Trauma is a major health problem and a leading cause of mortality and morbidity among young individuals in the world [
It is not currently possible to reliably predict the occurrence, timing, or type of complications in individual patients. However, identifying the subgroup(s) of patients (risk factors) that may develop complications may allow for preemptive rather than reactive therapy. In addition, identification of the epidemiology, patterns, and causes of complications following trauma may provide useful information for improving treatment strategies, outcomes, and costs ultimately enhancing the quality of the health system, especially in the area of trauma care (Level I Trauma Center) [
This study is a population-based retrospective cohort study. that was conducted using data from the University HealthSystem Consortium (UHC), an alliance of over 90% of academic medical centers and their affiliated hospitals in the United States [
This study included 11064 patients, 18 years of age or older, presenting with trauma and admitted to an intensive care unit (ICU) from May 2008 to April 2009. Trauma characteristics of patients were identified by selecting a specific group of ICD-9-Clinical Modification (ICD-9-CM) codes defined by the American College of Surgeons (ACS). Demographic and clinical data, including age, sex, mechanism of injury, procedures, hospital length of stay (LOS), complications, and inhospital mortality were obtained. For the purpose of the analysis, cause of admission was grouped into 4 major categories: internal injury (24.2%), traumatic CNS injury (23.7%), fracture (16.8%), and others (35.3%). All patients in our study had an invasive procedure performed such as surgery or vascular catherization.
The overall number of complications per patient was identified as well as the presence or absence of specific complications. A standardized manual for definitions of complications was used for reference [
The primary outcome of interest was presence or absence of complications. The secondary outcome of interest was mortality.
Exploratory analysis was carried out to determine the distribution of the demographic and clinical variables. Continuous variables are presented as mean (SD) or median (interquartile range). Distributions of categorical variables were presented as frequencies and percentages. The association between each continuous variable and complications or mortality was evaluated using the Mann-Whitney
Univariate logistic regression analysis was used to evaluate the prognostic ability of the demographic and clinical variables, individually, to predict the probability of development of complications or death. Crude odds ratios with 95% confidence intervals are presented. The
Variables associated with each outcome in the univariate analysis (
All hypothesis tests conducted were 2-tailed. A
The demographic and clinical data are summarized in Table
Summary of demographic and clinical data for trauma cases.
Age, years | |
Median (Q1–Q3) | 45.96 (29.3–61.7) |
Min–Max | 18–94 |
Gender, |
|
Female | 3196 (28.9%) |
Male | 7868 (71.1%) |
Mortality, |
|
Alive | 9076 (82.0%) |
Dead | 1988 (18.0%) |
Complication, |
|
No | 7613 (68.8%) |
Yes | 3451 (31.2%) |
Number of complications, |
|
1 | 2097 (60.8%) |
2 | 804 (23.3%) |
3 | 349 (10.1%) |
4 | 132 (3.8%) |
5 | 47 (1.4%) |
6 | 21 (0.61%) |
9 | 1 (0.03%) |
Among patients studied, 31.2% developed complications. The characteristics of patients with and without complications are listed in Table
Characteristics of patients with and without complications.
Characteristics | No complications |
Complications |
|
|
---|---|---|---|---|
Age: | Median (Q1–Q3) | 46.33 (29.58–62.67) | 45.17 (28.83–60) | 0.0009** |
Gender: F/M |
|
2306/5307 (30/70) | 890/2561 (26/74) | <0.0001*** |
Mortality: No/Yes |
|
6257/1356 (82/18) | 2819/632 (82/18) | 0.52 |
Significant differences are indicated with
Summary of complications for trauma cases.
