Objective The authors retrospectively reviewed their experience with e
xtracorporeal life support (ECLS) in 100 adult patients with severe re
spiratory failure (ARF) to define techniques, characterize its efficac
y and utilization, and determine predictors of outcome. Summary Backgr
ound Data Extracorporeal life support maintains gas exchange during AR
F, providing diseased lungs an optimal environment in which to heal. E
xtracorporeal life support has been successful in the treatment of res
piratory failure in infants and children. In 1990, the authors institu
ted a standardized protocol for treatment of severe ARF in adults, whi
ch included ECLS when less invasive methods failed. Methods From Janua
ry 1990 to July 1996, the authors used ECLS for 100 adults with severe
acute hypoxemic respiratory failure (n = 94): p(a)O(2)/FiO2 ratio of
55.7 +/- 15.9, transpulmonary shunt (Qs/Qt) of 52 +/- 22%, or acute hy
percarbic respiratory failure (n = 6): p(a)CO(2) 84.0 +/- 31.5 mmHg, d
espite and after maximal conventional ventilation. The technique inclu
ded venovenous percutaneous access, lung ''rest,'' transport on ECLS,
minimal anticoagulation, hemofiltration, and optimal systemic oxygen d
elivery. Results Overall hospital survival was 54%. The duration of EC
LS was 271.9 +/- 248.6 hours. Primary diagnoses included pneumonia (49
cases, 53% survived), adult respiratory distress syndrome (45 cases,
51% survived), and airway support (6 cases, 83% survived). Multivariat
e logistic regression modeling identified the following pre-ECLS varia
bles significant independent predictors of outcome: 1) pre-ECLS days o
f mechanical ventilation (p = 0.0003), 2) pre-ECLS p(a)O(2)/FiO2 ratio
(p = 0.002), and 3) age (years) (p = 0.005). Modeling of variables du
ring ECLS showed that no mechanical complications were independent pre
dictors of outcome, and the only patient-related complications associa
ted with outcome were the presence of renal failure (p < 0.0001) and s
ignificant surgical site bleeding (p = 0.0005). Conclusions Extracorpo
real life support provides life support for ARF in adults, allowing ti
me for injured lungs to recover. In 100 patients selected for high mor
tality risk despite and after optimal conventional treatment, 54% surv
ived. Extracorporeal life support is extraordinary but reasonable trea
tment in severe adult respiratory failure. Predictors of survival exis
t that may be useful for patient prognostication and design of future
prospective studies.