Introduction

Cardiopulmonary resuscitation (CPR) is vital for the survival of patients with sudden cardiac arrest (SCA). Chest compressions are the key component for providing flow of oxygenated blood to the brain and heart [1]. However, chest compressions are also traumatic. Previously reported CPR-related injuries range from non-complicated rib fractures to life-threatening injuries such as large pneumothorax or mediastinal haemorrhage [2,3,4].

Imaging after return of spontaneous circulation (ROSC) is being increasingly utilised to identify clinically occult causes of SCA. Although improving, the rate of survival to discharge remains low for those with ROSC following SCA. Only 10.8% of patients with out-of-hospital cardiac arrest and 22.3–25.5% of in-hospital cardiac arrest survive to discharge [1]. The goal of imaging is to identify a treatable cause of SCA and to rule out potentially complicating pathology, such as intracranial haemorrhage. Additionally, imaging may identify unexpected traumatic injuries related to CPR. It is unknown to what extent the injuries may affect survival in this patient population. The aim of this study is to assess traumatic injuries related to CPR in patients undergoing the non-traumatic SCA imaging protocol in our institution.

Materials and methods

The institutional review board (IRB) approved this HIPAA-compliant single-centre retrospective study with data acquired between January and September of 2016. The inclusion criteria were adult, non-traumatic, out-of-hospital SCA with ROSC who underwent SCA imaging protocol in our institution.

The SCA imaging protocol includes a WBCT including a retrospective ECG-gated CT arteriography of the chest, CT of the abdomen and pelvis in portal venous phase, and non-enhanced CT of the head. All CT images were acquired with a dual source dual energy CT, Siemens SOMATOM Force (Siemens Medical Systems, Erlangen, Germany). Chest CT was acquired in single energy mode and abdomen CT acquired in dual energy mode. Scan variables were as follows: 120 kVp (chest), Care KV modulation (abdomen); modulation mAs with reference of 160 (chest), 150 (abdomen); slice thickness 0.5mm (chest), 3mm (abdomen); rotation time 0.25s (chest), 0.5s (abdomen). Routinely, coronal and sagittal reconstructions were performed.

Study data was collected and managed using REDCap electronic data capture tools hosted at our institution [5]. REDCap (Research Electronic Data Capture) is a secure, web-based application designed to support data capture for research studies. The following data was collected from medical records: age, gender, height, weight, body mass index (BMI), use of anti-coagulants prior to arrest, cause of SCA, duration of CPR and survival to discharge. Statistical analysis was made using R Commander (Fox, J., and Bouchet-Valat, M. (2017). Rcmdr: R Commander. R package version 2.3-2). CT images were reviewed by two senior radiology residents together and by consensus to evaluate for traumatic injuries, including: location, number and type of rib fractures, sternal fracture, pneumothorax, haemothorax, aspiration, pneumomediastinum, mediastinal haemorrhage, haemopericardium, cardiac injuries, great vessel injuries and abdominal injuries. Rib fracture types included incomplete, complete non-displaced, and displaced. An incomplete fracture was defined as acute angulation without discrete fracture line. Displacement was defined by as at least one cortical width separation of fracture fragments.

Results

During the study period, 39 patients (13 female, 26 male) aged 23 through 83 years (mean 51.8) were enrolled. Twenty-three of the patients (59%) had non-cardiac cause of SCA and the remaining 16 (41%) had cardiac origin (Table 1). ROSC was achieved after less than 10 minutes of CPR in 17 patients (44%); 10 (26%) patients needed 10–20 min and in 11 (28%) more than 20 minutes of CPR was performed. According to clinical records, all patients underwent manual CPR. No use of mechanical chest compression devices was recorded. Eighteen patients (46%) survived to discharge and 21 (54%) deceased during hospitalisation. Detailed demographic variables are shown in Table 1.

Table 1 Patient demographics
Full size table

Rib fractures were the most common injuries (Fig. 1), present in 33 patients (85%). A total of 279 rib fractures were recorded. Rib fractures ranged from 0 to 13 per patient with a mean of 8.5 fractures per patient. Of these, 26 rib fractures (9%) were segmental. Patients with 7 or more fractures were significantly older than patients with less than 7 fractures, 56.16 vs. 46.57 years (p=0.046). Patients with segmental fractures were significantly older than patients without segmental fractures 67.5 vs 50.0 years (p=0.003). (Table 2).

