Prognostic value of lactate-to-albumin ratio and inflammatory indices in pediatric traumatic brain injury: A comparative study with PRISM III

Abstract

Traumatic brain injury (TBI) is an important cause of mortality and morbidity in pediatric patients. We investigated the prognostic value of systemic inflammatory indices and pediatric trauma scoring systems in predicting adverse outcomes in pediatric patients with TBI. This retrospective cross-sectional study assessed 98 children diagnosed with TBI who were admitted to our pediatric intensive care unit between January 2020 and June 2024. Patients were classified into survivor and non-survivor groups, and their clinical and laboratory parameters were compared. The median age was 54.5 months (range: 1–215 months), and 69.4% of the patients were male. The overall mortality rate was 9.2% (of 9/98 patients). Non-survivors showed a significant increase in pediatric risk of mortality (PRISM) III scores (P < .001), pediatric trauma scores (P = .013), and pediatric Glasgow Coma Scale scores (P < .001) were significantly lower in non-survivors. Serum lactate (P = .004) and procalcitonin (P = .023) levels were significantly elevated in the non-survivor group. Among the predictive indices, the systemic immune-inflammation index (P = .043), pan-immune-inflammation value (P = .024), neutrophil-to-lymphocyte ratio (P = .012), systemic inflammation response index (P = .006), and lactate-to-albumin ratio (LAR; P < .001) were notably higher in non-survivors. According to receiver operating characteristic curve analysis, pediatric Glasgow Coma Scale scores (area under the curve = 0.881) and LAR (area under the curve = 0.854) strongly predicted mortality among the scoring systems and inflammatory indices, respectively. In pediatric patients with TBI, both pGCS scores and admission LAR values demonstrated a strong prognostic value in predicting mortality. Additionally, systemic inflammation response index, neutrophil-to-lymphocyte ratio, procalcitonin, pan-immune-inflammation value, and systemic immune-inflammation index may serve as useful markers for assessing the risk of death in children with TBI.

1. Introduction

Traumatic brain injury (TBI) remains the leading cause of neurological morbidity and mortality in children and young adults worldwide. It is estimated that approximately 10 million children are affected by TBI each year, highlighting its significance as a global public health concern.^[1,2] In the United States, pediatric TBI accounted for 4.1% of all TBI-related deaths in 2018 and 2019.^[3] Although there are no comprehensive TBI-specific epidemiological datasets available for Turkey, the most recent statistics from the Turkish Statistical Institute indicate that trauma and poisoning rank among the top causes of mortality in the 0 to 14-year age group, accounting for approximately 7% of all pediatric deaths.^[4]

TBI is generally classified into 2 phases: primary and secondary. Primary TBI refers to immediate, irreversible damage to brain tissue caused by mechanical force. In contrast, secondary TBI arises from a cascade of delayed pathophysiological events that contributes to progressive neuronal damage and cell death, such as excitotoxicity, oxidative stress, and neuroinflammation.^[5,6] The immune response following trauma plays a critical role in the initiation and propagation of secondary injuries, particularly in cases of polytrauma. Hypoxia, hypotension, shock, and acidosis can trigger the release of proinflammatory cytokines, resulting in systemic inflammation and further aggravating brain injury.^[7] Neutrophils, the most abundant leukocytes in circulation, are key mediators of this inflammatory response. Depending on the phase and context of the injury, they may aid tissue repair or exacerbate tissue damage.^[6]

Timely and accurate prognostication of pediatric TBI is crucial for optimizing treatment strategies and improving outcomes. Over the years, various clinical scoring systems and laboratory markers have been proposed for mortality prediction, particularly in adult TBI populations.^[8-12] In pediatric intensive care units (PICUs), several scoring systems are routinely employed to assess illness severity and guide clinical decision making. The pediatric risk of mortality (PRISM) III score has been validated as a robust predictor of mortality in critically ill pediatric patients.^[13,14] The Glasgow Coma Scale (GCS), widely used to assess the level of consciousness, remains a cornerstone for evaluating TBI severity across all age groups.^[2] The pediatric trauma score (PTS), specifically developed for children, incorporates age-appropriate physiological parameters and is commonly used to assess trauma severity and prognosis.^[15]

However, existing studies on pediatric populations are limited in scope and often rely on single prognostic markers.^[2,6,16,17] Given the multifactorial nature of TBI and the growing interest in systemic inflammatory responses as prognostic tools, this study aimed to evaluate the predictive value of multiple systemic inflammatory indices along with established clinical scoring systems for mortality prediction in children with TBI.

