Secondary Logo

Journal Logo

Locoregional anaesthesia

Adverse events of postoperative thoracic epidural analgesia

A retrospective analysis of 7273 cases in a tertiary care teaching hospital

von Hösslin, Thomas; Imboden, Paul; Lüthi, Andreas; Rozanski, Marta J.; Schnider, Thomas W.; Filipovic, Miodrag

Author Information
European Journal of Anaesthesiology: October 2016 - Volume 33 - Issue 10 - p 708-714
doi: 10.1097/EJA.0000000000000446
  • Free



Thoracic epidural analgesia (tEDA) is an effective and well established technique for postoperative pain relief after major abdominal and thoracic surgery.1 Safety concerns include epidural haematoma and abscess which may cause transient or permanent neurologic deficit. The reported rates of these rare adverse events vary greatly amongst published studies and are based either upon general datasets (which were not specifically designed for documentation of pain management) or explicit studies with small numbers of patients and uncertain denominators.

We perform around 800 tEDAs per year. Data of all consecutive patients treated postoperatively by a dedicated ‘Acute Pain Service Team’ were entered prospectively in a specific data management system. This included patients treated with continuous EDA, patient controlled analgesia or continuous peripheral nerve blocks. The aim of our study was to evaluate the incidence and long-term outcomes of serious adverse events after tEDA-based postoperative pain management in a large tertiary care teaching hospital over a 10-year period.


Ethical approval for this study (Ethical Committee No. EKSG 12/157/U) was provided by Ethical Committee St. Gallen, St. Gallen, Switzerland (Chairperson Prof. Dr Ulrico Schmid) on 22 November 2012. Informed consent was sought from all patients approached by mail or telephone. For the current analyses, we retrieved prospectively, data collected from patients treated with tEDA for acute postoperative pain management for the 10-year period between January 2003 and December 2012. The data included information about demographics, the duration and quality of treatment and adverse events or unusual findings. The list was manually checked to find any deviations from the usual perioperative course. Thereafter, medical records of patients with suspected complications were reviewed. Serious adverse events were defined as spinal or epidural haemorrhage; spinal or epidural abscess or severe infections such as meningitis, discitis and sepsis; any permanent neurological deficit; EDA-associated cardiac arrest or death and incomplete removal of the EDA catheter.

All patients with documented or suspected adverse events were contacted by mail and were asked to complete a questionnaire to obtain long-term follow-up. Patients not sending back the questionnaire were contacted by phone and interviewed.

At our institution, all serious adverse events occurring during the perioperative course are registered in a ‘Serious Adverse Events Database’. This general database was additionally searched for adverse events linked to the placement of tEDA catheters. This additional approach was chosen because patients were entered in the tEDA-database only if the tEDA catheter was in use for postoperative pain control. By checking these two databases, we minimised the risk of missing a severe adverse event.

Thoracic epidural analgesia placement and management

According to institutional standards, epidural catheters (Perifix-Katheter, B. Braun, Melsungen, Germany) were placed in the middle or lower thoracic (T) spine (T5/6 for thoracic surgery and T8/9 or T9/10 for abdominal surgery, respectively) by the loss of resistance technique using an 18 gauge Tuohy needle. For all punctures a paramedian approach was used. For catheter placement, the catheter remained in its sterile plastic bag and was not touched. A test dose of 3 ml lidocaine (20 mg ml−1) CO2 and adrenaline (5 μg ml−1) was given to exclude intravascular or subarachnoid location of the catheter. Correct spread of the local anaesthetic was assured by loss of cold discrimination in the appropriate segments and the catheter was then sutured in place. A solution of bupivacaine (2.5 mg ml−1) and fentanyl (4 μg ml−1) was then administered continuously at a rate of 4 to 8 ml h−1 during and after surgery. All catheters were placed by a consultant anaesthetist or by a trainee under the supervision of a consultant anaesthetist.

Postoperative thoracic epidural analgesia management

Every patient receiving a tEDA catheter was admitted to the ICU or the post-anaesthesia care unit after surgery before transfer to a surgical ward. This transfer only occurred if the patient was able to give information about quality of pain control, had no motor blocks or other side-effects. Bupivacaine (2.5 mg ml−1) with fentanyl (4 μg ml−1) was given at any time during tEDA treatment. On the ward, a structured round was performed at least twice a day every 12 h by the acute pain service (APS) and 4 hourly by a ward nurse. The APS consists of a pain nurse and an anaesthesiologist or a resident in anaesthesia. The APS team is specially trained to manage regional anaesthesia within strict institutional standards.

