Is chest imaging relevant in diagnosing acute respiratory distress syndrome in polytrauma patients? A population-based cohort study

Purpose The definition of acute respiratory distress syndrome (ARDS) has often been modified with Berlin criteria being the most recent. ARDS is divided into three categories based on the degree of hypoxemia using PaO2/FiO2 ratio. Radiological findings are standardized with bilateral diffuse pulmonary infiltrates present on chest imaging. This study investigated whether chest imaging is relevant in diagnosing ARDS in polytrauma patients. Methods The 5-year prospective study included consecutive trauma patients admitted to a Level-1 Trauma Center ICU. Demographics, ISS, physiologic parameters, resuscitation parameters, and ARDS data were prospectively collected. Acute hypoxic respiratory failure (AHRF) was categorized as Berlin criteria without bilateral diffuse pulmonary infiltrates on imaging. Data are presented as median (IQR), p < 0.05 was considered significant. Results 267 patients were included. Median age was 45 (26–59) years, 199 (75%) males, ISS was 29 (22–35), 258 (97%) patients had blunt injuries. Thirty-five (13%) patients died. 192 (72%) patients developed AHRF. AHRF patients were older, more often male, had higher ISS, needed more crystalloids and blood products than patients without AHRF. They developed more pulmonary complications, stayed longer on the ventilator, in ICU and in hospital, and died more often. Fifteen (6%) patients developed ARDS. There was no difference in outcome between ARDS and AHRF patients. Conclusions Many patients developed AHRF and only a few ARDS. Patients with similar hypoxemia without bilateral diffuse pulmonary infiltrates had comparable outcome as ARDS patients. Chest imaging did not influence the outcome. Large-scale multicenter validation of ARDS criteria is warranted to investigate whether diffuse bilateral pulmonary infiltrates on chest imaging could be omitted as a mandatory part of the definition of ARDS in polytrauma patients. Electronic supplementary material The online version of this article (10.1007/s00068-019-01204-3) contains supplementary material, which is available to authorized users.


Introduction
Acute respiratory distress syndrome (ARDS) is clinically characterized by severe dyspnea, cyanosis refractory to oxygen therapy, loss of lung compliance, and diffuse alveolar infiltrates on chest radiograph [1]. This is caused by an increased permeability of the alveolar-capillary barrier resulting in lung edema with protein-rich fluid causing an impairment of arterial oxygenation. Lung edema, endothelial and epithelial injuries are accompanied by an influx of neutrophils into the interstitium and broncho-alveolar space. Activation and recruitment of neutrophils play an important role in progression of ARDS [2].
Over the years, the definition of ARDS has often been modified with Berlin criteria being the most recent [3,4]. To facilitate prognosis, ARDS was divided into three categories based on the degree of hypoxemia using the partial pressure of arterial oxygen and fraction of inspired oxygen (PaO 2 / FiO 2 ). Additionally, radiological findings were standardized and bilateral diffuse pulmonary infiltrates on chest X-ray or CT scan (without evidence of heart failure) have to be present [4][5][6]. Like previous definitions, the Berlin ARDS criteria also have been criticized, and several authors have questioned its usefulness [7][8][9]. Historically, ARDS has been a significant cause of trauma-related morbidity and mortality, with reported mortality rates up to 40% [10,11]. Many treatment strategies have been created; however, most of them had limited success [12]. Nowadays, the treatment is still limited and mainly consists of supportive mechanical ventilation with low tidal volume and inspiratory pressure ventilation. With improvement of trauma, critical-care mortality caused by ARDS has decreased in the last years [5,[13][14][15][16][17][18]. However, it still uses significant intensive care unit (ICU) resources.
In contrast to several studies reporting continuously high ARDS-related deaths [5,[13][14][15][16][17]19], we recently showed low incidence of ARDS and low mortality rates [18]. However, we did observe many polytrauma patients who developed some degree of acute hypoxic respiratory failure (AHRF) without the bilateral diffuse pulmonary infiltrates that characterize ARDS. We felt that these AHRF patients were similar to ARDS patients except for the bilateral diffuse pulmonary infiltrates. Therefore, we conducted a prospective study in polytrauma patients in which we compared outcome parameters of ARDS patients with similar degree of hypoxic respiratory failure, but without the classical radiological findings (AHRF patients). We hypothesized that there was no difference in outcome between both groups. Further, we hypothesized that chest imaging was only relevant to detect abnormalities that need additional treatment other than supportive mechanical ventilation with low tidal volume and inspiratory pressure ventilation.