2. Methods

This retrospective cross-sectional study involved children with TBI admitted to the PICU at the University of Health Sciences, Bağcilar Training and Research Hospital, a tertiary care facility, between January 2020 and June 2024.

Trauma mechanisms were defined in accordance with the “Guidelines for Field Triage of Injured Patients” and included high-energy falls and motor vehicle collisions.^[18] All pediatric patients admitted to the PICU due to TBI resulting from these mechanisms were included. Patients with known chronic illnesses (hepatic, genetic, hematologic, or metabolic disorders) or those who died within the first 24 hours of admission were excluded.

Data on age, sex, trauma type, therapeutic plasma exchange, intracranial involvement, requirement for mechanical ventilation within the first 24 hours, renal replacement therapy, neurosurgical procedures, and mortality status were collected. Additionally, laboratory parameters noted during the initial 24 hours of admission to the PICU included complete blood count (monocyte, lymphocyte, neutrophil, and platelet counts), serum albumin, C-reactive protein (CRP), procalcitonin (PCT), and arterial blood gas assessment (including lactate, bicarbonate, and pH levels).

The following systemic inflammatory indicators were calculated:

• Systemic immune-inflammation index (SII) = (neutrophil count [10³/μL] × platelet count [10³/μL])/lymphocyte count (10³/μL).

• Neutrophil-to-lymphocyte ratio (NLR) = neutrophil count (10³/μL)/lymphocyte count (10³/μL).

• Platelet-to-lymphocyte ratio (PLR) = platelet count (10³/μL)/lymphocyte count (10³/μL).

• Lactate-to-albumin ratio (LAR) = lactate (mmol/L)/albumin (g/dL).

• Pan-immune-inflammation value (PIV) = (neutrophil count [10³/μL] × platelet count [10³/μL] × monocyte count [10³/μL])/lymphocyte count (10³/μL).

• Systemic inflammation response index (SIRI) = (neutrophil count [10³/μL] × monocyte count [10³/μL])/lymphocyte count (10³/μL).

At the time of the initial assessment in the emergency department, trauma severity was evaluated using the PTS,^[15] and the level of consciousness was assessed using the pediatric Glasgow Coma Scale (pGCS).^[19] The PRISM III score was calculated 24 hours after admission to the PICU.^[20]

2.1. Sample size calculation

Post hoc power analysis was conducted using the mortality outcome as the primary endpoint. Based on the observed mortality rate of 9.2% in our cohort and an area under the curve (AUC) of 0.854 for LAR, a minimum sample size of 88 patients was required to achieve a statistical power of 80% (β = 0.20) with a 2-sided alpha of 0.05 and an expected effect size (AUC > 0.80). Our study included 98 patients, which was deemed adequate to detect significant differences in the mortality predictors.

2.2. Statistical analysis

The Shapiro–Wilk test was used to assess continuous variables. Data without normal distribution are presented as median and interquartile range (interquartile range: 25th to 75th percentiles) and compared between the survivor and non-survivor groups using the Mann–Whitney U test. Categorical variables were presented as numbers and percentages and were compared using the chi-square or Fisher’s exact test. Furthermore, receiver operating characteristic (ROC) curve analysis was used to assess the predictive ability of the inflammatory indices and clinical scoring systems for mortality. The AUC was determined for all parameters, and optimal cutoff values were established using the Youden index to evaluate specificity and sensitivity. Patients with missing data required for the calculation of prognostic indices or outcome analyses were also excluded. Imputation was not performed. Statistical significance was set at P < .05. SPSS 30 for Mac (IBMCorp., Armonk) was used for data analyses.

3. Results

During the study period, 1300 patients were admitted to the PICU, of whom 199 were trauma-related admissions. Based on the inclusion and exclusion criteria, 98 pediatric patients with confirmed TBI were included in the final analysis (Fig. 1). The overall mortality rate was 9.2% (9/98 patients).

Figure 1.

Flowchart of the selection process of study patients. PICU = pediatric intensive care unit, TBI = traumatic brain injury.

3.1. Demographic and trauma characteristics

The median age of the cohort was 54.5 months (range: 1–215 months), and 69.4% (68/98) of the patients were male. There were no significant differences in age or sex distribution between the survivors and non-survivors (P = .559 and P = .853, respectively).

The mechanism of trauma was a fall from a height in 67.3% of all cases and motor vehicle accidents in 32.7%. This distribution did not differ significantly between survivors and non-survivors (P = .429).

3.2. Clinical findings and interventions

Subarachnoid hemorrhage was significantly more frequent in the non-survivor group (66.7% vs 22.5%, P = .010). There were no significant differences between the groups in the presence of intracranial hematoma, cerebral edema, pneumocephalus, or need for neurosurgical intervention (P > .05).

