Portal de Periódicos da CAPES

Effect of peri‐ and postoperative epidural anaesthesia on pain and gastrointestinal function after abdominal hysterectomy

2001; Elsevier BV; Volume: 87; Issue: 4 Linguagem: Inglês

10.1093/bja/87.4.577

1471-6771

Hannah Joan Jørgensen, Jonna Storm Fomsgaard, Jesper Dirks, Jørn Wetterslev, B. Andreasson, J. B. Dahl,

Nausea and vomiting management

In a double blind study we have investigated the effects of epidural local anaesthesia (LA), when added to general anaesthesia (GA) and postoperative paracetamol and NSAID, on postoperative pain and gastrointestinal function in patients undergoing open hysterectomy. Sixty patients were randomized into three study groups: GA, and postoperative paracetamol and NSAID (GA, n=20); GA, paracetamol, NSAID, intraoperative epidural lidocaine and 24‐h postoperative epidural saline (Saline, n=20); or GA, paracetamol, NSAID, intraoperative epidural lidocaine and 24‐h postoperative epidural bupivacaine (Bupi, n=20). Patients were observed for 72 h postoperatively. Pain at rest, during cough, and mobilization, request for supplementary morphine, and time to first postoperative flatus, was reduced in patients receiving 24‐h postoperative epidural anaesthesia, compared with the two other groups. However, these effects of epidural LA, were not sustained beyond the period of infusion, and no differences in PONV, time to first postoperative defecation, mobilization or time to discharge from hospital were observed between groups. A 24 h postoperative epidural infusion with bupivacaine, when added to postoperative paracetamol and NSAID, reduces pain and opioid requirements, but has only limited effects on gastrointestinal function and patient recovery. In a double blind study we have investigated the effects of epidural local anaesthesia (LA), when added to general anaesthesia (GA) and postoperative paracetamol and NSAID, on postoperative pain and gastrointestinal function in patients undergoing open hysterectomy. Sixty patients were randomized into three study groups: GA, and postoperative paracetamol and NSAID (GA, n=20); GA, paracetamol, NSAID, intraoperative epidural lidocaine and 24‐h postoperative epidural saline (Saline, n=20); or GA, paracetamol, NSAID, intraoperative epidural lidocaine and 24‐h postoperative epidural bupivacaine (Bupi, n=20). Patients were observed for 72 h postoperatively. Pain at rest, during cough, and mobilization, request for supplementary morphine, and time to first postoperative flatus, was reduced in patients receiving 24‐h postoperative epidural anaesthesia, compared with the two other groups. However, these effects of epidural LA, were not sustained beyond the period of infusion, and no differences in PONV, time to first postoperative defecation, mobilization or time to discharge from hospital were observed between groups. A 24 h postoperative epidural infusion with bupivacaine, when added to postoperative paracetamol and NSAID, reduces pain and opioid requirements, but has only limited effects on gastrointestinal function and patient recovery. Abdominal hysterectomy is associated with moderate to severe postoperative pain, 1Kuhn S Cooke K Collins M Jones JM Mucklow JC. Perceptions of pain relief after surgery.BMJ. 1990; 300: 1687-1690Crossref PubMed Scopus (158) Google Scholar, 2Stanley G Appadu B Mead M Rowbotham DJ. Dose requirements, efficacy and side effects of morphine and pethidine delivered by patient‐controlled analgesia after gynaecological surgery.Br J Anaesth. 1996; 76: 484-486Crossref PubMed Scopus (63) Google Scholar, 3Eriksson‐Mjoberg M Svensson JO Almkvist O Olund A Gustafsson LL. Extradural morphine gives better pain relief than patient‐controlled i.v. morphine after hysterectomy.Br J Anaesth. 1997; 78: 10-16Crossref PubMed Scopus (31) Google Scholar significant nausea and vomiting 4Cruickshank RH Spencer A Ellis FR. Pretreatment with controlled‐release morphine for pain after hysterectomy.Anaesthesia. 