- Email: [email protected]
A comparison of initial lactate and initial base deficit as predictors of mortality after severe blunt trauma
j o u r n a l o f s u r g i c a l r e s e a r c h o c t o b e r 2 0 1 6 ( 2 0 5 ) 4 4 6 e4 5 5
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.JournalofSurgicalResearch.com
A comparison of initial lactate and initial base deficit as predictors of mortality after severe blunt trauma Stephen C. Gale, MD,a,* Jurek F. Kocik, MD,a Robert Creath, MD,b Jessica S. Crystal, MD,c and Viktor Y. Dombrovskiy, MD, PhD, MPHc a
Department of Surgery, Trauma Services, East Texas Medical Center, Tyler, Texas Department of Emergency Medicine, East Texas Medical Center, Tyler, Texas c Department of Surgery, Rutgers/RWJMS, New Brunswick, New Jersey b
article info
abstract
Article history:
Background: After injury, base deficit (BD) and lactate are common measures of shock.
Received 8 April 2016
Lactate directly measures anaerobic byproducts, whereas BD is calculated and
Received in revised form
multifactorial. Although recent studies suggest superiority for lactate in predicting
5 June 2016
mortality, most were small or analyzed populations with heterogeneous injury
Accepted 28 June 2016
severity. Our objective was to compare initial BD with lactate as predictors of inhospital
Available online 5 July 2016
mortality in a large cohort of blunt trauma patients all presenting with hemorrhagic shock.
Keywords:
Materials and methods: The Glue Grant multicenter prospective cohort database was
Lactate
queried; demographic, injury, and physiologic parameters were compiled. Survivors, early
Base deficit
deaths (24 h), and late deaths were compared. Profound shock (lactate 4 mmol/L) and
Blunt trauma
severe traumatic brain injury subgroups were identified a priori. Chi-square, t-test, and
Metabolic acidosis
analysis of variance were used as appropriate for analysis. Multivariable logistic regression
Shock
and area under the receiver operating characteristic curve analysis assessed survival
Resuscitation
predictors. P < 0.05 was significant. Results: A total of 1829 patients met inclusion; 289 (15.8%) died. Both BD and lactate were higher for nonsurvivors (P < 0.00001). After multivariable regression, both lactate (odds ratio [OR] 1.17; 95% confidence interval [CI]: 1.12-1.23; P < 0.00001) and BD (OR 1.04; 95% CI: 1.01-1.07; P < 0.005) predicted overall mortality. However, when excluding early deaths (n ¼ 77), only lactate (OR 1.12 95% CI: 1.06-1.19; P < 0.0001) remained predictive but not BD (OR 1.00 95% CI: 0.97-1.04; P ¼ 0.89). For the shock subgroup, (n ¼ 915), results were similar with lactate, but not BD, predicting both early and late deaths. Findings also appear independent of traumatic brain injury severity. Conclusions: After severe blunt trauma, initial lactate better predicts inhospital mortality than initial BD. Initial BD does not predict mortality for patients who survive >24 h. ª 2016 Elsevier Inc. All rights reserved.
* Corresponding author. Elmer G Ellis Level 1 Trauma Center, East Texas Medical Center, 1020 E. Idel St, Tyler, TX 75701. Tel.: 903-9693573; fax: 903-535-9217. E-mail address: [email protected] (S.C. Gale). 0022-4804/$ e see front matter ª 2016 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jss.2016.06.103
447
gale et al initial lactate predicts mortality
Introduction Both base deficit (BD) and serum lactate are common post-injury measures of metabolic acidosis.1e4 Serum lactate directly measures a byproduct of the anaerobic metabolism induced during periods of shock.5 In contrast, BD is a calculated value and is a more global assessment of acidebase status,6 being influenced by not only lactic acidosis but also other affecters such as minute ventilation and various therapies (i.e., sodium bicarbonate, intravenous fluids, and blood products7 and others). Although results of both tests are known to reflect a patient’s degree of shock and hypoperfusion, there is a continuing debate as to which initial measurement is more useful clinically.8e10 Specifically, it is unknown whether initial BD values or initial serum lactate levels, obtained during early resuscitation, better predict overall mortality for patients with severe injurydand whether any predictions apply to early or late mortality. Although recent studies confirm the utility of lactate clearance in predicting outcomes after injury,3,4,11 surprisingly few directly explore the predictive value of initial lactate compared with initial BD; furthermore, existing studies are often small5,12 or include very heterogeneous populations.6,13 In the present study, we used a large well-established database of very severely injured blunt trauma patients, all of whom presented in hemorrhagic shock, to investigate the predictive value of these two commonly used tests for both early (<24 h) and later
mortality in patients with and without severe metabolic derangement. We hypothesized that compared with initial BD, initial serum lactate is a better overall predictor of inhospital mortality after severe blunt trauma.
Materials and methods The patient population was selected from the Glue Grant Trauma-Related Database (TRDB): a large-scale multicenter collaborative research program entitled Inflammation and the Host Response to Injury conducted from 2002 to 2011 and supported by the National Institute of General Medical Sciences (NIGMS). This database is available online.14 The TRDB contains clinical, laboratory, and outcomes data for more than 2000 severely injured blunt trauma patients from eight major urban level 1 trauma centers. Inclusion criteria for the Glue Grant study, from which the TRDB derives, were: 1. Blunt trauma mechanism 2. Abbreviated Injury Scale (AIS) severity score >2 outside the head region 3. Emergency department (ED) arrival <6 h from time of injury 4. Systolic blood pressure (SBP) < 90 mm Hg or BD > 6 mEq/L (prehospital or within 60 min of arrival) 5. Blood transfusion within 12 h of injury 6. Intact cervical spinal cord
Table 1 e Study population characteristics.
Demographics Age, y (mean SD) Gender (males, %)
Total
Survivors
Nonsurvivors
1829
1540
289
42.8 18.7
41.4 17.8
50.4 21.6
P value <0.00001
66.6
66.5
67.1
0.83
White
77.3
77.3
76.8
0.31
Black
6.2
6.4
5.2
12.0
12.1
11.4
4.6
4.2
6.6
Motor vehicle crash
53.4
53.3
54.0
Pedestrian struck
16.1
15.3
20.4
Motor cycle crash
Race (%)
Hispanic Other Blunt mechanism (%)
0.23
15.0
15.5
12.8
Fall
8.6
9.0
6.6
Other
4.4
4.5
4.2
Assault
2.4
2.5
2.1
Initial parameters (mean SD) Head AIS > 3
22.9%
21.3%
31.5%
<0.0002
8.4 5.6
8.8 5.6
6.4 5.1
<0.00001
Lowest ED SBP (mm Hg)
84.0 23.9
86.0 22.9
73.7 26.2
<0.00001
ISS
38.9 14.0
37.7 13.5
45.2 15.1
<0.00001
Initial GCS
APACHE II
29.1 7.2
27.9 6.7
35.4 6.2
<0.00001
Initial BD (mEq/L)
8.81 4.80
8.45 4.38
10.75 6.10
<0.00001
Initial lactate (mmol/L)
4.57 2.86
4.20 2.48
6.54 3.78
<0.00001
SD ¼ standard deviation.
448
j o u r n a l o f s u r g i c a l r e s e a r c h o c t o b e r 2 0 1 6 ( 2 0 5 ) 4 4 6 e4 5 5
Fig. 1 e Initial BD and lactate levels by mortality class. In the study population, for both initial BD and initial lactate values, a significant progression was noted: early deaths > late deaths > survivors; (**P < 0.00001).
After receiving Category 4 Exempt status from the East Texas Medical Center Institutional Review Board and access approval from the Inflammation and the Host Response to Injury Investigators Large-Scale Collaborative Research Program, the TRDB was downloaded onto Excel spreadsheets (Microsoft, Redmond, WA) and queried for all patients whose study records contained both data points “Initial_Base_Deficit” and “ER_lactate” (ER ¼ emergency room) for comparison. Survival status, length of stay, demographics, injury severity scores (ISS), ED SBP, ED Glasgow Coma Scale (GCS), APACHE II scores, initial serum lactate levels, and initial BD values were compiled. The primary outcome was inhospital survival. The secondary outcome for nonsurvivors was early (24 h) or late (>24 h) mortality. Patients who survived to discharge were compared with nonsurvivors; nonsurvivors were further analyzed by early (24 h) or late (>24 h) death. Additional subgroup analysis was undertaken a priori for two subgroups: 1. Patients presenting with “profound shock”das indicated by “ER_lactate” 4.0 mmol/Ldto compare survival prediction for initial BD and lactate in the most physiologically deranged group of severely injured patients. 2. Patients with and without severe traumatic brain injury (TBI)das indicated with “Head AIS > 3”dto determine if severity of TBI influences the outcome prediction for initial BD and initial lactate. Statistical analysis was performed using SAS 9.4 (SAS Institute, Cary, NC) and Wizard for Mac (Evan Miller, Chicago, IL). For continuous data, presented as mean standard deviation (SD), comparisons were made using the Student’s t-test or analysis of variance where appropriate; chi-square analysis was used for categorical data. A multivariable logistic regression model (including age, gender, race, ISS, and APACHE II), computing odds ratio (OR) and 95% confidence interval (CI), was used to determine contributors to mortality.
