative enthesopathy and arthropathy (16), or arthritis associated with inflammatory bowel disease. Onset of disease was defined as the date on which arthritis or systemic features were documented by a physician.

Clinical examination and chart reviews.

The patients were examined by one of the authors (BF, GL, OV, or DS) at followup. Information about the onset of disease, its course, and treatment was obtained from medical records and interviews. The clinical examination included an assessment of actively involved (swollen or tender and mobility-restricted) and affected (swollen or mobility-restricted) joints (9) and physician's global assessment of disease activity (on a 5-point Likert scale [1 = inactive, 2 = mild, 3 = moderate, 4 = severe, and 5 = very severe disease activity]) (see Table 1 for study variables).

Table 1. Characteristics of 105 adolescents with early-onset JIA*
VariableJIA patients
  • *

    JIA = juvenile idiopathic arthritis; DMARDs = disease-modifying antirheumatic drugs.

Female, no. (%)80 (76)
Age at onset of JIA, mean ± SD years2.8 ± 1.4
Systemic onset, no. (%)15 (14)
Pauciarticular onset, no. (%)73 (70)
Polyarticular onset, no. (%)17 (16)
Age at followup, mean ± SD years17.0 ± 1.8
Weight, mean ± SD kg59.7 ± 10.7
Height, mean ± SD cm167.4 ± 9.7
Body mass index, mean ± SD kg/m221.3 ± 3.3
Calcium intake, mean ± SD mg/day1,053 ± 556
Vitamin D intake, mean ± SD μg/day4.5 ± 3.0
Disease duration, mean ± SD years14.2 ± 1.4
Remission, no. (%)44 (42)
IgM rheumatoid factor positive, no. (%)2 (2)
Antinuclear antibody positive, no. (%)45 (43)
DMARDs, current user, no. (%)22 (21)
DMARDs, ever user, no. (%)84 (80)
Prednisolone, current user, no. (%)5 (5)
Prednisolone, previous user, no. (%)25 (24)

The definition of remission was based on the preliminary American College of Rheumatology (formerly, the American Rheumatism Association) criteria for remission in rheumatoid arthritis, except that our time requirement was longer. Thus, 5 or more of the following criteria had been fulfilled for at least 2 years: morning stiffness not exceeding 15 minutes, no fatigue, no joint tenderness, no swelling in joints or tendon sheaths, and erythrocyte sedimentation rate (ESR) <20 mm/hour (17). A further criterion was that no antirheumatic medication had been given during the past 2 years. The Health Assessment Questionnaire (HAQ) was used to measure disability at followup (18).

Weight was measured, on a Seca (Hamburg, Germany) digital scale, to the nearest 0.1 kg, and height was measured to the nearest 0.5 cm. Pubertal development was determined according to the criteria described by Tanner (19). To evaluate growth disturbances in the adolescents with JIA, the mean weight and height of these boys and girls were compared with those of the Norwegian reference population (20).


At followup, 22 of the 105 adolescent JIA patients were receiving ≥1 disease-modifying antirheumatic drug (DMARD). Methotrexate was the most frequently used drug (n = 16), followed by hydroxychloroquine (n = 4), sulfasalazine (n = 1), and cyclophosphamide (n = 1). Oral corticosteroids were currently being taken by 5 patients. The number of ever users of DMARDs was 84, and hydroxychloroquine was the most frequently ever used drug (n = 79), followed by gold (n = 35), methotrexate (n = 28), penicillamine (n = 12), azathioprine (n = 7), cyclophosphamide (n = 4), sulfasalazine (n = 3), and cyclosporine (n = 1). Oral corticosteroids had previously been used by 25 patients.

Laboratory measures.

ESR, hemoglobin concentration, and serum levels of albumin, alkaline phosphatase (AP), phosphate, calcium, and parathyroid hormone were measured by routine laboratory methods. Bone formation was assessed by measuring serum levels of bone-specific AP and osteocalcin. Bone resorption was assessed by measuring serum levels of C-terminal type 1 telopeptide and urinary concentrations of deoxypyridinoline (D-Pyd). Vitamin D status was assessed by measuring serum levels of 25-hydroxyvitamin D (25[OH]D) and 1,25-dihydroxyvitamin D3 (1,25[OH]2D3). Positivity for IgM rheumatoid factor was defined as titers ≥1:64, as measured at least twice by the Rose-Waaler test. Positivity for antinuclear antibodies (ANA) was defined as titers ≥1:32, as measured at least twice by indirect immunofluorescence. The presence of HLA–B27 was determined by serologic testing. Venous blood samples were obtained before noon.

