Table 3.

Cross-sectional analyses of association between TRAF1/C5 locus variants and Larsen score or the presence of erosions using a zero-inflated negative binomial regression (ZINB) on all patients, a negative binomial regression (NBREG) on erosive patients, or a logistic regression (LOGIT) for the presence of erosions on all patients reveal that TRAF1 markers are associated with erosions in NOAR and with the extent of the radiological damage in erosive patients in ERAS.

Gene, SNP	Cohort	Year	Model	No. Patients	Estimate	Lo95	Up95	p
TRAF1/C5 rs2900180	ERAS	2	ZINB severity	356	0.42	0.05	0.93	0.025
			ZINB susceptibility		0.66	0.32	1.39	0.276
			NBREG	192	0.32	−0.04	0.81	0.092
			LOGIT	435	1.12	0.76	1.64	0.574
		3	ZINB severity	353	0.31	0.04	0.65	0.024
			ZINB susceptibility		1.14	0.58	2.24	0.700
			NBREG	224	0.43	0.12	0.81	0.004
			LOGIT	424	0.86	0.57	1.29	0.467
TRAF1/C5 rs10760130	ERAS	2	ZINB severity	356	0.41	0.03	0.92	0.030
			ZINB susceptibility		0.64	0.28	1.49	0.303
			NBREG	192	0.36	−0.02	0.88	0.069
			LOGIT	435	1.29	0.85	1.96	0.223
		3	ZINB severity	353	0.32	0.02	0.69	0.032
			ZINB susceptibility		1.08	0.52	2.26	0.832
			NBREG	224	0.30	0.01	0.68	0.046
			LOGIT	424	1.08	0.70	1.67	0.716
TRAF1/C5 rs2900180	NOAR	5	ZINB severity	608	0.15	−0.08	0.43	0.223
			ZINB susceptibility		1.88	1.11	3.19	0.020
			NBREG	267	0.06	−0.12	0.27	0.558
			LOGIT	611	1.53	1.10	2.12	0.012
TRAF1/C5 rs10760130	NOAR	5	ZINB severity	607	0.18	−0.07	0.50	0.165
			ZINB susceptibility		1.11	0.64	1.92	0.707
			NBREG	266	0.07	−0.12	0.30	0.513
			LOGIT	610	1.29	0.91	1.83	0.149

NOAR: Norfolk Arthritis Register; ERAS: Early Rheumatoid Arthritis Study; SNP: single-nucleotide polymorphism; Estimate: percentage increase in Larsen score for the severity part of the ZINB model and for the NBREG model, and OR for the susceptibility part of the ZINB model and for the LOGIT model; Lo95: lower 95% CI of the point estimate; Up95: upper 95% CI of the point estimate.