Study setting
The study was conducted at an urban major (Level-1) trauma center. From November 2013, a 5-year prospective population-based cohort study was undertaken including all consecutive trauma patients who were admitted to the intensive care unit (ICU) of the University Medical Center Utrecht. This major trauma center is the only Level-1 trauma center in the province of Utrecht and covers the central region of the Netherlands with a relatively small, but densely populated service area of 1500 km 2 and approximately 1.3 million residents. The service area for neurosurgery facilitates 2.1 million residents. Around 1300 trauma patients with full activation of a trauma team are annually admitted. Approximately, 375 of them are multiply injured (ISS > 15) [20]. Patients included in the study were all admitted to ICU either directly from the emergency department (ED) or postoperatively after urgent surgery was performed. Patients < 15 years of age, isolated injuries caused by asphyxiation, drowning, burns or isolated traumatic brain injury (TBI) were excluded. Patients who died within 48 h were excluded as well.

Data collection
All data were prospectively collected and included patient demographics, injury severity score (ISS), shock and resuscitation parameters. Admission arterial blood gas analysis, coagulation status, and temperature measurement were performed during resuscitation in ED as part of standard procedures. Arterial blood gas analysis and temperature measurement were repeated on arrival in ICU. Urinary output was measured in the first hour after arrival in ICU. Blood product (packed red blood cells (PRBC), fresh frozen plasma (FFP), and platelets (PLT)) use was recorded in the first 24 h following admission. The Denver multiple organ failure (MOF) scores and ARDS Berlin criteria were registered daily up until 28 days or discharge from ICU. Chest X-rays were performed on a daily basis while in ICU. If necessary, additional CTs of chest were performed. Chest X-rays were selected for review 2 days before and after the development of worst hypoxia. All imaging studies were reviewed by the radiologist on call who was blinded to the pulmonary condition. Primary outcome was development of AHRF or ARDS. Secondary outcomes were mortality, pulmonary complications (pneumonia, pulmonary embolism, thorax empyema), ventilator days, ICU length of stay (ICU-LOS), in-hospital length of stay (H-LOS), and multiple organ dysfunction syndrome (MODS).
Acute hypoxic respiratory failure (AHRF); hypoxemia divided in three categories similar to ARDS, but without classical bilateral diffuse pulmonary infiltrates on X-ray.
Multiple organ dysfunction syndrome (MODS) was defined by Denver multiple organ failure (MOF) scores of greater than 3, occurring more than 48 h after injury [9]. Denver MOF score was chosen over sequential organ failure assessment (SOFA) to avoid difficulties by including the Glasgow coma scale (GCS) in the organ failure score. GCS can be challenging to obtain in the trauma patient in ICU, because they are often sedated and intubated for extended periods. This could negatively influence the CNS organ failure score [21].
Urgent laparotomy was defined as a laparotomy that was performed in patients who were transported from ED directly (or via CT scan) to the operating room (OR).
Pneumonia was defined as the development of purulent sputum (with positive cultures of sputum) in conjunction with radiological evidence of a new or progressive pulmonary infiltrate.

Ethical approval
The local ethics committee approved this prospective observational study (reference number WAG/mb/16/026664).

Statistical analysis
Data were analyzed using IBM SPSS Statistics version 25.0 (Armonk, NY, USA). Graphs were prepared with GraphPad Prism version 7.04 (San Diego, CA, USA). Results are presented as median and interquartile range (IQR). Comparison of variables was done using Kruksal-Wallis test or Pearson's Chi-square test in dichotomous data. Statistical significance was defined as p < 0.05.
Patients who later developed AHRF/ARDS had lower PaO 2 levels, lower pH, and lower saturation both on arrival in ED and ICU, lower blood pressure and higher PaCO 2 in ED ( Table 2). In ED, base deficit (BD) was similar between both groups, whereas on arrival in ICU, BD was lower in patients who later developed AHRF/ARDS (Table 2).
Time to AHRF onset was early after trauma; AHRF developed 3 (2-5) days from injury with a length of 2 (1-4) days (Fig. 2a). Twenty-one patients (17%) had AHRF for more than 5 days and 11 patients (9%) had AHRF for more than 5 consecutive days (Fig. 2b). This early onset and short duration of AHRF is comparable to ARDS onset and duration in ARDS patients as we have demonstrated in a previous study [18].

