Introduction
Despite geographical differences, obesity has become a major burden to healthcare systems worldwide, especially in western countries. In North America, the prevalence of obesity (body mass index (BMI) ≥ 30), as defined by the World Health Organization [
1], exceeded 20% in all but two states in 2009, and obesity is reported to now account for up to 9% of healthcare costs [
2]. A similar trend with rising prevalence rates is observed in Europe, with obesity affecting between 4 and 28% of men and between 6 and 36% of women [
3]. In parallel to the rising prevalence of obesity in the general population, an increasing number of overweight and obese trauma patients are treated at major trauma centers.
A number of studies have sought to describe the effect of obesity on treatment and outcome in trauma patients. Several studies describe a higher risk of mortality in blunt trauma, a higher rate of organ dysfunction and failure, as well as a prolonged ICU and hospital stay [
4‐
7]. Ciesla and colleagues found a correlation between BMI and subsequent organ failure in their study of 716 trauma patients, irrespective of injury severity or the number of blood transfusions [
6]. Bercault and colleagues reported an increased surgical ICU mortality rate especially among younger obese patients using a matched cohort study design, and attributed the excess mortality mainly to secondary complications specific to ICU treatment [
4]. However, these findings are not unquestioned: neither Ciesla and colleagues or Duane and colleagues, in their series of 453 trauma patients, found differences in overall mortality in patients stratified by BMI [
6,
8] - only a slightly longer hospital stay was observed in obese patients, thereby raising doubts about the extent of obesity-related risks in trauma care.
The aims of our current study were threefold. Firstly, we wished to examine the distribution of normal-weight patients, obese patients, and also underweight patients in our patient population representative of a Western European trauma center and place our data in perspective with the mostly American literature. Secondly, we sought to compare admission physiologic status and treatment and in-hospital outcome parameters between underweight, normal-weight, and overweight patients. Finally, we wished to identify specific process measures that may help explain BMI-related outcome differences in trauma patients.
Materials and methods
Study design and setting
The study was conducted as a retrospective cohort analysis using a prospectively led single-center trauma database. The University of Zurich trauma database comprises all severe trauma patients admitted between 1996 and present. Inclusion criteria for this study are an Injury Severity Score (ISS) ≥ 16 in adult patients older than 16 years of age. Excluded from this study were patients with burn injuries and delayed referrals > 12 hours after trauma. BMI values were assessed using either information from recent hospital visits, information provided by the patient or their primary care physician, or (least preferred) by measurements taken during the patient's hospital stay. Patients with missing BMI values were excluded. A total of 1,084 patients have been enrolled into this study, which has been approved by the University of Zurich institutional review board. A waiver of individual patient consent was provided in adherence to institutional regulations, and the study was conducted according to good clinical practice guidelines.
Definition of obesity
Patients were grouped by BMI (body mass in kilograms divided by the square of the height in meters) into five groups as defined by the World Health Organization [
1]: underweight, all patients with BMI < 18.5; normal, all patients with BMI of 18.5 to 24.9; overweight, all patients with BMI of 25 to 29.9; obese, all patients with BMI of 30 to 39.9; and morbid obese, all patients with BMI ≥ 40. Because there were only 11 patients with BMI ≥ 40, we combined obese and morbidly obese patients into one group.
Process and outcome measures analyzed
Patient demographics, injury pattern and severity, prehospital resuscitation, and diagnostic measures upon admission and their results were recorded. The patients' trauma load and initial medical condition were assessed using the Abbreviated Injury Score, ISS and New ISS, Glasgow Coma Scale, and the Acute Physiologic and Chronic Health Evaluation score. Subsequent operations as well as their duration and daily routine laboratory parameters were recorded. We further analyzed the length of time between injury, trauma room admission, transfer to the operating theater, and admission to the trauma ICU. Outcome parameters included length of hospital stay and ICU stay, length of mechanical ventilation, overall mortality, cause of death and morbidity grouped into infectious complications and multiple organ dysfunction/multiple organ failure according to the Marshall score, the Goris score, the Murray score and Sequential Organ Failure Assessment (SOFA) scores [
9‐
11]. The Marshall and Goris scores vary in the number of organ systems evaluated and the points given to each - seven organs with 0 to 2 points each (maximum 14) in the Goris score, and six organs with 0 to 4 points each (maximum 24) in the Marshall score. The SOFA differs from the Marshall score in its definition of cardiovascular failure (for example, including amount of inotropes given). The Murray score is used to assess acute lung injury leading to acute respiratory distress syndrome [
11]. Systolic blood pressure < 90 mmHg was used as a substitute definition of shock in this study to allow for quantitative comparison between groups of different BMI.