After 3 to 4 days (8 days maximum), the tEDA-administered medication was reduced, stopped and the catheter was removed by the APS if pain control was satisfactory at an infusion rate of 3 ml h−1 or less. The dressing used for the catheter was changed every 24 h. Removal of catheters in anticoagulated patients occurred in accordance with international standards. Within 24 h after catheter removal, at least three follow-up visits were performed by a ward nurse and postoperative tEDA treatment was completed by a final check at 24 h after catheter removal by the APS. For thromboembolic prophylaxis, subcutaneous low-molecular weight heparin or i.v. heparin was given during the hospital stay.

Statistical analysis

Variables are shown as mean (mean ± SD) or rounded percentages. Excel for Windows (Microsoft, Redmond, USA) was used for the calculation of descriptive statistics. Confidence interval (CI) was calculated by using the webpage, last accessed on 14 December 2015.


The APS database contained 7430 patients. 7273 study participants met the inclusion criteria and were included in the final analyses. In total, 157 patients (2.1%) were excluded: one obstetric, 70 non-surgical cases, 25 non-EDA patients and 61 EDA-catheters which had not been placed by an anaesthesiologist. The mean age was 61 (±14.2) years (range, 13 to 100 years). There were more male than female patients (55.7% to 44.3%, respectively).

Surgical procedures were lower and upper abdominal interventions [n = 4857 (66.8%)], thoracic surgery [n = 923 (12.7%)] and urological surgery [n = 694 (9.5%)] and other procedures [n = 799 (11%)] (Fig. 1). Epidural drug administration lasted for 84.3 (± 39.3) hours and the mean infusion rate of the bupivacaine/opioid mixture was 3.5 (± 1.6) ml h−1.

Fig. 1
Fig. 1:
Overview illustrating the different surgical index procedures for which the patients received a thoracic epidural analgesia in our study.

In 6055 patients (83%) epidural drug administration ended after stepwise reduction of the infusion rate of the epidural medication to 3 ml h−1 or less and satisfactory pain control. In 1191 (16%) patients, the tEDA was discontinued ahead of schedule (Fig. 2). Among the most important reasons were unilateral, high level or unsatisfactory pain control in 338 (4.6%), catheter displacement, disconnections and other technical problems in 261 (3.3%) and fast-track surgery in 130 (1.8%) patients.

Fig. 2
Fig. 2:
Thoracic epidural analgesia has been discontinued earlier than planned in 1191 patients. The different reasons and numbers of patients are presented in this figure.

Recordings of 7207 patients were free of tEDA-related complications. Adverse events were reported in 66 patients (Table 1). Seven patients suffered a tEDA-associated serious adverse event (Tables 2 and 3). An epidural abscess (patient 1) treated surgically with complete recovery, a tEDA-associated cardiac arrest (patient 2) with complete recovery, a permanent neurological deficit at the L1/2-segment (patient 3) and incomplete removal of the EDA catheter (patients 4, 5, 6, 7). The fragments were not removed and caused no clinically evident problems.

Table 1
Table 1:
Table 2
Table 2:
Patients with major adverse events, demographics
Table 3
Table 3:
Patients with major adverse events, complications

Details of the most serious adverse events

Patient 1

A 59-year-old diabetic patient underwent low anterior resection for rectal carcinoma. Five days after surgery, the patient complained of discomfort and pain at the epidural puncture site (T8/9). The catheter was removed immediately and a subcutaneous abscess incised. One day later, fever and neck stiffness occurred. In an MRI, an epidural abscess at level C6 to T8/9 was diagnosed. Emergency laminectomy was performed and systemic antibiotic therapy was started. After surgery, the symptoms resolved and the recovery was complete and uneventful.

Patient 2

A 70-year-old female patient with rectal carcinoma underwent low anterior resection and left adnexectomy. Comorbidities included coronary and hypertensive heart disease with atrial fibrillation. On the first postoperative day, the patient was transferred to the regular ward with an epidural infusion rate of 3 ml h−1 (bupivacaine 2.5 mg ml−1 along with fentanyl 4 μg ml−1). After the transfer to the ward, low blood pressure and a bradycardia (heart rate around 50 bpm) were recorded. Three hours later, cardiac arrest occurred observed by healthcare professionals. The patient was successfully resuscitated. The epidural infusion was stopped immediately. No cerebrospinal fluid could be aspirated through the epidural catheter. The recovery was uneventful.

