ELDF Complications & Metanalysis Study


ELDF Complications & Metanalysis Study

Purpose of the Study

This study examines the morbidity and complications associated with the first 958 consecutive ELFs in a prospective study and subjected these results to a metanalysis of the safety of ‘conventional’ spinal surgical procedures. The study of the first 250 ELFs (General Outcome Study on page 242) reported a complication rate of 0.8% with only two patients experiencing notable complications (Knight et al 2000). This study has examined the prospective data recording of all complications in a larger longitudinal study.

Materials & Methods

Scope of Complications

A review of the literature provided a list of potential complications laid out as: systemic & neurological complications in Table 1, vascular & spinal complications in Table 2, Abdominal genito-urinary complications in Table 3,

Table 1: Systemic & Neurological Complications of Spinal Surgery
Systemic Complications  Neurological complications 
 Pneumonia   Foot Drop
Syndrome of Inappropriate ADH secretion  Arachnoiditis
 Anaphylaxis  Cauda Equina Syndrome
 Bedsores  Nerve Palsies
 Superficial wound Infection  
 Unknown Causes  


Table 2: Vascular & Spinal Complications of Spinal Surgery
Vascular Complications  Spinal Complications 
 Deep Venous Thrombosis  Spinal Complications
 Coronary Thrombosis  Recurrent disc protrusion
 Cardiac Arrythmias  Recurrent disc extrusion
 Angina  Recurrent disc sequestration
 Pulmonary Embolism  Infective Discitis
 Unknown Causes  Aseptic Discitis
 Basilar Insufficiency  Dural Tear
 Migraine  Intraspinal Bleeding
 Carotid Insufficiency  
 Cerebrovascular Attack (Left)  
Cerebrovascular Attack (Right)  


Table 3: Abdominal & Genitourinary Complications of Spinal Surgery

Abdominal Complications Genitourinary Complications
 Perforated Bowel  Urinary Retention
 Bowel Injury  Urinary Tract Infection
 Retroperitoneal Bleeding  

These tables list potential or recorded complications, which could arise in association with ELF or conventional surgical equivalents, discectomy, decompression, fusion and chemonucleolysis. 

Literature Review Inclusion Criteria

The review of the literature was restricted to RCTs where the level of supervision would be strict. Of these, only the following studies mentioned complications individually thus indicating that they were recorded. (Blankstein et al 1987, Burton 1978, Dauch 1986, Ferree and Wright 1993, Garfin 1989, Gill 1990, Johnson, RM and McGuire 1981, Kurz et al 1989, Montorsi and Ghiringhelli 1973, Noyes and Morrisseau 1982, Pilgaard 1969, Rothman and Simeone 1992, Shaw et al 1981, Stauffer and Coventry 1972).
Conventional techniques of spinal surgery were grouped as discectomy, including microdiscectomy and APLD, decompression, fusion (all types) and chemonucleolysis of the intervertebral disc. Using the randomised controlled trials of spinal surgery, selected by the Cochrane group, complication data for each of these procedures was compiled.

Table 4: Inclusion Criteria for Metanalysis
Criteria for inclusion of randomised controlled clinical trials in Cochrane review
Papers must be published in the English Language
Papers must involve a form of conventional spinal surgery
The trial must be recognised by the Cochrane review
Complications must be mentioned individually

In the RCTs comparing an intervention to placebo therapy (Bromley et al 1984, Dabezies et al, 1988, Fraser 1982, Hedtmann et al 1992, McGuire & Amundson 1993) only the complications related to the intervention group were collated and used in the metanalysis.

ELF Data Recording

Prospective records of each patient undergoing ELF, contained a full pre-operative assessment of symptoms, clinical and radiological examination and MRI scan reports. Post-operative records consisted of the Oswestry Disability Index, Visual Analogue Pain scores, psychological profiles and patient satisfaction scores A pain diary was given to patients that they were encouraged to fill in 3 times a day over the six weeks following the operation. This diary had a section for comments in which the patient or medical assessor would record any complications or change in symptoms with reasons thereto. Follow-up of the patients after ELF occurred at 1 day, 6 weeks, 3 months, 6 months and yearly thereafter. This study analysed all patients followed up at the 6-week review point with cohort integrity of 100%.

Per-operative complications were recorded by the attending nurse on the observation recording sheets completed throughout the aware state procedure and assessments made by the surgeon at the time of the operation and recorded in the immediate post-operative record.

These paper records were entered into the Spinal Foundation database and form a comprehensive account of all the patients’ consultations, operative and outcome findings.

Statistical Analysis

Differences between the overall complication rates for ELF and the four groups of conventional surgery were considered in three different ways. Firstly, the difference in proportion was considered using Chi square (or Fisher’s exact test as appropriate). Secondly, the magnitude of the population difference in proportion was assessed by 95% and 99% confidence intervals and lastly the relative risk of a complication with ELF compared to conventional surgery was calculated. The statistical packages SPSS (Statistical Package for Social Sciences, SPSS Inc, (Chicago) and CIA (Confidence Interval Analysis, Version 1.2, London) were used.