1996; 51: 1097-1101Crossref PubMed Scopus (11) Google Scholar, 5McIndoe AK Warwick P O'Connor M. A comparison of retrospective versus contemporaneous nausea scores with patient‐controlled analgesia.Anaesthesia. 1996; 51: 333-337Crossref PubMed Scopus (9) Google Scholar, 6Warriner CB Knox D Belo S Cole C Finegan BA Perreault I. Prophylactic oral dolasetron mesylate reduces nausea and vomiting after abdominal hysterectomy. The Canadian Dolasetron Study Group.Can J Anaesth. 1997; 44: 1167-1173Crossref PubMed Scopus (36) Google Scholar, 7Asantila R Eklund P Rosenberg PH. Continuous epidural infusion of bupivacaine and morphine for postoperative analgesia after hysterectomy.Acta Anaesthesiol Scand. 1991; 35: 513-517Crossref PubMed Scopus (39) Google Scholar and reduced gastrointestinal (GI) motility. Reduced GI motility is considered to be a result of nociceptive and sympathetic neural activity induced by surgery, and may last for 2–5 days, depending on the technique applied for anaesthesia and postoperative analgesia. 7Asantila R Eklund P Rosenberg PH. Continuous epidural infusion of bupivacaine and morphine for postoperative analgesia after hysterectomy.Acta Anaesthesiol Scand. 1991; 35: 513-517Crossref PubMed Scopus (39) Google Scholar, 8Wattwil M Thoren T Hennerdal S Garvill JE. Epidural analgesia with bupivacaine reduces postoperative paralytic ileus after hysterectomy.Anesth Analg. 1989; 68: 353-358Crossref PubMed Scopus (111) Google Scholar, 9Thorn SE Wattwil M Naslund I. Postoperative epidural morphine, but not epidural bupivacaine, delays gastric emptying on the first day after cholecystectomy.Reg Anesth. 1992; 17: 91-94PubMed Google Scholar, 10Jorgensen H Wetterslev J Moiniche S Dahl JB. Epidural local anaesthetics versus opioid‐based analgesic regimens on postoperative gastrointestinal paralysis, PONV and pain after abdominal surgery (Cochrane Review).in: The Cochrane Library. Update Software, Oxford2000Google Scholar Epidural local anaesthetics (LA) results in effective pain relief 11Jorgensen H Fomsgaard JS Dirks J Wetterslev J Dahl JB. Effect of continuous epidural 0.2% ropivacaine vs 0.2% bupivacaine on postoperative pain, motor block and gastrointestinal function after abdominal hysterectomy.Br J Anaesth. 2000; 84: 144-150Crossref PubMed Scopus (30) Google Scholar and improved GI motility compared with opioid‐based analgesia after hysterectomy. 8Wattwil M Thoren T Hennerdal S Garvill JE. Epidural analgesia with bupivacaine reduces postoperative paralytic ileus after hysterectomy.Anesth Analg. 1989; 68: 353-358Crossref PubMed Scopus (111) Google Scholar, 7Asantila R Eklund P Rosenberg PH. Continuous epidural infusion of bupivacaine and morphine for postoperative analgesia after hysterectomy.Acta Anaesthesiol Scand. 1991; 35: 513-517Crossref PubMed Scopus (39) Google Scholar, 12Thoren T Sundberg A Wattwil M Garvill JE Jurgensen U. Effects of epidural bupivacaine and epidural morphine on bowel function and pain after hysterectomy.Acta Anaesthesiol Scand. 1989; 33: 181-185Crossref PubMed Scopus (78) Google Scholar However, relative importance of spinal reflex activity induced by surgery/postoperative pain, and perioperative opioid administration, for the reduction in GI motility, is not settled. 13Holte K Kehlet H. Postoperative ileus: a preventable event.Br J Surg. 2000; 87: 1480-1493Crossref PubMed Scopus (488) Google Scholar Hence, the beneficial effect of epidural LA on postoperative GI function may be a result of neural block during and/or after surgery reduced opioid consumption, or both. Paracetamol and NSAIDs have been shown to reduce pain and the need for opioids after hysterectomy. 14Cobby TF Crighton IM Kyriakides K Hobbs GJ. Rectal paracetamol has a significant morphine‐sparing effect after hysterectomy.Br J Anaesth. 1999; 83: 253-256Crossref PubMed Scopus (75) Google Scholar, 15Blackburn A Stevens JD Wheatley RG Madej TH Hunter D. Balanced analgesia with intravenous ketorolac and patient‐ controlled morphine following lower abdominal surgery.J Clin Anesth. 