Introduction

The emerging understanding of early coagulopathies and their clinical consequences after severe trauma have created a search for better coagulation assays. Current routine screening coagulation tests (RSCTs), such as activated partial thromboplastin time (aPTT) and prothrombin time (PT), have limited utility to diagnose early trauma coagulopathies and direct their treatment. Neither test predicts the extent of bleeding in critically ill or trauma patients [1], and a recent systematic review concluded that they are inappropriate for trauma [2]. The cell-based understanding of hemostasis [3], emphasizing tissue factor (TF) as the initiator of coagulation and the role of platelets, has challenged the clotting cascade concept that underlies RSCTs. The cell-based model and the need for shorter turnaround time (TAT) for tests to guide transfusion in bleeding trauma patients have propelled interest in thromboelastography (TEG® ; Hemoscope Corporation, Niles, IL, USA) and thromboelastometry (ROTEM® ; Tem International GmbH).

TEG® and ROTEM® are based on the principle that the result of the hemostatic process is a clot whose physical properties determine patients’ hemostatic status. These tests provide global information on the dynamics of clot development, stabilization, and dissolution, reflecting in vivo hemostasis, and assess both thrombosis and fibrinolysis [4]. The additional information from TEG®/ROTEM® is based on their performance in whole blood [4], whereas RSCTs are performed in plasma, without the cellular components of platelets and tissue-bearing cells.

By systematically searching for relevant studies, we sought to evaluate the evidence that the use of TEG® and ROTEM® in adult traumatically injured patients (a) diagnoses trauma coagulopathies on admission to hospital, (b) guides transfusion, and (c) reduces mortality.

Materials and methods

This descriptive systematic review was reported in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [5].

Information sources and search technique

With the assistance of an experienced librarian, we searched MEDLINE (1946 to February 2014), EMBASE (1947 to February 2014), and Cochrane Controlled Trials Register (from inception to February 2014) to identify studies of thromboelastography and thromboelastometry in trauma. We used a sensitive search strategy combining MeSH headings and the key words “thromboelastography” AND “trauma,” “thromboelastometry” AND “trauma,” “thromboelastography” AND “injury,” “thromboelastometry” AND “injury,” TEG® AND “trauma,” TEG® AND “injury,” ROTEM® AND “trauma,” and ROTEM® AND “injury.” Search terms were defined a priori and by reviewing the MeSH terms of articles identified in preliminary literature searches. Two authors (LTL, AKS) independently reviewed the abstracts of all articles identified by the literature search and selected articles for detailed review if either reviewer considered them potentially relevant. We also searched the bibliographies of all articles selected for detailed review and all relevant published reviews to find any other studies potentially eligible for inclusion. We did not search conference proceedings. No language restrictions were imposed; we translated two studies in Spanish and Italian and engaged a medical student to translate one Chinese study that was ultimately excluded. Details of the search strategies are in Additional file 1.

Eligibility criteria and study selection

Studies were eligible for inclusion if they were observational studies or randomized controlled trials (RCTs) that evaluated TEG®/ROTEM® in adult trauma patients and reported outcomes related to diagnosis of coagulopathies (hypocoagulation, hypercoagulation, platelet dysfunction, hyperfibrinolysis (HF), TAT), transfusion management (prediction of massive transfusion (MT), and transfusion guidance), or mortality (prediction and reduction). Studies were excluded if they enrolled only burn patients or enrolled patients in other surgical specialties, or were case reports or case series. Two independent reviewers (LTL, AKS) reviewed all full-text versions of all potentially eligible studies. Agreement between reviewers was assessed by using the Cohen κ [6]. In case of disagreement, consensus was reached by discussion with a third author (BN, NKJA).

Data abstraction and analysis

We abstracted data from included studies on study objective, setting and study design, patient selection, clinical and demographic characteristics, TEG®/ROTEM® technique, RSCT technique, presence of comparison group, blood-product transfusion, and mortality. Two authors (LTL, AKS) independently assessed study methodology based on the Newcastle-Ottawa Scale for cohort studies [7] and QUADAS-2 [8] for quality assessment of diagnostic accuracy studies. For studies that did not report diagnostic accuracy, we supplemented the Newcastle-Ottawa scale by assessing the description of TEG®/ROTEM® performance. In applying the Newcastle-Ottawa scale, we considered management by TEG®/ROTEM® to be the relevant exposure and a nonexposed cohort to be one that was managed without TEG®/ROTEM®. We considered the following outcomes: (a) diagnostic performance of TEG®/ROTEM® parameters compared with RSCT (PT, aPTT, INR, platelet count, fibrinogen) for early coagulopathies, (b) utilization of blood products red blood cells (RBCs), fresh frozen plasma (FFP), platelets concentrate (PLT), fibrinogen concentrate (FC), cryoprecipitate, prothrombin complex concentrate (PCC), and (c) mortality. Because of clinical and methodologic heterogeneity among studies, we anticipated reporting results qualitatively instead of conducting meta-analyses.

