<<

Acute (2008) 10, 15—22

Indications and contraindications for thoracic epidural analgesia in multiply injured patients

Eileen M. Bulger a,∗, William T. Edwards b, Mario de Pinto b, Patricia Klotz a, Gregory J. Jurkovich a a Department of , University of Washington, Harborview Medical Center, Seattle, WA, USA b Department of , University of Washington, Harborview Medical Center, Seattle, WA, USA

Received 21 July 2007; received in revised form 17 October 2007; accepted 21 October 2007 Available online 21 December 2007 Presented at the American Society Regional and Pain Medicine, 2005 (poster)

KEYWORDS Summary Rib fractures; Background and objectives: Rib fractures are associated with significant morbid- Epidural analgesia; ity and mortality. Previous studies have demonstrated a significant reduction in Pain control pneumonia and duration of mechanical ventilation with epidural analgesia (EA) fol- lowing multiple rib fractures. There remains controversy regarding the appropriate indications and contraindications for EA in multiply injured patients. We sought to determine the factors underlying the decision to use or not to use EA in these patients. Methods: A survey was sent to the directors of services at all Amer- ican College of Committee on Trauma (ACS-COT) designated Level I trauma centers in the U.S. The survey queried their opinion regarding the appropriateness of 33 contraindications and eight indications for EA after rib fractures. Results: The response rate was 43% (81/188). Ninety-five percent of responding centers indicated that EA is used after rib fractures, but only 15% had guidelines defining the indications and contraindications. There was general agreement (>80%) regarding the indications for EA but disagreement regarding the contraindications. Contraindications were categorised based on the degree of agreement of respon- dents. The areas of greatest controversy involved minor spine injuries and minor . Conclusions: There is wide variability regarding contraindications employed for tho- racic epidural analgesia following rib fractures. These data support the need for evidenced based guidelines to define the use of EA in the multiply injured patient. © 2007 Elsevier B.V. All rights reserved.

∗ Corresponding author at: Department of Surgery, Box 359796, Harborview Medical Center, 325 9th Avenue, Seattle, WA 98104, USA. Tel.: +1 206 731 6448; fax: +1 206 731 3656. E-mail address: [email protected] (E.M. Bulger). 1366-0071/$ — see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.acpain.2007.10.019 16 E.M. Bulger et al.

1. Introduction relatively appropriate contraindication and com- monly applied (usually not ok to place epidural Rib fractures have been associated with signifi- ) and (D) inappropriate contraindication. cant morbidity and mortality following traumatic To develop a rank listing of contraindications, injury. The presence of three or more rib fractures answers A and C were pooled as general agree- has been associated with increased mortality and ment that a contraindication was appropriate and duration of ICU and hospital care [1,2]. A recent the percent of respondents in this category was analysis of the National Trauma Databank demon- listed in descending order. Indications were ranked strated a significant increase in mortality and as either appropriate or inappropriate. The indica- pulmonary morbidity for each additional rib frac- tions and contraindications included were based on ture [3]. The elderly are particularly susceptible our recent experience with a randomised controlled to complications of rib fractures with nosocomial trial of epidural analgesia in this patient popula- pneumonia rates reported as high as 31% [4]. The tion [5]. Respondents were not asked to suggest pain associated with rib fractures leads to impaired additional indications or contraindications but were ventilatory function and thus increases pulmonary asked to send their institutional guideline if avail- morbidity. Management of these patients is there- able. Review of these did not reveal any additional fore focused on achieving adequate analgesia and factors. Three mailings of the survey were sent to clearance of pulmonary secretions. Previous studies non-responders in an effort to improve the response have demonstrated that epidural analgesia provides rate, telephone calls to non-respondents were not superior pain relief, improved pulmonary function permitted. tests, reduced the risk of developing nosocomial pneumonia and led to a shorter duration of mechan- ical ventilation compared to intravenous 3. Results [5—9]. Despite these advantages only 2% of patients with rib fractures in the National Trauma Databank Responses were received from 81/188 (43%) of US received epidural analgesia [3]. This may be due Level 1 trauma centers. Seventy-seven of the cen- in large part to controversy regarding the indica- ters indicated that epidural analgesia is used at tions and contraindications for the use of epidural their institution for management of pain after rib analgesia in patients with multiple injuries. We fractures (95%). Twelve centers (15%) reported that sought to define the current practice in the United they had developed institutional guidelines regard- States by surveying pain service directors at Level ing the indications and contraindications for the 1 trauma centers regarding their current guidelines use of epidural analgesia in this patient population. for the use of epidural analgesia in this patient As illustrated in Table 1, there was general agree- population and to assess the level of agreement ment regarding the indications for the use of EA regarding indications and contraindications for its use. Table 1 Rank list based on % of respondents support- ing each indication 2. Methods Indication Support (%) A list of all Level 1 trauma centers in the United Inadequate pain control with 98 States was obtained from the American College systemic of Surgeons Committee on Trauma. A written Patient sedated by systemic 95 survey was sent to the pain service directors analgesics at all centers in September of 2004. Respon- ≥3 rib fractures 83 dents were asked whether epidural analgesia was Request from physician team 83 used for the management of pain following rib providing primary post trauma fractures and if so were there established insti- care IRB approved research study of 81 tutional guidelines regarding the indications and effectiveness of epidural contraindications for its use. They were asked analgesia for pain relief to consider seven potential indications and 33 Request from nurses providing 43 contraindications. They were asked to rank the con- primary post trauma care traindications on a four point scale as either: (A) Demand from physician team 36 appropriate contraindication, (B) relatively appro- providing primary post trauma priate contraindication but rarely applied (i.e. OK care to place epidural catheter most of the time), (C) Epidural analgesia in multiply injured patients 17

Table 2 Rank list based on % of respondents support- count <50,000, INR > 1.5, or administration ing each contraindication of low molecular weight heparin (LMWH) within the previous 8 h were appropriate contraindications to Contraindication Support (%) placement of a thoracic epidural catheter. Opinions Platelet count <50 k 93 were mixed regarding a platelet count <80,000 or Cellulitis at insertion site 92 administration of LMWH within the previous 12 h. injury 90 The majority of respondents did not believe that Epidural or spinal cord haematoma 90 INR between 1.0 and 1.5 was an appropriate con- INR > 1.5 90 traindication. LMW heparin w/in 8 h 88 Major TBI (GCS < 8) 83 Intracranial haemorrhage 81 3.2. Contraindications related to spinal Penetrating head injury 77 evaluation Haemodynamic instability 74 Unable to obtain consent 69 As many trauma patients admitted to the inten- LMW heparin w/in 12 h 69 sive care unit are still undergoing evaluation for Full spine precautions 68 potential spinal injury, we inquired whether the Platelet count <80 k 65 spine clearance status was an appropriate con- TBI (GCS 8—13) 64 Patient deeply sedated for vent 63 traindication to thoracic epidural placement. We Injury to thoracic aorta 62 included as contraindications an admission sta- Ligament disruption thoracic spine 62 tus of full spine precautions and a series of Ligament disruption cervical spine 57 questions regarding normal imaging at different Ligament disruption spine 52 spine levels but the inability to evaluate the Unstable pelvic fracture 47 patient clinically for pain that may be associ- Normal thoracic spine films, unable to 43 ated with ligamentous injury. As illustrated in examine Fig. 2, the majority of respondents felt that Transverse process fracture, thoracic 43 full spine precaution status was an appropri- spine ate contraindication, but there was disagreement Spinous process fracture, thoracic spine 40 among the other categories with most report- Normal lumbar spine films, unable to 40 examine ing these as inappropriate or relative but rarely Transverse process fracture, cervical 38 applied contraindications to epidural catheter spine placement. Spinous process fracture, cervical spine 37 Normal cervical spine films, unable to 35 3.3. Contraindications related to spine examine Respiratory failure w/o impaired 35 injury mechanics Spinous process fracture, lumbar spine 30 As illustrated in Fig. 3, we defined a series of con- Transverse process fracture, lumbar 29 traindications based on the degree of documented spine spinal injury at the cervical, thoracic, and lumbar 1.0 < INR < 1.5 24 spine levels. The majority of respondents agreed <3 rib fractures 18 that documented spinal cord injury or known spinal or epidural haematomas were appropriate con- traindications. However, there was wide variability in this setting (Table 1). Of the 33 contraindica- in the opinions regarding more minor spinal injuries tions evaluated, they were grouped based on the including ligamentous disruption and transverse or general level of agreement as shown in Table 2. spinous process fractures. In addition, specific survey results are presented for complications grouped by common classifica- 3.4. Contraindications related to associated tion. injuries

3.1. Contraindications related to As the majority of blunt traumatic injuries leading coagulopathy to rib fractures result from motor vehicle colli- sions, associated injuries are common and may As shown in Fig. 1, there were six contraindications impact the decision to place a thoracic epidural that addressed the issue of coagulopathy in trauma catheter. Fig. 4 illustrates the responses to con- patients. The majority of respondents felt that a traindications listed in this area. The majority of 18 E.M. Bulger et al.

Fig. 1 Contraindications related to coagulopathy. The distribution of responses regarding the appropriateness of contraindications to thoracic epidural placement relative to patient coagulopathy is illustrated. Response categories include: appropriate contraindication, relative contraindication commonly applied, relative contraindication uncom- monly applied, inappropriate contraindication, and no response. Abbreviations: INR, international ratio; LMW, low molecular weight. respondents agreed that significant traumatic brain 4. Discussion injury and haemodynamic instability were appro- priate contraindications. There was less agreement The use of epidural analgesia in the multiply regarding unstable pelvic fracture, injury to the injured patient with rib fractures is limited by thoracic aorta and the need for deep sedation for several potential contraindications to the place- ventilator support. ment of an epidural catheter. The risks of catheter

Fig. 2 Contraindications related to spine evaluation. The distribution of responses regarding the appropriateness of contraindications to thoracic epidural placement relative to patient level of spine evaluation is illustrated. Response categories include: appropriate contraindication, relative contraindication commonly applied, relative contraindica- tion uncommonly applied, inappropriate contraindication, and no response. Epidural analgesia in multiply injured patients 19

Fig. 3 Contraindications related to spine injury. The distribution of responses regarding the appropriateness of con- traindications to thoracic epidural placement relative to spine injury is illustrated. Response categories include: appropriate contraindication, relative contraindication commonly applied, relative contraindication uncommonly applied, inappropriate contraindication, and no response. Abbreviations: C, cervical, T, thoracic; L, lumbar; fx, fracture. placement must be weighed against the benefits infection that can lead to . Epidu- from improved pain control and the reduction in ral haematoma has been reported to occur in pulmonary morbidity reported in recent studies <1/100,000 procedures but is reported to be a [5—10]. The contraindications for thoracic epidu- greater risk in the setting of coagulopathy [11,12]. ral placement relate to the major complications In addition, there are concerns regarding the abil- that have been reported in association with its ity to assess the effectiveness of pain control or use including epidural haematoma and catheter the development of these complications in patients

Fig. 4 Contraindications related to associated injuries. The distribution of responses regarding the appropriateness of contraindications to thoracic epidural placement relative to associated injuries is illustrated. Response categories include: appropriate contraindication, relative contraindication commonly applied, relative contraindication uncom- monly applied, inappropriate contraindication, and no response. Abbreviations: TBI, traumatic brain injury; GCS, Glasgow coma score; vent, ventilator. 20 E.M. Bulger et al. with altered mental status and thus patients established for study purposes. The vast major- with associated traumatic brain injury are often ity of these were due to minor associated spine excluded. Finally, a high proportion of patients injuries. This survey was conducted to determine with rib fractures as a result of high-energy blunt the current consensus among pain service direc- trauma have associated spine injuries that may tors at Level 1 trauma centers in the United interfere with placement of the catheter or raise States regarding the indications and contraindi- concern regarding the development of neurologic cations for epidural catheter placement in these sequelae. In our previous randomised controlled patients. trial of epidural analgesia vs. intravenous opi- This report demonstrates general agreement oids we found that 69% of eligible patients met regarding the common indications for thoracic an exclusion criteria due to contraindications to epidural placement but significant disagreement placement of the epidural catheter that had been regarding a number of the contraindications listed.

Fig. 5 Local guideline for assessment and management of patients with multiple rib fractures. Epidural analgesia in multiply injured patients 21

A similar survey was recently reported regard- trauma centers and thus may not be representative ing the use of thoracic epidural analgesia in the of the practice in the general community. Level 1 United Kingdom [13]. While this study did not centers, however,generally treat the greatest num- focus on its use in multiply injured patients, it ber of severely injured patients and thus the fact did report little consensus regarding the degree that only 15% of respondents reported the existence of coagulopathy that would be tolerated prior of guidelines regarding the use of epidural analge- to catheter placement. This included controversy sia in multiply injured patients suggests that this regarding the timing of catheter placement relative has not been widely recognised as a practice issue. to a prophylactic dose of LMWH and disagreement We believe that our data point to the need for the over the appropriate INR and platelet count for development of evidence based guidelines for the catheter placement. A subsequent survey evalu- use of thoracic epidural analgesia for the manage- ated the use of epidural analgesia in an intensive ment of patients with rib fractures in the setting of care setting in England [14]. In this survey, the multiple injuries. presence of subclinical coagulopathy was consid- ered a contraindication to epidural placement by only 49% of the intensive care units surveyed. Conflict of interest statement A consensus document has been recently pro- duced regarding the use of regional anesthesia n There is no conflict of interest. anticoagulated patients, but this does not specifi- cally address patients with coagulopathy following injury [11] These reports coupled with our data highlight the current lack of consensus regard- References ing the appropriate contraindications for epidural [1] Lee RB, Morris Jr JA, Parker RS. Presence of three or more catheter placement among the critically ill or rib fractures as an indicator of need for interhospital trans- injured. fer. J Trauma 1989;29(6):795—9, discussion 799—800. It is our hope that this survey will stimulate the [2] Lee RB, Bass SM, Morris Jr JA, MacKenzie EJ. Three specialty societies to develop an evidence-based or more rib fractures as an indicator for transfer to a consensus guideline for the use of thoracic epidu- Level I trauma center: a population-based study. J Trauma 1990;30(6):689—94. ral analgesia for management of pain following [3] Flagel BT, Luchette FA, Reed RL, Esposito TJ, Davis in the multiply injured patient. We KA, Santaniello JM, et al. Half-a-dozen ribs: the break- have developed a local guideline for the providers point for mortality. Surgery 2005;138(4):717—23, discussion in our trauma center that is provided as Fig. 5, 723—715. but recognise that this has not been subject to a [4] Bulger EM, Arneson MA, Mock CN, Jurkovich GJ. Rib fractures in the elderly. J Trauma 2000;48(6):1040—6, dis- national consensus process. As shown, we consider cussion 1046—1047. all patients with ≥3 rib fractures as candidates for [5] Bulger EM, Edwards T, Klotz P, Jurkovich GJ. Epidural evaluation by our Acute Pain Relief Service (APS). analgesia improves outcome after multiple rib fractures. This service consists of anesthesiologist and mid- Surgery 2004;136(2):426—30. level practitioners specialising in the management [6] Mackersie RC, Karagianes TG, Hoyt DB, Davis JW. Prospec- tive evaluation of epidural and intravenous administration of acute pain and available 24/7 for consulta- of for pain control and restoration of ventila- tion and complex pain management. Patients are tory function following multiple rib fractures. J Trauma considered for ICU admission if they have any 1991;31(4):443—9, discussion 449—451. respiratory compromise, significant pulmonary co- [7] Moon MR, Luchette FA, Gibson SW, Crews J, Sudarshan morbidity or are >65 years of age. All patients G, Hurst JM, et al. Prospective, randomized comparison of epidural versus parenteral analgesia in thoracic receive respiratory therapy and initial pain manage- trauma. Ann Surg 1999;229(5):684—91, discussion 691—682. ment with intravenous opioids pending assessment [8] Wu CL, Jani ND, Perkins FM, Barquist E. Thoracic epidural for epidural analgesia. Our absolute and relative analgesia versus intravenous patient-controlled analgesia contraindications are noted in Fig. 5. Importantly, for the treatment of rib fracture pain after motor vehicle all patients with multiple rib fractures have their crash. J Trauma 1999;47(3):564—7. [9] Wisner DH. A stepwise logistic regression analysis of factors pain managed by APS regardless of whether they affecting morbidity and mortality after thoracic trauma: are candidates for epidural analgesia. effect of epidural analgesia. J Trauma 1990;30(7):799—804, The primary limitation to this survey is the discussion 804—795. response rate of 43%. There may be a selection [10] Block BM, Liu SS, Rowlingson AJ, Cowan AR, Cowan Jr JA, bias introduced by those who chose to respond to Wu CL. Efficacy of postoperative epidural analgesia: a meta- analysis. JAMA 2003;290(18):2455—63. our survey. Despite three mailings of the survey to [11] Horlocker TT, Wedel DJ, Benzon H, Brown DL, Enneking FK, non-responders, we were unable to improve this Heit JA, et al. Regional anesthesia in the anticoagulated response rate. The survey only included Level 1 patient: defining the risks (the second ASRA consensus con- 22 E.M. Bulger et al.

ference on neuraxial anesthesia and anticoagulation). Reg [13] O’Higgins F, Tuckey JP. Thoracic epidural anaesthesia and Anesth Pain Med 2003;28(3):172—97. analgesia: United Kingdom practice. Acta Anaesthesiol [12] Stenseth R, Breivik H, Grimstad J. Regional anaesthe- Scand 2000;44(9):1087—92. sia/analgesia in patients with bleeding disturbances or [14] Low JH. Survey of epidural analgesia management in gen- thromboprophylaxis. Norweigian Association of Anaesthe- eral intensive care units in England. Acta Anaesthesiol siologists NA Forum 1994;7:23—4. Scand 2002;46(7):799—805.

Available online at www.sciencedirect.com A comparison of regional and for total replacement of the hip or knee A META-ANALYSIS

S. Hu, We performed a meta-analysis to evaluate the relative efficacy of regional and general Z.-Y. Zhang, anaesthesia in patients undergoing total hip or knee replacement. A comprehensive search Y.-Q. Hua, for relevant studies was performed in PubMed (1966 to April 2008), EMBASE (1969 to April J. Li, 2008) and the Cochrane Library. Only randomised studies comparing regional and general Z.-D. Cai anaesthesia for total hip or knee replacement were included. We identified 21 independent, randomised clinical trials. A random-effects model was From the Second used to calculate all effect sizes. Pooled results from these trials showed that regional Military Medical anaesthesia reduces the operating time (odds ratio (OR) -0.19; 95% confidence interval (CI) University, Shanghai, -0.33 to -0.05), the need for transfusion (OR 0.45; 95% CI 0.22 to 0.94) and the incidence of People’s Republic of thromboembolic disease (deep-vein thrombosis OR 0.45, 95% CI 0.24 to 0.84; pulmonary China embolism OR 0.46, 95% CI 0.29 to 0.80). Regional anaesthesia therefore seems to improve the outcome of patients undergoing total hip or knee replacement.

In 2005, more than 375 000 patients underwent hypotension, motor blockade, urinary reten- total hip replacement (THR) and more than 530 tion and pruritus.16 Although there have been 000 had total knee replacement (TKR) in the many refinements to reduce such complica- United States.1 Despite the frequency with tions, there is still the potential for inadvertent which these procedures are carried out, there dural puncture and neurological injury which remains controversy as to whether they are best may make these techniques less acceptable.11 performed under regional anaesthesia using epi- In addition, many of these studies have been dural or spinal , or general compromised by small numbers of patients anaesthesia. In 2000, Rodgers et al2 published a and, in some cases, the relatively rare occur- meta-analysis which addressed the clinical mer- rence of complications. Consequently, it has its of neuraxial blockade for a variety of surgi- been difficult to draw any conclusions about cal procedures. However, its place in patients the effect of the choice of anaesthesia on the „ S. Hu, MD, Orthopaedic undergoing joint replacement remains uncer- outcome of joint replacement. „ Z.-Y. Zhang, MD, Professor tain. We carried out a meta-analysis to evaluate „ Y.-Q. Hua, MD, Orthopaedic Surgeon Regional anaesthesia in total joint replace- the relative efficacy of regional and general „ J. Li, MD, Orthopaedic ment is claimed to decrease the incidence of anaesthesia in patients undergoing joint Surgeon „ Z.-D. Cai, MD, Professor deep-vein thrombosis (DVT) and pulmonary replacement. We concentrated on THR and Department of Orthopaedics embolism and to reduce intraoperative bleed- TKR and restricted the regional techniques to Changhai Hospital, Second Military Medical University, ing, the need for transfusion and the length of epidural or spinal approaches, to limit the 3-9 168 Changhai Road, Shanghai hospital stay. It can also increase patient sat- number of confounding variables. 200433, People’s Republic of China. isfaction especially after one-stage bilateral THR or TKR.10 Epidural anaesthesia com- Materials and Methods Correspondence should be sent to Professor Z.-D. Cai; e-mail: bined with post-operative epidural analgesia A search of the literature was undertaken using [email protected] can reduce the physiological stress of surgery PubMed (1966 to April 2008), EMBASE (1969 ©2009 British Editorial Society and the incidence of complications as well as to April 2008) and the Cochrane Library data- of Bone and Joint Surgery improving the overall outcome of surgery.11 bases using the following keywords: hip doi:10.1302/0301-620X.91B7. 21538 $2.00 However, other studies have reported contra- replacement, hip arthroplasty, knee replace- dictory results.12-14 Moiniche et al15 have ment, knee arthroplasty, regional anaesthesia, J Bone Joint Surg [Br] 2009;91-B:935-42. shown that there may not be any advantage in epidural anaesthesia, and Received 16 July 2008; regional techniques. Moreover, spinal and epi- general anaesthesia. The terms regional, Accepted after revision 12 March 2009 dural anaesthesia and analgesia may cause epidural, spinal and general anaesthesia were

VOL. 91-B, No. 7, JULY 2009 935 936 S. HU, Z.-Y. ZHANG, Y.-Q. HUA, J. LI, Z.-D. CAI

Ta b l e I . Characteristics of the randomised controlled trials of regional versus general anaesthesia in total joint replacement of the lower limb contributing data to our meta-analysis

Source Joint studied* Regional anaesthesia technique Outcomes studied† Modig et al3 THR Continuous epidural Duration of surgery, blood loss, DVT, PE Modig et al4 THR Continuous epidural Duration of surgery, blood loss, DVT, PE Modig et al5 THR Continuous epidural Duration of surgery, blood loss, DVT, PE Davis et al6‡ THR Spinal Duration of surgery, blood loss, number of patients transfused, DVT, PE Jørgensen et al8 TKR Continuous epidural DVT, PE Nielsen et al9 TKR Continuous epidural DVT Brueckner et al12 THR Spinal Duration of surgery, blood loss Planes et al13 THR Spinal Duration of surgery, blood loss, DVT Mitchell et al14 TKR Continuous epidural DVT Borghi et al21‡ THR Continuous epidural Duration of surgery, blood loss, number of patients transfused Gonano et al24 THR or TKR Spinal Duration of surgery Lattermann et al25 THR Combined spinal and epidural Duration of surgery, blood loss Kita et al26 THR Continuous epidural Duration of surgery, blood loss, length of hospital stay Kudoh et al27 TKR Spinal Duration of surgery, blood loss, PONV Modig and Karlstrom28 THR Continuous epidural Duration of surgery, blood loss Wulf et al29 THR Continuous epidural Duration of surgery, length of hospital stay Donatelli et al30 THR or TKR Continuous epidural Duration of surgery, blood loss, number of patients transfused Hole et al31 THR Single- epidural Duration of surgery, PE Jones et al32 THR or TKR Spinal Duration of surgery, number of patients trans- fused, length of hospital stay, DVT, PE mortality Williams-Russo et al33 TKR Continuous epidural Duration of surgery, blood loss, DVT, PE Keith34 THR Single-injection epidural Blood loss, number of patients transfused * THR, total hip replacement; TKR, total knee replacement † DVT, deep-vein thrombosis; PE, pulmonary embolism; PONV, post-operative nausea and vomiting ‡ the data of the two publications of Davis et al (6, 20) were from the same group as were those presented by Borghi et al (21,22) in their two publications

linked with ‘or’ and combined using ‘and’ with each subse- Lastly, we asked authors if they knew of any other relevant quent term. Each publication was independently reviewed published studies. by two authors (SH, Z-DC) and the data collected entered Statistical analysis. We only included those outcome onto a standard sheet. We limited the search to articles pub- parameters which were presented in numerical format. lished in English. Only randomised, controlled trials which Continuous outcome parameters were expressed as the compared the outcome of elective THR and TKR carried mean and SD and dichotomous outcomes as the number of out under regional anaesthesia with those carried out under events. The level of significance for all tests was set at a general anaesthesia were included. We included all of these two-sided p-value of 0.05 and variances were not assumed despite the fact that some did not address all the aspects to be equal. Patients who had a were that we wished to study. Nonetheless, we read all the orig- treated as a control group, and those with regional anaes- inal articles carefully and extracted relevant data in order to thesia as an intervention group. The odds ratios (OR) and decrease publication bias. We also checked the references of the 95% confidence interval (CI) were calculated for each article for other studies which met all our inclusion dichotomous outcomes. The standardised mean difference criteria. We extracted the following outcome data from and the 95% CI were presented for continuous outcomes. each study if reported: the operating time, the intraopera- Heterogeneity among studies was tested using the chi- tive blood loss, the number of patients requiring blood squared test. We used a random-effects model to calculate transfusion, the number of patients with DVT or pulmo- all effect sizes since it was more conservative and included nary embolism, diagnosed radiologically or clinically, the both the random variation within the studies and the vari- length of hospital stay and any deaths. ation among the different studies. Moreover, the pooled The data collection was not blinded. A third reviewer estimates calculated by the random-effects and fixed-effects (Z-YZ) compared the two sets of data collection sheets and models were similar when there was minimal heterogeneity any differences were resolved by discussion and reconfirm- between studies. Whenever possible, subgroup analyses ing the data from the original paper. We attempted to con- were performed to detect and evaluate clinically significant tact the authors of a trial if they had published more than differences. All statistical analyses were performed using one report on the subject to confirm that the data in each of the freeware program Review Manager 4.2 (Cochrane their publications were from different groups of patients. Collaboration, Oxford, United Kingdom).

