−1.0; 95 % CI −1.78 to −0.23, p = 0.01, Fig. 5d). There
was a non-significant trend towards higher rates of suicide
by those family members who had witnessed resuscitation
when the data were analyzed on a per-protocol
basis, but this trend was not present on ITT analysis
(Fig. 5e). There was no evidence of statistical heterogeneity
in any of the above comparisons. Using GRADE,
outcomes based on RCTs start as “high”-quality evidence;
however, we downgraded the quality of evidence
for all outcomes due to imprecision because of the need
to account for clustering effects in the study by Jabre et
al. and the need to estimate mean and standard deviation
from median and interquartile range. There was
no evidence of inconsistency, or imprecision, and an insufficient
number of studies to assess for publication
bias (Table 3).
We were unable to conduct a meta-analysis in the
pediatric setting due to the lack of published data. Only
one randomized controlled trial [12] in the ER trauma
bay compared FPDR to usual care for pediatric patients,
demonstrating no evidence of interference by family
members resulting in changes in time or success of
major interventions, imaging procedures, or mortality.
The serious risk of bias due to poor randomization in
Dudley et al. [12] and the absence of any large effect
size, dose response, or plausible residual confounding