ECMO-/ECLS

collected prospectively for important variables, which can control for severity of illness.
Although valuable, these studies are limited to establishing associations between
variables or treatments and outcomes and cannot directly prove causality.
The traditional “gold” standard of clinical evidence is the randomized controlled
trial (RCT) which directly tests a treatment against a “control”. The major advantages of
a RCT are that randomization can control for selection bias and the design allows for a
causal link to be drawn between an intervention and changes in the primary outcome.
However, RCTs require significant financial resources for study staff, treatment
interventions, and data collection, monitoring, and analysis. In addition, for conditions
that occur infrequently, recruitment of enough patients to adequately power a trial may
not be feasible or may require a large multi-institutional effort,which would significantly
increase the cost. In addition, the generalizability of the results of a RCT may be limited
depending on the inclusion/exclusion criteria; studies that are too restrictive are not
generalizable and those that are too inclusive may not show a difference or may include
subgroups that did not benefit or were harmed by the investigated treatment. For
example, many RCTs of novel anti-inflammatory or anticoagulant therapies for sepsis
and septic shock demonstrated that these therapies were only beneficial in the most
severely ill patients with the highest risk of death and were potentially harmful in less
severely ill patients [ 6 - 11 ]. In addition, the results from RCTs performed in critically ill
adults may not be directly translatable to children as was demonstrated with clinical
trials of activated protein C (APC) [ 7 , 12 - 17 ].Therefore, it is critical to determine if the
patient you are treating is similar to the patient population studied in a RCT before
applying its results. Another important aspect to consider when evaluating a RCT is the