P < 0.05 was considered significant. Receiver operating characteristic (ROC) curves were also created to compare the area under the curve (AUC) for initial lactate and initial BD as predictors of mortality.
Results The TRDB query returned 1829 patients with both initial BD and serum lactate values for analysis. Of these, 289 (15.8%) died during their hospitalization. Table 1 summarizes the characteristics of the study population. Nonsurvivors were
Table 2 e Risk factors for inhospital mortality. n ¼ 1829
All patients Adjusted OR
95% CI
P value
Age
1.02
1.01-1.03
<0.0001
Gender
0.95
0.70-1.30
0.75
ISS
1.03
1.02-1.04
<0.00001
APACHE II
1.19
1.16-1.23
<0.00001
Initial lactate
1.17
1.12-1.23
<0.00001
Initial BD
1.04
1.01-1.07
<0.005
Patients with >24 h survival Adjusted OR
n ¼ 1752 95% CI
P value <0.00001
Age
1.02
1.01-1.03
Gender
0.92
0.65-1.29
ISS
1.04
1.02-1.05
<0.00001
0.62
APACHE II
1.17
1.13-1.20
<0.00001
Initial lactate
1.12
1.06-1.18
<0.00005
Initial BD
1.00
0.97-1.04
0.90
gale et al initial lactate predicts mortality
449
Fig. 2 e Inhospital mortality prediction for initial BD and initial lactate (AUC). (A) Initial BD versus total mortality (0.6135); (B) Initial lactate versus total mortality (0.7071); (C) Initial BD versus mortality after 24 h (0.5749); (D) Initial lactate versus mortality after 24 h (0.6726).
significantly older (50.4 21.6 versus 41.4 17.8, P < 0.00001) than survivors and were more likely to have a head AIS > 3 (P < 0.0002). Other demographic parameters and mechanism of injury distribution were similar between groups. All patients were severely injured with very high mean ISSs (38.9 14.0) and APACHE II (29.1 7.2) scores. As expected, all clinical parameters were significantly worse for nonsurvivors than for survivors. (ISS, APACHE II, initial BD, initial lactate, lowest ED SBP; initial GCS: P < 0.00001 for all). Of the 289 nonsurvivors, 77 (26.6%) died in the first day after injury. For the 212 patients who died later, average survival was 10.6 16.9 d. In Figure 1, the initial BD and initial lactate are compared between early deaths (24 h), later deaths (>24 h), and survivors. Significant differences are observed, between survivors and late deaths and between late and early deaths, for both BD and lactate levels.
Multivariable logistic regression analysis of inhospital mortality in the study population is summarized in Table 2 as ORs with 95% CI for inhospital mortality. After controlling for demographics (age, gender, race), ISS, and APACHE II scores, both initial BD and initial serum lactate significantly predict inhospital mortality: for each 1 meq/L increase in BD, mortality risk increased by 4%; for each 1 mmol/L increase in lactate, mortality risk increased by 17%. However, when the 77 patients who died within the first 24 h after injury were excluded, only initial serum lactate (12% increase in mortality per 1 mmol/L) was significantly associated with mortality. Initial BD did not predict mortality for patients who survived greater than 24 h. The ROC curves for both initial BD and initial serum lactate, as predictors of overall mortality, are presented in Figure 2. For the study population, the AUC for initial BD is 0.6135 (Fig. 2A)
450
j o u r n a l o f s u r g i c a l r e s e a r c h o c t o b e r 2 0 1 6 ( 2 0 5 ) 4 4 6 e4 5 5
Table 3 e Comparison of patients with and without shock (lactate ‡ 4).
Demographics Age, y (mean SD) Gender (males, %)
<4 mmol/L
4 mmol/L
914
915
43.9 18.9
41.7 18.5
63.9
69.3
White
81.3
73.2
Black
3.8
8.5
11.3
12.7
3.6
5.6
Motor Vehicle crash
53.8
53.0
Pedestrian struck
15.2
17.0
Motor cycle crash
Other
<0.02 <0.02 <0.00002
Race (%)
Hispanic
P value
Blunt mechanism (%)
0.64
14.9
15.2
Fall
9.1
8.1
Other
4.9
3.9
Assault
2.1
2.7
Initial physiology Head AIS > 3
19.1%
27.2%
<0.0005
Initial GCS
9.5 5.60
7.3 5.4
<0.00001
Lowest ED SBP (mm Hg)
87.9 22.0
80.1 25.0
<0.00001
ISS
36.8 13.4
41.0 14.3
<0.00001
APACHE II
26.7 6.9
31.4 6.7
<0.00001
Initial BD (mEq/L)
6.88 3.36
10.74 5.16
<0.00001
Initial lactate (mmol/L)
2.55 0.85
6.58 2.73
<0.00001
Inhospital
74 (8.1%)
215 (23.5%)
<0.00001
24 h
14 (18.9%)
63 (29.3%)
Mortality
0.08
SD ¼ standard deviation.
and for initial lactate is 0.7071 (Fig. 2B). When ROC curves are generated for those patients surviving >24 h, AUC for initial BD drops to 0.5749 (Fig. 2C), whereas the AUC for initial serum lactate is 0.6726 (Fig. 2D). These data and calculations demonstrate that although initial BD is elevated, both in patients who died early and in those who died later, this measurement is not a predictor of mortality for patients who survive past the first 24 h. In contrast, initial lactate levels remain a predictor of total inhospital mortality, independent of time frame, after severe blunt traumatic injury.
Shock subgroup (lactate 4 mmol/L) In subgroup analysis, we identified 915 (50.0%) patients who presented with initial serum lactate 4 mmol/L and were considered to have “profound shock.” Shock patients were significantly younger (P < 0.02), with a greater proportion of males (P < 0.02) and non-Caucasians (P < 0.00001). For shock patients, all physiologic parameters (GCS, SBP, ISS, APACHE II, initial BD, and initial lactate) were significantly worse than for patients in the “nonshock” subgroup (P < 0.00001 for all).
Mortality was also much higher for patients in the shock group than those without shock (23.5% versus 8.1%; P < 0.00001). Population characteristics and comparisons to the remaining “nonshock” population are summarized in Table 3. In Figure 3, for “shock” subgroup, the initial BD and initial lactate are compared between early deaths (24 h), later deaths (>24 h), and survivors. For initial BD, although significant differences are noted between early and late deaths (P < 0.0001), in this population, there is no difference between BD levels between late deaths and survivors (P ¼ 0.16). In contrast, for initial lactate, significant differences persist between early deaths, late deaths, and survivors (P < 0.00001 for all). In Table 4, multivariable logistic regression analysis of the subgroup is summarized as ORs with 95% CI for inhospital mortality. After controlling for demographics (age, gender, race), ISS, and APACHE II scores, both initial BD and initial lactate again predict inhospital mortality with initial lactate being a superior predictor: for each 1 meq/L increase in BD, mortality risk increased by 4%; for each 1 mmol/L increase in lactate, mortality risk increased by 15%. Similar to the larger study population, when those patients who died within the first 24 h after injury were excluded (n ¼ 63), only initial serum lactate (7% increase in mortality per 1 mmol/L) was significantly associated with mortalitydbut not initial BD. The ROC curves for initial BD and initial lactate for the profound “shock” subgroup are presented in Figure 4. For initial BD, the AUC is 0.5975 for all deaths and 0.5403 for mortality after 24 h; these values indicate overall poor prediction for mortality at any time point for BD in this subgroup. In contrast, for initial lactate, the AUC is 0.6591 for all deaths and 0.6100 for survival after 24 h. Although not as strong a predictor as APACHE II (AUC ¼ 0.7827), initial lactate does indeed predict both early and late mortality in the profound shock subgroupdand to greater degree than initial BD.