ARDS vs. AHRF patients
Patients who later developed ARDS had higher PaCO 2 in ICU (49 (44-56) vs. 43 mmHg (40-47), p = 0.004) than patients who developed AHRF but no ARDS. All other parameters such as demographics, injury severity score (ISS), shock and resuscitation parameters, pulmonary complications, and outcome showed no difference between ARDS and AHRF patients.
One of 15 ARDS patients (7%) died of ARDS. He developed severe ARDS 3 days after massive aspiration after fresh-water submersion after a motor vehicle accident with brain injury, maxillofacial, and several cervical spine fractures including dissection of a vertebral artery. He had ARDS for 4 days with lowest PaO 2 /FiO 2 ratio of 40 and despite venovenous extracorporal life support he died 7 days after admission. Further, two other patients died while having ARDS; one patient suffered from multiple injuries including severe brain injury, and treatment was withdrawn as it was considered as medically futile. The other patient    (20) 0.001* failed to wean from the ventilator after C2 cervical spine injury with myelum contusion resulting in tetraplegia. One patient with AHRF died of acute hypoxic respiratory failure (3%). He died 45 days after injury after a combination of aspiration and cardiac failure. Thirty-one other patients died while having AHRF; 27 patients died of severe brain injury, 1 of sepsis, 1 of MODS, and 2 patients failed to wean from the ventilator after high cervical spine injury with myelum contusion.

Relation chest imaging and AHRF
By definition, all ARDS patients had bilateral diffuse infiltrates on chest imaging. Fifty-eight patients (77%) who had no acute hypoxic respiratory failure had normal chest X-rays, whereas 15 patients without AHRF (23%) showed signs of pulmonary contusion or atelectasis on X-ray (Table 3). Hundred and twenty-five AHRF patients (89%) had abnormal chest X-rays, whereas 15 patients (11%) with AHRF had normal X-rays. Interestingly, 13 patients (38%) with pulmonary contusion on chest imaging had no hypoxemia, whereas all patients who developed pneumonia or signs of cardiac failure (bilateral pleural effusion) had some degree of hypoxemia (Table 3). Any discrepancy between the number of pneumonias registered as complication and the number of pneumonias on chest imaging can be explained by the fact that pneumonias, on imaging, were only calculated if present at the time of hypoxia. Sensitivity of chest imaging for diagnosing the cause of the hypoxic respiratory failure was 89%, specificity 77%, positive predictive value 88%, and negative predictive value 79%.

