Table 2 Quality of included randomized controlled trials using the Cochrane Collaboration tool for assessing risk of bias
| Â | Sequence generation | Concealment | Blinding | Incomplete outcome data | Selective outcome reporting | Other sources of bias | Study center | Percentage of patients lost to follow up | Source of study funding |
|---|---|---|---|---|---|---|---|---|---|
Montecalvo et al., 1992 [54] | Computer generated | Not described | No | No | No | No | Multiple ICUs, two centers | 0 | Not reported |
Kearns et al., 2000 [55] | Computer generated | Sealed envelope | No | No | No | No | Single | 0 | Ross Laboratories and the California Institute for Medical Research (partly) |
Chen et al., 2006 [56] | Not described | Not described | No | No | No | No | Two ICUs, one center | 0 | National Science Council |
Nguyen et al., 2007 [57] | Computer generated | Yes | Yes | No | No | No | Single | 0 | National Health and Medical Research Council (NHMRC) of Australia (partly) |
Desachy et al., 2008 [58] | Not described | Not described | No | No | No | No | Two ICUs | 0 | Not reported |
Hsu et al., 2009 [59] | Computer generated | Yes | No | No | No | No | Single | 0 | Kaohsiung Veterans General Hospital |
White et al., 2009 [60] | Computer generated | Yes (sealed opaque envelope) | No | Yes | No | No | Single | 0 | Not reported |
Montejo et al., 2010 [61] | Not described | Not described | No | No | No | No | Single | 0 | Not reported |
Acosta-Escribano et al., 2010 [62] | Central randomization | Yes | No | Yes | No | No | Multicenter | 0 | Novartis Consumer Health (Spain) |
Arabi et al., 2011 [14] | Computer generated | Yes | No | No | No | No | Single | 0 | King Abdulaziz City for Science and Technology |
Singer et al., 2011 [63] | Computer generated | Yes | No | No | No | No | Single | 0 | Not reported |
Rice et al., 2011 [15] | Not described | Yes (sealed opaque envelope) | No | No | No | No | Two ICUs, single center | 0 | National Institutes of Health (partly) |
Rice et al., 2012 [16] | Web-based system | Sealed envelope | No | No | No | No | Multicenter | 0 | National Heart, Lung, and Blood Institute |
Huang et al., 2012 [64] | Software-generated randomization | Not described | No | Yes Hospital mortality data were missing for some patients | No | No | Single | Hospital mortality data missing for 4/101 patients (4Â %) | Kaohsiung Veterans General Hospital |
Reignier et al., 2013 [65] | Computer-generated, interactive, web-response system | Yes | No | No | No | No | Multicenter | 0 | The Centre Hospitalier Departemental de la Vendee |
Rugeles et al., 2013 [23] | Computer-generated random allocations | No Sealed envelopes were used but one investigator knew patient allocation | No One investigator knew patient allocation. | Yes Mortality was reported as one of the secondary endpoints but was not reported | Yes | Yes Many patients were excluded from analysis with uncertainty if exclusion criteria were determined a priori. | Single | 0 | Lafrancol S.A |
Peake et al., 2014 [66] | Web-based system | Centralized, web-based randomization schedule | Yes | No | No | No | Multicenter | One patient in the intervention group was withdrawn and one patient in the control group was lost to follow up by day 90. | The Royal Adelaide Hospital and the Australian, New Zealand College of Anaesthetists and Fresenius Kabi |
Charles et al., 2014 [22] | Random number sequence | Opaque security envelopes | No | No | No | Yes The trial was stopped before achieving the target sample size because of slow enrolment | Single | 0 | The NIH |
Braunschweig et al., 2015 [67] | Computer-generated random allocations | Sealed envelopes | No | No | No | Yes The trial was stopped before achieving the target sample size because of higher mortality, a secondary outcome, in the intervention group | Single | 0 | The NIH/NHLBI |
Arabi et al., 2015 [68] | Computer-generated random allocations (blocks of variable size | Opaque sealed envelopes | No | No | No | No | Multicenter | 9 patients lost to follow up, 3 in the intervention group and 3 in the control group | King Abdullah International Medical Research Center |
Doig et al., 2015 [69] | Computer-generated random allocations (blocks of variable size) | Secure central randomization web server | No | No | No | No | Multicenter | 4 patients lost to follow up (90-day interview), 2 in each group. | National Health and Medical Research Council of Australia |