Table 3.

Adjusted relative risk estimates for serious infections after exposure to immunomodulatory treatments during pregnancy in women with autoimmune inflammatory conditions, Medicaid data 2000-10 and Optum Clinformatics data 2004-15

Comparison	Adjusted* hazard ratio (95% CI)
	Primary analysis	Analysis only including patients receiving monotherapy	Analysis truncating follow-up on index treatment discontinuation†
Non-biologics versus steroids in patients with autoimmune inflammatory conditions‡	0.81 (0.48 to 1.37)	0.85 (0.44 to 1.61)	0.71 (0.33 to 1.56)
TNF inhibitors versus steroids in patients with autoimmune inflammatory conditions‡ other than SLE	0.91 (0.36 to 2.26)	0.85 (0.28 to 2.56)	0.61 (0.12 to 3.16)
TNF inhibitors versus non-biologics in patients with autoimmune inflammatory conditions‡ other than SLE	1.36 (0.47 to 3.93)	1.44 (0.39 to 5.31)	—

TNF=tumor necrosis factor α; SLE=systemic lupus erythematosus.

^*Adjusted with propensity score weighting in Cox proportional hazard regression models. Variables with standardized differences >10 after propensity score weighting were also added to these models.

^†Discontinuation defined as no new filled prescription for one month after accounting for day supply of most recent prescription. This approach resulted in insufficient event counts for the TNF inhibitor versus non-biologics comparison to calculate hazard ratios.

^‡Autoimmune inflammatory conditions included ankylosing spondylitis, inflammatory bowel disease, psoriatic arthritis, rheumatoid arthritis, and SLE. TNF inhibitors are not indicated for the treatment of SLE and therefore patients with only SLE were excluded from comparisons concerning TNF inhibitors.