THE JOURNAL OF BONE AND JOINT SURGERY A COMPARISON OF REGIONAL AND GENERAL ANAESTHESIA FOR TOTAL REPLACEMENT OF THE HIP OR KNEE 937

Ta b l e I I . Results of the meta-analysis for all the outcomes of total joint replacement

Heterogeneity

Outcome* Number of studies SMD or OR (95% CI)† Z-value p-value Chi-squared test p-value I2(%)‡

Duration of surgery 17 SMD -0.12 (-0.23 to -0.02) 2.36 0.02 13.99 0.60 0.0 Intra-operative blood loss 12 SMD -0.50 (-0.96 to -0.04) 2.12 0.03 107.04 < 0.0001 91.6 Length of hospital stay 3 SMD -0.55 (-2.47 to +1.37) 0.57 0.57 4.31 0.12 53.6 Number of transfusions 5 OR 0.45 (0.22 to 0.94) 2.14 0.03 8.94 0.06 55.3 DVT 10 OR 0.45 (0.24 to 0.84) 2.48 0.01 28.88 0.0007 68.8 PE 8 OR 0.46 (0.21 to 1.02) 1.9 0.06 11.85 0.11 10.9 PONV 2 OR 0.27 (0.11 to 0.64) 2.96 0.003 0.19 0.66 0.0 Mortality 2 OR 0.94 (0.14 to 6.52) 0.06 0.95 0.83 0.36 0.0 * DVT, deep-vein thrombosis; PE, pulmonary embolism; PONV, post-operative nausea and vomiting † SMD, standardised mean difference; OR, odds ratio; 95% CI, 95% confidence interval ‡ inconsistency value, this represents the extent of the heterogeneity

Table III. Subgroup analysis of some of the outcome assessments

Heterogeneity

Outcome* Subgroup† Number of studies SMD or OR (95% CI)‡ Z-value p-value Chi-squared test p-value I2(%)§

Duration of surgery THR 12 SMD -0.19 (-0.33 to -0.05) 2.69 0.007 9.57 0.57 0.0 TKR 2 SMD -0.02 (-0.21 to +0.17)0.21 0.84 0.92 0.34 0.0

Intra-operative blood loss THR 10 SMD -0.65 (-1.24 to -0.06) 2.16 0.03 103.24 < 0.0001 91.3 THR¶ 8 SMD -0.48 (-1.13 to 0.17) 1.44 0.15 82.8 < 0.0001 91.6

Number of transfusions THR 4 OR 0.35 (0.18 to 0.67) 3.19 0.001 4.16 0.25 27.8 TKR None

DVT No TED drug 7 OR 0.27 (0.14 to 0.52) 3.88 0.0001 11.7 0.07 48.7 prophylaxis TED drug prophylaxis 3 OR 1.07 (0.58 to 1.97) 0.21 0.83 2.78 0.25 28.1

PE No TED drug 6 OR 0.26 (0.13 to 0.52) 3.87 0.0001 2.2 0.82 0.0 prophylaxis TED drug prophylaxis 2 OR 1.66 (0.61 to 4.52) 1.0 0.32 0.70 0.40 0.0 * DVT, deep-vein thrombosis; PE, pulmonary embolism † THR, total hip replacement; TKR, total knee replacement; TED, thromboembolic disease ‡ SMD, standardised mean difference; OR, odds ratio; 95% CI, 95% confidence interval § inconsistency value, this represents the extent of the heterogeneity ¶ two trials in which patients in the regional anaesthesia group had a lower blood pressure than those in the general anaesthesia group were excluded

Results operating time, intraoperative bleeding, rate of transfusion, We identified 25 randomised, controlled trials in the Eng- and the incidence of DVT and pulmonary embolism. lish language in which patients had been randomised to Whenever possible, we also performed several subgroup receive either regional or general anaesthesia. Of these, one analyses on the hips and knees independently and then in which the data were reported as medians and ranges was combined them to provide an overall estimate. The results excluded.17 Another three were also excluded. One did not of the meta-analysis are shown in Tables II and III. present data in a form which could be included in the Operating time. Data on the operating time were found in review18 and the other two included confounding factors, 17 studies with a total of 1483 patients.3-6,12,13,21,24-33 namely peri-operative hypotension and hip trauma.19,20 There was a decrease in the length of the operating time Two studies generated more than one publication each, but with regional compared with general anaesthesia (standard were counted only once.6,21-23 Finally, 21 studies were mean difference -0.12; 95% CI -0.23 to -0.021; Fig. 1). In included in the review3-6,8,9,12-14,21,24-34 (Table I). We the subgroup meta-analysis, the operating time for THR assessed the presence of publication bias by examining fun- under regional anaesthesia was significantly shorter than nel plots which displayed the studies included in the meta- that under general anaesthesia (OR -0.19; 95% CI -0.33 to analysis in a plot of effect size against sample size. Fortu- -0.05), but there was no significant difference between the nately, they showed no evidence of publication bias for the two for TKR (OR -0.02; 95% CI -0.21 to 0.17; Table III).

VOL. 91-B, No. 7, JULY 2009 938 S. HU, Z.-Y. ZHANG, Y.-Q. HUA, J. LI, Z.-D. CAI

Study Number of Regional anaesthesia Number of General anaesthesia Standardised mean difference Weight Standardised mean difference or sub-category patients Mean (SD) patients Mean (SD) (random) 95% CI (%) (random) (95% CI)

Hole et al31 29 190.00 (32.00) 31 207.00 (33.00) 3.94 -0.52 (-1.03 to 0.00) Modig et al3 15 147.00 (28.00) 15 161.00 (35.00) 1.99 -0 .43 (-1.15 to 0.30) Modig et al4 30 146.00 (22.00) 30 156.00 (26.00) 4.00 -0.41 (-0.92 to 0.10) Modig et al5 48 152.00 (25.00) 46 150.00 (26.00) 6.39 0.08 (-0.33 to 0.48) Modig and Karlstrom28 14 132.00 (23.00) 10 136.00 (30.00) 1.58 -0.15 (-0.96 to 0.66) Davis et al6 69 73.00 (13.00) 71 79.00 (21.00) 9.39 -0.34 (-0.67 to -0.01) Jones et al32 74 107.00 (24.00) 72 112.00 (28.00) 9.90 -0.19 (-0.52 to 0.13) Planès et al13 61 73.30 (2.30) 62 73.70 (2.40) 8.35 -0.17 (-0.52 to 0.19) Williams-Russo et al33 134 85.00 (33.00) 128 88.00 (32.00) 17.82 -0.09 (-0.33 to 0.15) Wulf et al29 43 145.00 (45.00) 45 136.00 (29.00) 5.95 0.24 (-0.18 to 0.66) Brueckner et al12 16 101.00 (23.00) 10 113.00 (39.00) 1.64 -0.39 (-1.19 to 0.41) Kudoh et al27 75 106.70 (31.50) 75 104.20 (11.80) 10.20 0.10 (-0.22 to 0.42) Borghi et al22 70 106.00 (29.00) 70 112.00 (41.00) 9.50 -0.17 (-0.50 to 0.16) Lattermann et al25 7 95.00 (22.00) 8 99.00 (26.00) 1.01 -0.16 (-1.17 to 0.86) Gonano et al24 20 109.00 (26.00) 20 111.00 (36.00) 2.72 -0.06 (-0.68 to 0.56) Donatelli et al30 30 175.00 (38.00) 30 168.00 (40.00) 4.07 0.18 (-0.33 to 0.68) Kita et al26 16 71.00 (34.00) 9 84.00 (34.00) 1.54 -0.37 (-1.19 to 0.45)

Total (95% Cl) 751 732 100.00 -0.12 (-0.23 to -0.02) Test for heterogeneity: chi-squared = 13.99, df = 16 (p = 0.60), l2 = 0% Test for overall effect: Z = 2.36 (p = 0.02)

-1.0 -0.5 0.0 0.5 1.0 Favours regional anaesthesia Favours general anaesthesia

Fig. 1

Chart showing comparison of the duration of surgery (mean, SD) between the regional and general anaesthesia groups for patients undergoing total joint replacement.

Study Number of Regional anaesthesia Number of General anaesthesia Standardised mean difference Weight Standardised mean difference or sub-category patients Mean (SD) patients Mean (SD) (random) 95% CI (%) (random) (95% CI)

Keith 34 10 342.00 (59.00) 9 648.00 (58.00) 6.24 -4.99 (-6.99 to -3.00) Modig et al5 48 1210.00 (490.00) 46 1680.00 (460.00) 14.32 -0.98 (-1.41 to -0.55) Modig and Karlstrom28 14 950.00 (300.00) 10 1140.00 (200.00) 12.17 -0.70 (-1.54 to 0.14) Davis et al6 69 512.00 (266.00) 71 706.00 (354.00) 14.66 -0.61 (-0.95 to -0.28) Planès et al13 61 523.00 (29.00) 62 508.00 (23.00) 14.58 0.57 (0.21 to 0.93) Borghi et al22 70 547.00 (99.00) 70 479.00 (107.00) 14.66 0.66 (0.32 to 100) Lattermann et al25 7 629.00 (525.00) 8 563.00 (306.00) 11. 11 0.15 (-0.87 to 1.16) Kita et al26 16 335.00 (381.00) 9 485.00 (383.00) 12.25 -0.38 (-1.20 to 0.44)

Subtotal (95% Cl) 295 285 100. 00 -0.48 (-1.13 to 0.17) Test for heterogeneity: chi-squared = 13.99, df = 7 (p < 0.00001), l2 = 91.6% Test for overall effect: Z = 1 .44 (p = 0.15)

-4 -2 0 2 4 Favours regional anaesthesia Favours general anaesthesia

Fig. 2

Chart showing comparison of intraoperative blood loss (mean, SD) between the regional and general anaesthesia groups for patients undergoing total joint replacement.

Intra-operative blood loss. Intraoperative blood loss However, when we excluded two trials in which patients in was reported in 12 studies which included 880 the regional anaesthesia group had a significantly lower patients.3-6,13,21,25-28,30,34 In five there was no statistical blood pressure than those in the general anaesthesia difference in the intraoperative blood loss between group,3,4 there was no significant difference between the regional and general anaesthesia. In another five there groups (Fig. 2). was reduced blood loss with regional anaesthesia. The Transfusion requirement. Five studies included data on the remaining two showed increased blood loss. These dif- number of patients transfused intraoperatively.6,21,30,32,34 ferences were statistically significant (standardised mean Meta-analysis showed that regional anaesthesia reduced difference -0.50, 95% CI -.96 to -0.04; Table II). the need for transfusion (OR 0.45; 95% CI 0.22 to 0.94; The pooled data from the ten studies on THR showed Table II). Subset modelling of the THR group showed that a statistically significant decrease in blood loss in regional anaesthesia reduced the need for transfusion (OR patients with regional anaesthesia (standardised mean 0.35; 95% CI 0.18 to 0.67; Table III). None of the studies difference -0.65; 95% CI -1.24 to -0.006; Table III). presented comparable data for the TKR group and

THE JOURNAL OF BONE AND JOINT SURGERY A COMPARISON OF REGIONAL AND GENERAL ANAESTHESIA FOR TOTAL REPLACEMENT OF THE HIP OR KNEE 939

Study Regional anaesthesia General anaesthesia Odds ratio (random) Weight (%) Odds ratio (random) or sub-category (n/N) (n/N) (95% CI)

No antithrombotic prophylaxis Modig et al3 5/15 11/15 7. 9 4 0.18 (0.0.4 to 0.87) Modig et al4 12/30 23/30 10.47 0.20 (0.03 to 0.62) Modig et al5 21/48 38/46 11.51 0.16 (0.06 to 0.42) Davis et al6 9/69 19/71 12.00 0.41 (0.17 to 0.99) Jones et al32 6/74 3/72 8.68 2.03 (0.49 to 8.44) Nielsen et al9 2/13 10/16 6.18 0.11 (0.02 to 0.67) Jørgensen et al8 3/17 13/22 8.24 0.15 (0.03 to 0.67)

Subtotal (95% CI) 266 272 65.66 0.27 (0.14, 0.521) Total events: 58 (regional anaesthesia), 117 (general anaesthesia) Test for heterogeneity: chi-squared = 11.70, df = 6 (p = 0.07), I2 = 48.7% Test for overall effect: Z = 3.88 (p = 0.0001)

TED prophylaxis Mitchell et al14 12/34 10/38 11.16 1.53 (0.56 to 4.18) Planés et al13 7/60 4/62 9.48 1.92 (0.53 to 6.91) Williams-Russo et al33 39/97 39/81 13.70 0.72 (0.40 to 1.31) Subtotal (95% CI) 191 181 34.34 1.07 (0.58 to 1.97) Total events: 58 (regional anaesthesia), 53 (General anaesthesia) Test for heterogeneity: chi-squared = 2.78, df = 2 (p = 0.25), I2 = 28.1% Test for overall effect: Z = 0.21 (p = 0.83)

Total (95% CI) 457 453 100.00 0.45 (0.24 to 0.84) Total events: 116 (regional anaesthesia), 170 (general anaesthesia) Test for heterogeneity: chi-squared = 28.88, df = 9 (p = 0.0007) I2 = 68.8% Test for overall effect: Z = 2.48 (p = 0.001)

0.1 0.2 0.5 1.0 2.0 5.0 10.0 Favours regional anaesthesia Favours general anaesthesia

Fig. 3

Chart showing comparison of the number of patients with deep-vein thrombosis in the regional and general anaesthesia groups with and without the administration of thromboembolic disease prophylaxis.

therefore the effect of regional anaesthesia on the need for Other outcome measurements. Data on mortality were transfusion in TKR could not be determined. presented in only two studies13,32 (OR 0.94; 95% CI 0.14 Thromboembolic disease. Ten studies presented data on the to 6.52; Table II). There was no obvious difference between incidence of DVT3-6,8,9,13,14,32,33 and eight on pulmonary regional and general anaesthesia. Meta-analysis was not embolism.3-6,8,31-33 These two meta-analyses showed that performed because of insufficient data. regional anaesthesia significantly lowered the incidence of Three studies recorded the length of hospital stay.29,32,33 thromboembolic disease when compared with general There was no discernible difference between the groups anaesthesia (DVT 95% CI 0.24 to 0.84; Fig. 3). Although a (standardised mean difference -0.55; 95% CI -2.47 to random-effects model did not show the difference +1.37; Table II). (pulmonary embolism 95% CI 0.21 to 1.02; Table II), it was identified using a fixed-effects model (pulmonary Discussion embolism 95% CI 0.29 to 0.80; Fig. 4). Our meta-analysis has shown several potential advantages of However, when subgroup analysis was performed of the regional over general anaesthesia. The former can be benefi- studies in which patients had been given anticoagu- cial in reducing the duration of surgery, the need for trans- lants,13,14,31,33 there was no significant difference in the fusion, post-operative nausea and vomiting and the incidence of thromboembolic disease between the two incidence of thromboembolic disease (DVT and pulmonary groups (DVT 95% CI 0.58 to 1.97; Fig. 3; pulmonary embolism) in patients undergoing total joint replacement, embolism 95% CI 0.61 to 4.52; Table III). particularly THR. There was insufficient evidence to support Post-operative nausea and vomiting. Data on the number or refute the use of regional anaesthesia in decreasing intra- of patients who suffered from this were only given in two operative blood loss, mortality or length of hospital stay. studies24,27 (OR 0.27; 95% CI 0.11 to 0.64; Table II). One Duration of surgery refers to the operating time and not noted that regional anaesthesia reduced its incidence,17 but to the duration of anaesthesia or recovery. Our meta- pooled data showed significant difference between the two analysis has shown that regional anaesthesia had only a groups. minimal effect in reducing the operating time. Sub-group

VOL. 91-B, No. 7, JULY 2009 940 S. HU, Z.-Y. ZHANG, Y.-Q. HUA, J. LI, Z.-D. CAI

Study Regional anaesthesia General anaesthesia Odds ratio (fixed) Weight Odds ratio (fixed) or sub-category (n/N) (n/N) (95% CI) (%) (95% CI)

Hole et al31 2/29 0/31 1.03 5.73 (0.26 to 124.51) Modig et al3 4/15 10/15 17.04 0.18 (0.04 to 0.87) Modig et al4 3/30 10/30 20.91 0.22 (0.05 to 0.91) Modig et al5 5/48 15/46 31.88 0.24 (0.08 to 0.73) Davis et al6 0/69 3/71 7. 9 6 0.14 (0.01 to 2.78) Jones et al32 2/74 2/72 4.58 0.97 (0.13 to 7.09) Jørgensen et al8 0/17 1/22 2.97 0.41 (0.02 to 10.69) Williamson-Russo et al33 10/97 6/81 13.63 1.44 (0.50 to 4.14)

Subtotal (95% CI) 379 368 100.00 0.48 (0.29 to 0.80) Total events: 26 ( regional anaesthesia), 47 (general anaesthesia) Test for heterogeneity: chi-squared = 11.85, df = 7 (p = 0.11), I2 = 40.9%) Test for overall effect: Z = 2.84 (p = 0.005)

0.1 0.2 0.5 1.0 2.0 5.0 10.0 Favours regional anaesthesia Favours general anaesthesia

Fig. 4

Chart showing comparison of the number of patients with pulmonary embolism in the regional and general anaesthesia groups.

analysis indicated that regional anaesthesia significantly between the groups as to the need for transfusion. We reduced the operating time for THR, a result which was believe that the reason for this may be that patients having consistent with that of other recent studies.35 However, this regional anaesthesia for a THR are given a plasma effect was not seen in patients undergoing TKR. The differ- expander more often than those who have general anaes- ence may relate to the use of a thigh tourniquet during thesia. Consequently, while general anaesthesia for THR TKR. The tourniquet gives an almost bloodless operating would appear to increase the need for , field which, in turn, may speed up the operation. patients who have regional anaesthesia either for THR or The pooled data showed a statistically significant TKR require significantly greater amounts of colloid to decrease in blood loss in the regional anaesthesia group. maintain their intravascular volume in order to compensate However, some of the studies were performed 20 years ago for sympathetic vasodilatation, thereby apparently reduc- and patients in the regional anaesthesia groups had a signif- ing the need for blood transfusions. icantly lower blood pressure.3,4 A subgroup analysis of the The prevention of thromboembolic disease is of concern eight THR studies, showed that there was no significant to all orthopaedic surgeons. Our meta-analysis showed a difference between the groups when this confounding fac- protective effect of regional anaesthesia against throm- tor was excluded. This is at odds with recent research by boembolic disease (DVT and pulmonary embolism) in Mauermann et al.35 The explanation for this discrepancy patients undergoing total joint replacement. A similar effect may be that the latter included studies in which patients in has been demonstrated by others.6,9 This finding was also the regional anaesthesia group had significant perioperative consistent with the results published by Rodgers et al2 hypotension. Therefore we think the result from our sub- which included over 9500 patients. There are many pos- group analysis is more appropriate and conclusive. In the sible reasons for this effect including altered coagulability, only study for which valid data about intra-operative blood increased volume flow to the lower limbs, an improved loss could be extracted there was no significant difference ability to breathe without pain and a reduction in the sur- between the two groups.27 They did not present the data in gical stress responses.4,37 However, most of the studies a form which could be included in the meta-analysis. which favour regional anaesthesia were carried out on Several other studies did not find any discernible difference patients who did not have heparin prophylaxis.10 These between the two groups.8,18,36 This is easily explained. findings must be viewed cautiously. A previous review38 Bleeding after TKR, which is performed under tourniquet, showed that regional anaesthesia was inadequate as sole occurs after the period of the anaesthetic effect, into surgi- prophylaxis, although it may reduce the incidence of cal drains. Although we cannot refute the possibility that thromboembolic disease by 50%.39 However, a recent regional anaesthesia reduces blood loss in TKR, we contend review showed that epidural anaesthesia without anticoag- that, if there is such an effect, it is probably very small. Data ulants resulted in a low incidence of venous thromboembo- on the need for transfusion were reported in five stud- lism in patients undergoing THR.40 We suggest that on this ies.6,21,30,32,34 We found that the use of regional anaesthesia basis regional anaesthesia, when used in conjunction with a reduced the risk of blood transfusion. The overall blood prophylactic , will be more beneficial in loss was not significantly different between the two meth- decreasing the rate of thromboembolism, especially in ods of anaesthesia, but significant differences were found those patients with risk factors such as varicose veins,