TBI subgroup analysis Additional subgroup analysis demonstrated that 419 patients (22.9%) presented with severe TBI. There were no differences in demographics between those with and without severe TBI; however, a significantly different mechanism of injury distribution was observed (P < 0.02). For severe TBI patients, most physiologic parameters (GCS, ISS, APACHE II, initial BD, and initial lactate) were significantly worse than those for patients with head AIS 3; lowest ED SBP was not different between groups. As expected, overall mortality was significantly higher for severe TBI patients (21.7% versus 14.0%; P < 0.0002); however, these patients were much less likely to die within 24 h of admission than patients without severe TBI (16.5% versus 31.3%; P < 0.01). Population characteristics and comparisons of patients with and without severe TBI are summarized in Table 5. In Table 6, multivariable logistic regression analysis of the TBI subgroups is summarized as ORs with 95% CI for inhospital mortality. For severe TBI patients, only initial lactate, but not initial BD (P ¼ 0.55), predicted inhospital mortality. Interestingly, when those who died within the first
gale et al initial lactate predicts mortality
451
Fig. 3 e Initial BD and lactate levelsdshock subgroup. In the shock subgroup, initial BD was higher for those who died early; however, there was no difference in initial BD between late deaths and survivors (P [ 0.15). For initial lactate values, a significant progression was again noted: early deaths > late deaths > survivors; (**P < 0.0001).
24 h after injury were excluded (n ¼ 15), neither initial lactate nor initial BD predicted inhospital mortality. For patients without severe TBI (n ¼ 1348), results of multivariable logistic regression mirrored those for the study population: both initial lactate and initial BD predict mortality when early deaths are included. However, when only late (>24 h) deaths are considered, initial lactate alone predicts mortality. This subgroup analysis demonstrates that outcomes related to severe TBI do not significantly influence the findings
Table 4 e Risk factors for inhospital mortalitydshock group. All patientsdlactate 4 mmol/L
n ¼ 915
Adjusted OR
95% CI
P value
Age
1.01
1.00-1.02
<0.01
Gender
0.96
0.67-1.42
0.88
ISS
1.02
1.01-1.04
<0.0002
APACHE II
1.19
1.15-1.24
<0.00001
Initial lactate
1.15
1.08-1.22
<0.00001
Initial BD
1.04
1.01-1.08
<0.03
Patients with >24 h survival Adjusted OR Age
1.01
n ¼ 852 95% CI
P value
1.00-1.03
<0.01
Gender
0.95
0.63-1.48
ISS
1.04
1.02-1.05
<0.00001
0.87
APACHE II
1.19
1.14-1.23
<0.00001
Initial lactate
1.08
1.00-1.16
<0.04
Initial BD
0.98
0.94-1.03
0.45
seen in the population as a whole: that initial serum lactate is a superior predictor of mortality compared with initial BD in severely injured blunt trauma patients.
Discussion In the present study, we demonstrate that initial lactate is a better predictor of inhospital mortality than initial BD for patients after severe blunt traumatic injury. Although BD levels were indeed higher in patients who died, those associations were weaker for inhospital mortality than for initial lactate and disappeared entirely for patients who survived past the first 24 h after injury. The utility of lactate as a predictor of mortality, in contrast, was shown to persist independent of the duration of survival or the degree of shock on presentation. The limited value of initial BD in predicting mortality after severe blunt trauma was further demonstrated with ROC curve analysis: for each different comparison, AUC was smaller for initial BD than for initial lactate. Importantly, for the sickest “shock” cohort, the AUC for initial BD was <0.6000 at all time points demonstrating it to be little better than a “coin flip” for predicting mortality in severely injured blunt trauma patients presenting with profound shock. Finally, these results are not significantly influenced by the TBI severity within the population. Our findings support the growing body of literature examining the utility of serum lactate measurements,3e5,11,13,15e18 and the superiority of lactate over BD,5,6,12 during resuscitation after major injury. Our work confirms that in the most severely injured patients, including those with profound shock at presentation, initial lactate is a reliable indicator of those who are at the greatest risk of inhospital death. Various approaches to serum lactate interpretation have been studied including initial lactate,4,5,13,15 lactate clearance,3,4,11,15,16 and
452
j o u r n a l o f s u r g i c a l r e s e a r c h o c t o b e r 2 0 1 6 ( 2 0 5 ) 4 4 6 e4 5 5
Fig. 4 e Inhospital mortality prediction for initial BD and initial lactatedshock subgroup (AUC). (A) Initial BD versus total mortality (0.5975); (B) Initial lactate versus total mortality (0.6591); (C) Initial BD versus mortality after 24 h (0.5410); (D) Initial lactate versus mortality after 24 h (0.6103).
lactate normalization time.17 Each of these parameters has utility in guiding resuscitation and in predicting outcomes after injury. In particular, initial lactate is important as a very early sign of metabolic derangement.4,15,18 Although serial lactate measurements are optimal as part of a goal-directed resuscitation,4 a significantly elevated initial lactate alerts surgeons to patients who require greater scrutiny, guides them to a more aggressive resuscitation strategy, and perhaps earlier surgical intervention. Our demonstration that initial BD poorly predicts outcome, especially in those who survive past the first day, is not surprising. In 1996, Mikulaschek et al.12 reported on a small series of injured patients and found no difference in BD between survivors and nonsurvivors and also that lactate and BD did not correlate. Martin et al.,6 demonstrated in 2006, in a mixed surgical intensive care unit cohort that included more
than 1000 injured patients, that the “discordance” between lactate and BD was significant: lactate levels predicted mortality “regardless of the associated BD level” and that BD is very limited as predictor of outcome. Husain et al.5 had similar findings in a smaller mixed surgical intensive care unit population and concluded that “base deficit should be used with caution as a marker of shock or resuscitation from shock.” The present study is the first large, multicenter comparison of post-injury BD and lactate. Furthermore, it is the first to make this comparison exclusively in a severely injured cohort. When combined with the extant body of literature, our findings conclusively demonstrate the superiority of initial lactate over BD in predicting inhospital mortality after blunt trauma. For the clinician, our data demonstrate that initial lactate, rather than initial BD, should be used to guide the
453
gale et al initial lactate predicts mortality
Table 5 e Traumatic brain injury subgroup analysis. Head AIS > 3 Demographics Age, y (mean SD) Gender (males, %)
Head AIS 3
Head AIS > 3
Adjusted OR
95% CI
All patients
P-value n ¼ 419
419
1410 42.7 18.5
0.59
Age
1.01
0.99-1.02
66.1
66.7
0.81
Gender
0.91
0.51-1.61
0.59
0.29
ISS
1.05
1.02-1.08
<0.0002
APACHE II
1.26
1.18-1.34
<0.00001
Initial lactate
1.11
1.02-1.22
<0.03
Initial BD
0.98
0.93-1.04
0.55
White
79.7
76.5
Black
4.3
6.7
11.7
12.1
4.3
4.7
Other
Table 6 e Risk factors for inhospital mortalitydby head AIS.
43.2 19.5
Race (%)
Hispanic
P value
>24-h survival
<0.02
Blunt mechanism (%)
0.25
n ¼ 409
Motor vehicle crash
55.4
52.8
Age
1.01
0.99-1.02
0.18
Pedestrian struck
19.6
15.1
Gender
0.82
0.45-1.48
0.51
Motor cycle crash
13.6
15.5
ISS
1.05
1.02-1.08
<0.001
Fall
7.9
8.8
APACHE II
1.24
1.16-1.33
<0.00001
Other
2.4
5.0
Initial lactate
1.08
0.98-1.19
0.12
Assault
1.2
2.8
Initial BD
0.96
0.90-1.03
0.25
5.6 4.4
9.3 5.6
83.3 25.7
84.2 23.3
Initial parameters (mean SD) Initial GCS Lowest ED SBP (mm Hg)
Head AIS 3 <0.00001 0.50
n ¼ 1410
All patients
<0.0001
Age
1.02
1.01-1.03
Gender
0.98
0.68-1.42
0.91
ISS
1.02
1.01-1.04
<0.0005
APACHE II
1.18
1.15-1.22
<0.00001
ISS
49.8 11.3
35.7 13.1
<0.00001
APACHE II
31.2 5.7
28.4 7.5
<0.00001
Initial base deficit (mEq/L)
9.23 4.72
8.69 4.77
<0.05
Initial lactate (mmol/L)
4.92 2.96
4.46 2.82
<0.005
Inhospital
91 (21.7%)
198 (14.0%)
<0.0002
Gender
0.99
0.64-1.51
24 h
15 (16.5%)
62 (31.3%)
<0.01
ISS
1.03
1.02-1.05
<0.00001
APACHE II
1.15
1.11-1.19
<0.00001
Initial lactate
1.14
1.06-1.22
<0.0002
Initial BD
1.02
0.97-1.06
0.44
Mortality
Initial lactate
1.20
1.13-1.27
<0.00001
Initial BD
1.07
1.03-1.10
<0.0005
>24-h survival Age
SD ¼ standard deviation.
