Early recognition and treatment of injury are critically important in the care of trauma patients. In addition to the primary and secondary surveys, a variety of trauma scores are available to help physicians assess the severity of the injured patient (1). Because these scores involve unwieldy calculations, they are often used later in the hospital course rather than early in the emergency department (ED) (2, 3). For that reason, many clinicians use specific serum blood values, such as arterial blood gas, base deficit, and lactate as rapid proxies for injury severity. Serum lactate is a by-product of anaerobic metabolism. It can be measured using a rapid assay in the ED via a venous puncture. We hypothesized that the initial serum lactate (ISL) level could have a prognostic value for acutely injured patients.
Several investigators have examined the prognostic value of lactate levels (venous or arterial) in trauma. Their conclusions have been limited by relatively small cohorts or lack of specific time definitions for on-admission lactate (3–7). Other available studies have focused on specific populations and are summarized in Table 1 (7–18).
MATERIALS AND METHODS
We received approval for this retrospective study from the Einstein Healthcare Network Institutional Review Board. Our institution is a Pennsylvania Trauma Systems Foundation–accredited Level 1 trauma center. We maintain a prospective database of all acutely injured patients who present to the institution. We reviewed 3,775 consecutive trauma patients admitted between January 2007 and June 2012, which was the available time frame with the most complete data. There was no significant change in our trauma care system within this period. We excluded 571 patients with unknown ISL and 72 patients younger than 18 years. We removed another 110 additional patients who presented to the ED more than 24 h after their initial injuries. After these adjustments, we focused on the timing of the ISLs. We excluded patients with the following issues with their ISL: measured outside the initial 24 h, unknown date, unknown time, or timing error. After applying these criteria, a total of 2,421 adult patients remained with reasonable timing between injury and ED admission, as well as ED admission and measured ISL. We felt it was important to include only ISLs measured early in the hospital course because that is most realistic when caring for actual patients. As such, the next step was minimizing the interval between the time of ISL and ED admission as much as possible. In reviewing the data, we found that 80.8% of the patients had their ISLs drawn within the first 35 min (Fig. 1). When that 35-min interval was further broken down into quintiles (7 min each), it became apparent that including ISLs greater than 35 min came with diminishing returns (Fig. 2). A total of 464 patients who had SL measured beyond the first 35 min after admission were excluded. Finally, 16 patients with incompletely documented injuries were excluded. A total of 1,941 patients remained for analysis (Fig. 3).
Our primary outcome was in-hospital mortality. Our secondary outcomes were intensive care unit admission (ICUA) and operative intervention (OI). We divided the cohort into two groups (survivors versus nonsurvivors) based on the defined primary outcome (in-hospital mortality).
We compared the two groups statistically. We tested all the data for normality and calculated medians (25th – 75th interquartile range [IQR]) for variables with non-normal distribution. Continuous variables were compared with a Mann-Whitney U test. Ordinal variables were tested with a chi-square test. Because mortality is a dichotomous variable, we conducted a logistic regression analysis to evaluate its potential predictors. For our primary outcome (mortality), we considered the following potential risk factors: age, gender, initial systolic blood pressure (SBP), heart rate (HR), respiratory rate, Glasgow Coma Scale (GCS), blood transfusion (packed red blood cells), and ISL. We included the Injury Severity Score (ISS), need for intubation, and mechanism of injury in the univariate logistic regression. Subsequently, we used a multivariate logistic regression model to examine the eligible variables (P < 0.2) derived from the univariate analysis. Goodness-of-fit was tested with a Hosmer-Lemeshow test. The colinearity among the covariates was tested and did not exist. In addition, we performed a receiver operating characteristic (ROC) analysis to better evaluate the discriminating ability of ISL on mortality as well as our two secondary outcomes.
For our secondary outcomes (ICUA and OI), we used the same univariate and multivariate logistic regression testing process as well as ROC analysis. We analyzed all data using IBM SPSS Statistics (Armonk, NY).
A total of 3,775 trauma patients existed in our electronic database in the study period. All records were unique, and there was no duplication. A total of 1,941 patients met the inclusion criteria. The on-admission information for the two groups (survivors vs. nonsurvivors) is tabulated in Table 2. In univariate logistic regression, all the covariates were statistically significant to predict mortality except for gender (P = 0.347). Subsequently, we excluded the gender and tested all the remaining variables simultaneously in a multivariate logistic regression model. In the multivariate logistic regression, SBP, HR, and blood transfusion lost significance but others including the ISL stayed significant (Table 3). The odds ratio for intubation was 6.57, which was the greatest observed odds ratio. This was not unexpected because the need for intubation is associated with poorer outcome and higher mortality. The associated P value and odds ratio for ISL were 0.015 and 1.010 (95% confidence interval, 1.002 – 1.019), respectively. We created an ROC curve with SL for mortality. The area under the curve was 0.63 (Fig. 4).
