Risk of Bias	Author’s Judgment Low Risk Unclear High Risk	Support for Judgment
Selection bias		Inclusion criteria: SIRS and suspected infection; only ICU patients
Bias in definition and measurement		Altered mentation defined by GCS, but did not specify at what level
Outcome measurement bias
Handling of missing data		Patients with missing value excluded, did not report the number of patients excluded
Confounding		Male-to-female ratio of 2:1
Bias of statistics or presentation of result		Logistic regression double counting variables Poor presentation of table margin