SNOM of the hemodynamically stable patient with splenic injury is widely accepted as the current standard of care. At our level 1 trauma centre, the proportion of severely injured trauma patients managed non-operatively increased over time, from 50% in 1996 to 78% in 2007. This finding is consistent with other published studies [22, 23]. It has been suggested that improved imaging and detection of splenic injuries has led to a decrease in the severity of splenic injuries that present to trauma centres over time [24, 25] which may account for an increase in the use and success of SNOM. We did not observe any trend in injury severity, with median ISS and maximum abdominal injury scores generally stable over time. We did however, observe an increasing proportion of patients receiving SAA over time which may, in part, account for the shift to SNOM as use of adjuvant modalities allows for continued SNOM in patients with ongoing hemorrhage .
The overall success rate of SNOM was 87% and associated mortality rate was 4%. Both these success and mortality rates are similar to other published series [6, 25, 27, 28]. Cocanour and colleagues (2000) had a similar patient population with a mean ISS of 20.5 ± 1.1 for all patients with blunt splenic injury (vs. a median ISS of 27 in our population). Their SNOM success rate was also 86% (for 198 patients undergoing SNOM) and the mortality rate was 5%. This study defined a failure of SNOM as requiring splenectomy or splenorrhaphy more than four hours after admission (as we did) but excluded patients who died within 24 hours of admission from massive injuries. Haan and colleagues (2005) found a higher (90%) success rate in SNOM. Their higher success rate may be because of lower overall severity of injuries in their patient population (overall ISS for the 648 patients in Haan and colleagues (2005) was 17 vs our overall ISS of 27) and their higher use of direct OM in their patient group (43% vs only 24% of our patients receiving OM). This difference in success may also relate to our lower than expected rate of adjunctive therapies. Approximately 20% of the patient population in Haan and colleagues (2005) received splenic artery embolization. Although only a small proportion of our SNOM patients were treated with SAA/SAE however, this proportion increased over time, a trend that we expect to continue, as SAA/SAE becomes more accessible to more trauma patients with the advent of interventional/trauma operating/resuscitation rooms .
Overall, approximately one third of SAA patient experienced complications (none requiring operative intervention). This is comparable to the rates of major complication (20% to 30%) reported in the literature [5, 27, 30–32]. Despite the widespread adoption of SNOM, patient selection criteria and the specifics of SNOM treatment protocols are not consistent between institutions [27, 33, 34]. Without a formal instiutional protocol, the decision to pursue SNOM is left to the discretion of the attending physician. Thus, the reported use and success rates of SNOM strategies for blunt spleen trauma are highly variable. Reviewing the CT findings of our early (< 48 hours) SNOM failures revealed six cases of seemingly unaddressed contrast extravasation or presumed traumatic pseudoaneurysm in otherwise hemodynamically stable patients who later required OM in the earlier years of our review. Although unproven, it is conceivable that early SAA/SAE may have obviated the need for OM, as it might have for the four patients who developed delayed extravasation more than 48 hours after hospital admission. This rationale has led to our introduction of education initiatives and a regional guideline for the management of splenic trauma which includes recommendations regarding addressing these findings in otherwise stable patients [Additional file 2 and Additional file 3].
As hemodynamic instability is the major factor in the decision to opt for OM of splenic injuries, we collected data on fluid and blood requirements during the first four hours of admission. We believe this was an appropriate proxy for the interval during which the decision is typically made to pursue SNOM or OM. As expected, there was a greater need for RBCs and fluid amongst those who received OM, likely due to the greater severity of their injuries. Balaa and colleagues (2004)  found that patients who failed SNOM did not require more blood overall than those who received OM initially, however, only 5 of 65 patients in their study failed SNOM. In our study, the 53 patients who failed SNOM required less RBCs and crystalloid fluids than those initially selected for OM, yet more than those who underwent successful SNOM. Demographically, the group of individuals who failed SNOM were slightly older with a median age of 38 (IQR, 25-52) than the successful SNOM patients. This is in agreement with previous studies .
