Article, Endocrinology

Blood glucose levels as an adjunct for prehospital field triage

a b s t r a c t

Objective: Elevated blood glucose levels (BGL) are known to be part of the physiologic response to stress following physical trauma. We aimed to study whether a measured BGL might help improve accuracy of field triage.

Methods: We conducted a retrospective study using the Israel Defense Forces Trauma Registry. BGLs were determined upon hospital arrival and were not available to medical providers in the field.

Results: There were 706 casualties in the registry who had a recorded BGL upon hospital arrival. Sixty percent (18/30) of casualties who had a BGL >=200 mg/dL had been triaged in the field as severely wounded, whereas 11% (71/651) of casualties who had a BGL b 200 mg/dL had been triaged as severely wounded. For predicting an Injury Severity Score N 15, the positive likelihood ratio using field triage of severe was 11, using BGL >=200 mg/dL was 8, and using a combination of the two tests was 26. For predicting the need for intensive care unit admission, the ratios were 8, 13, and 23, respectively.

Conclusions: Elevated BGL improved prediction of high Injury Severity Score and ICU use among casualties triaged as severe. If future research using BGL measured in the field yields similar results, combining BGL with standard field triage may allow for more accurate identification of casualties who need acute field intervention, have major injury, or require ICU admission.

(C) 2013

Introduction

Elevated glucose levels are known to be a physiological response to stress following physical trauma [1,2]. During the last decades, there has been increased interest in studying the correlation between blood glucose level (BGL) and prognosis of admitted Trauma victims. Less well studied, however, is the utility of BGL in the prehospital trauma setting.

The purpose of our study was to assess the utility of BGL as an adjunctive field triage tool, where we hypothesized that an elevated BGL would help identify the sickest patients over and beyond the field triage level. Secondarily, we hypothesized that elevated BGL would be associated with an increased need to perform prehospital Life-saving interventions (LSI). Because BGL was not available prior to field assessment and intervention and was only measured upon hospital arrival, we see our study as hypothesis generating.

? Financial support: None.

?? Declaration of interest: The authors report no conflicts of interest. The authors

alone are responsible for the content and writing of the paper.

? Not presented previously.

* Corresponding author. Tel.: +972 3 737 9284; fax: +972 3 737 7323.

E-mail address: [email protected] (E. Glassberg).

Methods

The research was performed at the Research Section of the Trauma & Combat Medicine Branch of the Israel Defense Forces (IDF) Surgeon General’s Headquarters. Data were drawn from the IDF Trauma Registry (ITR), maintained by the Research Section. The ITR has been active since 1997 and records data regarding trauma victims (the large majority is IDF personnel, but civilian casualties treated by IDF advanced life support teams are also included). The Trauma & Combat Medicine Branch receives data from military medical investigations, after-action reports conducted by the Medical Corps and airborne medical teams, hospitals, and forensic reports. Trained data abstrac- tors enter the data from the various sources (as available for any particular case) directly into a standardized, computerized abstrac- tion form. Periodic data reviews help ensure validity. The data include nature of injury, prehospital treatment, evacuation, treatment and diagnoses in-hospital, and additional details regarding the nature of the event. Only casualties who had BGLs recorded upon hospital arrival were included in the analyses.

This study was reviewed and approved by the Institutional Review Board (known locally as the Helsinki Committee) of the IDF Medical Corps.

Based on our literature review (see Discussion), BGL was dichotomized into high (>=200 mg/dL) and not-high levels (b 200 mg/dL). Injury severity, as determined by the senior provider in the

0735-6757/$ - see front matter (C) 2013 http://dx.doi.org/10.1016/j.ajem.2012.10.038

field, was defined as mild if no long-term injury was suspected, moderate if Long-term disability was anticipated, and severe if the injury was felt to be life-threatening. Note that in mid 2010, the IDF implemented a change to a binary injury severity system (urgent/not- urgent); subsequent casualties were not included in this study. Life- saving interventions were defined to include endotracheal intubation, cricothyroidotomy, Needle thoracostomy, tube thoracostomy, and tourniquet placement.