1995; 7: 103-108Abstract Full Text PDF PubMed Scopus (52) Google Scholar In a recent study, combined paracetamol, NSAID, and continuous 24 h infusion of epidural LA provided effective postoperative pain relief after hysterectomy. Median time to first passage of flatus and faeces was less than 48 h. It was not clear from this study, however, whether the relatively short duration of GI paralysis was because of administration of intra‐ and postoperative epidural LA, an opioid‐sparing effect of paracetamol and NSAID, or both. 11Jorgensen H Fomsgaard JS Dirks J Wetterslev J Dahl JB. Effect of continuous epidural 0.2% ropivacaine vs 0.2% bupivacaine on postoperative pain, motor block and gastrointestinal function after abdominal hysterectomy.Br J Anaesth. 2000; 84: 144-150Crossref PubMed Scopus (30) Google Scholar The aim of this study was to compare the effect of combined general anaesthesia and postoperative paracetamol and NSAID, with or without intraoperative, or intra‐ and 24 h postoperative epidural LA, on pain and GI function after abdominal hysterectomy. Sixty patients aged 18–75 yr undergoing elective abdominal total or supravaginal hysterectomy through a Phannenstiel or inferior median incision were included in the study. Informed written consent was obtained from all patients and approval was obtained from the local Ethics Committee and the Danish National Board of Health. Patients were recruited from the Department of Gynaecology, Herlev University Hospital, during the period January 1999 to September 2000. Patients were not included if they were unable to cooperate, had a body weight ≥ (1.3×body height (cm) – 100), moderate to severe heart or lung disease (ASA ≥III), a history of drug or alcohol abuse, chronic pain or daily intake of analgesics, or had contraindications to NSAIDs or to insertion of an epidural catheter. Preoperatively, patients were randomized by a computer program to general anaesthesia (the GA group), or combined general and epidural anaesthesia, using sealed envelopes. For patients randomized to combined general and epidural anaesthesia, a number referring to the postoperative epidural study drug, was stated in the envelope. One hour before surgery patients had sublingual triazolam 0.125 mg. Patients randomized to GA were induced with propofol 1.5–2.5 mg kg–1, alfentanil 1.0 mg, fentanyl 0.3 mg, and rocuronium 0.6 mg kg–1, and a tracheal tube was inserted. Propofol was given without nitrous oxide for maintenance of anaesthesia. If arterial pressure and/or heart rate increased to greater than 120% of initial values, fentanyl 0.1 mg was administered. Patients randomized to epidural treatment, had an 18‐gauge epidural catheter inserted through a Touhy needle and advanced 4–5 cm into the epidural space at the T10–11 level, and a 4 ml test dose of 2% lidocaine with ephinephrine was administered. A bolus dose of 16 ml of 2% lidocaine with epinephrine was administered, followed by continuous infusion of 8 ml h–1 of 2% lidocaine throughout the operation. General anaesthesia was induced with propofol 1.5–2.5 mg kg–1 and alfentanil 1.0 mg, and a laryngeal mask airway (LMA †LMA is the property of Intavent Limited. ) was inserted. Propofol without nitrous oxide was given for maintenance of anaesthesia. No additional doses of lidocaine or alfentanil were given. For all patients, hypotension was treated with i.v. infusion of isotonic sodium chloride, HAES 6% or ephedrine 5 mg i.v. in incremental doses, when systolic arterial pressure was below 90 mm Hg. Fluid and blood losses were replaced according to the prescriptions of the department. At skin closure, patients in the epidural group received bupivacaine 0.2% (the Bupi group) or saline (the Saline group) as a bolus of 8 ml and continuous infusion of 8 ml h–1 for 24 h. The study drugs were prepared by the hospital pharmacy in identical containers, marked with a number, the name of the project and the investigators name. This part of the study was double blinded. For the first 48 h after surgery, all patients received 2 G paracetamol rectally every 8 h, thereafter 1 G orally every 6 h. For the first 72 h, all patients received