Trauma is a major cause of mortality in children. Identifying potentially severe injuries can be complicated because children have a greater capacity for hemodynamic compensation than adults do. Although late shock is quickly recognizable in children, it may be difficult to detect early shock for several reasons. First, while tachycardia can be a sensitive indicator of shock, it is less applicable in children than in adults because of variations in resting heart rate at different ages and extreme lability of the heart rate due to anxiety or pain. Second, children often have normal mean arterial pressure during early shock due to catecholamine release, which increases diastolic pressure and obscures the effect of systolic pressure on mean arterial pressure. Third, with a smaller blood volume than adults, children are predisposed to unanticipated hemodynamic decompensation even from small blood volume losses.

Traumatic hemorrhage leads to tissue hypoxia that releases lactate, a biomarker for shock (1). There are unique challenges in interpretation of lactate concentrations in children. Lactate concentration may transiently increase by stress hormones released in fearful, anxious, or crying children (2,3). On the other hand, children have more efficient buffering systems and higher oxygen-carrying capacity than adults, resulting in a better response to tissue hypoxia (4,5). Therefore, the clinical utility of lactate in children has been questioned. Elevated values of admission lactate (6-8) and prehospital lactate (9) have been shown to be early prognostic indicators in adults with trauma. However, sparse evidence supports the utility of this biomarker in pediatric trauma. A promising association has been demonstrated in a few studies that are limited in size and scope. Shah et al. (10) found that prehospital lactate concentrations were higher in those requiring critical care, including those with normal prehospital vital signs and Glasgow Coma Scale (GCS). Ramanathan and colleagues (11) showed that elevated lactate values correlated with injury severity, hospital length of stay (LOS), morbidity, and mortality. Although elevated lactate concentration has not yet been universally defined in pediatric trauma, Ramanathan et al. (11) considered lactate > 4.7 mmol/L to be elevated.

We primarily aimed to determine the utility of admission lactate concentration in severe pediatric trauma by measuring the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of varying lactate values. The secondary objective was to assess the correlations of admission lactate with GCS, Injury Severity Score (ISS), and hospital LOS.

Table 1 shows the characteristics of the 566 patients included in this analysis. Motor vehicular collision was the most prevalent trauma mechanism (53.0%). Of the 53 patients (9.4%) who suffered severe trauma, 11 (1.9%) died, 32 (5.7%) underwent transfusions, and 32 (5.7%) did emergency surgical interventions. We applied mutually inclusive criteria for severe trauma, and as such the sum of patients in each classification exceeded the total number of patients with severe trauma (75 vs. 53). None of the patients had bleeding diseases such as hemophilia. The median lactate concentration was 1.7 mmol/L (interquartile range, 1.3-2.5) for the entire population. Patients with severe trauma had a higher median lactate value than those with non-severe trauma (2.5 mmol/L [1.9-3.8] vs. 1.7 mmol/L [1.2-2.4]; P < 0.001). There was a significant difference in the disposition of patients with and without severe trauma. Table 2 details the primary injury types. Head injuries were most prevalent (25.8%), followed by a similar distribution of injuries to the extremities, lungs, spine, and intraabdominal solid organs.

Table 3 shows the sensitivity, specificity, PPV, and NPV for different lactate cutoffs. A high cutoff had a higher specificity and NPV for severe trauma but did not perform as well for sensitivity and PPV. Fig. 1 depicts an ROC curve representing the sensitivity and specificity for varying lactate cutoffs with the optimal cutoff predictive of severe trauma. At an optimal cutoff of 1.9, the area under the curve was 0.69 (95% confidence interval, 0.61-0.77), resulting in a sensitivity of 75.4% and a specificity of 57.9%. ROC curves and optimal cutoffs for GCS and ISS are represented in Figs 2 and 3. Admission lactate was correlated negatively with GCS (Spearman’s rho = -0.134; P = 0.001) and positively with ISS (0.130; P = 0.002), while not correlated with hospital LOS (0.070; P = 0.096).

