Spinal epidural abscess (SEA) is a severe disease, which may have enormous implication on patient’s outcome and health related satisfaction [4-8,10,12,16-18]. Despite improvement in medical knowledge, technical and surgical techniques the management of SEA still remains challenging [2,15,31,33-36]. The total incidence of the disease amounts 2–3 cases per 10000 admissions. The average age at admission is above 50 years with a male predomination [8,29,33]. The most common abscess location is the lumbar spine, followed by the thoracic spine and the cervical spine [1,4,6,20,24]. Patients with severe pain or a neurological deficit, such as dysesthesia or motor weakness mostly undergo early surgery. However, controversial data have been published on the optimal timing for surgical treatment of SEA. More specifically, the literature disagrees on the beneficial effect of early surgery on SEA [3,15,25,30,32,35,36]. The aim of this study is to evaluate both clinical outcome and health-related quality of life (QOL) in patients following early versus late surgical treatment for SEA.

This study was approved by the Local Ethics Committee of Goethe University Frankfurt. A retrospective analysis was performed for all cases treated for SEA between 2007–2016. Patient charts were assessed for clinical, surgical, radiological and microbiological data. The neurological status was assessed using the American Spinal Injury Association Impairment Scale (AIS). This assessment was performed before surgery, postoperatively and at the time of discharge. According to our treatment algorithm, magnetic resonance imaging and computed tomography (CT) scans were performed to confirm the diagnosis, extent of infection and bony destruction. Indication for surgery were compression of nervous structures causing neurological deficits or severe pain, bony destruction and obvious signs of instability. Based on the clinical status at first presentation, the procedure was carried out either as an emergency surgery or as an elective surgery after admission. In this study we chose to evaluate the impact of early surgical treatment with an even shorter interval than 24 hours after admission being defined as 12 hours after admission. Early surgery was defined as surgery being performed within 12 hours after admission, late surgery was defined when the procedure was performed 12 hours after admission. Patients with severe neurological deficits, intractable pain and obvious predominant osteomyelitis or spondylodiscitis were assigned to early surgery group, whereas patients with mild symptoms, no symptoms and evident onset of neurological deficits longer than 72 days were assigned to the delayed surgery group. Thus, the onset of the symptoms and neurological deficits was another important factor for performing early or delayed surgery. Further, most reasons for delayed surgery was either failure of nonoperative treatment, preoperative work up and evaluation of the physical condition in cases of multimorbidity and delay in diagnosis.

Conservative therapy was either chosen by patients itself refusing surgery, harboring comorbidities with increased risk for perioperative complications or revealing no neurological deficits.

In the further follow-up period most patients with AIS E (mild or no symptoms) underwent surgery in order of severe pain and proven increase of bony destruction. Samples for bacteriological work-up were obtained by blood culture, during surgery or by CT-guided biopsy in conservatively treated patients. Calculated broad-spectrum antibiotics were administered until receiving the result of microbiological examination. Based on the result of the individual antibiogram the broad-spectrum antibiotics therapy was deescalated to pathogen specific antibiotics. The antibiotic regimen was routinely applied as intravenous treatment for 6 weeks followed by 6 weeks of oral antibiotics. This regimen was individually adapted to the microbiological findings, comprising of a minimum of 12 weeks’ total antibiotic therapy. The treatment was discontinued when C-reactive protein (CRP) and whole blood count normalized and clinical as well as radiological evaluation showed improvement. Surgical procedure ranged from open drainage, debridement and decompression of spinal cord and/or dura by laminectomy, hemilaminectomy, extended fenestration alone or debridement and placement of instrumentation. Instrumentation of the spine included dorsal, ventral or combined approaches. Postoperative pain management was carried out as published before [14]. QOL was evaluated prospectively with the Short-Form Health Survey 36 (SF-36) questionnaire. Intergroup comparison for detection of differences regarding patient´s QOL were performed in the follow up. Despite recommendation for early surgery, some patients denied surgery even in cases of radiologically evident osteomyelitis and favored conservative therapy with admission of antibiotics and underwent nonsurgical therapy.

Ninety-six patients with SEA were treated within a period of 7 years. Forty-nine patients (51%) underwent early surgery within 12 hours after admission (mean, 6.3 hours) and 47 (49%) delayed surgery later than 12 hours (mean, 51.6 hours). Mean age was 66.2 years (standard deviation [SD], 13.2) in the early surgery group and 65.8 years (SD, 13.1) in the delayed surgery group (p=0.9). The mainly affected region in both groups was the lumbar spine (50 patients; 20 early, 30 delayed, p=0.03) followed by the cervical spine (27 patients; 19 early and 8 late, p=0.02) and the thoracic spine (19 patients; 10 early and 9 late, p=1). The infection affected one level in 62 patients (30 early vs. 32 late; p=0.5), two levels in 19 patients (10 early vs. 9 late, p=0.9) and three or more affected levels in 16 patients (10 early vs. 6 late, p=0.4).