THE JOURNAL OF BONE AND JOINT SURGERY A COMPARISON OF REGIONAL AND GENERAL ANAESTHESIA FOR TOTAL REPLACEMENT OF THE HIP OR KNEE 941 malignancy and . old age and the use of oral 6. Davis FM, Laurenson VG, Gillespie WJ, et al. Deep vein thrombosis after total oestrogens.6 hip replacement: a comparison between spinal and general anaesthesia. J Bone Joint Surg [Br] 1989;71-B:181-5. Our data have shown that the incidence of post- 7. Sharrock NE, Haas SB, Hargett MJ, et al. Effects of epidural anesthesia on the operative nausea and vomiting was significantly lower in incidence of deep-vein thrombosis after total knee arthroplasty. J Bone Joint Surg the regional group than in the general anaesthesia group. [Am] 1991;73-A:502-6. 8. Jørgensen LN, Rasmussen LS, Nielsen PT, Leffers A, Albrecht-Beste E. Anti- Some authors have suggested that this is because regional thrombotic efficacy of continuous extradural analgesia after knee replacement. Br J anaesthesia provides significantly better post-operative Anaesth 1991;66:8-12. analgesia,41 which leads to a lower consumption of opioids. 9. Nielsen PT, Jørgensen LN, Albrecht-Beste E, Leffers AM, Rasmussen LS. Lower thrombosis risk with epidural blockade in knee arthroplasty. Acta Orthop Scand However, the sample size was small and the impact of 1990;61:29-31. regional anaesthesia on post-operative nausea and vomit- 10. Schäfer M, Elke R, Young JR, Gancs P, Kindler CH. Safety of one-stage bilateral ing was inconclusive. hip and knee arthroplasties under regional anaesthesia and routine anaesthetic mon- In our meta-analysis, we failed to confirm that regional itoring. J Bone Joint Surg [Br] 2005;87-B:1134-9. 11. Moraca RJ, Sheldon DG, Thirlby RC. The role of epidural anesthesia and analge- anaesthesia was associated with a reduced length of stay in sia in surgical practice. Ann Surg 2003;238:663-73. hospital or reduced mortality. 12. Brueckner S, Reinke U, Roth-Isigkeit A, et al. Comparison of general and spinal Despite the relatively large number of patients studied, anesthesia and their influence on hemostatic markers in patients undergoing total hip arthroplasty. J Clin Anesth 2003;15:433-40. there were several limitations to our study. First, we limited 13. Planès A, Vochelle N, Fagola N, Feret J, Bellaud M. Prevention of deep vein our meta-analysis to articles in English. Although the effect thrombosis after total hip replacement: the effect of low-molecular-weight heparin of excluding non-English trials on the results of a meta- with spinal and general anaesthesia. J Bone Joint Surg [Br] 1991;73-B:418-22. analysis is unclear, exclusion of such trials may have little 14. Mitchell D, Friedman RJ, Baker JD 3rd. Prevention of thromboembolic disease following total knee arthroplasty: epidural versus general anesthesia. Clin Orthop effect on the summary effects of treatment and may actually 1991;269:109-12. give a more conservative estimate of the effect of treat- 15. Moiniche S, Hjortso NC, Hansen BL, et al. The effect of balanced analgesia on ment.42 Secondly, although the data were weighted by trial early convalescence after major orthopaedic surgery. Acta Anaesthesiol Scand 1994;38:328-35. size, they were not weighted by the quality of the ran- 16. Gedney JA, Liu EH. Side-effects of epidural infusions of opioid mix- domised controlled trials included nor were they assessed in tures. Anaesthesia 1998;53:1148-55. a blinded fashion. We believe that the quality of the trials 17. Chu CP, Yap JC, Chen PP, Hung HH. Postoperative outcome in chinese patients having primary total knee arthroplasty under general anaesthesia/intravenous was generally similar, since all were randomised and con- patient-controlled analgesia compared to spinal epidural anaesthesia/analgesia. trolled. We also used a random-effects model for meta- Hong Kong Med J 2006;12:442-7. analysis, which assumed some heterogeneity between stud- 18. Nielson WR, Gelb AW, Casey JE, et al. Long-term cognitive and social sequelae of general versus regional anesthesia during arthroplasty in the elderly. Anesthesiology ies and was thus more stringent in assigning statistical sig- 1990;73:1103-9. nificance than a fixed-effect model. Lastly, although these 19. Thorburn J, Louden JR, Vallance R. Spinal and general anaesthesia in total hip data represent most of the randomised evidence available, replacement: frequency of deep vein thrombosis. Br J Anaesth 1980;52:1117-21. the CIs were wide for many outcomes and the summary 20. Davis FM. Anesthesia for hip surgery. Comparative studies of spinal anaesthesia and general anaesthesia for elective total hip arthroplasty and internal fixation of frac- estimates of effect (OR and standardised mean difference) tures of the femoral neck in the elderly. University of Otago, 1987 [MD Thesis]. should be interpreted with caution. 21. Borghi B, Casati A, Iuorio S, et al. Frequency of hypotension and bradycardia dur- In conclusion, our data support the recent trend towards ing general anesthesia, epidural anesthesia, or integrated epidural-general anesthe- sia for total hip replacement. J Clin Anesth 2002;14:102-6. the increased use of regional anaesthesia. Furthermore, epi- 22. Borghi B, Casati A, Iuorio S, et al. Effect of different anesthesia techniques on red dural anaesthesia/analgesia has been shown to improve the blood cell endogenous recovery in hip arthroplasty. J Clin Anesth 2005;17:96-101. post-operative outcomes by relieving pain, reducing pulmo- 23. Davis FM, McDermott E, Hickton C, et al. Influence of spinal and general anaes- nary complications, allowing early mobilisation and short- thesia on haemostasis during total hip arthroplasty. Br J Anaesth 1987;59:561-71. ening the length of hospital stay.11,17,43 24. Gonano C, Leitgeb U, Sitzwohl C, et al. Spinal versus general anesthesia for orthopedic surgery: anesthesia drug and supply costs. Anesth Analg 2006;102:524-9. No benefits in any form have been received or will be received from a commer- 25. Lattermann R, Belohlavek G, Witmann S, et al. The anticatabolic effect of cial party related directly or indirectly to the subject of this article. neuraxial blockade after hip surgery. Anesth Analg 2005;101:1202-8. 26. Kita T, Maki N, Song YS, et al. Caudal epidural anesthesia administered intraoper- References atively provides for effective postoperative analgesia after total hip arthroplasty. J Clin Anesth 2007;19:204-8. 1. No authors listed. Health and Human Services. Number of patients, number of pro- cedures, average patient age, average length of stay of National Hospital Discharge 27. Kudoh A, Takase H, Takazawa T. A comparison of anesthetic quality in propofol- Survey 1998-2005. American Academy of Orthopaedic Surgeons, 2008. spinal anesthesia and propofol-fentanyl anesthesia for total knee arthroplasty in elderly patients. J Clin Anesth 2004;16:405-10. 2. Rodgers A, Walker N, Schug S, et al. Reduction of postoperative mortality and 28. Modig J, Karlstrom G. Intra- and post-operative blood loss and haemodynamics in morbidity with epidural or spinal anaesthesia: results from overview of randomised total hip replacement when performed under lumbar epidural versus general anaes- trials. BMJ 2000;321:1493. thesia. Eur J Anaesthesiol 1987;4:345-55. 3. Modig J, Hjelmstedt A, Sahlstedt B, Maripuu E. Comparative influences of epi- 29. Wulf H, Biscoping J, Beland B, et al. Ropivacaine epidural anesthesia and anal- dural and general anaesthesia on deep venous thrombosis and pulmonary embolism gesia versus general anesthesia and intravenous patient-controlled analgesia with after total hip replacement. Acta Chir Scand 1981;147:125-30. in the perioperative management of hip replacement. Anesth Analg 4. Modig J, Borg T, Karlstrom G, Maripuu E, Sahlstedt B. Thromboembolism after 1999;89:111-16. total hip replacement: role of epidural and general anesthesia. Anesth Analg 30. Donatelli F, Vavassori A, Bonfanti S, et al. Epidural anesthesia and analgesia 1983;62:174-80. decrease the postoperative incidence of insulin resistance in preoperative insulin- 5. Modig J, Maripuu E, Sahlstedt B. Thromboembolism following total hip replace- resistance subjects only. Anesth Analg 2007;104:1587-93. ment: a prospective investigation of 94 patients with emphasis on the efficacy of lum- 31. Hole A, Terjesen T, Breivik H. Epidural versus general anaesthesia for total hip bar epidural anesthesia in prophylaxis. Reg Anesth 1986;11:72-9. arthroplasty in elderly patients. Acta Anaesthesiol Scand 1980;24:279-87.

VOL. 91-B, No. 7, JULY 2009 942 S. HU, Z.-Y. ZHANG, Y.-Q. HUA, J. LI, Z.-D. CAI

32. Jones MJ, Piggott SE, Vaughan RS, et al. Cognitive and functional competence 38. Feller JA, Parkin JD, Phillips GW, et al. Prophylaxis against venous thrombosis after anaesthesia in patients aged over 60: controlled trial of general and regional after total hip arthroplasty. Aust N Z J Surg 1992;62:606-10. anaesthesia for elective hip or knee replacement. BMJ 1990;300:1683-7. 39. Prins MH, Hirsh J. A comparison of general anaesthesia and regional anaesthesia 33. Williams-Russo P, Sharrock NE, Haas SB, et al. Randomized trial of epidural ver- as a risk factor for deep vein thrombosis following hip surgery: a critical review. sus general anesthesia: outcomes after primary total knee replacement. Clin Orthop Thromb Haemost 1990;64:497-500. 1996;331:199-208. 40. Daniel J, Pradhan A, Pradhan C, et al. Multimodal thromboprophylaxis following 34. Keith I. Anaesthesia and blood loss in total hip replacement. Anaesthesia primary hip arthroplasty: the role of adjuvant intermittent pneumatic calf compres- sion. J Bone Joint Surg [Br] 2008;90-B:562-9. 1977;32:444-50. 41. Block BM, Liu SS, Rowlingson AJ, et al. Efficacy of postoperative epidural anal- 35. Mauermann WJ, Shilling AM, Zuo Z. Comparison of neuraxial block versus gen- gesia: a meta-analysis. JAMA 2003;290:2455-63. eral anesthesia for elective total hip replacement: a meta-analysis. Anesth Analg 2006;103:1018-25. 42. Juni P, Holenstein F, Sterne J, Bartlett C, Egger M. Direction and impact of lan- guage bias in meta-analyses of controlled trials: empirical study. Int J Epidemiol 36. Kohro S, Yamakage M, Arakawa J, et al. Surgical/tourniquet pain accelerates 2002;31:115-23. blood coagulability but not fibrinolysis. Br J Anaesth 1998;80:460-3. 43. Ballantyne J, Carr D, de Ferranti S, et al. The comparative effects of postopera- 37. Modig J, Borg T, Bagge L, Salden T. Role of extradural and of general anaesthesia tive analgesia therapies on pulmonary outcome: cumulative meta-analysis of random- in fibrinolysis and coagulation after total hip replacement. Br J Anaesth 1983;55:625-9. ized, controlled trials. Anesth Analg 1998;86:598-612.

THE JOURNAL OF BONE AND JOINT SURGERY Analgesia following total knee arthroplasty.

Knee reconstruction

Current Opinion in Orthopedics. 18(1):76-80, February 2007. Trueblood, Andrew a,b; Manning, David W a,b Abstract: Purpose of review: Conventional methods of postoperative pain control for total knee arthroplasty include continuous epidural infusions of and patient-controlled intravenous narcotics. Dose- dependent side effects are common with these modalities and may result in limited effectiveness and decreased patient satisfaction. This review intends to detail recent advances in postoperative pain management after total knee arthroplasty.

Recent findings: Multimodal, or balanced, techniques for postoperative pain control have been described that simultaneously utilize multiple analgesics with different mechanisms of action, additive effects and decreased dose-related side effects. Narcotic side-effect-sparing modalities include the scheduled use of acetaminophen, nonsteroidal anti-inflammatory drugs, tramadol, cryotherapy, and the use of sustained-release oral narcotics. In addition, regional block is a reproducible technique that provides more consistent and efficacious analgesia than patient- controlled intravenous narcotic therapy alone and is associated with fewer side effects than continuous epidural analgesia.

Summary: Regional coupled with nonnarcotic analgesics and combined long and short-acting oral narcotics decreases postoperative pain, allows for faster functional recovery, and decreases medication- related side effects. It may also shorten hospitalization and enhance patient satisfaction when compared with traditional approaches to postoperative pain management.

British Journal of Anaesthesia 103 (3): 335–45 (2009) doi:10.1093/bja/aep208 Advance Access publication July 23, 2009

REVIEWARTICLE Does regional anaesthesia improve outcome after total hip arthroplasty? A systematic review

A. J. R. Macfarlane12, G. A. Prasad1, V. W. S. Chan1 and R. Brull1*

1Department of Anaesthesia and Pain Management, Toronto Western Hospital, University Health Network, 399 Bathurst Street, Toronto, ON, Canada M5T 2S8 2Present address: Department of Anaesthesia, Glasgow Royal Infirmary, 84 Castle Street, Glasgow G4 0SF, UK *Corresponding author. E-mail: [email protected]

Total hip arthroplasty (THA) is amenable to a variety of regional anaesthesia (RA) techniques that may improve patient outcome. We sought to answer whether RA decreased mortality, cardiovascular morbidity, deep venous thrombosis (DVT) and pulmonary embolism (PE), blood loss, duration of surgery, pain, opioid-related adverse effects, cognitive defects, and length of stay. We also questioned whether RA improved rehabilitation. To do so, we performed a sys- tematic review of the contemporary literature to compare general anaesthesia (GA) and RA and also systemic and regional analgesia for THA. To reflect contemporary surgical and anaes- thetic practice, only randomized controlled trials (RCTs) from 1990 onward were included. We identified 18 studies involving 1239 patients. Only two of the 18 trials were of Level I quality. There is insufficient evidence from RCTs alone to conclude if anaesthetic technique influenced mortality, cardiovascular morbidity, or the incidence of DVT and PE when using thromboprophylaxis. Blood loss may be reduced in patients receiving RA rather than GA for THA. Our review suggests that there is no difference in duration of surgery in patients who receive GA or RA. Compared with systemic analgesia, regional analgesia can reduce postopera- tive pain, morphine consumption, and nausea and vomiting. Length of stay is not reduced and rehabilitation does not appear to be facilitated by RA or analgesia for THA. Br J Anaesth 2009; 103: 335–45 Keywords: anaesthesia, general; anaesthetic techniques, regional; analgesia, postoperative; techniques, regional; surgery, orthopaedic

Total hip arthroplasty (THA) is amenable to a variety of include the training required to develop the necessary regional anaesthesia (RA) techniques. Central neuraxial technical skills for successful RA and, more recently, the blockade (CNB) can provide excellent intraoperative expense of ultrasound equipment as this method of nerve anaesthesia and prolonged postoperative analgesia. localization increases in popularity. Finally, many patients Peripheral nerve blockade (PNB) avoids some of the are fearful of RA and may have misconceptions about the unwanted adverse effects of CNB and allows for targeted technique involved.43 analgesia of the operative limb.66465The use of continu- Although RA is increasingly used, the results of large ous PNB has increased as it has the advantage of longer meta-analyses and randomized controlled trials (RCTs) postoperative analgesia compared with a single-injection comparing general anaesthesia (GA) and RA for major technique.83153 lower limb orthopaedic surgery often conflict.71844566977 Despite a low rate of complications and apparent It is not uncommon for the results of large RCTs to dis- benefits in certain orthopaedic procedures, including better agree with those of meta-analyses.30 This can be due to postoperative analgesia, improved rehabilitation, and a the inclusion of small studies, publication bias, and reduced length of hospital stay, there are disadvantages of sample heterogeneity between different trial populations RA.2 – 4 10 14 15 22 49 60 72 76 There is an inherent block and meta-analysis bias.32 40 46 More importantly, many failure rate, even in expert hands.51 Operating theatre trials included in recent meta-analyses were originally delays and a perceived risk of increased liability are criti- published more than 30 yr ago and do not reflect modern cisms also often directed at RA.38 52 Other limiting factors anaesthetic or surgical practice. Landmark articles which

# The Author [2009]. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please email: [email protected] Macfarlane et al. compared GA and RA for hip surgery used drugs that are recorded. The specific outcomes sought in each article no longer available.47 48 In the past two decades, surgical were: (i) mortality, (ii) cardiovascular morbidity (myocar- techniques and postoperative patient care have improved dial infarction, arrhythmia, and hypotension), (iii) DVT, considerably, new thromboembolic prophylaxis regimes (iv) PE, (v) blood loss, (vi) duration of surgery, (vii) have been introduced, and RA has advanced as a result of pain (pain scores and morphine consumption), (viii) enhanced needle technology, block placement techniques, opioid-related adverse effects (nausea, vomiting, pruritis, catheter design, and infusion pumps.25 27 33 58 63 We have sedation, urinary retention, and respiratory depression), re-examined existing data66 for relevance and application (ix) cognitive defects, (x) length of stay, and (xi) rehabili- to modern anaesthetic practice. tation (range of motion and ambulation). It was noted We have performed a systematic review of the literature, whether each outcome was primary or secondary. Each published from 1990 onwards, to ascertain if either RA outcome was then evaluated qualitatively for each inter- was superior to GA or regional analgesia was superior to vention and comparator and the data were recorded in systemic analgesia for THA. The specific questions we tables. Because there were a limited number of studies sought to answer were whether, when compared with GA with homogenous design for each outcome, meta-analysis or systemic analgesia, RA or regional analgesia for THA was not performed. decreased: (i) mortality, (ii) cardiovascular morbidity, (iii) The methodological quality of each trial was assessed deep venous thrombosis (DVT) and pulmonary embolism using several criteria. The likelihood of methodological (PE), (iv) blood loss, (v) duration of surgery, (vi) pain, bias of each RCT was assessed using the Jadad score,34 (vii) opioid-related adverse effects, (viii) cognitive defects, which assigns points based on three factors. One point was and (ix) length of stay. We also examined whether or not given to randomized studies, an additional point was given RA or regional analgesia improved rehabilitation. if the method of randomization was described and appro- priate, and one point was deducted if randomization was inappropriate. One point was given if a study was double- blind, and an additional point was given if the blinding Methods procedure was described and appropriate. One point was Two of the authors (G.A.P. and R.B.) searched the elec- deducted if blinding was inappropriate. One point was tronic databases MEDLINE, EMBASE, and the Cochrane given if the numbers and reasons for withdrawals were Central Register of Controlled Clinical Trials (from described. The maximum score is 5; trials scoring 3 or January 1990 to October 2008) using the following popu- more are generally regarded as having satisfactory metho- lation search terms: ‘total hip replacement’ OR ‘total hip dologic quality. Allocation concealment, which helps arthroplasty’ OR ‘hip operation’. These search results were eliminate selection bias, was assessed and defined as ade- combined with ‘anaesthesia’ OR ‘analgesia’ using the quate, unclear, or inadequate. Finally, whether patient Boolean search operator AND. The references of retrieved follow-up rates were ,80% was recorded. articles were hand searched for any relevant articles not After abstraction of information, a level of evidence was identified in the original search. assigned to the outcomes of each RCT (Level I is a high- The study selection criteria were limited to include only quality RCT; Level II is a lesser quality RCT, e.g. ,80% RCTs, English language, and human adults. Each abstract follow-up, no blinding, or improper randomization). Two was screened to identify studies that had randomized authors (A.J.R.M. and R.B.) independently reviewed and patients to compare GA or RA for surgery. RCTs comparing scored each RCT using this methodology. systemic or regional techniques for postoperative analgesia were also included. Studies were excluded if surgery other than a joint arthroplasty was performed, or if both hip and Results knee arthroplasty were treated as one single study popu- In total, 18 RCTs were identified that compared either GA lation and data on the patients undergoing knee surgery vs RA, systemic vs regional analgesia, or both for THA were not presented separately in the results.926293573 (Fig. 1). Ten of these had a Jadad score of 2 or less. Studies using opioid-only neuraxial techniques were Allocation concealment was unclear in 13 trials and excluded.24 71 Finally, studies were excluded if the primary inadequate in one. Follow-up was adequate in all trials outcome was not included in the list described above.13 57 (Table 1). Two RCTs were considered to provide Level I The data were extracted onto a templated evidence- evidence.37 65 In total, the studies included 1239 patients. based medicine literature review form to assist in the sys- A summary of the outcomes reported in each trial is pro- tematic review of full-text articles and data collection. vided in Table 2. Data extracted for comparison included year of publi- cation, author, total number of subjects, mean patient age, per cent male, and co-morbidity. The intervention (specific Mortality RA, regional analgesia technique, or both) and comparator There were no trials primarily designed to assess differ- (GA, specific systemic analgesia technique, or both) were ences in mortality after GA or RA for THA. Only two

336 GA vs RA for THA

Deep venous thrombosis and pulmonary embolism Total hip arthroplasty OR Anaesthesia AND Total hip replacement OR OR analgesia The incidence of DVT and PE after THA was evaluated as Hip operation a secondary outcome in two trials comparing GA with 20 55 639 citations RA. In both of these trials, all patients received RCTs only chemical thromboprophylaxis. When comparing EA and GA with GA alone, there was no difference in the inci- 213 abstracts dence of DVT on postoperative days one to nine, with Foreign language, non-adult subjects administered to all patients after operation (Level 20 166 abstracts II). Spinal anaesthesia was compared with GA using GA vs RA, three groups which differed with respect to administration systemic vs regional of thromboprophylaxis.55 First, the GA group received pre- analgesia, or both 27 full-text articles operative enoxaparin, unlike the two spinal anaesthesia Data not exclusive for groups. Secondly, one of the spinal anaesthesia groups THA, did not meet inclusion criteria received enoxaparin 1 h after operation. All three groups 18 full-text articles received daily enoxaparin from 12 h after operation. Compared with the GA group, distal DVTs (measured Fig 1 Flow chart of screened, excluded and analysed papers. 12–15 days after operation) were more frequent in the spinal anaesthesia group who did not received enoxaparin trials documented mortality as a secondary outcome at 1 h (11% vs 0%, P¼0.013, Level II).55 The incidence (Level II). The follow-up period was 48 h in one study of proximal DVTs did not differ significantly between the and 16 days in the other.55 78 There was only one death three groups (Level II). This was the only study that exam- reported in 276 patients included in these two studies; no ined the incidence of PE, again as a secondary outcome, difference in mortality was detected between the groups in but as this was zero in both RA and GA groups, no differ- either study. ence was detected (Level II).55

Blood loss Cardiovascular morbidity Ten studies measured intraoperative blood loss. In four of Six trials examined cardiovascular morbidity, although this these, intraoperative blood loss was reduced (ranging from was a primary outcome in only one trial.11 Hypotension 118 to 595 ml) by RA compared with either GA or sys- was most commonly studied. GA combined with epidural temic analgesia, whereas in the six others, there was no 1219–21555961646567 anaesthesia (EA), when compared with GA alone, resulted significant difference. Postoperative in significantly more frequent hypotensive episodes, both at blood loss was recorded in seven trials. In two of these, induction (41% vs 23%, P¼0.0049) and intraoperatively postoperative blood loss was reduced (ranging from 140 to (54% vs 35%, P¼0.04). Intraoperative mean arterial 517 ml) in the RA group compared with GA or systemic pressure (80 vs 88 mm Hg, P¼0.037) was also lower in the analgesia, whereas in the five others, there was no differ- 12 19 – 21 55 65 67 combined group (Level II).11 20 However, the effect of epi- ence. Blood transfusion was reduced (range dural analgesia on hypotension after operation varied, and 1.3–1.7 units) in the RA group in two out of the five 19–215561 in three trials, the frequency was either less (0% vs 33%, studies that measured this outcome. All studies P¼0.001, Level I), greater (40% vs 13%, P¼0.01, Level were Level II evidence except one Level I study which II), or no different (Level II) when compared with patients demonstrated reduced intra- and postoperative blood loss 65 receiving systemic analgesia.37 61 78 The incidence of post- using combined GA and LPB compared with GA alone. operative hypotension with continuous In no case was blood loss significantly greater in the RA was significantly less (0% vs 40%, P¼0.01, Level II) when group compared with GA or systemic analgesia. (For compared with CEA.61 Continuous lumbar plexus block further information, see Supplementary Table 4.) (CLPB), compared with systemic analgesia, did not affect the incidence of hypotension (Level II), again when studied Duration of surgery as a secondary outcome however.5 The incidence of brady- The duration of surgery was not influenced by the type of cardia was assessed in two studies578and was not signifi- anaesthetic for THA (Level II).6 11 12 19 – 21 37 50 55 65 67 68 cantly different with either CEA or CLPB when compared with systemic analgesia (Level II). Myocardial ischaemia or significant arrhythmias other than bradycardia was inves- Postoperative analgesia tigated in one study comparing GA with or without lumbar We identified 11 RCTs comparing systemic and regional EA.20 There was no difference between the groups, up to analgesia for THA and in all but one,61 regional analgesia 48 h after operation (Level II). (For further information, see reduced pain scores or morphine consumption.5637455059 Supplementary Table 3.) 64 65 68 78 There was good evidence that, compared with