ketorolac 30 mg i.m. every 8 h. If additional analgesic was required, morphine 0.125 mg kg–1 was given. Patients were given ondansetron 4 mg i.v., if they demanded antiemetics. If an additional antiemetic was required before another dose of ondansetron could be administered (8 h), metoclopramide 10 mg i.v. was given. All patients were instructed preoperatively on the use of the visual analogue scale (VAS) pain score, and to request supplementary analgesic and antiemetics if needed. Pain scores were assessed by the patients at 4, 6, 22, 24, 26, 30, 48, and 72 h after the operation on a VAS (0 mm=no pain, 100=worst pain imaginable) at rest, on coughing, and on mobilization from the supine to the sitting position. The number of morphine doses was recorded at the same times. Levels of sensory block to pinprick were assessed bilaterally at 6, 24, 26, and 30 h after operation; in the event of asymmetrical block the most extensive spread was recorded. Motor block was assessed using a four‐point modified Bromage scale (0=no motor block, 1=inability to raise extended legs, 2=inability to flex knees, 3=inability to flex ankle joints) at 6, 22, 24, 26, and 30 h after operation. Ability to walk on the floor (0=no difficulty, 1=little difficulty, 2=very difficult, 3=impossible, 4=difficult or impossible for other reasons than motor block) was assessed at 6, 22, 24, 26, 30, and 48 h after operation. The actual duration of mobilization ( 4 h) during the 24–48 and 48–72 h intervals was assessed at 48 and 72 h after the operation. Patients who were not able to mobilize at >4 h were asked for the primary reason for this. At all visits, gastrointestinal function was monitored by asking the patients if and when they had first passage of flatus and faeces, and if they had experienced nausea (0=no, 1=light, 2=moderate, 3=severe) and/or vomiting since last assessment. Patients were encouraged to eat and drink whenever possible after the operation. Ingestion of predefined quantities of food and beverages was assessed at 24, 48, and 72 h after the operation. Every 24 h after the operation, patients were assessed as ready or not ready for discharge from hospital, by four discharge criteria: (1) Normal defecation and no urinary retention: yes/no; (2) able to mobilize and dress: yes/no; (3) need for opioid: yes/no; (4) surgical complication requiring patient hospitalized: yes/no. When the patient scored yes on the two former, and no on the two latter questions, they were assessed ready for discharge from hospital. The primary end‐points of this study were pain during coughing, and time to first passage of faeces. Power calculation for the pain score was based on results from an unpublished pilot study performed at our institution (VAS pain scores during cough, in the general anaesthesia group, 24 h after the operation: 40 mm; type 1 error: 5%; type 2 error: 20%; minimal difference not to be overlooked: 20 mm reduced pain score; sample size needed: 16 patients in each group). Power calculation for time to first postoperative defecation was also based on the results from the unpublished pilot study in hysterectomy patients. (On day 3 after the operation, 17% have defaecated. Other type 1 error: 5%; type 2 error: 20%; minimal difference not to be overlooked: 50% reduction in time to first postoperative defecation: sample size needed 14 patients in each study group.) Data are presented as medians with upper and lower quartiles. Statistical analyses were performed using the Mann–Whitney rank sum test for unpaired data, Wilcoxon signed rank sum test for paired data, and Fisher's exact test for dichotomous data, where appropriate. If multiple testing was performed, significant P values were corrected with a Bonferroni factor for multiple comparisons. In addition, a longitudinal analysis of VAS pain scores was performed. In order to make the normality assumption more reasonable, we performed a square root transformation of the VAS scores before analysis. The model allowed for correlation between measurements on the same patient, in