No significant differences were observed regarding prevalence of obesity, time from admission to discharge, diabetes, prior known spinal infection and bony destruction (Table 1).

The management of SEA still remains controversial. Despite improvement in surgical techniques, deeper understanding of the pathophysiology and broader availability of medical services, the best timing on evacuation of SEA is still unclear [8-11,24-30]. Several mostly retrospective data previously published favor early surgical treatment. However, the literature on the optimal treatment of SEA displays various limitations: next to the retrospective nature of most publication, another limitation lies within the lack of a consensus of “early” and “late” procedures, as observed in other neurosurgical diseases [13]. Connor et al. [3] reported their series consisting of 77 patients using a 3-day cutoff. In this series, the patients undergoing “early” surgery demonstrated a better neurological improvement without reaching the statistical significant level. Another operative series published by Ghobrial et al. [15] focused on timing of surgery in patients with SAE. The authors did not demonstrate a statistically significant benefit of early surgery (<24 hours after admission) over delayed surgery, as well. Ghobrial and colleagues [15] concluded that surgery within 24 hours seems to be beneficial, with the limitation of not having observed a statistical significance. The importance of early surgery in assurance of a good clinical outcome was assumed by several authors. However, the data justifying this hypothesis could not be provided so far.

The increased accessibility of medical services, early diagnosis and improved surgical techniques guarantee the possibility of early surgery. In this series, surgery within 12 hours after admission could be performed in 49 patients, whereas a high number of patients had to be operated later than 12 hours after admission. Therefore, early surgery could not be performed in these patients. The reason for delayed surgery in symptomatic patients were mostly the delay in diagnosis or poor general condition at admission with subsequent needed therapy to improve the health condition and enabling these patients undergoing surgery. Most patients with spinal infection and epidural abscess were referred from secondary or tertiary hospitals. Here, we could see in some cases a significant delay in establishing the diagnosis.

Most characteristics such as age, sex, diabetes, prior history of surgery, and obesity did not differ significantly in both cohorts. We were able to show that patients undergoing early surgery improved significantly compared to patients undergoing delayed surgery. Another predicting factor of good outcome is the initial clinical status. In the early surgery cohort, all patients with AIS grade D and E improved to AIS grade E. With the exception of two patients with neurological deterioration in the delayed surgery group (from AIS grade E to AIS grade D and AIS grade D to AIS grade B) and six patients all patients improved or remained unchanged in regard to their neurological status. Contrarily, of the patients admitted with poor initial neurological condition in both groups (AIS grade A and B, n=41) only 17 improved to AIS grade D and E.

Nevertheless, the initial neurological status, especially the presence of neurological deficits (AIS grade A–D) did not differ in both groups, as well as predicting factors, such as obesity, diabetes, bony destruction and prior history of spinal infection. Moreover, the clinical course of patients with pure epidural abscesses without osteomyelitis or discitis is certainly different from those with predominant osteomyelitis and may affect the long-term outcome of patients. The distribution of predominant osteomyelitis or discitis was without significant difference (p=0.5) between both groups. Based on these data and comparing almost identical cohorts we now provide evidence, that patients undergoing early surgery reveal a clear benefit regarding neurological status/outcome.

Another benefit of early surgery was the increased rate of identification of the infectious disease-causing pathogen. In the early group and after initial surgery the pathogen was identified almost twice as often when compared to the delayed group. Treatment with antibiotics during an initially nonoperative therapy and/or optimizing the clinical condition for surgery may be the reason for the fact, that pathogens were more often identified in the early surgery group.

To verify our obtained data, which were provided by physicians through physical examination at admission and before discharge, we conducted a prospective collection of data focusing on patient satisfaction undergoing both therapy modalities. The self-reported QOL in both groups was assessed after a mean follow up time of 58 months. Here, we were able to demonstrate that patients undergoing early surgery reveal a better QOL than after delayed evacuation of the SEA. We were therefore able to prove and reevaluate our initial findings after discharge in the follow up period. The beneficial effect of early surgery was reproducible in the follow up period, as well. More specifically, the important outcome parameters physical functioning and bodily pain were significantly better in the early surgery cohort. Role limitation, which is one item of the psychological score was also significantly better in the early surgery cohort, as well. The overall patient satisfaction was better in six of eight items in the early surgery cohort.