Conventional Spinal Surgery Complications

The Cochrane Review examined 41 randomised controlled trials in the field of spinal surgery. Of these 41 trials, 6 were written in a language other than English (Bontoux et al 1990, Bourgeois et al 1988, Feldman et al 1986, Hedtmann et al 1992, Lagarrigue and Chaynes, 1994, Lavignolle et al 1987) and were excluded on this basis. One paper was concerned with the effects of electromagnetic fields on lumbar interbody fusion (Mooney 1990) and was excluded, as it was unclear as to whether the complications reported had occurred because of the operative element of the study. Two further papers were radiological studies (Jensen, T et al 1996, MacKay et al 1995) and as such were excluded because they did not comment on the gamut of complications arising from the procedure used in the trial.

      Patient Totals

Table 5 Incidence of complications with ELF and conventional Surgery

Four trials made no comment as to complications encountered with the procedure (Chatterjee et al 1995, Herkowitz and Kurz 1991, Schwetschenau et al 1976, Weber, H 1983)
Overall, the results of 22 papers were reviewed. These commented on the complications that arose during the reported procedure. The breadth and incidence of these complications is summarised in Table 5 above.


Complications encountered in ELF
Discitis (Infected)
9 (1)
Dural Tear
Deep Wound Infection
Transient Foot Drop (Sustained)
2 (1)
Myocardial Infarction
Erectile Dysfunction
Panic Attacks following surgery
Clinically significant residual disc herniation

Table 6 Complications reported in 958 ELFs

In the consecutive series of 958 ELFs, 24 complications occurred in 23 patients as shown.

Table 6 above records the range and incidence of these complications.

This reveals an overall complication rate of 2.4% following ELF. The incidence of the more serious complications is highlighted in red and amounts to an incidence of 1.1%.
The total complication rate for ELF was compared to that reported in the RCTs of conventional spinal surgery as tabulated in Table 5 above.

versus ELF
Confidence Level
Confidence Interval
-0.106   to –0.354
-0.103   to 0.000564
-0.0654 to –0.00406
-0.132   to –0.0550

Table 7 Confidence Interval analysis of ELF versus Conventional Procedures

Versus ELF
Chi-Square Test
P value
Pearson Chi-Square
Fishers Exact Test
Pearson Chi-Square
Pearson Chi-Square

Table 8 Chi-Square Analysis of ELF versus Conventional Procedures


The metanalysis was confined to RCTs because the results obtained would be based upon closely scrutinised prospective studies with control of bias. RCTs tend to be performed in specialist centres to exacting criteria. The ELF was studied using stringent prospective data collection and comparison to RCT data would seem appropriate.

At six weeks follow-up, it can be seen that ELF displays an overall complication rate lower than that evident in conventional spinal surgery. At 99% confidence intervals, the magnitude of this difference in proportion is found to be between 0.1 and 0.35 for chemonucleolysis, 0.6 to 0.13 for fusion and 0.004 to 0.07 for discectomy. There were relatively few cases and complications for decompression and the 95% confidence interval is 0 to 0.10 (Table 7 above).

Chi squared tests showed a significant difference in proportion between ELF and the four separate conventional treatments (Table 8 above).
ELF has a lower relative risk of complications than all four conventional procedures (Table 6).

The ideal study of post-operative complications would be a RCT directly comparing ELF with specific conventional procedures with specific well-defined criteria focused upon the detection of complications. Such a study is not available at this time and the matter has been investigated by metanalysis but this technique suffers shortcomings. Whilst the ELF study concentrates on recording all adverse events, other reports focusing on clinical outcomes make cursory reference to complications. These reports are subject to the surgeon’s concept of what constitutes a complication. Many events may be deemed a complication, some more significant than others. This paper sets out a comprehensive list of criteria concerning complications and set these against ELF on a prospective basis. It cannot be surmised that even the RCT reports represent a comprehensive record and underestimation is likely. By contrast, a comprehensive list of complications was used to evaluate the morbidity of ELF. Comparisons with conventional spinal surgery may still be drawn because ELF has a lower rate of complications despite using wider inclusive complication criteria.

Complications of Fusion

The nature of fusion leads to hardware problems contributing to specific complications with the procedure. This however is a complication and though this is not experienced in the other procedures it is still appropriate to be considered as such. ELF has been used for CLS, FBS and spondylolytic spondylolisthesis in this series for patients who otherwise would have been treated by fusion in many cases.

Complications of Chemonucleolysis

Chemonucleolysis is a minimally invasive technique, even less invasive than ELF as it involves only the insertion of a needle into the intervertebral disc. It does however have a complication uncommon in other forms of spinal surgery, namely that of anaphylaxis. Additionally, Chemonucleolysis reactions cause prolonged painful back spasm postoperatively. Anaphylaxis is a major complication that is difficult to minimise even with pre-operative sensitivity tests.

Complications of Discectomy and Decompression

The complications encountered in discectomy and decompression are in general those attributed to all forms of spinal surgery. In this review, no complications were found to have been reported that were peculiar to discectomy or decompression setting it apart from spinal surgery as a whole.