Radiographic examinations.

Radiographs of the hips and ankles of all patients were obtained at followup. Radiographs of other affected joints were obtained when clinically indicated. The radiographs of peripheral joints were graded independently by 2 radiologists (VJ and KD) according to a radiographic classification system for JRA in which grade 0 = normal joints, grade 1 = suspicious changes, grade 2 = growth disturbances, grades 3 and 4 = joint erosions, and grade 5 = mutilating abnormalities (21).

Physical activity assessment.

Patients and controls matched for age, sex, and county of residence completed a self-report that included questions about smoking habits, fractures, and leisure-time physical activity. The main part of the questionnaire had been developed, evaluated for validity and reliability, and used in larger studies by the Research Center for Health Promotion, University of Bergen, Norway (22). The leisure-time physical activity measures have been found to distinguish between active and inactive subjects and are associated with fitness status. The questionnaire included all types of leisure-time activities during non-school hours: outdoor recreational activities, play, games, and sports. The level of measurement was ordinal (e.g., once a week, less than once a week) and was recorded in the data analysis so that the values reflected interval level variables (e.g., once weekly = 1.0, less than once weekly = 0.5) (23). Weight-bearing activities were defined as activities in which the heel touches the ground.

Dietary intake assessment.

Food and nutrition intake were assessed by a standardized quantitative food frequency questionnaire evaluated for reproducibility and validity in adolescents (24). Nutrient calculations were computed in accordance with the Norwegian Food Composition Table (25).

Bone mass measurements.

Bone mineral was measured in the total body, lumbar spine, hip, and forearm by means of the same dual x-ray absorptiometry (DXA) equipment (Lunar Expert-XL; GE Lunar, Madison, WI), by 1 of 2 operators. Subjects were positioned according to standard techniques. Daily measurements were made on a spine phantom to determine the variations in measurements over time. The coefficient of variation (over the year of investigation) was 0.5%. Analyses were performed using Expert-XL software version 1.63 (GE Lunar). In this study, all scan analyses were read by one of the authors (GL). The level of radiation exposure from our DXA equipment was measured, found to be very low, and was approved by the Norwegian Radiation Protection Authority. A total of 103 participants underwent all of the DXA scans. In 2 subjects, measurements of the total body and hip were not obtained (because of a prosthetic hip and technical problems, respectively). Lean mass in kilograms was measured, and BMD, BMC, and bone area were calculated for the total body, the lumbar spine at L2–L4, the femoral neck, and one-third of the distal radius.

The Lunar Expert-XL (fan-beam) produces results that correlate highly with the Lunar DPX series (pencil-beam), and the reference population data for adults can be equally applied to the 2 systems (26). For children, correlations between data from the 2 systems had not been published when we started our investigation. We therefore conducted a study to determine the correlation between osteodensitometry measurements for children, as obtained within a few days by both Lunar DPX and Expert-XL. Total-body and lumbar spine scans of 31 children between the ages of 6 and 19 years were performed. The inter-item correlation coefficients ranged from 0.98 to 0.99 (27). A Bland Altman plot was also performed (data not shown). The results indicate that the reference population data for children can be equally applied to the 2 systems.

BMD and BMC T and Z scores.

Low bone mass (osteopenia) in adult women is defined by the World Health Organization (WHO) as a value for BMD or BMC (expressed as a T score) between 1 and 2.5 SD below the mean value for young adults. Osteoporosis is defined as a BMD or BMC value greater than 2.5 SD below the mean value for young adults (28). The WHO definition cannot be applied to children, because they have yet to achieve peak bone mass. For children, the BMD value is usually interpreted as a Z score in terms of the number of SDs above or below the age-specific mean for healthy individuals (29). At present, Norwegian reference values for pediatric DXA measurements are not available. We chose to use age- and sex-specific numeric data from adolescents provided by the manufacturer (GE Lunar) to calculate Z scores for BMD and BMC using the following formula: Z score = (subject's measurement − mean measurement of the reference population)/SD of the reference population. BMC reference values for the femoral neck were available only for females, and no reference data for the forearm were available.

