Examining the association of elevated initial serum lactate with mortality and morbidity in trauma patients: a retrospective study

Abstract

Background

Elevated initial serum lactate (iSL) levels are frequently employed to assess trauma severity, but their predictive value for mortality and morbidity remains inconsistent. We evaluated the association of iSL with mortality and morbidity at Puerto Rico Trauma Hospital (PRTH).

Methods

This IRB-approved retrospective study included trauma patients ≥ 18 years with iSL measured within the first 48 h of admission to PRTH (July 2014-June 2019). Patients were classified as normal (4.5–19.8 mg/dL) or elevated (≥ 19.9 mg/dL) iSL levels. Group comparisons were conducted using t-tests, Wilcoxon rank-sum tests, Pearson’s chi-squared, or Fisher’s exact tests. Associations were evaluated with regression and ROC analyses.

Results

Among 536 patients, 54.3% had elevated iSL levels. Initially, elevated iSL was associated with in-hospital mortality (OR: 2.18, 95%CI: 1.36–3.51, p < 0.001), traumatic intensive care unit (TICU) admission (OR: 2.06, 95%CI: 1.46–2.92, p < 0.001), and need for mechanical ventilation (MV) (OR: 2.80, 95%CI: 1.97–3.98, p < 0.001). However, adjusted analyses showed no significant associations (Mortality-AOR: 1.72, 95%CI: 0.97–3.04, p = 0.06; TICU-AOR: 1.11, 95%CI: 0.71–1.75, p = 0.65; MV-AOR: 1.49, 95%CI: 0.89–2.49, p = 0.13). Both iSL (AUC: 0.59, 95%CI: 0.54–0.64) and ISS (AUC: 0.59, 95%CI: 0.54–0.64) demonstrated limited ability to predict mortality, with no statistically significant difference between them (p > 0.99). Patients with elevated iSL experienced prolonged hospital and TICU stays and severe injuries.

Conclusions

Elevated iSL levels may not independently predict mortality, TICU admission, or the need for MV in trauma patients. However, their rapid availability supports their use alongside other clinical markers to guide trauma care decision-making and improve trauma outcomes.

Background

Elevated serum lactate levels are early biomarkers for hypoxia, disease progression, bacterial load, sepsis, cancer, and cellular stress, providing critical insights into a patient’s metabolic state [1,2,3,4]. In critical illness care, lactate measurements help identify organ hypoperfusion and oxygen deficits resulting from anaerobic metabolism [5, 6]. The use of a quick, sensitive, and affordable blood test to screen lactate levels between 0.5 and 48 h after trauma admissions [7, 8] can potentially improve clinical decision-making, patient management, resuscitation, and treatment, consequently reducing clinical complications [5, 6, 9]. However, a standardized protocol for screening has yet to be established.

Elevated initial serum lactate (iSL) levels are frequently employed to stratify trauma patients in acute care settings [7, 10]. They are also associated with multi-organ failure, a higher need for mechanical ventilation (MV), signs of severe injury [6], hospitalization for more than 48 h [2], and intensive care unit (ICU) admission [4] in the emergency department (ED). Furthermore, elevated iSL levels are associated with a 34% 30-day mortality rate in non-trauma patients [9] and mortality of up to 28% in trauma patients [1, 7, 10]. Nevertheless, previous studies on the relationship of iSL with mortality and ICU admission have yielded mixed results [2, 3, 5]. According to some studies, ICU admission [11] and mortality in trauma patients [5, 12] have no association with elevated iSL levels.

These inconsistencies highlight the need to evaluate the role of iSL levels in clinical practice, especially regarding mortality and ICU admission. This study wants to analyze data from the Puerto Rico Trauma Hospital (PRTH), the only trauma center in the Caribbean, providing valuable insights into the utility of iSL within a unique healthcare context and a specific demographic. Our primary aim was to examine the predictive ability of iSL levels related to mortality at PRTH compared to individuals with normal iSL levels. The secondary aim was to describe the morbidity (ICU admission and MV needs) and evaluate their association with iSL levels in trauma patients. In this retrospective analysis, we hypothesized that trauma patients with elevated iSL levels would have higher in-hospital mortality and worse morbidity than patients with normal iSL levels. These findings may lead to the incorporation of early iSL assessment as a standard procedure for identifying critically ill patients, guiding patient care, optimizing hospital management, and enhancing hospital outcomes in trauma settings.