Results

The electronic search identified 1,352 potentially relevant studies. After evaluating 82 full-text manuscripts, 55 met inclusion criteria (Figure 1). An excellent agreement was reached between the reviewers for study inclusion (κ = 0.82). References for the excluded studies are in Additional file 1.

Figure 1
figure 1

Flow of studies through the systematic review.

Full size image

Study characteristics

All 55 studies were observational (Table 1). Forty studies [9]-[48] addressed the use of TEG®/ROTEM® in diagnosing early coagulopathies; 25 studies [9],[15],[16],[22],[23],[25],[26],[29],[31],[32],[35]-[37],[41],[45],[47]-[56] examined associations with transfusion; and 24 studies [14],[16],[21],[23],[29],[32],[33],[36]-[38],[40],[45],[47],[48],[51],[54],[56]-[63] examined associations with mortality. Only three [40],[41],[61] were conducted in multiple centers. Thirty-eight studies were prospective cohorts [9]-[14],[16],[17],[19],[20],[24]-[28],[30],[32],[33],[36],[38],[40]-[48],[50],[52],[53],[55]-[58],[61],[62], two were before-after [37],[63], and 15 were retrospective cohorts [15],[18],[21]-[23],[29],[31],[34],[35],[39],[49],[51],[54],[59],[60]. The techniques used for TEG® and ROTEM® varied: in 28 studies [9],[11]-[14],[17],[20],[22],[26],[28],[30],[33],[38]-[41],[44],[45],[47],[49],[50],[52]-[55],[58],[59],[61],[63], the tests were done at 37°C; in nine [10],[15],[16],[18],[19],[21],[37],[43],[57], at the patient’s temperature; in 16, there was no description of the temperature [23]-[25],[29],[31],[32],[34]-[36],[42],[46],[48],[51],[56],[60],[62]; and one study performed the test at different temperatures [27]. Rapid TEG® (r-TEG® ), a technique with fresh whole blood using a solution containing TF as the coagulation trigger, was used in 12 studies [17],[18],[21],[25],[33]-[35],[37],[43],[49],[52],[54]; ROTEM® , in 18 studies [12],[14],[20],[22],[23],[26],[28],[31],[32],[38],[40],[42],[50],[51],[55],[58]-[60]; TEG® , in 23 studies [9]-[11],[13],[15],[16],[19],[24],[27],[29],[30],[36],[39],[41],[44],[45],[53],[56]-[58],[61]-[63]; and two studies performed TEG® and r-TEG® [17],[46] in the same cohort of patients.

Table 1 Characteristics of the studies included in the systematic review
Full size table

Demographic data

The 55 studies included 12,489 patients. The mean or median age of the patients ranged from 24 to 74 years (mean of mean or median age across all studies, 40 years). The majority were male (9,858 patients, 78.9%), and the mean injury severity score (ISS) [64] ranged from 9 to 55 (mean of means in all studies, 25.3). Studies included trauma with or without traumatic brain injury (TBI) (n = 49), isolated TBI (n = 4), and burns with all types of trauma (n = 2). The median (interquartile range (IQR)) sample size of the included studies was 87 (52 to 235).

Methodologic quality

The overall methodologic quality of the studies was moderate (Table 2). Most studies (n = 36) included consecutive patients [10],[12]-[14],[16],[19],[21]-[27],[30]-[35],[37]-[41],[45],[46],[50]-[52],[54]-[56],[58]-[60],[63]. Fifty-two (94.5%) studies were not controlled [9]-[11],[13]-[36],[38]-[50],[52]-[62]. The three (5.5%) controlled studies [37],[51],[63] had comparable control groups managed without TEG®/ROTEM®. Eleven (20%) studies used healthy [12],[13],[19],[27],[41],[42],[44],[61] or other hospitalized trauma [24],[32],[60] controls to examine associations between TEG®/ROTEM® abnormalities and outcomes. Nearly all (n = 53, 96.4%) studies had adequate follow-up. The mean Newcastle-Ottawa score (n = 55 studies) was 6.07 (SD, 0.49; possible range, 1 to 9).