the form of a random level and an autoregressive error over time. Apart from 'time itself' (regarded as a categorical variable to allow for maximal flexibility) and treatment group, we considered the covariates 'duration of operation', 'age of patient' and 'Phannenstiel or median inferior incision'. The analyses were made for three different VAS pain scores: at rest, during cough and during mobilization, and each 24 h interval was treated separately. A Cox regression analysis with time to first postoperative faeces and flatus as dependent variables was performed. Covariates chosen beforehand were: treatment group, duration of operation, milligram of morphine before the event and age. P<0.05 was considered statistically significant. Calculations were performed using SPSS 9.0 for Windows and SAS 6.12 (PROC MIXED procedure). We considered 173 patients for inclusion in the study. One hundred and six patients were not included: 10 were aged more than 75 yr; four were unable to communicate properly; 13 were adipose; six had moderate to severe heart or lung disease or disseminated cancer; 10 used analgesics or suffered alcohol abuse; three had other surgical procedures in addition to hysterectomy; two needed anaesthetic procedures other than allowed in the protocol; 11 had contraindications to NSAIDs; 43 did not want to participate either because they wanted (n=13) or did not want (n=30) insertion of an epidural catheter; and four patients were not included because of lack of time. Sixty patients were included in the study, but seven of these were excluded: five after randomization but before operation (two because of withdrawal of consent after randomization to epidural catheter, two because it was not possible to insert an epidural catheter and one because of change in the surgical procedure); and two at 8 h (one because of haemorrhage and reoperation and one because of withdrawal of consent). The hospital pharmacy produced a new series of sealed envelopes and blinded medicine in replacement for the seven excluded patients, and seven new patients were included. Thus data from 60 patients were included and analysed. There were no significant differences between groups for patient characteristics or operative data, except for peroperative fentanyl and ephedrine (Table 1). Eleven and 10 patients in groups 2 and 3, respectively, needed i.v. ephedrine.Table 1Patient and perioperative data (median (range) or number)GroupGA groupSaline groupBupi groupNo. of patients202020Age (yr)54 (40–75)49.5 (32–70)50.5 (41–75)Weight (kg)68(50–90)73 (50–95)70 (54–90)Height (cm)167(160–175)165 (150–176)164(153–180)Phannenstiel/inferior median incision15/514/617/3Duration of surgery (min)88 (47–151)81 (55–141)75 (59–110)Perioperative propofol (mg)1285 (782–2123)1075 (607–2270)1185 (769–1629)Perioperative fentanyl (mg)0.5 (0.3–0.9)0 (0–0)0 (0–0)Perioperative haemorrhage (ml)275 (50–1600)200 (100–1200)200 (20–1000)Perioperative sodium chloride (ml)1350 (800–2400)1200 (500–2500)1200(1000–2200)Perioperative 6% HAES (ml)500 (0–1000)500 (0–1000)500 (0–1000)Perioperative ephedrine (mg)0 (0–0)5 (0–25)2.5 (0–60) Open table in a new tab VAS pain scores were significantly lower both at rest (P=0.006), during cough (P=0.01), and during mobilization (P=0.01) until 24 h after the operation in the Bupi group, compared with both the GA group and the Saline group (Fig. 1a–c). Following the first 24 h after the operation there were no significant differences. Neither the duration of operation, age of patient, nor type of incision had significant influence on pain scores according to the longitudinal analysis.Fig 1Median visual analogue scale (VAS) pain scores (a) at rest, (b) during coughing, and (c) during mobilization from the supine to the sitting position. VAS pain scores were significant lower both at rest, during cough and during mobilization until 24 h after the operation in the Bupi group, compared with both the GA group and the Saline group.View Large Image Figure ViewerDownload (PPT)Fig 1Median visual analogue scale (VAS) pain