aggressiveness of early resuscitation and the degree of vigilance a critically injured patient might require. We believe that because BD is influenced by so many other factors, it does not, as reliably, predict post-injury shock. Furthermore, we show that elevated initial lactate levels have both qualitative and quantitative values whereby the degree of elevation is important. Although in the present study, the AUC for initial lactate in the various subgroups is not considered high (<0.75 for all), for our study population, it was the highest among all data available to the clinician during early resuscitation (i.e., age, GCS, BD, head AIS, initial SBP, and so forth). Before APACHE scores can be calculated, and even before all injuries are identified, initial lactate identifies the severity of shock and predicts outcome. The best interpretation of these moderate correlations is that a very high lactate predicts, but does assure a poor outcomedfurther emphasizing the potential clinical impact of early aggressive intervention. Of course, to maximize the clinical utility of serum lactate, these initial values should not be interpreted in a vacuum. Although a single value should influence the aggressiveness of resuscitation or the speed of intervention, the success of
1.03
n ¼ 1348 1.01-1.04
<0.00001 0.96
resuscitation is best judged, along with other end points, by “serial lactate” testing to follow lactate clearance and to assure timely lactate normalization. Unfortunately, the current body of literature, examining serial lactate measurements for trauma or sepsis,19,20 presents a mixture of clearance percentages and normalization time frames, which are confusing to the surgical critical care community. Therefore, future studies, using this data set and others, should focus on identifying the specific target end points, which best predict outcomes: that is. which clearance percentage and time to normalization are associated with the least morbidity and mortality? Our data may also have more broad implications for laboratory utilization and costs of care. In the past, routine lactate measurement was limited by availability and the need for arterial samples. Currently, however, nearly all hospitals can perform this test routinely and quickly (often with point-of-care technology), and venous lactate is known to
454
j o u r n a l o f s u r g i c a l r e s e a r c h o c t o b e r 2 0 1 6 ( 2 0 5 ) 4 4 6 e4 5 5
correlate well with arterial values.21 If lactate is superior in identifying shock and predicting outcomes, clinicians must ask: “what does base deficit add to patient care?” Certainly lactate, both initially and with serial testing, is a better marker of metabolic acidosis. To extend the argument further: what is diagnostic yield of routine arterial blood gas (ABG) analysis, in general, during trauma resuscitation? With the widespread ability to monitor both minute-to-minute oxygenation (pulse oximetry) and ventilation (end tidal CO2) noninvasively, clinicians might conclude that routine ABG measurements add little or nothing to post-injury care that is not either duplicated or supplanted by newer technologies. As effort toward cost-containment increases, with bundled payments and other payer-driven utilization initiatives, our data argue against routine ABG testing during trauma resuscitation and management in favor of newer technologies.
by the National Institute of General Medical Sciences (U54GM062119). Also, they are grateful for the influence and mentorship of Ms. Susette Coyle, RN, and Drs Stephen Lowry (deceased), Siobhan Corbett, and Steven Calvanodall from the Robert Wood Johnson Medical School, a Glue Grant “participating institution.” They were instrumental in understanding the potential that this data set holds for clinical investigation and in helping formulate the questions investigated in this and other studies. Finally, they thank Mr Elmer G. Ellis for institutional support at East Texas Medical Center in Tyler, Texas. Authors’ contributions: S.C.G., J.S.C., and J.K.F. contributed to study conception/design. S.C.G. contributed to data acquisition. S.C.G. and V.Y.D. contributed to data analysis/ interpretation. S.C.G., R.C., and J.S.C. contributed to drafting of the article. All the authors contributed to critical revision of the article.
Limitations As with all database analyses, for the present study, the greatest limitation is its retrospective nature. Fortunately, the Glue Grant project involved a highly standardized protocol for patient selection and significant quality assurance mechanisms for reporting and inclusion in the data set. The database is relatively homogenous in terms of its trauma population, being comprised only of severely injured blunt trauma victims and excluding patients with isolated TBI. Because the Glue Grant study was a multicenter trial, it also does not suffer from other limitations that may reduce the applicability of single-center studies. In that the Glue Grant study included only blunt trauma victims, there may be a limitation in applying our findings after penetrating trauma. That all Glue Grant patients exhibited either hypotension or required transfusion as part of the study’s inclusion criteria, does broaden the applicability, however, to patients with the hemorrhagic shock more common after penetrating trauma.
Conclusions After severe blunt trauma, initial serum lactate is superior to initial BD in predicting inhospital survival in patients with and without shock. Initial BD does not predict mortality for patients whose survival is longer than 24 h. Lactate measurements should replace calculated BD for guiding early post-injury resuscitation. Furthermore, highly elevated initial lactate levels should prompt greater scrutiny, more aggressive treatment, and a higher index of suspicion. Future studies to identify optimal clearance end points are needed to clarify the timing of serial lactate measurements during resuscitation. Routine ABG testing to guide resuscitation and management in critically ill trauma patients should also be scrutinized.
Acknowledgment The authors would like to acknowledge the contribution of Inflammation and the Host Response to Injury investigators in the Large-Scale Collaborative Research Program, supported
Disclosure The authors have no conflicts of interest to disclose.
references
1. Davis JW, Kaups KL, Parks SN. Base deficit is superior to pH in evaluating clearance of acidosis after traumatic shock. J Trauma. 1998;44:114e118. 2. Kincaid EH, Miller PR, Meredith JW, Rahman N, Chang MC. Elevated arterial base deficit in trauma patients: a marker of impaired oxygen utilization. J Am Coll Surg. 1998;187:384e392. 3. Abramson D, Scalea TM, Hitchcock R, Trooskin SZ, Henry SM, Greenspan J. Lactate clearance and survival following injury. J Trauma. 1993;35:584e588. discussion 588e589. 4. Odom SR, Howell MD, Silva GS, et al. Lactate clearance as a predictor of mortality in trauma patients. J Trauma Acute Care Surg. 2013;74:999e1004. 5. Husain FA, Martin MJ, Mullenix PS, Steele SR, Elliott DC. Serum lactate and base deficit as predictors of mortality and morbidity. Am J Surg. 2003;185:485e491. 6. Martin MJ, FitzSullivan E, Salim A, Brown CV, Demetriades D, Long W. Discordance between lactate and base deficit in the surgical intensive care unit: which one do you trust? Am J Surg. 2006;191:625e630. 7. Li K, Xu Y. Citrate metabolism in blood transfusions and its relationship due to metabolic alkalosis and respiratory acidosis. Int J Clin Exp Med. 2015;8:6578e6584. 8. Ouellet JF, Roberts DJ, Tiruta C, et al. Admission base deficit and lactate levels in Canadian patients with blunt trauma: are they useful markers of mortality? J Trauma Acute Care Surg. 2012;72:1532e1535. 9. 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. 2006;60:583e587. discussion 587e589. 10. 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. 2011;77:1337e1341. 11. Regnier MA, Raux M, Le Manach Y, et al. Prognostic significance of blood lactate and lactate clearance in trauma patients. Anesthesiology. 2012;117:1276e1288.
gale et al initial lactate predicts mortality
12. Mikulaschek A, Henry SM, Donovan R, Scalea TM. Serum lactate is not predicted by anion gap or base excess after trauma resuscitation. J Trauma. 1996;40:218e222. discussion 222e214. 13. 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. 2009;66:1040e1044. 14. Trauma Related Database (TRDB). Inflammation and Host Response to Injury; 2011. http://www.gluegrant.org. 15. Dezman ZD, Comer AC, Smith GS, Narayan M, Scalea TM, Hirshon JM. Failure to clear elevated lactate predicts 24-hour mortality in trauma patients. J Trauma Acute Care Surg. 2015;79:580e585. 16. McNelis J, Marini CP, Jurkiewicz A, et al. Prolonged lactate clearance is associated with increased mortality in the surgical intensive care unit. Am J Surg. 2001;182:481e485.
455
17. Galkova K, Vrabelova M. Normalization of blood lactate as early end-point of polytrauma treatment. Bratislava Med J. 2013;114:637e641. 18. 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. 2008;77:363e368. 19. Marty P, Roquilly A, Vallee F, et al. Lactate clearance for death prediction in severe sepsis or septic shock patients during the first 24 hours in intensive care unit: an observational study. Ann Intensive Care. 2013;3:3. 20. Casserly B, Phillips GS, Schorr C, et al. Lactate measurements in sepsis-induced tissue hypoperfusion: results from the Surviving Sepsis Campaign database. Crit Care Med. 2015;43:567e573. 21. 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. 2000;190:656e664.