The data were imported into a commercially available statistical package (Stata 10, StataCorp LP, College Station, TX) for analysis. Apart from blood glucose levels which are reported using means and standard deviations, data are reported using medians and interquar- tile ranges (IQR). Means were compared using Student’s t test or the Kruskal-Wallis test, as appropriate. Unequal population variance was assumed for testing mean BGL versus survival. For categorical variables, Fisher’s exact test was used. Nonparametric equality-of- medians tests were used for comparing lengths of stay. P b .05 was considered statistically significant.

In addition to studying the associations between BGL and injury severity, we also analyzed the associations between BGL and demographic characteristics, injury mechanism, need for LSI, hospital length of stay, and survival.

Results

Between the years 1997 to 2010, 5553 records were entered into the ITR, of whom 706 (13%) were identified that had a recorded BGL upon hospital arrival. The characteristics of the study population are presented in the left side of Table 1. As expected in a trauma setting involving primarily soldiers, the vast majority of casualties are young (90% between 17 and 29 years old; N = 645 [i.e., 61 missing]) and

male (94%; N = 706).

The primary mechanisms of injury are shown in Table 1. The 292 penetrating cases included injuries from shrapnel (61% of 292; N = 178), firearms (21%; N = 61), and stabbings (13%; N = 39), with the

Table 1

Characteristics of the study population for all subjects and grouped by blood glucose level (b or >=200 mg/dL)

All subjects N = 706

BGL b200 mg/dL N = 668

(94.6%)

BGL >=200 mg/dL N = 38

(5.4%)

P

Age [y], median

20 (19-22)

20 (19-22)

21 (19-31)

.06

(IQR)

Male, N (%)

663 (94%)

626 (94%)

37 (97%)

.7

Mechanism

Penetrating, N (%)

292 (41%)

270 (41%)

22 (58%)

b.001*

Blunt, N (%)

257 (36%)

254 (39%)

3 (7.9%)

Burn / inhalation,

N (%)

56 (7.9%)

55 (8.4%)

1 (2.6%)

Crush, N (%)

26 (3.7%)

18 (2.8%)

8 (21%)

Other, N (%)

59 (8.4%)

55 (8.4%)

4 (11%)

Prehospital Severity

Severe, N (%)

89 (13%)

71 (11%)

18 (60%)

b.001*

Moderate, N (%)

208 (30%)

198 (30%)

10 (33%)

Mild, N (%)

384 (56%)

382 (59%)

2 (7%)

LSI, N (%)

114

90 (13%)

24 (63%)

b.001*

BGL [mg/dL], mean

120 (55)

110 (28)

287 (116)

N/A

(SD)

ISS, median (IQR)

4 (1-9)

4 (1-9)

12 (6-22)

b.001*

LOS [days], median

2 (1-6)

2 (1-5)

17 (3-36)

b.001*

(IQR)

Died, N (%)

6 (0.8%)

1 (0.1%)

5 (13%)

b.001*

P values computed comparing characteristics across blood glucose level groups. LOS, length of stay in hospital; *significant at 95% Confidence level.

Number of missing subjects by variable (BGL b200 mg/dL): age, 52; mechanism, 16; prehospital severity, 17; ISS, 22; LOS, 39.

Number of missing subjects by variable (BGL >=200 mg/dL): age, 9; prehospital severity, 8; ISS, 2; LOS, 11.

remaining penetrating mechanisms designated as “other penetrating” (4.8%; N = 14). The “other” mechanisms shown in Table 1 include environmental (envenomation, hyperthermia, hypothermia, drown- ing) (3.7% of 706; N = 26); blast (2.1%; N = 15); extremity

amputation (1.3%; N = 9); electrocution (0.6%; N = 4); and assorted

other mechanisms (0.7%; N = 5). In 50 of the 706 cases (7.1%), a secondary mechanism was listed (ie, there was a mixed mechanism). The distribution of prehospital triage categories includes 13% severe, 30% moderate, and 56% mild cases (N = 681), and life-saving interventions were performed in 114 (16%; N = 706) of the casualties.