Fig. 1.
figure 1

Schematic demonstrating cardio pulmonary resuscitation (CPR)-related injury distribution

Full size image
Table 2 Comparison of age for total rib fractures and presence of segmental fractures
Full size table

CT findings of CPR-related rib fractures follows a spectrum of severity and can be further classified in three categories (Fig. 2): a) incomplete rib fractures, 206 (74%); complete non-displaced fracture, 43 (15%); displaced fractures, 27 (10%). Also, 5 (2%) chondral fractures were noted. Distribution of the rib fractures is summarised in Table 3.

Fig. 2.
figure 2

Spectrum of rib fracture types. Axial contrast-enhanced CT from three different patients shows incomplete rib fracture (a, arrow), complete non-displaced fracture (b, arrow), and displaced fracture (c, arrow)

Full size image
Table 3 Rib fracture distribution
Full size table

Additional chest injuries were present including: 12 patients (31%) had sternal fracture, 5 (13%) mediastinal haemorrhage, 4 (10%) pneumothorax, 3 (8%) pneumomediastinum and 1 haemothorax. Also, abdominal injuries were found in three patients (8%). All three had liver injuries: one had American Association for the Surgery of Trauma (AAST) [6] grade II laceration, one a grade III laceration and the other one a grade III subcapsular haematoma with active bleeding and haemoperitoneum (Fig. 3). All detected injuries are included in Table 4.

Fig. 3.
figure 3

50-year old male underwent 10-20 minutes of CPR for sudden cardiac arrest (SCA) with return of spontaneous circulation (ROSC). Axial (a) and Coronal (b) contrast-enhanced CT demonstrates hepatic subcapsular haematoma (asterisk) with multiple foci of active extravasation (arrows) and adjacent displaced rib fracture (arrowhead). Patient survived to discharge

Full size image
Table 4 Injury summary
Full size table

Discussion

Traumatic injuries frequently result from CPR, most commonly rib and sternal fractures. A review by Hoke et al. [7] reported an incidence of rib fractures ranging from 13 to 97% and sternal fractures from 1 to 43%. More recent studies [3, 8,9,10] utilising post-mortem or post-resuscitation CT also show a wide variability with rib fractures present in 26.8–70% and sternal fractures in 4.2–30%. The reason for the variability is not known and likely multifactorial. Kim et al. [8] showed wide variation in the frequency of rib fractures among different hospitals. The incidence and outcome of cardiac arrest has been found to have significant regional differences [11]. The contributing variables such as training level of emergency medical services (EMS), time from call to EMS arrival, distance to hospital, prevalence of bystander CPR, and aetiology of SCA may account for some of the variability in CPR injury incidence.

A factor not addressed in prior studies is rib fracture type, which may contribute to the variability in reported rib fracture incidence. We found rib fractures in 84.6% of patients, which is at the upper end of the previously reported ranges. The majority of these fractures (74%) were incomplete and identified by focal increased rib angulation without discrete fracture line (Fig. 2). This can be a subtle radiologic finding and may not be reported as a fracture. Indeed, numerous fractures were only identified on a single plane of the multiplanar reconstructions (MPR). Use of curved planar reformats may be helpful with reported improved rib fracture detection sensitivity and reduced read time compared to MPR [12]. Although the clinical significance of these incomplete fractures has yet to be determined, traumatic rib fractures in general are independent risk factors for morbidity and mortality, particularly in the elderly [13], or when there are three or more [14].

When evaluating the number of rib fractures per patient, there were two general groups, those with no fractures and those with multiple. Of those with rib fractures only 15% (5/33) had one to five fractures while 76% (25/33) had eight or more fractures. Given this frequency of numerous fractures and the possible subtle radiologic findings, if a rib fracture is identified, there should be a high index of suspicion for multiple additional fractures. Similar to prior studies [8, 15], we found that patients with multiple rib fractures were significantly older.

