Abstract
Background
Rotational thromboelastometry (ROTEM) is a blood test that measures hemostatic parameters to guide hemostatic therapy. ROTEM outputs can be cognitively challenging to interpret, which may limit adherence in trauma care. Our objective was to assess hemostatic therapy administration adherence to local ROTEM recommendations.
Study Design and Methods
We conducted a retrospective cohort study of trauma patients receiving ROTEM testing at a level 1 trauma center between January 1st 2017, and December 31st 2021. Adherence to local ROTEM best practices was determined by comparing the blood products patients received after a patient's first ROTEM test to those that should have been administered based on their ROTEM results. Multivariable logistic regression models were used to determine the association between clinical and patient covariates with ROTEM adherence and between ROTEM adherence and in‐hospital mortality.
Results
Only 46.6% (n = 208/446) of patients had complete adherence to ROTEM recommendations. Product‐specific adherence was lower when product initiation was recommended (vs. not) by ROTEM. A greater number of ROTEM abnormalities (odds ratio [OR]: 0.11, 95% confidence interval [CI]: 0.05–0.19) and a higher injury severity score (OR: 0.96, 95% CI: 0.94–0.98) reduced adherence. Adherence to ROTEM did not reduce in‐hospital mortality (OR: 0.71, 95% CI: 0.35–1.41). The number of ROTEM abnormalities was associated with in‐hospital mortality (OR: 3.07, 95% CI: 2.01–4.77).
Discussion
We found moderate adherence to admission ROTEM recommendations with lower adherence for more severely injured patients. The number of ROTEM abnormalities increased the odds of in‐hospital mortality. Quantifying adherence is valuable for understanding ROTEM implementations in trauma care.
1. INTRODUCTION
Major blood loss, or hemorrhage, is estimated to be responsible for half of the roughly 5 million injury deaths that occur globally. 1 A significant determinant of massive bleeding, and bleeding‐related mortality, is trauma‐induced coagulopathy. 2 Providing appropriate and timely administration of hemostatic therapies to critically injured patients is an essential aspect of trauma resuscitation. The current standard of care in a massive hemorrhage protocol (MHP) is to transfuse patients with blood components at a ratio of 1:1:2 or 1:1:1 of plasma, platelets, and red blood cells (RBCs). 3 , 4 , 5 A harmonized regional MHP has been recently developed in Ontario, Canada, to simplify training and increase uptake of evidence‐based interventions to improve patient outcomes. 6 However, the etiology of coagulopathy is multifactorial and heterogeneous, which means that the current one‐size‐fits‐all approach to resuscitation may not be suitable for every patient. 7
Viscoelastic testing (VET) provides a point‐of‐care determination of a patient's hemostatic capacity (their ability to form blood clots/stop bleeding) from their whole blood sample within minutes of the blood draw in the trauma bay. The outputs of VET can then be used to tailor hemostatic therapy to the patient's specific coagulopathy profile. 8 , 9 VET can detect coagulopathy earlier compared to standard coagulation testing, as the use of whole blood in VET circumvents the need to transport samples to hospital laboratories for centrifugation and enables testing to occur within the trauma bay. The proposed benefits of VET also include identification of hemostatic defects not detected by standard coagulation testing (i.e., fibrinolysis), 8 tailored administration of blood products, and potential reductions in blood products utilized. 10
Despite the potential benefits of VET and its recommendation by clinical guidelines, 10 it has not been shown to improve patient outcomes, including mortality. The ITACTIC trial, a multicenter, randomized controlled trial comparing outcomes in trauma patients who received empiric MHPs alongside either VET or conventional coagulation tests, failed to demonstrate a difference in overall mortality or serious adverse outcomes between study arms 9 While VET, such as rotational thromboelastometry (ROTEM), has been implemented at hospitals, its use remains low in North America 11 , 12 Assessment of adherence to trauma transfusion practices for deployed VET and evaluation of its impact on patient outcomes is important for understanding clinical utility.
The ROTEM program has been operational at St. Michael's Hospital (SMH) for over 10 years, with local ROTEM recommendations developed by local hemostasis experts based on contemporary international VET consensus recommendations. However, the adherence to local ROTEM recommendations and their impact on patient outcomes at SMH is unknown. Thus, our primary objective was to describe the hemostatic therapy administration adherence to local ROTEM recommendations based on the first‐admission ROTEM results. Our secondary objectives were to explore the associations of patient characteristics on ROTEM adherence and to determine the association between the number of abnormalities on first‐admission ROTEM and in‐hospital mortality.
2. STUDY DESIGN AND METHODS
2.1. Study design and data sources
This was a retrospective cohort study of trauma patients who received ROTEM testing at a level 1 trauma center between January 1, 2017, and December 31, 2021. The study institution, SMH, is a level 1 trauma center in Toronto, the 4th most populous city in North America, with >500 inpatient beds, >1500 trauma patients annually, and > 172,000 annual emergency department visits. As a downtown hospital, SMH provides trauma care to a diverse patient population, in addition to supporting smaller hospitals across Ontario through referral of severely injured patients by land and air ambulance.
Data sources included data from the SMH trauma registry, transfusion medicine laboratory information system (LIS), and electronic health records (EHRs). The SMH trauma registry is an established and validated database that has been leveraged in previous studies. 13 , 14 Maintenance of the database is a requirement for verification as a level 1 trauma center, and these data are further provided to the Ontario Ministry of Health. 15 In the use of patient data for research, any previously expressed wishes by patients regarding the use of their information were strictly adhered to. There was no linkage of individual patient data beyond SMH data holdings.