Injury Severity Score (ISS), hospital length of stay, and survival are also presented in Table 1.

The average blood glucose level upon hospital arrival among all 706 subjects was 120 mg/dL, of whom 668 (95%) had a BGL b 200 mg/ dL, and 38 (5.4%) had a BGL >=200 mg/dL. The middle and right-hand side of Table 1 present the characteristics of this dichotomized population. The difference in ages between the two groups approached statistical significance, whereas gender did not. All other variables demonstrated a statistically significant difference between the 2 groups. (BGL is not reported in Table 1 as the two groups are defined by this variable.) The 4 subjects with a mechanism of “Other” and elevated BGL had all suffered traumatic amputations. The 2 subjects who were triaged as mild and who had elevated BGL were 31 and 42 years old.

The relationship between field triage category and average BGL, and between ISS and average BGL, are shown in the left and right sides of the Fig., respectively. Casualties who had BGLs >=200 mg/dL upon hospital arrival were much more likely to have been triaged in the field as severe as compared to casualties with BGLs b 200 mg/dL (Table 1). Similarly, they were much less likely to be triaged as mild, and were about equally likely to be triaged as moderate. Further, they underwent more LSIs, had higher ISSs, had longer lengths of hospital stay, and were more likely to die.

Among the casualties with elevated BGL, 18 (47%) required at least 1 day in the intensive care unit (ICU), whereas among those with normal BGL, 27 (4.0%) required at least 1 day in the ICU (P b .001). Examining only those casualties who spent at least 1 day in the ICU, the median ICU length of stay was 4 days (IQR 2-8 days) if the BGL was elevated and 2 days (IQR 2-6 days) if the BGL was not elevated (P = .1). Across all 237 (34%) subjects for whom data on transport times were available, the median time from injury to hospital arrival was 116 minutes (IQR 82-192 minutes). Of these, 31 (13%) were in the BGL

>=200 mg/dL group with a median transport time of 100 minutes (IQR 85-135 minutes), and 206 (87%) were in the BGL b 200 mg/dL group with a median of 120 minutes (IQR 81-220 minutes). The P value for the difference in medians is .02. Every 10-minute increase in transport time decreased the odds of having BGL >=200 mg/dL by 6% (95% CI - 1% to 2.6%).

We then evaluated how well field triage category (severe versus not severe) and BGL (>=200 mg/dL vs b 200 mg/dL) could predict major injury (defined as ISS N 15) and the need for intensive care (ie, any nonzero number of days in the ICU). We also evaluated the combination of these 2 tests in predicting either outcome. While ISS is approximately a gold standard (inasmuch as it has been shown to correlate with severity of injury [3] and is very unlikely influenced by the BGL), ICU admission is more complicated in that it is more likely influenced by the presence of hyperglycemia.

Major injury (ie, ISS N 15) was present in 8.7% of the dataset (59 subjects; N = 682). Table 2 (top) displays the 2 x 2 tables for ISS versus the three tests. For example, of 87 subjects triaged as severe, 44 (51%) had an ISS N 15. Of 17 subjects who were both field triaged as severe and who had a BGL >=200 mg/dL, 12 (71%) had an ISS N 15. These percentages define their respective positive predictive values. This test characteristic, as well as sensitivity, specificity, negative predictive value, and the positive and negative likelihood ratios, are shown in Table 2 (bottom) with their 95% confidence intervals (CI). It

Fig. Mean blood glucose level (mg/dL) by field triage category and ISS. The number of subjects in each category is shown in parentheses. The 95% upper limit of the SE of each mean is displayed as a whisker.

is apparent that, compared to a field triage category of severe, BGL

>=200 mg/dL has poorer sensitivity but somewhat higher specificity. The positive likelihood ratios of field triage of severe and BGL >=200 mg/dL are both reasonable, though field triage of severe is higher. Importantly, any given casualty has a prior probability (ie, before triage) of 8.7% of having major injury, based on the prevalence of major injury in the data set. If the casualty is then triaged as severe, the probability increases to 51%; if the casualty is triaged as severe and has an elevated BGL, the probability of major injury rises to 71%.