Table 2 Newcastle-Ottawa score[6]for the cohort studies included in the systematic review
Full size table

We assessed 47 studies of diagnostic accuracy by using QUADAS-2 (Table 3). Considering the domains of patient selection, index test, reference standard, and flow and timing, only three studies (6.4%) [32],[37],[39] had low risk of bias in all; 24 (51.1%) [10],[16],[21]-[26],[30],[31],[33]-[35],[38],[40],[44],[45],[47],[50],[52],[54],[55],[58],[62] had low and unclear risks; and 20 (42.5%) [9],[11]-[15],[17]-[20],[29],[36],[42],[48],[49],[53],[57],[59]-[61] had high risk of bias in at least one domain. In the applicability section, 37 studies (78.8%) [9]-[11],[16]-[18],[21]-[26],[29]-[40],[44],[45],[47],[48],[50],[52]-[55],[57]-[59],[62] had low concerns, and 10 studies (21.2%) [12]-[15],[19],[20],[42],[49],[60],[61] had at least one domain with high concern. The remaining eight studies that that did not evaluate diagnostic performance against a reference standard are represented in Table 4.

Table 3 Assessment of studies of diagnostic performance of TEG®/ROTEM® by using the QUADAS-2[8]tool
Full size table
Table 4 Details of TEG®/ROTEM® for studies that did not evaluate diagnostic performance against a reference standard
Full size table

Outcomes

The included studies collectively reported on hypocoagulability, hypercoagulability, platelet dysfunction, and hyperfibrinolysis. We summarize the parameters of TEG®/ROTEM® used for diagnosis, turnaround times (TATs), and results concerning prediction, reduction, or guidance of transfusion and prediction or reduction of mortality. Details of all studies without a control group (for clinical outcomes) or a reference standard (for diagnostic performance) are given in Table 5.

Table 5 Main findings of the included studies
Full size table

Diagnosis of early trauma coagulopathies

Hypercoagulability

Six studies demonstrated hypercoagulability in trauma patients not detected by RSCT. Hypercoagulability was defined mostly by the manufacturer of both TEG® and ROTEM® devices, and the reference standards (where reported) were Doppler ultrasound, CT angiography, or a surgical procedure demonstrating thrombus. One study [11] demonstrated that 62% of trauma patients were hypercoagulable according to TEG® (R <3.7) on the first day after injury with normal RSCT values. Another study [19] showed higher TEG® α-angle (which reflects the degree of fibrin cross-linking) and MA (maximal amplitude) in trauma compared with RSCT (PT and aPTT) (P <0.05), indicating a hypercoagulable state. A study [24] detected increased G (shear elastic modulus strength (5,000 – MA)/(100 – MA), which reflects clot strength) in a cohort of trauma patients after splenectomy, who had more thromboembolic events, compared with patients treated nonoperatively. Another study of rapid TEG® (r-TEG® ) [18] showed that increased G was associated with thromboembolic complications (OR, 1.25; 95% CI, 1.12 to 1.39), after controlling for thromboprophylaxis and using the reference standards discussed earlier. When TEG® [27] was performed 24 hours after injury, trauma patients were more hypercoagulable compared with healthy volunteers across a broad range of temperatures (32°C to 38°C).

Finally, a cohort study [34] suggested an association between admission MA and pulmonary embolism (OR, 3.5 for MA >65; 95% CI, 1.69 to 7.23; and OR, 5.8 for MA >72; 95% CI, 2.85 to 11.77), after controlling for gender, race, age, and ISS.

Hyperfibrinolysis (HF)

Only one study [14] compared HF detected with ROTEM® with a laboratory gold standard and showed that ROTEM® had satisfactory diagnostic properties for HF, defined by laboratory measurement of euglobin lysis time (ELT). However, the sample size was very small (n = 23, of which five had HF), limiting the strength of inferences.

