Epidural fentanyl/bupivacaine infusions often are limited to high dependency units or intensive care units. One thousand fourteen patients receiving epidural fentanyl/bupivacaine infusions for analgesia after major surgery who were managed in the general surgical ward were prospectively surveyed.
Patients leaving the recovery room with an epidural catheter in situ were assessed three times a day by acute pain service personnel for quality of pain relief, using a rating scale that accounted for pain on movement. The presence of side effects and complications was assessed.
Data were collected from February 1990 to May 1993. The average duration of infusion was 3 days. A patient's pain relief was rated as good to excellent on 82.6% of visits. Side effects possibly attributable to fentanyl included sedation (7.4%), pruritus (10.2%), nausea and vomiting (3.1%), and respiratory depression (1.2%). Respiratory depression commonly was associated with sedation and was detected easily on the postsurgical ward, with only four patients requiring naloxone (0.4%). Side effects possibly related to bupivacaine included unpleasant sensory block (2.6%), significant lower limb motor block (3.0%), and hypotension (6.6%). There were two cases of epidural hematoma. Inflammation at the epidural catheter insertion site occurred in 3.8% (38), of which 42% (16) had some cutaneous purulence detected. There were no epidural space infections. Mechanical problems, including dislodgment of the catheter, accounted for 18.7% of infusion discontinuations within the first 72 h.
Postoperative epidural fentanyl/bupivacaine infusions are effective and can be managed readily in general postsurgical wards with minimal complications provided that appropriate patient observations are performed.
Methods: Patients leaving the recovery room with an epidural catheter in situ were assessed three times a day by acute pain service personnel for quality of pain relief, using a rating scale that accounted for pain on movement. The presence of side effects and complications was assessed.
Results: Data were collected from February 1990 to May 1993. The average duration of infusion was 3 days. A patient's pain relief was rated as good to excellent on 82.6% of visits. Side effects possibly attributable to fentanyl included sedation (7.4%), pruritus (10.2%), nausea and vomiting (3.1%), and respiratory depression (1.2%). Respiratory depression commonly was associated with sedation and was detected easily on the postsurgical ward, with only four patients requiring naloxone (0.4%). Side effects possibly related to bupivacaine included unpleasant sensory block (2.6%), significant lower limb motor block (3.0%), and hypotension (6.6%). There were two cases of epidural hematoma. Inflammation at the epidural catheter insertion site occurred in 3.8% (38), of which 42% (16) had some cutaneous purulence detected. There were no epidural space infections. Mechanical problems, including dislodgment of the catheter, accounted for 18.7% of infusion discontinuations within the first 72 h.
Conclusions: Postoperative epidural fentanyl/bupivacaine infusions are effective and can be managed readily in general postsurgical wards with minimal complications provided that appropriate patient observations are performed.
Key words: Analgesia: postoperative. Analgesics: fentanyl. Anesthetic techniques: epidural. Complications. Epidural analgesia: methods. Pain: postoperative. Prospective survey.
The provision of good quality analgesia in the postoperative period is important, not only to ease patient suffering and improve well-being, but because poor pain control can lead to cardiovascular and respiratory complications. Analgesia for high-risk patients via the epidural route may result in shorter intensive care unit (ICU) stays. 
After surgery, benefits of good quality epidural analgesia include improved respiratory function, [2-6]decreased perioperative cardiac complications, earlier mobilization and return of bowel function, and shorter hospital stays. [4,6]
These benefits are not without their risks. The problem of respiratory depression in particular has raised issues of safety, with some authors limiting the use of epidural morphine to the high dependency unit (HDU), or ICU, where appropriate monitoring can occur. [8-12]Other serious concerns related to epidural analgesia include profound hypotension, migration of the epidural catheter into intravenous or subarachnoid positions, epidural abscess formation, and epidural hematoma.