Results are remarkably similar for predicting the need for ICU admission (Table 3). Note that there is only partial overlap between need for ICU admission and major injury: only 34% (15/44) of the casualties with major injury required ICU admission, and 52% of those sent to the ICU (15/29) had major injury. Unlike in the ISS N 15 setting, here BGL >= 200 mg/dL has a higher positive likelihood ratio than field triage severe, though again, used together the tests have a considerably higher positive likelihood ratio.

Table 2

Predicting major injury (defined by injury severity score N 15) using field triage of severe, blood glucose level cut-off of 200 mg/dL, and their combination

We subsequently examined whether adding BGL to a field triage of moderate would help identify casualties with major injury or who would require ICU admission; the data are presented in Table 4. Subjects who were triaged as moderate had a 5% chance of having major injury and a 6% chance of requiring at least 1 day in the ICU. If they also had a BGL >=200 mg/dL, they had an 11% chance of having major injury (odds ratio = 2.3, 95% CI 0-16) and a 50% chance of requiring at least 1 day in the ICU (odds = 24, 95% CI 6-94). We note that the 5 casualties with a high BGL who went to the ICU were all young, healthy soldiers. Because the numbers of casualties in some of the table cells are fairly small, the confidence intervals are wide. A BGL b 200 mg/dL did not help decrease the likelihood of having major injury or of requiring an ICU bed among those triaged as moderate.

We then developed a logistic regression model for predicting major injury among all casualties, including independent variables age, field triage level, time from injury to hospital arrival, and BGL (b 200 or >=200 mg/dL). Age and time from injury to hospital arrival were generally not informative. Field triage was most informative, increasing the odds of major injury by an average of 15 for each increase in triage level. Within each field triage level, a BGL >=200 mg/ dL increased the odds of major injury by 2.6 times (95% CI 1-7).

We now present some additional findings. Among the subjects less

than 17 years old, between 17 and 29 years old, and older than 29

Field triage Blood glucose level

Combined

years old, the mean BGLs were 127 mg/dL, 115 mg/dL, and 135 mg/dL, respectively. The mean glucose was 120 mg/dL for males vs 111 mg/dL

Severe Mild/ moderate

>=200

mg/dL

b200

mg/dL

Positive Negative

for females. Among non-survivors, the mean BGL was 342 mg/dL versus 118 mg/dL among the surviving casualties. (The single death in the BGL b 200 mg/dL group had head and pelvic trauma from military rifle wounds, with exposed brain matter. His BGL upon hospital arrival was 168 mg/dL.)

The mean BGL was highest among victims during military operational activities and lowest among those injured while involved in military training exercises (135 mg/dL [N = 258] vs. 105 mg/dL [N = 73], respectively), while terror attack victims and routine military duty victims’ BGLs were between those values (120 mg/dL [N = 5] and 110 mg/dL [N = 302], respectively).

ISS N 15

44

12

15

44

12

44

ISS b15

43

559

21

602

5

597

Sensitivity

79 (66-88)

25 (15-38)

21 (12-34)

(%)

Specificity

93 (90-95)

97 (95-98)

99 (98-100)

(%)

PPV (%)

51 (40-62)

42 (26-59)

71 (44-90)

NPV (%)

98 (96-99)

93 (91-95)

93 (91-95)

LR+

11 (8-15)

8 (4-14)

26 (9-71)

LR-

0.2 (0.1-0.4)

0.8 (0.7-0.9)

0.8 (0.7-0.9)

The prevalence of major injury in the dataset was 8.7%. Test characteristics are given with their 95% confidence intervals.