337 Table 1 Patient characteristic data, design, and breakdown of Jadad score for each trial. *Randomization groups. Age is presented as mean (SD) or median (range) where available. Alloc Conceal, allocation concealment; SA, systemic analgesia; CPB, continuous psoas block; CS, continuous spinal; CLPB, continuous lumbar plexus block; SNB, ; GA, general anaesthesia; EA, epidural anaesthesia; FNB, femoral nerve block; OB, obturator nerve block; CFNB, continuous femoral nerve block; CSE, combined spinal epidural; CPB, continuous psoas block; R, ropivacaine; PCA, patient-controlled analgesia; CEA, continuous epidural analgesia; N/A, not available; ASA, American Society of Anesthesiologists; †See Remarks

Study n Anaesthesia Analgesia Age (yr) % Male Co-morbidity Alloc Conceal Jadad score Remarks

Becchi and colleagues5 37 SA CPB* 69 (8) 43 ASA I–II; age 61–82 Adequate 11 001 (3) Both CPB and i.v. infusion 36 SA Systemic infusion*† 71 (11) 43 continued for 48 h; †morphine and ketorolac; observors and patients Macfarlane adequately blinded but not

338 investigators Siddiqui and colleagues59 17 GA CLPB* 59 (12) 41 ASA I–III; age 18–80; body mass Adequate 11 001 (3) CLPB

17 GA I.V. PCA* 52 (13) 53 index .40 excluded al. et Stevens and colleagues64 22 SA Fascia iliaca* 69 (10) 68 ASA I–III Unclear 11 101 (4) Modified fascia iliaca 22 SA Placebo* 67 (9) 50 block—injection 1 cm above the inguinal ligament. and bupivacaine 0.5% used for block Eroglu and colleagues21 20 EA* CEA 64 (13) 20 ASA I–III; patients with unstable Unclear 11 000 (2) Hypotensive anaesthesia in 20 GA*† CEA 62 (10) 30 angina, both groups (mean arterial valvular disease, neurologic, or pressure 50–60 mm Hg); cerebrovascular †total i.v. anaesthesia disease excluded Singelyn and colleagues61 15 GA CFNB* 67 (11) 33 ASA I–II; age 18–80 Unclear 11 000 (2) CFNB and CEA continued 15 GA CEA* 61 (13) 47 for 48 h 15 GA I.V. PCA* 66 (15) 40 Borghi and colleagues12 70 EA* CEA 66 (10) 46 ASA I–III Adequate 11 000 (2) Drop-outs replaced by a 70 EAþGA* CEA 64 (11) 36 new randomized 70 GA* I.V. opioid infusion 64 (10) 33 participant Biboulet and colleagues6 16 GA FNB* 53 (12) 63 N/A Unclear 10 101 (3) 15 GA Psoas block* 57 (13) 40 14 GA I.V. PCA* 58 (14) 64 Maurer and colleagues45 34 CS anaesthesia CS analgesia * 74 (9) 38 ASA I–III; age 50–85; weight 40– Inadequate 11 001 (3) CS analgesia continued for 34 SA I.V. PCA* 70 (11) 55 100 kg 24 h Borghi and colleagues11 70 EA* CEA 66 (10) 46 ASA I–III Adequate 11 000 (2) Same population as in 70 EAþGA* CEA 64 (11) 36 Borghi and colleagues12 70 GA* I.V. opioid infusion 64 (10) 33 Kampe and colleagues37 12 EA CEAþplacebo i.v. PCA* 66 (9) 25 ASA I–III; age 18–80; weight 50–90 Unclear 10 111 (4) CEA continued for at least 11 EA Placebo CEAþi.v. PCA* 62 (6) 33 kg 44 h (duration of study); four patients in placebo CEA vs two in CEA received GA Stevens and colleagues65 30 GA LPB* 66 (10) 50 ASA I–III Unclear 10 111 (4) 30 GA I.V. PCA* 66 (10) 50 Wulf and colleagues78 43 EA* CEA N/A N/A ASA I–III; age older than 18 yr Unclear 10 001 (2) Epidural infusion 45 GA* I.V. PCA N/A N/A continued for 24 h; boluses thereafter if required; removed at 48 h D’Ambrosio and colleagues19 15 EAþGA*I CEA 62 (9) 53 ASA I–II Unclear 10 000 (1) CEA continued for 48 h; 15 EAþGA*II CEA 67 (7) 60 I. Aprotinin 500 000 KIU 15 GA*III Systemic opiates 67 (9) 47 preop and 500 000 KIU 15 GA*IV Systemic opiates 61 (13) 47 h21 intraoperatively; II. No aprotinin (placebo); III. Aprotinin 500 000 KIU preop and 500 000 KIU h21 intraoperatively; IV. No aprotinin (placebo) Dauphin and colleagues20 20 EAþGA* I.V./I.M. opioid 71 (7) 35 ASA I–III Unclear 11 001(3) Epidural for intraoperative 17 GA* I.V./I.M. opioid 66 (14) 41 period only Moiniche and colleagues50 11 EA* CEA* 73 (range 67–82) 18 No apparent exclusions Unclear 11 000 (2) CEA continued for 48 h GA

11 GA* I.M. morphine* 71 (range 61–85) 9 vs 68

Uhrbrand and colleagues 93 GA ‘3 in 1’ block* 71 (range 48–85) 47 N/A Unclear 10 001 (2) THA for RA 339 89 GA Systemic opioid* 70 (range 53–88) 48 Planes and colleagues55 65 SA*I Acetaminophen 67 (1) 46 ASA I–III; age .45 yr; weight .45 Unclear 10 001 (2) I. No additional 61 SA*II Acetaminophen 67 (1) 52 kg enoxaparin; II. Enoxaparin 62 GA*III Acetaminophen 67 (1) 45 20 mg 1 h after SA; III. Enoxaparin 40 mg 12 h before GA; enoxaparin 12 h after operation and daily thereafter Twyman and colleagues67 10 GAþLPB* Undisclosed N/A 0 Women only Unclear 10 000 (1) 10 GA Undisclosed N/A 0 Table 2 Outcomes and levels of evidence for each study included in the review. Arrows indicate whether the outcome is no different ($), better (#), or worse (") with RA, regional analgesia, or both compared with GA, systemic analgesia, or both. *Randomization groups. Mort, mortality; CVS, cardiovascular morbidity; DVT, deep venous thrombosis; PE, pulmonary embolus; Cog Def, cognitive deficit; LOS, length of stay; Rehab, rehabilitation; SA, spinal anaesthesia; CS, continuous spinal; CLPB, continuous lumbar plexus block; SNB, sciatic nerve block; GA, general anaesthesia; EA, epidural anaesthesia; FNB, femoral nerve block; CFNB, continuous femoral nerve block; R, ropivacaine; PCA, patient-controlled analgesia; CSE, combined spinal–epidural; CEA, continuous epidural analgesia; MC, morphine consumption; B, bupivacaine

Study Anaesthesia Analgesia Outcomes (levels of evidence)

Mort CVS DVT PE Blood loss Duration of surgery Pain Adverse Cog Def LOS Rehab effects

Becchi and colleagues5 SA CPB* $ (II) # (II) # (II) SA I.V. infusion* Siddiqui and colleagues59 GA CLPB* $ (II) # (II); #MC (II) # (II) GA I.V. PCA* Stevens and colleagues64 SA Fascia iliaca* $ (II) $ (II); #MC (II) $ (II) SA Placebo* Eroglu and colleagues21 EA* CEA # (II) $ (II) GA* CEA Singelyn and colleagues61 GA CFNB* " (II) $ (II) $ (II) $ (II) $ (II) $ (II) GA CEA* GA I.V. PCA* Borghi and colleagues12 EA* CEA $ (II) $ (II)

GAþEA* CEA Macfarlane GA* I.V. opioid infusion Biboulet and colleagues6 GA FNB* $ (II) # (II); #MC (II); $ (II) $ (II) $ (II) 340 GA Psoas block* psoas block only GA I.V. PCA* tal. et Maurer and colleagues45 CS anaesthesia* CS analgesia* # (II) # (II) SA* I.V. PCA* Borghi and colleagues11 EA* CEA " (II); EAþGA group GAþEA* CEA GA* I.V. opioid infusion Kampe and colleagues37 EA CEAþplacebo i.v. PCA* # (I) $ (II) # (I) # (I) EA Placebo CEAþi.v. PCA* Stevens and colleagues65 GA LPB* # (I) $ (II) # (I); #MC (I) $ (II) GA I.V. PCA* Wulf and colleagues78 EA* CEA $ (II) " (II) $ (II) # (II) # (II) # (II) $ (II) GA* I.V. PCA D’Ambrosio and colleagues19 EAþGA* CEA # (II) $ (II) EAþGA* CEA GA* Systemic opiates GA* Systemic opiates Dauphin and colleagues20 EAþGA* I.V./I.M. opioid " (II) $ (II) # (II) $ (II) GA* I.V./I.M. opioid Moiniche and colleagues50 EA* CEA* $ (II) # (II); $MC (II) $ (II) $ (II) GA* I.M. morphine* Uhrbrand and colleagues68 GA ‘3 in 1’ block* $ (II) # MC (II) GA Systemic opioid* Planes and colleagues55 GA CFNB* $ (II) " (II) $ (II) $ (II) $ (II) GA I.M. opioid* Twyman and colleagues67 GAþLPB* Undisclosed # (II) $ (II) GA* Undisclosed GA vs RA for THA systemic analgesia, single-shot lumbar plexus block (LPB) systemic analgesia (Level II). These were all secondary out- reduced pain scores and morphine consumption but only comes, and two of the trials were also completely unblinded. for a short duration (Level I).665Continuous LPB, (For further information, see Supplementary Table 8.) however, was of benefit for up to 48 h compared with sys- temic analgesia (Level II).559In three37 50 78 out of four trials,61 epidural analgesia provided superior pain relief compared with systemic analgesia. The analgesic benefit Discussion lasted from only 8 (Level I) to 48 h (Level II). Compared The aim of this systematic review was to examine contem- with a placebo injection, a modified fascia-iliaca block porary RCTs, published from 1990 onwards, comparing reduced morphine consumption up to 24 h, but not pain either GA and RA or regional and systemic analgesia for scores (Level II).64 Similarly, a ‘3 in 1’ block also reduced THA. We found insufficient evidence from RCTs alone to 24 h opioid consumption (Level II) compared with sys- conclude whether anaesthetic technique influenced mor- temic opioid alone.68 Femoral nerve block alone, however, tality, cardiovascular morbidity, duration of surgery, or the either as a single shot or continuous infusion was of no incidence of DVT and PE in the setting of routine throm- benefit.661Continuous spinal analgesia reduced pain boprophylaxis. Blood loss may be reduced in patients scores compared with systemic analgesia for the 24 h dur- receiving RA rather than GA for THA. RA does, however, ation of the infusion (Level II).45 (For further information, reduce postoperative pain and also nausea and vomiting. see Supplementary Table 5.) Length of stay is not reduced and rehabilitation does not appear to be facilitated by RA or analgesia for THA. Adverse effects Before further considering the implications of our review, we accept that there are several limitations. First, Although all adverse effects were analysed as secondary for practical reasons, we chose to include only English outcomes, significant reductions in postoperative nausea language trials. Although this may have introduced bias, it (relative risk reduction 85–93%, Level II)54559and vomit- has been suggested that excluding trials not published in ing (relative risk reduction of 100%, Level I)37 45 were English has little effect on summary treatment effect esti- observed in favour of regional analgesia in four of the mates.36 Secondly, we found that 10 of the 18 RCTs evalu- nine studies examining this outcome.5637455961646578 ated herein had Jadad scores of 2 or less. This was In the other five studies, there was no difference between primarily due to lack of any form of blinding in a number systemic and regional analgesia. There was no difference of the trials. Double-blinding can be difficult to institute detected in the incidence of pruritis, sedation, urinary when studying peripheral nerve blocks for systemic analge- retention, or respiratory depression in the studies that exam- sia, especially in awake patients, as sham nerve blocks are ined these as secondary outcomes.5637455961(For further often not performed for ethical reasons. The latter notwith- information, see Supplementary Table 6.) standing the Jadad score is effectively reduced automati- cally by 2 points for the lack of a double-blinded study Cognitive deficit design. Thirdly, there were no large (n.1000) trials ident- In the only trial that commented on postoperative cogni- ified. The studies included in the present review had sample tive deficit after THA, those patients who received EA had sizes varying from only 22 to 210 patients. In trials with ‘superior mental clarity and cooperativeness’ compared small numbers of subjects, the absence of ‘significant’ with the GA group.78 The method of assessment for cogni- differences in secondary outcomes must be interpreted with tion was unclear. caution as these studies are often inadequately powered to detect such differences.42 We took care to highlight second- Length of stay ary outcomes in the summary tables and these shortcom- ings are also reflected in the level of evidence scores (i.e. Four studies measured length of stay (some including time by definition, Level II). Finally, the purpose of this review spent in a rehabilitation centre) as either a primary or a was not to provide recommendations on the preferred mode secondary outcome, but there was no difference between of anaesthesia for THA. To do so would have required an regional (FNB,6 CFNB,61 and CEA50 61) and systemic assessment of harm and consideration of other information analgesia (Level II). (For further information, see such as costs, quality of life, and feasibility. One major Supplementary Table 7.) concern with RA for instance is the risk of nerve injury compared with GA. This is difficult to quantify and has Rehabilitation been addressed elsewhere in the contemporary literature.14 Three studies examined rehabilitation. Outcome measures As with any anaesthetic technique, patient co-morbidity included daily hours of ambulation and activity scores,50 and preference and also expertise of the anaesthesiologist time to first ambulation,61 or the different ranges of move- must also be considered. ment about the hip joint.661In all three RA, regional analge- The lack of difference in mortality between GA and RA sia, or both provided no benefit compared with GA and for THA is unsurprising given the safety of modern

341 Macfarlane et al. anaesthetic and surgical practice. Much greater numbers thromboprophylaxis. Although it has been suggested that than those included in the two RCTs that we identified CNB may decrease the incidence of DVT either directly would be required to demonstrate any difference.55 78 by enhancing lower extremity venous blood flow or by Furthermore, neither trial followed up patients beyond the attenuating the prothrombotic effects of the stress study period (48 h and 15 days). A large meta-analysis response, further work is required to ascertain whether RA comparing GA and CNB alone for a variety of surgery offers any additive benefit when used in combination with types found that overall mortality was reduced by contemporary routine thromboprophylaxis. Indeed, one one-third [odds ratio (OR) 0.70, 95% confidence interval trial included in our review where all patients received (CI) 0.54–0.90] in patients allocated to CNB.56 When chemical thromboprophylaxis, but which did not have each group in this meta-analysis was analysed according DVT as an outcome, found no difference in laboratory to type of surgery, there was decreased mortality only in measurements of coagulation between patients who the orthopaedic subgroup.56 Interestingly, overall mortality received GA and RA.21 This lack of difference was simi- was reduced whether or not CNB was continued after larly reflected in a further trial that did not meet our operation. Conversely, combined intraoperative GA and inclusion criteria but compared haemostatic markers in CNB negated the mortality benefit of CNB alone. In a ret- patients undergoing THA with either GA or spinal anaes- rospective review examining postoperative analgesia for thesia.13 Again, patients received chemical thrombopro- THA, there was no difference in mortality when compar- phylaxis. Although no difference occurred between GA ing epidural analgesia and systemic analgesia.77 These and RA in the one trial that examined the incidence of PE, data suggest that intraoperative CNB alone may confer the both the method of randomization and diagnosis of PE most benefit. Indeed, in a meta-analysis comparing CNB were unclear and there was no blinding of observers.55 with GA for hip fracture repair, CNB was shown to reduce Four trials demonstrated reduced blood loss in THA with 1 month mortality in patients undergoing hip fracture RA, whereas in six there was no difference. A recently pub- surgery.69 We recognize, however, that the latter lished meta-analysis found that CNB for THA reduced the meta-analysis of urgent hip fracture repair, arguably in a likelihood of transfusion by three-quarters (OR 0.25, 95% generally sicker group of patients, is not necessarily appli- CI 0.11–0.53), although seven of the 10 trials included in cable to patients undergoing elective THA. this analysis were published before 1990 and may not reflect Only three trials examined cardiovascular morbidity current surgical or anaesthetic practice.28 Interestingly, in other than hypotension, comparing either CLPB or epi- this meta-analysis, no difference in intraoperative blood loss dural analgesia vs systemic analgesia, or EA and GA vs was detected between RA and GA. This contrasts with a GA alone. Although there was no difference in the inci- meta-analysis which found a reduction in intraoperative dence of ischaemia or arrhythmia between the groups, this blood loss by 275 ml per case (95% CI 180–371 ml) in the could have been the result of inadequate numbers of RA group.44 Again, however, of the eight (some quasirando- patients. In the meta-analysis described above, there was a mized only) trials included, seven were published before reduction in the incidence of in the 1990. With the variety of anaesthetic, analgesic, and throm- epidural group, although the CI just reached zero.23 This boprophylaxis regimes in the 10 trials included in our significant difference was detected only when all surgical review, results were not homogeneous enough to pool. Three groups were combined and did not specifically apply to trials did compare combined EA and GA vs GA, but even orthopaedic patients alone. Further RCTs with large here there was a confounding factor, such as a statistically numbers are required to examine whether RA reduces significant difference in arterial pressure between the groups. serious cardiovascular morbidity in elective THA. With the use of restrictive transfusion strategies now com- We found no difference in the frequency of proximal monplace, and many patients now participating in preopera- DVT in the two trials that compared RA and GA, although tive blood conservation therapies, it may become more both trials suffered from a combination of inadequate difficult to demonstrate a significant reduction in transfusion blinding, underpowering, or both.20 55 Although one trial requirements in primary arthroplasty if the prevalence of found an increased incidence of distal DVT in the RA blood transfusion decreases, particularly if this endpoint is group compared with GA, the two groups were not truly studied as a secondary outcome. comparable because enoxaparin was administered at In contrast to our findings, a recent meta-analysis different times in each group. Furthermore, the clinical demonstrated a modest, but significant, decrease in surgi- importance of distal DVTs is arguably less than that for cal time with CNB compared with GA for THA.44 It is proximal clots. Our findings contrast with the results of a noteworthy that the definition of operating theatre time recent meta-analysis where patients undergoing THA with may vary, and the majority of the studies we reviewed did CNB compared with GA had a significantly lower risk of not consider the total time including anaesthetic interven- DVT (OR 0.27, 95% CI 0.17–0.42) and PE (OR 0.26, tion. A ‘block room’ or anaesthesia induction room, where 95% CI 0.12–0.56).44 Importantly, however, among the RA is performed before patient transfer, can increase five trials included in this meta-analysis, all were carried efficiency and boost cost-savings by reducing out before 1989 and none utilized routine anaesthesia-related time.1627475

342 GA vs RA for THA

Our review found that regional analgesia reduced post- In conclusion, we found insufficient evidence from operative pain in THA. The purpose of our review was not RCTs alone to conclude whether anaesthetic technique to specifically compare the different RA or analgesia tech- influenced mortality, cardiovascular morbidity, duration of niques for THA. This has been addressed elsewhere.23 surgery, or the incidence of DVT and PE in the setting of Choi and colleagues18 recently published a meta-analysis routine thromboprophylaxis. Our systematic review does comparing postoperative epidural analgesia with systemic suggest that blood loss may be reduced in patients receiv- analgesia after THA or total knee arthroplasty (TKA) and ing RA rather than GA for THA. Regional analgesia does, concluded that epidural analgesia provided better pain however, reduce postoperative pain and also nausea and relief for up to 6 h after operation compared with systemic vomiting. Length of stay is not reduced and rehabilitation analgesia. All patients, however, whether THA or TKA, does not appear to be facilitated by RA or analgesia for were analysed together, despite important differences in THA. Sixteen of the 18 RCTs identified were of Level II the severity of postoperative pain between these two surgi- quality and therefore further work is required to investigate cal procedures. Although when THA is examined indepen- whether RA influences the majority of outcomes after dently, we found that pain scores were reduced for up to THA. 48 h,50 78 the sole trial of Level I quality demonstrated that the benefit of epidural analgesia was in fact only evident for 8 h.37 LPBs appeared to be of benefit but are arguably associated with more serious complications than Supplementary material 16 17 other peripheral nerve blocks such as femoral block. Supplementary material is available at British Journal of The lack of benefit of femoral block alone, however, com- Anaesthesia online. pared with either fascia iliaca or ‘3 in 1’ blocks is not unsurprising because this block does not provide anaesthe- sia to the lateral cutaneous nerve of the thigh which innervates the site of skin incision in THA. Although con- References tinuous spinal anaesthesia can provide reliable, titratable 1 Armstrong KP, Cherry RA. Brachial plexus anesthesia compared anaesthesia and analgesia, this is not yet used widely due to general anesthesia when a block room is available. Can J to frequent technical difficulties, requirement for closer Anaesth 2004; 51:41–4 ward , and fears of complications such as cauda 2 Aromaa U, Lahdensuu M, Cozanitis DA. Severe complications equina syndrome.39 70 associated with epidural and spinal anaesthesias in Finland 1987– Like postoperative pain, opioid-related adverse effects, 1993. A study based on patient insurance claims [see comment]. especially nausea and vomiting, are a major concern to Acta Anaesthesiol Scand 1997; 41: 445–52 3 Auroy Y, Benhamou D, Bargues L, et al. Major complications of patients and can delay discharge from hospital.41 54 regional anesthesia in France: The SOS Regional Anesthesia Unfortunately, however, opioid-related adverse effects are Hotline Service. Anesthesiology 2002; 97: 1274–80 almost never investigated as a primary outcome. Despite 4 Auroy Y, Narchi P, Messiah A, Litt L, Rouvier B, Samii K. Serious this we found, primarily Level II evidence, that single shot complications related to regional anesthesia: results of a prospec- and continuous LPB, continuous spinal analgesia, and epi- tive survey in France. Anesthesiology 1997; 87: 479–86 dural analgesia (Level I) reduced postoperative nausea and 5 Becchi C, Al Malyan M, Coppini R, Campolo M, Magherini M, vomiting.5374559A reduction in morphine consumption Boncinelli S. Opioid-free analgesia by continuous psoas compart- ment block after total hip arthroplasty. A randomized study. Eur J was not always associated with a reduction in adverse Anaesthesiol 2008; 25: 418–23 effects, but this may be due to inadequate powering. The 6 Biboulet P, Morau D, Aubas P, Bringuier-Branchereau S, Capdevila five RCTs in which no benefit was observed were all X. Postoperative analgesia after total-hip arthroplasty: compari- graded as Level II. These were either inadequately son of intravenous patient-controlled analgesia with morphine powered or had poor methodological quality. and single injection of femoral nerve or psoas compartment We found that RA or regional analgesia does not appear block. A prospective, randomized, double-blind study. Reg Anesth to significantly shorten length of stay in hospital or hasten Pain Med 2004; 29: 102–9 7 Block BM, Liu SS, Rowlingson AJ, Cowan AR, Cowan JA, Jr, Wu postoperative rehabilitation for THA. This apparent lack of CL. Efficacy of postoperative epidural analgesia: a meta-analysis. sustained benefit in the setting of THA is quite different J Am Med Assoc 2003; 290: 2455–63 from RCTs examining the TKA population, which is 8 Boezaart AP. Perineural infusion of local anesthetics. likely due to rapidly decreasing pain after THA in com- Anesthesiology 2006; 104: 872–80 parison with TKA.15 60 Furthermore, postoperative phy- 9 Bogoch ER, Henke M, Mackenzie T, Olschewski E, Mahomed NN. siotherapy in TKA can exacerbate pain severity. Finally, Lumbar paravertebral nerve block in the management of pain many centres performing major joint surgery now after total hip and knee arthroplasty: a randomized controlled clinical trial. J Arthroplasty 2002; 17: 398–401 implement clinical pathway protocols. It may become 10 Borgeat A, Ekatodramis G, Kalberer F, Benz C. Acute and nona- increasingly difficult to detect differences in length of stay cute complications associated with interscalene block and based on mode of anaesthesia as streamlined care often shoulder surgery: a prospective study. Anesthesiology 2001; 95: results in all patients being discharged at similar times. 875–80