Platelet dysfunction

With TEG® platelet-mapping (PM) test, a study [13] showed that patients with TBI had more platelet dysfunction on admission, measured by lower platelet response to arachidonic acid (AA) but not to adenosine diphosphate (ADP), compared with non-TBI trauma patients, alcohol abusers, and healthy volunteers (P <0.001). Another study [41], using the same technique, found lower platelet response to both AA and ADP in trauma patients versus healthy volunteers (P <0.0001). With ROTEM® , one study [42], comparing healthy volunteers with trauma patients, speculated that an observed difference in clot strength arose from platelet dysfunction. A related study [65] of ROTEM® at emergency department (ED) admission demonstrated significantly lower values of platelet component of clot elasticity (MCE EXTEM, ROTEM® extrinsically activated test with TF and MCE FIBTEM, ROTEM® fibrin-based extrinsically activated test with TF and the platelet inhibitor cytochalasin D) in trauma nonsurvivors vs. survivors (P = 0.0012).

Hypocoagulability

Six studies directly compared TEG® or ROTEM® with RSCT, with variable results for diagnostic performance. One study [9] demonstrated that TEG® detected hypocoagulability in 45 (85.5%) of 52 patients, whereas only one (1.9%) of 52 was hypocoagulable by elevated PT/aPTT and two (3.8%) of 52 were hypercoagulable by elevated platelet count. In a cohort [12] of trauma patients and healthy volunteers, EXTEM A15 ≥ 32 mm (amplitude at 15 minutes) and FIBTEM A10 ≥ 5 mm were sensitive (87% and 91%) and specific (100% and 85%) for detection of PT >1.5 of control value and fibrinogen <1 g/L, respectively. Doran [20] found that 16 (64%) of 25 patients were hypocoagulable by ROTEM® trace, and only 10% had abnormal RSCT (P = 0.0005). In contrast, another study [39] (n = 219) found that TEG® -R (R, reaction time, defined as the time until a clot firmness of 2 mm is achieved, corresponding to CT, clotting time, in ROTEM® ) performed worse than INR for the diagnosis of vitamin K deficiency in trauma patients (clotting factor activity used as gold standard). TEG® -R (compared with INR >1.5) had a sensitivity of 33% (67% for INR), specificity of 95% (98% for INR), positive predictive value (PPV) of 47% (84% for INR), and negative predictive value (NPV) of 92% (96% for INR). In another cohort [26] (n = 300), ROTEM® parameters of CFT (clot-formation time), α-angle, A5 (clot amplitude 5 minutes after CT) and MCF (maximum clot firmness) were significantly different in the group with coagulopathy, defined by INR >1.2. A5 ≥ 35 mm had a sensitivity of 77% and specificity of 87% for the detection of coagulopathy. Finally, a recent large study [40] (n = 517) found that ROTEM® EXTEM, and FIBTEM measures of A5 and MCF were significantly correlated with fibrinogen levels (EXTEM A5, r2 = 0.35, and MCF, r2 = 0.26; FIBTEM A5, r2 = 0.44, and MCF, r2 = 0.27; all P <0.001). The sensitivity/specificity of EXTEM A5 < 36 mm (FIBTEM A5 < 9.5 mm) for discriminating patients with admission fibrinogen <1.5 g/L were 53%/87% (78%/70%; ROC AUC 0.8, 95% CI 0.7 to 0.9 for both).

One study [46] comparing the same parameters measured by r-TEG® and conventional kaolin-activated TEG® found strong correlation for MA (marker of platelet function; r = 0.80); moderate correlation for G (overall clot strength; r = 0.70), k (speed of clot formation; r = 0.66), and α-angle (degree of fibrin cross-linking; r = 0.38); and poor correlation for LY30 (degree of fibrinolysis; r = 0.19).

Although TEG® and r-TEG® may be moderately sensitive in detecting abnormal clot strength, they have not differentiated between fibrinogen and platelet contributions to clot integrity. Recent studies of the TEG® -based functional fibrinogen assay (FF) have examined the relative contribution each. By using Kaolin TEG® MA to define coagulopathy, Johansson [45] showed that TEG® -FF MA and G were lower in hypocoagulable patients and significantly higher in hypercoagulable patients compared with patients with normal Kaolin TEG® MA (P <0.001). In another study [47], coagulopathic patients (INR ≥1.3) had lower admission fibrinogen contribution to MA than did noncoagulopathic patients (24.7% versus 31.2%; P <0.05). Platelet contribution to MA was higher than fibrinogen at all time points, decreased over time, and stabilized at 72 hours (69.4% at 0 hours, 56.2% at 72 hours). In contrast, fibrinogen contribution to MA increased over time and stabilized at 72 hours (30.6% at 0 hours, 43.8% at 72 hours).