scores (a) at rest, (b) during coughing, and (c) during mobilization from the supine to the sitting position. VAS pain scores were significant lower both at rest, during cough and during mobilization until 24 h after the operation in the Bupi group, compared with both the GA group and the Saline group.View Large Image Figure ViewerDownload (PPT) Patients in the Bupi group requested a significantly smaller amount of supplementary morphine from 0–24 h compared with the other two groups. There were no significant differences in the number of patients requesting supplementary morphine, or in the number of doses, between groups at 24–48 or 48–72 h after the operation (Table 2).Table 2Number of patients who requested morphine, and number of doses administered. ns=not significantGroupPostoperative periodGA groupSaline groupBupi groupP0–24 hNo. of patients receiving morphine19/2020/2014/200.02No. of doses in patients receiving morphine87124350.00124–48 hNo. of patients receiving morphine7/206/209/20nsNo. of doses in patients receiving morphine121111ns48–72 hNo. of patients receiving morphine4/205/203/20nsNo. of doses in patients receiving morphine1078ns Open table in a new tab Preoperatively patients receiving epidural treatment had a sensory block extending from at least Th6 to L3 (median 13 dermatomes). Preoperatively and at 4 h after the operation, there were no significant differences in the spread of sensory block between the Saline and the Bupi group. At 6, 22, 24, and 26 h after the operation, patients in the Saline group had a smaller spread than those in the Bupi group (P 0.05); at 30 h the corresponding numbers were 75, 80, and 85% (P>0.05) (Fig. 2). At 24 h after the operation, the ability to walk could not be evaluated for different reasons. In the GA group, patients stated dizziness (15%) and nausea (15%) as primary reasons, in the Saline group pain (20%) and nausea (10%), and in the Bupi group motor block (20%) and dizziness (20%). There were no differences in the actual duration of mobilization (Fig. 3). The primary reasons for mobilizing <4 h were pain and fatigue both in the 24–48 and 48–72 h intervals.Fig 3Patient assessed actual duration of mobilization: 4 h. Columns show percentages of patients with a particular score in the GA‐, the Saline‐ and the Bupi groups at the 24–48 and 48–72 h intervals. No significant differences between groups.View Large Image Figure ViewerDownload (PPT) Time to first postoperative passage of flatus was significantly reduced in the Bupi group compared with the other two groups (P=0.009) whereas time to first postoperative defecation was not different between groups (P=0.70) (Fig. 4a and b). Time to first postoperative flatus and defecation was not influenced by duration of operation, amount of postoperative morphine, or age of patient according to the Cox regression analysis.Fig 4Fraction of patients without flatus (a) and defecation (b) in the postoperative period. Time to first postoperative passage of flatus was significantly reduced in the Bupi group compared with the other two groups (P=0.009) whereas time to first postoperative defecation was not different between groups (P=0.70)View Large Image Figure ViewerDownload (PPT) The three groups had a similar incidence of nausea and vomiting (Table 3). There were no significant differences in the number of patients requesting antiemetics or the number of doses between groups at 0–24, 24–48, and 48–72 h after the operation (P>0.2 and P>0.2) (Table 4). Side effects were of a mild nature in most patients, except for nine (three in the GA and six in the Saline group) who experienced severe nausea.Table 3Percentage of patients with light to severe nausea and percentage who vomited since last assessment in the three groupsTime after operation (h)0–44–66–2222–2424–2626–3030–4848–72Light to severe nauseaGA group2530402030151515Saline group4515602515255020Bupi group251530510251020Vomited since last assessmentGA group15102051510155Saline group1020501500100Bupi group0515055105 Open table in a new tab Table 4Number of