Background

Trauma is a major cause of mortality, with traumatic injures estimated to result in over five million deaths annually worldwide [1]. Severe hemorrhage is responsible for approximately 55% of in-hospital deaths after trauma [2, 3]. Therefore, hypotension and hypoperfusion should be promptly addressed. [4,5,6] Management of hemorrhagic shock calls for expeditious hemorrhage control, volume resuscitation with crystalloids and blood products, and treatment of coagulopathy. Accordingly, it has been demonstrated that pre-hospital resuscitation with intravenous crystalloid improves survival in trauma patients [6]. Despite the benefits of volume replenishment in trauma previous studies showed that patients with uncontrolled sources of bleeding would likely benefit from low-volume resuscitation [7,8,9,10]. However, the ideal volume of crystalloid infusion in hemorrhagic shock is a matter of debate. A recent multicenter randomized controlled trial was unable to show lower mortality with restrictive use of crystalloid fluid and preferential use of blood products in hemorrhagic shock trauma patients [11].

Injudicious use of crystalloid during resuscitation has been shown to exacerbate the systemic insult in response to hemorrhagic shock, including worsened cardiac function, pulmonary complications, gastric dysmotility, abdominal compartment syndrome, coagulopathy, immunological and inflammatory dysfunction [12,13,14]. Excessive fluid resuscitation is also linked to reperfusion injury and up-regulation of various inflammatory processes indicated by leukocyte adhesion, capillary leakage and edema [3, 15]. In keeping with these principles, a retrospective trial demonstrated a survival advantage in elderly patients receiving limited-volume crystalloid resuscitation [16]. Additionally, a single-centre retrospective trial in the United States demonstrated higher mortality in trauma patients receiving more than 1.5L of crystalloid in the emergency department [17]. In the present study, we set forth to investigate whether large-volume crystalloid resuscitation of the adult trauma patient was associated with higher mortality. We also investigated the incidence of complications and duration of mechanical ventilation.

Methods

Study population

A retrospective cohort analysis of all adult trauma patients admitted to St. Michael’s Hospital (SMH) between December 2011 and December 2012 was performed. Patients with traumatic hanging injuries, and those who died within 1h of hospital admission were excluded from data analysis. The study was approved by the research ethics board (REB) of St. Michael’s Hospital, Toronto, Canada.

Data source

The trauma registry of our institution captures data on individual trauma patients admitted to the hospital’s Emergency Department (ED) up to the day of discharge. The following data were obtained from the registry: demographic information, Injury Severity Score (ISS), initial laboratory parameters (hemoglobin, platelet count, international normalized ratio (INR), partial thromboplastin time (PTT), lactate, pH, fibrinogen), total volume of crystalloid and blood products administered during the first 24h of injury (including pre-hospital phase), intensive care unit (ICU) stay and hospital length of stay (LOS), ventilator days, in-hospital complications, and in-hospital mortality. Data from pre-hospital fluid infusion data were extracted from ambulance report forms. Crystalloid infusion was calculated as the total volume of normal saline (NS) and/or lactated Ringer’s (LR) received during the first 24h, starting from the time of injury. Quality of data was assessed for accuracy and completeness. Any outliers were verified by revisiting patient records. Missing values from the trauma registry were obtained from the hospital charts when available. Due to a high percentage of missing data for pH values (258/903 missing) and lactate values (475/903 missing), these variables were not included in the analysis. Furthermore, blood pressure on arrival was not collected and missing values of base excess (608/903 missing) precluded their inclusion in the analysis.

The primary outcome was in-hospital mortality. This was recorded in the registry as either death during hospital stay or alive at discharge. Secondary outcomes included: development of in-hospital complications (IHC) defined by pneumonia and/or sepsis) and total number of ventilator-days (measured as total number of days the patients required mechanical ventilation). In our study, large-volume resuscitation was defined as infusion of more than 5l of crystalloids in the first 24h.