Hypotension was significantly more common among non-survivors (P < .001).

Intensive care interventions, such as renal replacement therapy, therapeutic plasma exchange, and mechanical ventilation, were significantly more frequent in non-survivors (P < .001, P < .001, and P = .002, respectively). Non-survivors also required longer mechanical ventilation durations (P = .009). However, the PICU length of stay did not differ significantly between the groups (P = .848; Table 1).

Table 1.

Clinical and laboratory parameters of survivor and non-survivor patients.

Parameter	Survivor (n = 89)	Non-survivor (n = 9)	Total (n = 98)	P value
Sex, male, % (n)	69.7% (62/89)	66.7% (6/9)	69.4% (68/98)	.853*
Age (mo), median (IQR)	54 [25–94]	68 [26–124]	55 [25–104]	.559†
Trauma type, % (n)
Fall	68.5% (61/89)	55.6% (5/9)	67.3% (66/98)	.429*
Motor vehicle	31.5% (28/89)	44.4% (4/9)	32.7% (32/98)
Radiological imaging, % (n)
Subarachnoid hemorrhage	22.5% (20/89)	66.7% (6/9)	26.5% (26/98)	.010 *
Pneumocephalus	50.6% (45/89)	55.6% (5/9)	51.0% (50/98)	.775*
Cerebral edema	20.2% (18/89)	33.3% (3/9)	21.4% (21/98)	.296*
Intracranial hematoma	80.9% (72/89)	88.9% (8/9)	81.6% (80/98)	.555*
Clinical parameters, % (n)
Need for surgery	29.2% (26/89)	11.1% (1/9)	27.6% (27/98)	.247*
Hypotension	6.7% (6/89)	66.7% (6/9)	12.2% (12/98)	<.001 *
Clinical scores, median (IQR)
PRISM III score	3 [0–9]	30 [12–38]	3 [0–10]	<.001 †
PTS	7 [4–9]	2 [0–5]	6 [4–9]	.013 †
pGCS	14 [9–15]	3 [3–5]	13 [7–15]	<.001 †
Laboratory parameters, median (IQR)
Neutrophils (103/µL)	10.6 [7.6–14.4]	16.5 [15.6–25.5]	10.9 [7.6–15.6]	.023 †
Lymphocytes (103/µL)	3.1 [1.6–6.60]	1.70 [1.22–2.50]	3.01 [1.70–6.55]	.027 †
Platelets (103/µL)	335 [252–415]	209 [199–363]	327 [245–413]	.110†
Mean platelet volume (fL)	9.40 [8.70–10.10]	9.80 [9.40–10.10]	9.45 [8.70–10.10]	.260†
Monocytes (103/µL)	0.70 [0.55–1.04]	0.93 [0.64–1.24]	0.72 [0.55–1.07]	.313†
Lactate (mmol/L)	2.50 [1.70–3.40]	9.20 [3.50–12.00]	2.60 [1.70–3.70]	.004 †
Albumin (g/dL)	3.99 [3.57–4.30]	2.60 [2.25–2.98]	3.88 [3.20–4.26]	<.001 †
C-reactive protein (mg/L)	6.23 [1.08–28.83]	18.86 [2.11–23.70]	7.06 [1.08–28.83]	.453†
Procalcitonin (ng/mL)	0.43 [0.12–3.27]	4.42 [0.70–24.16]	0.50 [0.12–3.90]	.023 †
Inflammation indices, median (IQR)
Lactate/albumin ratio	0.65 [0.42–0.97]	4.09 [1.09–4.35]	0.70 [0.42–1.09]	<.001 †
Neutrophil/lymphocyte ratio	3.51 [1.30–5.95]	10.70 [6.23–19.99]	3.77 [1.31–7.20]	.012 †
SIRI	2.45 [0.99–4.92]	13.93 [3.99–16.81]	2.58 [1.01–5.25]	.006 †
SII	1088 [444–2124]	2452 [896.–4284]	1131 [446–2183]	.043 †
PIV	751 [285–1624]	2420 [887–5056]	799 [317–1723]	.024 †
Treatment in PICU
Massive transfusion	2.2% (2/89)	0.0% (0/9)	2.0% (2/98)	.689*
Therapeutic plasma exchange	0.0% (0/89)	22.2% (2/9)	2.0% (2/98)	<.001 *
Renal replacement therapy	1.1% (1/89)	33.3% (3/9)	4.1% (4/98)	<.001 *
PICU stay (d)	5 [2–11]	5 [3–6]	5 [2–10]	.848†
Mechanical ventilation	33.7% (30/89)	88.9% (8/9)	38.8% (38/98)	.002 *
MV duration (d)	0 [0–3]	4 [2–6]	0 [0–4]	.009 †

IQR = interquartile range, MV = mechanical ventilation, pGCS = pediatric Glasgow Coma Scale, PICU = pediatric intensive care unit, PIV = pan-immune-inflammation value, PRISM = pediatric risk of mortality, PTS = pediatric trauma score, SII = systemic immune-inflammation index, SIRI = systemic inflammation response index.