PPV, positive predictive value; NPV, negative predictive value; LR+, positive likelihood ratio; LR-, negative likelihood ratio.

The Combined test is positive if field triage is severe and BGL >=200 mg/dL, and negative if either field triage is not severe or BGL b200 mg/dL or both.

BGL and ISS data are jointly available for more subjects (682) than are Field Triage and ISS data (658), and the Combined data are necessarily restricted to their minimum (658).

Among the 292 casualties who experienced penetrating mecha- nisms, the mean BGL was 144 mg/dL with shrapnel, 124 mg/dL with firearms, 107 mg/dL with stabbings, and 106 mg/dL with other penetrating mechanisms. Among the 257 casualties who experienced Blunt mechanisms, the mean BGL was 106 mg/dL. The highest mean BGLs were seen among those who suffered a Crush injury or an extremity amputation (177 mg/dL for both; N = 35). All other mechanisms had mean BGLs b 112 mg/dL.

Table 3

Predicting admission to intensive care unit using field triage of severe, blood glucose level cut-off of 200 mg/dL, and their combination

Field triage

Blood glucose level

Combined

Severe

Mild/moderate

>=200 mg/dL

b200 mg/dL

Positive

Negative

ICU Admission

30

13

18

27

11

32

ICU Non-admission

59

579

20

641

7

631

Sensitivity (%)

70 (54-83)

40 (26-56)

26 (14-41)

Specificity (%)

91 (88-93)

97 (95-98)

99 (98-100)

PPV (%)

34 (24-44)

47 (31-64)

61 (36-83)

NPV (%)

98 (96-99)

96 (94-97)

95 (93-97)

LR+

8 (6-10)

13 (8-23)

23 (10-57)

LR-

0.3 (0.2-0.5)

0.6 (0.5-0.8)

0.8 (0.6-0.9)

The prevalence of intensive care unit admission in the dataset was 6.3%. Test characteristics are given with their 95% confidence intervals. PPV, positive predictive value; NPV, negative predictive value; LR+, positive likelihood ratio; LR-, negative likelihood ratio.

The Combined test is positive if field triage is severe and BGL >=200 mg/dL, and negative if either field triage is not severe or BGL b200 mg/dL or both. BGL and ICU admission data are jointly available for more subjects (706) than are Field Triage and ICU admission data (681), and the Combined data are necessarily restricted to their minimum (681).

Life-saving interventions performed in the field included endotracheal intubations (N = 46), tube thoracostomies (N = 20), cricothyroidotomies (N = 6), and tourniquet placements (N = 96). Mean BGL was 167 mg/dL in the group in which at least one LSI was performed versus 111 mg/dL in the group in which no LSI was performed.

Finally, because only 13% of the ITR had recorded glucose levels on hospital arrival, we compared important variables between the entire dataset (N = 5553) and our subset (N = 706) as a rough assessment of selection bias. The entire dataset had 92% males; median age 20 years (IQR 19-22 years); penetrating mechanism in 36%; prehospital triage levels of 17% severe, 28% moderate, and 55% mild; median ISS of 4 (IQR 1-9); median hospital days 2 (IQR 1-5); median ICU days among those admitted to the ICU of 3 (IQR 2-7); and 4% mortality (most of whom were killed in action, and thus did not receive care in a hospital). (See Table 1 to compare with the subset that constitutes the present study.) Primary settings includED operations (44%), routine activities (42%), and training (8%).