Information from the EHRs was accessed through chart extraction by GP, VH, and FA. After extraction, data was de‐identified and accessible only by the research team through a secure data access platform. To ensure no errors in the transfer of data when collecting data on the standardized case report form (CRF) tool, random checks of 5% of CRF against medical charts were conducted throughout the study. This study received Research Ethics Board (REB) approval from St. Michael's Hospital (REB 22‐234 and 24‐052).
2.2. Study population
The study population consisted of SMH trauma patients (Tier 1 and 2) with dates of admission between January 1, 2017, and December 31, 2021. We included those who had ROTEM testing done while in the trauma bay and were transfused at least one unit of RBCs. The receipt of at least one unit of RBCs was used as a surrogate for suspected hemorrhage by the treating trauma team. The exclusion criteria were non‐trauma patients (e.g., not included in the trauma registry), trauma patients who did not have ROTEM testing while in the trauma bay, those who died prior to arrival at SMH or prior to ROTEM testing, and patients objecting to blood transfusions.
2.3. Institutional massive hemorrhage protocol implementation
MHP activation at SMH is a physician judgment decision, in our case from the trauma team leader (TTL). Institutional guidelines suggest optimal MHP triggering criteria that include high shock index, Assessment of Blood Consumption score greater than or equal to two, high resuscitation intensity (substantial and rapid blood loss not responsive to greater than or equal to 4 units of fluid within the first 30 min), and anticipated massive and rapid blood loss requiring transfusion of greater than or equal to 6 units of RBCs. Of note, whole blood did not have Health Canada approval during the study period and was not used in the MHP.
Once an MHP is called, blood products are delivered from the Transfusion Medicine department in a series of cooler shipments. The first shipment contains 4 units of packed RBCs. The second shipment contains 4 units of RBCs, 4 units of frozen plasma (FP), and 1 unit of platelets. The third shipment contains 4 units of RBCs, 2 units of FP, and 4 g of fibrinogen concentrate. The fourth shipment contains 4 units of RBCs and 2 units of FP. The Transfusion Medicine department aims to deliver each shipment in sequential 30‐minute periods, anticipating that the prior shipment will be transfused in the interim. After the fourth shipment, Transfusion Medicine prepares 4 units of RBCs and 2 units of FP every 30 minutes until the MHP is terminated. Empiric management is performed at a fixed 2(RBCs):1(plasma):1(platelets) ratio. The TTL can deviate from the empiric MHP schedule and call for other products, including with ROTEM guidance.
2.4. Institutional ROTEM implementation
ROTEM has been implemented and widely used at SMH for over 10 years as an augment to the standard MHP. ROTEM is ordered in all trauma patients as part of initial blood work drawn upon arrival in the trauma bay as the standard of care. During the study period, substantial educational efforts, as well as the involvement of senior clinical champions, distribution of pocket‐size ROTEM guides, and incorporation of ROTEM into policies have been undertaken at SMH in attempts to improve utilization of ROTEM outputs. While the testing is performed in the central hospital laboratory, the ROTEM readout (temogram and numerical results with reference values) is displayed in real time on a large screen in the trauma bay to promote awareness among the trauma team. The EXTEM and FIBTEM temograms typically begin to display in the trauma bay within 10 min of the blood being drawn or 5 min of the arrival of the blood to the hospital laboratory. Numerical results and reference values are then displayed along the temogram as they become available. Final ROTEM results are then posted on EMR. Local ROTEM value thresholds and treatment recommendations at SMH are outlined in Appendix S1. As our study was retrospective, we used the first ROTEM results reported in patients' charts.
Two ROTEM assays are routinely performed for trauma patients: EXTEM (assessment of the extrinsic system of coagulation) and FIBTEM (assessment of fibrinogen function). The EXTEM assay captures platelet quantitative or qualitative dysfunction and fibrinolysis and includes variables of clotting time (CT), amplitude 10 min after CT (A10), maximum clot firmness (MCF), and maximum lysis. FIBTEM variables include A10 and MCF. Temograms, in conjunction with the results of the complete blood count, international normalized ratio, activated partial thromboplastin time, and Claus fibrinogen assay for all patients, are interpreted independently by the medical director of the coagulation laboratory and reported in the patient's chart.
2.5. Assessment of hemostatic therapy administration
To determine the hemostatic therapy administered to each patient, we conducted a chart review using EHR data, pulling transfusion records, TTL notes, and discharge documentations. In our definition of hemostatic therapy administration, we included the following: plasma, platelets, fibrinogen concentrate, and additional doses of tranexamic acid (TXA, defined as ≥2 g). These hemostatic therapies were selected as their administration is informed by ROTEM outputs. To focus on the impact of ROTEM in the initial resuscitation window, we included only data on hemostatic therapy administered in the first 4 h at SMH. In extracting the data, we included only hemostatic therapies with evidence of administration (i.e., time stamped administration documentation from nursing).
2.6. Outcomes
Our primary outcome was complete adherence to ROTEM (yes/no), which was assessed at the patient level and was determined by comparing the patients' hemostatic therapy administration to the products that should have been received based on their initial ROTEM readout (i.e., first‐admission ROTEM) and local ROTEM treatment recommendations (Appendix S1). ROTEM adherence was defined as the correct administration of all 4 hemostatic therapies according to local ROTEM recommendations (within 4‐hours); adherence also included non‐administration of hemostatic therapy if not recommended. We also examined adherence to individual hemostatic therapies (TXA, plasma, platelets, fibrinogen concentrate). Given the retrospective nature of the study, we cannot be certain whether providers reviewed in real time ROTEM results, were able to interpret them, or made transfusion decisions based on ROTEM output. Thus, “adherent patients” refer to scenarios whereby physicians followed all ROTEM recommendations, knowingly or unknowingly.
