Results
Of the 128 patients with VA-ECMO support, the median age was 60 years (interquartile range [IQR]: 49.5-68.5) and 81 (63%) were males. ECMO indications included post-cardiotomy shock (39.8%), cardiac arrest (29.7%), and cardiogenic shock (26.6%). Of 128, 66 (51.6%) experienced ABI.
Within first 24 hours, we found 8,423 temperature measurements(Supplemental Figure 1) with a median of 71 readings per patient. Mild hypothermia of varying durations was observed in 88/128 patients (68.8%) within the first 24 hours of ECMO support, and 18/128 patients (14.2%) had a mean temperature less than 36°C (range: 32.4-35.9°C). Mean 24-hour temperature was 36.4°C within the post-cardiotomy shock patients, 36.0°C within the cardiac arrest patients, and 36.5°C within the cardiogenic shock patients.
Baseline characteristics of patients with vs. without mild hypothermia were comparable (Table 1 ). 5/18 (27.8%) patients who had good neurologic outcome with mild hypothermia, while 21/110 patients who had good neurologic outcome without mild hypothermia (19.1%) (p=0.40) (Supplemental Figure 2 ).
In a multivariable model adjusting for age, SOFA score, ABI, systemic hemorrhage, mean 24-hour PaO2, and post-cardiotomy shock, mean 24-hour temperature was not associated with good neurological outcome (OR=0.70, 95%CI=0.18-2.73, p=0.61). Using the same model, longer duration of mild hypothermia was independently associated with good neurological outcome (OR=1.17, 95%CI=1.04-1.31, p=0.01) (Figure 1, Table 2 ).
In a sensitivity analysis, neither duration of mild hypothermia (OR=0.93, 95%CI=0.84-1.03, p=0.17) nor mean temperature (OR=0.78, 95%CI=0.29-2.08, p=0.62) was significantly associated with mortality. Similarly, duration of mild hypothermia (p=0.47) and mean 24-hour temperature (p=0.76) were not significantly associated with frequency of systemic hemorrhages (Supplemental Results, Supplemental Table 1-4) . After excluding patients who experienced cardiac arrest prior to VA-ECMO (n=58), longer duration of mild hypothermia was associated with improved neurologic outcome at discharge, however this finding was not statistically significant (OR=0.81, 95% CI=0.64-1.01, p=0.07). Mortality was not associated with either mean 24-hour temperature (OR=0.77, CI=0.28-2.09, p=0.62)(Supplemental Table 1) or duration of mild hypothermia (OR=0.93, CI=0.85-1.03, p=0.17) (Supplemental Table 2) after adjusting for age, SOFA score, ABI, mean 24-hour PaO2, and post-cardiotomy shock.
Overall, 68 (54%) experienced systemic hemorrhages. Patients with systemic hemorrhage were more likely to have higher SOFA scores (p=0.003), more likely to be centrally cannulated (p=0.02), and more likely to have post-cardiotomy shock (p=0.04). In a multivariable model adjusting for SOFA scores, central cannulation, and post-cardiotomy shock, mean 24-hour temperature (OR=0.91, CI=0.53-1.57, p=0.73)(Supplemental Table 3) and duration of hypothermia (OR=0.97, CI=0.90-1.05, p=0.48) (Supplemental Table 4) was not associated with increased odds of systemic hemorrhage. Eight (6.3%) had ICH and mean temperature (p=0.30) and duration of hypothermia was not associated with ICH (p=0.99).