Volume therapy data are provided as crystalloid, colloid, and total volumes administered over time. Volume corrected per BMI is defined as the mean amount divided by the mean BMI in each group, thereby controlling for BMI. Arterial lactate is shown as the admission value, the percentage of patients with admission lactate above a defined threshold of 2.5 mmol/l, and the course of arterial lactate over time as a function of BMI.
Admission day (day 0) trauma care protocols
All patients were resuscitated and diagnosed according to advanced trauma life support guidelines with routine basic imaging containing plain chest and pelvic radiography as well as focused abdominal sonography in trauma by the trauma resident. Our protocol of intravenous access provides for the use of an intraosseus cannula after the third failure to achieve intravenous access. In hemodynamically stable patients with no need for immediate operative intervention, computed tomography scanning was performed. In hemodynamically unstable patients, life-saving surgery either in the trauma room or in the adjacent operating theater was performed without prior computed tomography scanning. Our operating tables are certified for patients weighing ≤ 250 kg, thereby rarely necessitating specialized tables for bariatric surgery (for which our hospital is certified). Following initial stabilization, all patients with severe trauma were transferred to a specialized trauma ICU. Hemodynamic monitoring using an arterial radial line was used in patients with anticipated fluctuations in intravascular volume, those with pre-existing cardiac and pulmonary constraints, and in patients with traumatic brain injury. For non-invasive monitoring, an oversized cuff was used in obese patients or those with corpulent upper arms.
Statistical analysis
Univariate comparisons between groups were performed to detect differences between the four BMI groups; analysis of variance was used to detect differences in normally distributed continuous data, and the Kruskal-Wallis-test for non-normally distributed parameters. To compare nominal data, we utilized the Pearson chi-square test or the Fisher's exact test. To correct for multiple testing among different subgroups, these tests were followed by Bonferroni correction and significance was set at P < 0.0083. Data are presented as the median (range) or mean ± standard deviation, as appropriate. P values in the tables indicate comparison of all groups demonstrating an overall effect of BMI. Error bars in the figures indicate the standard error of the mean. Independent predictors for death were determined using a stepwise logistic regression analysis. For the logistic regression only, (severe) trauma to a distinct organ was defined by Abbreviated Injury Score ≥ 3.
All statistical analyses were performed using IBM SPSS Statistics 18.0 software (SPSS Inc., Chicago, IL, USA). Graphical visualization was performed using SigmaPlot 9.0 (Systat Software Inc., Richmond, CA, USA).
Discussion
Trauma in obese patients has become a regular occurrence in western trauma centers and the topic has recently been given broad attention in the North American literature. By nature, the percentage of obese patients varies greatly between studies (5 to 31%), with a higher incidence in more recent reports [
5‐
7,
12‐
18]. However, fewer data exist on the incidence of obesity in European trauma centers. The first aim of our study was therefore to survey the incidence of obesity in our trauma population, which appears comparably low at a mean of 8.3% from the years 1996 to 2009. In fact, only 11 patients (1%) of our population fall into the category of morbid obesity. At the same time, our percentage of overweight patients (35% with BMI of 25 to 29.9) is similar to that in the existing literature [
5‐
7,
12,
15,
16,
18].
In essence, our study shows that obesity is an independent predictor for overall mortality after trauma (Table
2). Unadjusted mortality rates are highest in obese patients and lowest in normal-weight patients. Underweight patients (BMI < 18.5) had nonsignificantly higher mortality, which has been described earlier [
19,
20]. Interestingly, the rate of suicide was almost threefold higher in underweight patients when compared with normal-weight patients. The association between BMI and trauma-related mortality is discussed controversially in the literature; most authors report higher mortality rates in obese trauma patients, such as Neville and colleagues who surveyed 242 blunt trauma patients and observed a mortality rate of 32% in obese patients compared with 16% in normal-weight patients [
14]. Similarly, Choban and colleagues reported a high (42.1%) mortality rate among extremely overweight (BMI ≥ 40) patients compared with only 5% among those with BMI < 27 [
21]. Bercault and colleagues showed an increased mortality in obese patients in a large study of 1,927 patients (32% vs. 17%) [
4]. A recent analysis of 5,766 patients from the German Trauma Registry also showed a significant correlation between BMI and mortality in both overweight and underweight patients [
19].