The secondary outcome was in‐hospital mortality (yes/no), defined as death during the initial injury‐related admission, captured in the discharge disposition statement of the SMH trauma registry.
2.7. Exposures and covariates
The main exposure for both multivariable models was ROTEM adherence, as specified above. Additional covariates included patient demographics, injury characteristics, and details on available hospital environment/healthcare resources. The patient demographics were age (years) and sex (female vs. male). The injury characteristics were time and date of injury, mechanism of injury (categorized as blunt or penetrating), vital signs on arrival to the trauma bay (heart rate, systolic blood pressure, Glasgow Coma Scale), injury severity score (ISS), arrival source (from scene vs. referring hospital), method of arrival (air ambulance, land ambulance, and other), and level of trauma activation (tier 1 or 2). Additional hospital environment covariates included time of arrival at SMH (categorized as 0:00–7:59, 8:00:15:59, 16:00–23:59), date of arrival at SMH (categorized as weekday vs. weekend), initiation of MHP (yes/no), and total hospital length of stay (days).
2.8. Statistical analysis
Descriptive statistics on patient demographics, injury characteristics, and hospital resources were used to summarize the included cohort (all trauma patients in study window), stratified by ROTEM adherence (adherence vs. nonadherence). Differences between the ROTEM adherence groups were tested using standardized mean difference (SMD), with a SMD >10% indicating meaningful differences between groups. 16
A multivariable logistic regression (LR) model was used to identify the association between clinical and patient factors (sex, age, type, ISS, scene vs. transfer patient, and number of ROTEM abnormalities) and ROTEM adherence. A priori selection of these variables reflected maximized inclusion of the clinical covariates without overparameterizing the model. A second multivariable LR model was created to determine whether ROTEM adherence was associated with in‐hospital mortality, when accounting for the aforementioned clinical and patient covariates. In both models, adequate fit was assessed by using the Omnibus test, Hosmer‐Lemeshow test (comparing observed cases vs. predicted cases), and a receiver operating characteristic (ROC) curve, defining acceptable discrimination as an area under the curve (AUC) ≥ 0.7. Multicollinearity was assessed by calculating the Variance Inflation Factor (VIF) of each variable. There were no variables that had a VIF exceeding .5, which was our predefined threshold for collinearity. LR outputs were presented as odds ratios (ORs) with 95% confidence intervals (CI).
Statistical analyses were performed in R (version 4.3.3) and Python (version 3.7). All tests were two‐sided, with p‐values less than 0.05 indicating statistical significance.
3. RESULTS
3.1. Cohort characteristics
In total, throughout the study period, there were 4750 trauma patients receiving a ROTEM test in the trauma bay, and 446 had at least one blood transfusion in the trauma bay. The median age of the cohort was 39 years (IQR: 28–58 years), of whom 76% were male (Table 1). Included admissions mostly occurred between 16:00 and 23:59 (42%) and on weekdays (68%). Most patients arrived directly from the scene (68%) and by land ambulance (88%). Blunt injuries accounted for the predominant mechanism (66%), with 34% as penetrating injuries. The median ISS score was 25 (IQR: 14–34) and 55% had MHPs called (Table 1).
TABLE 1.
Cohort characteristics stratified by adherence to local rotational thromboelastometry recommendations (N = 446).
| Total cohort (N = 446) | Complete adherence (N = 208) | Nonadherence (N = 236) | |
|---|---|---|---|
| Sex, n (%) | |||
| Female | 109 (24.4) | 53 (25.5) | 56 (23.7) |
| Male | 337 (75.6) | 155 (74.5) | 180 (76.3) |
| Age (years) | |||
| Median (IQR) | 38.5 (28–58) | 41 (29–60) | 38 (25–56) |
| Time of admission, n (%) | |||
| 0:00–7:59 | 143 (32.1) | 68 (32.7) | 75 (31.8) |
