Table 4.

Predictors of new work disability in patients with RA who were working at the start of anti-TNF therapy

	Univariate analysis		Multivariate analyses
	OR (95% CI)	P-value	OR (95% CI)	P-value	FMI
Age	1.02 (1.00, 1.04)	0.100	1.02 (0.99, 1.04)	0.188	0.001
Gender, female vs male	1.01 (0.65, 1.58)	0.961
Disease duration at baseline	1.01 (0.99, 1.04)	0.232
HAQ score	2.79 (1.89, 4.12)	<0.001	2.86 (1.90, 4.30)	<0.001	0.016
DAS-28 remission at 6 months	0.55 (0.31, 0.99)	0.046	0.75 (0.41, 1.39)	0.361	0.109
DAS-28 good response at 6 months	0.66 (0.42, 1.04)	0.073
Manual job	2.31 (1.52, 3.52)	<0.001	2.53 (1.64, 3.91)	<0.001	0.000

FMI: fraction of missing information. This measures the relative increase in the uncertainty about the coefficient of interest due to missing data. For example, had we had complete data on all subjects for all variables, the variance of the estimate of the effect of age would decrease by nearly 0.1%, whilst the variance of the estimate of the effect of DAS-28 remission would decrease by 11%. In these analyses, only those working patients at baseline (n = 1108) who were either working or work disabled at follow-up were included in the analyses.