One of the important questions in treatment of patients with discitis, osteomyelitis and even with epidural collection of infectious mass remain the effectiveness of conservative therapy. In this study we were able to evaluate the effect of conservative therapy in this patients’ population. Twenty-seven patients underwent conservative therapy with long term antibiotics and immobilization. The basic parameters between all groups, such as age, sex, number of affected spine levels, bony destructions, diabetes, prior spinal infections, predominant discitis/osteomyelitis and rate of died patients in the follow up, did not differ significantly. In conservatively treated patients the infection was mostly located in the lumbar spine. Patients with severe neurological deficits, who had to undergo early surgery revealed often a cervical affection of the spine. Conservatively treated had a shorter hospital stay. In cases of conservative management, patients were mostly sent back to the primary hospital for antibiotic therapy. That means that hospital stay in the authors institution was significantly shorter than patients, who were operated on their spinal infection. The hospital stay in all patients undergoing surgery was essentially longer. The main advantage of surgery, especially of early surgery was the identification of the infectious causing pathogens and initiation of the pathogen specific antibiotic therapy. Here, we could show that the rate of germ-proof in the early surgery group was significantly higher than in the other two groups (79.8% vs. 48.9% vs. 22.2%). The effectiveness of conservative therapy in our investigated patients’ population was not effictive. The number of patients without neurological deficit was significantly higher in the conservative therapy group. Despite adequate therapy we could see an increase of failure of conservative therapy. From 15 patients with AIS grade E at admission only nine remained unchained, whereas the number of patients with neurological deficits (AIS grade A–D) increased over time (44.4% vs. 66.7%).

Another evidence for the inferiority of conservative therapy is the fact that almost 30% of patients were readmitted to our hospital in the follow up and almost 20% underwent subsequent surgery. The reason for the surgery in follow up was a marked increase of the bony destruction with deterioration of the pain situation and occurrence of neurological deficits due to spinal instability.

Moreover, comparing patients without and with evident neurological deficit treated conservatively and surgically revealed that patients undergoing conservative therapy show a decline in neurological status, whereas patients undergoing late surgery reveal an improvement in neurological status but not as effectively as undergoing early surgery, Table 2.

Another concern that could arise by reading the results of the study might be the definition of early surgery, which was defined as 12 hours after admission. In our believe early surgery seems to be beneficial for patients with SEA, the results of our data support the early surgical approach, as well. Nevertheless, until now there is no evidence for any time point performing surgery. Moreover, the effect of conservative treatment is mentioned very rare in the pertaining literature [19,23,27].

Instrumentation and fusion in patients with spinal infection is another often discussed aspect. Inserting of foreign bodies in an infectious situs seems not to be appropriate, whereas many authors published and publish data supporting the beneficial effect of fusion surgery. It provides an adequate surgical debridement and enables immediate stabilization. Surgical treatment with fusion involves debridement of necrotic or infected tissue and irrigation with an antibiotic regimen tailored to treat the responsible pathogen. Furthermore, instrumentation is necessary to bridge the large bony defects and stabilize the spine. Indeed, fusion with graft and internal fixation for refractory pyogenic discitis and vertebral osteomyelitis has been shown to be efficacious in resolving infection, restoring functional mobility, and reducing pain [9,21,22,26].

An important limitation of the present study is that it is a retrospective analysis. It should be considered that most of the surgical cohort comprises patients presenting complications of their infection and/or inadequate response to the conservative therapy. Both groups are difficult to compare, especially because conservative therapy was recommended in cases either presenting with mild manifestation of symptoms or with severe presentation as on the contrary harboring contraindications for surgery. It is to mention that there are some differences affecting the location of infection and severity of the disease. Nevertheless, the aim of this study is to provide an attempt of comparison between the different groups. The Time of surgery was measured from admission to our institution. Actually, it would be better to compare patients from onset of their symptoms than rather from admission. But the onset was very variable in both patients collective and could not evaluated in all cases. In addition, it should be considered that the evaluation of QOL might be influenced by comorbidity and even by social aspects and financial situation. Despite these limitations, patients in the early surgical group revealed a much better QOL. Furthermore, it will be interesting to know how the QOL in these patients was before the manifestation of the spinal infection. A prospective study is necessary to validate the obtained results of our study and to mitigate the occurrence of potential unidentified confounders and selection bias. Hence, over interpretation of the data must carefully be avoided. Since surgeons prefer to perform early surgery on patients with neurological deficits in order to relieve the pressure causing by SEA, we cannot exclude that the inclusion of these patients may elevates the better profile of early surgery.

The results obtained within this study demonstrate a beneficial impact of early surgical treatment of SEA within 12 hours after admission on both the neurological status and the QOL in affected patients. An important prognostic factor of good clinical outcome in our investigated cohort was the initial neurological status. Conservative therapy seems to be inferior to surgical therapy.