343 Macfarlane et al.

11 Borghi B, Casati A, Iuorio S, et al. Frequency of hypotension and 30 Halpern S. Why meta-analysis? Reg Anesth Pain Med 2002; 27: bradycardia during general anesthesia, epidural anesthesia, or 3–5 integrated epidural-general anesthesia for total hip replacement. 31 Hebl JR, Kopp SL, Ali MH, et al. A comprehensive anesthesia pro- J Clin Anesth 2002; 14: 102–6 tocol that emphasizes peripheral nerve blockade for total knee 12 Borghi B, Casati A, Iuorio S, et al. Effect of different anesthesia and total hip arthroplasty. J Bone Joint Surg Am 2005; 87:63–70 techniques on red blood cell endogenous recovery in hip arthro- 32 Higgins MS, Stiff JL. Pitfalls in performing meta-analysis: I. plasty. J Clin Anesth 2005; 17: 96–101 Anesthesiology 1993; 79: 405 13 Brueckner S, Reinke U, Roth-Isigkeit A, Eleftheriadis S, 33 Huo MH, Parvizi J, Bal BS, Mont MA. What’s new in total hip Schmucker P, Siemens HJ. Comparison of general and spinal arthroplasty. J Bone Joint Surg Am 2008; 90: 2043–55 anesthesia and their influence on hemostatic markers in patients 34 Jadad AR, Moore RA, Carroll D, et al. Assessing the quality of undergoing total hip arthroplasty. J Clin Anesth 2003; 15: 433–40 reports of randomized clinical trials: is blinding necessary? Control 14 Brull R, McCartney CJ, Chan VW, El-Beheiry H. Neurological Clin Trials 1996; 17: 1–12 complications after regional anesthesia: contemporary estimates 35 Jones MJ, Piggott SE, Vaughan RS, et al. Cognitive and functional of risk. Anesth Analg 2007; 104: 965–74 competence after anaesthesia in patients aged over 60: controlled 15 Capdevila X, Barthelet Y, Biboulet P, Ryckwaert Y, Rubenovitch J, trial of general and regional anaesthesia for elective hip or knee d’Athis F. Effects of perioperative analgesic technique on the sur- replacement. Br Med J 1990; 300: 1683–7 gical outcome and duration of rehabilitation after major knee 36 Juni P, Holenstein F, Sterne J, Bartlett C, Egger M. Direction and surgery. Anesthesiology 1999; 91: 8–15 impact of language bias in meta-analyses of controlled trials: 16 Capdevila X, Coimbra C, Choquet O. Approaches to the lumbar empirical study. Int J Epidemiol 2002; 31: 115–23 plexus: success, risks, and outcome. Reg Anesth Pain Med 2005; 30: 37 Kampe S, Randebrock G, Kiencke P, et al. Comparison of con- 150–62 tinuous epidural infusion of ropivacaine and with intra- 17 Capdevila X, Macaire P, Dadure C, et al. Continuous psoas com- venous patient-controlled analgesia after total hip replacement. partment block for postoperative analgesia after total hip arthro- Anaesthesia 2001; 56: 1189–93 plasty: new landmarks, technical guidelines, and clinical 38 Katz J. A survey of anesthetic choice among anesthesiologists. evaluation. Anesth Analg 2002; 94: 1606–13 Anesth Analg 1973; 52: 373–5 18 Choi PT, Bhandari M, Scott J, Douketis J. Epidural analgesia for 39 Lambert DH, Hurley RJ. syndrome and continuous pain relief following hip or knee replacement. Cochrane Database spinal anesthesia. Anesth Analg 1991; 72: 817–9 Syst Rev 2003; CD003071 40 LeLorier J, Gregoire G, Benhaddad A, Lapierre J, Derderian F. 19 D’Ambrosio A, Borghi B, Damato A, D’Amato G, Antonacci D, Discrepancies between meta-analyses and subsequent large ran- Valeri F. Reducing perioperative blood loss in patients undergoing domized, controlled trials. N Engl J Med 1997; 337: 536–42 total hip arthroplasty. Int J Artif Organs 1999; 22:47–51 41 Macario A, Weinger M, Carney S, Kim A. Which clinical anesthe- 20 Dauphin A, Raymer KE, Stanton EB, Fuller HD. Comparison of sia outcomes are important to avoid? The perspective of patients. general anesthesia with and without lumbar epidural for total hip Anesth Analg 1999; 89: 652–8 arthroplasty: effects of epidural block on hip arthroplasty. J Clin 42 Mariano ER, Ilfeld BM, Neal JM. ‘Going fishing’—the practice of Anesth 1997; 9: 200–3 reporting secondary outcomes as separate studies. Reg Anesth 21 Eroglu A, Uzunlar H, Erciyes N. Comparison of hypotensive epi- Pain Med 2007; 32: 183–5 dural anesthesia and hypotensive total intravenous anesthesia on 43 Matthey PW, Finegan BA, Finucane BT. The public’s fears about intraoperative blood loss during total hip replacement. J Clin and perceptions of regional anesthesia. Reg Anesth Pain Med 2004; Anesth 2005; 17: 420–5 29: 96–101 22 Fanelli G, Casati A, Garancini P, Torri G. Nerve stimulator and mul- 44 Mauermann WJ, Shilling AM, Zuo Z. A comparison of neuraxial tiple injection technique for upper and lower limb blockade: failure block versus general anesthesia for elective total hip replacement: rate, patient acceptance, and neurologic complications. Study a meta-analysis. Anesth Analg 2006; 103: 1018–25 Group on Regional Anesthesia. Anesth Analg 1999; 88:847–52 45 Maurer K, Bonvini JM, Ekatodramis G, Serena S, Borgeat A. 23 Fischer HB, Simanski CJ. A procedure-specific systematic review Continuous spinal anesthesia/analgesia vs. single-shot spinal and consensus recommendations for analgesia after total hip anesthesia with patient-controlled analgesia for elective hip replacement. Anaesthesia 2005; 60: 1189–202 arthroplasty. Acta Anaesthesiol Scand 2003; 47: 878–83 24 Fournier R, Weber A, Gamulin Z. Intrathecal sufentanil is more 46 McKenzie PJ. Pitfalls in performing meta-analysis: II. Anesthesiology potent than intravenous for postoperative analgesia after total-hip 1993; 79: 406–8 replacement. Reg Anesth Pain Med 2005; 30: 249–54 47 McKenzie PJ, Wishart HY, Smith G. Long-term outcome after 25 Geerts WH, Heit JA, Clagett GP, et al. Prevention of venous repair of fractured neck of femur. Comparison of subarachnoid thromboembolism. Chest 2001; 119: 132S–75S and general anaesthesia. Br J Anaesth 1984; 56: 581–5 26 Gonano C, Leitgeb U, Sitzwohl C, Ihra G, Weinstabl C, Kettner 48 McLaren AD, Stockwell MC, Reid VT. Anaesthetic techniques for SC. Spinal versus general anesthesia for orthopedic surgical correction of fractured neck of femur. A comparative surgery: anesthesia drug and supply costs. Anesth Analg 2006; study of spinal and general anaesthesia in the elderly. Anaesthesia 102: 524–9 1978; 33:10–4 27 Grossi P, Urmey WF. Peripheral nerve blocks for anaesthesia 49 Moen V, Dahlgren N, Irestedt L. Severe neurological compli- and postoperative analgesia. Curr Opin Anaesthesiol 2003; 16: cations after central neuraxial blockades in Sweden 1990–1999. 493–501 Anesthesiology 2004; 101: 950–9 28 Guay J. The effect of neuraxial blocks on surgical blood loss and 50 Moiniche S, Hjortso NC, Hansen BL, et al. The effect of balanced blood transfusion requirements: a meta-analysis. J Clin Anesth analgesia on early convalescence after major orthopaedic 2006; 18: 124–8 surgery. Acta Anaesthesiol Scand 1994; 38: 328–35 29 Haas SB. Effects of epidural anesthesia on incidence of venous 51 Nielsen KC, Steele SM. Outcome after regional anaesthesia in thromboembolism following joint replacement. Orthopedics 1994; the ambulatory setting—is it really worth it? Best Pract Res Clin 17:18–20 Anaesthesiol 2002; 16: 145–57

344 GA vs RA for THA

52 Oldman M, McCartney CJ, Leung A, et al. A survey of orthopedic 66 Thorburn J, Louden JR, Vallance R. Spinal and general anaesthesia surgeons’ attitudes and knowledge regarding regional anesthesia. in total hip replacement: frequency of deep vein thrombosis. Br J Anesth Analg 2004; 98: 1486–90 Anaesth 1980; 52: 1117–21 53 Pagnano MW, Hebl J, Horlocker T. Assuring a painless total hip 67 Twyman R, Kirwan T, Fennelly M. Blood loss reduced during hip arthroplasty: a multimodal approach emphasizing peripheral arthroplasty by lumbar plexus block. J Bone Joint Surg Br 1990; 72: nerve blocks. J Arthroplasty 2006; 21:80–4 770–1 54 Pavlin DJ, Rapp SE, Polissar NL, Malmgren JA, Koerschgen M, 68 Uhrbrand B, Jensen TT, Bendixen DK, Hartmann-Andersen JF. Keyes H. Factors affecting discharge time in adult outpatients. Perioperative analgesia by 3-in-one block in total hip arthroplasty. Anesth Analg 1998; 87: 816–26 Prospective randomized blind study. Acta Orthop Belg 1992; 58: 55 Planes A, Vochelle N, Fagola M, Feret J, Bellaud M. Prevention of 417–9 deep vein thrombosis after total hip replacement. The effect of 69 Urwin SC, Parker MJ, Griffiths R. General versus regional anaes- low-molecular-weight heparin with spinal and general anaesthesia. thesia for hip fracture surgery: a meta-analysis of randomized J Bone Joint Surg Br 1991; 73: 418–22 trials. Br J Anaesth 2000; 84: 450–5 56 Rodgers A, Walker N, Schug S, et al. Reduction of postoperative 70 Van Gessel E, Forster A, Gamulin Z. A prospective study of the mortality and morbidity with epidural or spinal anaesthesia: feasibility of continuous spinal anesthesia in a university hospital. results from overview of randomised trials. Br Med J 2000; 321: Anesth Analg 1995; 80: 880–5 1493 71 Viscusi ER, Martin G, Hartrick CT, Singla N, Manvelian G. 57 Salo M, Nissila M. Cell-mediated and humoral immune responses Forty-eight hours of postoperative pain relief after total hip to total hip replacement under spinal or general anaesthesia. Acta arthroplasty with a novel, extended-release epidural morphine Anaesthesiol Scand 1990; 34: 241–8 formulation. Anesthesiology 2005; 102: 1014–22 58 Sharrock NE, Cazan MG, Hargett MJ, Williams-Russo P, Wilson 72 Watson MW, Mitra D, McLintock TC, Grant SA. Continuous PD, Jr. Changes in mortality after total hip and knee arthroplasty versus single-injection lumbar plexus blocks: comparison of the over a ten-year period. Anesth Analg 1995; 80: 242–8 effects on morphine use and early recovery after total knee 59 Siddiqui ZI, Cepeda MS, Denman W, Schumann R, Carr DB. arthroplasty. Reg Anesth Pain Med 2005; 30: 541–7 Continuous lumbar plexus block provides improved analgesia 73 Weller R, Rosenblum M, Conard P, Gross JB. Comparison of epi- with fewer side effects compared with systemic opioids after hip dural and patient-controlled intravenous morphine following joint arthroplasty: a randomized controlled trial. Reg Anesth Pain Med replacement surgery. Can J Anaesth 1991; 38: 582–6 2007; 32: 393–8 74 Williams BA, Kentor ML, Vogt MT, et al. Economics of nerve 60 Singelyn FJ, Deyaert M, Joris D, Pendeville E, Gouverneur JM. block pain management after anterior cruciate ligament recon- Effects of intravenous patient-controlled analgesia with morphine, struction: potential hospital cost savings via associated post- continuous epidural analgesia, and continuous three-in-one block anesthesia care unit bypass and same-day discharge. Anesthesiology on postoperative pain and knee rehabilitation after unilateral 2004; 100: 697–706 total knee arthroplasty. Anesth Analg 1998; 87:88–92 75 Williams BA, Kentor ML, Williams JP, et al. Process analysis in 61 Singelyn FJ, Ferrant T, Malisse MF, Joris D. Effects of intravenous outpatient knee surgery: effects of regional and general anesthe- patient-controlled analgesia with morphine, continuous epidural sia on anesthesia-controlled time. Anesthesiology 2000; 93: analgesia, and continuous femoral nerve sheath block on rehabili- 529–38 tation after unilateral total-hip arthroplasty. Reg Anesth Pain Med 76 Williams-Russo P, Sharrock NE, Haas SB, et al. Randomized trial 2005; 30: 452–7 of epidural versus general anesthesia: outcomes after primary 62 Smith MP, Sandberg WS, Foss J, et al. High-throughput operating total knee replacement. Clin Orthop Relat Res 1996; 331: 199–208 room system for joint arthroplasties durably outperforms routine 77 Wu CL, Hurley RW, Anderson GF, Herbert R, Rowlingson AJ, processes. Anesthesiology 2008; 109:25–35 Fleisher LA. Effect of postoperative epidural analgesia on morbid- 63 Steele SM, Klein SM, D’Ercole FJ, Greengrass RA, Gleason D. A new ity and mortality following surgery in medicare patients. Reg continuous catheter delivery system. Anesth Analg 1998; 87:228 Anesth Pain Med 2004; 29: 525–33 64 Stevens M, Harrison G, McGrail M. A modified fascia iliaca com- 78 Wulf H, Biscoping J, Beland B, Bachmann-Mennenga B, Motsch J. partment block has significant morphine-sparing effect after total Ropivacaine epidural anesthesia and analgesia versus general hip arthroplasty. Anaesth Intensive Care 2007; 35: 949–52 anesthesia and intravenous patient-controlled analgesia with mor- 65 Stevens RD, Van Gessel E, Flory N, Fournier R, Gamulin Z. phine in the perioperative management of hip replacement. Lumbar plexus block reduces pain and blood loss associated with Ropivacaine Hip Replacement Multicenter Study Group. Anesth total hip arthroplasty. Anesthesiology 2000; 93: 115–21 Analg 1999; 89: 111–6

345 “I would have everie man write what he knowes and no more.”–Montaigne BRITISH JOURNAL OF ANAESTHESIA Volume 101, Number 3, September 2008

British Journal of Anaesthesia 101 (3): 291–3 (2008) doi:10.1093/bja/aen232 Editorial I

Haematoma and abscess after epidural analgesia

The survey by Meikle and colleagues1 published in this haematoma appear frequently in such cases. In contrast to month’s British Journal of Anaesthesia highlights the epidural haematomas, which usually present during the epi- uncommon but catastrophic complication of epidural hae- dural infusion,7 abscesses often present late and after the matoma. Although the incidence of symptomatic epidural patient has left hospital,3 8–10 but for those which present haematoma may appear small, amounting, they suggest, to before discharge and for the large majority of epidural hae- one case every 2 yr in the UK, it appears that the use of matomas, our experience is that there is usually a significant epidural infusions is increasing, alongside an increase in delay before diagnosis because junior surgical trainees, prophylactic anticoagulation and low-dependency care of inexperienced anaesthetists, or both wrongly attribute the patients with indwelling epidural .2 The results of onset of weakness and increasing numbness to the effects this survey indicate that the incidence of epidural haema- of the local anaesthetic. The infusion is usually continued toma may, in fact, be significantly higher than suggested for hours,7 and sometimes for several days, before the by reported cases, and the ongoing national audit by the opinion of an experienced anaesthetist is sought, resulting Royal College of Anaesthetists may give us a better indi- in permanent neurological damage. It is impossible to cation of the true risk of this complication. educate all trainee surgeons and nurses to recognize the Similar catastrophic injury may result also from epidural significance of these clinical signs, and we agree that strict abscess, which has been reported in one UK hospital to protocols offer the best to early diagnosis, investi- occur with a frequency of 1 in 800, when epidural catheters gation, and treatment. were inserted for postoperative pain management.3 Our Although reporting-bias affects the incidence calculated experience from forensic practice is that Meikle (and from reported cases, examination using closed-claim previous authors) underestimate the incidence of epidural analysis has similar flaws. Closed-claim analysis allows a haematoma considerably, but that the experience of Phillips glimpse of cases that would usually not be reported. and colleagues overestimates the risk of abscess nationally. However, the technique must also underestimate the true The incidence of epidural haematoma and abscess probably incidence of complications because not all patients who varies in different patient populations; Scott and Hibbard4 suffer complications sue. Patients who decide to sue have reported two haematomas and one abscess after 505 000 usually suffered a significant and long-lasting injury, and epidural blocks in obstetric practice. The incidence in have been able to secure funding for their claim. In higher risk patients is likely to be greater. For example, addition, the true incidence of a complication which leads Ngan Kee and colleagues5 reported a higher incidence after to litigation cannot be estimated accurately from the thoracic epidural analgesia and Okano and colleagues6 experience of two individuals. The number of other noted 11 out of 30 patients with epidural abscesses had an patients who have entered litigation proceedings in relation underlying illness or were receiving therapy. to epidural haematoma or abscess in the UK in the last Two of us (A.R.A. and J.G.H.) have encountered, in 10–11 yr is unknown. medicolegal practice, 11 cases of epidural haematoma in an Another method of estimating the extent of damage 11 yr period (the events occurred between 1997 and 2007) caused by specific anaesthetic techniques is to consider the and eight cases of epidural abscess in a 10 yr period costs of claims handled by large insurance or indemnity (between 1995 and 2004). Failure to discontinue epidural organizations. Between April 1995 and October 2005, the infusions after the presentation of new neurological signs Clinical Negligence Scheme for Trusts (CNST), which (especially leg weakness) and failure to recognize the deals with litigation for all NHS hospitals in England, urgency of diagnosis and surgery after suspicion of epidural handled 251 claims associated with epidural blocks.11

# The Board of Management and Trustees of the British Journal of Anaesthesia 2008. All rights reserved. For Permissions, please e-mail: [email protected] Editorial I

The claims had a total value of £32 346 737 (an average prevent the development of permanent and disabling of £128 871 each). There were lower incidences of brain neurological injury in a considerably larger number of damage and fatality in claims related to epidural block patients than they suggest. than in those associated with general anaesthesia. However, there were higher incidences of nerve damage, N. M. Bedforth1 paraplegia, partial paralysis, spinal damage, and unnecess- A. R. Aitkenhead2 ary pain. We have been unable to establish how many of J. G. Hardman2* these claims were related to delay in recognizing the 1Department of Anaesthesia symptoms and signs of epidural haematoma or abscess, Queen’s Medical Centre but the size of the settlements suggests that the proportion Nottingham NG7 2UH of these 251 patients suffering debilitating neurological UK injury was not inconsiderable. 2 1 University Department of Anaesthesia Meikle and colleagues recommend that patients should Queen’s Medical Centre receive neurological observations at least every 4 h and Nottingham NG7 2UH that these observations should continue for at least 24 h UK after removal of the epidural catheter. This recommen- dation seems valid in view of the previously reported *E-mail: [email protected] cases of haematoma and abscess formation after catheter removal.5812Every department should have readily avail- able written guidelines regarding the use of neuraxial tech- References niques in patients with potentially altered coagulation.13 1 Meikle J, Bird S, Nightingale JJ, White N. Detection and manage- The authors also recommend the cessation of the epidural ment of epidural haematomas related to anaesthesia in the UK: infusion after the presentation of new neurological signs, a national survey of current practice. Br J Anaesth 2008; 101: with suspicion of epidural haematoma if these signs do 400–4 2 Burstal R, Wegener F,Hayes C, Lantry G. Epidural analgesia: prospec- not resolve. The authors do not recommend a minimum tive audit of 1062 patients. Anaesth Intensive Care 1998; 26:165–72 time interval between the suspicion of haematoma or the 3 Phillips JM, Stedeford JC, Hartsilver E, Roberts C. Epidural abscess cessation of the infusion and MRI scanning; we suggest complicating insertion of epidural catheters. Br J Anaesth 2002; 89: that no more than 4 h should elapse between the onset of 778–82 new neurological signs and MRI scanning, and this scan 4 Scott DB, Hibbard BM. Serious non-fatal complications associated (and ideally, the patient) should be assessed by an expert. with extradural block in obstetric practice. Br J Anaesth 1990; 64: Should there be a delay in stopping the epidural infusion 537–41 5 Ngan Kee WD, Jones MR, Thomas P, Worth RJ. Extradural abscess after the presentation of new signs, then MRI scanning complicating extradural anaesthesia for . Br J may need to take place before the local anaesthetic effect Anaesth 1992; 69: 647–52 of the epidural may be expected to resolve. 6 Okano K, Kondo H, Tsuchiya R, Naruke T, Sato M, Yokoyama R. In hospitals without expertise to carry out surgical Spinal epidural abscess associated with epidural catheterization: decompression, protocols and procedures need to be in report of a case and a review of the literature. Jpn J Clin Oncol place to ensure that patients are transferred to a unit where 1999; 29:49–52 surgery can be performed within 12 h of the onset of 7 Christie IW,McCabe S. Major complications of epidural analgesia after surgery: results of a six-year survey. Anaesthesia 2007; 62: 335–41 weakness or increasing numbness to optimize the chance 8 Hearn M, Roberts C. Epidural abscess complicating insertion of of recovery. In the absence of focal neurological signs, epidural catheters. Br J Anaesth 2003; 90: 706–7 conservative management of epidural abscess may be 9 Strong WE. Epidural abscess associated with epidural catheteriza- successful14 –17 but frequently urgent surgical evacuation tion: a rare event? Report of two cases with markedly delayed is required.918–22 presentation. Anesthesiology 1991; 74: 943–6 Owing to the low incidence of epidural haematoma, we 10 Sowter MC, Burgess NA, Woodsford PV, Lewis MH. Delayed will never be in a position to introduce true, evidence- presentation of an extradural abscess complicating thoracic extra- dural analgesia. Br J Anaesth 1992; 68: 103–5 based practice. Even if studies of sufficient size are 11 Catchpole K, Bell D, Johnson S, Boult M. Reviewing the evidence conducted, their relevance will be limited by constant of patient safety incidents in anaesthetics. Internal Report. The evolution in practice. Thus, we are obliged to apply National Patient Safety Agency, 2006 common-sense and learning to help our patients avoid 12 Yin B, Barratt SM, Power I, Percy J. Epidural haematoma after a life-damaging event. The relative rarity of epidural hae- removal of an epidural catheter in a patient receiving high-dose matoma and abscess means that expensive and laborious enoxaparin. Br J Anaesth 1999; 82: 288–90 additions to current practice are not appropriate, but the 13 Prophylaxis of venous thromboembolism; spinals and epidurals. SIGN publication No 62, October 2002. ISBN 1899893 03 simple measures suggested by Meikle and colleagues need 2. Available from http://www.sign.ac.uk/guidelines not be expensive or laborious, and we wholly commend 14 Dysart RH, Balakrishnan V. Conservative management of extra- them to practising anaesthetists in the UK. It is likely that dural abscess complicating spinal-extradural anaesthesia for the introduction of strict protocols would minimize or Caesarean section. Br J Anaesth 1997; 78: 591–3

292 Editorial II

15 Leys D, Lesoin F, Viaud C, et al. Decreased morbidity from acute 19 Del Curling O, Jr, Gower DJ, McWhorter JM. Changing concepts bacterial spinal epidural abscesses using computed tomography in spinal epidural abscess: a report of 29 cases. Neurosurgery and nonsurgical treatment in selected patients. Ann Neurol 1985; 1990; 27: 185–92 17: 350–5 20 Hlavin ML, Kaminski HJ, Ross JS, Ganz E. Spinal epidural abscess: 16 Hanigan WC, Asner NG, Elwood PW. Magnetic resonance a ten-year perspective. Neurosurgery 1990; 27: 177–84 imaging and the nonoperative treatment of spinal epidural 21 Lange M, Tiecks F, Schielke E, Yousry T, Haberl R, Oeckler R. abscess. Surg Neurol 1990; 34: 408–13 Diagnosis and results of different treatment regimens in 17 Mampalam TJ, Rosegay H, Andrews BT, Rosenblum ML, Pitts LH. patients with spinal abscesses. Acta Neurochir (Wien) 1993; 125: Nonoperative treatment of spinal epidural infections. J Neurosurg 105–14 1989; 71: 208–10 22 Danner RL, Hartman BJ. Update on spinal epidural abscess: 35 18 Borum SE, McLeskey CH, Williamson JB, Harris FS, Knight AB. cases and review of the literature. Rev Infect Dis 1987; 9: 265–74 Epidural abscess after obstetric epidural analgesia. Anesthesiology 1995; 82: 1523–6