Turnaround times

Four studies reported on the use of TEG® and ROTEM® as POC devices. Carroll [16] demonstrated no statistical difference in the diagnosis of acute trauma coagulopathies when collecting blood on site or 1 hour after admission in the emergency department (ED), except for a small but statistically significant change in MA (60.6 (SD 11.1) mm on site and 63.4 (SD 12.1) mm in ED; P = 0.014). A cohort study [17] demonstrated that r-TEG® had a shorter TAT (time to MA) by a median (IQR) of 10.8 (1.1 to 18.5) minutes compared with TEG® . In another cohort [26], laboratory PT had a median TAT of 78 minutes (IQR, 62 to 103 minutes), whereas ROTEM® A5 was available by 5 minutes.

Finally, in another study [25], early r-TEG® values (ACT [activated clotting time], k-time, and r-value) were available within 5 minutes, late r-TEG® values (MA and α-angle) within 15 minutes, and RSCTs within 48 minutes (P <0.001 for all comparisons with r-TEG® ).

Blood transfusion

Prediction of massive transfusion (MT) and any transfusion

Several studies compared ROTEM® and TEG® parameters with RSCTs for prediction of MT (defined by most studies as transfusion of ≥10 RBC units within 24 hours of trauma). Davenport [26] found better sensitivity (71% versus 42%) for ROTEM® A5 ≥ 35 mm versus INR >1.2, but worse specificity (85% versus 94%). Another study [31] found that FIBTEM (0 to 3 mm) had the highest AUC (0.84; 95% CI, 0.79 to 0.88) among ROTEM® tests for prediction of MT, but hemoglobin (AUC 0.87; 95% CI, 0.83 to 0.91) and PT (AUC, 0.87; 95% CI, 0.83 to 0.90) were better predictors.

A third study [22] showed multiple ROTEM® tests to be associated with MT; in a multivariable analysis limited by few events, hemoglobin ≤100 g/L (OR, 18.18; 95% CI, 2.73 to 125.00) was a stronger predictor of MT than abnormal MCF (OR, 8.47; 95% CI, 1.19 to 62.50). Similarly, RCSTs and TEG® had similar abilities to predict MT (G AUC 0.89; 95% CI, 0.89 to 0.96; INR AUC, 0.92; 95% CI, 0.86 to 0.98; PTT AUC 0.90; 95% CI, 0.83 to 0.97) [54].

ROTEM® parameters significantly associated with MT include increased CFT [22],[31], decreased MCF [22],[31], prolonged EXTEM and INTEM CT (intrinsically activated test using ellagic acid, clotting time) [31] and FIBTEM A10 (ROC AUC, 0.83; 95% CI, 0.78 to 0.87) [31]. For TEG® , statistically significant differences in α angle, MA, K, G (at 1 hour), and estimated lysis according to transfusion need (minimal, moderate, or massive) were reported [21]. Another study [36] found that patients with HF, defined by estimated lysis >15%, had a greater need for MT (76.9% versus 8.7%; adjusted OR, 19.1; 95% CI, 3.6 to 101.3).

Finally, a recent study demonstrated that a TEG® LY30 (percentage decrease in clot amplitude at 30 minutes after MA) of 3% or greater had poor sensitivity (31%) but high specificity (91%) for predicting MT [56].

Investigators have also compared TEG®/ROTEM® with RCST for prediction of any blood-product transfusion, generally guided by RSCT or a massive transfusion protocol (MTP). One study [52] found that r-TEG® α-angle <74.7 degrees had higher sensitivity (84%) but lower specificity (57%) to predict transfusion of any blood product versus INR >1.2 (38%, 88%), INR >1.5 (19%, 96%), or aPTT >60 (5%, 98%); fibrinogen <3 g/L had the highest sensitivity (90%) and similar specificity (48%). Another study [26] showed a similar pattern for ROTEM® A5 ≥ 35 mm in prediction of any RBC (sensitivity 33%, specificity 88%) or FFP (36%, 87%) compared with INR >1.2 (17%, 96% for RBC; 21%, 96% for FFP).

