Effectiveness and safety of antifibrinolytic agents in the management of pulmonary haemorrhage: A systematic review with narrative synthesis

Table 1 Study-level risk-of-bias assessment with domain-level justification

Study	Tool	Overall judgement	Justification
Wand et al[15]	RoB 2	Low risk	Double-blind, placebo-controlled, adequate randomisation and allocation concealment reported; outcome assessment blinded; low attrition; pre-specified outcomes reported
Gopinath et al[4]	RoB 2	Some concerns	Open-label design means lack of blinding of participants/clinicians could influence co-interventions and subjective outcomes (e.g., decision to embolise); randomisation and outcome data otherwise adequate
Bellam et al[22]	RoB 2	Some concerns	Single-blind pilot design with small sample size; underpowered for the pre-specified intervention-rate outcome, raising risk of bias from imprecision and possible selective emphasis on significant secondary endpoints
Al-Samkari et al[16]	NOS	Moderate risk	Before-after pathway design without concurrent control; patients act as their own historical comparator, so secular trends and regression to the mean cannot be excluded; outcome ascertainment was consistent within the pathway
Kinoshita et al[17]	NOS	Moderate risk	Large propensity-matched administrative cohort with robust ascertainment of mortality and length of stay; however, administrative coding cannot capture bleeding severity, indication for treatment, or unmeasured confounders, so residual confounding by indication remains likely
Bethuel et al[5]	NOS	Serious/critical risk	Retrospective multicentre cohort with strong likely confounding by indication (nebulised tranexamic acid preferentially used in more severely bleeding or deteriorating patients); treatment selection not randomised or adjusted for all relevant severity markers; mortality finding should not be read as a causal estimate
Alkazemi et al[21]	NOS	Moderate/serious risk	Retrospective matched cohort with a small treated group (n = 14), limiting power to detect true differences; matching reduces but does not eliminate confounding by indication
Alabdrabalnabi et al[20]	NOS	Serious/critical risk	Uncontrolled case series of three patients with no comparator; high risk of selective reporting and very limited generalisability
O'Neil et al[18]	NOS	Serious risk	Single-centre retrospective observational study without a comparator group; small sample (n = 19); outcome ascertainment based on clinical record review with no blinding
Singleton et al[19]	NOS	Serious risk	Retrospective ECMO-specific cohort without a comparator group; highly selected, anticoagulated population with multiple competing risks for bleeding outcomes

NOS: Newcastle-Ottawa Scale; RoB 2: Cochrane Risk of Bias tool, version 2; ECMO: Extracorporeal membrane oxygenation.