Methods

The PRHT is the only medical center certified as a tertiary Level II trauma facility that treats patients with multiple traumatic injuries from urban and rural regions in Puerto Rico and the Caribbean. Participating in the US National Trauma Data Bank (NTDB) enables standardized data collection and benchmarking against other trauma centers. This NTDB was our primary database source, which included the demographics, injury-related parameters, and hospital courses of patients admitted between July 2014 and June 2019. The PRTH Medical Center Laboratory provided data on iSL, albumin, and creatinine levels in our study.

Patient population

The study considered all trauma patients meeting the following inclusion criteria: (a) aged 18 years and older in the registry; (b) patients who underwent iSL-measured tests performed within the first 48 h of admission, as determined by clinical judgment; (c) patients with penetrating or blunt injuries. We excluded patients with the following criteria: (a) Patients with missing information on mortality status, admission, or discharge data; (b) patients without iSL-measured tests or patients with iSL levels < 4.5 mg/dL; (c) patients with burn or other type of injuries. Patients with iSL levels < 4.5 mg/dL were excluded to minimize confounding by omitting cases with distinctly lower metabolic stress, making them physiologically different from the normal and elevated groups [4]. Although this exclusion introduces some selection bias, it ensures the comparability of the study groups. Notably, the study had minimal missing data on mortality status, admission, or discharge data, making the impact of these exclusions negligible.

Patients were categorized into two groups based on iSL levels: normal (4.5–19.8 mg/dL) and elevated (≥ 19.9 mg/dL). The threshold for elevated iSL levels (2.2 mmol/L or 19.9 mg/dL) was adopted based on the Régnier et al.’ study, highlighting its clinical relevance for predicting poor outcomes among trauma patients [13]. We selected the normal range for iSL levels (0.5–2.2 mmol/L or 4.5–19.8 mg/dL) because it reflects a balanced metabolic state, indicating proper metabolic function and adequate oxygen delivery to tissues, according to clinical and diagnostic guidelines for measuring lactic acid [14].

Outcomes and variables

The primary outcome was in-hospital mortality, compared between normal and elevated iSL groups (exposure of interest). Secondary outcomes included morbidity indicators, including trauma intensive care unit (TICU) admission and need for MV, evaluated based on iSL levels.

We categorized additional variables into three domains: sociodemographic, injury-related, and hospital courses. For the sociodemographic domain, we considered parameters such as sex, age [18–39 years, 40–64 years, or 65 years or older], and health insurance status [private, public, or uninsured]. Data from the injury-related profiles included the mechanisms of injury [motor vehicle collisions (MVC), gunshot wounds (GSW), stab wounds (SW), falls, pedestrians, or others], type of injury [blunt or penetrating trauma], and Injury Severity Score (ISS) [1–14, 15–24, or 25–75]. In addition, the study collected data on the Glasgow Coma Scale (GCS) [mild (13–15), moderate (9–12), or severe (3–8)], injured body regions [head and neck, chest, abdomen, extremities, and external (skin injury)], and vital signs and clinical profiles of the patients [systolic blood pressure (SBP), respiratory and heart rates, creatinine, and albumin]. Finally, the hospital course domain included hospital length of stay (LOS), trauma intensive care unit (TICU) LOS, and days on MV.

Statistical analysis

Descriptive statistics included mean, standard deviation (SD), median, 25^th percentile, and 75^th percentile for continuous data, whereas we used frequencies and percentages for categorical variables. Comparisons between groups (elevated vs. normal iSL) were performed using the unpaired t-test for continuous data (or Wilcoxon rank-sum test for non-normally distributed continuous data) and Pearson’s chi-square test (or Fisher’s exact test for small sample sizes or expected cell counts below 5) for categorical data, as appropriate. We performed univariable and multivariable logistic regression analysis to examine associations between elevated iSL and outcomes, including TICU LOS, MV requirements, and in-hospital mortality. Variables statistically significant in the bivariate analyses were selected and incorporated into the multivariable logistic regression models, using a stepwise elimination process to identify the most relevant factors while controlling for potential confounders. We evaluated the predictive capability of iSL levels for in-hospital mortality through receiver operating characteristic (ROC) curve analysis, with the ISS serving as the widely recognized gold standard for assessing trauma severity. This study determined sensitivity and specificity based on the area under the curve (AUC) with a 95% confidence interval. Statistical significance was p < 0.05 for bivariate and multivariable analyses. The software used for the evaluation was STATA version 14 (STATA Corp., College Station, Texas, US).