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.JournalofSurgicalResearch.com
A comparison of initial lactate and initial base deficit as predictors of mortality after severe blunt trauma Stephen C. Gale, MD,a,* Jurek F. Kocik, MD,a Robert Creath, MD,b Jessica S. Crystal, MD,c and Viktor Y. Dombrovskiy, MD, PhD, MPHc a
Department of Surgery, Trauma Services, East Texas Medical Center, Tyler, Texas Department of Emergency Medicine, East Texas Medical Center, Tyler, Texas c Department of Surgery, Rutgers/RWJMS, New Brunswick, New Jersey b
article info
abstract
Article history:
Background: After injury, base deficit (BD) and lactate are common measures of shock.
Received 8 April 2016
Lactate directly measures anaerobic byproducts, whereas BD is calculated and
Received in revised form
multifactorial. Although recent studies suggest superiority for lactate in predicting
5 June 2016
mortality, most were small or analyzed populations with heterogeneous injury
Accepted 28 June 2016
severity. Our objective was to compare initial BD with lactate as predictors of inhospital
Available online 5 July 2016
mortality in a large cohort of blunt trauma patients all presenting with hemorrhagic shock.
Keywords:
Materials and methods: The Glue Grant multicenter prospective cohort database was
Lactate
queried; demographic, injury, and physiologic parameters were compiled. Survivors, early
Base deficit
deaths (24 h), and late deaths were compared. Profound shock (lactate 4 mmol/L) and
Blunt trauma
severe traumatic brain injury subgroups were identified a priori. Chi-square, t-test, and
Metabolic acidosis
analysis of variance were used as appropriate for analysis. Multivariable logistic regression
Shock
and area under the receiver operating characteristic curve analysis assessed survival
Resuscitation
predictors. P < 0.05 was significant. Results: A total of 1829 patients met inclusion; 289 (15.8%) died. Both BD and lactate were higher for nonsurvivors (P < 0.00001). After multivariable regression, both lactate (odds ratio [OR] 1.17; 95% confidence interval [CI]: 1.12-1.23; P < 0.00001) and BD (OR 1.04; 95% CI: 1.01-1.07; P < 0.005) predicted overall mortality. However, when excluding early deaths (n ¼ 77), only lactate (OR 1.12 95% CI: 1.06-1.19; P < 0.0001) remained predictive but not BD (OR 1.00 95% CI: 0.97-1.04; P ¼ 0.89). For the shock subgroup, (n ¼ 915), results were similar with lactate, but not BD, predicting both early and late deaths. Findings also appear independent of traumatic brain injury severity. Conclusions: After severe blunt trauma, initial lactate better predicts inhospital mortality than initial BD. Initial BD does not predict mortality for patients who survive >24 h. ª 2016 Elsevier Inc. All rights reserved.
* Corresponding author. Elmer G Ellis Level 1 Trauma Center, East Texas Medical Center, 1020 E. Idel St, Tyler, TX 75701. Tel.: 903-9693573; fax: 903-535-9217. E-mail address: [email protected] (S.C. Gale). 0022-4804/$ e see front matter ª 2016 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jss.2016.06.103
447
gale et al initial lactate predicts mortality
Introduction Both base deficit (BD) and serum lactate are common post-injury measures of metabolic acidosis.1e4 Serum lactate directly measures a byproduct of the anaerobic metabolism induced during periods of shock.5 In contrast, BD is a calculated value and is a more global assessment of acidebase status,6 being influenced by not only lactic acidosis but also other affecters such as minute ventilation and various therapies (i.e., sodium bicarbonate, intravenous fluids, and blood products7 and others). Although results of both tests are known to reflect a patient’s degree of shock and hypoperfusion, there is a continuing debate as to which initial measurement is more useful clinically.8e10 Specifically, it is unknown whether initial BD values or initial serum lactate levels, obtained during early resuscitation, better predict overall mortality for patients with severe injurydand whether any predictions apply to early or late mortality. Although recent studies confirm the utility of lactate clearance in predicting outcomes after injury,3,4,11 surprisingly few directly explore the predictive value of initial lactate compared with initial BD; furthermore, existing studies are often small5,12 or include very heterogeneous populations.6,13 In the present study, we used a large well-established database of very severely injured blunt trauma patients, all of whom presented in hemorrhagic shock, to investigate the predictive value of these two commonly used tests for both early (<24 h) and later
mortality in patients with and without severe metabolic derangement. We hypothesized that compared with initial BD, initial serum lactate is a better overall predictor of inhospital mortality after severe blunt trauma.
Materials and methods The patient population was selected from the Glue Grant Trauma-Related Database (TRDB): a large-scale multicenter collaborative research program entitled Inflammation and the Host Response to Injury conducted from 2002 to 2011 and supported by the National Institute of General Medical Sciences (NIGMS). This database is available online.14 The TRDB contains clinical, laboratory, and outcomes data for more than 2000 severely injured blunt trauma patients from eight major urban level 1 trauma centers. Inclusion criteria for the Glue Grant study, from which the TRDB derives, were: 1. Blunt trauma mechanism 2. Abbreviated Injury Scale (AIS) severity score >2 outside the head region 3. Emergency department (ED) arrival <6 h from time of injury 4. Systolic blood pressure (SBP) < 90 mm Hg or BD > 6 mEq/L (prehospital or within 60 min of arrival) 5. Blood transfusion within 12 h of injury 6. Intact cervical spinal cord
Table 1 e Study population characteristics.
Demographics Age, y (mean SD) Gender (males, %)
Total
Survivors
Nonsurvivors
1829
1540
289
42.8 18.7
41.4 17.8
50.4 21.6
P value <0.00001
66.6
66.5
67.1
0.83
White
77.3
77.3
76.8
0.31
Black
6.2
6.4
5.2
12.0
12.1
11.4
4.6
4.2
6.6
Motor vehicle crash
53.4
53.3
54.0
Pedestrian struck
16.1
15.3
20.4
Motor cycle crash
Race (%)
Hispanic Other Blunt mechanism (%)
0.23
15.0
15.5
12.8
Fall
8.6
9.0
6.6
Other
4.4
4.5
4.2
Assault
2.4
2.5
2.1
Initial parameters (mean SD) Head AIS > 3
22.9%
21.3%
31.5%
<0.0002
8.4 5.6
8.8 5.6
6.4 5.1
<0.00001
Lowest ED SBP (mm Hg)
84.0 23.9
86.0 22.9
73.7 26.2
<0.00001
ISS
38.9 14.0
37.7 13.5
45.2 15.1
<0.00001
Initial GCS
APACHE II
29.1 7.2
27.9 6.7
35.4 6.2
<0.00001
Initial BD (mEq/L)
8.81 4.80
8.45 4.38
10.75 6.10
<0.00001
Initial lactate (mmol/L)
4.57 2.86
4.20 2.48
6.54 3.78
<0.00001
SD ¼ standard deviation.
448
j o u r n a l o f s u r g i c a l r e s e a r c h o c t o b e r 2 0 1 6 ( 2 0 5 ) 4 4 6 e4 5 5
Fig. 1 e Initial BD and lactate levels by mortality class. In the study population, for both initial BD and initial lactate values, a significant progression was noted: early deaths > late deaths > survivors; (**P < 0.00001).
After receiving Category 4 Exempt status from the East Texas Medical Center Institutional Review Board and access approval from the Inflammation and the Host Response to Injury Investigators Large-Scale Collaborative Research Program, the TRDB was downloaded onto Excel spreadsheets (Microsoft, Redmond, WA) and queried for all patients whose study records contained both data points “Initial_Base_Deficit” and “ER_lactate” (ER ¼ emergency room) for comparison. Survival status, length of stay, demographics, injury severity scores (ISS), ED SBP, ED Glasgow Coma Scale (GCS), APACHE II scores, initial serum lactate levels, and initial BD values were compiled. The primary outcome was inhospital survival. The secondary outcome for nonsurvivors was early (24 h) or late (>24 h) mortality. Patients who survived to discharge were compared with nonsurvivors; nonsurvivors were further analyzed by early (24 h) or late (>24 h) death. Additional subgroup analysis was undertaken a priori for two subgroups: 1. Patients presenting with “profound shock”das indicated by “ER_lactate” 4.0 mmol/Ldto compare survival prediction for initial BD and lactate in the most physiologically deranged group of severely injured patients. 2. Patients with and without severe traumatic brain injury (TBI)das indicated with “Head AIS > 3”dto determine if severity of TBI influences the outcome prediction for initial BD and initial lactate. Statistical analysis was performed using SAS 9.4 (SAS Institute, Cary, NC) and Wizard for Mac (Evan Miller, Chicago, IL). For continuous data, presented as mean standard deviation (SD), comparisons were made using the Student’s t-test or analysis of variance where appropriate; chi-square analysis was used for categorical data. A multivariable logistic regression model (including age, gender, race, ISS, and APACHE II), computing odds ratio (OR) and 95% confidence interval (CI), was used to determine contributors to mortality.