However, several well-designed trials have been unable to confirm higher mortality rates in obese patients [
6,
8]. In a prospective study on 716 trauma patients, Ciesla and colleagues found an almost twofold higher incidence of early multiple organ failure and a prolonged hospital stay in obese patients, but no difference in overall mortality [
6]. Similarly, Duane and colleagues in their retrospective analysis of 338 trauma victims did not record a higher mortality in obese patients [
8]. A recent meta-analysis on the effects of obesity on survival in critically ill surgical patients by Akinnusi and colleagues has shown equal mortality in obese and nonobese patients, which the authors attributed to several factors. First, they argued that the abundance of adipose tissue may diminish the consequences of the severe catabolic state in prolonged systemic illness [
22]. Alternatively, obesity may already contribute to greater alertness among ICU staff and expedite the correction of physiologic derangements such as stress hyperglycemia. Akinnusi and colleagues' study also found that obese patients required a prolonged length of ICU stay and mechanical ventilation. The authors attribute these findings to a diminished respiratory reserve in obese patients and a subsequent risk of respiratory failure following even mild pulmonary or systemic insults [
22]. In fact, only three of the examined 13 studies did not show a prolonged ICU length of stay in obese patients. Our data are in accordance with these findings in that we recorded (nonsignificantly) longer lengths of stay in obese and overweight patients, which may be explained by a higher rate of comorbidities (Acute Physiologic and Chronic Health Evaluation II scores) and more frequent complications such as superficial wound infections.
Similarly, the occurrence of multiple organ dysfunction and multiple organ failure appears to be associated with obesity (Table
3): the Marshall score, the Murray score, or the SOFA score show higher values with increasing BMI either at admission or as a peak value, indicating a moderate yet significant level of organ dysfunction. These findings are comparable with a recent study of 1,543 trauma patients by Newell and colleagues in which the authors found an odds ratio of 13.5 for the development of renal failure in morbidly obese patients [
23]. Hence, the higher incidence of renal failure may be due to an unrecognized need for higher fluid requirements in obese trauma victims.
Another finding of our study is that obesity is associated with increased mortality from hemorrhagic shock, despite this being a rare occurrence. In fact, no underweight patients and only 10 patients with a normal BMI died from this cause. Our data suggest that, once controlled for BMI, the initial volume therapy may be inadequate: arterial pH values showed optimal courses only in patients with normal BMI. In the same way, we observed a higher rate of hypotension on admission and a greater percentage of obese patients with lactate levels above 2.5 mmol/l, although neither difference reached statistical significance.
Initial volume therapy is guided by parameters such as hemodynamic monitoring, arterial lactate, diuresis, and visible and anticipated blood loss. Monitoring cardiac output/index using Swan-Ganz or peripherally inserted central catheters may help to assess individual hemodynamics and distinguish hypovolemic from distributive shock, but their use in trauma care is debatable [
23]. Our data indicate that metabolic derangements in obese patients are not reflected in significantly lower systolic or mean arterial blood pressures, thereby facilitating inadvertent under-resuscitation. Difficulty in intravenous access with a need for intraosseous cannulation may also further impair resuscitation, but this was not analyzed in our study.
In their study on 625 trauma patients, Belzberg and colleagues showed reduced cardiac index and impaired tissue oxygenation with early organ failure in obese nonsurvivors when compared with obese survivors [
24]. They emphasized the need for continuous monitoring in obese patients using the cardiac index, central venous pressure, and peripheral tissue oxygenation as markers of adequate resuscitation. Correspondingly, Winfield and colleagues concluded in their study of 877 trauma patients that morbidly obese patients are inadequately resuscitated during the first 48 hours [
18]. These patients remained in metabolic acidosis longer, resolved their base deficit more slowly, and developed multiple organ failure at a higher rate than normal-weight patients. In this study population, nearly 90% of obese patients with ongoing metabolic acidosis developed multiple organ failure. The authors argue that pH may provide a more reliable marker due to the large variations inherent in central venous pressure and the influence of inorganic acids on metabolic acidosis, rendering lactate a less suitable parameter during the early resuscitation period.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
JN carried out the data acquisition, participated in subsequent data analyses, and provided a first manuscript draft. ATB helped in data acquisition and conducted initial statistical analyses, provided the figures and tables, and helped draft the manuscript. BS participated in the study design/supervised all statistical analyses and performed additional analyses where necessary, and helped author the statistics section. VN assisted in designing the study as well as analyzing and interpreting the data, and critically revised the final manuscript. OT helped in data interpretation and critically revised the manuscript. CH assisted in data interpretation and analysis as an expert in anesthesiology and intensive care medicine, and helped to write the manuscript. MT conceived the study, supervised data acquisition and analysis, and wrote the final manuscript. All authors read and approved the final version of the manuscript.