Table 3

Effect of bDMARDs on the change in mSASSS and on the likelihood of progression: comparative studies with variable exposure to treatment

StudyFollow-up yearsN patients/intervals*ComparisonOutcomeHandling of pretreatment time-varying confoundersTime-varying
mediator
Total effect
(95% CI)
Mediation analysis
Molnar et al, 201744Up to 10432/616TNFi versus no TNFi≥2 mSASSSAdjust. (mSASSS, ASDAS and NSAID+baseline variables)ASDAS
(start interval)
OR: 0.5 (0.3 to 0.9)Indirect effect +
Direct effect
Gensler et al, 2018424519/NRTNFi versus no TNFi†mSASSSAdjust. (ASDAS+baseline variables) Strat. (NSAID index)NAΔ: −3.3 (4.0 to –2.6)NA
Park et al, 201945NR215/518TNFi versus NSAIDmSASSSNo TV confounders. Adjust. BL (age, CRP, smoking and syndesmophytes)CRP
(aver interval)
β: −0.9 (−1.5 to −0.3)Indirect effect +
Direct effect
Koo et al, 202043Up to 18338/2364TNFi on versus TNFi offmSASSSIPTW (mSASSS, ESR, CRP, BASDAI and comedication+baseline variables)NAβ: −0.1 (0.2 to 0.0)NA
Sepriano et al, 202146Up to 10314/442TNFi versus no TNFimSASSSAdjust. (mSASSS)‡+PS. Adjust. BL (gender, symptom duration, ASDAS, HLA-B27, mSASSS, EMM and NSAIDs)ASDAS
(start interval)
β: −0.8 (1.4 to −0.2)Indirect effect +
Direct effect +
  • Total effect: Effect of TNFi on radiographic progression at the end of each interval adjusting for pretreatment confounders but not for the mediator.

  • Direct effect: Effect of TNFi on radiographic progression at the end of each interval adjusting for pretreatment confounders and for the mediator.

  • Indirect effect: Effect of TNFi on radiographic progression at the end of each interval through the mediator (also adjusting for pretreatment confounders).

  • *2-year intervals in all studies.

  • †If NSAID index ≥50.

  • ‡NSAIDs not kept in the final model because it did not prove to confound the association of interest.

  • adjust, adjustment; ASDAS, Ankylosing Spondylitis Disease Activity Score; aver, average; BASDAI, Bath Ankylosing Spondylitis Disease Activity Index; bDMARDs, biological disease-modifying antirheumatic drugs; BL, baseline; CRP, C reactive protein; EMM, extramusculoskeletal manifestations (uveitis, inflammatory bowel disease and psoriasis); ESR, erythrocyte sedimentation rate; HLA, human leukocyte antigen; IPTW, inverse probability treatment weighting; mSASSS, modified Stoke Ankylosing Spondylitis Spinal Score; NR, not reported; NSAIDs, non-steroidal anti-inflammatory drugs; PS, propensity score; Strat, stratification; TNFi, tumour necrosis factor alpha inhibitor; TV, time-varying.