In this pediatric study, we found that admission lactate has a good specificity, i.e., rule-in capacity, for severe trauma at the cost of limited sensitivity, i.e., rule-out capacity. As expected, the specificity improved with increasing lactate cutoffs. A cutoff higher than 3.5 mmol/L had a specificity higher than 91%. Conversely, the sensitivity tends to be better at a lower cutoff (12); the overall sensitivities were relatively low (13.2%-96.2%) at lactate cutoffs ranging from 1 to 6 mmol/L. This means that some patients with severe trauma did not have high lactate concentrations. When ruling out severe trauma, lactate concentration should be used with caution, and in conjunction with other screening tools. The optimal sensitivity and specificity at a lactate value of 1.9 mmol/L (Fig. 1) indicate that any value above this cutoff, the upper limit of normal lactate value, should raise concern about severe trauma.

In a prospective study on 277 children, Ramanathan and colleagues (11) showed that a lactate concentration above 4.7 mmol/L strongly suggested severe injury, while a lower value is reassuring for not having an injury. In our report, higher lactate values around 4.7 mmol/L (thresholds of 4.5 and 5.0 mmol/L) were highly specific for severe trauma. Contrary to the prospective study, a lactate value lower than 4.7 mmol/L was not necessarily reassuring against severe injury. At the time of writing, the authors found an article describing similar initial lactate cutoffs in pediatric trauma; 5.1 mmol/L for mortality (AUC, 0.87 [95% confidence interval, 0.84-0.90]); sensitivity, 65.8% [48.6-80.4]; specificity, 92.7% [90.1-94.8]; and odds ratio, 6.43), 3.2 mmol/L for transfusion (AUC, 0.73 [0.69-0.77]; sensitivity, 51.9% [45.3-58.4]; specificity, 84.4% [79.9-88.3]; and odds ratio, 2.82), and 2.9 mmol/L for surgical intervention (AUC, 0.60 [0.56-0.65]; sensitivity, 49.4% [38.1-60.7]; specificity, 67.2% [62.8-71.5]; and odds ratio, 2.82) (13).

The PPV improved with increasing lactate cutoffs, but overall, PPVs were 10.9%-53.9% while NPVs were 91.7%-95.5%. Unlike sensitivity and specificity, PPV and NPV vary with prevalence levels. The prevalence of severe trauma in our cohort was low (9.4%), which means that a higher NPV does not necessarily prove that the test is beneficial in deciding whether the trauma is severe. In effect, some patients without severe trauma may still have high lactate values.

In our cohort, head injury was the most common type of injury (Table 2), which is also the most prevalent cause of trauma mortality in relevant studies (14-16). We found correlations of admission lactate positive with ISS and negative with GCS, indicating that patients with severe head trauma tended to have higher lactate values. Fu et al. (16) studied admission lactate values in 213 children with moderate and severe traumatic brain injuries. They reported that lactate value was an independent risk factor for mortality (odds ratio, 1.19; 95% confidence interval, 1.002-1.41) and that children with a lactate of 2 mmol/L or smaller had a 3.2% mortality, whereas those with a lactate higher than 2 mmol/L had a 46.6% mortality (P < 0.001).

This retrospective study had inherent limitations in data analysis and interpretation. The patients were all from a single center with modest pediatric trauma capacity, which may restrict the generalizability to large multispecialty pediatric institutions. However, this limitation might be counterbalanced by the extensive period and substantial sample size. An important limitation of the study is that we did not account for factors affecting lactate concentration, such as the timing between initial blood draw and injury, lack of information about prehospital resuscitation, or occurrence of sampling errors with hemolyzed specimens. Finally, the sensitivity and specificity might be skewed by the substantial proportion of patients with only mild injury.

To establish a more conclusive role of serum lactate as a prognostic biomarker for severe trauma, future research should include larger cohorts of severely injured children with adjustments made for different patient characteristics as well as specific clinical and injury features. It will be helpful to address the challenges of interpreting serum lactate concentration in children by excluding hemolyzed samples or including data on the time of blood draw relative to the traumatic event. Two adult trauma studies exploring lactate clearance yielded favorable findings, wherein poor lactate clearance was highly predictive of mortality (17,18). Hence, multiple sequential testing may be clinically relevant in accounting for factitious elevation or temporal changes in lactate values from the prehospital phase, ED, and hospital stay. Integrating serum lactate into different nonspecific trauma scoring systems and comparing their predictive performance to traditional prediction models may shed more light on the utility of lactate in pediatric trauma evaluations.

In conclusion, admission lactate demonstrates high specificity, a good rule-in capacity, for severe trauma in children at the cost of inadequate sensitivity. It should be used in conjunction with other screening tools when ruling out severe trauma.