Complications of ELF

Within the set of 958 patients, two deaths occurred within the six-week period. The first was an elderly patient who died of natural causes, as attributed by a coroner’s inquest. The second was of a middle-aged person who developed pneumonia six weeks after surgery and despite treatment, the patient subsequently died. This was not considered attributable to the operation as onset occurred six weeks following the ELF in a fully mobile individual.

One patient developed panic attacks following surgery. The ELF is an ‘aware state’ procedure with the patients being semi-conscious and responsive during the procedure. The patient in question became mentally ‘traumatised’ following the procedure reporting the onset of her symptoms a few weeks postoperatively every time she thought of the operation. It is interesting to note that this patient had not complained of any problems during the procedure, only encountering the symptoms a number of weeks later. The psychological implications of an ‘aware-state’ procedure are important to consider. Patients must consider this eventuality and must be fully informed verbally and in writing prior to the procedure. We recommend that patients have the opportunity to discuss the operation with other patients who have undergone the procedure, if they have anxieties. This psychological sequel has been minimised by clear pre-operative explanation and maintenance of surgeon-patient confidence throughout the procedure.

Out of 36 RCTs, only three comment on perineural scarring as a finding after surgery on the lumbar spine. Current evidence suggests that scarring can be clinically inconsequential being a finding in both symptomatic and asymptomatic patients (Jensen, T et al 1996). Scarring may become a problem once inflammation occurs alongside it, leading to pain and recurrent neurological sequelae. Scarring may cause symptoms if it leads to tethering of the nerve in the foramen, and as such reduces the capability of the nerve to move when corresponding movement of the spine occurs. Whether perineural scarring is a complication or an outcome of any spinal surgery is debateable but the occurrence of clinically related symptoms would be included as a complication in this study. However, symptoms arising from perineural scarring usually arise at about 3 – 5 months from surgery and lie out with the range of this study. In the light of the low reportage in the metanalysis, this offers little bias because the majority of surgical complications occur within six weeks of surgery.

In many studies, residual or recurrent disc herniation has been treated as a failure of outcome rather than operative procedure but in this study, it has been included as a complication. The incidence in textbooks is stated to range from 2 – 11% (Garfin 1989, Garfin et al 1988, Suk et al 2001). Patients were determined to have a residual disc herniation if the symptoms were present and ongoing during the first six weeks after surgery. Relief from symptoms for a period of more than six weeks followed by recurrence was deemed a recurrent disc herniation.

A review of infective discitis following lumbar discectomy reveals an incidence of 0 - 4% (Rhode et al 1998, Tai et al 2002), after Microdiscectomy 0.9 - 2.5% (McCulloch 1987, McCulloch 1992, Meyer et al 1995) and fusion 2 – 6% (Hodges et al 1998). However, after Arthroscopic Microdiscectomy Kambin et al (1998) reported an incidence of 0.3% in 169 patients in keeping with the low incidence following ELF. Controversy surrounds aseptic discitis concerning inception with or without a bacterial origin (Fraser et al 1996). The literature concerning aseptic discitis is obscured by the difficulty in distinguishing between aseptic (chemical) discitis and infective discitis. Agre et al (1984) stated that the incidence of aseptic discitis was 1% in a collated review of 29,075 patients undergoing chemonucleolysis, but this entity is seldom reported in conventional surgery perhaps because it was deemed an accepted outcome rather than a complication. It has been included in the complications attributable to ELF and occurred in nine cases (0.9%). This and the incidence of residual disc protrusion accounts for the preponderance of spinal complications following ELF and the higher incidence compared to conventional decompression or discectomy.

The clinical outcome following ELF is promising (see General Outcome Study, Conclusion on page 29). It is also pivotal to note that in the first 250 ELF studied for outcome, 142 of these patients were assessed prior to ELF and advised that conventional methods of surgery were inappropriate for their condition and that 75 had had previous surgery. ELF in contradistinction to conventional spinal surgery was used to address CLS, spondylolytic spondylolisthesis, the infirm, elderly, FBS syndrome and those deemed ‘untreatable’. Despite application in this clinically challenging group, the complication rate was 3 times safer than fusion or decompression, more than twice as safe as discectomy, and four times as safe as Chemonucleolysis.


In summary ELF is safer than conventional surgery for the treatment of CLS, spondylolytic spondylolisthesis and FBS, back pain and sciatica (P<0.01) and as such this ‘day-case’ procedure provides a low risk minimally intrusive treatment for the patient.

This study highlights the lack of comprehensive reporting of complications following conventional surgery and the need for the re-definition of residual or recurrent herniation and aseptic discitis as complications of surgery rather than reallocating these problems to outcome analysis.

This prospective study has high cohort integrity but provides only a short-term (6-week) review. In the United Kingdom, the coroner shall be informed about any death occurring within a year of surgery as a means of catching the majority of operation related sequelae.

This study will miss the incidence of symptomatic perineural scarring presenting after about 3 – 5 months, but the other studies reviewed made no reference to postoperative scarring so the data is not biased in this regard. This study could miss the occurrence of late infection and the presence of postoperative settlement but again other studies have not reported such data and again this study is not biased adversely in this regard.