^*Pearson chi-square.

^†Mann-Whitney U test. Significant P values are bold.

3.3. Laboratory parameters

Non-survivors had significantly higher neutrophil counts (P = .023), lactate levels (P = .004), and PCT levels (P = .023), whereas lymphocyte counts were higher in survivors (P = .027). The serum albumin levels were significantly lower in the non-survivor group (P < .001).

No significant differences were observed in CRP levels (P = .453), CRP-to-albumin ratio (P = .212), platelet count, mean platelet volume, or monocyte count.

3.4. Inflammatory indices and clinical scoring systems

Inflammatory indices, such as the SIRI (P = .006), LAR (P < .001), NLR (P = .012), SII (P = .043), and PIV (P = .024) were significantly higher in non-survivors.

No significant differences were found in the PLR (P = .479) or the CRP/albumin ratio (P = .212).

3.5. Predictive value for mortality (ROC analysis)

ROC curve analysis demonstrated that the pGCS (AUC = 0.881) and PRISM III had the highest discriminatory power for predicting mortality (AUC = 0.861). Among the inflammatory markers, LAR showed a comparable performance (AUC = 0.854), followed by SIRI (AUC = 0.777), NLR (AUC = 0.755), PCT (AUC = 0.730), PIV (AUC = 0.729), and SII (AUC = 0.707; Table 2, Fig. 2).

Table 2.

Diagnostic accuracy of inflammatory indices, pGCS and PRISM III score for predicting mortality in pediatric TBI patients.

Predictor	Cutoff value	Sensitivity (%)	Specificity (%)	AUC (95% CI)	P value
pGCS	<7.0	88.9	82.0	0.881 (0.770–0.992)	<.001
PRISM III	>14.0	88.9	82.0	0.861 (0.775–0.947)	<.001
LAR	>0.68	88.9	78.0	0.854 (0.762–0.946)	<.001
SIRI	>2.45	77.8	70.0	0.777 (0.642–0.911)	.004
NLR	>5.20	77.8	68.3	0.755 (0.604–0.905)	.006
PCT	>1.15	66.7	76.8	0.730 (0.576–0.884)	.024
PIV	>580	66.7	73.2	0.729 (0.569–0.888)	.021
SII	>1450	66.7	70.7	0.707 (0.542–0.872)	.030

Cutoff values were determined based on ROC analysis in this cohort using the Youden index.

AUC = area under the curve, CI = confidence interval, LAR = lactate-to-albumin ratio, NLR = neutrophil-to-lymphocyte ratio, PCT = procalcitonin, pGCS = pediatric Glascow coma scale, PIV = pan-immune-inflammation value, PRISM = pediatric risk of mortality, ROC = receiver operating characteristic, SII = systemic immune-inflammation index, SIRI = systemic inflammation response index, TBI = traumatic brain injury.

Figure 2.

ROC curves demonstrating the predictive accuracy of inflammatory indices, pediatric Glasgow Coma Scale and PRISM III score for mortality in pediatric TBI patients. PRISM = pediatric risk of mortality, ROC = receiver operating characteristic, TBI = traumatic brain injury.

4. Discussion

In this study, we evaluated pediatric patients with TBI and found that the LAR at admission had a significant prognostic value in predicting mortality. Notably, LAR was the inflammatory index most closely associated with mortality risk, showing a predictive power comparable to that of PRISM III and pGCS. Other inflammatory indices, including SIRI, NLR, PCT, PIV, and SII, also demonstrated strong performance in identifying patients at a higher risk of death.

The PRISM III score was significantly higher in the non-survivors, whereas both the pGCS and PTS scores were significantly lower. ROC curve assessment indicated that pGCS had the highest discriminatory ability for predicting mortality (AUC = 0.881), consistent with its performance in other critically ill pediatric populations. To date, relatively few studies have evaluated the prognostic utility of scoring systems specifically for pediatric TBI, and those that exist have used a wide range of tools.^[21,22] For example, a study of 97 pediatric polytrauma patients, including 35 with TBI, compared several scoring systems, including GCS, PTS, Injury Severity Score, and National Advisory Committee for Aeronautics score, and found that Injury Severity Score and National Advisory Committee for Aeronautics at admission were more effective than PTS or GCS in predicting outcomes.^[21] In contrast, our study found that the pGCS score had a superior predictive performance compared to both the PTS and PRISM III scores in children with TBI.