Bone mass in children increases with body size, and size considerations are important when assessing children with JIA, because the disease may have a negative effect on growth (30). We compared the height and weight of the GE Lunar reference population with those of the adolescent JIA patients (ages 13–19 years), but found no significant differences for either boys or girls (data not shown).

Total-body BMC.

Total-body measurements that include all skeletal regions have been recommended for studies in children. Moreover, because BMD only partially corrects for bone size and does not correct for changes in shape or thickness, it is considered preferable to use BMC (31). In this study, we chose total-body BMC as the outcome variable in analyses of factors associated with low bone mass.

Statistical analysis.

The independent samples t-test for continuous variables was used to compare growth between patients and the reference populations. Differences between patients and matched controls for fracture rates, physical activity, and smoking habits were tested by the paired samples t-test for normally distributed continuous variables, the Wilcoxon's signed rank sum test for non–normally distributed continuous variables, and McNemar's test for categorical variables. Pearson's correlation coefficient was calculated in order to measure the association between 2 normally distributed variables. Within the patient cohort, differences between 2 groups were tested by the independent sample t-test or the Mann-Whitney test for continuous variables, and by the chi-square test or Fisher's exact test for categorical variables. Corrections for the number of comparisons between patients with low total-body BMC and those with normal total-body BMC were not made, because anthropometric measurements, nutrition, disease, and treatment variables have previously been reported in the literature to be associated with variations in bone mass.

The variation in total-body BMC was analyzed by multiple regression analyses. Patient and disease characteristics were assumed to be possible predictors and were included in the model if P was less than 0.2 in unadjusted linear regression analyses. The numbers of independent variables were restricted in order to make an acceptable model in relation to sample size (32). Highly intercorrelated independent variables (r > 0.7) in the multiple variable model were avoided. To avoid the possibility of size-related artifacts, bone area, weight, and height were included in the multiple regression models, as proposed by Prentice et al (33). Forward and backward stepwise regression methods were used.

For all analyses, P values less than or equal to 0.05 (2-tailed tests) were considered significant. The statistical analysis was performed using SPSS version 10.0 (Chicago, IL).


Demographic and disease characteristics.

The characteristics and disease variables of the 105 adolescent JIA patients assessed at followup are shown in Table 1. Growth of the patients was comparable to that of the Norwegian reference population (20). Twenty (21%) of the 95 adolescent JIA patients who completed the self-report for physical activity and 16 (17%) of the 95 matched controls reported previous fractures, but the difference was not statistically significant. Leisure-time weight-bearing activities (reported as times per week) were less frequent in the JIA group than in the control group (mean ± SD 7.9 ± 4.4 versus 9.6 ± 4.8; P = 0.015) (data not shown).

Bone mineral density and bone mineral content.

BMD and BMC values (except for spine BMD) for adolescent JIA patients with active disease were statistically significantly lower than those for adolescent JIA patients in remission (Table 2).

Table 2. Bone mineral density and bone mineral content in adolescents with active juvenile idiopathic arthritis (JIA) or JIA in remission*
VariableAll patients (n = 105)Patients in remission (n = 44)Patients with active disease (n = 61)P
  • *

    Values are the mean ± SD.

  • For total body and femoral neck, n = 103.

  • Bone mineral density and bone mineral content reference populations for the adolescent radius were not available.

  • §

    A bone mineral content reference population for the femoral neck was available only for females.

