This was a prospective cohort study of consecutive deliveries at Washington University in St. Louis Medical Center from 2009 to 2014. The study was approved by the Washington University School of Medicine Human Research Protection Office.
Inclusion criteria were singleton pregnancies at term, vertex presentation, and labor prior to delivery. Multiple gestations and pregnancies with fetal anomalies were excluded. Term pregnancy was defined as gestational age ≥37 weeks. Pregnancies were dated by a woman’s last menstrual period and confirmed with first or second trimester ultrasound [14 (link)]. Demographic information, medical and surgical history, obstetric and gynecologic history, prenatal history and detailed labor and delivery information were abstracted from patients’ charts by trained research nurses.
Umbilical cord blood was collected immediately after infant delivery, prior to knowledge of neonatal outcomes as previously described [7 (link)]. Briefly, a policy of universal umbilical cord gas and lactate measurement was instituted prior to the study. Both arterial and venous blood samples are obtained from a clamped segment of cord immediately after delivery. Umbilical blood lactate is measured from whole blood using an automated benchtop analyzer (DXC-800 Automated Chemistry Analyser, Beckman Coulter). As previously reported the coefficient of variation of the lactate assay in our laboratory is 2.9% [7 (link)]. Umbilical arterial blood samples were validated to be arterial or venous by ensuring that pH was at least 0.02 lower in the artery than the vein [15 (link)].
The outcome measures were arterial lactic acidemia and a composite neonatal outcome. Umbilical arterial lactic acidemia was defined as arterial lactate >3.9 mmol/L based on a prior study in our institution [7 (link)]. The composite neonatal outcome was made up of neonatal death and any of a number of neonatal morbidities including endotracheal intubation, mechanical ventilation, meconium aspiration syndrome, hypoxic-ischemic encephalopathy, and therapeutic hypothermia as previously reported [7 (link)]. Components of the composite were diagnosed by the attending neonatologist, without knowledge of the umbilical cord lactate levels. Meconium aspiration syndrome was diagnosed based on the presence of meconium stained amniotic fluid, neonatal respiratory distress, and characteristic radiographic abnormalities [16 (link)]. Hypoxic-ischemic encephalopathy was diagnosed based on the National Institute of Child Health and Human Development (NICHD) criteria [17 (link)]. Administration of therapeutic hypothermia was indicated for neonates meeting the following criteria per institutional protocol: ≥36 weeks gestational age at birth, moderate to severe hypoxic-ischemic encephalopathy with or without seizures, and any one of 10-minute Apgar score <5, prolonged resuscitation at birth, severe acidosis (pH < 7.1) on cord or neonate blood gas analysis within 60 minutes of birth, or base deficit (>12 mmol/L) on cord or neonate blood gas analysis within 60 minutes of birth [18 (link)]. Only one morbidity was counted per patient for the composite.
Baseline characteristics were calculated for the entire cohort and compared between women with and without the composite neonatal outcome. Continuous variables were compared using the Student’s t test while categorical variables were compared using the chi-square or Fisher’s exact test as appropriate. Normality of distribution of the continuous variables was evaluated using the Kolmogorov-Smirnov test.
We used linear regression analysis to examine the relationship between umbilical cord arterial and venous lactate. We constructed receiver-operating characteristics (ROC) curves to assess the predictive ability of umbilical venous lactate for arterial lactic acidemia, and to compare the predictive ability of venous and arterial lactate for the composite neonatal outcome. The ‘optimal’ cut-point of venous lactate for predicting arterial lactic acidemia was estimated based on the maximal Youden index [19 (link)]. The cut-point corresponding to the maximal Youden index maximizes the correct classification of subjects [20 (link)]. The area under the ROC curves for venous and arterial lactate were compared using the method described by Delong et al. [21 (link)]. We calculated and compared predictive characteristics (sensitivity, specificity, positive and negative predictive values, and positive and negative likelihod ratios) of arterial and venous lactate for neonatal morbidity based on the ‘optimal’ cut-points. Sensitivities and specificities were compared using an extension of the McNemar test [22 (link)]. To explore the effects of using different cut-points of venous lactate on prediction of the composite neonatal outcome, we calculate predictive characteristics for different venous lactate cut-points. These included the 95th and 99th percentile thresholds from the current cohort as well as venous lactate thresholds estimated from arterial lactate thresholds in the literature [2 (link), 23 (link)].
We did not estimate the sample size a priori; all consecutive patients meeting the inclusion criteria during the study period were included. All statistical tests were 2-tailed and P<0.05 was considered significant. Analyses were conducted using STATA software package, version 12, Special Edition (College Station, TX).
