Methods for analyzing observational longitudinal prognosis studies for rheumatic diseases: a review & worked example using a clinic-based cohort of juvenile dermatomyositis patients

Table 1 Predictors identified from the joint multivariate model of MDAS and SDAS

Predictor	Outcome	Predictor Estimate	Standard Error	p
bDAS	MDAS	0.0363	0.0316	0.25
bDAS	SDAS	0.0214	0.0342	0.53
Time^a*bDAS	MDAS	1.0140	0.1321	<0.0001
Time^a*bDAS	SDAS	0.0362	0.1261	0.77
Steroid^b	MDAS	-0.0919	0.0753	0.22
Steroid^b	SDAS	0.1211	0.0688	0.08

bDAS Baseline DASm measurement; Time *bDAS denotes the crossing of a time term with bDAS
^aThe shapes of the MDAS and SDAS models are defined using 2 time terms each (fractional polynomials). MDAS and SDAS crossed with their common time form (p = –1,) of their respective fractional polynomials (see Additional file 1: Appendix)
^bSteroid treatment from 3 months before each occasion of DASm measurement, i.e., it is a time-varying predictor. Significant results have been bolded