We followed the same process to test our two secondary outcomes (ICUA and OI). In our cohort, 1,063 (54.7%) of 1,941 patients were admitted to the ICU. The median ISL for patients who had an ICUA was 23 mg/dL (IQR, 14 – 36) vs. 20 mg/dL (IQR, 14 – 31) mg/dL for patients who did not have an ICUA (P = 0.002). Among the previously mentioned variables, ISL, age, GCS, ISS, blood transfusion, and gender were statistically significant in the univariate analysis. However, only age, ISS, and intubation stayed significant in the multivariate analysis, whereas ISL lost significance (P = 0.358). The area under the curve in the ROC analysis was 0.54 (Fig. 5).
Finally, OI was evaluated as the other secondary outcome. In this cohort, 851 (43.8%) of 1,941 patients were taken to the operating room. The median ISL was 24 mg/dL (IQR, 17 – 38) for patients who needed OI versus 19 mg/dL (IQR, 13 – 30) for patients who did not need OI (P < 0.001). Of the previously listed variables, HR and GCS were not statistically significant in the univariate analysis, but ISL was. In the multivariate analysis, ISL was still significant for OI (P = 0.033). The area under the curve in the ROC analysis was 0.60 (Fig. 6).
This study is a general evaluation of the prognostic significance of ISL. It includes all adult trauma patients admitted to our center for 5 years. All the included lactates were venous samples measured within 35 min of admission. Therefore, our study has a time-based (albeit arbitrary) definition for ISL. Our primary outcome was in-hospital mortality. The median survival for nonsurvivors was 1 day (IQR, 1 – 6). The rate of death in this cohort was 121 (6.2%) of 1,941.
The median ISL was statistically different between survivors (21 mg/dL) and nonsurvivors (32 mg/dL). The ISL was associated with a significant P value (P = 0.015) to predict in-hospital mortality when tested with multivariate logistic regression. However, the odds ratio was only 1.010. Therefore, ISL was an independent, but weak, predictor for mortality. This is consistent with the results of Guyette et al. (2), although their focus was prehospital lactate. Guyette and colleagues (2) reviewed prehospital venous or capillary SL levels. They studied 1,168 patients with a primary outcome of in-hospital mortality.
Their patients were predominantly blunt trauma patients of higher acuity. Their frequency of penetrating injury was approximately 4%. In contrast, we studied in-hospital ISL in a larger cohort and with a more balanced ratio of penetrating-blunt injury (22.8% – 77.1%). Of note, Guyette et al. considered a threshold value of greater than 2 mmol/L (18 mg/dL) for lactate. We considered all serum lactates, regardless of value or the normal range, to minimize the possibility of any selection bias. In our ROC analysis, the area under the curve for ISL was only 0.63 when it was measured for mortality (Fig. 4). This result is very similar to the analysis performed by Callaway et al. (10), leading us to agree that ISL is unable to discriminate for mortality. To be a strong predictor, the area under the curve would need to be between 0.8 and 0.9.
The lack of discrimination for mortality (hence, being a weak predictor) can be gleaned from Table 2, where the IQRs of the ISL for nonsurvivors and survivors overlap. Interestingly, the same overlap was observed by Guyette et al. (2). They reported an odds ratio of 1.23 for lactate to predict mortality. Concordantly, Pal et al. (5) discussed the existence of an overlap for on-admission ISL between survivors and nonsurvivors while finding a statistically significant lower mean for lactate in survivors (P < 0.001). They confirmed their finding on ROC analysis.
In our study, we found some associations among some variables (respiratory rate, blood transfusion, ISS, GCS, mechanism of injury, etiology, localization of injury, and need for intubation) and mortality (Table 2). Although all of these variables have a value of P < 0.05, they do not rise to the level of predictors when put through a logistic regression. A similar finding was described by Guyette et al. (2), who found some associations between age, mechanism of injury, transport type, localization of injury, GCS score less than 15, heart rate, and peripheral serum lactate. As Pal et al. (5) remarked, association and predictive value are not synonymous.
Pal et al. (5) studied admission SL and tried to determine its predictive value on 48-h mortality. They studied 5,995 patients and demonstrated lack of predictive value for elevated ISL. However, they evaluated the overall predictive value of lactate using ROC analysis and left the multivariate logistic regression for certain subpopulations, including advanced age, gender, ISS greater than 20, mechanism of injury, deaths occurring within 48 h of admission, and patients with GCS score of 7 or higher (5). Although they tried to evaluate the usefulness of SL using their ROC analysis, they did not apply a specific time interval for SL.