A more lipophilic drug, such as fentanyl, should reduce the risk of respiratory depression, because its rapid absorption into the spinal cord and nearby blood vessels decreases the concentrations in cerebrospinal fluid (CSF) more rapidly than morphine and reduces the risk from cephalad CSF spread. [5,12]This advantage has been confirmed in clinical studies. [5,8,13,14]Large series describing the use of epidural opioids have not dealt with the use of bupivacaine and fentanyl mixtures in the general surgical ward. In 1985, Stenseth et al. presented a series of 1,085 patients who were treated with bolus epidural morphine in the postoperative care unit, with a low incidence of side effects. Ready et al., in 1991, were able to demonstrate in 1,106 patients that bolus epidural morphine was safe in the surgical ward. This was confirmed for continuous infusion of epidural morphine and bupivacaine in a study of 4,227 surgical cancer patients by de Leon-Casasola et al. Chien et al. presented a series of 517 patients who received fentanyl alone by infusion on two specially trained surgical wards. This study is a prospective audit of more than 1,000 cases of epidural fentanyl/bupivacaine infusions used for postoperative analgesia in general surgical wards. We investigated the efficacy of the analgesic regimen and incidence of side effects.
Methods and Materials
We aimed to prospectively collect data from 1,000 patients receiving epidural bupivacaine and fentanyl infusion for postoperative pain relief. Institutional ethics committee approval was not considered necessary.
Epidural catheters were placed preoperatively at a spinal level appropriate for the proposed surgery and were used to provide both intraoperative and postoperative analgesia. A 16-G Tuohy needle and an 18-G catheter (Portex) were used, and a standard intravenous infusion pump (IMED) was used to deliver the drugs epidurally via an IMED "9630 Accuset-Closed system non-Poly Vinyl Chloride (PVC) fluid path" set. This infusion set has no injection ports and is a different color than standard intravenous infusion sets, which precludes confusion with conventional intravenous infusion sets.
Solutions were prepared by either pharmacy (0.1% bupivacaine and 5 micro gram *symbol* ml sup -1 fentanyl (solution 0.1/5)) or the anesthesiologist who placed the epidural catheter. During the period audited, the nature of the epidural fentanyl bupivacaine mixture changed. Initially, mixtures of 0.1% bupivacaine with 1 micro gram *symbol* ml sup -1 fentanyl (solution 0.1/1) and 0.1% bupivacaine with 10 micro gram *symbol* ml sup -1 fentanyl (solution 0.1/10) were used, but it became apparent that a third solution was required with an intermediate concentration of fentanyl, and toward the end of 1990, a solution of 0.1/5 was introduced (see Discussion). Although the former two solutions still were used toward the end of the study period, the majority of fentanyl/bupivacaine solutions tended to be the latter. The initial rates (day 0) were chosen by the anesthesiologist for the case and were based on an estimate of patient's postoperative requirements. These rates were subsequently modified, as necessary, by acute pain service (APS) personnel. The management of inadequate analgesia required an assessment by the APS resident, usually followed by increasing the infusion rate by 2-4 ml/h with a 5-ml bolus of the infusion mixture, if required. Supplemental analgesia was limited to nonsteroidal antiinflammatory drugs (NSAIDs) because of concern with adding additional parenteral opioids to an existing epidural opioid infusion. The use of supplemental analgesics was recorded for the majority of the audit period.
In the early part of the study, catheter fixation was achieved with a clear adhesive dressing (Tegaderm) placed over the site and "windowed" with adhesive tape (Sleek). Unfortunately, any fluid collection (perspiration, blood, or local anesthetic solution) resulted in the clear dressing lifting, which enabled catheter movement and possible displacement. Since late 1992, the clear adhesive dressing was covered with a semi-opaque, firmer woven dressing (Mefix), with adhesive tape covering the catheter up the patient's back and over their shoulder.
Nursing standard orders for epidural observations and reporting are listed in Appendixes 1-4. These are printed on the ward "special analgesia observation chart" for each patient.
Patients were nursed on standard surgical wards unless they were required for medical or surgical reasons to spend 1 or 2 days in the ICU or HDU.