| 8:00–15:59 | 116 (26.0) | 44 (21.2) | 72 (30.5) |
| 16:00–23:59 | 187 (41.9) | 96 (46.2) | 89 (37.7) |
| Date of admission, n (%) | |||
| Weekday | 302 (67.7) | 139 (66.8) | 162 (68.6) |
| Weekend | 144 (32.3) | 69 (33.2) | 74 (31.4) |
| Admission source, n (%) | |||
| Referring hospital | 145 (32.3) | 51 (24.5) | 94 (39.8) |
| Scene | 3031 (67.4) | 157 (75.5) | 142 (60.2) |
| Transportation on arrival | |||
| Air ambulance | 20 (4.5) | 7 (3.4) | 13 (5.8) |
| Land ambulance | 400 (89.7) | 191 (93.6) | 207 (91.6) |
| Other | 12 (2.7) | 6 (2.9) | 6 (2.7) |
| Missing | 14 (3.1) | 4 (1.9) | 10 (4.2) |
| Injury severity score (ISS) | |||
| Mean (SD) | 25.5 (15.2) | 20.3 (12.7) | 30.4 (15.6) |
| Mechanism of injury | |||
| Blunt | 293 (65.7) | 124 (59.6) | 169 (71.6) |
| Penetrating | 153 (34.3) | 84 (40.4) | 67 (28.4) |
| Heart rate | |||
| Mean (SD) | 103.0 (27.6) | 100.8 (26.4) | 105.2 (28.7) |
| Systolic blood pressure | |||
| Mean (SD) | 107.1 (33.3) | 114.0 (31.0) | 100.9 (34.3) |
| Glasgow coma scale, n (%) | |||
| <8 | 52 (14.3) | 14 (7.4) | 38 (22.0) |
| 8–12 | 42 (11.5) | 16 (8.5) | 26 (15.0) |
| 13–15 | 270 (74.2) | 159 (84.1) | 109 (63.0) |
| Trauma tier, n (%) | |||
| Tier 1 | 226 (50.7) | 81 (38.9) | 144 (61.0) |
| Tier 2 | 220 (49.3) | 127 (61.1) | 92 (39.0) |
| MHP called, n (%) | |||
| Yes | 245 (54.9) | 70 (34.3) | 173 (73.3) |
| No | 197 (44.2) | 134 (65.7) | 63 (26.7) |
| Missing | 4 (0.9) | 4 (1.9) | 0 (0.0) |
| Post‐ED disposition, n (%) | |||
| Discharged home | 2 (0.4) | 2 (1.0) | 0 (0.0) |
| Operating room | 171 (38.3) | 60 (28.8) | 111 (47.0) |
| Special care unit | 200 (44.8) | 91 (43.8) | 109 (46.2) |
| Ward | 73 (16.4) | 55 (26.4) | 16 (6.8) |
| Length of stay (days) | |||
| Median (IQR) | 14 (5–35) | 17 (6–43) | 10 (4–25) |
| Missing | 2 (0.0) | 2 (1.0) | 0 (0.0) |
| Discharge status, n (%) | |||
| Alive | 361 (80.9) | 190 (91.3) | 169 (71.6) |
| Dead | 85 (19.1) | 18 (8.7) | 67 (28.4) |
3.2. ROTEM adherence
In total, 47% (n = 208/446) had complete adherence to ROTEM recommendations. As shown in Table 1, adherent patients were of lower injury severity (median ISS: 20 vs. 30), more commonly had penetrating mechanisms of injury (40% vs. 28%), arrived from the scene (76% vs. 60%), and less commonly had MHPs activated (34% vs. 73%) than those whose providers did not adhere completely.
Complete ROTEM adherence decreased with increasing number of ROTEM abnormalities (Figure 1), with adherence only in 9.6% (n = 12/125) of patients with one or more ROTEM abnormalities. Across the study cohort, the most recommended hemostatic therapy was plasma (26%, n = 116/446) and fibrinogen concentrate (8%, n = 34/446). Additional tranexamic acid and platelets were much less often recommended by ROTEM; only 3% (n = 12/446) and 0.45% (n = 2/446) of patients had ROTEM indicate administration of these products, respectively.
FIGURE 1.
Number of rotational thromboelastometry (ROTEM) abnormalities per patient on first ROTEM read across cohort (N = 446).
With respect to individual blood components, overall adherence with ROTEM recommendations was 68% (n = 300/446) for plasma, 90% (n = 403/446) for tranexamic acid, 73% (n = 324/446) for platelets, and 83% (n = 371/446) for fibrinogen concentrate (Table 2 and Figure 2). Across all products, ROTEM adherence was highest when ROTEM did not indicate the need for administration. For each product, adherence to negative action recommendations (i.e., correctly not administering that product) was 92% (n = 402/434) for tranexamic acid, 67% (n = 220/328) for plasma, 73% (n = 323/442) for platelets, and 86% (n = 353/412) for fibrinogen concentrate (Table 2). In contrast, when a product was indicated by ROTEM, adherence (i.e., correctly administering that product) was much lower, with 8% (n = 1/12) for tranexamic acid, 69% (n = 80/116) for plasma, 50% (n = 1/2) for platelets, and 53% (n = 18/34) for fibrinogen concentrate (Table 2).
TABLE 2.
Rotational thromboelastometry (ROTEM) values and adherence to local ROTEM recommendations (N = 446).
