Table 2. Variables associated with VIP serum levels during follow-up of patients with early arthritis.

	Model 1		Model 2
	β coeff ± SE	P	β coeff ± SE	P
Gender
Male	Reference		Reference
Female	0.108±0.067	0.106	0.105	0.086
Diagnosis
RA	n.i.	n.s.	n.i.	n.s.
UA
ACPA	n.i.	n.s.	n.i.	n.s.
Age (by 10 yr)	0.05±0.019	0.013	0.05±0.018	0.008
GDA Pat	n.i.	n.s.	n.i.	n.s.
GDA Phy	n.i.	n.s.	n.i.	n.s.
HAQ	n.i.	n.s.	n.i.	n.s.
SJC	n.i.	n.s.	n.i.	n.s.
MS	n.i.	n.s.	n.i.	n.s.
DAS28	−0.027±0.014	0.045	–	–
Disease activity
Remission	–	–	Reference
Low			−0.094±0.043	0.029
Moderate			−0.187±0.049	<0.001
High			−0.113±0.057	0.046
Leflunomide(mg/d)	n.i.	n.s.	n.i.	n.s.
TNF blockers	0.183±0.075	0.01	0.165±0.065	0.011

The longitudinal analysis was performed with data (logarithmic transformation of censored VIP levels; see figure S2) from 340 visits corresponding to the 88 patients with all information available in, at least, two visits. Model 1 was fitted using the continuous value of DAS28 as measure of disease activity while model 2 includes the categorical variable based in cut-off values of this index. The average number of visits by patient was 3.9. The table shows all the variables reaching p<0.15 at the bivariate analysis (see Methods section for further information on multivariable analysis modeling). Coeff: coefficient; CI: confidence interval; RA: rheumatoid arthritis; UA: undifferentiated arthritis; ACPA: anti-citrullinated peptide antibodies; yr, year; GDA: global disease assessment; Pat: patient; Phy: physician; HAQ: health assessment questionnaire; SJC: swollen joint count; MS: morning stiffness; DAS28, 28-joint Disease Activity Score; TNF: tumor necrosis factor; n.s.: not significant; n.i.: not included.