It is
more likely that we have not fully controlled for confounding
by indication, despite a 17-factor adjustment. In other
words, the need for tracheostomy itself defines, or is a
marker for, a risk for adverse outcome that is not wholly
captured by the risk factors included. These limitations of
our statistical models do not change the importance of
our results. Clinicians considering a tracheostomy for an individual
patient could supplement clinical status and history
with these data to help parents grasp potential longterm
outcomes.