A data sheet about each patient was kept by the APS. Demographic data about the patient and their surgery, catheter site, and depth to the epidural space were recorded, along with details of infusion type and rate. Also recorded were details of complications or side effects and the need for supplemental analgesics.
Each day the duty resident and/or the attending consultant on call for the APS visited the patients, usually at 8 AM, 4 PM, and 11 PM. During these rounds, the patient's pain status was assessed and recorded and the analgesic regime modified, if necessary. Pain relief was scored by the observer after questioning the patient and asking them to breathe deeply, cough, and move about. It was rated on a scale of 1 to 5 according to the pain relief rating (PRR; Appendix 4).
Any problems, complications, or side effects were recorded and managed appropriately. Unpleasant sensory loss or unpleasant motor block was recorded. Pruritus was defined as itch of sufficient severity requiring specific management. Hypotension was defined as a systolic blood pressure of less than 100 mmHg and was recorded regardless whether it was believed to be due to the epidural. Nausea and vomiting were recorded when abnormally persistent or severe or when requiring adjustment in the analgesic regime. Respiratory depression was defined arbitrarily as a respiratory rate of less than or equal to 8 breaths/min independent of the patient's conscious state. Excessive sedation was defined as the inability to respond to spoken command or gentle shake. Naloxone was given for severe respiratory depression, usually associated with excessive sedation, as judged by the APS resident.
The epidural site was inspected daily, particularly for signs of infection at the skin, and the patient's progress was discussed with the patient's medical and nursing team.
Termination of the epidural infusion was recorded as "elective" if the catheter remained in situ for at least 3 postoperative days or discontinuation occurred because there was no further need for epidural infusion. Patient pyrexia higher than 38.5 degrees Celsius, later than 24 h postoperatively, was considered an indication for catheter removal if, on consultation with the surgical unit, the risks of retaining an epidural catheter in the presence of possible bacteremia outweighed the benefits. Other reasons for termination of infusion were noted.
Data were entered into a microcomputer database (Microsoft Excel), and statistical analysis was carried out using "Statview" (Abacus Concepts, Los Angeles, CA). Continuous data are expressed as the mean plus/minus SD. Nonparametric data were analyzed using Fisher's exact test. A P value of less than 0.05 was taken to represent statistical significance.
Data were collected over the 40 months from February 1990 to May 1993. During this time, 1,297 epidural infusions were managed on the APS, of which 1,014 patients received fentanyl and bupivacaine mixtures, and 283 received solutions of either local anesthetic alone or other mixtures.
Patient ages ranged from 5 to 94 yr with an average age of 56.9 (plus/minus 17.4 SD) yr. The duration of infusion ranged from < 1 to 13 days, with an average of 3 days and a mode of 3 days (Figure 1). Overall, this amounted to 3,005 patient-days of epidural infusion. Catheters were inserted at a segmental level appropriate to the surgical wound, which involved a large proportion of thoracic-level catheters (Figure 2). The type of surgery the patients received is listed in Table 1, along with the duration of infusion, catheter insertion level, and patient age for each operative category. Nonsurgical patients included analgesia for chest trauma (fractured ribs) and pancreatitis.
Of the 1,014 patients, the majority (89%) returned directly from the recovery room to the normal surgical wards. The remaining 11% (109 patients) were admitted to the ICU or HDU for postoperative monitoring. The decision regarding ICU or HDU admission was made independently of the presence of epidural analgesia. Sixty-eight percent of these patients were discharged to the normal surgical wards after 24 h or less of monitoring, with the epidural infusion continuing.
Quality of analgesia, as recorded on the ward rounds, was high overall. Terms are defined in Appendix 4. On day 0 (the day of surgery), 60.4% of patients had continuous pain relief, with an additional 24.8% having satisfactory relief, and 14.8% breakthrough pain. This was similar for day 1; however, as can be seen from Figure 3, the proportion of continuous relief patients increased to 81.3% and 87% on days 2 and 3, respectively, whereas the breakthrough pain groups decreased to 7.1% and 4.3%, respectively. Overall, pain relief ratings of 4 or 5 were achieved on 83% of visits, a rating of 3 on 11% of visits and a rating of 1 or 2 on 6% of visits.