| Overall cohort (N = 446) | Cohort by adherence status | |||
|---|---|---|---|---|
| Adherence (N = 208) | Nonadherence (N = 236) | Standardized mean difference | ||
| ROTEM EXTEM ML, mean (SD) | 7.75 (16.27) | 6.14 (3.65) | 9.17 (22.04) | 0.192 |
| ROTEM EXTEM CT, mean (SD) | 85.92 (159.58) | 66.34 (14.02) | 103.37 (217.71) | 0.240 |
| ROTEM EXTEM A10, mean (SD) | 51.13 (11.79) | 56.47 (7.39) | 46.43 (12.96) | 0.952 |
| ROTEM EXTEM MCF, mean (SD) | 59.45 (10.15) | 63.66 (6.15) | 55.73 (11.51) | 0.859 |
| ROTEM FIBTEM A10, mean (SD) | 13.23 (7.13) | 15.94 (7.25) | 10.80 (6.13) | 0.765 |
| ROTEM FIBTEM MCF, mean (SD) | 14.46 (8.04) | 17.08 (7.89) | 12.13 (7.47) | 0.643 |
| Number ROTEM Recommendations, mean (SD) | 0.37 (0.65) | 0.06 (0.25) | 0.64 (0.76) | 1.041 |
| Tranexamic acid, n (%) | ||||
| Indicated and given | 1 (0.2) | 0 (0.0) | 1 (0.4) | 0.092 |
| Indicated and not given | 11 (2.5) | 0 (0.0) | 11 (4.7) | 0.313 |
| Not indicated and given | 32 (7.2) | 0 (0.0) | 32 (13.6) | 0.560 |
| Not indicated and not given | 402 (90.1) | 208 (100.0) | 192 (81.4) | 0.677 |
| Fresh frozen plasma, n (%) | ||||
| Indicated and given | 80 (17.9) | 11 (5.3) | 69 (29.2) | 0.668 |
| Indicated and not given | 36 (8.1) | 0 (0.0) | 36 (15.3) | 0.600 |
| Not indicated and given | 108 (24.2) | 0 (0.0) | 108 (45.8) | 1.299 |
| Not indicated and not given | 220 (49.3) | 197 (94.7) | 23 (9.7) | 3.234 |
| Platelet, n (%) | ||||
| Indicated and given | 1 (0.2) | 0 (0.0) | 1 (0.4) | 0.092 |
| Indicated and not given | 1 (0.2) | 0 (0.0) | 1 (0.4) | 0.092 |
| Not indicated and given | 119 (26.7) | 0 (0.0) | 119 (50.4) | 1.426 |
| Not indicated and not given | 323 (72.4) | 208 (100.0) | 115 (48.7) | 1.451 |
| Fibrinogen concentrate, n (%) | ||||
| Indicated and given | 18 (4.0) | 1 (0.5) | 17 (7.2) | 0.355 |
| Indicated and not given | 16 (3.6) | 0 (0.0) | 16 (6.8) | 0.381 |
| Not indicated and given | 59 (13.2) | 0 (0.0) | 59 (25.0) | 0.816 |
| Not indicated and not given | 353 (79.1) | 207 (99.5) | 144 (61.0) | 1.105 |
FIGURE 2.
Distribution of rotational thromboelastometry (ROTEM) outputs across cohort, stratified by adherence.
3.3. Predictors of Complete ROTEM Adherence
Results of the multivariable LR model predicting ROTEM adherence (AUC = 0.81) are displayed in Table 3. The number of ROTEM abnormalities (OR: 0.11, 95% CI: 0.05–0.19) and ISS (OR: 0.96, 95% CI: 0.94–0.98) were significant predictors of nonadherence, while the source of arrival (scene vs. referring hospital) was associated with increased adherence (OR: 1.74, 95% CI: 1.05–2.91). The other modeled covariates (sex, age, and injury type) were not predictors of ROTEM adherence.
TABLE 3.
Predictors of rotational thromboelastometry (ROTEM) adherence (N = 446).
| Predictor | Adjusted odd ratio (95% confidence interval) | Test statistic | p‐value |
|---|---|---|---|
| Omnibus likelihood ratio (x2 (df), p‐value) | 142.86 (6) | <0.01 | |
| Sex (male vs. female) | 0.93 (0.55, 1.59) | 0.07 | 0.80 |
| Injury type (penetrating vs. blunt) | 0.66 (0.36, 1.19) | −0.42 | 0. 17 |
| Age (years) | 1.00 (0.99, 1.02) | 0.00 | 0. 59 |
| Injury severity score | 0.96 (0.94, 0.98) | −0.04 | <0.01 |
| Source of arrival (scene vs. referring hospital) | 1.74 (1.05, 2.91) | 0.56 | 0.03 |
| Number of ROTEM abnormalities | 0.11 (0.05, 0.19) | −2.25 | <0.01 |
3.4. Association between ROTEM adherence and in‐hospital mortality
The multivariable LR model predicting in‐hospital mortality demonstrated strong fit (AUC: 0.86) with no evidence of multicollinearity (Table 4). After adjusting for the patient and clinical covariates, complete adherence to ROTEM was not significantly associated with increased odds of in‐hospital mortality (OR: 0.71, 95% CI: 0.35–1.41). Regarding other covariates, the number of ROTEM abnormalities (OR: 3.07, 95% CI: 2.01–4.77), ISS (OR: 1.05, 95% CI: 1.03–1.08), and age (OR: 1.03, 95% CI: 1.01–1.05) were also associated with increased odds of in‐hospital mortality. The other covariates (source of arrival, injury type, and sex) were not associated with increased odds of in‐hospital mortality.
TABLE 4.
Multivariable logistic regression of the association between rotational thromboelastometry (ROTEM) adherence and in‐hospital mortality (N = 446).
| Predictor | Adjusted odd ratio (95% Confidence interval) | Test statistic | p‐value |
|---|---|---|---|
| Omnibus likelihood ratio (x2 (df), p‐value) | 124.7 (7) | <0.01 | |
| Complete ROTEM adherence (yes vs. no) | 0.71 (0.35, 1.41) | −0.34 | 0.33 |
| Sex (male vs. female) | 0.96 (0.51, 1.86) | −0.04 | 0.90 |
| Injury type (penetrating vs. blunt) | 0.52 (0.18, 1.32) | −0.66 | 0.19 |
| Age (years) | 1.03 (1.01, 1.05) | 0.03 | <0.01 |
| Injury severity score | 1.05 (1.03, 1.08) | 0.05 | <0.01 |
| Source of arrival (scene vs. referring hospital) | 0.88 (0.48, 1.64) | −0.13 | 0.69 |
| Number of ROTEM abnormalities | 3.07 (2.01, 4.77) | 1.12 | <0.01 |
4. DISCUSSION
We evaluated the adherence of trauma teams to their local ROTEM recommendations at a level one trauma center in Ontario, Canada. Adherence to ROTEM recommendations was low, with only half of the cohort having adhered to ROTEM recommendations, most often in patients with normal ROTEM values. Nonadherence tended to occur among patients with more complex, higher severity injuries. Adherence to ROTEM recommendations did not significantly reduce in‐hospital mortality; however, the number of ROTEM abnormalities was a significant predictor of in‐hospital mortality, suggesting its utility in identifying coagulopathy and that there may be cognitive overload or other barriers to adherence in severely injured patients. In our study, awareness of ROTEM recommendations could have changed transfusion management in 53% of patients. Taken together, these findings provide evidence for the need to enhance ROTEM interpretability and other barriers to adherence.