patients who requested antiemetics (ondansetron or metoclopramide), and number of doses administeredGroupPostoperative periodGA groupSaline groupBupi groupP0–24 hNo. of patients receiving antiemetics13/2011/2010/20nsNo. of doses in patients receiving antiemetics332516ns24–48 hNo. of patients receiving antiemetics4/207/204/20nsNo. of doses in patients receiving antiemetics8124ns48–72 hNo. of patients receiving antiemetics1/204/201/20nsNo. of doses in patients receiving antiemetics151ns Open table in a new tab There were no significant differences between groups with respect to amount and time to ingestion of food and beverages, from 0–24, 24–48, or 48–72 h after the operation (data not shown). Most patients had an indwelling bladder catheter, which was removed on the morning after surgery. After removal, two patients, one from the Saline group and one from the Bupi group, complained of urinary retention, lasting a few hours. Patients fulfilled discharge criteria at median 3 (quartiles 3–4), 4 (3–6), and 5 (3–7) days after operation in the GA‐, Saline‐ and Bupi groups, respectively (ns). The challenge of modern anaesthesia and perioperative medicine is to create efficient treatment regimens with an optimal balance between protective and unwanted effects, in order to ensure patient safety and comfort, and to facilitate recovery. The main problems of abdominal hysterectomy in the immediate perioperative period are pain, PONV, and gastrointestinal paralysis, which may postpone recovery and discharge from hospital. The optimal anaesthetic regimen for this procedure should carry a low risk, along with effective pain relief and minimal, if not protective, effects on the gastrointestinal dysfunction and PONV induced by surgery. Hence, anaesthetics and analgesics, which produce effective analgesia with a low potential for gastrointestinal side effects, should be advantageous. This study was designed to investigate whether addition of intraoperative, or intraoperative plus 24 h postoperative epidural analgesia to an opioid‐sparing regimen with paracetamol and NSAID, was able to further reduce pain, improve gastrointestinal function, and enhance recovery. Results showed a significant reduction in pain and need for supplementary morphine during epidural treatment with bupivacaine. This effect was, however, not sustained beyond the time of the epidural infusion. Gastrointestinal function was only marginally improved, with a minor reduction in time to first postoperative flatus. Side effects, including PONV, were similar in the three study groups, as was recovery as measured by ability to ambulate, oral ingestion of food and fluid, and fulfilment of discharge criteria. Intraoperative epidural analgesia per se had no influence on postoperative pain or gastrointestinal function compared with the GA group. Hence, no protective analgesic effect of neither a short lasting, nor a 24‐h epidural regimen with LA was observed in this study. Previous studies, in patients undergoing abdominal hysterectomy 8Wattwil M Thoren T Hennerdal S Garvill JE. Epidural analgesia with bupivacaine reduces postoperative paralytic ileus after hysterectomy.Anesth Analg. 1989; 68: 353-358Crossref PubMed Scopus (111) Google Scholar and colonic surgery, 16Bredtmann RD Herden HN Teichmann W et al.Epidural analgesia in colonic surgery: results of a randomized prospective study.Br J Surg. 1990; 77: 638-642Crossref PubMed Scopus (115) Google Scholar, 17Liu SS Carpenter RL Mackey DC et al.Effects of perioperative analgesic technique on rate of recovery after colon surgery.Anesthesiology. 1995; 83: 757-765Crossref PubMed Scopus (432) Google Scholar, 18Scheinin B Asantila R Orko R. The effect of bupivacaine and morphine on pain and bowel function after colonic surgery.Acta Anaesthesiol Scand. 