To refine this methodology, we used a time-based ISL in our ROC analysis and performed a multivariate analysis for all patients. Hence, we did not limit the analysis to specific subpopulations. We evaluated the prognostic value of the lactate for the two secondary outcomes (i.e., ICUA and OI), which were not addressed by Pal et al.
In another study, Régnier et al. (19) reviewed 586 patients to evaluate prognostic significance of blood lactate and lactate clearance in trauma patients. They stressed the importance of lactate clearance within the first 2 h after admission. All of their subjects received prehospital mobile ICU care because injury severity was high enough to warrant it. Therefore, their patients were partially treated before presenting to the ED. In contrast, none of our trauma patients received any prehospital care. In addition, our cohort was at least three times larger.
Callaway et al. (10) studied SL and base deficit as predictors of mortality in 588 elderly trauma patients versus 1,188 younger trauma patients. They demonstrated that both lactate and base deficit were associated with significantly increased mortality in normotensive elderly blunt trauma patients. Their study was focused on a specific age group (elderly) and mechanism (blunt trauma). We included all adult ages with both blunt and penetrating injuries.
Lavery et al. (4) considered a specific time interval for SL in their study of 375 patients—limiting it to within the first 10 min after admission. Their cohort was relatively small and was mostly composed of motor vehicle accident victims (51%). Our study covered more patients with diverse etiologies. However, their incorporation of a defined time frame for the collection of SL strengthened their methodology and conclusions. It was with this in mind that we also decided to set a fixed and realistic time frame for the collection of ISL.
We designed our study with the existing literature in mind to try to replicate their methodological strengths and avoid their limitations. Accordingly, we reviewed the serum lactates in a large cohort of patients and within a defined time frame. We studied all patients presenting to the ED, regardless of mechanism of injury. In addition, we reviewed all serum lactates, regardless of the normal range or any other arbitrary cutoff.
Based on our results, ISL can continue to be a useful component of our diagnostic armament to improve the identification of patients who are at risk for in-hospital mortality. However, it is not a strong predictor by itself. With regard to our secondary outcomes, ISL predicted OI in the multivariate logistic regression (P = 0.033). However, the area under the curve was 0.60, which, similar to mortality, indicated a lack of good prediction. The ISL did not predict ICUA in the multivariate analysis (P = 0.358). The area under the curve was 0.54. In addition, the curve partially overlapped the diagonal line (Fig. 5). Although ICUA is subjective rather than objective and is generally decided by the surgeon, we chose to consider it as one of our end points because it is an important outcome.
It is important to recognize some limitations of this study. First, while every effort was made to ensure accuracy of the data set, we acknowledge that a retrospective study is subject to errors including data entry errors, inaccuracies, inconsistencies, and omissions. Furthermore, 571 (15.1%) of 3,775 of the patients had an unknown SL (Fig. 1). By design, they were excluded from the analysis. This could represent a selection bias. Of these 168 (29.4%) of 571 patients, which comprised only 168 (4.4%) of 3,775 of the entire cohort died on the same day as ED admission or up to 10 days later. One hundred fifty-nine (94.6%) of 168 patients died on the same day as ED admission. Twenty-five (15.7%) of 159 died within the first 3 min after ED arrival. One hundred thirty-four (84.2%) of 159 died within 4 min to 5 h after ED admission. Figure 7 illustrates the pattern of death for trauma patients without ISL. The median of time to death for patients without ISL was 10 (0 – 13,582) min. Our step-by-step exclusions were necessary to refine the data and make the information ready for analysis. However, we experimented further to evaluate how these necessary exclusions might have influenced the findings. The mortality rate for excluded patients was 229 (12.4%) of 1,834. We compared it with the mortality rate for the included patients, which was 121 (6.2%) of 1,941. The associated P value was significant (P < 0.001), which could suggest the possibility of some selection bias as we acknowledged earlier. We also compared the ISS between the included and excluded patients. There was no statistical difference (P = 0.068) between the median ISS for included (10 [1 – 75]) versus excluded patients (9 [1 – 75]). Accordingly, it is reasonable to infer that the two populations (excluded and included) were similar in terms of their severity of injury.
We do not routinely obtain arterial blood gas results from every trauma patient, so an accurate (not calculated) base deficit is absent in the vast majority of our sample patients. Therefore, we did not include base deficit as part of our analysis. In addition, we did not include lactate clearance because serial lactate measurements were only available in some patients. We do not collect coagulation profile values within the database, and so they were not available for analysis. This represents another potential limitation. Finally, our cohort is different from other reports in that a large number of our patients (43.8%) did require OI. This could be a result of the relatively large number of penetrating trauma patients we evaluated in our institution.