Bone mineral density    
 Total-body, gm/cm21.06 ± 0.121.10 ± 0.121.04 ± 0.120.005
  Z score−0.51 ± 1.420.02 ± 1.41−0.90 ± 1.300.001
 Spine, L2–L4, gm/cm21.14 ± 0.181.17 ± 0.161.12 ± 0.190.127
  Z score0.01 ± 1.620.49 ± 1.39−0.34 ± 1.700.007
 Femoral neck, gm/cm21.00 ± 0.161.05 ± 0.160.97 ± 0.160.010
  Z score−0.27 ± 1.540.16 ± 1.55−0.59 ± 1.460.015
 One-third distal radius, gm/cm20.63 ± 0.080.65 ± 0.090.61 ± 0.070.003
Bone mineral content    
 Total-body, gm2,275.87 ± 450.862,491.43 ± 440.442,115.10 ± 390.03<0.001
  Z score−0.58 ± 1.490.10 ± 1.53−1.10 ± 1.26<0.001
 Spine, L2–L4, gm47.69 ± 10.6551.24 ± 9.7245.14 ± 10.620.003
  Z score0.19 ± 1.600.84 ± 1.39−0.28 ± 1.58<0.001
 Femoral neck, gm4.91 ± 0.925.33 ± 0.864.60 ± 0.84<0.001
  Z score§−0.33 ± 1.550.45 ± 1.37−0.73 ± 1.490.001
 One-third distal radius, gm1.56 ± 0.321.72 ± 0.311.44 ± 0.26<0.001

Frequency of reduced BMD and BMC.

Low BMD (i.e., Z score less than −1 SD according to the reference population) was more frequent in adolescents with early-onset JIA than in the comparison group of young adult patients with later-onset JIA that was selected from the same cohort (P < 0.001 for total-body, P = 0.030 for lumbar spine L2–L4, and P = 0.001 for femoral neck) (Figure 1). The distribution of total-body, lumbar spine at L2–L4, and femoral neck BMD Z scores in adolescents with active JIA or JIA in remission, young adults with active JIA or JIA in remission, and healthy young adults is shown in Figure 2. Five (5%) of the adolescent JIA patients were found to be osteoporotic (BMD Z score ≤ −2.5) in the total body, 5 (5%) in the lumbar spine at L2–L4, and 8 (8%) in the femoral neck. Low total-body BMC was observed in 42 (41%) of the 103 adolescents who underwent total- body imaging, and low lumbar spine (L2–L4) BMC was observed in 23 adolescents (22%). Low femoral neck BMC was observed in 24 (30%) of the girls. Reference data for femoral neck BMC in boys were not available.

Figure 1.

Frequency of reduced total-body, lumbar spine, and femoral neck bone mineral density (BMD) in adolescent patients with early-onset juvenile idiopathic arthritis (JIA), young adults with later-onset JIA, and healthy young adults. Reduced BMD is defined as a Z score < −1 SD according to an age- and sex-matched reference population. The frequencies of low T scores in young adult JIA patients were similar to the frequencies of low Z scores for lumbar spine (15%) and femoral neck (16%) BMD. For total-body BMD, 16% of young adult JIA patients had low T scores, and 14% had low Z scores. ∗ = P < 0.05 and ∗∗ = P ≤ 0.001 versus JIA patients ages 20–31 years and healthy young adults; † = P < 0.05 and ‡ = P ≤ 0.001 versus healthy young adults ages 20–30 years.

Figure 2.

Scatter plots showing total-body, lumbar spine, and femoral neck bone mineral density (BMD) Z scores. JIA = juvenile idiopathic arthritis. Horizontal bars show the mean values.

Factors associated with low total-body BMC.

Adolescent patients with low total-body Z scores for BMC had, on average, lower body weight, height, and lean mass, and higher levels of disease activity and severity than did those with normal Z scores for BMC (P values ranged from <0.001 to 0.033) (Table 3). Only 5 patients were current users of prednisolone, and 4 of these had low total-body BMC. The frequency of previous users of prednisolone was no higher in the patient group with low total-body BMC than in the group with normal total-body BMC (P = 0.709). Of the 74 patients who had never used prednisolone, 29 (39%) had low bone mass. The levels of bone formation markers and bone resorption markers were higher in patients with low BMC than in patients with normal BMC, but the differences were not statistically significant.

Table 3. Anthropometric measurements, disease variables, and bone markers in adolescent patients with juvenile idiopathic arthritis*
VariableLow total-body BMC (n = 42)Normal total-body BMC (n = 61)P
  • *

    Unless indicated otherwise, values are the mean ± SD. A low Z score is defined as less than 1 SD below the mean in an age- and sex-matched reference population. BMC = bone mineral content; HAQ = Health Assessment Questionnaire; ESR = erythrocyte sedimentation rate; ANA = antinuclear antibodies.