1987; 31: 161-164Crossref PubMed Scopus (118) Google Scholar have shown that an epidural regimen with LA reduces postoperative pain compared with regimens based on systemic opioids. The present study confirms these results. A longitudinal analysis of variance components, with pain score as the dependent variable, was performed in this study. The covariates, other than treatment group: duration of operation, age of patient, and type of incision were included in the analysis with a hypothesis generating purpose, but none of these showed significant influence on postoperative pain scores. However, it may be that the insignificant results could also be a result of the limited range of covariates and study size. In contrast to a meta‐analysis of previous studies, 8Wattwil M Thoren T Hennerdal S Garvill JE. Epidural analgesia with bupivacaine reduces postoperative paralytic ileus after hysterectomy.Anesth Analg. 1989; 68: 353-358Crossref PubMed Scopus (111) Google Scholar, 16Bredtmann RD Herden HN Teichmann W et al.Epidural analgesia in colonic surgery: results of a randomized prospective study.Br J Surg. 1990; 77: 638-642Crossref PubMed Scopus (115) Google Scholar, 19Ahn H Bronge A Johansson K Ygge H Lindhagen J. Effect of continuous postoperative epidural analgesia on intestinal motility.Br J Surg. 1988; 75: 1176-1178Crossref PubMed Scopus (107) Google Scholar, 20Riwar A Schar B Grotzinger U. Effect of continuous postoperative analgesia with peridural bupivacaine on intestinal motility following colorectal resection.Helv Chir Acta. 1992; 58: 729-733PubMed Google Scholar, 21Wallin G Cassuto J Hogstrom S Rimback G Faxen A Tollesson PO. Failure of epidural anesthesia to prevent postoperative paralytic ileus.Anesthesiology. 1986; 65: 292-297Crossref PubMed Google Scholar showing a reduction in time of postoperative gastrointestinal paralysis of about 40 h, 10Jorgensen H Wetterslev J Moiniche S Dahl JB. Epidural local anaesthetics versus opioid‐based analgesic regimens on postoperative gastrointestinal paralysis, PONV and pain after abdominal surgery (Cochrane Review).in: The Cochrane Library. Update Software, Oxford2000Google Scholar the effect of epidural LA on gastrointestinal function in the present study was only small. In contrast to previous studies (except for ref. 17), patients in the GA (control) group in our study received both paracetamol and NSAID, and this opioid‐sparing regimen may have reduced gastrointestinal paralysis and thus blunted the effect of epidural anaesthesia. This suggestion, however, has to be confirmed in controlled studies. No differences in the incidence of side effects was found in this study, but the relative small number of patients in each study group, does not allow firm conclusions. To be able to evaluate effects of the different regimens on recovery itself, criteria concerning ambulation, oral ingestion of food and fluid, and discharge from hospital were standardized. It should be emphasized, though, that fulfilling discharge criteria did not necessarily mean that patients were actually discharged from hospital. Thus, patients in the GA‐, the Saline‐, and the Bupi group were actually discharged from hospital on the fourth, fifth, and fifth day, respectively. The most frequent reason for this discrepancy was either a different assessment by the surgeon, or social causes concerning the patient. Still, no differences were observed between groups, and improved analgesia during the first postoperative 24 h was not followed by faster recovery in the Bupi group. In conclusion, a 24 h postoperative epidural infusion with bupivacaine, when added to postoperative paracetamol and NSAID, reduced pain and opioid requirements, but had no analgesic effects beyond the time of the epidural infusion, and only limited effects on gastrointestinal function and patient recovery after abdominal hysterectomy. Epidural anaesthesia may be superfluous in this surgical procedure. The study was funded by a grant from Snedkermester Axel Wichmann og fru Else Wichmann's fond. We thank Statistician Lene Theil Skovgaard, Department of Biostatistics, University of Copenhagen, for much help performing the analysis of variance components. Also we thank Research nurse Lene Henkel for much help and flexibility in assessing patients different times a day.