British Journal of Anaesthesia 101 (3): 293–5 (2008) doi:10.1093/bja/aen233 Editorial II

NICE and warm

Inadvertent perioperative hypothermia, defined as core vague and giving flexibility at the expense of clear body temperature 36.08C, is a common consequence of guidance.8 anaesthesia. Its adverse effects are well known to anaes- Recognizing the significance of inadvertent periopera- thetists and include greater intraoperative blood loss and tive hypothermia and the deficiencies in current practice in consequent blood transfusion.1 After operation, inadvertent the UK, the National Institute for Clinical Excellence perioperative hypothermia can lead to an increased rate of (NICE) convened a guideline development group to wound infection,2 morbid cardiac events,3 and pressure address the issue. This culmination of the group’s work sores,4 and also a longer stay in both recovery and came with the publication of the ‘Management of inadver- hospital.5 These are apart from the subjective discomfort tent perioperative hypothermia in adults’ guideline.10 and wound pain which cold and shivering may cause the The guidance is divided into the pre-, intra-, and post- patient. Significantly, maintaining normothermia periopera- operative phases. Before operation, the key recommen- tively can modify these adverse effects. dations are that a formal assessment of the risk of Despite this knowledge, implementation of warming hypothermia should be undertaken for each patient and strategies remains patchy. An audit in the hospital of one that patients themselves should be empowered by being of the authors (C.M.H.) indicated that there is an incidence given information that will help them minimize that risk. of inadvertent perioperative hypothermia in the region of Another important element is that the temperature should 20% and that there is inconsistency in the methods of be measured in the hour before surgery. Should it be warming used. There are no active temperature manage- ,36.08C, unless the operation is life or limb saving, active ment protocols and, as with anything that may cost money, warming should be initiated until such time as the patient there is resistance to more aggressive prevention of inad- is normothermic. vertent perioperative hypothermia on economic grounds. Intraoperatively, the recommendations are that forced-air In the USA, where there are guidelines,6 compliance warming is commenced as early as possible, preferably in remains poor. It has been suggested that there are a the anaesthetic room, for any patient having surgery with number of factors contributing to this: a misguided belief an anaesthetic time (i.e. from first anaesthetic intervention that forced-air warming can increase the rates of infection,7 to arrival in recovery) of .30 min, or who has two or surgeons’ complaints of discomfort, inconsistent monitoring more risk factors for inadvertent perioperative hypother- (hindered by the inconsistency between different ther- mia. I.V. fluids should be warmed when .500 ml is to be mometers and sites of measurement), and a simple lack of given.11 12 These recommendations therefore encompass appreciation of the causes and consequences of inadvertent the majority of operations and infusions. perioperative hypothermia.8 Additionally, even where Monitoring is an integral part of perioperative thermal there are standards such as those of the American Society management and one that remains neglected.13 The guide of Anesthesiologists (ASA),9 they are criticized for being recommends that core temperature should be recorded at

293 Anesthesiology 2005; 103:126–9 © 2005 American Society of Anesthesiologists, Inc. Lippincott Williams & Wilkins, Inc. A Comparative Study of Sequential Epidural Technique and Continuous Epidural Infusion Kenichi Ueda, M.D.,* Wasa Ueda, M.D., Ph.D.,† Masanobu Manabe, M.D., Ph.D.‡

Background: In this randomized, double-blind study, the infusion (CEI) to achieve a similar quality of epidural authors compared the effectiveness of a sequential epidural analgesia.5 Further, Sia and Chong6 found that PCEA bolus (SEB) technique versus a standard continuous epidural could provide satisfactory epidural analgesia even when infusion (CEI) technique of local anesthetic delivery. Both tech- niques used the same hourly dose of local anesthetic. the background infusion was eliminated. These findings Methods: Sixteen gynecologic patients undergoing abdomi- caused us to hypothesize that an intermittent bolus ad- nal surgery received postoperative epidural analgesia using ministration with regular intervals independent of pa- 0.75% ropivacaine at a dose of 22.5 mg (3 ml) per hour. Patients tient demand could be superior to CEI for producing were randomly assigned to one of two groups. In the SEB group epidural sensory block with better quality. To test this patients received one third of the hourly dose every 20 ,(8 ؍ n) the hourly dose was new epidural delivery mode, we provided postoperative ,(8 ؍ min as a bolus. In the CEI group (n administered as a continuous infusion. Analgesia was assessed epidural analgesia by two different techniques, using by rest pain scored by a visual analog scale and pinprick to identical hourly doses of local anesthetic. One group of determine the number of separately blocked spinal segments patients received the standard CEI technique. The other on each side of the body. Doses of rescue medication for pain group received a sequential epidural bolus technique were also recorded. Results: The median number of blocked spinal segments was (SEB) in which one third of the hourly dose was admin- 19.5 (range, 18–24) in the SEB group and 11.5 (range, 10–18) in istered every 20 min. the CEI group (P < 0.001). The median difference in the number of blocked segments between the right and left sides was 0 (range, 0–1) in the SEB group and 2 (range, 0–6) in the CEI group (P < 0.04). No patients in the SEB group but one patient Materials and Methods in the CEI group required rescue medication for pain. The visual analog scale pain score was 0 in both groups except for one The study protocol was approved by the Ethics Com- patient in the CEI group during the study period. mittee of the Kochi Medical School (Kochi, Japan) and Conclusion: The SEB technique with ropivacaine provides written informed consent was obtained from each pa- superior epidural block compared with an identical hourly dose tient. The study group consisted of 16 gynecologic pa- administered as a continuous infusion. tients with American Society of Anesthesiologists physi- cal status I or II who were scheduled to undergo lower CONTINUOUS epidural analgesia now plays an impor- abdominal surgery. None of the patients had previous 1 tant role in postoperative pain control. In the standard experience with epidural analgesia. continuous epidural infusion technique (CEI), the initial All patients received standardized anesthetic care by dose establishes the requisite extent of analgesia, and the same anesthesiologist for the surgery. Oral diazepam, then continuous infusion preserves it. Over time, how- 10 mg, was administered 1 h before surgery. The epi- ever, some patients experience diminution of analgesia, dural catheter (18 gauge; Portex, Kent, United Kingdom) characterized by a reduction in the number of blocked was placed before induction of anesthesia, with the spinal segments and development of unequal right and patient in the right lateral decubitus position, at the left blockade. In such cases, additional top-up doses are T11–T12 or T12–L1 spinal interspace using the parame- necessary. The development of patient-controlled epi- dian approach. General anesthesia included mask induc- dural analgesia (PCEA) permitted patients to superim- tion with sevoflurane, , and mainte- pose a limited volume of bolus dosing on continuous nance with 1.0% end-tidal concentration of isoflurane in 2 infusion. PCEA was first applied to obstetrics and was oxygen. Muscle relaxation was facilitated with vecuro- then found to be effective in postoperative pain control nium. No opioids were administered. After induction of 3,4 that requires a steady level of analgesia. general anesthesia, an initial epidural dose of 8 ml lido- Interestingly, patients with PCEA required less local caine, 2%, with 1:200,000 epinephrine was administered. anesthetic than did patients with continuous epidural Thereafter, 1 ml , 2%, with 1:200,000 epinephrine was given as a bolus every 10 min until the end of the surgery. If a patient showed any reaction to the surgical * Assistant Professor, ‡ Professor, Department of Anesthesia, † Professor, Department of Perioperative Care and Medicine, Kochi Medical School. maneuvers that required more than 1.2% end-tidal concen- Received from the Department of Anesthesia, Kochi Medical School, Kochi, tration of isoflurane to control, we eliminated the case from Japan. Submitted for publication October 4, 2004. Accepted for publication the study because of failed epidural block. February 28, 2005. Support was provided solely from institutional and/or depart- mental sources. At the end of surgery, patients were given either SEB or Address reprint requests to Dr. Ueda: Department of Anesthesiology, Kochi CEI according to a computerized random-number gener- Medical School, Nankoku, Kochi, Japan 783-8505. Address electronic mail to: [email protected]. Individual article reprints may be purchased through the ator. Both groups received 0.75% ropivacaine at a dose Journal Web site, www.anesthesiology.org. of 3 ml/h. In the SEB group (n ϭ 8), ropivacaine was

Anesthesiology, V 103, No 1, Jul 2005 126 COMPARISON OF EPIDURAL DOSING METHODS 127 given as a sequential epidural bolus of 1 ml every 20 min. Table 1. Demographic Data of the Patients Studied In the CEI group (n ϭ 8), ropivacaine was given as a SEB Group CEI Group continuous epidural infusion. Patients were blinded to the drug administration technique. On arriving at the n88 postanesthesia care unit, patients were asked to rate Age, yr 33.5 (21–48) 33.5 (20–50) Height, cm 154 (150–159) 156 (150–162) their pain experience on a visual analog scale (VAS). Body weight, kg 54 (45–68) 53.5 (48–65) Only rest pain was assessed by nurses who were not ASA PS, I/II 7/1 7/1 aware of this study. The patients stayed in the postanes- thesia care unit for3htoconfirm that the pain control Values are presented as mean or n, with range in parentheses. was adequate. In the ward, the patients were allowed to ASA PS ϭ American Society of Anesthesiologists physical status. take diclofenac as rescue medication from ical status between the two groups. Surgical epidural an independent nurse. block was successful in all patients. No patient in either An independent nurse in the ward assessed vital signs group had hypotension, bradycardia, or bradypnea dur- and rest pain score by VAS every 4 h. If a patient expe- ing the study. rienced hypotension, bradycardia, or bradypnea, epi- Figure 1 shows the extent of sensory blockade ob- dural infusion was discontinued and the acute pain ser- tained after the 15 h of epidural infusion in the study. vice team was called for assessment. Hypotension was SEB produced a larger band of sensory blockade. The defined as a systolic blood pressure of less than 90 median number of blocked segments in the mmHg, bradycardia was defined as a heart rate of less SEB group was 19.5 (range, 18–24), and that in the CEI than 50 beats/min, and bradypnea was defined as a group was 11.5 (range, 10–18) (P Ͻ 0.001). SEB also respiratory rate of less than 10 breaths/min. significantly improved equal right and left distribution of Anesthesiologists who were unaware of the infusion sensory blockade. The median difference in the number mode visited the patients after the 15 h of ropivacaine of blocked spinal segments between the right and left infusion. Using a chart,7 the extent of anal- sides, expressed by spinal nerve segment, in the SEB gesia was judged by the number of blocked spinal seg- group was 0 (range, 0–1), and that in the CEI group was ments as determined by the pinprick method. 2 (range, 0–6) (P Ͻ 0.04). To produce the SEB mode, we modified the program One patient in the CEI group required rescue medica- of the standard (SP-80RS; Nipro, Osaka, tion for left lower abdominal pain. In this patient, there Japan) so that the pump divided the hourly dose into was a significant difference between the right and left three parts and infused each every 20 min as a bolus side analgesia, with segments T8–L2 blocked on the (bolus rate of 500 ml/h).

Statistics A pilot study with 12 patients showed the number of blocked spinal nerve segments to be 21 Ϯ 3 (mean Ϯ SD) in the SEB group and 13 Ϯ 3 in the CEI group. The sample size to give a power of 0.8 with an ␣ of 0.05 was 4 when the expected difference was 8 Ϯ 3. The difference in the number of blocked segments be- tween the right and left sides was 0 Ϯ 0intheSEB group and 3 Ϯ 2 in the CEI group. The sample size to give a power of 0.8 with an ␣ of 0.05 was 7 when the expected difference was 2 Ϯ 1. Therefore, we decided to use a sample size of 16. The measured data were rejected by normality test, so we compared the two groups by Mann–Whitney rank sum test. Statistical analysis with SigmaStat 3.0 (SPSS Inc., Chicago, IL) yielded a significant probability value of less than 0.05. Fig. 1. Extent of sensory blockade after 15 h of epidural infu- ,(8 ؍ Continuous data are expressed as medians, with the sion. In the sequential epidural bolus (SEB) group (n ranges given in parentheses. 0.75% ropivacaine was given as a sequential epidural bolus of 1 ml every 20 min. In the continuous epidural infusion (CEI) ropivacaine was given as a continuous %0.75 ,(8 ؍ group (n epidural infusion. The median number of blocked spinal seg- Results ments was 19.5 (range, 18–24) in the SEB group and 11.5 (range, 10–18) in the CEI group (P < 0.001). The median difference in the number of blocked segments between the right and left Table 1 shows the demographic data of the patients sides was 0 (range, 0–1) in the SEB group and 2 (range, 0–6) in studied. There was no statistical difference in the phys- the CEI group (P < 0.04).

Anesthesiology, V 103, No 1, Jul 2005 128 UEDA ET AL.

Fig. 2. Spread of contrast medium after injection of 1 ml iotrolan, 240 mg I/ml. Epidurograms A and B were taken from different patients to locate the epidural catheters, which were placed for chronic pain management. The length of the ar- row represents right- or left-sided spread of contrast medium. (Courtesy of Ma- sataka Yokoyama, M.D., Department of Anesthesiology, School of Medicine, Okayama University, Okayama, Japan.)

right side but only T9–T11 on the left side. Accordingly, block.10 If this occurs, a larger dose of local anesthetic is the pain score at rest by VAS was 0 in both groups, required to restore analgesia. If instead the same dose of except for one patient in the CEI group who scored 3 as anesthetic is used before decay of analgesia, the extent the maximum VAS score during the study. of analgesia may increase.11 In the current SEB, the bolus administration might have occurred at regular intervals before decay of sensory neural block, thereby increasing Discussion the number of sensory-blocked spinal segments. In this study, local anesthetic administered as a sequen- Ropivacaine, 0.75%, produced profound neural block tial epidural bolus (SEB) provided greater longitudinal sufficient to control pain without the help of opioids in extension of sensory blockade than continuous epidural the current study. Epidural analgesia has commonly infusion (CEI). The SEB technique also improved equal been induced with local anesthetic combined with opi- right and left distribution of sensory blockade compared oids to increase the quality of analgesia. Adverse effects with CEI. We propose some possible explanations. of opioids, however, have the potential to cause compli- First, the bolus technique may have a different mode of cations.12 To avoid such adverse effects and also to spread through the as compared with clearly define the range of sensory neural block, we continuous infusion. It is known that some of the epi- performed epidural block exclusively by local anes- dural local anesthetic is captured by perineural tissue or thetic. The concentration was decided according to the removed by blood vessels, making it unavailable for recommendation by Sakura et al.13; concentration deter- 8 nerve block. A bolus of local anesthetic likely spreads mines the intensity of sensory block such that a high further longitudinally and bilaterally before such capture concentration and a small volume, rather than a low occurs than does continuous infusion. concentration and a high volume, are suitable when It seems unlikely that only a 1-ml bolus of local anes- profound neural blockade is required. The volume given thetic would make such a significant difference as com- to the current patients was to achieve sufficient analge- pared with continuous infusion. Nonetheless, 1 ml con- sia with CEI and thereby caused no difference in VAS trast medium (iotrolan, 240 mg I/ml) when given as a pain score between the SEB and CEI groups. Because the bolus can clearly appreciate an extended area in the range of blocked segments was so extended with SEB, it epidural space (fig. 2). This is impossible by continuous infusion of the medium. Yokoyama et al.9 showed that may be possible to reduce the dose of the anesthetic in the behavior of the local anesthetic and contrast medium SEB. Further studies are needed to establish the optimal inside the epidural space were well correlated. There- timing, concentration, and volume of anesthetic for SEB fore, bolus dosing has the potential to increase the in various conditions. spread of local anesthetic in the epidural space, and a There are limitations to our study design. First, we bolus as small as 1 ml, as used in the current study, might evaluated the epidural-blocked segments at only a single have contributed to the success of SEB. data point (15 h). We did not study a longer duration of Second, proper timing of epidural dosing is key to infusion. This may have been insufficient to ascertain the maintain effective analgesia. If diminution of the epi- effects of SEB. Secondary, the lidocaine used for surgical dural blockade occurs, sensory neural input to the spinal anesthesia could have affected our results. We believe cord has been shown to accelerate the decline of the this is unlikely because effective surgical analgesia by 2%

Anesthesiology, V 103, No 1, Jul 2005 COMPARISON OF EPIDURAL DOSING METHODS 129 lidocaine with epinephrine lasts for 2–3 h at the most.14 esthesia, University of Iowa Hospitals and Clinics, Iowa City, Iowa), for helpful comments in the preparation of the manuscript. After 15 h, the residual effect of lidocaine is negligible. Another issue is whether SEB is economical. SEB re- quires a special infusion pump. To produce the SEB References mode, we modified the operating program of a comput- 1. Block BM, Lieu SS, Rowlingson AJ, Cowan AR, Cowan Jr, JA Wu CL: Efficacy er-controlled pump for this study. Multipurpose of postoperative epidural analgesia. JAMA 2003; 290:2455–63 2. Gambling DR, Yu P, Cole C, McMorland GH, Palmer L: A comparative study infusion pumps equipped with various infusion modes of patient controlled epidural analgesia (PCEA) and continuous infusion epidural currently available allow SEB in clinical practice. Such analgesia (CIEA) during labour. Can J Anesth 1988; 35:249–54 3. Silvasti M, Pitkaenen M: Patient-controlled epidural analgesia versus contin- pumps are more expensive than pneumatic infusion uous epidural analgesia after total knee arthroplasty. Acta Anaesthesiol Scand pumps. However, an electrical pump uses a disposable 2001; 45:471–6 4. Standl T, Burmeister M-A, Ohnesorge H, Wilhelm S, Striepke M, Gottschalk plastic reservoir with tubing, which is less expensive A, Horn E-P, Esch JS: Patient-controlled epidural analgesia reduces analgesic and, as a medical waste product, less bulky than a dis- requirements compared to continuous epidural infusion after major abdominal surgery. Can J Anesth 2003; 50:258–64 posable pneumatic infuser. As a whole, the expense 5. Antok E, Bordet F, Duflo F, Lansiaux S, Combet S, Taylor P, Pouyau A, associated with the infusion pump may not be a draw- Paturel B, James R, Allaouchiche B, Chassard D: Patient-controlled epidural analgesia versus continuous epidural infusion with ropivacaine for postoperative back of SEB. analgesia in children. Anesth Analg 2003; 97:1608–11 SEB would have the potential to replace CEI in various 6. Sia AT, Chong JL: Epidural 0.2% ropivacaine for labour analgesia: Parturient- controlled or continuous infusion? Anesth Intensive Care 1999; 27:154–7 situations. SEB can be applied to epidural anesthesia 7. Barash PG, Cullen BF, Stoelting RK: Clinical Anesthesia, 4th edition. Phila- during surgery. Also, it can be combined with PCEA. delphia, JB Lippincott, 2001, p 692 8. Bromage PR: Mechanism of action of extradural analgesia. Br J Anesth 1975; Still, the possibility of local anesthetic overdose by pro- 47:199–212 grammed dose administration necessitates careful selec- 9. Yokoyama M, Hanazaki M, Fujii H, Mizobuchi S, Nakatsuka H, Takahashi T, Matsumi M, Takeuchi M, Morita K: Correlation between the distribution of tion of dose and interval and continued assessment by contrast medium and the extent of blockade during epidural anesthesia. ANES- the anesthesiologist. THESIOLOGY 2004; 100:1504–10 10. Ueda W, Sagara Y, Hirakawa M: Role of afferent neural input in regression In conclusion, SEB improved the quality of sensory of sensory paralysis during epidural block. Anesth Analg 1992; 74:358–61 blockade as compared with continuous epidural block in 11. Cousins MJ, Bromage P: Epidural neural blockade, Neural Blockade, 2nd edition. Edited by Cousins MJ, Bridenbaugh PO. Philadelphia, JB Lippincott, this study. This infusion mode has the potential to in- 1988, pp 253–360 crease the extent of sensory blockade as well as to 12. Jorgensen H, Formsgaard JS, Dirks J, Wetterslev J, Andreasson B, Dahl JB: Effect of epidural bupivacaine vs combined epidural bupivacaine and morphine on decrease unilateral blockade, thereby improving the re- gastrointestinal function and pain after major gynecological surgery. Br J Anesth liability of epidural analgesia in postsurgical pain control. 2001; 87:727–32 13. Sakura S, Sumi M, Kushizaki H, Saito Y, Kosaka Y: Concentration of lidocaine affects intensity of sensory block during lumbar epidural anesthesia. The authors thank Masataka Yokoyama, M.D. (Associate Professor of Anesthe- Anesth Analg 1999; 88:123–7 siology, Okayama University School of Medicine, Okayama, Japan), for his help in 14. Camann W, Loferski B, Fanciullo G, Stone M, Datta S: Does epidural providing epidurographic pictures. The authors also thank William Hammonds, administration of butorphanol offer any clinical advantage over the intravenous M.D. (Professor), and Alan Ross, M.D. (Associate Professor, Department of An- route? ANESTHESIOLOGY 1992; 76:216–20

Anesthesiology, V 103, No 1, Jul 2005 Anesthesiology 2008; 109:664–74 Copyright © 2008, the American Society of Anesthesiologists, Inc. Lippincott Williams & Wilkins, Inc. Population Pharmacokinetic–Pharmacodynamic Modeling of Epidural Anesthesia Erik Olofsen, M.Sc.,* Antonius G. L. Burm, Ph.D.,† Mischa J. G. Simon, M.D., Ph.D.,‡ Bernadette Th. Veering, M.D., Ph.D.,‡ Jack W. van Kleef, M.D., Ph.D.,§ Albert Dahan, M.D., Ph.D.†