Results

Among all patients hospitalized at the PRTH, 536 had their iSL levels assessed within 48 h of arrival; of these, 291 (54.3%) had elevated iSL levels. Both groups were predominantly male, with 8% more males in the high iSL group, a statistically significant difference (82.8% vs. 74.3%; p = 0.02). This group had predominantly patients between 18 and 39 years (50.2% vs. 43.6%; p = 0.32). Most patients had private health insurance regardless of their iSL level (60.5% vs. 63.7%; p = 0.73). MVC was the most prevalent mechanism of injury in both groups, with a lower proportion observed in the elevated iSL group (38.8% vs. 42.9%; p < 0.01). However, the proportion of patients with GSW doubled in this group (26.5% vs. 13%; p < 0.01), making it the second most common mechanism among patients with elevated iSL. Although blunt trauma was the most prevalent type of injury, the proportion of individuals with penetrating trauma was substantially higher in the elevated iSL group than in their counterparts (31.3% vs. 19.6%; p < 0.01) (Table 1).

Compared to the control group, approximately 7% more patients with elevated iSL had higher ISS values (Category: 25–75) (22.3% vs. 14.7%; p < 0.01). Moreover, patients with elevated iSL levels showed a noticeably higher percentage of severe GCS scores, almost twice that of their counterparts (25.6% vs. 13.1%; p < 0.01). As for the incidence of injury by body region, more than half of the patients in both groups presented with chest (61.9% vs. 58.8%; p = 0.47) and abdominal injuries (64.6% vs. 50.2%; p < 0.01), with the latter being significantly more prevalent among elevated iSL patients. Similarly, this group had a higher rate of external trauma (43.3% vs. 32.7%; p = 0.01) compared to the control group (Table 1).

High iSL was associated with tachypnea, tachycardia, elevated creatinine levels, and reduced albumin levels, with 10–20% more in each category (p < 0.01). This group was also marginally associated with SBP < 90 mmHg (p = 0.07). Furthermore, the group with elevated iSL levels was likelier to require admission to the TICU (65% vs. 47.4%; p < 0.01). They spent more days in the TICU, with a median of 17 days, five to six days longer than those with normal iSL (p < 0.01). Although both groups required MV for approximately 10 days (p = 0.12), the elevated iSL group demonstrated a higher frequency of MV utilization (67% vs. 42%; p < 0.01). They also had a higher hospital LOS with a median of 19 days, eight days longer than the control group (p < 0.01). The in-hospital mortality rate nearly doubled among the elevated iSL patients (22.7% vs. 11.8%; p < 0.01) (Table 1).

Table 2 presents the results of the univariable and multivariable analyses. The unadjusted regression showed that patients with elevated iSL had 2.2 times the odds of in-hospital mortality (95%CI: 1.36–3.51, p < 0.001). Evaluating morbidity indicators, this group had around 2-fold increased odds of TICU admission (OR: 2.06, 95%CI: 1.46–2.92, p < 0.001) and 3-times the odds of requiring MV (OR: 2.80, 95%CI: 1.97–3.98, p < 0.001). After adjusted analysis, the association found with mortality lost statistical significance when adjusting for age, mechanism of injury, ISS, GCS, SBP, respiratory and heart rates, creatinine, and albumin (AOR: 1.72, 95%CI: 0.97–3.04, p = 0.06). Also, the associations found with TICU admission (AOR: 1.11, 95%CI: 0.71–1.75, p = 0.65) and the need for MV (AOR: 1.49, 95%CI: 0.89–2.49, p = 0.13) dissipated after adjusted analyses.

Table 3 shows the ROC analysis to assess the predictive value of iSL levels for in-hospital mortality and compare it to the ISS. In the predictive power for in-hospital mortality, iSL levels (cut-off of 19.9 mg/dL) had an area under the curve (AUC) of 0.59 (95% CI: 0.54–0.64) with a sensitivity of 69.47% and a specificity of 48.98%. Also, the ISS (> 15) had an AUC of 0.59 (95% CI: 0.54–0.64) with a sensitivity of 70.53% and specificity of 47.85% (Fig. 1). Both predictors had a poor discriminatory predictive value for mortality with no considerable power of prediction differences between ISS and iSL levels (p > 0.99).

Table 1 Sociodemographic, injury-related profile, and hospital course between patients with elevated and normal initial serum lactate levels (iSL) at the Puerto Rico Trauma Hospital (July 2014 – June 2019) (N = 536)

Table 2 Univariable and multivariable logistic regression analyses of the effects of elevated initial serum lactate levels (iSL) on morbidity and in-hospital mortality among trauma patients admitted to the Puerto Rico Trauma Hospital (July 2014 – June 2019) ^±

Fig. 1

ROC curve comparison of injury severity score (ISS), and initial serum lactate levels (iSL) for predicting in-hospital mortality in traumatic patients admitted to the Puerto Rico Trauma Hospital (July 2014 – June 2019)

