Cut-offs derived from neonatal outcome variables
Table 2 presents AUCs and different optimal cut-off levels for lactate for different outcomes, with corresponding sensitivity, specificity, negative and positive likelihood ratios. Four of the ROC curves are presented in Figure 1. AUC-MA was 0.87 (95% CI: 0.77-0.97), sensitivity 82.4% (95% CI: 56.6-96.2), specificity 80.1% (95% CI: 76.0-83.7) with the optimal cut-off ≥5.7mmol/L. For all the outcome variables, the optimal cut-off varied between 3.6 mmol/L to 5.8 mmol/L. There were too few cases with AS < 4 at 5 minutes (0-25 minutes n = 4, whole cohort n = 10)) and HIE (0-25 minutes n = 2, whole cohort n = 4) for statistical analysing. From Sahlgrenska University Hospital cases with simultaneously measured pH and SSLX values were obtained. Due to the small number of cases within 25 minutes (n = 41) ROC curves for the predictive ability of pH respective lactate for MA were calculated for cases within 60 minutes (n = 208). The AUC-MA for scalp blood pH was 0.47 (95% CI: 0.26-0.67) and for scalp SSLX lactate 0.66 (95% CI: 0.49-0.82).
To decide which cut-off should be suggested, we present the predictive ability for three different cut-offs calculated only from cohort 1 together with the expected intervention rates for the cut-off when applied on the whole cohort (Table 3).
Table 4, found in the supplementary material, represents the babies from cohort one born with metabolic acidosis. Only two of the babies were diagnosed with HIE. Remarkably, three of the babies had scalp blood values under recommended cut-off for intervention, but none of those had a very low AS at one or five minutes and no need for transfer to NICU.