In summary, our results showed that ISL, measured within 35 min, is a statistically significant risk factor for mortality and operative intervention in a strictly controlled cohort. The ISL was not statistically significant for ICUA.
1. Kaplan LJ, Kellum JA: Comparison of acid-base models for prediction
of hospital mortality
29 (6): 662–666, 2008.
2. Guyette F, Suffoletto B, Castillo JL, Quintero J, Callaway C, Puyana JC: Prehospital serum lactate
as a predictor of outcomes in trauma
patients: a retrospective observational study. J Trauma
70 (4): 782–786, 2011.
3. Sammour T, Kahokehr A, Caldwell S, Hill AG: Venous glucose and arterial lactate
as biochemical predictors of mortality
in clinically severely injured trauma
patients—a comparison with ISS and TRISS. Injury
40 (1): 104–108, 2009.
4. Lavery RF, Livingston DH, Tortella BJ, Sambol JT, Slomovitz BM, Siegel JH: The utility of venous lactate
to triage injured patients in the trauma
center. J Am Coll Surg
190 (6): 656–664, 2000.
5. Pal JD, Victorino GP, Twomey P, Liu TH, Bullard MK, Harken AH: Admission serum lactate
levels do not predict mortality
in the acutely injured patient. J Trauma
60 (3): 583–587, 2006.
6. Aslar AK, Kuzu MA, Elhan AH, Tanik A, Hengirmen S: Admission lactate
level and the APACHE II score are the most useful predictors of prognosis following torso trauma
35 (8): 746–752, 2004.
7. Caputo N, Fraser R, Paliga A, Kanter M, Hosford K, Madlinger R: Triage vital signs do not correlate with serum lactate
or base deficit, and are less predictive of operative intervention in penetrating trauma
patients: a prospective cohort study. Emerg Med J
30 (7): 546–550, 2013.
8. Dunne JR, Tracy JK, Scalea TM, Napolitano LM: Lactate
and base deficit in trauma
: does alcohol or drug use impair their predictive accuracy? J Trauma
58 (5): 959–966, 2005.
9. Neville AL, Nemtsev D, Manasrah R, Bricker SD, Putnam BA: Mortality
risk stratification in elderly trauma
patients based on initial arterial lactate
and base deficit levels. Am Surg
77 (10): 1337–1341, 2011.
10. Callaway DW, Shapiro NI, Donnino MW, Baker C, Rosen CL: Serum lactate
and base deficit as predictors of mortality
in normotensive elderly blunt trauma
patients. J Trauma
66 (4): 1040–1044, 2009.
11. Cerović O, Golubović V, Spec-Marn A, Kremzar B, Vidmar G: Relationship between injury severity and lactate
levels in severely injured patients. Intensive Care Med
29 (8): 1300–1305, 2003.
12. Husain FA, Martin MJ, Mullenix PS, Steele SR, Elliott DC: Serum lactate
and base deficit as predictors of mortality
and morbidity. Am J Surg
185 (5): 485–491, 2003.
13. Herbert HK, Dechert TA, Wolfe L, Aboutanos MB, Malhotra AK, Ivatury RR, Duane TM: Lactate
: a poor predictor of mortality
in the setting of alcohol ingestion. Am Surg
77 (12): 1576–1579, 2011.
14. Vandromme MJ, Griffin RL, Weinberg JA, Rue LW 3rd, Kerby JD: Lactate
is a better predictor than systolic blood pressure for determining blood requirement and mortality
: could prehospital measures improve trauma
triage? J Am Coll Surg
210 (5): 861–869, 2010.
15. Siegel JH, Rivkind AI, Dalal S, Goodarzi S: Early physiologic predictors of injury severity and death in blunt multiple trauma
. Arch Surg
125 (4): 498–508, 1990.
16. Paladino L, Sinert R, Wallace D, Anderson T, Yadav K, Zehtabchi S: The utility of base deficit and arterial lactate
in differentiating major from minor injury in trauma
patients with normal vital signs. Resuscitation
77 (3): 363–368, 2008.
17. Sobhian B, Kröpfl A, Hölzenbein T, Khandem A, Redl H, Bahrami S: Increased circulating D-lactate
levels predict risk of mortality
after hemorrhage and surgical trauma
in baboons. Shock
37 (5): 473–477, 2012.
18. Wacharasint P, Nakada TA, Boyd JH, Russell JA, Walley KR: Normal-range blood lactate
concentration in septic shock is prognostic and predictive. Shock
38 (1): 4–10, 2012.
19. Régnier MA, Raux M, Le Manach Y, Asencio Y, Gaillard J, Devilliers C, Langeron O, Riou B: Prognostic significance of blood lactate
clearance in trauma
117 (6): 1276–1288, 2012.