Since January 1991, the use of supplemental analgesics was recorded. Over this time, 23% of patients received NSAIDs in addition to an epidural fentanyl/bupivacaine infusion. Indomethacin suppository use declined after the introduction of intramuscular ketorolac in mid 1992. The main areas of NSAID use were in thoracic (34%) and upper abdominal (36%) patient groups, whereas vascular patients had the least use (11% of patients).
Infusion rates for the epidural solutions increased slightly by day 1 (Table 2). In 7% (68) of cases, the initial infusion solution was changed to another solution to improve analgesia or reduce specific side effects. On 34 occasions, this change involved switching to a nonopioid epidural infusion solution.
Termination of epidural analgesic therapy was elective in 59% of patients, with a further 3% being at the request of the surgical unit (Table 3). Technical catheter- or infusion pump-related problems occurred in 19% of infusions--two-thirds of these were due to catheter dislodgment within the first 72 h. In 13 cases, the epidural catheter was reinserted to continue the infusion from a different level. Since late 1992, after the change in the catheter fixation technique (see Methods), the incidence of catheter displacement had decreased from 15% to 8% (overall rate for the study period 13%). A fever higher 38.5 degrees Celsius, later than 24 h postoperatively, accounted for 74 (8%) of catheter removals, 17 of which were associated with catheter insertion site inflammation (see below).
Inadequately controlled analgesia accounted for 8% (76) of cases of cessation of epidural analgesic therapy. The commonest surgical group was urology patients (18 cases of 135 (13%)), and the least common were in vascular patients (0 cases of 44). Such patients were managed with alternate therapies and accounted for 23% of patients treated on each of days 1 to 3, postoperatively.
Side effects resulted in the cessation of 44 infusions (Table 4). Hypotension accounted for 17 of these (39%), although hypotension occurred in 67 patients overall and was adequately managed in the remainder without discontinuing the infusion. Pruritus occurred in 104 patients (10%) but required cessation of therapy in only 12 cases. Confusion or sedation, usually in elderly patients (average age 65.7 plus/minus 19.6 yr), was present in 75 patients (7%), and the epidural opioid infusion was stopped or changed to local anesthetic alone in 11 of these. One patient was given a single dose of naloxone for excessive drowsiness (when the respiratory rate was 10 breaths/min). Other side effects included anesthesia (not liked by the patient) in 3% of cases and undesirable motor block of one or both legs in 3%. Undesired anesthesia was significantly more common in the 0.1/1 solution (which was infused at higher rates; Table 2; P < 0.05) than in the 0.1/5 solution, although this did not create significantly more limb motor block nor hypotension. In comparing the 0.1/1 solution with the 0.1/5 solution, there is a significantly greater proportion of confusion or drowsiness and pruritus (P < 0.01, for both), which may relate to the larger fentanyl dose in the latter group. This was not apparent, however, when comparing with the smaller number of patients receiving the 0.1/10 solution.
Respiratory depression occurred in 12 patients (1.2%; Table 5). No patient had a respiratory arrest. The average age of these patients was 65.5 plus/minus 17.2 yr (mean plus/minus SD), which is not statistically different from the average age of patients without respiratory depression (56.8 plus/minus 17.1, P = 0.10). There was a wide range of catheter insertion levels. Respiratory depression occurred within hours of emergence from anesthesia to 6 days postoperatively, with the commonest time being 2 days. All events were detected by nursing staff following the normal observation protocols (see Methods). Inappropriate sedation was present in ten of these cases. All cases were assessed promptly by a medical practitioner, and a decision to use naloxone was made in four patients (0.4%). The other patients were managed by a reduction or temporary cessation of the infusion in six patients, a change to nonopioid epidural infusion in one case, and discontinuation of the infusion in the remainder. Two patients were transferred to the ICU or HDU for further monitoring, one of whom received a continuous naloxone infusion for 12 h. One patient had experienced respiratory failure, and although small pupils were noted, she also had congestive cardiac failure and pneumonia and was admitted to the ICU for ventilator support.