Patient 3

A 43-year-old otherwise healthy male patient was admitted to the hospital for repair of an incisional hernia. During insertion of the epidural catheter at the T9/10 level, he complained of a sharp pain sensation affecting the whole body. He also reported pain in the right inguinal region. The catheter was then removed and a new puncture was successfully performed at the T8/9 level. Surgery and the first postoperative days were uneventful. After regular removal of the catheter, pain in the right inguinal region and scrotum occurred. The patient was seen by a consultant neurologist who made the diagnosis of a neuralgiform pain syndrome involving the genitofemoral nerve. The MRI investigations of the spine and the pelvic region were free of any pathological finding and particularly showed no abnormalities of the genitofemoral, the ilioinguinal or any other pelvic nerve. A treatment with pregabalin was started according to the neurologist's suggestion. In total, 40 months after the index hospitalisation, the patient still complained of a persistent hypoaesthesia in the genitofemoral nerve area.

The incidence of each of the most serious adverse events (epidural abscesses, persistent neurological deficit and cardiac arrest) was 1.4 : 10 000 (= 0.014%, 95% CI, 0 to 0.08%). The incidence of retained catheter fragments after removal was 5.5 : 10 000 (= 0.055%, 95% CI, 0.01 to 0.14%). In total, 52 patients had a minor EDA-associated adverse event (e.g. transient sensorimotor blocks resolving after reduction of the application of local anaesthetics, subcutaneous abscesses and, a pleural tap or local infections) (Table 1). Three of the seven serious events were found in the APS database (patients 1, 2, 4), whereas four were discovered in the Serious Adverse Events Database (patients 3, 5, 6, 7).


The retrospective study of a large database about tEDA at a single institution revealed a relatively low incidence of serious adverse events of tEDA-based pain management following major surgery which was comparable to several previously reported rates.

For epidural abscess, studies have reported incidences between 1 : 800 to 1 : 85 000.2–5 Our patient had several risk factors for local infection such as a history of diabetes and rectal cancer, and multiple insertion attempts were performed using the same catheter, which constituted a deviation from our institutional standards namely that the catheter must not be touched directly and replaced after an unsuccessful insertion attempt. The epidural abscess was diagnosed 5 days after insertion of the epidural catheter which accords with the median of 5 days and range of 2 days to 5 weeks reported in the literature.6

For cardiac arrest associated with neuraxial anaesthesia, a rate of 1.3 to 18 : 10 000 has been reported.7 The study included spinal and epidural anaesthesia in patients undergoing mainly orthopaedic and urologic procedures. Our patient became hypotensive and bradycardic at a low epidural infusion rate and the observed cardiac arrest from which the patient was successfully resuscitated occurred after transfer to the ward. The most plausible explanation of the cardiac arrest is migration of the epidural catheter to the subarachnoid space causing high spinal anaesthesia and hypotension. Epidural catheter tips may penetrate the dura while changing position (e.g. from left lateral to supine position), which underlines the need for repeated negative aspiration tests and continuous surveillance.8,9 The catheter could migrate into the subdural epiarachnoid space which is known to produce extensive spread of small volumes of local anaesthetic solutions.10,11 Local anaesthetic systemic toxicity is a very unlikely cause for the cardiac arrest in our case because the administrated dosage was very small (3 ml h−1).12 Other potential causes for the cardiac arrest such as a pulmonary embolism, acute coronary syndrome or arrhythmia were ruled out.

We identified one patient with a persistent neurological deficit. Pain and paraesthesia occurred during the placement of the EDA catheter at T8/9 but the deficit affected the genitofemoral nerve (L1/2). MRI revealed no pathological findings of the vertebral column, genitofemoral, ilioinguinal or iliohypogastric nerves. A causative relationship between the deficit and the placement of the EDA cannot be ruled out with certainty13 but seems implausible because the single concerned nerve is not affected by the epidural access route. However, the level of puncture was determined by surface anatomy and might have been mistaken by one or two segments.14 In addition, the origin of the ilioinguinal and the iliohypogastric nerves can be as high as segment T11.15 More plausible is that the hernia itself or the surgical treatment caused isolated nerve damage. A recent study showed one persisting neurological deficit in 5083 patients.16