P < 0.05 was considered significant. Receiver operating characteristic (ROC) curves were also created to compare the area under the curve (AUC) for initial lactate and initial BD as predictors of mortality.
Results The TRDB query returned 1829 patients with both initial BD and serum lactate values for analysis. Of these, 289 (15.8%) died during their hospitalization. Table 1 summarizes the characteristics of the study population. Nonsurvivors were
Table 2 e Risk factors for inhospital mortality. n ¼ 1829
All patients Adjusted OR
95% CI
P value
Age
1.02
1.01-1.03
<0.0001
Gender
0.95
0.70-1.30
0.75
ISS
1.03
1.02-1.04
<0.00001
APACHE II
1.19
1.16-1.23
<0.00001
Initial lactate
1.17
1.12-1.23
<0.00001
Initial BD
1.04
1.01-1.07
<0.005
Patients with >24 h survival Adjusted OR
n ¼ 1752 95% CI
P value <0.00001
Age
1.02
1.01-1.03
Gender
0.92
0.65-1.29
ISS
1.04
1.02-1.05
<0.00001
0.62
APACHE II
1.17
1.13-1.20
<0.00001
Initial lactate
1.12
1.06-1.18
<0.00005
Initial BD
1.00
0.97-1.04
0.90
gale et al initial lactate predicts mortality
449
Fig. 2 e Inhospital mortality prediction for initial BD and initial lactate (AUC). (A) Initial BD versus total mortality (0.6135); (B) Initial lactate versus total mortality (0.7071); (C) Initial BD versus mortality after 24 h (0.5749); (D) Initial lactate versus mortality after 24 h (0.6726).
significantly older (50.4 21.6 versus 41.4 17.8, P < 0.00001) than survivors and were more likely to have a head AIS > 3 (P < 0.0002). Other demographic parameters and mechanism of injury distribution were similar between groups. All patients were severely injured with very high mean ISSs (38.9 14.0) and APACHE II (29.1 7.2) scores. As expected, all clinical parameters were significantly worse for nonsurvivors than for survivors. (ISS, APACHE II, initial BD, initial lactate, lowest ED SBP; initial GCS: P < 0.00001 for all). Of the 289 nonsurvivors, 77 (26.6%) died in the first day after injury. For the 212 patients who died later, average survival was 10.6 16.9 d. In Figure 1, the initial BD and initial lactate are compared between early deaths (24 h), later deaths (>24 h), and survivors. Significant differences are observed, between survivors and late deaths and between late and early deaths, for both BD and lactate levels.
Multivariable logistic regression analysis of inhospital mortality in the study population is summarized in Table 2 as ORs with 95% CI for inhospital mortality. After controlling for demographics (age, gender, race), ISS, and APACHE II scores, both initial BD and initial serum lactate significantly predict inhospital mortality: for each 1 meq/L increase in BD, mortality risk increased by 4%; for each 1 mmol/L increase in lactate, mortality risk increased by 17%. However, when the 77 patients who died within the first 24 h after injury were excluded, only initial serum lactate (12% increase in mortality per 1 mmol/L) was significantly associated with mortality. Initial BD did not predict mortality for patients who survived greater than 24 h. The ROC curves for both initial BD and initial serum lactate, as predictors of overall mortality, are presented in Figure 2. For the study population, the AUC for initial BD is 0.6135 (Fig. 2A)
450
j o u r n a l o f s u r g i c a l r e s e a r c h o c t o b e r 2 0 1 6 ( 2 0 5 ) 4 4 6 e4 5 5
Table 3 e Comparison of patients with and without shock (lactate ‡ 4).
Demographics Age, y (mean SD) Gender (males, %)
<4 mmol/L
4 mmol/L
914
915
43.9 18.9
41.7 18.5
63.9
69.3
White
81.3
73.2
Black
3.8
8.5
11.3
12.7
3.6
5.6
Motor Vehicle crash
53.8
53.0
Pedestrian struck
15.2
17.0
Motor cycle crash
Other
<0.02 <0.02 <0.00002
Race (%)
Hispanic
P value
Blunt mechanism (%)
0.64
14.9
15.2
Fall
9.1
8.1
Other
4.9
3.9
Assault
2.1
2.7
Initial physiology Head AIS > 3
19.1%
27.2%
<0.0005
Initial GCS
9.5 5.60
7.3 5.4
<0.00001
Lowest ED SBP (mm Hg)
87.9 22.0
80.1 25.0
<0.00001
ISS
36.8 13.4
41.0 14.3
<0.00001
APACHE II
26.7 6.9
31.4 6.7
<0.00001
Initial BD (mEq/L)
6.88 3.36
10.74 5.16
<0.00001
Initial lactate (mmol/L)
2.55 0.85
6.58 2.73
<0.00001
Inhospital
74 (8.1%)
215 (23.5%)
<0.00001
24 h
14 (18.9%)
63 (29.3%)
Mortality
0.08
SD ¼ standard deviation.
and for initial lactate is 0.7071 (Fig. 2B). When ROC curves are generated for those patients surviving >24 h, AUC for initial BD drops to 0.5749 (Fig. 2C), whereas the AUC for initial serum lactate is 0.6726 (Fig. 2D). These data and calculations demonstrate that although initial BD is elevated, both in patients who died early and in those who died later, this measurement is not a predictor of mortality for patients who survive past the first 24 h. In contrast, initial lactate levels remain a predictor of total inhospital mortality, independent of time frame, after severe blunt traumatic injury.
Shock subgroup (lactate 4 mmol/L) In subgroup analysis, we identified 915 (50.0%) patients who presented with initial serum lactate 4 mmol/L and were considered to have “profound shock.” Shock patients were significantly younger (P < 0.02), with a greater proportion of males (P < 0.02) and non-Caucasians (P < 0.00001). For shock patients, all physiologic parameters (GCS, SBP, ISS, APACHE II, initial BD, and initial lactate) were significantly worse than for patients in the “nonshock” subgroup (P < 0.00001 for all).
Mortality was also much higher for patients in the shock group than those without shock (23.5% versus 8.1%; P < 0.00001). Population characteristics and comparisons to the remaining “nonshock” population are summarized in Table 3. In Figure 3, for “shock” subgroup, the initial BD and initial lactate are compared between early deaths (24 h), later deaths (>24 h), and survivors. For initial BD, although significant differences are noted between early and late deaths (P < 0.0001), in this population, there is no difference between BD levels between late deaths and survivors (P ¼ 0.16). In contrast, for initial lactate, significant differences persist between early deaths, late deaths, and survivors (P < 0.00001 for all). In Table 4, multivariable logistic regression analysis of the subgroup is summarized as ORs with 95% CI for inhospital mortality. After controlling for demographics (age, gender, race), ISS, and APACHE II scores, both initial BD and initial lactate again predict inhospital mortality with initial lactate being a superior predictor: for each 1 meq/L increase in BD, mortality risk increased by 4%; for each 1 mmol/L increase in lactate, mortality risk increased by 15%. Similar to the larger study population, when those patients who died within the first 24 h after injury were excluded (n ¼ 63), only initial serum lactate (7% increase in mortality per 1 mmol/L) was significantly associated with mortalitydbut not initial BD. The ROC curves for initial BD and initial lactate for the profound “shock” subgroup are presented in Figure 4. For initial BD, the AUC is 0.5975 for all deaths and 0.5403 for mortality after 24 h; these values indicate overall poor prediction for mortality at any time point for BD in this subgroup. In contrast, for initial lactate, the AUC is 0.6591 for all deaths and 0.6100 for survival after 24 h. Although not as strong a predictor as APACHE II (AUC ¼ 0.7827), initial lactate does indeed predict both early and late mortality in the profound shock subgroupdand to greater degree than initial BD.
