1a. Was a consecutive or random sample of patients enrolled?

YES ‐ if this was clearly stated in the methods section or if the study stated that 'all' eligible patients were enrolled.
NO ‐ if it was clear that the clinician made the selection.
UNCLEAR ‐ if we could not decide between YES or NO.

1b. Was a case‐control design avoided?

This was scored YES for all included studies, as case‐control designs were excluded.

1c. Did the study avoid inappropriate exclusions?

YES ‐ if, for example, only solid tumour transplants were excluded or if no one was excluded.
NO ‐ if, for example, exclusion was done based on EORTC criteria or index test results. Studies excluding 'possible IA' patients were excluded.
UNCLEAR ‐ if it was not clear whether excluded patients were inappropriately excluded.

RISK OF BIAS:
HIGH ‐ when at least one question was answered with 'NO'.
LOW ‐ when at least two questions were answered with 'YES'.
UNCLEAR: all other situations.

CONCERNS RE. APPLICABILITY: We made an inventory of whether participants were inpatients or outpatients, the age groups of the participants and the cause of their increased risk for IA (neutropenia, corticosteroids etc).
HIGH CONCERN ‐ xxx
LOW CONCERN ‐ xxx
UNCLEAR CONCERN ‐ xxx

2a. Were the index test results interpreted without knowledge of the results of the reference standard?

YES ‐ if the index test was done before the reference standard was assessed, or when the authors clearly stated that the assessment of the index test was blinded.
NO ‐ if the authors stated that the assessment of the index test was not blinded.
UNCLEAR ‐ all other situations.

2b. If a threshold was used, was it pre‐specified?

YES ‐ if the threshold was mentioned in the methods section, or if the authors stated that they followed the manufacturer's directions (which include guidance on the threshold).
NO ‐ if the authors drew a ROC plot, or if multiple thresholds (outside the commonly reported ones of 0.5, 1.0 and 1.5) were reported, or if the authors stated that the threshold they used was the optimal threshold in their study.
UNCLEAR ‐ all other situations.

RISK OF BIAS:
HIGH ‐ when at least one question was answered with 'NO'.
LOW ‐ when only 2b was answered 'YES' and 2a with 'UNCLEAR' (as the galactomannan test is a laboratory test) or when both questions were answered with 'YES'.
UNCLEAR: if both were answered 'UNCLEAR'.

CONCERNS RE. APPLICABILITY:
HIGH CONCERN ‐ if a threshold was used that was not one of the more commonly used thresholds (i.e. 0.5 or 1.0 or 1.5).
LOW CONCERN ‐ if a threshold of 0.5, 1.0 or 1.5 was used; all studies used a commercially available and thus relatively standard and commonly available test.
UNCLEAR CONCERN ‐ if the threshold used was not clear, or if it was not entirely clear whether the authors evaluated the Platelia® kit.

3a. Is the reference standards likely to correctly classify the target condition?

As we regarded the EORTC/MSG criteria as being acceptable, this item was always fulfilled by all included studies. We did, however, register whether the authors of the primary study used the exact criteria of the EORTC/MSG and (if reported) how they were interpreted. If they only mentioned that they did divide their patients into categories, but did not explain on what basis, we scored this item as 'unclear'.

3b. Were the reference standard results interpreted without knowledge of the results of the index tests?

YES ‐ when the authors clearly stated that the assessment of the index test was blinded.
NO ‐ if the authors stated that the assessment of the index test was not blinded, or if the index test formed part of the reference standard (incorporation bias).
UNCLEAR ‐ all other situations.

RISK OF BIAS:
HIGH ‐ when at least one question was answered with 'NO'.
LOW ‐ when both questions were answered 'YES'.
UNCLEAR ‐ if one was answered 'UNCLEAR' and the other 'YES' or if both were answered 'UNCLEAR'.

CONCERNS RE. APPLICABILITY:
HIGH CONCERN ‐ if the EORTC criteria were used in a different way than described in the reference papers from 2002 and 2008 (e.g. if the authors used five or three categories instead of four).
LOW CONCERN ‐ if the EORTC criteria were used in the same way as described in the reference papers from 2002 and 2008.
UNCLEAR CONCERN ‐ if it was not clear how the EORTC criteria were implemented.

4a. Was there an appropriate interval between index test and reference standard?

The calculation of the diagnostic accuracy of a test is more reliable when the time between the Platelia test and the final diagnosis is not too long. If the galactomannan test is negative on day 1 and the patient is diagnosed as having IA on day 20, this test result will be regarded as a false negative result. The patient's true status on day 1, however, was not known in this case and the false negative result may have been a true negative result at that moment. We judged a time interval of less than 15 days as appropriate (YES).

4b. Did all patients receive the same reference standard?

The reference test was in most studies a composite reference while the index test was often used as screening tool to monitor whether patients developed IA. So some patients fulfilled more criteria than others. However, we considered the EORTC criteria as one reference standard. Partial verification would have been a problem in studies were only autopsy is used as reference standard, because it is only done when a patient dies and his or her family gives permission.

4c. Were all patients included in the analysis?

YES ‐ if all included patients also ended up in the 2 x 2 table.
NO ‐ if there was a discrepancy between these two numbers.
UNCLEAR ‐ if it was unclear whether patients were missing or not.

RISK OF BIAS:
HIGH ‐ when at least one question was answered with 'NO'.
LOW ‐ when at least two questions were answered 'YES' and the remaining one with 'UNCLEAR'.
UNCLEAR ‐ all other situations.