3. Jansen 2004: 'Risk of bias' assessment for observational studies using the Newcastle‐Ottawa Scale (Wells 2013).

Risk of Bias	Assessment	Support for judgement
Selection (one star each, maximum four stars)	2 stars
Representativeness of the exposed cohort	0 stars	This study only included participants with AML as opposed to all patients with haematological malignancies
Selection of the non‐exposed cohort	0 star	Participants in the control were from a second haematology centre, but there was no information to reassure that this cohort was drawn from the same community as exposed cohort
Ascertainment of exposure	1 star	Secondary analyses from HOVON 29, prospective randomised controlled trial
Demonstration that outcome of interest was not present at start of study	1 stars	The primary outcome and other outcomes were defined and were based on events that occurred after the study started
Comparability of cohort on design and analyses (maximum of two stars) Recognition of at least 75% of the main potential confounding factors (2 stars) Recognition of 50% to 75% of the main potential confounding factors (1 star)	0 stars	< 50% of potential cofounders considered and sex, age and AML type were adjusted for in the multiple regression model. There was no discussion on previous severe bleeding, use of anticoagulation, use of radiotherapy in addition to chemotherapy, previous cardiovascular disease, previous alloimmunisation or performance status
Outcome (one star each, maximum of four stars)	2 stars
Assessment of outcome	0 stars	Not described
Was follow‐up long enough for outcomes to occur?	1 star	Yes, 31 days from chemotherapy
Adequacy of follow‐up of cohorts	0 star	Reported for all, except unclear for infection, mean Hb during follow‐up, total number of platelet/red blood cell units received
Follow‐up equal between groups for primary and secondary outcomes?	1 star	Follow‐up not significantly different
Additional concerns		None
Overall assessment	4 stars

AML = acute myeloid leukaemia
 Hb = haemoglobin