178 A . G . Beshish et al .: J Extra Corpor Technol 2024 , 56 , 174 – 184 ( p = 0.0008 ). Comparisons between different FSS domains on admission and discharge are presented in Table 4 . The mean FSS score difference between the two groups was 0.4 ( 95 % CI : �0.9 , 1.6 , p = 0.512 ). Of the 27 overall survivors , 19 / 27 ( 70 %) were from the non-hyperoxia group and 8 / 27 ( 30 %) were from the hyperoxia group . We failed to identify an association between designation as “ hyperoxia ” and new morbidity , or unfavorable outcome ( Table 5 ).
Outcomes analysis
Figure
2 . Receiver operating characteristic curve ( ROC ) identifying the optimal discriminatory cut-point for operative mortality was PaO 2 = 182 mmHg ( sensitivity 68 %, and specificity 70 %).
the first 4-hours of ECLS ( 152.7 vs . 124.1 , p < 0.05 ). Although the CICU , postoperative , and overall hospital length of stay was shorter in the hyperoxia group , the mortality rate was significantly higher ( 77 % vs . 39 %, p = 0.005 ) Table 2 . Thecauses of comorbidity and death are listed in Table 3 . Of the 65 patients who required post-Norwood ECLS , only 6 patients ( 9.2 %) required reintervention [ 5 aortic arch augmentation / revision of DKS ( Damus Kaye Stansel ), and 1 patient required shunt revision ].
Patient demographics and characteristics for overall cohort stratified by timing of ECLS initiation
We then stratified patients into two groups based on the timing of initiation of ECLS post-Norwood (< 5 days post- Norwood vs . 5 days post-Norwood ). Patients in the early group had higher median PaO 2 in the first 48 h of ECLS ( 274.6 mmHg , IQR 166.24 , 313.12 vs . 101.79 mmHg , IQR 67.70 , 201.60 , p < 0.001 ), had higher rates of central vs . peripheral cannulation ( 85 % vs . 37 %, p < 0.0001 ), had higher serum lactate within 2 h of ECLS initiation 12.85 ( IQR 5.12 , 15.23 ) vs . 5.74 ( IQR 3.0 , 12.20 ), p = 0.02 , and had shorter CICU length of stay ( LOS ), postoperative LOS , and overall hospital LOS . A full comparison between the 2 groups is shown in Supplemental Table 1 .
Functional Status Scale ( FSS )
The mean total FSS score for survivors in the non-hyperoxia group increased from 6 ( SD 0 ) on admission / baseline to 8.4 ( SD 1.2 ) at discharge ( p < 0.0001 ), while the mean total FSS score for survivors in the hyperoxia-group increased from 6 ( SD 0 ) on admission / baseline to 8.8 ( SD 1.4 ) at discharge
In univariable analysis , using the ROC curve , PaO 2 > 182 mmHg was associated with higher odds of mortality [ OR 5.2 ( 95 % CI : 1.8 – 15.0 ), p = 0.003 ]. However , the association was insignificant when controlling for the source of pulmonary blood flow , CPB time , and post-Norwood VIS score at 48 h [ OR 3.1 ( 95 % CI : 0.8 – 12.3 ), p = 0.104 ]. No difference in stage II or III AKI or PPLOS was detected between the hyperoxia and non-hyperoxia-groups ( Table 6 ). The association of average PaO 2 and CPB time is graphically demonstrated in Figure 3 , with a correlation coefficient of 0.4 ( 95 % CI : 0.1 – 0.6 , p = 0.003 ).
Discussion
Despite advances in cardiac surgery and extracorporeal technology , morbidity and mortality persist . Many of the risk factors for poor outcomes , including weight and surgical complexity are not modifiable . Thus , there is value in identifying practice-based risk factors to improve outcomes . We describe the relationship between hyperoxia in the first 48-hours while on ECLS and mortality in an unadjusted analysis with an OR of 5.15 . However , this association did not persist when adjusting for confounding variables ( source of pulmonary blood flow , CPB time , and post-Norwood VIS score at 48 h ). It is possible that this lack of association was due to inadequate sample size . There was no significant association seen with AKI or PPLOS . Although oxygen administration was not standardized , there were important differences in the treatment groups . The hyperoxia group had different operative strategies and higher markers of illness , however received earlier ECLS with greater support .
In other critical illness settings , an association between excessive oxygen delivery with poor clinical outcomes has been reported . In patients requiring ECLS for cardiac arrest , hyperoxia ( as defined by a mean PaO 2 > 193 Torr ) was associated with 30-day mortality and the need for dialysis [ 1 , 21 , 32 ]. Several reports of neonates with asphyxia have demonstrated an association between hyperoxia and a risk of brain injury and mortality [ 1 , 33 , 34 ]. In a prior report we showed that a substantial portion of infants undergoing cardiac surgery using CPB were exposed to hyperoxia and patients in the hyperoxia group had four-fold greater odds of mortality within 30 days of surgery [ 14 ]. This current report supports earlier findings that hyperoxia is likely associated with worse outcomes . However , understanding of particular populations at risk remains unclear as some studies fail to demonstrate an association between hyperoxia and mortality [ 35 ].