Beyond rib and sternal fractures, multiple CPR related injuries have been described [2]. As expected based on proximity to the chest compressions, injuries most frequently arise in the chest and upper abdomen. We found a similar rate of thoracic injuries as previously described. The most common thoracic injuries are those associated with rib and sternal fractures including mediastinal haematoma (10.2–20.7%) and pneumothorax (2.5–12.2%) [16, 17]. We identified liver injuries in 3 patients (7.7%), which is higher than the reported incidence at autopsy of 0.6% [18] - 2.1% [19]. Other abdominal visceral injuries rarely occur with the spleen and stomach the next most commonly injured [19]. Most studies utilising CT have focused on thoracic injuries [3, 8,9,10, 16, 20] and were either performed without intravenous contrast or were limited to the thorax. However, Koga et al. [17] found secondary signs of abdominal injury, including haemoperitoneum in 14% of patients following CPR using post-mortem CT. Thus CPR related abdominal injuries on CT could be more common than previously described at autopsy. Larger studies utilising whole body contrast enhanced CT are required to better delineate the incidence of abdominal injuries.

All patients in our study underwent manual compression and bystander assisted CPR. No use of automated mechanical chest compression devices was recorded. Published data regarding CPR-related injuries after using mechanical chest compression devices shows that traumatic injuries are more common with use of these devices [21, 22], including abdominal injuries [17]. A recently published meta-analysis does not support that routine mechanical CPR improves survival or neurologic outcome [23]. A potential advantage of mechanical CPR is that WBCT can be acquired with minimal interruption to continued resuscitation [24].

No cardiovascular injuries were identified in our study. The most common cardiac injuries related to CPR include pericardial injury, haemopericardium, myocardial contusion, and conduction system injuries [16]. With the exception of haemopericadium, these injuries may not be identifiable on CT. Also our inclusion criteria of ROSC selects out most patients with severe cardiovascular injuries as the injuries may preclude spontaneous circulation. It is, however, important to be aware of these injuries, as they would require immediate intervention. Indeed successful repair of CPR-related ventricular rupture has been described [25]. Cardiac rupture occurs in up to 0.6% [16] of patients and although usually associated with prior ischemia/infarct, may arise without underlying wall weakness [25]. Traumatic vascular injuries including laceration, transection, pseudoaneurysm, and occlusion occur in the vessels immediately adjacent to the chest compressions. Coronary artery injury or His bundle haemorrhage has been found in up to 44% [26] of patients at autopsy. The aorta and internal mammary arteries are less commonly injured.

Bedside focused ultrasound to assess pericardial and peritoneal fluid plays a valuable role in screening trauma patients. Similarly, it may play a role in patients who underwent CPR to detect potential life-threating injuries like haemopericardium [16]. Benefits of ultrasound are well known: it is widely available, can be used at the bedside, avoids ionising radiation and is less expensive than CT. The use in the setting of CPR should be further explored.

Out of hospital CPR with ROSC carries a poor prognosis with only 10.8% [1] of patients surviving to discharge. Although we found a higher survival of 46%, the sample size is too small to determine any significance. The purpose of the SCA imaging protocol is to identify the SCA aetiology and direct early intervention. Whether this has been successful will be evaluated after larger patient enrolment. An unanticipated role of the imaging protocol has been to identify a variety of CPR related injuries. The effect of the injuries on outcome is not (yet) known and thus the role of routine imaging for CPR related injuries remains unclear. Only a small fraction of the identifiable injuries necessitate emergent intervention. However, with as many as 84.6% suffering rib fractures, the potential clinical impact should be considered. Indeed each additional rib fracture in the elderly increases the mortality risk by 19% and the risk of pneumonia by 27% [13]. Also, as up to 76% of those with rib fractures had eight or more, there may be a role for surgical intervention. Early rib plating for severe rib fracture patterns, such as flail chest, has demonstrated improved morbidity including reduced time on ventilation, days in ICU and rate of pneumonia [27]. Further studies are required to delineate the impact of CPR-related injuries on outcomes and possible early intervention. However, radiologists should be aware of the spectrum of CPR-related injuries, as well as their known and potential clinical impact.

Conclusion

CPR-related injuries on WBCT after ROSC are common, with serial rib fractures detected most commonly. An unexpectedly high rate of abdominal injuries was detected on SCA-WBCT. As more institutions adopt SCA imaging protocols, radiologists should be aware of the spectrum of CPR related injuries.