TBI subgroup analysis Additional subgroup analysis demonstrated that 419 patients (22.9%) presented with severe TBI. There were no differences in demographics between those with and without severe TBI; however, a significantly different mechanism of injury distribution was observed (P < 0.02). For severe TBI patients, most physiologic parameters (GCS, ISS, APACHE II, initial BD, and initial lactate) were significantly worse than those for patients with head AIS 3; lowest ED SBP was not different between groups. As expected, overall mortality was significantly higher for severe TBI patients (21.7% versus 14.0%; P < 0.0002); however, these patients were much less likely to die within 24 h of admission than patients without severe TBI (16.5% versus 31.3%; P < 0.01). Population characteristics and comparisons of patients with and without severe TBI are summarized in Table 5. In Table 6, multivariable logistic regression analysis of the TBI subgroups is summarized as ORs with 95% CI for inhospital mortality. For severe TBI patients, only initial lactate, but not initial BD (P ¼ 0.55), predicted inhospital mortality. Interestingly, when those who died within the first
gale et al initial lactate predicts mortality
451
Fig. 3 e Initial BD and lactate levelsdshock subgroup. In the shock subgroup, initial BD was higher for those who died early; however, there was no difference in initial BD between late deaths and survivors (P [ 0.15). For initial lactate values, a significant progression was again noted: early deaths > late deaths > survivors; (**P < 0.0001).
24 h after injury were excluded (n ¼ 15), neither initial lactate nor initial BD predicted inhospital mortality. For patients without severe TBI (n ¼ 1348), results of multivariable logistic regression mirrored those for the study population: both initial lactate and initial BD predict mortality when early deaths are included. However, when only late (>24 h) deaths are considered, initial lactate alone predicts mortality. This subgroup analysis demonstrates that outcomes related to severe TBI do not significantly influence the findings
Table 4 e Risk factors for inhospital mortalitydshock group. All patientsdlactate 4 mmol/L
n ¼ 915
Adjusted OR
95% CI
P value
Age
1.01
1.00-1.02
<0.01
Gender
0.96
0.67-1.42
0.88
ISS
1.02
1.01-1.04
<0.0002
APACHE II
1.19
1.15-1.24
<0.00001
Initial lactate
1.15
1.08-1.22
<0.00001
Initial BD
1.04
1.01-1.08
<0.03
Patients with >24 h survival Adjusted OR Age
1.01
n ¼ 852 95% CI
P value
1.00-1.03
<0.01
Gender
0.95
0.63-1.48
ISS
1.04
1.02-1.05
<0.00001
0.87
APACHE II
1.19
1.14-1.23
<0.00001
Initial lactate
1.08
1.00-1.16
<0.04
Initial BD
0.98
0.94-1.03
0.45
seen in the population as a whole: that initial serum lactate is a superior predictor of mortality compared with initial BD in severely injured blunt trauma patients.
Discussion In the present study, we demonstrate that initial lactate is a better predictor of inhospital mortality than initial BD for patients after severe blunt traumatic injury. Although BD levels were indeed higher in patients who died, those associations were weaker for inhospital mortality than for initial lactate and disappeared entirely for patients who survived past the first 24 h after injury. The utility of lactate as a predictor of mortality, in contrast, was shown to persist independent of the duration of survival or the degree of shock on presentation. The limited value of initial BD in predicting mortality after severe blunt trauma was further demonstrated with ROC curve analysis: for each different comparison, AUC was smaller for initial BD than for initial lactate. Importantly, for the sickest “shock” cohort, the AUC for initial BD was <0.6000 at all time points demonstrating it to be little better than a “coin flip” for predicting mortality in severely injured blunt trauma patients presenting with profound shock. Finally, these results are not significantly influenced by the TBI severity within the population. Our findings support the growing body of literature examining the utility of serum lactate measurements,3e5,11,13,15e18 and the superiority of lactate over BD,5,6,12 during resuscitation after major injury. Our work confirms that in the most severely injured patients, including those with profound shock at presentation, initial lactate is a reliable indicator of those who are at the greatest risk of inhospital death. Various approaches to serum lactate interpretation have been studied including initial lactate,4,5,13,15 lactate clearance,3,4,11,15,16 and
452
j o u r n a l o f s u r g i c a l r e s e a r c h o c t o b e r 2 0 1 6 ( 2 0 5 ) 4 4 6 e4 5 5
Fig. 4 e Inhospital mortality prediction for initial BD and initial lactatedshock subgroup (AUC). (A) Initial BD versus total mortality (0.5975); (B) Initial lactate versus total mortality (0.6591); (C) Initial BD versus mortality after 24 h (0.5410); (D) Initial lactate versus mortality after 24 h (0.6103).
lactate normalization time.17 Each of these parameters has utility in guiding resuscitation and in predicting outcomes after injury. In particular, initial lactate is important as a very early sign of metabolic derangement.4,15,18 Although serial lactate measurements are optimal as part of a goal-directed resuscitation,4 a significantly elevated initial lactate alerts surgeons to patients who require greater scrutiny, guides them to a more aggressive resuscitation strategy, and perhaps earlier surgical intervention. Our demonstration that initial BD poorly predicts outcome, especially in those who survive past the first day, is not surprising. In 1996, Mikulaschek et al.12 reported on a small series of injured patients and found no difference in BD between survivors and nonsurvivors and also that lactate and BD did not correlate. Martin et al.,6 demonstrated in 2006, in a mixed surgical intensive care unit cohort that included more
than 1000 injured patients, that the “discordance” between lactate and BD was significant: lactate levels predicted mortality “regardless of the associated BD level” and that BD is very limited as predictor of outcome. Husain et al.5 had similar findings in a smaller mixed surgical intensive care unit population and concluded that “base deficit should be used with caution as a marker of shock or resuscitation from shock.” The present study is the first large, multicenter comparison of post-injury BD and lactate. Furthermore, it is the first to make this comparison exclusively in a severely injured cohort. When combined with the extant body of literature, our findings conclusively demonstrate the superiority of initial lactate over BD in predicting inhospital mortality after blunt trauma. For the clinician, our data demonstrate that initial lactate, rather than initial BD, should be used to guide the
453
gale et al initial lactate predicts mortality
Table 5 e Traumatic brain injury subgroup analysis. Head AIS > 3 Demographics Age, y (mean SD) Gender (males, %)
Head AIS 3
Head AIS > 3
Adjusted OR
95% CI
All patients
P-value n ¼ 419
419
1410 42.7 18.5
0.59
Age
1.01
0.99-1.02
66.1
66.7
0.81
Gender
0.91
0.51-1.61
0.59
0.29
ISS
1.05
1.02-1.08
<0.0002
APACHE II
1.26
1.18-1.34
<0.00001
Initial lactate
1.11
1.02-1.22
<0.03
Initial BD
0.98
0.93-1.04
0.55
White
79.7
76.5
Black
4.3
6.7
11.7
12.1
4.3
4.7
Other
Table 6 e Risk factors for inhospital mortalitydby head AIS.
43.2 19.5
Race (%)
Hispanic
P value
>24-h survival
<0.02
Blunt mechanism (%)
0.25
n ¼ 409
Motor vehicle crash
55.4
52.8
Age
1.01
0.99-1.02
0.18
Pedestrian struck
19.6
15.1
Gender
0.82
0.45-1.48
0.51
Motor cycle crash
13.6
15.5
ISS
1.05
1.02-1.08
<0.001
Fall
7.9
8.8
APACHE II
1.24
1.16-1.33
<0.00001
Other
2.4
5.0
Initial lactate
1.08
0.98-1.19
0.12
Assault
1.2
2.8
Initial BD
0.96
0.90-1.03
0.25
5.6 4.4
9.3 5.6
83.3 25.7
84.2 23.3
Initial parameters (mean SD) Initial GCS Lowest ED SBP (mm Hg)
Head AIS 3 <0.00001 0.50
n ¼ 1410
All patients
<0.0001
Age
1.02
1.01-1.03
Gender
0.98
0.68-1.42
0.91
ISS
1.02
1.01-1.04
<0.0005
APACHE II
1.18
1.15-1.22
<0.00001
ISS
49.8 11.3
35.7 13.1
<0.00001
APACHE II
31.2 5.7
28.4 7.5
<0.00001
Initial base deficit (mEq/L)
9.23 4.72
8.69 4.77
<0.05
Initial lactate (mmol/L)
4.92 2.96
4.46 2.82
<0.005
Inhospital
91 (21.7%)
198 (14.0%)
<0.0002
Gender
0.99
0.64-1.51
24 h
15 (16.5%)
62 (31.3%)
<0.01
ISS
1.03
1.02-1.05
<0.00001
APACHE II
1.15
1.11-1.19
<0.00001
Initial lactate
1.14
1.06-1.22
<0.0002
Initial BD
1.02
0.97-1.06
0.44
Mortality
Initial lactate
1.20
1.13-1.27
<0.00001
Initial BD
1.07
1.03-1.10
<0.0005
>24-h survival Age
SD ¼ standard deviation.