4.1. ROTEM adherence is low
Importantly, we demonstrated low levels of ROTEM adherence, with adherence to local recommendations only occurring for under half of the trauma patients receiving blood products in the trauma bay. While ROTEM testing has been implemented at SMH for over 10 years, its impact on clinical decision‐making has remained limited. Low adherence could be due to several factors. For example, ROTEM results may be perceived as difficult to interpret as they contain several complex parameters that clinicians need to parse through in an already cognitively challenging environment. 17 , 18 Improving interpretability may increase integration into clinical care by improving interrater reliability and reducing practice variations that may compromise outcomes in a patient population already at high risk of morbidity and mortality. 19 Other factors that limit ROTEM adherence may include latency in receiving ROTEM results, poor team communication of results, or lack of clarity on when to deviate from MHP and follow ROTEM, leading to empiric management and physician skepticism in the value of ROTEM‐guided transfusion. Positive nonadherence (i.e., giving products when not indicated), such as in the case of plasma or fibrinogen concentrate, may be due in part to an availability heuristic whereby TTLs decide to transfuse products as they see them arriving in coolers at set time intervals as per the MHP. Our results suggest that traditional educational interventions and clinical champions may be insufficient to promote adherence to VET in trauma care. One avenue may be decision support in computerized physician order entry. Future research should also focus on the human‐technology interface of ROTEM and identify ways in which we can improve the interpretability and clinical utility of VET readouts.
European centers tend to have higher adoption of ROTEM compared to those in North America, which could be due to differences in the strength of recommendations in trauma resuscitation guidelines to follow a goal‐directed approach guided by VET. 10 , 20 , 21 In contrast, empiric approaches to MHP predominate in North America. A 2016 survey of 125 American Association for the Surgery of Trauma institutions found that while 98% of institutions had MHPs in place, only 9% regularly used thromboelastography (TEG) or ROTEM within their MHPs. 11 Furthermore, there was considerable heterogeneity between surveyed institutions in how empiric MHPs were structured, with the number of coolers and units of blood products within coolers being highly variable. 11 In Canada, 65% of level 1 and level 2 trauma centers did not have access to VET. 12 Our findings show that even when a VET is in place, transfusion practices commonly deviate from VET recommendations. Future research characterizing VET adherence internationally is warranted to better understand the interplay between VET technology and the transfusion paradigm that it is implemented within.
4.2. ROTEM abnormalities predict mortality
Interestingly, we show that the number of ROTEM abnormalities was a predictor of in‐hospital mortality (3‐fold increase in the odds of in‐hospital mortality). In the existing literature, however, the effect of ROTEM adherence on mortality is unclear. 22 , 23 , 24 , 25 Inconsistent results are likely due, at least in part, to differences in study definitions of ROTEM adherence. For example, a recent Canadian study found that ROTEM‐guided hemostatic therapy administration was associated with a nonsignificantly decreased mortality rate; however, this study did not evaluate ROTEM adherence (defining ROTEM guidance as all instances in which ROTEM testing was “ran” vs. “followed”) and used a small sample size of 35 patients. 26 The Implementing Treatment Algorithms for the Correction of Trauma Induced Coagulopathy (ITACTIC) trial compared the use of a traditional MHP with either VET or conventional coagulopathy testing in 390 trauma patients and failed to demonstrate that MHP augmented with VET improved mortality or reduced the use of massive transfusion in the first 24 hours. However, it did not differentiate adherence levels within the VET arm, nor delineate the impact of ROTEM from MHPs. 9 A secondary, subgroup analysis of the ITACTIC trial found that while the VET group received a higher rate of goal‐directed therapy and coagulopathy correction versus the conventional group, these rates were low in both groups. Moreover, there was a considerable time delay in transfusing goal‐directed blood products in both groups. While our findings on mortality should be considered exploratory due to a relatively small sample size, both our work and the ITACTIC secondary analysis demonstrate the need for future studies to understand and mitigate the drivers of nonadherence that perhaps delay goal‐directed therapy and lead to poorer outcomes. Such work, like our own, should consider both overall and component‐specific ROTEM adherence, as their drivers may differ.