There were 38 cases (3.8%) of insertion site inflammation or infection. Twenty-one infusions were terminated because of site inflammation or local purulence without fever. No patient experienced an epidural space infection. Of the 38 cases, 2 had swelling and tenderness present at the catheter insertion site; in 17, there was erythema at the site; and in 16, small amounts of expressible pus was present (half of these patients had another systemic focus of infection). Three cases had subcutaneous abscesses. Two required drainage in the ward, one growing Pseudomonas aeruginosa (the same organism that had been cultured from the blood 1 day previously), the other growing Enterobacter and Acinetobacter species. One abscess developed in a neutropenic patient with Crohn's disease who required drainage in the operating room (Staphylococcus aureus was cultured). The duration of catheter insertion in these patients ranged from 1 to 6 days, with a mode of 3 days. Of the 16 purulent sites cultured, 1 had no growth, 5 had sensitive S. aureus, 5 grew S. epidermidis, and the remainder grew miscellaneous skin flora. No patient developed long-term morbidity from these infections, although one patient showed a cholestatic jaundice, which may have been due to an antibiotic used to treat a site infection.
Two patients had epidural hematomas. One patient developed an epidural hematoma after removal of the catheter. It presented as leg and back pain occurring within minutes of catheter withdrawal (after 5 days of infusion), which was variable in intensity and transiently associated with leg weakness. Neurosurgical assessment and computed tomography scan confirmed the presence of the hematoma; but with evidence of rapid resolution of symptoms and signs, conservative management resulted in complete recovery over the next week. There was no history of recent anticoagulant use, apart from indomethacin. This case is being prepared for submission as a case report. The second case of epidural hematoma was in an 82-yr-old patient who was transferred to our ICU moribund from another hospital with an epidural catheter in situ after a major operative blood loss from abdominal surgery. On wakening, when the patient was removed from artificial ventilation 2 days later, it was apparent that she could not move her legs. A magnetic resonant imaging (MRI) scan confirmed the presence of a hematoma at the level of the epidural catheter entry site. Because of the course of events and the condition of the patient, no benefit was considered likely from operative intervention. She subsequently was transferred back to her previous nursing home with a permanent neurologic deficit.
Other complications included a grand mal convulsion in a 61-yr-old patient on day 1 postoperatively, who was receiving an epidural infusion of the 0.1/10 solution at 4 ml/h. Neurologic consultation and investigation failed to reveal a cause. One patient had an area of anesthesia on his buttock persisting on discharge on day 8. Another patient had persistent hypotension and extensive block in recovery and, after restarting the infusion (0.1/5 solution) in the ward, developed a very extensive block. The catheter tip was found to be subarachnoid and was removed without further incident.
This prospective audit shows that epidural fentanyl and bupivacaine infused at appropriate dermatomal levels for the surgery is efficacious in the general surgical ward and is associated with few serious complications. The PRR incorporates an assessment of pain relief at rest and on activity. Our aim was to achieve a PRR of 4 or 5, which would allow mobilization and cooperation with chest physiotherapy. A score of 3 was considered equivalent pain relief to that provided by intramuscular opioids. It is recognized that such observer-generated analgesic scoring may introduce some bias; however, it is a practical and widely used method for evaluating pain in the clinical setting. The quality of pain relief achieved was "continuous" (PRR always greater than or equal to 4) on 82.6% of visits (Figure 3). This improved progressively from day 0 to day 3. The dropout rate due to inadequate analgesia was low and does not account for the improvement over time. Many factors may explain such improvement, including optimizing the rate of infusion for individual patients, adding supplemental NSAIDs, and diminishing analgesic requirements in the postoperative period. In cases of poor pain control, it is important to be certain that the epidural catheter is not malpositioned or dislodged. The quality of pain relief varied between surgical groups. The high incidence of catheter removal for inadequate analgesia in urology patients mostly occurred in postnephrectomy patients who had difficulties with shoulder tip pain due to diaphragm irritation. The vascular patients had the least number of catheters removed for inadequate pain relief, which may be explained partly by the decreasing analgesic requirement with age (Table 1).