Weight, kg51.9 ± 8.065.2 ± 9.1<0.001
Height, cm160.8 ± 9.2171.6 ± 7.4<0.001
Lean mass, kg37.7 ± 7.447.1 ± 7.6<0.001
Duration of active disease, years10.3 ± 5.47.6 ± 5.30.013
Calcium intake, mg/day1,069 ± 6761,034 ± 4780.897
Vitamin D intake, μg/day4.4 ± 3.04.5 ± 3.00.887
Remission, no. (%)12 (29)32 (52)0.016
Physician's global assessment (range 1–5)2.0 ± 1.01.4 ± 0.6<0.001
Patient's global assessment (range 1–5)2.0 ± 1.01.5 ± 0.70.010
HAQ score (range 0–3)0.4 ± 0.60.1 ± 0.30.011
No. of active joints (range 0–69)2.8 ± 6.30.3 ± 1.1<0.001
No. of joints with restricted mobility7.9 ± 11.71.5 ± 3.90.001
Erosions, no. (%)17 (40)6 (10)<0.001
Radiographic changes (range 0–5)2.5 ± 1.21.8 ± 0.80.001
ESR, mm/hour13.7 ± 13.67.7 ± 7.40.033
ANA positive, no. (%)24 (57)19 (31)0.010
Serum bone-specific alkaline phosphatase, units/liter48.8 ± 38.241.9 ± 35.70.089
Serum osteocalcin, nmoles/liter4.8 ± 3.34.4 ± 3.70.334
Serum C-telopeptide 1, μg/liter10.0 ± 4.79.0 ± 4.90.077
Urinary deoxypyridinoline, nM/mM creatinine14.7 ± 7.912.8 ± 7.80.081
Prednisolone, current user, no. (%)4 (10)1 (2)0.156
Prednisolone, previous user, no. (%)9 (21)15 (25)0.709
Prednisolone, never used, no. (%)29 (69)45 (74)0.659

There were no statistically significant differences between the 2 groups in calcium and vitamin D intake (Table 3), serum levels of 25(OH)D, 1,25(OH)2D3, and parathyroid hormone, Tanner pubertal stages, or weight-bearing physical activity (data not shown).

Thirty-one (43%) of 72 patients with pauciarticular-onset, 9 (56%) of 16 patients with polyarticular-onset, and 2 (13%) of 15 patients with systemic-onset JIA who underwent total-body imaging had low total-body BMC. Low BMC was less frequent in the group with systemic onset compared with the 2 other groups (P = 0.040 and 0.023), but the difference between pauciarticular and polyarticular onset was not statistically significant (data not shown).

Determinants of total-body BMC in adolescent JIA patients.

Multiple linear regression analyses were performed to identify the most important demographic, disease-related, and biochemical variables explaining the variance in total-body BMC (Table 4). Sex, age at followup, bone area, weight and height, duration of active disease, number of joints with restricted mobility, number of active joints, physician's global assessment, patient's global assessment, HAQ score, radiographic changes, ESR, levels of urinary D-Pyd, levels of serum osteocalcin, months receiving prednisolone, and calcium intake were included in the model as independent variables.

Table 4. Patient and disease characteristics predicting total-body BMC by linear regression analysis in 103 adolescent juvenile idiopathic arthritis patients*
VariableUnivariate analysisMultiple linear regression analysis
Coefficient95% CIPCoefficient95% CIP
  • *

    BMC = bone mineral content; 95% CI = 95% confidence interval; HAQ = Health Assessment Questionnaire; ESR = erythrocyte sedimentation rate. Adjusted R2 value (88%) is the percentage of the variance of total-body BMC explained by variables found significant in the final model.

  • Final model identifying significant predictors of total-body BMC by forward stepwise regression analysis.

  • Regression coefficient from univariate and multiple regression analyses.