Discussion

Elevated blood glucose levels among admitted patients have been shown to be associated with worse outcomes. Sung et al reported a relationship between an admission preoperative BGL greater than 200

Table 4

Predicting major injury (defined by injury severity score N 15) or ICU admission among casualties triaged in the field as moderate using blood glucose level cut-off of 200 mg/dL

Blood glucose level Blood glucose level

>=200 mg/dL b200 mg/dL >=200 mg/dL b 200 mg/dL

ISS N 15 1 9

ISS b 15 9 184

ICU admission 5 8

ICU nonadmission 5 190

Sensitivity (%)

10 (0.25-44)

39 (14-68)

Specificity (%)

95 (91-98)

97 (94-99)

PPV (%)

10 (0.25-44)

50 (19-81)

NPV (%)

95 (91-98)

96 (92-98)

LR+

2 (0.3-15)

15 (5-45)

LR-

0.9 (0.8-1.2)

0.6 (0.4-1.0)

Major injury was present in 4.9% and intensive care unit admission was required in 6.2% of these casualties. Test characteristics are given with their 95% confidence intervals. PPV, positive predictive value; NPV, negative predictive value; LR+, positive likelihood ratio; LR-, negative likelihood ratio.

Casualties for whom ISS is missing do not appear in the upper left corner table; numbers will thus not necessarily sum to those in Table 1.

mg/dL and an increase in infection rate, intensive care and overall hospital lengths of stay, and mortality [4]. These correlations were demonstrated after adjustment for age and ISS, and known diabetics were excluded. A later study at the same institution found that an acute elevation in glucose also portends infection; diabetics were included in this population [5].

Laird et al reported a positive correlation between high BGL during the first and second days of admission to the ICU and mortality [6]. Their cohort consisted of 516 non-diabetic trauma patients admitted to the trauma ICU. Using a multiple regression model, they found that a BGL of 200 mg/dL or greater was an independent predictor of both infection and mortality, but no such relationship was found with threshold levels of 110 mg/dL or 150 mg/dL.

Paladino et al found that non-Diabetic patients with Major injuries had significantly higher BGL on admission to the emergency department, concluding that high glucose levels were as good as Base deficit and lactate levels in differentiating major from Minor injury [7]. They added that a BGL cutoff of 200 mg/dL performed better than the Revised Trauma Score in identifying major injury.

In the setting of Traumatic head injury, Jeremitsky et al found hyperglycemia to be among the leading factors of secondary injury following traumatic brain injury. They reported that 52% of their patients who suffered from at least one episode of hyperglycemia (N 200 mg/dL) did not survive, whereas only 28% of the patients who did not experience an episode of hyperglycemia did not survive [8]. Other authors have similarly argued that elevated BGL (N 200 mg/dL) upon admission probably reflected extensive brain injury and correlated with poor neurological outcome and mortality in head injured patients, in both pediatric and adult populations [9-11].

In the battlefield setting, Savic et al analyzed 59 military gunshot and missile wound casualties in former Yugoslavia [12]. A positive correlation between admission BGL and ISS was demonstrated, though correlation with the Red Cross Wound Classification was less clear. The authors concluded that glucose measurement could be a useful supplementary tool for rapidly assessing severely wounded patients. Currently, there are conflicting studies [13-15], including in the trauma setting [16], addressing whether tight glucose control improves outcomes in critical patients. These research efforts, should they ultimately demonstrate a benefit, may provide additional

evidence that elevated glucose is related to worse outcomes.

Our study findings are consistent with BGL being a possible tool for improving field triage, at least in the military setting. The relationship we identified between high BGL and the need for prehospital life- saving interventions further supports BGL as a marker of injury severity. Our data are also in accord with the previous authors’ findings [4,6], supporting the use of blood glucose levels for in- hospital measurement of severity of injury.

The study of military casualties offers several unique characteris- tics. Most casualties are fit, young adults with no significant disease; specifically, the prevalence of diabetes, as well as other chronic illnesses, is extremely low. Conversely, the evacuation time to a surgical facility is longer than would be found in most civilian trauma. While the former helps reduce the heterogeneity of the study population, the latter makes the use of a hospital-based BGL measurement as a proxy for a field-based measurement more problematic. With regard to the latter, it is encouraging that the high BGL group is over-represented among those with shorter transport times, as this may indicate that there was less time for a fluctuation in BGL during transport of this group. This phenomenon is likely due to the more expedient evacuation of more severely injured patients, as determined by field triage level. As we have shown, sicker patients tend to have higher BGLs, and thus high BGL (though unknown in the field) would be correlated with shorter transport times.

