In this retrospective analysis we present data on 30-day survival in 53 patients treated with ECMO at a Level 1 trauma center. Our data shows significantly higher survival patients treated with VV than with VA ECMO. There was also a significant difference in survival based on the indication for ECMO support, and no difference was observed among patients cannulated at a referring hospital and then transferred to our hospital compared to those cannulated in-house. Furthermore, we found no difference in survival outcomes throughout the study period. According to the ELSO Registry, overall survival for adult ECMO patients in North America and Europe over the last 5 years is 53% [28]. This is comparable to our results of 51%, and results previously reported by others [29–30]. Hence, our results provide further evidence that ECMO can be used successfully to support patients with very severe trauma, in particular, patients with primary lung pathologies.

Major trauma is the primary cause of death among young adults. The use of ECMO in trauma, while still controversial, has become more prevalent over the past decades. However, most reports are case studies or retrospective analyses with a limited number of patients, usually less than 20 subjects [29]. With the increased interest in ECMO for trauma there is a need for clearer indications to identify the patients most likely to benefit from this support.Trauma patients are heterogeneous both with respect to comorbidities, age and mechanism of injury. The mechanical impact of a motor vehicle accident has a different impact from a stab wound or a drowning but the common denominator is the problem of oxygen delivery, for which ECMO provides an unique approach.

To quantify the severity of trauma in this patient population we chose the widely used ISS and NISS for easy comparison to previous studies. Our sample population, although small, was similar to the trauma population in Sweden with respect to mechanism of injury with the most common cause being traffic accidents (ca. 50%) followed by falls. However, the sex distribution in our sample was more heavily skewed towards males than that which is reported by the Swedish trauma registry (i.e. roughly two thirds). Furthermore, our sample was comparatively young with a median age of 24 compared to 37 for the trauma population at large [31]. Furthermore, the trauma severity in our population was quite high. The mean NISS and ISS assessed in this study were 46 and 44, respectively, thus representing a very high risk of mortality as shown by Copes et al., and Ghorbani [11, 32]. Overall, 41% of the patients had an ISS greater than 50, which otherwise was associated with almost 100% mortality rate [11]. According to the Swedish trauma registry roughly 12,000 patients are brought to hospital due to trauma per year, of which about 2–3% have an NISS > 40 [31]. Even though ECMO is not a typical supportive modality in the trauma patient, the results showed more than half of the patients were still alive at follow-up. Furthermore, we divided both the ISS and NISS into intervals according to Copes et al. showing that there was similar distribution with a slight tendency towards the higher bins within the VA group, but with a higher proportion 30-day survival for the patients on VV ECMO (Tables 3 and 4). The cause of higher mortality in the VA group might be partially explained by the fact that most patients with cardiac arrest and ongoing CPR, a subgroup with a very high mortality rate, were placed on VA ECMO.

Almost two thirds of the patients were placed on VA ECMO in contrast to earlier studies where pure respiratory support was more common due to ALI/ARDS [29–30]. The ELSO Registry reported an overall survival rate of 43% for all VA ECMO patients, and 30% for patients treated with extracorporeal cardiopulmonary resuscitation [28]. Previous studies on the use of VV ECMO in the trauma population show a survival range of to 56–89% compared to 42–63% for VA ECMO [29]. This was similar to our findings of 75% 30-day survival in the VV ECMO group, and 36% for VA ECMO. This difference may be explained by physiological differences in the patients, whereby patients requiring VA ECMO had severe hemorrhagic shock, cardiac arrest, cardiac failure or a combined cardio-pumlmonary failure secondary to injuries. Trauma patients in cardiac arrest had a particularly poor survival outcome, and after the exclusion of the patients with CA, survival from VA ECMO was 61% as compared to 75% on VV ECMO. Taken together, this may reflect that trauma patients who present with lung problems, and who are then placed on VV ECMO are more likely to benefit from ECMO support.

ECMO support increases the risk of bleeding complications and subsequently the risk of cerebral hemorrhage [33]. Hence, most clinicians may feel apprehensive to offer ECMO to trauma patients with head injuries. In this study, most of the patients were multi-trauma, and nearly half presented with head injuries. Three patients were categorized as isolated head injuries, which is a relative contraindication for ECMO. Despite this, one of these patients survived. Of the 26 patients with head trauma, 54% survived 30 days after discharge, a survival rate, similar to the overall survival rate for the study population (51%). Previous studies have shown benefits in patients with traumatic brain injury (TBI), both from VV and VA ECMO [34–35]. Considering the patients with TBI in both this study and previous studies had an outcome comparable to the typical trauma patient supported on ECMO, the risk-benefit of ECMO use in patients with TBI should be re-evaluated. ECMO can safely be used for several days without anticoagulation, particularly in the coagulopathic trauma patient, and can therefore be considered in patients with massive hemorrhage and TBI [36].

Another area of limited knowledge is the outcome of trauma patients transported on ECMO. We observed no difference in mortality between those patients cannulated and then transported on ECMO compared to those who were cannulated in house. The mobile ECMO team in this study has extensive experience including both in-hospital cannulations and multiple ECMO transports of non-trauma patients, and has only recorded three deaths during transport over a 30-year period, of which one was a trauma patient [37,38,39]. The analysis showed that cannulations at referring hospitals were generally performed at a later stage (> 12 h post trauma), and the patients still alive had been stabilized by local intensive care for several hours and thus more stable than the patients brought directly to our university hospital from the trauma site. This finding may be subject to selection bias influenced by more strict selection criteria where the time factor and prognosis would strongly impact a decision to dispatch the mobile ECMO team. Nonetheless, the high survival rate (69%) suggests that with strict selection criteria a mobile ECMO team that transports patients to a larger ECMO center may be a life-saving strategy for patients suffering from severe trauma.

Limitations of this work include the sample size, which was relatively low from a general perspective but rather high given the targeted ECMO population from a single center. The applicability of findings from a high-volume center may not extend to institutions with less experience, potentially limiting their generalizability. Our ECMO management has developed over the years as a slow continuous process based on our own experiences and influences from international collaborations, others’ experiences, and research. Our unit does not manage ECMO according to any preset treatment protocol since extracorporeal life support (ECLS) is what we do every day being one of extremely few intensive care units (ICU) worldwide dedicated exclusively to ECLS. This fact brings a limitation regarding generalizability of the results from this study. Other limitations may be patient heterogeneity, mechanism of injury and age (1–72 years). Further limitations include the time period of the study, 22 years, during which changes in practice occurred, and that patients recovered from other hospitals may have been subjects of selection bias. The key strengths of this study were the Swedish Civic registration number, which allowed for long-term follow-up of patients, and the local databases for trauma and ECMO support.