Inclusion criteria were singleton pregnancies at term, vertex presentation, and labor prior to delivery. Multiple gestations and pregnancies with fetal anomalies were excluded. Term pregnancy was defined as gestational age ≥37 weeks. Pregnancies were dated by a woman’s last menstrual period and confirmed with first or second trimester ultrasound [14 (link)]. Demographic information, medical and surgical history, obstetric and gynecologic history, prenatal history and detailed labor and delivery information were abstracted from patients’ charts by trained research nurses.
Umbilical cord blood was collected immediately after infant delivery, prior to knowledge of neonatal outcomes as previously described [7 (link)]. Briefly, a policy of universal umbilical cord gas and lactate measurement was instituted prior to the study. Both arterial and venous blood samples are obtained from a clamped segment of cord immediately after delivery. Umbilical blood lactate is measured from whole blood using an automated benchtop analyzer (DXC-800 Automated Chemistry Analyser, Beckman Coulter). As previously reported the coefficient of variation of the lactate assay in our laboratory is 2.9% [7 (link)]. Umbilical arterial blood samples were validated to be arterial or venous by ensuring that pH was at least 0.02 lower in the artery than the vein [15 (link)].
The outcome measures were arterial lactic acidemia and a composite neonatal outcome. Umbilical arterial lactic acidemia was defined as arterial lactate >3.9 mmol/L based on a prior study in our institution [7 (link)]. The composite neonatal outcome was made up of neonatal death and any of a number of neonatal morbidities including endotracheal intubation, mechanical ventilation, meconium aspiration syndrome, hypoxic-ischemic encephalopathy, and therapeutic hypothermia as previously reported [7 (link)]. Components of the composite were diagnosed by the attending neonatologist, without knowledge of the umbilical cord lactate levels. Meconium aspiration syndrome was diagnosed based on the presence of meconium stained amniotic fluid, neonatal respiratory distress, and characteristic radiographic abnormalities [16 (link)]. Hypoxic-ischemic encephalopathy was diagnosed based on the National Institute of Child Health and Human Development (NICHD) criteria [17 (link)]. Administration of therapeutic hypothermia was indicated for neonates meeting the following criteria per institutional protocol: ≥36 weeks gestational age at birth, moderate to severe hypoxic-ischemic encephalopathy with or without seizures, and any one of 10-minute Apgar score <5, prolonged resuscitation at birth, severe acidosis (pH < 7.1) on cord or neonate blood gas analysis within 60 minutes of birth, or base deficit (>12 mmol/L) on cord or neonate blood gas analysis within 60 minutes of birth [18 (link)]. Only one morbidity was counted per patient for the composite.
Baseline characteristics were calculated for the entire cohort and compared between women with and without the composite neonatal outcome. Continuous variables were compared using the Student’s t test while categorical variables were compared using the chi-square or Fisher’s exact test as appropriate. Normality of distribution of the continuous variables was evaluated using the Kolmogorov-Smirnov test.
We used linear regression analysis to examine the relationship between umbilical cord arterial and venous lactate. We constructed receiver-operating characteristics (ROC) curves to assess the predictive ability of umbilical venous lactate for arterial lactic acidemia, and to compare the predictive ability of venous and arterial lactate for the composite neonatal outcome. The ‘optimal’ cut-point of venous lactate for predicting arterial lactic acidemia was estimated based on the maximal Youden index [19 (link)]. The cut-point corresponding to the maximal Youden index maximizes the correct classification of subjects [20 (link)]. The area under the ROC curves for venous and arterial lactate were compared using the method described by Delong et al. [21 (link)]. We calculated and compared predictive characteristics (sensitivity, specificity, positive and negative predictive values, and positive and negative likelihod ratios) of arterial and venous lactate for neonatal morbidity based on the ‘optimal’ cut-points. Sensitivities and specificities were compared using an extension of the McNemar test [22 (link)]. To explore the effects of using different cut-points of venous lactate on prediction of the composite neonatal outcome, we calculate predictive characteristics for different venous lactate cut-points. These included the 95th and 99th percentile thresholds from the current cohort as well as venous lactate thresholds estimated from arterial lactate thresholds in the literature [2 (link), 23 (link)].
We did not estimate the sample size a priori; all consecutive patients meeting the inclusion criteria during the study period were included. All statistical tests were 2-tailed and P<0.05 was considered significant. Analyses were conducted using STATA software package, version 12, Special Edition (College Station, TX).