Discussion
In this population of severely injured patients, 72% of patients developed some degree of AHRF. These AHRF patients were older, more often male, had higher ISS, needed more crystalloids and blood products than patients who did not develop AHRF. Further, they developed more often pulmonary complications, stayed longer on the ventilator, longer in ICU and in hospital, and died more often. Most patients died of brain injury even though we excluded patients with isolated TBI. The percentage of brain/spinal cord injury-related deaths was higher than most reports about the cause of death in polytrauma [22,23]. We have shown this in previous studies [18,24], and this could be partly explained by the fact that our level-1 trauma center is the only referral center for brain and spinal cord injuries in the state [20].
In contrast to high incidence of AHRF, the incidence of ARDS was low (6%). When comparing AHRF patients with ARDS patients, only PaCO 2 in ICU was different, with ARDS patients having a higher PaCO 2 . All other parameters including resuscitation and outcome were comparable. In other words, the only other difference between ARDS and AHRF patients was bilateral infiltrates on chest X-ray since this is part of the definition of ARDS. One could only speculate why ARDS patients had higher first-measured PaCO 2 in ICU; is it because the influx of neutrophils into the lung has already started and ARDS is already in progress? Or do these patients develop ARDS, because they did not receive optimal low tidal volume and inspiratory pressure ventilation from the start?
Another interesting finding was that 11% of patients who developed grade 2 or 3 AHRF did not have any abnormalities on their chest X-ray at the time of hypoxia. Further, AHRF developed early (3 days) after injury. Since early AHRF may be related to fluid overload rather than complications such as pneumonia, a possible explanation could be that early pulmonary edema was already causing hypoxia, but was insufficient to be seen on X-ray. We feel that it would be more useful to use criteria of hypoxemia without the bilateral infiltrates on chest imaging. The fact that a patient is identified as having ARDS rather than AHRF in general does not influence ventilator settings since lung-protective ventilation nowadays is the standard of care anyway. In this study, comparison between patients with and without AHRF showed that AHRF alone has the discriminative power to separate sick from the not-so-sick patients. Another argument for dismissing imaging out of the definition is the fact that many trauma patients sustain pulmonary contusions. Early after injury, it is very difficult to differentiate between bilateral infiltrates caused by ARDS or by bilateral pulmonary contusion. Vice versa, there is also the potential danger of labeling a patient as having ARDS based upon bilateral infiltrates which originate from a different source (such as pulmonary contusions). Obviously, chest imaging itself is still very valuable to diagnose entities such as pneumothorax and hemothorax that need additional treatment to lung-protective ventilation. We, therefore, do not suggest to reduce the use or abandonment of chest imaging in the trauma patients.
Further, a recent study showed that inter-observer reliability of diagnosing ARDS was only moderate (kappa 0.50). The disagreement between clinicians was explained by differences in how chest imaging studies were interpreted [9]. Further, Chung et al. have demonstrated that CT findings cannot differentiate between pathology-proven diffuse alveolar damage as can be seen in ARDS and histopathological features of pneumonia [25].
Like others [4,5,19,26], we have also shown that ventilator days, days in ICU and in hospital increased with increasing grade of hypoxemia. The question is whether grading of hypoxemia according to the Berlin criteria is  really relevant. It might be helpful for predicting prognosis, but not for diagnosis of ARDS since treatment strategies are limited and not reserved for a specific grade of hypoxemia. This has also been previously addressed by others [6]. Also, both autopsy study [27] and biopsy study [28] demonstrated that the Berlin criteria did not correlate with the presence of diffuse alveolar damage in more than 50% of patients categorized as moderate or severe ARDS.
These arguments raise the question whether the definition of ARDS is specific enough. Although the definition of ARDS, which is based on consensus criteria, is useful for screening, it is less suitable as a diagnostic test. ARDS is a heterogeneous syndrome and it is unlikely that it will be amenable to a single intervention. If consensus definitions of a heterogeneous syndrome are not specific enough, patient selection for interventional trials could be incorrect resulting in false-negative results. This issue has also been addressed by Laffey and Kavanagh in their comments on many negative trials in critical care [29].
AHRF patients received more crystalloids < 8 h and < 24 h after admission than no-AHRF patients; however, it was noted that crystalloid infusion was high in all the included patients. During the first 24 h after admission most patients were in ICU where intensivists decided on resuscitation, since there is a closed-format ICU in our hospital. During the beginning of the studies, there was an ongoing debate between trauma surgeons and intensivists on resuscitation in trauma patients. However, in recent years, there has been an increased awareness of the detrimental effects of excessive amount of crystalloids in severely injured patients. Nowadays, hemostatic resuscitation is aimed for in all polytrauma patients.
One of the limitations of this study is that it was conducted in polytrauma patients only. ARDS is a syndrome that occurs in a wide spectrum of patients of whom the majority are not trauma patients. In some patient groups, other than trauma, the finding of bilateral pulmonary infiltrates is relevant for diagnostic and prognostic purposes.
We acknowledge that our data only address a small subset of patients suffering from ARDS.
Further, this study was performed at a single institution in which the clinical treatment and research were conducted by the same clinicians. Patients with isolated head injuries were excluded, because of possible different physiologic response to trauma, although we did include polytrauma patients who had associated head injuries. We chose to include these patients, since many of our severely injured patients have associated brain injury and they are also prone to ARDS. Another limitation is that we neither prospectively collected data on tidal volume and plateau pressures nor did we collect data on comorbidities that could have contributed to hypoxia such as COPD or preexisting heart failure.

Conclusions
In conclusion, in this polytrauma population, many patients developed AHRF and only a few ARDS. Chest imaging did not seem to influence treatment strategies and outcome, since patients with similar hypoxemia but without bilateral diffuse pulmonary infiltrates had similar outcome as ARDS patients. Large-scale multicenter validation of ARDS criteria is warranted to investigate whether diffuse bilateral pulmonary infiltrates on chest imaging could be omitted as a mandatory part of the definition of ARDS in polytrauma patients.
Author contributions All authors have made substantial contributions: KW and LL have designed the study. KW has performed acquisition of data. KW has done the analysis and interpretation of data. KW has drafted the article. KW and LL have revised it critically for important intellectual content. KW and LL have given final approval of the version to be submitted.

Funding
The authors have not received any funding for this work. Availability of data and material The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Compliance with ethical standards
Conflict of interest The authors declare no conflict of interest.
Ethical approval and consent to participate The local ethics committee approved this prospective observational study (reference number WAG/ mb/16/026664). This information is also included in the methods and materials section.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.