Definitions of complications

Complications were defined by the Ontario Trauma Registry Comprehensive Data Set (OTR CDS) data dictionary (May 2014) [18]. The OTR CDS provides detailed information on complications experienced by traumatically injured patients in the province of Ontario (see Appendix for definitions).

Statistical analysis

Descriptive statistics were performed for all variables. Continuous variables were reported as median and interquartile range (IQR), as the variables did not follow a normal distribution. Categorical variables were reported as absolute numbers and percentages. Univariate in-hospital mortality (non-survivors versus survivors) and complications (complications versus no-complications) were compared using Wilcoxon Rank-Sum test for continuous variables and Chi-square test for categorical variables. To assess the independent relationship of large-volume crystalloid transfusion to mortality and complications, we performed a multivariable logistic regression analysis using a backward selection process while controlling for variables selected a priori. The results were reported as adjusted odds ratio (aOR) with 95% confidence interval (CI).

A multivariable Poisson regression analysis was performed to assess the independent association between administration of ≥5L crystalloid/24h with respect to number of ventilator-days while controlling for variables selected a priori. Prior to multivariable modeling, the variables were assessed for multicollinearity using tolerance statistics. Appropriate numbers of variables were used so as not to overfit our model. Presence of variance inflation factor (VIF) of > 0.4 was considered to be multicollinear. Model diagnostics were performed, and a p-value of < 0.05 was considered statistically significant. All data analyses were carried out using Statistical Software SAS (v.9.3 SAS Institute, Carry, NC).

Results

Patient characteristics

During the study period, 987 adult trauma patients were admitted to St. Michaels Hospital. A total of 970 patients were included in the final analysis, 17 were excluded because of inadequate data. Patient demographics, fluid resuscitation volume, clinical and laboratory admission parameters and blood product transfusion data are shown in Table 1. Patients were categorized according to the volume of crystalloid: ≥5L or < 5L of crystalloids within 24h of the injury (Table 1). Of the 970 patients, 264 (27%) received ≥5 L of crystalloids in the first 24h following injury while 706 (73%) received < 5L of crystalloids. Apart from gender, all other parameters were significantly different (p < 0.05) between the patients according to the volume of crystalloid received. Patients who received ≥5L of crystalloids within 24h were, for the most part, more severely injured (median ISS17 (IQR 9–25)), had worse physiologic parameters during admission, received more blood products (except for cryoprecipitate), had higher complication rate and ultimately higher mortality (Table 1).

Table 1 Clinical characteristics of trauma patients included in the analysis (n = 970) and patients receiving ≥5L versus < 5L crystalloids in 24 h of injuries
Full size table

Mortality

Univariate analysis between non-survivors and survivors showed that, aside from gender, all other demographics were significantly different between survivors and non-survivors (Table 2). Physiologic variables were also noted to be worse in the non-survivors group (Table 2).

Table 2 Univariate comparison between non-survivors and survivors
Full size table

The non-survivor group received a median resuscitative crystalloid volume of 5.3L in the first 24h (IQR 2.1–10) versus 2.2 L (IQR 1–5) in the survivor group. Larger volumes of blood products were also administered in the first 24 h to the non-survivor group as compared to the survivor group (Table 2).

Multivariable logistic regression analysis showed that older age, higher ISS, infusion of ≥5L crystalloids within 24 h of injury, and lower temperature were independent predictors of mortality (Table 3). Ultimately, patients receiving ≥5L crystalloids within 24 h of injury died as a result of multi organ dysfunction (MOD) associated with recalcitrant coagulopathy of trauma. The regression model had excellent discrimination (receiver operating characteristic: 0.87) and calibration (Hosmer-Lemeshow goodness-of-fit statistic 0.28).