Table 3 Comparison of sensitivity, specificity, and AUC of injury severity score (ISS) and initial serum lactate levels (iSL) in predicting in-hospital mortality among trauma patients at the puerto rico trauma hospital (July 2014 – June 2019)

Discussion

In clinical trauma scenarios, the predictive utility of elevated iSL levels remains inconsistent across studies [3, 5, 7, 9,10,11, 15]. This research used data from the PRTH, the only trauma center in the Caribbean, to explore the utility of iSL levels within a unique healthcare system and population. The study primarily examined the predictive ability of elevated iSL levels for mortality. For secondary outcomes, we evaluated the association with morbidity at PRTH. Consequently, this study assessed the predictive ability of elevated iSL levels for mortality in comparison to normal iSL levels. Contrary to our hypotheses, we did not find a significant association between elevated iSL levels and mortality, TICU admission, or MV needs after adjusting for sociodemographic and injury-related parameters. Moreover, the evaluated iSL levels demonstrated suboptimal performance in predicting mortality, exhibiting high sensitivity but low specificity. Nevertheless, we noted that more patients with elevated iSL levels had higher ISS scores, elevated creatinine levels, and prolonged hospitalizations.

Initially, we observed a 2.2-fold increase in the odds of in-hospital mortality among patients with high iSL levels. However, after adjusting for confounding factors, this association became non-significant, consistent with previous studies that found no link between these parameters after multivariable analysis [3, 5, 12, 15]. Conversely, some studies have found a significant relationship between elevated iSL levels and mortality, with odds ratios ranging from 1.17 to 5.7 [1, 7, 10, 13]. Notably, our study defined elevated iSL levels as ≥ 2.2 mmol/L, which differs from other investigations that used thresholds of ≥ 3 [10], ≥ 4 [1], or ≥ 5 mmol/L [13]. Some studies focused on specific mortality timelines (e.g., 24 h, 48 h, or 30 days) and conducted regression analyses with different confounding variables, such as sex, race, arterial pressure score, revised trauma score, or APACHE II [1, 7, 10, 13]. In contrast, our focus remained on overall in-hospital mortality and included sociodemographic and injury-related variables. Differences in results may arise from variations in cut-off values, multivariable analyses, and comparison groups employed in different studies. For instance, Dezman et al. reported higher mortality among patients with iSL levels of ≥ 7 mmol/L compared to those with 3–4 mmol/L [7], Venkatesan et al. evaluated iSL levels in patients with hip fractures [10], and Gale et al. compared iSL levels with base deficit [1].

ROC analysis showed iSL had limited predictive ability for in-hospital mortality (AUC = 0.59, 95% CI: 0.54–0.64) using a 19.9 mg/dL cut-off. By comparison, Régnier et al.. reported a stronger predictive value (AUC = 0.77, 95% CI: 0.60–0.87) with a similar threshold [13]. This discrepancy may arise from differences in study design, patient demographics, and outcome definitions. However, our findings are consistent with those of Pal et al., who found that lactate levels (cut-off: 2 mmol/L) were weak predictors of outcomes in trauma patients, even after adjusting for age in their ROC curve and logistic regression analyses [5]. In our study, iSL and ISS exhibited poor discriminative ability for in-hospital mortality (AUC: 0.59, 95% CI: 0.54–0.64). Nevertheless, iSL demonstrated high sensitivity (69.47%) for predicting mortality, comparable to the ISS sensitivity (cut-off > 15: 70.53%) for the same outcome. Importantly, iSL levels are quicker to assess than ISS, making them a valuable clinical tool for identifying at-risk patients in situations requiring rapid decision-making. Despite its high sensitivity, the low specificity of iSL suggests a tendency to overestimate mortality risk in survivors.

Regarding the need for MV or TICU hospitalization, we found no significant association with elevated iSL levels. This contrasts with prior studies, such as Régnier et al. [13] and Davis et al. [16], which identified elevated iSL as a potential predictor of these outcomes in trauma patients. Régnier et al. demonstrated that higher lactate levels were significantly associated with increased ICU admissions, reporting an AUC 0.70 (95% CI: 0.65–0.74) for predicting critical outcomes, while Davis et al.. suggested that iSL, although less predictive than base deficit, was still a marker of severity requiring resuscitative efforts. Lavery et al., in their adjusted analysis, identified a correlation between high lactate levels and surgical ICU admission [11]. The discrepancy between our study and that of Lavery et al. may be due to variations in the blood sample collection sites (i.e., arterial vs. venous lactate), a factor not considered in our research. Our study employed an adjusted multivariable approach to reduce the effect of confounder variables, potentially explaining the discrepancy with prior findings. Also, these discrepancies may stem from variations in patient populations, timing of lactate measurement, and differences in defining critical outcomes.