Our analysis of BGL as a potential prehospital diagnostic test rests on the assumption that the BGL measured upon hospital arrival correlates well with the BGL that would have been measured in the field (with respect to whether it is above or below 200 mg/dL).

Our data suggest that elevated BGL may find use as an adjunctive prehospital triage tool for predicting ISS and ICU usage. The latter may be important both in evacuation decisions and receiving hospital preparation. We note that hyperglycemia very likely influences the decision to admit the injured to the ICU. However, the prediction of ICU use, even if in some sense self-fulfilling, is still an accurate indicator for resource planning. It is important we mention that based on our data, BGL should not be used in lieu of, but rather in conjunction with, field triage by trained providers.

If field triage is not severe, the likelihood of major injury is considerably reduced. (The sensitivity seems only fair, but it is combined with a low prevalence of major injury yielding a reasonable negative predictive value.) If field triage is severe and the BGL is positive (ie, >=200 mg/dL), then it is quite likely that the casualty will have major injury or (independently) require ICU admission. BGL improved the identification of casualties with major injury or in need of ICU admission beyond that available from field triage (severe) alone.

If field triage is moderate, then the addition of BGL is helpful in identifying casualties requiring ICU care (even if for “self-fulfilling” reasons as described above). While a BGL >=200 mg/dL more than doubles the odds that a casualty triaged as moderate has major injury, the absolute change is less impressive (5%-10%), and we thus see BGL as less useful in identifying major injury in this group.

A strong association was found between our dichotomized BGL and both mortality and injury severity as reflected by the ISS, consistent with the literature cited above [3,5]. That the BGL was positively associated with the severity of the casualties as determined by the field medical providers may be seen as indirect evidence of accurate triage. Previous work has shown that IDF triage has low under- and acceptable over-triage rates [17].

A positive correlation not previously reported was found between elevated BGL and prehospital life-saving interventions. The odds of requiring LSI were 11 times higher among those casualties with BGL >= 200 mg/dL. This of course may be a marker of the severity of injury and the need to perform a LSI, but it may also reflect the stress involved with undergoing the LSI itself, among other Prehospital factors.

We found that mean BGL was increased in the N 29-year-old group versus the 17- to 29-year-old group. In addition, the only two casualties who were triaged as mild (and indeed had low ISS) but had high BGL were over 29 years old. This might suggest caution should be employed when analyzing BGLs of casualties over the age of 29, among whom 15% had a BGL >=200 mg/dL. While we noted that average BGL was also somewhat higher in the b 17 year old population, we have very few data points in that age range.

The higher BGLs among casualties wounded while engaged in operational duties might be explained by higher stress levels and especially the more lethal weapons involved. Terror victims, with the next highest BGLs, could be explained similarly. In terms of mechanisms, the highest mean BGLs were seen among casualties suffeRing crush injury or extremity amputation, followed by Penetrating injuries from shrapnel and firearms. Apart from crush injury, which was primarily from motor-vehicle accidents, almost all of the casualties experiencing these mechanisms were involved in military operations.

We cannot identify casualties with diabetes mellitus (pre- existing or otherwise) in our dataset. While this is less of a concern in a military casualty setting, we would expect the test character- istics in the Tables to be similar in other populations with similar casualty characteristics, including prevalence of diabetes among the treated casualties.

Our study suffers several important limitations in addition to those discussed above, including being retrospective in nature and dealing with a mixed cohort of patients (different scenarios, evacuation times, causes of injuries, etc.). In addition, the study population underwent treatment in multiple different hospitals, differing in their capabilities (a mixture of level 1 and 2 trauma centers), which may influence, especially, length of stay and survival.