Table 3 Multivariable logistic regression model associated with hospital mortality
Full size table

In-hospital complications

A total of 118 patients (12%) developed complications. Univariate association of selected risk factors with in-hospital complications status are described in Table 4. Patients who experienced in-hospital complications were older (53(IQR 33–65) vs 44(27–64), p = 0.004); severely injured (26 (IQR 19–35) vs. 16 [8,9,10,11,12,13,14,15,16,17, 19,20,21,22,23,24,25,26], p < 0.0001) and received more blood products (PRBCs, platelets and fibrinogen). Total duration of ICU stay (15 (IQR 9–24) vs 0 (IQR 0–1) p < 0.0001); duration of mechanical ventilation ((13.5 (8–21)vs0 (0–1), p < 0.0001) and total hospital stay ((26 (IQR 17–45)vs. 4 (IQR 1–9), p < 0.0001) were significantly longer in patients who experienced complications compared to the patients without (Table 4). Furthermore, mortality was significantly higher in patients with complications compared to patients without complications (14% versus 6%, p < 0.0001). The variables found to be independently associated with in hospital complications in multivariable logistic regression were: older age (aOR1.03 (95% CI: 1.01–1.05), p < 0.001); longer ICU days (aOR 1.58 (1.46–1.7), p < 0.0001) and administration of platelet units within 24 h (aOR 1.36 (1.00–1.84), p = 0.044) (Table 5).

Table 4 Univariate comparison of patients with and without complications
Full size table
Table 5 Multivariable logistic regression model associated with complications
Full size table

Ventilator-days

A multivariable Poisson regression estimation of the association between patient factors and the need for ventilator support was assessed. Results showed that administration of ≥5L crystalloid in the first 24h was associated with a significant increase in ventilator-days (RR 2.31 (95%CI 1.81–2.96), p < 0.0001. Other factors included higher ISS (1.02 (95% CI 1.01, 1.03), p < 0.0001), development of complications (RR1.91 (95%CI 1.48, 2.48), p < 0.0001) and low initial temperature (0.87 (95% CI 0.80, 0.96), p = 0.0039) (Table 6).

Table 6 Multivariable Poisson regression estimating the association of administration of ≥5L crystalloid/24h of injury and duration of ventilator stay
Full size table

Discussion

In our retrospective cohort study, we found that trauma patients who received ≥5L of crystalloids in the first 24 h of injury had significantly higher mortality rates (26%), when compared to patients who received < 5L of crystalloids within the same period (7%). This increase in mortality represents an odds ratio of 2.55. Moreover, ≥5L fluid resuscitation in the first 24 h of injury was an independent predictor of mortality, along with older age, higher ISS, and lower admission temperature, accentuating the overall detrimental effect of injudicious fluid infusion in trauma patients. A previous study showed higher mortality related to increased crystalloid administration in the emergency department [17]. However, that study did not include pre-hospital and total fluid administration in 24 h, which we believe to be an important and potentially less biased measure of total resuscitation.

The importance of fluid resuscitation to maintain tissue perfusion in hemorrhagic shock has been well stablished and should not be underestimated. However, studies showed that resuscitation regimens aimed at decreasing crystalloid infusion and delivering higher ratios of blood products led to fewer complications associated with excessive fluids, less coagulopathy and ultimately increased survival [19,20,21,22].

In keeping with those reports, our findings provide further support for a more conservative approach to fluid resuscitation.

At the microvasculature and cellular levels, large volume resuscitation provokes generalized increase in interstitial fluid and cellular edema. These changes have been linked to organ dysfunction [12]. Excessive LR infusion, particularly the D-isomer of lactate, is known to increase the expression of inflammatory genes, neutrophil inflammatory burst, and neutrophil adhesion molecules [23, 24].