Background: In previous studies, the authors reported on the dent on pharmacokinetic factors. The rate of systemic absorption and disposition kinetics of and absorption of local anesthetics gives some indication of ropivacaine. The current study was designed to develop a pop- the relation between neural blockade and the amount of ulation pharmacokinetic–pharmacodynamic model capable of linking the kinetic data to the analgesic effects of these local drug remaining at or near the site of injection. In the past anesthetics (i.e., sensory neural blockade). two decades, we indirectly assessed epidural drug con- Methods: A disposition compartmental model was fitted to centrations by estimation of the time course of systemic concentration data of the intravenously administered deuteri- absorption from the epidural space in humans.1–5 um-labeled anesthetics, and a model consisting of two parallel In our previous studies, the pharmacokinetic data absorption compartments and the identified disposition compartments was fitted to concentration data of the con- were analyzed for each individual separately. However, comitantly epidurally administered unlabeled anesthetics. population pharmacokinetic–pharmacodynamic (PK- The epidural segments were modeled by individual central PD) analysis of epidural anesthesia is important because and peripheral absorption compartments and effect sites, it enables the description of both within- and between- which were fitted to the simultaneously acquired pinprick subject variability, enables the development of predic- data. A covariate model incorporated the effects of age. Results: The threshold for epidural anesthesia increased tive models, and, consequently, improves therapeutic 6 7 from the lower to the higher segments. The central effect com- outcome. Schnider et al. developed a population phar- partment equilibration half-lives were approximately 15 min macodynamic model to describe the time course and for levobupivacaine and 25 min for ropivacaine. For levobupi- blockade level of spinal anesthesia. The objective of the vacaine, age reduced the equilibration half-lives at all segments; current study was to develop a population PK-PD model for ropivacaine, age increased the anesthetic sensitivity at seg- ments T12 and higher. of epidural anesthesia based on pharmacokinetic absorp- Conclusions: A population pharmacokinetic–pharmacody- tion profiles. In a first attempt, we applied the model to namic model was developed that quantitatively described sen- levobupivacaine and ropivacaine data and determined sory blockade during epidural anesthesia, including the effects the impact of age on the model parameters. of age. The model may be useful to individualize dose require- ments, to predict the time course of sensory blockade, and to study new local anesthetics. Materials and Methods EPIDURAL anesthesia is obtained by the injection of a Study Design local anesthetic drug into the epidural space. The clini- For the development of the epidural PK-PD model, cal characteristics of the ensuing sensory neural block- data were used from two previous studies3,5 on the ade, such as onset time, intensity, and duration, depend epidural injection of levobupivacaine and ropivacaine directly on the changes in the concentration of the local (see table 1 for patient characteristics). In these studies, anesthetic at the axonal membrane, which are depen- arterial blood samples were obtained before, during, and after the epidural injection of the local anesthetic (at the  Additional material related to this article can be found on the L3–L4 interspace; table 2). The maximum duration of ANESTHESIOLOGY Web site. Go to http://www.anesthesiology sampling was 24 h. When after epidural injection satis- .org, click on Enhancements Index, and then scroll down to find the appropriate article and link. Supplementary material factory anesthetic conditions (i.e., the presence of a can also be accessed on the Web by clicking on the “Arti- bilateral sensory blockade, assessed by pinprick) were clePlus” link either in the Table of Contents or at the top of obtained (usually 15–25 min after the injection), the the Abstract or HTML version of the article. 2 same but now deuterium ( H3)–labeled local anesthetic was administered intravenously (table 2). Blockade as- sessments were made every 5 min during the first 30 * Research Associate, † Professor of Anesthesiology, ‡ Staff Anesthesiologist, min, every 15 min for the next 3.5 h, and subsequently § Professor of Anesthesiology and Chairman. The authors regret the untimely death of Dr. Toine Burm. They dedicate this work to his memory. every 30 min for the next 6 h. Refer to our previous Received from the Department of Anesthesiology, Leiden University Medical publications for further details. For the current study, Center, Leiden, The Netherlands. Submitted for publication September 11, 2007. there were no additional protocols that needed approval Accepted for publication June 18, 2008. Support was provided solely from institutional and/or departmental sources. of an institutional review board. Address correspondence to Dr. Dahan: Department of Anesthesiology, Leiden University Medical Center, P5-Q, P.O. Box 9600, 2300 RC Leiden, The Nether- lands. [email protected]. Information on purchasing reprints may be found at Development of the Pharmacokinetic Model www.anesthesiology.org or on the masthead page at the beginning of this issue. The reanalysis of the pharmacokinetic data from the ANESTHESIOLOGY’s articles are made freely accessible to all readers, for personal use only, 6 months from the cover date of the issue. previous studies had two aims: (1) to obtain popula-

Anesthesiology, V 109, No 4, Oct 2008 664 PK-PD MODELING OF EPIDURAL ANESTHESIA 665

Table 1. Patient Characteristics

Anesthetic Sex, No. F/M Age, yr Weight, kg Height, cm No. PK No. PD*

Levobupivacaine 8/19 55.9 Ϯ 20.0 76.9 Ϯ 16.4 173 Ϯ 9.82 27 24 Ropivacaine 2/22 54.3 Ϯ 16.8 79.7 Ϯ 12.1 177 Ϯ 8.16 24 24 All 10/41 55.1 Ϯ 18.4 78.2 Ϯ 14.5 175 Ϯ 9.19 51 48

Data are mean Ϯ SD or number of patients where appropriate. * For some of the patients, there were pharmacokinetic (PK) but no pharmacodynamic (PD) data available. tion pharmacokinetic models for the local anesthetics ments (fig. 1). After a unit dose, the amount of anesthetic levobupivacaine and ropivacaine together with a co- in a central absorption compartment Ac(t), is given by variate analysis (to enable individualization at the A ͑t͒ ϭ F · exp͑Ϫk · t͒ ϩ F · exp͑Ϫk · t͒, (1) pharmacokinetic level) and (2) to obtain optimal indi- c 1 a1 2 a2 ϭ ϭ vidual absorption parameters for a pharmacodynamic with ka1 log(2)/t½,a1 and ka2 log(2)/t½,a2 for conve- model that describes the effects of those anesthetics nience. The concentration Cc(t) in a central absorption on sensory blockade (and to enable individualization compartment after a dose consisting of amount A (frac- on the pharmacodynamic level). The pharmacokinetic tion of total dose; see Main Assumptions section) of analysis was performed in four steps: (1) A three- anesthetic is then described by compartment structural model was fitted to the con- C ͑t͒ ϭ f · exp͑Ϫk · t͒ ϩ f · exp͑Ϫk · t͒, (2) centration data of the intravenously administered la- c 1 a1 2 a2 ϭ ⅐ ϭ ⅐ beled local anesthetics (fig. 1). (2) A model including with f1 F1 A/Vc and f2 F2 A/Vc, where Vc is the covariate age was developed. (3) A model, consisting volume of that central absorption compartment. Effect of two parallel absorption compartments and the sites were postulated to exist at each segment to account model of disposition developed in step 2, was fitted to for lags between the absorption compartment concen- concentration data of the epidurally administered un- tration and effect, quantified by equilibration rate con- labeled local anesthetic (fig. 1; note that the disposi- stant ke0. The effect site concentration Ce(t; ke0) is then tion parameters were fixed to their empirical bayesian given by [see Statistical Analysis section] values obtained in C ͑t; k ͒ step 2). (4) A model including covariate age was e e0 developed. f · k ϭ 1 e0 Ϫ Ϫ Ϫ Ϫ · ͑exp͑ ka1 · t͒ exp͑ ke0 · t͒͒ The absorption compartments were characterized by ke0 ka1 the following parameters: F , F , t , and t , denot- 1 2 ½,a1 ½,a2 f · k ing the fractions absorbed and absorption half-lives of ϩ 2 e0 Ϫ Ϫ Ϫ Ϫ · ͑exp͑ ka2 · t͒ exp͑ ke0 · t͒͒ (3) the two compartments. We used a parameterization ke0 ka2 = ϭ ⅐ = ϭ Ϫ = with F and F1 where F1 F F1 and F2 F ·(1 F1), Epidural blockade (analgesia with respect to pinprick) because F, the total absorption, should be around 1, and was assumed to occur when Ce(t; ke0) exceeds a con- interindividual variability in F is likely to be counter- 1 centration threshold Cthr (fig. 1). acted by approximately the same variability in F2. Types of Observations. The acquired pharmacody- namic data contained three types of observations: Development of the Pharmacodynamic Model Ն 1. an interval (ton, toff) with Ce(ton; ke0) Cthr where a The Epidural Segments. Each segment was modeled Ͻ blockade occurred, and Ce(toff; ke0) Cthr where the by its own central and peripheral absorption compart- blockade disappeared; 2. as type 1, but at the last assessment time a blockade Table 2. and Amounts of Local Anesthetics Administered in the Two Studies on Which the Current was still present; and Analysis Is Based 3. there was no blockade at all assessment times.

Unlabeled LA Although this might seem counterintuitive, a type 3 observation does provide information, because no block- Concentration, Amount, Dose, Labeled LA ade occurs because C is relatively high, V is large, Study % ml mg Dose, mg thr c and/or A is small. With only type 3 observations in the Levobupivacaine3 0.75 18 126 23 data set, only a lower limit of C can be determined for Ropivacaine5 1 15 125 20 thr a fixed ke0. Together with type 1 and type 2 observa- The local anesthetics (LAs) were administered as short infusions: the unla- tions, type 3 observations do allow for more accurate beled LAs epidurally, the deuterium-labeled LAs intravenously. estimates of Cthr and ke0. Evidently, also the reverse is

Anesthesiology, V 109, No 4, Oct 2008 666 OLOFSEN ET AL.

Fig. 1. Schematic diagram of the pharma- cokinetic–pharmacodynamic model. By intravenous administration of a labeled local anesthetic (LA), its disposition is identified and described by pharmacoki-

netic parameters V1 (central volume) and kij (distribution rate constants between the compartments; V2 and V3 are derived parameters). By of the unlabeled LA, its absorption is iden-

tified and described by parameters F1, F2, ؍ ka1, and ka2. Assumption 1: At time T 0, the unlabeled LA administered at the L3–L4 interspace spreads instantaneously across all segments (for segments T9 and T10, model details are given). Conse- quently, there is only drug distribution between the central absorption compart- not for T > 0(dashed ,0 ؍ ments for T arrows), and there is no distribution be- tween the effect sites via the cerebrospi- nal fluid (CSF) (dotted arrows). Assump- tion 2: For T > 0, the local absorption profiles are identical to the identified global absorption profile. Consequently, the two parallel absorption compartments are equivalent with the central and pe- ripheral absorption compartments as de-

scribed by parameters Vc, kabs, kcp, and kpc (Vp is a derived parameter). When the ef- fect site concentration Ce(t) is above a threshold Cthr, a stimulus S is not per- ceived because it is blocked (e.g., at T10);

when Ce(t) is below Cthr, it is not blocked and can be perceived (e.g., at T9). true: When type 3 observations would be discarded, profile would display such a clear biphasic pattern.1,8 estimates of Cthr would be biased downward. Furthermore, distribution via the cerebrospinal fluid Main Assumptions. According to the pharmacody- can be assumed to be minimal.9 We therefore believe namic model, each segment is described by the follow- that the postulated assumptions are reasonable. ing six parameters: F1, F2, ka1, ka2, ke0, and Cthr.In After the initial spread, the amounts (A) of anes- addition, rate constants should be defined that describe thetic present in each of the segments, and the central the transport of the anesthetic between segments. How- absorption volumes (Vc) at those segments, are not ever, there were at most only two informative observa- simultaneously identifiable. Only a combination param- ϭ ⅐ tions per dermatome, and the pharmacokinetic analysis eter Athr Cthr Vc/A can be estimated. For parameter provides global rather than local absorption process pa- estimation, we therefore used (cf. equations 1 and 3): rameters. So there are more parameters than observa- tions. The following assumptions allowed us to proceed Ae͑t; ke0͒ (fig. 1): F · k ϭ 1 e0 Ϫ Ϫ Ϫ 1. The longitudinal spread of the anesthetic solution Ϫ · ͑exp͑ ka1 · t͒ exp͑ ke0 · t͒͒ ke0 ka1 across the segments of the epidural space occurs F · k instantaneously. ϩ 2 e0 Ϫ Ϫ Ϫ Ϫ · ͑exp͑ ka2 · t͒ exp͑ ke0 · t͒͒ (4) 2. The parameters F1, F2, ka1, and ka2 are equal for each ke0 ka2 segment, albeit that F1 and F2 are interpreted with respect to the fraction of the dose that is present in and assumed that a blockade occurs when A (t; k ) Ն each segment after the initial spread. e e0 Athr. Parameter Athr is a measure of the anesthetic Under these assumptions, both the local (segmental) sensitivity. With two observations per dermatome, the and global (total epidural; obtained by adding all local remaining unknown parameter, ke0, can be estimated. absorption profiles) processes are described by the Parameters Athr and ke0 were assumed to be lognor- same parameters as those obtained from the pharma- mally distributed across the population. Because of cokinetic analysis. If the local absorption or transport the assumptions postulated, they can be estimated for processes would be subject to large intrasegmental each of the segments independently. Further details variability, it is unlikely that the global absorption are given in the appendix.

Anesthesiology, V 109, No 4, Oct 2008 PK-PD MODELING OF EPIDURAL ANESTHESIA 667

Covariate Analysis covariate age via forward selection was based on The pharmacokinetic and pharmacodynamic parame- Akaike’s Information-Theoretic Criterion (AIC).11 The ters (␪, indexed by i) were multiplied by factors contain- best model was defined as the one with the lowest value ing log-transformed and centered (at 50) covariate age: of the criterion, where AIC ϭ MVOF ϩ 2p, where MVOF is the Ϫ2 log-likelihood and p is the number of parame- ␪ ϭ ␪ · exp͑␣ · c ϩ␩͒, (5) i pop,i AGE,i i ters. The model with the lowest value of the criterion is ␪ ϭ 12,13 where pop,i denote population (typical) values; c assumed to have the best predictive properties. ␣ ␩ log(age/50), AGE,i denote covariate coefficients; and i However, to present a classic measure of the significance denote residual interindividual variabilities. of difference from zero of each of the final covariate coefficients of the pharmacokinetic models, likelihood ⌬ Ͼ Statistical Analysis ratio tests were performed ( objective function value Statistical analysis of the pharmacokinetic data were 3.84 corresponds to P Ͻ 0.05; ⌬objective function value performed with the NONMEM version VI software pack- Ͼ 6.63 corresponds to P Ͻ 0.01). To present a classic age (a data analysis program for nonlinear mixed effects measure of significance of the selected covariate coeffi- modeling)10 using its routine ADVAN11 for three-com- cients of the pharmacodynamic models, their 95% con- partmental models of the intravenous data and analytical fidence intervals were graphically displayed together expressions (for processor time reduction) in $PRED for with zero ordinate lines. = the five-compartmental model of the epidural data. Non- Except for F1, lognormal distributions were postulated ␩ = zero variance terms of the s were determined for the for all parameters. The value of F1 was constrained be- population model without covariate age; the structure was tween 0 and 1, which was accomplished by using the assumed to be the same for the model with the covariate inverse logit transformation on its ␩. included (mainly a decrease in variances is to be expected). With distributions of Athr and ke0 available, the prob- Constant relative errors were assumed for the concentra- ability of blockade at each segmental level can be tion measurements. All numerical results are given with determined as a function of time and age. Blockade three significant digits (conform NONMEM output). The probabilities were computed with respect to a stan- empirical bayesian estimates (provided by NONMEM) of dardized dose of 100 mg for ropivacaine and 125 mg the disposition parameters were used for the estimation of for levobupivacaine. the absorption parameters; the empirical bayesian esti- mates of the absorption parameters were used for the estimation of the pharmacodynamic parameters. Statistical analysis of the pharmacodynamic data was Results performed using software written in the computer lan- Pharmacokinetic Analysis guage C by one of the authors (E.O.) using the free GNU Best, median, and worst model fits of the labeled local Scientific Library (GSL).࿣ Additional information regard- anesthetic concentration data (panels 1, 2, and 3, respec- ing this is available on the ANESTHESIOLOGY Web site at tively) according to the coefficient of determination, http://www.anesthesiology.org. With distributions of together with the corresponding model fits of the unla- parameters A and k postulated, the probability of thr e0 beled local anesthetic concentration data, are shown in observing the data per dermatome can be computed by figure 2. The worst model fits of the unlabeled local integration across the (Athr, ke0) plane where it is possi- ble that the data are as observed. The likelihood of anesthetic data still provide adequate inputs for the phar- observing the data are the product of the probabilities of macodynamic models. all observations per dermatome. By maximizing this like- Levobupivacaine and ropivacaine disposition parame- lihood, the parameters of the lognormal distributions of ter estimates are presented in table 3. The only parame- ter that differed significantly between the two anesthet- Athr and ke0 can be obtained. (Note that NONMEM can- not be used when the integrand used for integration ics was k31. A significant effect of age on V1, and across the (Athr, ke0) plane is not continuous.) The therefore on elimination clearance, was detected. In boundaries of the 95% confidence intervals of the param- addition, a significant effect of age on intercompartmen- eter estimates were determined by those parameter val- tal distribution rate constant k12 was detected. ues that increase the objective function by 3.84 points Levobupivacaine and ropivacaine absorption parame-

(i.e., the likelihood profile method). Further details are ter estimates are presented in table 4. Parameters t½,a1 given in the appendix. and t½,a2 were significantly different between the two = The improvement of the model fit by inclusion of anesthetics. Significant effects of age on t½,a1 and F1 were separate parameters for the different anesthetics and detected. Parameter F was approximately 10% higher than 1 for both anesthetics (and had small interindividual ࿣ Available at: http://www.gnu.org/software/gsl. Accessed April 1, 2008. variability around it).

Anesthesiology, V 109, No 4, Oct 2008 668 OLOFSEN ET AL.

Fig. 2. Best, median, and worst model fits (filled circles; panels 1, 2, and 3, respec- tively) according to the coefficient of de- termination (R2) for unlabeled levobupi- vacaine and ropivacaine concentration data (panels A and B, respectively), to- gether with the corresponding model fits of the labeled local anesthetic concentra- tion data (open circles). The numbers in the subplots denote patient identification number/R2 unlabeled/R2 labeled.

Pharmacodynamic Analysis distributions across the population can be estimated.

Figure 3 presents the population estimates of t1⁄2,ke0 The probabilities of blockade (after a dose of 100 mg (converted from the ke0 values for convenience of inter- ropivacaine and 125 mg levobupivacaine) as a function pretation) and Athr with their 95% confidence intervals of time and age of the patient are presented in figure 5. for each segment separately. Both parameters were sig- The effects of increasing age are clearly visible at the nificantly different between levobupivacaine and ropiva- higher segments (T9 and higher) where the probability caine. At the L2 segment level, parameter t1⁄2,ke0 was of block dramatically increases, and by the increase of approximately 5 min for levobupivacaine and 9 min for duration of block. To facilitate the interpretation of fig- ropivacaine, which approximately doubled at higher and ure 5, a three-dimensional view of the blockade proba- tripled at lower dermatome levels. Parameter Athr in- bilities for levobupivacaine in a patient aged 50 yr is creased from S5 to T5, with the highest overall level for given in figure 6. Additional information regarding this is levobupivacaine. The variabilities in ke0 and Athr were available on the ANESTHESIOLOGY Web site at http://www approximately 45% and 25%, respectively (data not .anesthesiology.org. shown). Figure 4 presents the effects of age on the population values. For levobupivacaine, age increase the equilibra- Discussion tion rate constants ke0 at all segments. For ropivacaine, age increased the anesthetic sensitivity Athr at segments We developed a predictive population PK-PD model T12 and higher. of epidural anesthesia and determined its parameters, The nature of the pharmacodynamic data are such that taking into account the effects of age, for the local the model fits exactly for each individual and dermatome anesthetics levobupivacaine and ropivacaine. The (with two measurements (onset and offset of effect), and pharmacokinetic analysis yielded similar population two parameters (t1⁄2,ke0 and Athr)); only their probability parameter estimates with respect to our previous stud-

Table 3. Disposition Pharmacokinetic Model Parameter Estimates for Levobupivacaine and Ropivacaine

␻2 ␣ Parameter Value SE SE AGE SE Ϫ V1 6.85 0.404 0.0352 0.00973 0.193* 0.0949 k10 0.0522 0.00352 0.0724 0.0315 k12 0.152 0.0183 0.0357 0.0188 0.579† 0.155 k21 0.0884 0.00670 0.0241 0.0111 k13 0.0519 0.00508 0.147 0.0399 k31,levo 0.00732‡ 0.000563 0.0669 0.0185 Ϫ0.213 0.137 k31,ropi 0.0137‡ 0.000782 ␴2 0.00991 0.000904

* P Ͻ 0.05, † P Ͻ 0.01, ‡ P Ͻ 0.01 between anesthetics. ϭ ϭ ␻2 ϭ ␴2 ϭ kij rate constant between compartments i and j; V1 volume of central compartment; variance of first-level random effect; variance of second-level random effect.

Anesthesiology, V 109, No 4, Oct 2008 PK-PD MODELING OF EPIDURAL ANESTHESIA 669

Table 4. Absorption Pharmacokinetic Model Parameter Estimates for Levobupivacaine and Ropivacaine

␻2 ␣ Parameter Value SE SE AGE SE t½,a1,levo, min 5.76‡ 0.466 0.147 0.0281 Ϫ0.666† 0.129 t½,a1,ropi, min 9.66‡ 0.837 t½,a2,levo, min 425‡ 27.4 0.0783 0.0150 t½,a2,ropi, min 237‡ 12.1 F 1.08 0.0122 0.00475 0.00168 = F1,levo 0.193 0.0112 0.127 0.0338 Ϫ0.337* 0.142 = F1,ropi 0.221 0.0134 ␴2 0.0149 0.00141

* P Ͻ 0.05, † P Ͻ 0.01, ‡ P Ͻ 0.01 between anesthetics. ϭ = ϭ ϭ ␻2 ϭ F total fraction absorbed; F 1 fraction absorbed in compartment 1; t½,a1 and t½,a2 absorption half-lives (see text); variance of first-level random effect; ␴2 ϭ variance of second-level random effect. ies. The PK-PD model was able to predict the proba- vidual patients were adequately described by fitting bility of block and duration of anesthesia per segment. directly the aggregated model of two parallel first- The covariate analysis showed that age had a signifi- order absorption compartments and the disposition cant and clinically important effect on the spread and profile (a two- or three-compartmental model). These duration of analgesia. analyses allowed the determination of the profiles of We previously determined absorption kinetics after the different local anesthetics and the effects of age. epidural administration of bupivacaine, levobupiva- However, these individual analyses may be ham- caine, and ropivacaine.1–5 Stable isotope–labeled ana- pered by interindividual variability, caused by sex dif- logs were intravenously infused to obtain disposition ferences and genetic, environmental, and pathophys- kinetics. Subsequently, with the use of the disposition iologic factors.14–16 A population approach of these kinetics and the plasma concentration of the epi- data may be attractive because it can explain a part of durally administered unlabeled local anesthetic, the the wide variability by incorporating covariates, such absorption kinetics could be obtained by deconvolu- as the type of local anesthetic and age. In addition, it tion. The plasma concentration–time curves of indi- enables the development of models that are able to

Fig. 3. Equilibration half-lives and absorp- tion thresholds, with their 95% confi- dence intervals, at segments S5–T5 of the two local anesthetics under study. To guide the eye, dashed lines were drawn at the parameter values at level L2 (epidural injection was at the L3–L4 interspace).

Anesthesiology, V 109, No 4, Oct 2008 670 OLOFSEN ET AL.

Fig. 4. Effect of age on ke0 and Athr at seg- ments S5–T5 as quantified by the respec- tive covariate coefficients (equation 5) together with their 95% confidence inter- vals, for the two local anesthetics under study.

make predictions about certain clinically important efficient (see Discussion, Pharmacodynamic Analysis endpoints, such as maximal level and duration of section). blockade. This may, consequently, improve therapeu- Parameter F, which represents the fraction of the ad- tic outcome in future patients.6 ministered dose that is present in the patient’s blood, should be 1. It was greater than 1 for levobupivacaine Pharmacokinetic Analysis and ropivacaine. Possible explanations are given in our The current analysis yielded similar results with re- previous articles3–5; in brief, bioavailability may be over- spect to our previous studies; small differences, how- estimated because of the finite experiment time from ever, were apparent. For example, we now observed which the disposition kinetics (in particular elimination that the clearances of the two anesthetics decrease sig- clearance) are determined. nificantly with increasing age. The observed differences between the absorption Pharmacodynamic Analysis parameters of the two anesthetics can be explained We observed that the sensitivity to the two anesthet- from their vasoconstrictory and vasodilatory proper- ics, quantified by an increase of parameter Athr,de- ties.5 The effects of those differences on the concen- creased with segment height. Assuming that the sen- tration in the epidural space after a standardized bolus sitivity of the to the presence of an anesthetic, dose are shown in figure 7. Levobupivacaine concen- and the volume of the central absorption compart- tration levels are the highest, and this anesthetic could ments do no differ between segments, this decrease is be called the most efficient in terms of concentration. due to a smaller amount of anesthetic reaching the

This observation has to be linked to the pharmacody- higher segments. The overall Athr is highest for namic potency to determine whether it is also effect levobupivacaine. So although the absorption charac-

Fig. 5. Blockade probabilities for the two local anesthetics (ropivacaine dose 100 mg, levobupivacaine dose 125 mg) and the effect of age. The size of the dots is pro- portional to the probability of blockade; in addition, 90%, 75%, and 50% isoeffect lines were drawn (solid, heavily, and lightly dashed lines, respectively).