Complication | Overall |
Survivors |
Deceased |
|
---|---|---|---|---|
Postoperative pulmonary compromise | 1733 (30.7%) | 1299 (27.9%) | 434 (43.8%) | <0.0001*** |
Venous thrombosis/pulmonary embolism | 627 (11.1%) | 568 (12.2%) | 59 (6.0%) | <0.0001*** |
Other complications of procedures | 464 (8.2%) | 411 (8.8%) | 53 (5.4%) | 0.0002** |
Mechanical complications due to device or implant | 356 (6.3%) | 306 (6.6%) | 50 (5.1%) | 0.0588 |
Cellulitis or decubitus ulcer | 329 (5.8%) | 306 (6.6%) | 23 (2.3%) | <0.0001*** |
Postprocedural hemorrhage or hematoma | 312 (5.5%) | 261 (5.6%) | 51 (5.2%) | 0.4951 |
Postoperative pneumonia | 309 (5.5%) | 244 (5.2%) | 65 (6.6%) | 0.1772 |
Reopening of surgical site | 262 (4.6%) | 218 (4.7%) | 44 (4.4%) | 0.6748 |
Wound infection | 262 (4.6%) | 242 (5.2%) | 20 (2.0%) | <0.0001*** |
Miscellaneous complications | 254 (4.5%) | 228 (4.9%) | 26 (2.6%) | 0.0016* |
Procedure-related perforations or lacerations | 192 (3.4%) | 155 (3.3%) | 37 (3.7%) | 0.7043 |
Postoperative infections not pneumonia/wound | 187 (3.3%) | 171 (3.7%) | 16 (1.6%) | 0.0010* |
Postoperative GI hemorrhage or ulceration | 84 (1.5%) | 71 (1.5%) | 13 (1.3%) | 0.6494 |
Postoperative stroke | 82 (1.5%) | 67 (1.4%) | 15 (1.5%) | 1.000 |
Postoperative AMI | 61 (1.1%) | 39 (0.84%) | 22 (2.2%) | 0.0004** |
Postoperative cardiac abnormality | 47 (0.83%) | 14 (0.30%) | 33 (3.3%) | <0.0001*** |
Shock or cardiorespiratory arrest | 26 (0.46%) | 10 (0.21%) | 16 (1.6%) | <0.0001*** |
Aspiration pneumonia | 24 (0.42%) | 20 (0.43%) | 4 (0.40%) | 1.000 |
Postoperative urinary tract complication | 18 (0.32%) | 14 (0.30%) | 4 (0.40%) | 0.8703 |
Postoperative physical and metabolic derangements | 15 (0.27%) | 11 (0.24%) | 4 (0.40%) | 0.5881 |
Central or peripheral nervous system | 3 (0.05%) | 2 (0.04%) | 1 (0.10%) | 1.000 |
Septicemia | 2 (0.04%) | 1 (0.02%) | 1 (0.10%) | 0.7956 |
Complications related to anesthetic agents/CNS agents | 1 (0.02%) | 1 (0.02%) | 0 (0.00%) | 1.000 |
|
||||
TOTAL | 5650† | 4659 | 991 |
Significant differences are indicated with
Most patients had a single complication (60.8%) (Table
Eight hundred and four patients (23.3%) developed two complications. Interestingly, more than 50% of the patients with 2 complications presented a pulmonary complication.
Individual logistic regression models examining the strength of association between each clinical and demographic variable and the development of complications were constructed. This analysis showed that several characteristics predict complication after trauma (Table
Unadjusted odds ratios of clinical and demographic characteristics for predicting complications in trauma patients, using univariate logistic regression.
Variable | OR (95% CI) |
|
|
---|---|---|---|
Age group | 5 | ||
Young adults | Reference | ||
Middle age | 1.04 (0.95–1.15) | 0.40 | |
Elderly | 1.02 (0.91–1.15) | 0.71 | |
Advanced seniority | 0.68 (0.59–0.77) | <0.0001*** | |
Gender | 30 | ||
Female | Reference | ||
Male | 1.25 (1.37–1.42) | <0.0001*** | |
Traumatic CNS injury | 10 | ||
No | Reference | ||
Yes | 1.161 (1.160–1.273) | 0.001* | |
Diagnosis on admission | 2 | ||
Other | Reference | ||
Fracture | 1.179 (1.012–1.374) | 0.035* | |
Internal injury | 1.350 (1.172–1.556) | <0.0001*** | |
Traumatic CNS injury | 1.306 (1.132–1.506) | 0.0002** |
OR = odds ratios.
Forward stepwise logistic regression analysis identified patient age, gender, and presence of CNS injury as predictors for complications (see Methods section). When we explored interactions, we found that the interaction between CNS injury and gender was also significant. Our final model included patient characteristics age and gender as well as the presence of CNS injury and the interaction between CNS injury and gender as covariates. Adjusting for all other variables in the model, analysis of complications demonstrated that patients in the advanced seniority age group have odds of developing complications which are 30% less than that among young adults (adjusted OR 0.70, 95% CI 0.61 to 0.80,
Univariate binary logistic regression analysis showed that several characteristics were strongly associated with death after trauma (Table
Unadjusted odds ratios of clinical and demographic characteristics for prediction of death in trauma patients, using univariate logistic regression.
Variable | OR (95% CI) |
|
|
---|---|---|---|
Age | 1.024 (1.022–1.027) | <0.0001*** | 57 |
Age group | 60 | ||
Young adults | Reference | ||
Middle age | 1.36 (1.20–1.55) | <0.0001*** | |
Elderly | 2.01 (1.76–2.33) | <0.0001*** | |
Advanced seniority | 4.02 (3.51–4.60) | <0.0001*** | |
Gender | 10 | ||
Female | Reference | ||
Male | 0.869 (0.783–0.966) | 0.009* | |
Traumatic CNS injury | 12 | ||
No | Reference | ||
Yes | 4.549 (4.106–5.040) | <0.0001*** | |
Complications | 1 | ||
No | Reference | ||
Yes | 0.967 (0.871–1.073) | 0.52 | 0.1 |
Number of complications | 0.983 (0.933–1.036) | 0.52 | |
Diagnosis on admission | 15 | ||
Other | Reference | ||
Fracture | 0.351 (0.277–0.444) | <0.0001*** | |
Internal injury | 0.578 (0.478–0.699) | <0.0001*** | |
Traumatic CNS injury | 3.314 (2.816–3.900) | <0.0001*** |
OR = odds ratios.