In our study, the difference in fluid volume between survivors and non-survivors was more than double. Nonetheless, we speculate that higher volume differences would have had an even more pronounced effect in the parameters assessed in our study [25,26,27]. Small differences in intravenous fluid volumes between study groups has been a major limitation in studies designed to investigate low-volume resuscitation and permissive hypotension in trauma patients. For instance, a randomized prospective clinical trial terminated prematurely because it failed to demonstrate decreased mortality with intraoperative hypotensive resuscitation at a target mean arterial pressure of 50 mmHg compared to 65 mmHg [28]. The lack of difference in mortality in the aforementioned study could be explained by the modest amounts of intravenous crystalloid infusion overall (median volume of 2–2.2 l) and the negligible difference in the total volumes of fluid received between the groups (approximately 1 l).

As expected, our study showed that patients who developed in-hospital complications had significantly longer ICU and hospital LOS. However, volume infusion within the first 24h was not an independent predictor of the aforementioned findings. Nevertheless, infusion of ≥5L of crystalloids in the first 24h was associated with a significant increase in the duration of mechanical ventilation. Thus, it is conceivable that higher volume infusion could have a primary effect on pulmonary function interfering with weaning from mechanical ventilation. Moreover, other detrimental effects linked to high volume resuscitation also contributed to more prolonged ICU LOS. Accordingly, previous reports showed that mechanical ventilation is a known risk factor for in-hospital complications linked to significant personal and economic costs [8, 29, 30].

The extent of physiologic derangement as reflected in laboratory variables was associated with higher crystalloid resuscitation volume. This was demonstrated by higher levels of INR, PTT, and lactate, together with lower fibrinogen and pH; previous studies corroborate these findings [12, 19]. Younger patients and patients with higher injury severity scores (ISS) also tended to receive more fluids in our study. The effect of intravenous resuscitation volume on mortality of elderly trauma patients compared to nonelderly has been previously addressed [17]. It was shown that mortality was almost three times more likely in elderly patients who received 3 L of crystalloid in the emergency department compared to nonelderly patients receiving similar volumes. In addition, older age (> 80) and Glasgow coma scale (GCS) < 8 were also significantly related to mortality. Certainly, the combination of excessive fluid administration and advanced age can exacerbate complications after resuscitation.

The use of vasopressors in severe trauma patients has been described as an option to decrease fluid resuscitation volumes [31, 32]. However, strong support for the use of this strategy in the exsanguinating patient is still lacking [32]. Additionally, the relationship between cardiac output, volume status, and venous return in the hemorrhaging patient has restricted the use of this strategy [33]. Moreover, concerns have been raised about arteriolar constriction and reduced micro-circulatory flow to the splanchnic circulation with vasopressor therapy in hemorrhagic shock. However, experimental studies suggest that the doses of vasopressor required in hemorrhagic shock resuscitation are, for the most part, insufficient to compromise splanchnic microcirculation [34].

Our study has several limitations inherent in retrospective data collection. Despite a cohort of 970 patients the study population is considerably heterogeneous, even though attempt was made to match these groups as well as possible. However, in a single center study like ours the effect of group differences is considerably important in the data analysis. Another important limitation was the inability to include lactate, pH, admission blood pressure and base excess data in the multivariate logistic regression analysis because of significant missing data. Lactate and base excess are important markers of tissue perfusion, thus a reliable marker of the resuscitation efforts. Furthermore, admission blood pressure of severely injured patients help guide clinical decision making with respect to the need for ongoing volume replacement. Finally, we assessed crystalloid administration during the first 24 h after injury. It is conceivable that the effect of fluids administered during other periods of the hospital admission were not considered in the results. Nonetheless, our objective was to determine the effect of fluids received during the initial 24 h on the outcome of the patients.

Conclusion

Large-volume fluid resuscitation (≥5L) with crystalloids in the first 24h after traumatic injury is associated with higher mortality rates and prolonged need of mechanical ventilation. In addition, age, higher ISS, and presenting temperature were independently predictive of in-hospital mortality. The current study supports the principles of judicious fluid resuscitation in trauma. These findings also highlight the need for revision of current resuscitation guidelines for fluid administration to traumatically-injured patients.