Considered in isolation, many TEG®/ROTEM® abnormalities have been associated with transfusion of specific blood products. Reduced TEG® MA was associated with transfusion of RBC [15],[29], FFP [29], and PLT [29]; TEG® ACT predicted RBC, FFP, and PLT transfusions within the first 2 hours [25]; α-angle <56 predicted MT of RBC, FFP, PLT, and cryoprecipitate [35]; and patients with r-TEG® -defined maximum rate of thrombin generation (MRTG) ≥9.2 mm/min at 3 hours received significantly less RBC, FFP, and cryoprecipitate in the first 6 hours [37]. A study combining several markers [53] showed that patients considered coagulopathic based on TEG® (R, K, α-angle, and MA) received more RBC (10 versus 0), FFP (7 versus 0), and PLT (3 versus 0) in the first 24 hours (P <0.05). More recently, the inhibition of ADP function was correlated with transfusion in the first 6 hours (59.6% inhibition (0 RBC) versus 96.1% inhibition (>1 RBC), P = .025) [41], and TEG® -defined hypocoagulable patients required more uncross-matched blood (adjusted P = 0.004) and FFP (adjusted P <0.001) at 24 hours compared with normocoagulable and hypercoagulable patients [48].

Finally, patients requiring FFP had a significantly lower admission fibrinogen contribution to MA by the FF assay (26.6% versus 30.6%; P <0.05). In contrast, one study [16] showed no significant differences in TEG® parameters (R, K, α-angle, MA, LY60 (percentage decrease in clot amplitude at 60 minutes after MA) in patients transfused with RBC, FFP, or PLT versus not; only MA-ADP, a measure of ADP-platelet activation, correlated with transfusion of any blood product (P = 0.004).

Similarly, nonsurviving patients with isolated TBI had more ROTEM® -defined coagulation abnormalities and received more RBCs (P = 0.016), FC (P = 0.01) and PCC (P <0.001) than survivors [23], and higher FIBTEM MCF was associated with decreased RBC transfusion (adjusted OR, 0.92 per 1-unit increase, 95% CI, 0.87 to 0.98) [32]. Patients with HF [55], based on plasmin-antiplasmin complex levels and ROTEM® maximum lysis (ML) >15%, required more transfusions of RBC, FFP, PLT, and cryoprecipitate (P <0.05 for all comparisons).

Few data exist on TEG®/ROTEM® to monitor coagulopathy in response to transfusion. Rourke et al.[40] showed that administration of high doses of FC (6 to 12 g) or cryoprecipitate (30 U) restored EXTEM and FIBTEM A5 and MCF to the level of patients with minor injuries; fibrinogen levels were maintained but not augmented after FC administration. Changes in ROTEM® -determined CT, A5, and MCF after the transfusion of 4 units of RBC, dependent on the RBC/FFP ratio (≥1:1, 1:2 to 3:4, <1:2) [50].

TEG®/ROTEM®-guided transfusions versus conventional guidance

A modeling study [49] (n = 44) suggested that transfusion guided by r-TEG® versus RSCT would reduce the proportion of patients needing blood-product transfusion from 73.1% to 53.9% (P = 0.03), driven by reductions in FFP administration (61.5% with INR trigger versus 26.9% with r-TEG® -ACT trigger, P = 0.003). No difference was predicted for transfusion of PLT (P = 0.32) or cryoprecipitate (P = 0.18). A cohort study reported lower exposure to blood products in 80 patients with ROTEM® -guided FC and PCC compared with a historical group (n = 601) for whom FFP and PLT transfusions were guided by clinical decision (generally RSCT) [51]. Transfusion of RBC (PLT) was avoided in 29% (91%) of patients in the ROTEM® -guided FC and PCC group compared with 3% (56%) in the RSCT-guided FFP and PLT groups (P <0.001).

Mortality

Prediction

TEG® and ROTEM® parameters have been compared with RCSTs for prediction of mortality. For prediction of coagulopathy-related death, G had a similar adjusted AUC (0.93, 95% CI, 0.87 to 0.98) compared with INR (AUC, 0.88; 95% CI, 0.80 to 0.97; P = 0.11) and aPTT (AUC, 0.89; 95% CI, 0.81 to 0.97; P = 0.19) [54]. In another analysis, r-TEG® G, MRTG, and total thrombin generation (TG) (P = 0.03 for each) discriminated between survivors and nonsurvivors, in contrast to INR at 6 hours (P = 0.10) [37]. Another study found that TEG® -defined hypercoagulable patients had lower 24-hour mortality (0 versus 5.5% (normocoagulable) versus 27.8% (hypocoagulable), 10 deaths total, adjusted P <0.001) and hospital mortality (11.1% versus 5.5% versus 38.9%, 20 deaths total, adjusted P <0.001) [48]. With ROTEM® , a study [23] showed similar discrimination for mortality in TBI by using EXTEM with cytochalasin D (FIBTEM) MCF (AUC, 0.77; 95% CI, 0.66 to 0.85) and aPTT (AUC, 0.79; 95% CI, 0.69 to 0.87). A second study (n = 334; 26 early deaths) [32] using RSCTs, ROTEM® , and clinical judgment to guide transfusion, demonstrated significant correlations among PT, aPTT, fibrinogen, platelet count, and ROTEM® measurements (all |Spearman r| >0.5). In separate logistic regression analyses, each adjusted for hemoglobin and base excess, PT, aPTT, CT, CFT, MCF, LI (all EXTEM) and FIBTEM MCF were associated with 24-hour mortality. However, the predictive abilities of ROTEM® measurements and RSCTs were not directly compared.