aggressiveness of early resuscitation and the degree of vigilance a critically injured patient might require. We believe that because BD is influenced by so many other factors, it does not, as reliably, predict post-injury shock. Furthermore, we show that elevated initial lactate levels have both qualitative and quantitative values whereby the degree of elevation is important. Although in the present study, the AUC for initial lactate in the various subgroups is not considered high (<0.75 for all), for our study population, it was the highest among all data available to the clinician during early resuscitation (i.e., age, GCS, BD, head AIS, initial SBP, and so forth). Before APACHE scores can be calculated, and even before all injuries are identified, initial lactate identifies the severity of shock and predicts outcome. The best interpretation of these moderate correlations is that a very high lactate predicts, but does assure a poor outcomedfurther emphasizing the potential clinical impact of early aggressive intervention. Of course, to maximize the clinical utility of serum lactate, these initial values should not be interpreted in a vacuum. Although a single value should influence the aggressiveness of resuscitation or the speed of intervention, the success of
1.03
n ¼ 1348 1.01-1.04
<0.00001 0.96
resuscitation is best judged, along with other end points, by “serial lactate” testing to follow lactate clearance and to assure timely lactate normalization. Unfortunately, the current body of literature, examining serial lactate measurements for trauma or sepsis,19,20 presents a mixture of clearance percentages and normalization time frames, which are confusing to the surgical critical care community. Therefore, future studies, using this data set and others, should focus on identifying the specific target end points, which best predict outcomes: that is. which clearance percentage and time to normalization are associated with the least morbidity and mortality? Our data may also have more broad implications for laboratory utilization and costs of care. In the past, routine lactate measurement was limited by availability and the need for arterial samples. Currently, however, nearly all hospitals can perform this test routinely and quickly (often with point-of-care technology), and venous lactate is known to
454
j o u r n a l o f s u r g i c a l r e s e a r c h o c t o b e r 2 0 1 6 ( 2 0 5 ) 4 4 6 e4 5 5
correlate well with arterial values.21 If lactate is superior in identifying shock and predicting outcomes, clinicians must ask: “what does base deficit add to patient care?” Certainly lactate, both initially and with serial testing, is a better marker of metabolic acidosis. To extend the argument further: what is diagnostic yield of routine arterial blood gas (ABG) analysis, in general, during trauma resuscitation? With the widespread ability to monitor both minute-to-minute oxygenation (pulse oximetry) and ventilation (end tidal CO2) noninvasively, clinicians might conclude that routine ABG measurements add little or nothing to post-injury care that is not either duplicated or supplanted by newer technologies. As effort toward cost-containment increases, with bundled payments and other payer-driven utilization initiatives, our data argue against routine ABG testing during trauma resuscitation and management in favor of newer technologies.
by the National Institute of General Medical Sciences (U54GM062119). Also, they are grateful for the influence and mentorship of Ms. Susette Coyle, RN, and Drs Stephen Lowry (deceased), Siobhan Corbett, and Steven Calvanodall from the Robert Wood Johnson Medical School, a Glue Grant “participating institution.” They were instrumental in understanding the potential that this data set holds for clinical investigation and in helping formulate the questions investigated in this and other studies. Finally, they thank Mr Elmer G. Ellis for institutional support at East Texas Medical Center in Tyler, Texas. Authors’ contributions: S.C.G., J.S.C., and J.K.F. contributed to study conception/design. S.C.G. contributed to data acquisition. S.C.G. and V.Y.D. contributed to data analysis/ interpretation. S.C.G., R.C., and J.S.C. contributed to drafting of the article. All the authors contributed to critical revision of the article.
Limitations As with all database analyses, for the present study, the greatest limitation is its retrospective nature. Fortunately, the Glue Grant project involved a highly standardized protocol for patient selection and significant quality assurance mechanisms for reporting and inclusion in the data set. The database is relatively homogenous in terms of its trauma population, being comprised only of severely injured blunt trauma victims and excluding patients with isolated TBI. Because the Glue Grant study was a multicenter trial, it also does not suffer from other limitations that may reduce the applicability of single-center studies. In that the Glue Grant study included only blunt trauma victims, there may be a limitation in applying our findings after penetrating trauma. That all Glue Grant patients exhibited either hypotension or required transfusion as part of the study’s inclusion criteria, does broaden the applicability, however, to patients with the hemorrhagic shock more common after penetrating trauma.
Conclusions After severe blunt trauma, initial serum lactate is superior to initial BD in predicting inhospital survival in patients with and without shock. Initial BD does not predict mortality for patients whose survival is longer than 24 h. Lactate measurements should replace calculated BD for guiding early post-injury resuscitation. Furthermore, highly elevated initial lactate levels should prompt greater scrutiny, more aggressive treatment, and a higher index of suspicion. Future studies to identify optimal clearance end points are needed to clarify the timing of serial lactate measurements during resuscitation. Routine ABG testing to guide resuscitation and management in critically ill trauma patients should also be scrutinized.
Acknowledgment The authors would like to acknowledge the contribution of Inflammation and the Host Response to Injury investigators in the Large-Scale Collaborative Research Program, supported
Disclosure The authors have no conflicts of interest to disclose.
references
1. Davis JW, Kaups KL, Parks SN. Base deficit is superior to pH in evaluating clearance of acidosis after traumatic shock. J Trauma. 1998;44:114e118. 2. Kincaid EH, Miller PR, Meredith JW, Rahman N, Chang MC. Elevated arterial base deficit in trauma patients: a marker of impaired oxygen utilization. J Am Coll Surg. 1998;187:384e392. 3. Abramson D, Scalea TM, Hitchcock R, Trooskin SZ, Henry SM, Greenspan J. Lactate clearance and survival following injury. J Trauma. 1993;35:584e588. discussion 588e589. 4. Odom SR, Howell MD, Silva GS, et al. Lactate clearance as a predictor of mortality in trauma patients. J Trauma Acute Care Surg. 2013;74:999e1004. 5. Husain FA, Martin MJ, Mullenix PS, Steele SR, Elliott DC. Serum lactate and base deficit as predictors of mortality and morbidity. Am J Surg. 2003;185:485e491. 6. Martin MJ, FitzSullivan E, Salim A, Brown CV, Demetriades D, Long W. Discordance between lactate and base deficit in the surgical intensive care unit: which one do you trust? Am J Surg. 2006;191:625e630. 7. Li K, Xu Y. Citrate metabolism in blood transfusions and its relationship due to metabolic alkalosis and respiratory acidosis. Int J Clin Exp Med. 2015;8:6578e6584. 8. Ouellet JF, Roberts DJ, Tiruta C, et al. Admission base deficit and lactate levels in Canadian patients with blunt trauma: are they useful markers of mortality? J Trauma Acute Care Surg. 2012;72:1532e1535. 9. 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. 2006;60:583e587. discussion 587e589. 10. 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. 2011;77:1337e1341. 11. Regnier MA, Raux M, Le Manach Y, et al. Prognostic significance of blood lactate and lactate clearance in trauma patients. Anesthesiology. 2012;117:1276e1288.
gale et al initial lactate predicts mortality
12. Mikulaschek A, Henry SM, Donovan R, Scalea TM. Serum lactate is not predicted by anion gap or base excess after trauma resuscitation. J Trauma. 1996;40:218e222. discussion 222e214. 13. 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. 2009;66:1040e1044. 14. Trauma Related Database (TRDB). Inflammation and Host Response to Injury; 2011. http://www.gluegrant.org. 15. Dezman ZD, Comer AC, Smith GS, Narayan M, Scalea TM, Hirshon JM. Failure to clear elevated lactate predicts 24-hour mortality in trauma patients. J Trauma Acute Care Surg. 2015;79:580e585. 16. McNelis J, Marini CP, Jurkiewicz A, et al. Prolonged lactate clearance is associated with increased mortality in the surgical intensive care unit. Am J Surg. 2001;182:481e485.
455
17. Galkova K, Vrabelova M. Normalization of blood lactate as early end-point of polytrauma treatment. Bratislava Med J. 2013;114:637e641. 18. 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. 2008;77:363e368. 19. Marty P, Roquilly A, Vallee F, et al. Lactate clearance for death prediction in severe sepsis or septic shock patients during the first 24 hours in intensive care unit: an observational study. Ann Intensive Care. 2013;3:3. 20. Casserly B, Phillips GS, Schorr C, et al. Lactate measurements in sepsis-induced tissue hypoperfusion: results from the Surviving Sepsis Campaign database. Crit Care Med. 2015;43:567e573. 21. 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. 2000;190:656e664.