Sex (male = 1, female = 2)−261.9−465.0, −58.90.012
Age at followup, years72.726.2, 119.30.00327.54.7, 50.40.019
Height, cm34.728.6, 40.7<0.0017.71.2, 14.30.021
Weight, kg35.631.3, 39.9<0.00116.910.8, 23.0<0.001
Bone area, cm21.71.5, 1.9<0.0010.60.1, 1.00.011
Duration of active disease, years−31.3−46.4, −16.3<0.001−8.9−15.6, −2.20.010
No. of joints with restricted mobility−20.3−29.8, −10.7<0.001−6.1−10.3, −1.90.005
Urinary deoxypyridinoline, nM/mM creatinine−18.0−29.5, −6.40.003−8.8−14.2, −3.50.001
Serum osteocalcin, nmole/liter−18.9−44.9, 7.10.151
Physician global assessment (range 1–5)−272.4−360.6, −184.2<0.001
Patient global assessment (range 1–5)−171.7−274.7, −68.7<0.001
HAQ score (range 0–3)−381.8−578.7, −185.0<0.001
Radiographic changes (range 0–5)−141.5−223.1, −60.0<0.001
ESR, mm/hour−12.4−20.6, −4.10.004
No. of active joints−35.3−55.0, −15.50.001
Prednisolone use, months−4.0−7.5, −0.50.027
Calcium intake, mg/day0.1−0.0, 0.30.151

Some of the details in Table 4 can be explained by considering the variable duration of active disease. It can be seen that this variable was significant in both the univariate model (P < 0.001) and the adjusted model after controlling for age, height, weight, bone area, and disease variables. The multiple linear regression analysis (adjusted model) showed that a 1-year duration of active disease resulted in an expected reduced total-body BMC of 8.9 gm. In the final model, using forward stepwise regression analysis, age at followup (P = 0.019), bone area (P = 0.011), height (P = 0.021), weight (P < 0.001), duration of active disease (P = 0.010), number of joints with restricted mobility (P = 0.005), and levels of urinary D-Pyd (P = 0.001) were statistically significant predictors of total-body BMC, and explained 88% of the variance. Backward stepwise regression analysis gave essentially similar results. According to the model, increasing duration of active disease, increasing numbers of joints with restricted mobility, and increasing levels of urinary D-Pyd were associated with decreased BMC, while increasing age at followup, increasing bone area, increasing height, and increasing weight were associated with increased BMC.


In this long-term outcome study, 41% of the adolescents with early-onset JIA had low total-body BMC at followup. Low total-body BMD was observed in 34% of the patients, low lumbar spine BMD was observed in 25%, and low femoral neck BMD was observed in 31%. Low BMD was more frequent in adolescents with early-onset JIA than in young adult patients with later-onset JIA. Seventy-one percent of the patients were not being treated with corticosteroids, and 39% of these had low bone mass. Total-body BMC was related to duration of active disease, disease severity, weight, and height. Our results are in accordance with the frequencies of low total-body bone mass in non–corticosteroid-treated postpubertal females reported by Henderson et al (34). Those investigators observed low BMC in 11 (31%) of 36 patients. To our knowledge, there are no other studies in which the frequencies of low bone mass in cohorts of adolescent JIA patients are reported.

The frequencies of low BMD in our cohort were significantly higher among the adolescents with a median age at JIA onset of 2.4 years than in the young adults with a median age at JIA onset of 10.3 years. These results are in accordance with those from another study involving our cohort, showing that young age at onset was a predictor of unfavorable outcome in JRA (10). Our findings corroborate data from a study by Badley and Ansell showing unfavorable outcome, with a higher incidence of fractures, in female JRA patients in whom the disease commenced before age 5 years compared with patients with an older age at disease onset (35).

Among the 216 young adult JIA patients, 33 (15%) were found to have low BMD of the total body, 32 (15%) had low BMD of the spine at L2–L4, 33 (15%) had low BMD of the femoral neck, and 53 patients (25%) had low BMD in one-third of the distal radius. In a population-based study in Rochester, Minnesota, French et al (2) reported results in accordance with ours for BMD of the radius and total body, but they observed higher frequencies of low bone mass in the spine (28%) and femoral neck (32%). Zak et al (3) and Bartram et al (4), in cross-sectional long-term followup studies, reported frequencies of low bone mass in the spine and hip of 40–52%. Different patient populations and differences in disease activity and disease duration may explain the conflicting results of these studies.