Both fentanyl and bupivacaine have a segmental effect when administered spinally. [20,21]Therefore, the epidural catheters were placed at sites appropriate for the site of the surgery, and this has resulted in the large numbers of thoracic epidural placements (Figure 2and Table 1). This was done to improve the quality of analgesia. In addition, this has been shown to reduce the amount of fentanyl infused and reduce side effects. When compared to the lumbar route, the thoracic route requires less fentanyl for postthoracotomy pain. [4,19]If the catheter is placed away from the level of surgery, low-dose bupivacaine has no added advantage over fentanyl alone. 
The optimal concentration of fentanyl for epidural infusion is not known, although the majority of papers dealing with epidural fentanyl report the use of fentanyl in 10 micro gram *symbol* ml sup -1 concentration. [3,23-31]After initially using this concentration with 0.1% bupivacaine, we found that we were able to reduce the concentration of fentanyl to 5 micro gram *symbol* ml sup -1, without a proportionate increase in infusion rate, and we were able to achieve the same analgesia with less fentanyl. The 1-micro gram *symbol* ml sup -1 solution with bupivacaine was frequently inadequate and required high infusion rates, resulting in extensive areas of anesthesia.
The safety of epidural fentanyl infusions depends primarily on the reduction of risk of respiratory depression, when compared with other parenteral opioids, and with epidural morphine. The reported incidence of respiratory depression with epidural morphine ranges from 0.07% to 0.9%, [10,15-17]which is lower than our incidence of 1.2%. There is wide variation in defining respiratory depression and respiratory rate, arterial carbon dioxide tension, and ventilatory response to inhaled carbon dioxide have been used. Most of the above citations have defined respiratory depression as that requiring naloxone. Using this criteria, only 4 (0.4%) of our patients had respiratory depression, and thus our incidence is similar to epidural morphine. The exception is the study by de Leon-Casasola, with a reported incidence of respiratory depression of 0.07%, defined as a respiratory rate of less than 10 breaths/min. Although all patients received naloxone, none were transferred to the ICU. The routine monitoring of our patients by nursing staff in the surgical ward, involving hourly observations of respiratory rate and 1- to 2-hourly observations of level of sedation, was sufficient to detect these patients. All the episodes of respiratory depression in this survey occurred during the time of the infusion but at varying times after the start of the infusion (Table 5). This highlights the need for monitoring throughout the infusion period. Ten of the patients also had a depressed conscious state. Although it is possible to have respiratory depression without a low respiratory rate, [8,16,19,32]sedation should be apparent. There were no cases in which respiratory depression, or sedation, occurred after cessation of the infusion. Hydrophilic drugs, such as morphine, may be associated with late respiratory depression up to 24 h after an epidural dose. [8,12]This has important implications for safety and ease of patient management. Late respiratory depression probably is caused by rostral spread of CSF-containing opioid. Lipophilic opioids such as fentanyl are less likely to remain in the CSF for long enough after cessation of administration to cause this problem. The incidence of respiratory depression in Ready's study was 0.2%, and the two episodes occurred 8.5 and 19.5 h after the initial dose of morphine. Both these patients required naloxone. In Stenseth's study wherein 0.9% had respiratory depression, episodes occurred 1.5-7 h after a previous dose of epidural morphine. Naloxone was given to 8 of the 10 patients with respiratory depression, with most requiring more than one dose. 