Considering TEG®/ROTEM® measurements alone, weak clot strength [16],[21],[29],[32],[37],[45],[47],[54],[57],[58],[61],[62] and HF [14],[16],[21],[23],[32],[33],[35],[36],[38],[40],[45],[55],[56],[60] have been associated with morbidity and mortality. For example, ROTEM® -defined HF was associated with multiorgan dysfunction syndrome (MODS) (n = 115; 63.2% in patients with HF versus 24.6% in patients without HF, P = 0.004) [38] and hospital mortality (n = 115; 52.2% versus 12.9%, P <0.001) [38]; five of five versus nine of 82, P <0.05 [14]). Similarly in another study [36], TEG® -defined HF (estimated percentage lysis >15%) was associated with increased 24-hour mortality (69.2% versus 1.9% without HF; adjusted OR, 55.8; 95% CI, 7.2 to 432.3), but the number of deaths (n = 12) was extremely small. Holcomb [35] showed that r-TEG® LY-30, along with most other r-TEG® parameters, was associated with mortality; the logistic regression models were not described. More recently, a study [47] examined associations of the FF assay with mortality and demonstrated that a higher admission contribution of FF to MA predicted lower mortality (unadjusted hazard ratio, 0.815; P <0.001; 95% CI and adjusted model not reported).

Effect of TEG®/ROTEM®-guided transfusion

Several observational studies have examined whether TEG®/ROTEM®-guided transfusion reduced mortality after trauma and found no consistent effect. A small before/after study (n = 68) found lower crude mortality in patients in whom resuscitation was guided by r-TEG® (29% versus 65%) but did not report statistical testing or an adjusted analysis [37]. In a retrospective cohort study of massively bleeding patients (n = 131) transfused FC, PLT, and PCC using ROTEM® guidance, the observed mortality was 24.4%, which was lower than the expected mortality by TRISS (33.7%; P = 0.032) but similar to expected mortality by RISC (28.7%, P >0.05) [59]. Similarly, another cohort study of trauma patients (n = 681) found no difference in mortality between those treated in one center with ROTEM® -guided administration of FC and PCC (7.5%) and a multicenter control group treated with plasma transfusion (10%, P = 0.69) guided by the usual RSCT-guided clinical practice [51]. A before/after study [63] (n = 289) compared outcomes in trauma patients transfused with at least 6 U RBC in the first 24 hours according to TEG® -driven practice to a later period when a MTP not guided by TEG® was used. Overall, unadjusted mortality was unchanged, and MTP versus TEG® -directed care was not associated with mortality in multivariable analysis.

Discussion

Main findings

Our systematic review found 55 studies of TEG®/ROTEM® examining the diagnosis of trauma coagulopathies, including hypocoagulation, hypercoagulation, platelet dysfunction and fibrinolysis; guidance of blood-product administration; and associations with mortality. To our knowledge, this review is the first to summarize the literature on the use of TEG® and ROTEM® in trauma. The overall methodologic quality of included studies was moderate. No RCTs were reported; most cohort studies lacked clinically similar control groups managed without TEG®/ROTEM®, and standard measures of diagnostic accuracy were inconsistently reported. Observational data suggest that TEG® and ROTEM® may have adequate diagnostic properties for abnormalities identified by RSCTs and may identify additional coagulation disorders. However, the effect of these tests on the need for blood-product transfusion and mortality is unclear.

Studies in different but related clinical settings, not included in this systematic review, have also investigated TEG® and ROTEM® . Two studies of ROTEM® in a mixed population with shock [66] and noncardiac surgery [67] demonstrated rapid and useful results to guide decisions in hemostatic resuscitation. Other studies with mixed trauma and nontrauma populations [68]-[71] demonstrated good ability of ROTEM® to predict the need for MT, a clinically useful outcome regardless of management approach (laboratory abnormality-directed versus blood-product ratio based). Another possible advantage of these tests is that timely results may avoid FFP transfusion and subsequent FFP-related adverse events. However, relevant RCTs to test these hypotheses are lacking.