In this group of adolescent patients with early-onset JIA, duration of active disease, the number of joints with restricted mobility, height, weight, bone area, and age at followup were found to be the most important variables explaining the variance in total-body BMC by multiple linear regression analyses. Patients with low bone mass had higher ESRs, higher frequencies of joint erosions, and greater disability compared with patients with normal bone mass. These results are in accordance with those of several other studies showing a relationship between bone loss and disease activity and between bone loss and number of involved joints in JIA patients (8, 34, 36–41).

The urinary level of D-Pyd was another important variable that explained the variance of total-body BMC by multiple linear regression analyses. If this finding of a higher urinary concentration of D-Pyd is considered together with our findings of higher serum levels of bone-specific AP, osteocalcin, and C-telopeptide 1 in patients with low total-body BMC compared with patients with normal BMC (although these findings were not statistically significant), it might indicate an increased bone turnover in these patients. Similar results have been reported by Henderson et al, who found significantly higher levels of osteocalcin and C-telopeptide 1 in patients with low total-body BMC (34). In contrast, Pepmueller et al (8) and Hillman et al (42) found decreased levels of bone-specific AP, osteocalcin, and the bone resorption marker tartrate-resistant acid phosphatase in children with active JRA, suggesting reduced bone turnover. The differences between the bone marker findings may be attributable to differences in age, corticosteroid therapy, and disease activity, all of which may complicate the interpretation of the findings.

Four of the 5 adolescent patients who were currently receiving corticosteroids had low bone mass. Previous users of corticosteroids did not have increased frequencies of low bone mass. The cumulative dose of corticosteroids, expressed as months of prednisolone treatment, did not contribute significantly to the variation in total-body BMC as assessed in the multiple regression analyses. Our data are in accordance with the results of a meta-analysis by van Staa et al, who found strong evidence suggesting that corticosteroid-induced osteoporosis is reversible after corticosteroid therapy has been stopped (43). The information about prednisolone dose schedules in our study may have been imprecise, because it was obtained from patient interviews and medical records and not from prospective assessments.

Twenty-one percent of the adolescent JIA patients reported having had 1 or more fractures, compared with 17% of their matched controls, but the difference was not statistically significant. Few studies have reported frequencies of fractures in JIA, but a study by Badley and Ansell revealed that 12% of their JRA patients had sustained at least 1 fracture (35).

The adolescent JIA patients engaged in leisure-time weight-bearing activities significantly less frequently than did the Norwegian controls. Weight-bearing activities were not found to be statistically significant as an explanatory variable for the variance in total-body BMC in the multiple linear regression analyses. However, studies of healthy children (44, 45) have shown that weight-bearing activities may be of importance in the development of bone mass.

Our study investigated the long-term outcome for 105 adolescent patients with early-onset JIA, who belonged to a cohort consisting of all the new patients with JIA who were admitted to our hospital between 1980 and 1985. Patients who are admitted to hospital tend to have more severe disease than those who are recruited from the general population. However, the age, sex, and disease onset type of the patients in the cohort were similar to those of patients in previous epidemiologic studies (46–49), indicating that the cohort and the adolescent participants were representative of patients with diagnosed JIA. Because of the health care system in Norway, which provides regular free checkups for all children, most children who had chronic rheumatic disease during the period 1980–1985 have probably been hospitalized. It has been estimated that during this period our hospital admitted ∼70% of Norwegian children with diagnosed JIA (50).

We found that in adolescents with early-onset JIA who were close to the age of skeletal maturation and expected peak bone mass, bone mass in cortical and trabecular bone was reduced. Low bone mass was related to the duration of active disease and disease severity. The possibility of skeletal catch-up growth will influence the future fracture risk of these young people. We know very little about expected catch-up growth for these patients or what factors affect the phenomenon. Studies of the osteoporotic process during different periods of the life course are needed if we are to gain further insight into the development of low bone mass, and particularly to find out whether it is possible to prevent future fractures.


We thank Howard S. Barden (GE Lunar) for help in providing us with the normative data for the reference population, and Gunn J. Hovland and Bente M. Solberg for conducting the DXA scans.