The dose of opioid is said to be important. Weightman reported three cases of respiratory depression in patients who received more than 1.5 micro gram *symbol* kg *symbol* sup -1 h sup -1 fentanyl. In their cases, the concentration of fentanyl was high (20-25 micro gram *symbol* ml sup -1). In our series, the average infusion rate for patients who had respiratory depression was 0.74 micro gram *symbol* kg *symbol* sup -1 h sup -1. However, a wide range of infusion rates was used, and we are unable to confirm a dose relationship for respiratory depression. Other factors that may be important in respiratory depression include increasing age and concomitant opioid use. In this study, the average age of patients with respiratory depression was not statistically different from those without. Standing orders for patients receiving epidural opioids include the avoidance of other parenteral opioids. This may be of less concern in the light of de Leon-Casasola's study, whose protocol allowed for 2-4 mg intravenous morphine to be given to supplement continuous epidural morphine/bupivacaine infusion, with a resultant low respiratory depression rate. 
Sedation or confusion was observed in 74 (7.4%) of patients, without alteration in respiratory rate. It is possible that some of these patients had some degree of respiratory depression not manifested as a decrease in respiratory rate. Management consisted of infusion rate change or changing the solution to a nonopioid-containing epidural solution. Naloxone was given to one patient, who was difficult to rouse but did not appear have respiratory depression.
Pruritus was noted in 10.3% of cases and was the most common side effect. Ready's series had a 25% incidence and Stenseth's series 11%. Our incidence represents the number of cases that required active management. This consisted of a change in infusion rates, antihistamines, low-dose naloxone, or, in 12 patients, change to a nonopioid-containing solution. Nausea and vomiting was recorded in 31 cases (3.1%), and this probably was underreported, but many factors contribute to postoperative nausea and vomiting. Urinary retention was not recorded frequently, and many patients had a urinary catheter inserted as part of the operative procedure.
Mechanical problems accounted for 19% of catheter discontinuations with dislodgment of the epidural catheter being the most common mechanical problem (13%). Two of the large series auditing epidural analgesia did not mention mechanical problems. [15,18]However, Ready's series had a 3% premature dislodgment, and de Leon-Casasola's series had a 1.6% dislodgment rate. Of note, Sawchuck had one-third of his thoracic epidural catheters dislodge prematurely versus none in the lumbar group. Epidural catheters placed higher on the back may be more prone to pressure and friction, which may explain the higher dislodgment rate. We were able to reduce our dislodgment rate to 8% by reinforcing the dressing with Mefix. Other mechanical problems (6.1%) were due to kinking, occlusion, disconnection, or infusion pump malfunction and will be reduced by improved education of staff regarding prevention and management of these problems. One case of subarachnoid placement of an epidural catheter occurred because of unrecognized initial intrathecal placement, although catheter migration into subarachnoid space can occur. The presence of bupivacaine makes motor block likely if subarachnoid placement occurs, and this aids early detection of catheter migration. Migration into an epidural vein should not cause concern, because the low bupivacaine concentration is unlikely to cause toxicity, and the fentanyl doses used (Table 2) are low for intravenous analgesia with fentanyl. [4,27]However, Chien et al. has reported one case of respiratory depression due to intravascular migration of the catheter.
Inflammation of the catheter insertion site occurred in 38 (3.8%) of the cases, including 16 (1.58%) that were purulent, and is of concern. The inflammation rate in de Leon-Casasola's series was 0.57% with a similar proportion having a purulent catheter site (12/24). In our study, the duration of therapy was not related to the occurrence of site infection, because infection/inflammation occurred from day 1 to day 6. The reason for the high inflammation rate in our study is not clear. Relevant factors may include different patient populations, different surgical cases, and a greater proportion of thoracic catheters in our study compared to de Leon-Casasola's study. Other causes include lapses in aseptic technique during catheter insertion, the security of the dressing (mentioned above), catheter movement at the site, and the presence of other sites of infection, as was the case in eight of the patients with purulent catheter sites. Despite the presence of pus at the catheter insertion site, there were no deep infections.
