Glucocorticoid exposure and the risk of serious infections in rheumatoid arthritis: a marginal structural model application.

"to specify one outcome model over all follow-up periods we fixed the length of exposure history to three years and set the value of gc exposure before the study to zero adjusting for actual gc use before baseline in a separate variable (details in supplementary data s1 available at rheumatology online).; because conditioning on time-updated confounder values could introduce bias in the presence of exposure-confounder feedback we subsequently adjusted for confounding and selection bias (via informative censoring) in msm using stabilized ipw (details in supplementary data s2 available at rheumatology online) [ ].; the method is described in more detail in supplementary data s3 available at rheumatology online.; in a sensitivity analysis we investigated the 'missingness patterns' approach where it was assumed that treatment decision was only influenced by variables measured recently ( supplementary data s3 available at rheumatology online)."

"Disclosure statement : J.A. has received research grants and consulting fees from: AbbVie, AstraZeneca, BMS, MSD, Eli Lilly, Pfizer, Roche, Samsung Bioepis, Novartis and UCB. A.B, A.S., B.D. and T.F. have no conflicts of interest to declare."

"Funding This work was supported by the Swedish Research Council (grant number: 2016–01355), and agreements between Karolinska Institutet (J.A. as principal investigator) and Abbvie, BMS, MSD, Eli Lilly, Pfizer, Roche, Samsung Bioepis, and UCB, mainly regarding the safety monitoring of bDMARDs in rheumatology. Disclosure statement: J.A. has received research grants and consulting fees from: AbbVie, AstraZeneca, BMS, MSD, Eli Lilly, Pfizer, Roche, Samsung Bioepis, Novartis and UCB. A.B, A.S., B.D. and T.F. have no conflicts of interest to declare."