Metabolic response to trauma also results in elevated lactate levels. Several studies have shown that elevated lactate levels on admission are associated with increased morbidity and mortality in children with TBI.^[23] However, lactate levels can be influenced by numerous other clinical conditions, including alcohol or drug use, toxins, shock, post-cardiac arrest, diabetic ketoacidosis, seizures, thiamin deficiency, liver failure, and mitochondrial disorders.^[24-26] In recent years, studies in adult populations have demonstrated that LAR has greater prognostic value than lactate alone in predicting mortality in TBI.^[10,11] In a study of 273 adults with moderate-to-severe TBI, Wang et al^[11] reported that LAR outperformed lactate alone in predicting mortality (AUC_lactate = 0.733 vs AUC_LAR = 0.780). Another study involving 460 adults with TBI found that LAR was associated with 24-hour mortality and the need for massive transfusion.^[10] No study has specifically evaluated the prognostic significance of LAR in pediatric TBI. Our study is the first to address this gap by identifying LAR as the most powerful predictor of mortality among the inflammatory indices analyzed (AUC = 0.854). Given that LAR can easily be calculated from routine laboratory tests performed on admission, its strong predictive performance is particularly noteworthy. Compared to the more complex and time-dependent PRISM III score, LAR appears to be a more accessible and practical tool for early risk stratification in clinical settings.

The systemic inflammatory response triggered by TBI contributes to secondary brain damage. Neuroinflammation begins almost immediately after injury and typically peaks within 24 hours.^[2,7,16] Neuronal damage initiates the release of cytokines and chemokines, which leads to neutrophil activation. In addition, T cells, platelets, macrophages, and monocytes are activated. These cells, through their pro-inflammatory roles, can contribute to both tissue repair and exacerbation of neuronal injury.^[6] Recently, the prognostic utility of systemic inflammatory markers derived from hematological parameters has been increasingly investigated in patients with TBI.^[2,6,9] A study of 161 adults with moderate-to-severe TBI found that PLR, NLR, and SII were useful predictors of mortality.^[9] In a pediatric study by Marchese et al,^[2] NLR was associated with both positive cranial computed tomography findings and the presence of neurological symptoms in 219 children aged 2 to 18 years with TBI. Similarly, in a study of 374 pediatric patients with moderate-to-severe TBI, higher NLR values were observed in those with poor clinical outcomes.^[6]

Unlike earlier studies that focused on 1 or 2 markers, our study comprehensively compared a wide range of inflammatory indices in relation to pediatric TBI outcomes. Although LAR demonstrated the strongest performance, we also identified SIRI (AUC = 0.777), NLR (AUC = 0.755), PIV (AUC = 0.729), and SII (AUC = 0.707) as effective mortality predictors. However, our results suggest that the PLR is not a reliable indicator of mortality risk in children with TBI.

The most important limitation of this study was its single-center and retrospective design. Additionally, the number of non-survivors in the cohort was relatively small (n = 9), which may limit the statistical power and generalizability of our findings regarding the mortality predictors. Therefore, these results should be interpreted with caution and further validation in larger multicenter prospective studies is necessary. Another limitation was the variability in the time interval between the TBI event and blood sample collection at PICU admission, although in Turkey, trauma-exposed children are typically rapidly transferred to tertiary care centers.

Despite these limitations, the major strength of this study is that it is the first comprehensive analysis of multiple systemic inflammatory indices in the context of pediatric TBI mortality prediction.

5. Conclusion

This study demonstrated that LAR at admission, along with the pGCS score, is a valuable predictor of mortality in pediatric patients with TBI. In addition, other readily available, cost-effective inflammatory indices, such as SIRI, NLR, PIV, and SII, have been used to forecast mortality among pediatric patients with TBI. These findings support the use of inflammatory markers as accessible tools to aid in guiding clinical decision making and early risk stratification in pediatric TBI.

Author contributions

Conceptualization: Özlem Bostan Gayret, Abdulrahman Özel.

Data curation: Harun Çatak, Selen Mandel Işikli.

Formal analysis: Abdulrahman Özel, Servet Yüce.

Investigation: Harun Çatak, Selen Mandel Işikli.

Methodology: Özlem Bostan Gayret, Abdulrahman Özel, Servet Yüce.

Supervision: Servet Yüce, Meltem Erol.

Visualization: Özlem Bostan Gayret, Meltem Erol.

Writing – original draft: Özlem Bostan Gayret, Abdulrahman Özel.