Other randomized [72]-[74] and observational [75]-[77] studies in cardiac surgery, burns, and mixed perioperative settings found reduced blood product transfusion and improved clinical outcomes after implementation of POC coagulation-management algorithms guided by ROTEM® . In contrast, existing observational studies in trauma patients do not suggest an effect of TEG®/ROTEM®-based transfusion protocols on clinically important outcomes, including mortality.

Strengths and weaknesses of this study

Major limitations of this review are related to the quality of the included studies. Reproducible technical standards for the performance of TEG®/ROTEM® were lacking, and inconsistent reporting of 2 × 2 tables precluded calculation of summary diagnostic test-performance measures and exploration of threshold effects. A major problem faced by diagnostic studies of trauma coagulopathy is the ambiguous nature of the gold standard, given that RSCTs may not provide an adequate description of all associated abnormalities. No RCTs exist in trauma patients, aside from one that enrolled burns patients exclusively [73], and the quality of the observational studies is modest. Studies differed in the use of TEG® or ROTEM® , and the few studies [78],[79] that compared TEG® and ROTEM® concluded that these methods are not interchangeable.

Studies also examined different patient populations, transfusion triggers, and transfusion protocols, limiting direct comparisons and generalizability. Clinical differences between many included studies and contemporary practice include substitution of FFP for clotting factors concentrate such as PCC, FC, and cryoprecipitate. Other methods for analysis of platelet dysfunction have been developed, such as platelet function analyzer (PFA-100) and multiple platelet function analyzer (Multiplate). These analyzers monitor different aspects of platelet function and appear to be technically reliable and practical POC devices, despite limitations [65],[80]-[82].

Two systematic reviews of TEG®/ROTEM® exist for nontrauma populations. A Cochrane review [83] included nine RCTs, mostly in cardiac surgery, that compared transfusion guided by TEG®/ROTEM® with transfusion guided by clinical judgment, RSCTs, or both in severely bleeding patients. The review found that TEG®/ROTEM® reduced blood loss by a mean of 85 ml (95% CI, 29 to 141 ml) but had no effect on mortality.

Another systematic review [84] included 16 observational studies and two RCTs in patients with sepsis and concluded that TEG®/ROTEM® (compared with RSCTs) detect impaired fibrinolysis, which may help to discriminate between sepsis and systemic inflammatory response syndrome (SIRS). However, limitations of data prevented conclusions regarding the value of TEG®/ROTEM® to identify patients with sepsis who could benefit from anticoagulants.

Conclusions

In summary, our systematic review demonstrated limited but rapidly growing observational evidence on the use of TEG® and ROTEM® in trauma. Both methods may be useful for diagnosis of early trauma coagulopathies, specifically hypocoagulability, hypercoagulability, hyperfibrinolysis, and platelet dysfunction. They may also be used to direct blood and blood-product transfusion; effects on patient-important outcomes are uncertain. The existing literature helps clinicians to appreciate the potential impact of these novel methods on transfusion guidance and outcomes in trauma. However, adequately powered and methodologically sound RCTs will be required to prove positive effects on blood-product transfusion and patient-important outcomes.

Key messages

The literature on TEG® and ROTEM® in trauma is limited by the lack of randomized controlled trials and the moderate quality of observational studies.

TEG® and ROTEM® may be superior to routine screening coagulation tests to promptly diagnose early trauma coagulopathy, including hypocoagulability, hyperfibrinolysis, hypercoagulability, and platelet dysfunction.

Many TEG® and ROTEM® abnormalities predict the need for massive transfusion and predict death, but predictive performance is not consistently superior to routine screening coagulation tests.

Limited evidence from one observational study suggests that a ROTEM® -based transfusion algorithm reduces the amount of blood and blood products transfused.

TEG® and ROTEM® -based resuscitation for bleeding trauma patients is not associated with lower mortality in most observational studies, but the question requires evaluation in randomized trials.

Authors’ contributions

LTL participated in the study design, data collection, data analysis, drafting of the manuscript, and revision of the manuscript. BN participated in the study design and manuscript revision. AKS participated in data collection and data analysis. SR participated in study design and manuscript revision. NKJA participated in the study design, data collection, data analysis, drafting of the manuscript, and revision of the manuscript. All authors approved the final manuscript.

Additional file