Session Information
Date: Tuesday, November 7, 2017
Title: Pediatric Rheumatology – Clinical and Therapeutic Aspects Poster III: Juvenile Arthritis
Session Type: ACR Poster Session C
Session Time: 9:00AM-11:00AM
Background/Purpose: We compare the effectiveness of two approaches to treat Juvenile idiopathic arthritis (JIA): early combination of biologic and non-biologic disease modifying anti rheumatic drugs (b+nbDMARDs) vs. non-biologic DMARDs (nbDMARDs) alone using a novel statistics causal inference method applied to an existing observational study. Other commonly used causal inference methods are also applied and the results are compared.
Methods: A study of health related quality of life (HRQoL) study enrolled 193 children age 2-16 with newly diagnosed (< 6 month) JIA treated at a large pediatric rheumatology clinic in United States between October 2008 and August 2012, and prospectively followed the participants at 6 and 12 months after enrollment. The current study retrospectively extracted medication prescriptions to the HRQoL participants from the date of diagnosis to the 12 month follow up visit. Patient with polyarticular forms of JIA (pJIA) and on DMARDs at the study baseline were eligible for the current study. The study endpoint was Juvenile Arthritis Disease Activity Score (JADAS) at the 6 months of follow up visit. Different statistics causal inference methods are considered, including propensity score sub-classification, inverse treatment probability weighting (IPTW), regression adjustment, and a newly proposed Bayesian Gaussian Process (GP) causal inference method that is designed to model outcome and while matched on pre-treatment factor simultaneously.
Results: At the baseline visit, 111 children were treated with DMARDs: 32 (29%) with b+nbDMARDs, 66 (59%) nbDMARD and 13 (12%) bDMARDs only. Table 1 provides the baseline characteristics of patients by treatment group. Patients on b+nbDMARDs had higher JADAS scores at the baseline, with mean±SD of 18.66±11.69 vs.11.46±8.54 (Student P value <.02), and higher rate of RF positive (28.1% vs. 10.6%, Chisq P<.05). Controlled for treatment selection bias, Bayesian GP method find children treated with b+nbDMARDs show 3.83 points improvement (95% confidence interval of 0.14-7.53) in JADAS than those treated with nbDMARDs at 6 month. Other causal inference methods suggested similar effectiveness but with more uncertainty.
Conclusion: The early initiation of b+nbDMARDs approach improves clinical outcomes at 6 months more effectively than the nbDMARDs alone strategy in children with newly onset of pJIA.
Table 1. Baseline Patient Characteristics by Treatment Group
|
nbDMARDs (N = 66) |
b+nbDMARDs (N = 32) |
bDMARDs (N = 13) |
|||||||||||
|
N |
Col% |
Row% |
N |
Col% |
Row% |
N |
Col% |
Row% |
|||||
|
JIA Subtype* |
|||||||||||||
|
Poly RF- |
39 |
59.1 |
60.0 |
18 |
56.3 |
27.7 |
8 |
61.5 |
12.3 |
||||
|
Poly RF+ |
7 |
10.6 |
43.8 |
9 |
28.1 |
56.3 |
0 |
0.0 |
0.0 |
||||
|
Oligo |
14 |
21.2 |
82.4 |
2 |
6.3 |
11.8 |
1 |
7.7 |
5.9 |
||||
|
Other |
6 |
9.1 |
46.2 |
3 |
9.4 |
23.1 |
4 |
30.8 |
30.8 |
||||
|
White Race |
60 |
90.9 |
57.7 |
31 |
96.9 |
29.8 |
13 |
100.0 |
12.5 |
||||
|
Private Insurance |
50 |
75.8 |
58.1 |
24 |
75.0 |
27.9 |
12 |
92.3 |
14.0 |
||||
|
Mean |
Std |
Min |
Max |
Mean |
Std |
Min |
Max |
Mean |
Std |
Min |
Max |
||
|
biological age (year) |
10.00 |
4.30 |
2 |
17 |
10.01 |
4.48 |
2.20 |
16.60 |
11.37 |
3.71 |
3.17 |
16.18 |
|
|
onset age (year) |
8.92 |
4.37 |
1.3 |
15.9 |
9.04 |
4.53 |
1.1 |
15.8 |
9.85 |
3.19 |
2.7 |
13.6 |
|
|
diagnosis age (year) |
9.80 |
4.32 |
1.8 |
16.9 |
9.81 |
4.47 |
2 |
16.4 |
11.19 |
3.70 |
3 |
15.9 |
|
|
MD Global** |
2.87 |
2.49 |
0 |
10 |
4.40 |
2.57 |
0.5 |
10 |
3.41 |
2.64 |
0 |
7.5 |
|
|
Patient Wellbeing |
2.65 |
2.65 |
0 |
10 |
3.20 |
2.33 |
0 |
8 |
2.85 |
2.79 |
0 |
9 |
|
|
Active Joint Count** |
5.33 |
6.21 |
0 |
26 |
9.88 |
9.33 |
0 |
32 |
6.62 |
9.87 |
0 |
30 |
|
|
Lost Range of Motion |
4.73 |
6.78 |
0 |
26 |
6.84 |
6.89 |
0 |
26 |
7.46 |
10.69 |
0 |
32 |
|
|
C-HAQ+ |
0.53 |
0.72 |
0 |
2.75 |
0.85 |
0.84 |
0 |
2.5 |
0.55 |
0.66 |
0 |
2 |
|
|
JADAS* |
11.46 |
8.54 |
0 |
37.5 |
18.66 |
11.69 |
2 |
41.6 |
11.40 |
10.80 |
0 |
40.6 |
|
|
global pain rating |
3.04 |
2.73 |
0 |
9 |
3.77 |
2.71 |
0 |
8 |
3.15 |
2.67 |
0 |
8 |
|
|
ANA |
0.62 |
0.49 |
0 |
1 |
0.52 |
0.51 |
0 |
1 |
0.67 |
0.49 |
0 |
1 |
|
|
ESR+ |
16.33 |
17.87 |
2 |
107 |
23.03 |
20.72 |
5 |
88 |
14.23 |
12.11 |
1 |
41 |
|
Statistics comparisons performed for the b+nbDMARDs patients vs. the nbDMARDS only patients using Chi-square for categorical and student T for continuous measures: ** P value <.01; * P value <.05; + P value <.1;
To cite this abstract in AMA style:
Huang B, Morgan E, Chen C, Liu J, Adams M, Beukelman T, Brunner HI, Lovell DJ. Comparing Effectiveness of Early Initiation of Biologic Treatment for Newly Diagnosed Juvenile Idiopathic Arthritis Using a Novel Statistics Causal Inference Method Applied to Observational Data [abstract]. Arthritis Rheumatol. 2017; 69 (suppl 10). https://acrabstracts.org/abstract/comparing-effectiveness-of-early-initiation-of-biologic-treatment-for-newly-diagnosed-juvenile-idiopathic-arthritis-using-a-novel-statistics-causal-inference-method-applied-to-observational-data/. Accessed .ACR Meeting Abstracts - https://acrabstracts.org/abstract/comparing-effectiveness-of-early-initiation-of-biologic-treatment-for-newly-diagnosed-juvenile-idiopathic-arthritis-using-a-novel-statistics-causal-inference-method-applied-to-observational-data/