Out of more than 5000 cases in the ITR, only 706 records included the glucose level, which may result in significant selection bias, though we have reason to suspect that it is primarily missing at random. A comparison of the entire dataset and the subgroup used in the present study shows no evidence consistent with selection bias, though the decreased mortality required explanation. While this is reassuring, we cannot of course exclude selection bias of this nature. Finally, we have no information on casualties who were not evaluated in a hospital (either because they had minor injuries not requiring evacuation or because they died before hospital arrival), an additional source of possible selection bias as we cannot know what the glucose levels were in this group. On the other hand, their triage

status is likely clear.

Conclusion

Blood glucose levels measured on hospital arrival were associated with major injury and ICU usage among casualties triaged as severe. Combining BGL with standard field triage may allow for more accurate identification of casualties who need acute field intervention, have major injury, or require ICU-level care. Future research should determine whether BGLs measured in the prehospital setting are indeed useful.

References

  1. McNamara JJ, Molot M, Stremple JF, et al. Hyperglycemic response to trauma in combat casualties. J Trauma 1971;11:337-9.
  2. Desborough JP. The stress response to trauma and surgery. Br J Anaesth 2000;85: 109-17.
  3. Maurer A, Morris Jr JA. Injury severity scoring. In: Moore EE, Feliciano DV, Mattox KL, editors. Trauma. 5th ed. New York: McGraw-Hill; 2004. p. 87-91.
  4. Sung J, Bochicchio GV, Joshi M, et al. Admission hyperglycemia is predictive of outcome in critically ill trauma patients. J Trauma 2005;59:80-3.
  5. Bochicchio GV, Bochicchio KM, Joshi M, et al. Acute glucose elevation is highly predictive of infection and outcome in critically injured trauma patients. Ann Surg 2010;252:597-602.
  6. Laird AM, Miller PR, Kilgo PD, et al. Relationship of early hyperglycemia to mortality in trauma patients. J Trauma 2004;56:1058-62.
  7. Paladino L, Subramanian RA, Nabors S, et al. Triage hyperglycemia as a prognostic indicator of major trauma. J Trauma 2010;69:41-5.
  8. Jeremitsky E, Omert L, Dunham CM, et al. Harbingers of poor outcome the day after severe brain injury: hypothermia, hypoxia, and hypoperfusion. J Trauma 2003;54: 312-9.
  9. Young B. Relationship between admission hyperglycemia and neurologic outcome of severely brain-injured patients. Ann Surg 1989;210:466.
  10. Merguerian PA. Persistent Nonketotic hyperglycemia as a grave prognostic sign in head-injured patients. Crit Care Med 1981;9:838-40.
  11. Asilioglu N, Turna F, Paksu MS. Admission hyperglycemia is a reliable outcome predictor in children with severe traumatic brain injury. J Pediatr (Rio J) 2011;87:325-8.
  12. Savic J, Cernak I, Jevtic M, et al. Glucose as an adjunct triage tool to the Red Cross Wound Classification. J Trauma 1996;40:S144-7.
  13. van den Berghe G, Wouters P, Weekers F, et al. Intensive insulin therapy in critically ill patients. N Engl J Med 2001;345:1359-67.
  14. Wiener RS, Wiener D, Larson RJ. Benefits and risks of tight glucose control in critically ill adults: a meta-analysis. JAMA 2008;300:933-44.
  15. Kavanagh BP. Glucose in the ICU-evidence, guidelines, and outcomes. N Engl J Med 2012;367:1259-60.
  16. Scalea TM, Bochicchio GV, Bochicchio KM, et al. Tight Glycemic control in critically injured trauma patients. Ann Surg 2007;246:605-10.
  17. Katzenell U, Yegorov Y, Tapia AL, et al. Triage in the military setting-the experience of IDF [Submitted].