4.3. Strengths and limitations
These findings provide new insights on the implementation of ROTEM testing and its impact on patient outcomes. Compared to previous studies, this cohort is larger in size, and our evaluation of adherence considers how physician actions align with ROTEM output—both overall and component‐specific ROTEM recommendations. Our work also has some important limitations. We use EHR data, a source that can contain inaccuracies and incompleteness. However, for trauma care in Ontario, maintenance of the hospital‐based trauma registry is a requirement for verification as a lead trauma hospital, and thus, SMH is obligated to ensure completeness and quality of data. As a single‐center study, our findings do not directly generalize to other trauma sites given differences in hospital dynamics and patient populations. Our limited sample size did not allow us to account for all possible confounders that may influence in‐hospital mortality without overparameterizing the models. Moreover, the data used do not contain provider characteristics which may confound the relationship between ROTEM adherence and mortality (e.g., discipline of TTL, years of experience, sex, and age); further work could extend on our findings by developing or using data with provider characteristics and developing mixed models, clustering by providers. 27 , 28 We also presume that all clinicians saw the ROTEM output and attempted to interpret and act on the results, regardless of MHP activation, though this cannot be confirmed. We also did not capture variables reflective of institutional processes, such as the time between when the first ROTEM test was reported and when the patient arrived in the trauma bay or the time until the first blood products were transfused. An analysis of these times and other process factors would be an important direction for future work, especially across multiple institutions. Lastly, in our models, we did not account for practice changes (e.g., COVID‐19 pandemic) or the release of new information (e.g., ITACTIC trial 9 ) within our study time window. Importantly, local ROTEM recommendations did not change during the study period.
5. CONCLUSION
We found low levels of overall adherence to local ROTEM recommendations, particularly among patients with complex, high‐severity injuries. Adherence to ROTEM recommendations that encouraged hemostatic therapy was much lower than adherence to ROTEM recommendations that did not indicate administration, across all hemostatic therapies. The number of ROTEM abnormalities increased the odds of in‐hospital mortality. These findings emphasize the need to improve the interpretability of ROTEM recommendations and determine whether improved interpretability enhances patient outcomes, particularly in sicker patients. Further research should explore methods to foster complete adherence for ROTEM interpretation as well as evaluate the utility of ROTEM adherence across multiple centers.
FUNDING INFORMATION
There was no funding for this project.
CONFLICT OF INTEREST STATEMENT
The authors have no relevant conflicts of interest to disclose.
Supporting information
Data S1.
ACKNOWLEDGMENTS
The analyses, conclusions, opinions and statements expressed herein are solely those of the authors and do not reflect those of the funding or data sources; no endorsement is intended or should be inferred.
Harish V, Postill G, Al‐Haimus F, McGowan M, Pavenski K, Beckett A, et al. Adherence to local rotational thromboelastometry recommendations in the care of trauma patients: A retrospective cohort study. Transfusion. 2025;65(9):1716–1727. 10.1111/trf.18349
DATA AVAILABILITY STATEMENT
The dataset from this study is held securely in Unity Health remote servers. Due to patient privacy, it is not publicly available.
REFERENCES
- 1. Gruen RL, Brohi K, Schreiber M, Balogh ZJ, Pitt V, Narayan M, et al. Haemorrhage control in severely injured patients. The Lancet. 2012;380(9847):1099–1108. 10.1016/S0140-6736(12)61224-0 [DOI] [PubMed] [Google Scholar]
- 2. Moore EE, Moore HB, Kornblith LZ, Neal MD, Hoffman M, Mutch NJ, et al. Author Correction: Trauma‐induced coagulopathy. Nat Rev Dis Primer. 2022;8(1):25. 10.1038/s41572-022-00360-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Young PP, Cotton BA, Goodnough LT. Massive transfusion protocols for patients with substantial hemorrhage. Transfus Med Rev. 2011;25(4):293–303. 10.1016/j.tmrv.2011.04.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Miller TE. New evidence in trauma resuscitation ‐ is 1:1:1 the answer? Perioper Med. 2013;2(1):13. 10.1186/2047-0525-2-13 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Holcomb JB, Tilley BC, Baraniuk S, Fox EE, Wade CE, Podbielski JM, et al. Transfusion of Plasma, Platelets, and Red Blood Cells in a 1:1:1 vs a 1:1:2 Ratio and Mortality in Patients With Severe Trauma: The PROPPR Randomized Clinical Trial. JAMA. 2015;313(5):471. 10.1001/jama.2015.12 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Callum JL, Yeh CH, Petrosoniak A, McVey MJ, Cope S, Thompson T, et al. A regional massive hemorrhage protocol developed through a modified Delphi technique. CMAJ Open. 2019;7(3):E546–E561. 10.9778/cmajo.20190042 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Goobie SM, Shander A. One size does not fit all in treating massive hemorrhage. Anesth Analg. 2020;131(2):480–482. 10.1213/ANE.0000000000004690 [DOI] [PubMed] [Google Scholar]
- 8. Whiting D, DiNardo JA. TEG and ROTEM: technology and clinical applications. Am J Hematol. 2014;89(2):228–232. 10.1002/ajh.23599 [DOI] [PubMed] [Google Scholar]
- 9. Baksaas‐Aasen K, Gall LS, Stensballe J, Juffermans NP, Curry N, Maegele M, et al. Viscoelastic haemostatic assay augmented protocols for major trauma haemorrhage (ITACTIC): a randomized, controlled trial. Intensive Care Med. 2021;47(1):49–59. 10.1007/s00134-020-06266-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Rossaint R, Bouillon B, Cerny V, Coats TJ, Duranteau J, Fernández‐Mondéjar E, et al. The European guideline on management of major bleeding and coagulopathy following trauma: fourth edition. Crit Care. 2016;20(1):100. 10.1186/s13054-016-1265-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Etchill E, Sperry J, Zuckerbraun B, Alarcon L, Brown J, Schuster K, et al. The confusion continues: results from an American Association for the Surgery of Trauma survey on massive transfusion practices among United States trauma centers. Transfusion (Paris). 2016;56(10):2478–2486. 10.1111/trf.13755 [DOI] [PubMed] [Google Scholar]
- 12. Tastad K, Sholzberg M, Beckett A, McGowan M, Nolan B. Viscoelastic testing at Canadian trauma centres: A cross‐sectional survey of current practices, barriers, and facilitators. Available from: https://www.traumacanada.org/wp-content/uploads/2024/04/TAC-Final-Program1.pdf.