In our study, we found a rather high rate of broken catheters, one catheter was cut through during removal and another catheter was damaged by a suture which was applied too tightly. Three potential mechanisms for catheter disruptions are discussed in the literature: catheter shearing by the epidural needle, kinking or knotting while inserting and entrapment of the catheter by ligaments, bony structures and the consecutive breaking during removal.17 Complications such as these have been described in case reports18–20 and in larger studies which reported rates of catheter disruption of 1 : 60 000 (0.002%)18 and 1 : 2250 (0.04%),21 respectively. The true incidence, mechanisms and clinical consequences of catheter disruptions are only partially explored. In the literature most catheter disruptions occurred at lumbar level.22–24 This phenomenon is explained by differences in the anatomy of the lumbar and the thoracic epidural space, for example, the different segmentation of the epidural space and differences in the angle of the needle entrance into the space.24 In our study all catheter fragments were left in situ which is in accordance with current recommendations.17 Catheter fragments are sterile and chemically inert and thus, are expected to cause no harm. During the study period we found no evidence for any clinically relevant problems caused by the retained catheter fragments although long-term follow-up could not be obtained as all of these patients had died within 10 months after surgery and one was suffering dementia.

Fifty-nine minor events which resolved completely during index hospitalisation were identified in the APS database (see Table 1). Of those events, seven were assessed as not being EDA-associated and 52 as being potentially EDA-associated (transient sensory-motor blocks resolving after reduction of the application of local anaesthetics, local and subcutaneous infections and a pleural tap; Table 1). In the literature, there are few case reports describing accidental placement of an ‘epidural’ catheter into the pleural space.25 In our patient, the incorrect position of the catheter was not detected until the third postoperative day, when she experienced severe pain and a partial motor block of the lower limb. A computer tomography of the spine revealed the malposition of the catheter in the pleural space. The catheter was removed thereafter. The motor block disappeared after a short time but might not have been associated with pleural application of local anaesthetics. Other complications after unintentional pleural puncture such as haemothorax26 or pneumothorax27 have been described, but were not evident in our patient. The patient had difficult anatomy for performance of an epidural block because of obesity and excessive skin folds and thus two attempts were needed for catheter placement.

Limitations of our study

The population in our study consisted of patients undergoing major non-cardiac surgery and the EDA was placed in thoracic segments. Hence, our data may not be valid in other populations, for example, obstetric or chronic pain patients.

We cannot exclude the possibility of undetected adverse events. This might especially be true in cases where tEDA could not be placed successfully and surgery was performed without tEDA, and a possible complication remained undetected by the anaesthesiologist performing the (unsuccessful) tEDA placement.

As in every cohort study we cannot completely rule out selection and report bias and so we used two independent institutional databases for determination of adverse events. One of the seven events was recorded in both data sets, whereas three were recorded in the APS database and four in the adverse events data set. We could not determine retrospectively the reasons for reporting events in only one or other database but by using two databases we were able to improve our numerator data.

It is possible that complications might have remained undetected during index hospitalisation but become evident after discharge or after transfer to a peripheral hospital and not all patients in our study could be followed up.


In our single-centre study of thoracic epidural analgesia, serious adverse events occurred in 0.1% cases (1 : 1000) while long-term outcome was compromised in 0.014% (1.4 : 10 000) which is similar to the serious adverse event rates and outcomes reported in the current literature.

Acknowledgements relating to this article

Assistance with the study: none.

Financial support and sponsorship: none.

Conflicts of interest: none.

Presentation: none.