Forward stepwise logistic regression analysis, including patient characteristics (age, gender) and trauma characteristics (diagnosis on admission) as covariates, identified patient age and diagnosis on admission as predictors of death. Complementary to the previous model, stepwise logistic regression analysis, including patient characteristics (age, gender) and traumatic CNS injury as covariates, identified patient age and traumatic CNS injury as predictors of death. Gender was not significant in the full model. It is likely that the gender’s effect in the simple model was related to age. No significant interactions were found. In multivariate analysis of young adults with no traumatic CNS injury as the references, patients in advanced seniority (adjusted OR 4.30, 95% CI 3.72–4.97,
Complications following admission for traumatic injury are common and have been shown to increase morbidity, length of stay, and costs in a level I trauma center [
The objective of this study was to describe epidemiologic features, risk factors for acquisition, and outcome of complications that can occur after trauma in a cohort of 11,064 patients who were admitted to the ICU in Level I Academic Trauma Centers. The findings from our study show that (1) age, gender, diagnosis on admission, and CNS injury were associated with higher incidence of complications; (2) occurrence and number of complications correlated with LOS but not with mortality; and (3) mortality and complications are associated with different risk factors.
Our data suggest that there is a gender-related difference in complication rates. In particular, we demonstrated that male patients had substantially higher incidences of complications. Supporting these observations, multivariable logistic regression analysis identified gender as an independent predictor, with men exhibiting higher odds of developing complications when compared to female patients. In line with our findings, there is an increasing number of experimental and human studies supporting a gender-related differences among trauma patients in developing complications [
Additionally, an interaction analysis was undertaken to evaluate whether gender impacts the association between complications and CNS injury. The results of our analysis showed that if a patient sustained a traumatic CNS injury, there were predictive gender differences. Men having a traumatic CNS injury were found to have a 24% higher odd of developing complications over those without a CNS injury. In contrast, the presence of CNS injury showed no significant increase in the odds of complications as compared to females without CNS injury. This finding supports previous experimental and clinical evidence showing gender-related differences in outcome after a neurotrauma [
Another interesting finding is that, despite this gender-related difference in developing complications, there was no difference in survival which is consistent with epidemiological studies and clinical experience [
In line with previous research [
Interestingly, consistent with previous reports [
Not all variables were diametrically opposite for prediction of either complications or mortality. For example, traumatic CNS injury was significantly associated with both complications and mortality. Traumatic CNS injury has been demonstrated a primary cause of death in previous reports [
It should also be noted that when stratifying complication type with mortality, complications related to a cardiopulmonary process were more significantly associated with mortality; while blood borne complications such as infection and deep vein thrombus were more associated with those who survived. While the severity of some complications are more lethal, the frequency of nonlethal complications account for the majority of ICU complications (Table
Our study has some limitations. The uncertainty about the timing of complication onset did not allow us to investigate the temporal distribution of the events. In particular, we were not able to establish a temporal relationship (and potentially causal-effect relationship) between complications and mortality. However, the intent of our studies was to investigate the general features of complications in trauma population and to compare risk factors associated with the onset of complications with those associated with death. An additional constraint was the use of an administrative database. Administrative databases are an important source of information, and they are especially convenient in studying low frequency events. However, their main limitation is related to the level of detail required for clinical interventions at the bedside for the different conditions and diseases especially in an ICU setting [
A main strength of this article is that it represents one of the largest groups of ICU patients in which complications were evaluated. This provides the statistical power to capture even rare clinical events. Furthermore, this is a multicenter study representing over 90% of academic centers in the United States. This avoids the bias that can be present in studies using a single center. Additionally, our study includes Level 1 trauma centers that consistently provide the highest level of surgical care and ICU management and a full spectrum of patients. Therefore, these results should be representative and can be extrapolated towards the general population of trauma patients.
The current research of >11,000 patients has provided characterization of patients and their complications which develop after trauma. This valuable information may help identify subjects at high risk for the future development of complications and be applied in clinical practice for preemptive strategies. Furthermore, using multivariable logistic regression analysis we have shown that complications and mortality among ICU trauma patients are associated with different risk factors. This means that modifying factors influencing occurrence of complications do not necessarily offer survival advantage after trauma. Therefore, before embarking on large expensive clinical trials targeting or manipulating specific variables, it is of paramount importance to conduct thorough studies adequately addressing the role and interactions of various risk factors.
The authors declare that there is no conflict of interests regarding the publication of this paper.
The authors are grateful to Dr. Andrea Gabrielli for his valuable advice for the paper and Dr. Samuel Hohmann of the University Health System Consortium who provided the data.