Generally, failure of SNOM is reported to occur within the first 48 hours [33, 35, 36], and we encountered several cases of failure only hours into the hospital admission. These very early failures preclude making comprehensive recommendations regarding early screening for vascular injuries or extravasation amenable to percutaneous therapies, other than to mandate vigilant monitoring and to ensure that an immediate surgical or combined surgical/angiographic response is available. Even if these early failures can be averted however, late failure can occur several days and even weeks or months following injury [19, 35]. Of the seven cases of late failure that we experienced, five patients required splenectomy and two were managed with SAA and proximal coil SAE. Our seven cases of delayed splenic rupture occurred five days to two months following the initial injury, four during their hospitalization and three after hospital discharge. Whether vascular lesions pre-disposing to late failure were present in these late failures however, is uncertain due to our limited use of follow-up imaging prior to discharge.
Thus 2% percent of our SNOM patients residing within Calgary were readmitted within 6 months with a delayed splenic complication, and all required either a splenectomy or angioembolization. This rate of late failure of SNOM is particularly concerning given the trend toward early discharge from hospital. These concerns are corroborated by another study by Zarzaur and colleagues (2009)  who examined Tennessee hospital readmission data for blunt spleen trauma patients and found that 1.4% of patients (n = 1932) who received SNOM and were discharged home, were readmitted for splenectomy within six months of discharge. Other small case series have described results consistent with our readmission and surgical intervention rate [16, 37].
The need for routine follow up imaging for blunt spleen trauma has been widely debated. We found that patients with higher grade injuries were more likely to receive follow-up imaging during their hospital admission. In our subset analysis of SNOM patients with abdominal CT imaging available for grading (total 256 patients), overall 50% (129 patients) received follow-up imaging during their hospital admission. There is good evidence to suggest that follow-up imaging in the two to five days following injury is valuable for detection of splenic pseudoaneurysm [11, 13, 17]. In our institution we now recommend that repeat CT imaging is obtained within 72 hours in all grades of splenic injury, based on reports documenting the detection of splenic artery pseudoaneurysms even in low Grade injuries . Our protocol also permits the use of dedicated ultrasound exams in selected young patients Grade I and II injuries instead of CT scan based on clinician judgment. It is questionable whether the cost and stochastic risks associated with late abdominal CT imaging is warranted in stable, SNOM patients after discharge from hospital when imaging typically confirms resolution of injury. Other studies have found that CTs taken more than ten days following presentation are generally not used to determine patient treatment or add clinically relevant information [38–40]. Thaemert and colleagues (1997)  found that improvement in the appearance of the spleen was evident in all 33 follow up CT scans taken more than 10 days following the injury. Post-discharge imaging was obtained less frequently among our patients (about 17%), with the majority of results confirming improvement or resolution of the spleen injury. Interestingly, SAA patients are overrepresented amongst those who received imaging post-discharge. Eight of the patients in our cohort who received follow-up CTs during the post-discharge period had received SAA on their initial admission. Post-discharge follow-up imaging has been advocated in certain patients to guide return to certain activities and at our institution this is left to the discretion of the attending surgeon. Of the three patients who experienced late-failure in the post-discharge period, all late failure splenic ruptures were detected following onset of patient symptoms. Nevertheless, late follow up imaging may be of value in patients considering return to contact sports or other activities where they may be at risk of re-injury.
The retrospective nature of the data is the greatest limitation of our study. As the decision to treat operatively or not was rarely explicitly noted in the medical record, the definition of what did and did not constitute a decision to treat a patient non-operatively was debatable (i.e., if no surgery within four hours) and based on commonly reported literature. Some patients requiring surgery within this interval may have been selected for SNOM but experienced an unexpected deterioration leading to operative intervention. Alternatively, some patients who went to the operating room outside of the four hour window may have been receiving stabilizing treatment for other injuries, with spleen surgery planned as soon as possible. Accordingly, our estimates of the use of SNOM may not be completely precise. However, it is unlikely that such misclassification was differential over time, and the trend towards increased use of SNOM over time is likely valid. Furthermore, our results may not be generalizable to patients with less severe injuries or patients with isolated spleen injuries since the Alberta Trauma Registry identifies all trauma patients with injury severity scores greater than 12.
At our level 1 trauma centre, the majority of severely injured patients with blunt splenic injuries are currently managed non-operatively with an overall success rate approximating 87%. The low utilization of SAA/SAE potentially offers opportunities to improve this rate in the future. The majority of SNOM failures occur within 24 hours of hospital admission, however, we did observe seven cases of delayed splenic rupture requiring intervention. Although low, rate of delayed splenic rupture and risk of complications resulting from these occurrences suggests that the need for patient education around symptoms of delayed splenic rupture may prove to be increasingly important in our move toward early discharges.