Table 5.

Multivariable logistic regression model for anti-MDA5 autoantibody-associated juvenile myositis compared with other myositis autoantibody groups

Variablea	Odds ratio	95% confidence interval	P-value
Anti-MDA5 (n =34) autoantibody-positive vs anti-TIF1 autoantibody-positive (n =155) JIIM
Weight loss	9.2	1.3, 17.2	<0.0001
Arthritis	6.5	2.2, 24.6	0.002
Arthralgia	4.2	3.9, 17.2	0.026
Dysphagia	0.19	0.06, 0.55	0.004
Falling episodes	0.26	0.08, 0.74	0.017
Likelihood ratio chi-square 64.9, P<0.0001, percent concordant 86.9, c statistic 0.884
Anti-MDA5 autoantibody-positive (n =31) vs anti-NXP2 autoantibody-positive (n =74) JIIM
Weight loss	13.6	3.9, 58.8	0.0001
Arthritis	9.5	2.5, 46.2	0.002
Dysphagia	0.12	0.02, 0.61	0.018
Likelihood ratio chi-square 58.4, P<0.0001, percent concordant 88.8, c statistic 0.893
Anti-MDA5 autoantibody-positive (n =23) vs anti-synthetase autoantibody-positive (n =14) JIIM
Interstitial lung disease	0.058	0.002, 0.57	0.032
Serum creatine kinase level	0.063	0.003, 0.54	0.027
Delay to diagnosis	0.063	0.003, 0.54	0.029
Likelihood ratio chi-square 25.9, P<0.0001, percent concordant 88.8, c statistic 0.916
Anti-MDA5 autoantibody-positive (n =33) vs MSA/MAA-negative (n =54) JIIM
Arthritis	6.3	1.4, 35.6	0.023
Arthralgia	6.1	1.4, 34.9	0.025
Weight loss	5.6	1.6, 21.7	0.008
Periungual capillary changes	5.2	1.2, 27.9	0.039
Likelihood ratio chi-square 49.2, P<0.0001, percent concordant 88.8, c statistic 0.896

^aThe top variables from the pruned Random Forests models were used in the logistic regression models. Only subjects with complete data were used in the analysis. Note that 70 juvenile myositis patients with other autoantibodies were not included in this analysis.

ARS: aminoacyl-tRNA synthetase; IQR: interquartile range; JIIM: juvenile myositis; MAA: myositis associated autoantibody; MSA: myositis specific autoantibody; MDA5: melanoma differentiation-associated gene 5; NXP2: nuclear matrix protein 2; TIF1: transcriptional intermediary factor 1.