1. Pöpping DM, Elia N, Van Aken HK, et al. Impact of epidural analgesia on mortality and morbidity after surgery: systematic review and meta-analysis of randomized controlled trials. Ann Surg 2014; 259:1056–1067.
2. Phillips JMG, Stedeford JC, Hartsilver E, Roberts C. Epidural abscess complicating insertion of epidural catheters. Br J Anaesth 2002; 89:778–782.
3. Wang LP, Hauerberg J, Schmidt JF. Incidence of spinal epidural abscess after epidural analgesia: a national 1-year survey. Anesthesiology 1999; 91:1928–1936.
4. Cameron CM, Scott DA, McDonald WM, Davies MJ. A review of neuraxial epidural morbidity: experience of more than 8000 cases at a single teaching hospital. Anesthesiology 2007; 106:997–1002.
5. Aromaa U, Lahdensuu M, Cozanitis DA. Severe complications associated with epidural and spinal anaesthesias in Finland 1987–1993. A study based on patient insurance claims [see comment]. Acta Anaesthesiol Scand 1997; 41:445–452.
6. Moen V, Dahlgren N, Irestedt L. Severe neurological complications after central neuraxial blockades in Sweden 1990-1999. Anesthesiology 2004; 101:950–959.
7. Kopp SL, Horlocker TT, Warner ME, et al. Cardiac arrest during neuraxial anesthesia: frequency and predisposing factors associated with survival. Anesth Analg 2005; 100:855–865.
8. Philip JH, Brown WU. Total spinal anesthesia late in the course of obstetric bupivacaine epidural block. Anesthesiology 1976; 44:340–341.
9. Jiang X, Huang W, Lin X. Ropivacaine-induced cardiac arrest and paraplegia after epidural anesthesia. Minerva Anestesiol 2012; 78:1309–1310.
10. Reina MA, Collier CB, Prats-Galino A, et al. Unintentional subdural placement of epidural catheters during attempted epidural anesthesia: an anatomic study of spinal subdural compartment. Reg Anesth Pain Med 2011; 36:537–541.
11. Collier CB, Reina MA, Prats-Galino A, Machés F. An anatomical study of the intradural space. Anaesth Intensive Care 2011; 39:1038–1042.
12. Neal JM, Bernards CM, Butterworth JF, et al. ASRA practice advisory on local anesthetic systemic toxicity. Reg Anesth Pain Med 2010; 35:152–161.
13. Dahlgren N, Törnebrandt K. Neurological complications after anaesthesia. A follow-up of 18 000 spinal and epidural anaesthetics performed over three years. Acta Anaesthesiol Scand 1995; 39:872–880.
14. Holmaas G, Frederiksen D, Ulvik A, et al. Identification of thoracic intervertebral spaces by means of surface anatomy: a magnetic resonance imaging study. Acta Anaesthesiol Scand 2006; 50:368–373.
15. Klaassen Z, Marshall E, Tubbs RS, et al. Anatomy of the ilioinguinal and iliohypogastric nerves with observations of their spinal nerve contributions. Clin Anat 2011; 24:454–461.
16. Kang X-H, Bao F-P, Xiong X-X, et al. Major complications of epidural anesthesia: a prospective study of 5083 cases at a single hospital. Acta Anaesthesiol Scand 2014; 58:858–866.
17. Kasivisvanathan R, Sodhi M, Setty S. The broken epidural catheter: to remove or not to remove? Br J Hosp Med (Lond) 2014; 1–2.
18. Collier C. Epidural catheter breakage: a possible mechanism. Int J Obstet Anesth 2000; 9:87–93.
19. Staats PS, Stinson MS, Lee RR. Lumbar stenosis complicating retained epidural catheter tip. Anesthesiology 1995; 83:1115–1118.
20. Blanchard N, Clabeau JJ, Ossart M, et al. Radicular pain due to a retained fragment of epidural catheter. Anesthesiology 1997; 87:1567–1569.
21. Crawford JS. Some maternal complications of epidural analgesia for labour. Anaesthesia 1985; 40:1219–1225.
22. Schummer W, Schummer C. Another cause of epidural catheter breakage? Anesth Analg 2002; 94:233.
23. Hobaika AB, de S. Breakage of epidural catheters: etiology, prevention, and management. Rev Bras Anestesiol 2008; 58:227–233.
24. Dounas M, Peillon P, Lebonhomme JJ, Guittard Y. Difficulties in the removal and rupture of a peridural catheter. Ann Fr Anesth Reanim 2002; 21:600–602.
25. Sundary MT. Epidural catheter misplaced into the thoracic cavity: Utilized to provide interpleural analgesia. Anesth Essays Res 2015; 9:121–123.
26. Iida Y, Kashimoto S, Matsukawa T, Kumazawa T. A hemothorax after thoracic epidural anesthesia. J Clin Anesth 1994; 6:505–507.
27. Miura K, Tomiyasu S, Cho S, et al. Pneumothorax associated with epidural anesthesia. J Anesth 2004; 18:138–140.
© 2016 European Society of Anaesthesiology