Our results align more closely with those of Dekker et al. [8], who found that iSL levels did not outperform lactate clearance metrics or base deficits in predicting ICU admission or MV. Similarly, Pal et al. [5]. concluded that lactate levels alone were inadequate predictors of trauma outcomes when adjusted for confounding factors such as GCS. Although our study did not establish a significant relationship between iSL levels and the need for MV or TICU admission, it underscores the complex nature of trauma prognostication. Elevated iSL levels may still provide valuable insights when combined with other clinical indicators, such as GCS or ISS, as suggested by Wang et al. [17] in their evaluation of lactate levels alongside GCS-pupils scores for traumatic brain injury. Future research should focus on integrating multiple predictors to improve lactate measurement accuracy and clinical utility in trauma settings.

Our findings are consistent with previous studies reporting that elevated iSL levels have extended hospitalizations, higher ISS, penetrating injuries, and GSW, underscoring their association with critical illness severity [1, 2, 4, 6,7,8, 16, 18]. However, we found a more extended hospitalization LOS for individuals with elevated iSL than previously reported [2], emphasizing the importance of thoroughly monitoring and managing these patients. Additionally, we noted elevated creatinine levels in patients with elevated iSL, an association that, to our knowledge, has not been previously observed in this context. These findings suggest that monitoring creatinine levels could serve as a critical indicator for assessing the severity of illness as reported by Lautrette [19].

Lactate levels represent dynamic physiological responses triggered by multiple cellular mechanisms in response to injury [2, 17]. After adjusting for variables such as age, mechanism of injury, ISS, and vital signs, our results show that iSL levels are not independently associated with mortality. Although other studies have stated that lactate levels are reliable indicators of poor outcomes in trauma patients in triage settings [2, 3, 13, 18], our findings did not support this statement for the parameters studied (e.g., need MV and TICU admission). However, healthcare practitioners should consistently monitor lactate levels in patients and use iSL levels alongside other biomarkers and clinical assessments to evaluate the severity and progression of trauma-injured patients [5]. A standardized point-of-care for iSL assessment is necessary to identify patients at risk of poor outcomes and to ensure optimal clinical care for trauma patients.

There are several limitations in our study. First, we used the PRTH registry and Medical Center Laboratory as retrospective data sources, which could introduce information bias and potentially underestimate the association. Second, our cohort consisted exclusively of patients treated at the PRTH, which mainly receives cases of moderate to severe trauma either directly from the scene or as referrals from other hospitals. As a result, our study may not fully represent all trauma patients on the island, excluding those who died at the scene or those with less severe injuries. Third, our generalization could be potentially impacted because we excluded patients with lower iSL levels from our analysis. We excluded lower iSL levels from our analyses to better understand the relationship between normal and elevated iSL levels with mortality and morbidity. Also, we acknowledge the absence of data on patients with lower iSL levels and note that only 0.03% of cases had missing data for the specified variables; however, these exclusions were unlikely to affect the study population. Fourth, this study only included individuals with measured lactate levels within the first two days of admission, which could impact the generalizability of our findings. Although other research indicated that lactate levels within the first 0.5-2 admission hours may predict mortality in trauma patients [7, 8, 12, 13], no standard protocol has been established for iSL-level measurements. Fifth, our study only considered iSL levels and did not evaluate other options, such as lactate clearance and multiple lactate timing measurements, that may better describe the role of serum lactate levels in mortality based on prior studies [1, 7, 13]. To the best of our knowledge, this study is one of the few that has conducted regression evaluations for hospital outcomes [11] to minimize the potential impact of confounding variables (including sociodemographic and injury-related parameters), as previous studies assessed hospital outcomes using only bivariate analyses [8, 13, 16].

Conclusion

This study, conducted at the only trauma center in the Caribbean, found that elevated iSL levels had limited value in predicting in-hospital mortality and were not linked to morbidity outcomes, such as TICU admission or the need for MV. However, their relationship with higher ISS scores, prolonged hospital stays, and elevated creatinine levels highlights their potential as a supplementary marker of trauma severity. While iSL levels independently lack sufficient discriminatory power, their rapid availability makes them a valuable tool when integrated with other clinical parameters, supporting faster decision-making in trauma care. Standardized protocols should be established to enhance the clinical applicability of iSL measurements, considering the timing of assessments and appropriate cut-off thresholds. Future research should explore the combined predictive power of iSL levels with established scoring systems to improve the accuracy of mortality and morbidity predictions in trauma patients, particularly within unique healthcare settings like the PRTH.