Pyrexia during postoperative epidural infusion raises significant management issues. It is not uncommon for patients to have temperatures greater than 38.5 degrees Celsius in the first 24 h postoperatively. Persistent or recurrent fever may be due to bacteremias, which could seed to the epidural catheter or to a small epidural hematoma at the catheter insertion site. For this reason, such patients were discussed with the surgical team and a decision to remove the catheter was made on an individual risk versus benefit basis, bearing in mind that good analgesia would help with chest physiotherapy and that postoperative chest infection is a common cause of postoperative fever.
Of some concern are the two epidural hematomas that occurred during the study period. Although rare, they can lead to serious morbidity. The first case demonstrates that there is a risk associated with catheter removal, and correction of coagulopathy has been advised before catheter withdrawal. In this case, indomethacin was the only recently administered agent that may have had an effect on coagulation. The spontaneous resolution demonstrates a lack of severity of the bleed, but it must be emphasized that conservative management in these circumstances requires frequent observation, and was only done in this case with ongoing neurosurgical consultation. In the second case, a complex scenario involving large blood loss, patient transport, and ICU care meant that it is difficult to identify the primary cause of the problem. She was included in our audit because the ICU referred her to our service when she was discharged to the ward, making the assumption that her leg immobility was due to a local anesthetic effect. By this time, it unfortunately was too late for surgical intervention. The use of a low concentration of epidural bupivacaine is associated infrequently with motor block of the lower limbs.
On discontinuation of epidural analgesia, patients were changed to other modes of analgesia, usually to opioids by alternate routes or oral NSAIDs. Many of our epidural infusions were supplemented using NSAIDs. In the early part of this study, this was usually indomethacin suppositories. Intramuscular ketorolac has been shown to be a useful adjuvant to epidural analgesia after radical retropubic prostatectomy, and in our practice, it was useful for procedures resulting in pain or discomfort well away from the operative site such as shoulder tip pain following thoracotomy or nephrectomy. These agents have proven to be a most effective analgesic supplement, although careful patient selection is required to minimize the incidence of side effects.
Since 1990, the APS has been involved in upgrading the quality of care of patients receiving epidural infusions. This has been achieved through education, standardized infusion and management protocols, and the establishment of formal pain rounds. We believe that the safe management of epidural fentanyl in the general surgical ward is contributed significantly to by the activities of the APS.
The opioid-related side effects of pruritus, nausea and vomiting, sedation, and respiratory depression may be reduced by further decreasing the concentration of fentanyl in our infusion mixtures. Although the 0.1/1 solution resulted in high infusion rates and local anesthetic side effects, a concentration in the range of 2-3 micro gram *symbol* ml sup -1 with 0.1% bupivacaine may be effective, provided it does not result in a proportional dose-related increase in infusion rates. We currently are evaluating a concentration of 2.5 micro gram *symbol* ml sup -1.
The use of epidural fentanyl and bupivacaine for postoperative analgesia is an effective method of providing pain relief. We believe that, with careful patient and drug selection and with the use of epidural catheters placed appropriately for the proposed surgery, that high-quality postoperative pain relief can be offered. When compared to reviews of epidural morphine or epidural morphine and bupivacaine, the side effects spectrum of epidural fentanyl with bupivacaine is similar. A formalized APS is a significant factor in the quality of pain relief and the reduction of complications. Patients must be managed by suitably trained nursing and medical staff using protocols for the detection and management of any side effects or complications. The incidence of local anesthetic complications was low. Of concern, however, is the incidence of catheter-related mechanical problems and catheter site inflammation. Improved fixation techniques may improve this and reduce site inflammation by decreasing movement of the catheter at the skin entry site. The respiratory depression incidence of 1.2% (0.4% requiring naloxone) was detected early and easily by nursing staff observations of respiratory rate and sedation level. The opioid-related side effects of epidural fentanyl/bupivacaine solutions may be reduced by decreasing the fentanyl concentrations in these mixtures. No episodes of respiratory depression occurred after cessation of the epidural infusions.