to the offering of FPDR to those randomized to usual care.
For per-protocol analysis, we compared family members
who actually were present during resuscitation to family
members who were not present. For outcomes where
FPDR was expected to be of benefit, such as psychological
outcomes in family members, data from an ITT analysis is
presented, as it is more likely to provide a conservative estimate
of effect. For outcomes where FPDR may potentially
be of harm, such as patient mortality, duration of
resuscitation, or time to critical intervention, per-protocol
data was used where available, as it gives a less conservative
estimate of risk.
Risk of bias across studies and across outcomes
Quality of evidence supporting each outcome was assessed
using the Grading of Recommendations Assessment,
Development and Evaluation (GRADE) approach. GRADE
takes into account study risk of bias, publication bias, imprecision,
inconsistency, and indirectness of the evidence
[8]. GradePRO software (www.guidelinedevelopment.org)
was used to generate summary of findings tables including
data from RCTs or observational studies if no RCTs were
available for outcomes of interest.