- 13. Nolan B, Tien H, Sawadsky B, Rizoli S, McFarlan A, Phillips A, et al. Comparison of helicopter emergency medical services transport types and delays on patient outcomes at two level i trauma centers. Prehosp Emerg Care. 2017;21(3):327–333. 10.1080/10903127.2016.1263371 [DOI] [PubMed] [Google Scholar]
- 14. Nolan B, Ackery A, Nathens A, Sawadsky B, Tien H. Canceled to be called back: a retrospective cohort study of canceled helicopter emergency medical service scene calls that are later transferred to a trauma center. Air Med J. 2018;37(2):108–114. 10.1016/j.amj.2017.11.015 [DOI] [PubMed] [Google Scholar]
- 15. Trauma Association of Canada . Trauma System Accreditation Guidelines. Canada: Trauma Association of Canada; 2011. Fourth Revision. Available from: https://www.traumacanada.org/wp-content/uploads/2019/05/Accreditation_Guidelines_2011.pdf [Google Scholar]
- 16. Austin PC. Using the standardized difference to compare the prevalence of a binary variable between two groups in observational research. Commun Stat ‐ Simul Comput. 2009;38(6):1228–1234. 10.1080/03610910902859574 [DOI] [Google Scholar]
- 17. Szulewski A, Howes D, Van Merriënboer JJG, Sweller J. From theory to practice: the application of cognitive load theory to the practice of medicine. Acad Med. 2021;96(1):24–30. 10.1097/ACM.0000000000003524 [DOI] [PubMed] [Google Scholar]
- 18. Hicks C, Petrosoniak A. The human factor: optimizing trauma team performance in dynamic clinical environments. Emerg Med Clin North Am. 2018;36(1):1–17. 10.1016/j.emc.2017.08.003 [DOI] [PubMed] [Google Scholar]
- 19. Chin V, Cope S, Yeh CH, Thompson T, Nascimento B, Pavenski K, et al. Massive hemorrhage protocol survey: Marked variability and absent in one‐third of hospitals in Ontario, Canada. Injury. 2019;50(1):46–53. 10.1016/j.injury.2018.11.026 [DOI] [PubMed] [Google Scholar]
- 20. Curry NS, Davenport R, Pavord S, Mallett SV, Kitchen D, Klein AA, et al. The use of viscoelastic haemostatic assays in the management of major bleeding: A British Society for Haematology Guideline. Br J Haematol. 2018;182(6):789–806. 10.1111/bjh.15524 [DOI] [PubMed] [Google Scholar]
- 21. Stanworth SJ, Dowling K, Curry N, Doughty H, Hunt BJ, Fraser L, et al. Haematological management of major haemorrhage: a British Society for Haematology Guideline. Br J Haematol. 2022;198(4):654–667. 10.1111/bjh.18275 [DOI] [PubMed] [Google Scholar]
- 22. Veigas PV, Callum J, Rizoli S, Nascimento B, Da Luz LT. A systematic review on the rotational thrombelastometry (ROTEM®) values for the diagnosis of coagulopathy, prediction and guidance of blood transfusion and prediction of mortality in trauma patients. Scand J Trauma Resusc Emerg Med. 2016;24(1):114. 10.1186/s13049-016-0308-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Da Luz LT, Nascimento B, Shankarakutty AK, Rizoli S, Adhikari NK. Effect of thromboelastography (TEG®) and rotational thromboelastometry (ROTEM®) on diagnosis of coagulopathy, transfusion guidance and mortality in trauma: descriptive systematic review. Crit Care. 2014;18(5):518. 10.1186/s13054-014-0518-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Bugaev N, Como JJ, Golani G, Freeman JJ, Sawhney JS, Vatsaas CJ, et al. Thromboelastography and rotational thromboelastometry in bleeding patients with coagulopathy: Practice management guideline from the Eastern Association for the Surgery of Trauma. J Trauma Acute Care Surg. 2020;89(6):999–1017. 10.1097/TA.0000000000002944 [DOI] [PubMed] [Google Scholar]
- 25. Casaer MP, De Jong A, Perner A. Clinical validation of precision medicine protocols: the last mile is the longest. Intensive Care Med. 2021;47(1):80–82. 10.1007/s00134-020-06301-1 [DOI] [PubMed] [Google Scholar]
- 26. Salehi M, Bola R, De Jong N, Shih AW, Garraway N, Dawe P. Guided blood transfusion of trauma patients with rotational thromboelastometry: a single‐center cohort study. World J Emerg Surg. 2023;18(1):40. 10.1186/s13017-023-00508-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Chen Y, Patel MB, McNaughton CD, Malin BA. Interaction patterns of trauma providers are associated with length of stay. J Am Med Inform Assoc. 2018;25(7):790–799. 10.1093/jamia/ocy009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Kelley KC, Alers A, Bendas C, Thomas PG, Cipolla J, Hoey BA, et al. Emergency Trauma Providers as Equal Partners: From ‘Proof of Concept’ to ‘Outcome Parity’. Am Surg. 2019;85(9):961–964. 10.1177/000313481908500936 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1.
Data Availability Statement
The dataset from this study is held securely in Unity Health remote servers. Due to patient privacy, it is not publicly available.