Anesthesiology, V 109, No 4, Oct 2008 PK-PD MODELING OF EPIDURAL ANESTHESIA 671

Fig. 6. Blockade probabilities for levobupi- vacaine for age 50 yr as a function of time and segment level.

teristics cause this anesthetic to be concentration ef- the nerve block characteristics after epidural adminis- ficient, this is counteracted by its lower sensitivity tration of local anesthetics. A declining number of (figs. 3 and 7). Ropivacaine was found to have a lower neurons, deterioration in myelin sheaths in the dorsal speed of onset and offset (as expressed by t1⁄2,ke0)than and ventral roots, changes in the anatomy of the spine, levobupivacaine. This may be explained by the fact and intervertebral foramina may contribute to altered that the physiochemical properties of ropivacaine are nerve block characteristics after epidural anesthe- similar to those of levobupivacaine, apart from its sia.18,19 Furthermore, the number of in periph- lower lipid solubility. Because lipid solubility is re- eral nerves decreases with advancing age, and the lated to potency, ropivacaine may be less potent.17 conduction velocity diminishes, particularly in motor 19–21 The smaller t1⁄2,ke0 in the neighborhood of the L2 seg- nerves. With increasing age, changes in the con- ment is probably related to differences in the distri- nective tissue ground substances and epidural fat may bution of the anesthetics in the epidural space close to result in changes in local distribution, i.e.,inthe the site of injection. distribution rate of the local anesthetic from the site of Anatomical and physiologic changes associated with injection (the epidural space) to the sites of action.18 advancing age may affect the pharmacokinetics and We observed important age effects for ropivacaine

and levobupivacaine, respectively: (1) Athr decreased at the higher segment levels (T12 and higher); and (2) onset and offset of the sensory blockade, as expressed

by parameter t1⁄2,ke0 was faster with increased age. The most probable mechanism of an increased anesthetic sensitivity with age is an age-dependent change in the longitudinal spread of the anesthetics in the epidural space. This is evident because we observed an in- creased sensitivity with age mainly at the higher seg- ments. The reduced loss of anesthetic via sclerotic intervertebral foramina during its rostral spread with increasing age may explain this observed age effect. Fig. 7. Comparison of absorption profiles of levobupivacaine and ropivacaine in the epidural space (absorption parameters The higher level of analgesia in older patients may as from table 4). well be attributed to a greater sensitivity (a lower Athr)

Anesthesiology, V 109, No 4, Oct 2008 672 OLOFSEN ET AL. such that with the same local anesthetic concentra- tion). In reality, the local anesthetic spreads with a cer- tions at higher thoracic segments, blockade occurs in tain delay. Incorporation of segment-dependent delay is older but not in younger patients. Consequently, to not feasible, because we have only two measurements obtain comparable epidural blocks, smaller doses of per dermatome (onset and offset times of blockade). local anesthetic solutions should be administered to Note that in reality t1⁄2,ke0 is not the delay to the segments older as compared with younger patients. The effect but the delay from the segment to the effect site (i.e., the of age on the speed of onset and offset of effect is spinal nerve roots). Consequently, we may have slightly probably related to changes in epidural fat. Uptake overestimated the value of t1⁄2,ke0. into extraneural tissues, such as epidural fat, limits the Furthermore, the results that we present here are valid rate and extent of drug distribution to the nerves and for the specific volume of the local anesthetic given as thereby changes the time profile of clinical potency.22 well as the location of the epidural puncture (L3–L4 interspace). It has been shown that in contrast to dose, PK-PD Models of Spinal Anesthesia volume per se has little or no effect on sensory block To the best of our knowledge, this is the first PK-PD height and quality of anesthesia.25,26 We expect, how- model developed for lumbar epidural administration ever, that the site of injection affects the parameter of local anesthetics. There are, however, PK-PD mod- values of our model. els for spinal anesthesia in pigs and in humans.7,23,24 The way covariate age was included in the models Shafer et al.23 performed PK-PD modeling of intrathe- allows for a covariate effect in one way only (either a cal neostigmine using cerebrospinal fluid samples and decrease or an increase). In our previous articles, pa- analgesia (visual analog scale) scores. The pharmaco- tients were assigned to age groups; however, this al- kinetic part of their model included a component to lowed for physiologically unrealistic covariate effects. account for the spatial dimension in the intrathecal There was no intraindividual variability in the pharma- administration–analgesia relation. The pharmacoki- codynamic data because of the nature of the binary netic model developed by Ummenhofer et al.24 con- assessments. There was, however, intraindividual uncer- sists of four spatially interconnected subunits of four tainty in the actual times of onset and offset of effect, serially connected compartments to describe drug dis- depending on the sampling times according to the pro- tribution in each subunit. From microdialysis data, it tocol; this uncertainty will bias the estimates of interin- ␻2 ␻2 was not possible to estimate all model parameters for dividual variability ( Athr and ke0) upward. all subunits, and some additional assumptions had to be made. For example, exchange between spinal lev- els was assumed to occur only via the cerebrospinal Conclusions fluid and not via the epidural space. For our model of In conclusion, we developed a predictive PK-PD epidural anesthesia, we assumed that no exchange model of epidural anesthesia in humans. We were able 9 occurs via the cerebrospinal fluid and epidural to demonstrate the importance of age on the spread of space, except in the latter during epidural administra- the sensory blockade as well as on the speed of onset/ 7 tion. Schnider et al. developed a population pharma- offset of blockade. The model allows the study of codynamic model of spinal anesthesia from pinprick various important factors in epidural anesthesia, such assessments. Although their model is applicable to as the site of injection, the volume of the injected epidural anesthesia, it is unable to predict sensory fluid, combined spinal–epidural injections, anesthetic- blockade at levels lower than the maximum. Further- opioid interaction, and pregnancy. Furthermore, the model more, because they did not take into account pharma- may be used in the development and study of new local cokinetic data (on disposition and absorption) and anesthetics for epidural use. hence the local anesthetic concentration, their phar- macodynamic analysis is affected by pharmacokinetic variability. Although this seems of limited importance References when treating and predicting epidural anesthesia in a 1. Burm AGL, Vermeulen NPE, van Kleef JW, de Boer AG, Spierdijk J, Breimer new patient, the inclusion of covariates in the phar- DD: Pharmacokinetics of lignocaine and bupivacaine in surgical patients follow- ing epidural administration: Simultaneous investigation of absorption and dispo- macokinetic model will improve the prediction of sition kinetics using stable isotopes. Clin Pharmacokinet 1987; 13:191–203 sensory anesthesia in this particular patient. 2. Veering BT, Burm AGL, Vletter AA, van den Heuvel RPM, Onkenhout W, Spierdijk J: The effect of age on the systemic absorption, disposition and phar- macodynamics of bupivacaine. Clin Pharmacokinet 1992; 22:75–84 Critique on Methods 3. Simon MJG, Veering BT, Stienstra R, van Kleef JW, Burm AGL: Effect of age on the clinical profile and systemic absorption and disposition of levobupiva- A potential drawback of our model is the underlying caine after epidural administration. Br J Anaesth 2004; 93:512–20 essential assumption that rostral and caudal spread of the 4. Simon MJG, Veering BT, Stienstra R, van Kleef JW, Williams SGP, McGuire GM, Burm AGL: The systemic absorption and disposition of levobupivacaine 0.5% anesthetic in the epidural space is instantaneous and after epidural administration in surgical patients: A stable-isotope study. Eur J subsequently remains unchanged (apart from absorp- Anaesth 2004; 21:460–70

Anesthesiology, V 109, No 4, Oct 2008 PK-PD MODELING OF EPIDURAL ANESTHESIA 673

5. Simon MJG, Veering BT, Vletter AA, Stienstra R, van Kleef JW, Burm AGL: Effect of age on the systemic absorption and systemic disposition of ropivacaine after epidural administration. Anesth Analg 2006; 102:276–82 6. Olofsen E, Dahan A: Population pharmacokinetics/pharmacodynamics of anesthetics. AAPS J 2005; 7:E383–9 7. Schnider TW, Minto CF, Bruckert H, Mandema JW: Population pharmaco- dynamic modeling and covariate detection for central neural blockade. ANESTHE- SIOLOGY 1996; 85:502–12 8. Tucker GT, Mather LE: Pharmacokinetics of local anesthetic agents. Br J Anaesth 1975; 47:213–24 9. Hogan Q, Toth J: Anatomy of soft tissues of the . Reg Anesth Pain Med 1999; 24:303–10 10. Beal SL, Sheiner LB, Boeckman AJ, editors: NONMEM User’s Guides. Ellicott City, Maryland, Icon Development Solutions, 1989–2006 11. Akaike H: A new look at the statistical model identification. IEEE Trans Automat Contr 1974; 19:716–23 12. Burnham KP, Anderson DR: Multimodel inference: Understanding AIC and BIC in model selection. Sociol Meth Res 2004; 33:261–304 Fig. 8. Central and effect site absorption profiles in the epidural 13. Olofsen E: The performance of model selection criteria in the absence of space (solid and hatched curves, respectively) given by equa- a fixed-dimensional correct model (abstract 1198). Abstracts of the Annual Meeting tions 1 and 4 with parameter estimates from patient 4210 sub- of the Population Approach Group in Europe. Copenhagen, Denmark, 2007. stituted. The dots denote the effect site absorption Ae(t; ke0)at 14. Sarton E, Olofsen E, Romberg R, DenHartigh J, Kest B, Nieuwenhuijs D, onset and offset times t , t , t , and t , respectively. Burm A, Teppema L, Dahan A: Sex differences in morphine analgesia: An exper- on,1 on,2 off,1 off,2 The anesthetic sensitivity A (horizontal line) and central ef- imental study in healthy volunteers. ANESTHESIOLOGY 2000; 93:1245–54 thr 15. Pe´rez-Urizar J, Granados-Soto V, Flores-Murrietta FJ, Castan˜eda-Hernandez fect site equilibration rate constant ke0 were approximated G: Pharmacokinetic-pharmacodynamic modeling: Why? Arch Med Res 2000; from the average (ton, toff) (see text). 31:539–45 16. Wright PMC: Population based pharmacokinetic analysis: Why do we need the (double) integral of the probability distributions of Athr and ke0 it; what is it; and what has it told us about anaesthetics. Br J Anaesth 1998; across this area: 80:488–501 17. Casati A, Putzu M: Bupivacaine, levobupivacaine and ropivacaine: Are they ϭ ͐ke0,max ͐Athr,max͑ke0͒ ␮ ␻2 ␮ ␻2 clinically different? Best Pract Res Clin Anaesthesiol 2005; 19:247–68 Li NL͑Athr; Athr, Athr͒NL͑ke0; ke0, ke0͒dAthrdke0 k A ͑k ͒ 18. Bromage PR: Epidural Analgesia. Philadelphia, Saunders, 1978, pp 10–45, e0,min thr,min e0 263 (6) 19. Ferrer-Brechner T: Spinal and epidural anesthesia in the elderly. Semin Anesth 1986; 5:54–61 where NL denotes the fact that Athr and ke0 are normally distributed in 20. Jacob JM, Love S: Qualitative and quantitative morphology of human sural the log-domain and ␮ and ␻2 denote, in NONMEM parlance, typical nerve at different ages. Brain 1985; 108:897–924 21. Dorfman LJ, Bosley TM: Age related changes in peripheral and central population values and variabilities. The integration interval for Athr nerve conduction in man. Neurology 1979; 29:38–44 depends on ke0, and the integration interval for ke0 is determined by 22. Burm AGL: Clinical pharmacokinetics of epidural and spinal anesthesia. nonempty Athr integration intervals (fig. 9). The integrals are evaluated Clin Pharmacokinet 1989; 16:283–311 in the log-domain; the inner integral can be numerically approximated 23. Shafer SL, Eisenach JC, Hood DD, Tong C: Cerebrospinal fluid pharmaco- kinetics and pharmacodynamics of intrathecal neostigmine methylsulfate in hu- by one of the GSL gsl_cdf_normal routines, and the outer integral can mans. ANESTHESIOLOGY 1998; 89:1074–88 be numerically approximated by one of the GSL gsl_integration_qag 24. Ummenhofer WC, Arends RH, Shen DD, Bernards CM: Comparative spinal routines, depending on whether the ke0 range is finite or infinite. The distribution and clearance kinetics of intrathecally administered morphine, fen- likelihood of the set of observations is given by the product of equation tanyl, alfentanil, and sufentanil. ANESTHESIOLOGY 2000; 92:739–53 25. Dernedde M, Stadler M, Bardiau F, Boogaerts JG: Continuous epidural 7 for each individual. infusion of large concentration/small volume versus small concentration/large For a type 2 observation, a blockade is still present at the last volume of levobupivacaine for postoperative analgesia. Anesth Analg 2003; ϭ Ն assessment time, so Ae(tend) Ae(toff,1) Athr and toff,2 is not observed, 96:796–801 which means that the information about A and k is not bounded by 26. Duggan J, Bowler GMR, McClure JH, Wildsmith JAW: Extradural block thr e0 with bupivacaine: Influence of dose, volume, concentration and patient charac- the line Ae(toff,2, ke0) in figure 9, and the area of integration therefore 2 teristics. Br J Anaesth 1988; 61:324–31 also includes the horizontally shaded part (with ke0 0). Ͻ For a type 3 observation, Ae(t) Athr for all assessment times; only Ͻ the maximum value of Ae(tmax) gives information, because if Ae(tmax) A , A (t) Ͻ A for all t t (the observations are not independent, Appendix thr e thr max and function Ae(t) has only one maximum). Likelihood Function for the Set of Observations For a type 1 observation, pinprick assessment times yield intervals

(ton,1, ton,2) and (toff,1, toff,2) that constitute the closest intervals around the actual onset and offset times of effect (fig. 8). The remaining assessments do not contain any additional information, Յ Ն because they were always 0 for t ton,1 and t toff,2, and 1 for ton,2 Յ Յ t toff,1 (a result of the choice of assessment times with respect to the rate of change in the effect site absorption profile). There- fore, the uncertainty about whether there is a blockade cannot be estimated and has to be neglected, but there is uncertainty with respect to the actual onset and offset times. This uncertainty is taken into account by the postulated normal probability distribu- ␻2 ␻2 tions for the logarithms of Athr and ke0 with variances Athr and ke0, respectively. Thus, they are a measure of the combination of intra- Fig. 9. The area of integration is the area bounded by the effect individual and interindividual variabilities (which cannot be sepa- site absorption profiles for ton,1, ton,2, toff,1, and toff,2 (when rated). Figure 9 depicts the (A , k ) area for which it is possible thr e0 observed) as functions of ke0 (given by equation 4 with fixed that the observed intervals (ton,1, ton,2) and (toff,1, toff,2) are ob- absorption parameters). The dot in the center denotes the (Athr, served. The likelihood of an observation in individual i is given by ke0) estimate from the average (ton, toff).

Anesthesiology, V 109, No 4, Oct 2008 674 OLOFSEN ET AL.

The implementation of the integrals in the C program were checked Blockade Probabilities using Monte Carlo simulation, i.e., by counting the number of samples Once the distributions of Athr and ke0 are known (their parameters from the lognormal distributions of Athr and ke0 that fall within the estimated), the probability of blockade at time t is given by: region of integration. The model parameters were estimated by the method of maximum P ͕ blockade at t ͖ likelihood, i.e., they are given by those values that maximize L ϭ͟L i i ϭ ͕ ͑ ͒ Ն ͖ (for all individuals i, equation 6); the maximum was numerically de- P Ae t; ke0 Athr termined by the GSL routine gsl_multimin_fminimizer_nmsimplex. ϱ A ϭ ͐ ͐ e͑t;ke0͒ ␮ ␻2 ␮ ␻2 NL͑Athr; Athr, Athr͒NL͑ke0; ke0, ke0͒dAthrdke0 (7) 0 0 Initial Parameter Values Suppose that t Ϸ t and t Ϸ t , so that they can be approx- on,1 on,2 off,1 off,2 ␮ ␮ where the population values Athr and ke0 are functions of age as imated by their averages ton and toff. Then there exists a unique relation given by equation 5, and the parameters of Ae(t; ke0) (equation 4) between (ton, toff) and (Athr, ke0); the latter are given by those values that ϭ ϵ are given by population estimates of F1, F2, ka1, and ka2, which are satisfy Ae(ton; ke0) Ae(toff; ke0) Athr (using equation 4). They are unique 1 1 2 1 also functions of age. When the population distributions of the because Ae(ton; ke0) when ke0 while Ae(toff; ke0) when ke0 . The equality is numerically found by the GSL root-finding routine gsl_root_fsol- absorption parameters are postulated correctly and/or their vari- ver_brent. Initial parameter values were calculated as the means and variances ances are small, the bias caused by taking their population estimates of the logarithms of the Athr and ke0 estimates at each dermatome for all instead of in addition integrating across their respective distribu- individuals from the type 1 observations. tions is assumed to be negligible.

Anesthesiology, V 109, No 4, Oct 2008 Are epidurals worthwhile in vascular surgery? Bernadette T. Veering

Department of Anesthesiology, Leiden University Purpose of review Medical Center, Leiden, The Netherlands Patients undergoing major vascular surgery are at increased risk for postoperative Correspondence to Bernadette T. Veering, MD, PhD, complications due to the high incidence of comorbidities in this population. Department of Anesthesiology, Leiden University Medical Center, P5-Q, P.O. Box 9600, 2300 RC Epidural anaesthesia provides potential benefits but its effect on morbidity and mortality Leiden, The Netherlands is unclear. Tel: +31 71 52622301; fax: +31 71 5266230; e-mail: [email protected] Recent findings Existing studies fail to demonstrate improved clinical outcome and reduced mortality for epidural anaesthesia or combined epidural/general techniques compared with Current Opinion in Anaesthesiology 2008, 21:616–618 general anaesthesia. Postoperative epidural analgesia provides better pain relief and reduces the duration of postoperative mechanical ventilation. Summary Optimization of perioperative care rather than the anaesthetic technique may have potential benefit in improving postoperative outcome.

Keywords epidural anaesthesia, outcome, vascular surgery

Curr Opin Anaesthesiol 21:616–618 ß 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins 0952-7907

suppressing the surgical stress response [2]. Major surgery Introduction is associated with a hypercoagulable state that persists Major vascular surgery is considered to be a high-risk into the postoperative period. The incidence of throm- procedure. It is often applied in elderly patients, who boembolic complications is lower in patients operated exhibit many comorbidities such as ischaemic heart upon under epidural anaesthesia compared to those given disease, renal insufficiency and diabetes. Many of these general anaesthesia [3]. Modulation of the surgically patients are at increased risk for perioperative compli- induced changes in coagulation or fibrinolysis by epidural cations. Consequently, these patients would benefit from anaesthesia might be due to an increased blood flow anaesthetic techniques reducing those risks. The ques- with subsequent reduced venous stasis, to effects of tion whether epidural or general anaesthesia is preferable the local anaesthetics themselves and to blunting the for vascular surgery has been debated for years. stress response [3,4]. The present review will focus on the probable advantages Epidural anaesthesia extending above T12 is associated of epidural anaesthesia for vascular surgery. with splanchnic sympathetic nervous blockade, resulting in reduced inhibitory gastrointestinal tone and increased intestinal blood flow. Perioperative intestinal hypoperfu- Physiologic effects sion is a major contributing factor leading to organ dys- The use of epidural anaesthesia provides physiologic function. TEA during and after major surgery has been benefits and may attenuate the pathophysiology that shown to protect the gut from decreased microvascular occurs following surgery. This may be associated with perfusion and to improve the mucosal blood flow, even a reduction in and morbidity under conditions of decreased perfusion pressure [5,6]. when compared with general anaesthesia. TEA with postoperative thoracic epidural analgesia has Thoracic epidural anaesthesia (TEA) covering the car- been shown to shorten the time to first bowel movement diac segments (T1–T4) improves an ischaemia-induced [7] and the duration of postoperative colonic ileus [8]. left ventricular dysfunction; reduces electrocardio- graphic, echocardiographic and angiographic signs of coronary insufficiency; decreases the incidence of arrhy- Vascular surgery thmias and provides relief of ischaemic chest pain [1]. Various comorbidities associated with advanced age, such Local anaesthetics have the capability to block afferent as ischaemic heart disease, renal insufficiency and dia- and efferent signals to and from the spinal cord, thus betes are robust, predictors of cardiac complications in

0952-7907 ß 2008 Wolters Kluwer Health | Lippincott Williams & Wilkins DOI:10.1097/ACO.0b013e32830c6ec8

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. Are epidurals worthwhile in vascular surgery? Veering 617

the vascular patient [9]. Cardiac and pulmonary com- analgesia in reducing postoperative myocardial infarction. plications remain the main causes of perioperative mor- These findings suggest that thoracic epidural anaesthesia bidity associated with major vascular surgery [10]. may have a favorable effect on cardiac morbidity, however more randomized clinical trials are warranted. A commonly asked question is whether the mentioned physiologic benefits ascribed to epidural anaesthesia will A part of postoperative complications is caused by respir- make a difference in the outcome of vascular surgical atory problems during the early postoperative recovery patients. period. A meta-analysis by Ballantyne et al. [21] found that postoperative epidural analgesia was associated with Aortic surgery a decrease in the incidence of atelectasis, pneumonia, Major vascular surgical procedures are amenable to overall pulmonary complications and an increase in the epidural anaesthesia combined with general anaesthesia arterial partial pressure of oxygen, compared with syste- (EA þ GA). Aortic cross clamping and unclamping are mic opioid administration. These findings suggest that associated with complex cardiovascular, renal, humoral postoperative epidural analgesia has a beneficial effect on and haemostatic changes [11]. During reconstructive aortic pulmonary morbidity. surgery, clamping and unclamping is usually well tolerated but may produce severe haemodynamic disturbances. A recent systemic review showed that epidural analgesia provides better pain relief and reduces the duration of Patients with TEA þ GA to have a more stable haemo- tracheal intubation after open abdominal aortic surgery dynamic course during the critical times of aortic clamp- compared to systemic opioid-based relief [22]. These ing and unclamping with fewer fluctuations in blood findings suggest that epidural pain relief may be advan- pressure and heart rate [12,13]. tageous for pulmonary patients undergoing abdominal aortic surgery. With cross clamping both with epidural anaesthesia þ general anaesthesia and with general anaesthesia alone, Peripheral vascular surgery signs of myocardial ischaemia did occur. So the favourable Studies of Christopherson et al. [23] and of Bode et al. [24] factor of epidural anaesthesia on the myocardial oxygen reported no difference in myocardial ischaemia, major supply–demand relationship did not influence the inci- cardiac morbidity or operative mortality when comparing dence of intraoperative myocardial ischaemia [14]. epidural anaesthesia and analgesia with general anaes- thesia followed by intravenous patient-controlled analge- Discrepancies exist between studies that have evaluated sia (PCA) in patients undergoing lower extremity vascular the effect of epidural anaesthesia þ general anaesthesia surgery. The limitation of both studies was that they were and postoperative analgesia in comparison with general inadequately powered to detect differences between anaesthesia and postoperative systemic opioid-based pain anaesthetic techniques. However, a benefit of epidural relief on the postoperative cardiac outcome in patients anaesthesia over general anaesthesia was a reduction in who underwent elective abdominal aortic surgery. Some the incidence of graft occlusion [4,15,25]. studies [15,16] involving high-risk (e.g. aortic reconstruc- tion) surgery patients have reported significant reductions Lower extremity revascularization is performed in patients in cardiac morbidity associated with the use of intraopera- with a high incidence of coexistent diseases. Krupski et al. tive and postoperative epidural anaesthesia/analgesia [26,27] compared the difference in early and late cardiac using local anaesthetics and opioids. morbidity in patients undergoing suprainguinal aortic operations versus patients undergoing infrainguinal vas- Others[13,17–19]didnotdemonstrateadecreaseincardiac cular operations. The incidence of early postoperative morbidity in patients undergoing aortic reconstructions cardiac complications was the same in both groups, how- treated with intraoperative TEA þ light general anaesthe- ever cardiac morbidity rates at 2 years were greater for sia. However, epidural anaesthesia and analgesia was infrainguinal versus abdominal aortic procedures. This associated with shortened intubation times and ICU stay. may be due to a greater prevalence of diabetes and cardiac morbidities in patients with peripheral vascular diseases. A meta-analysis of 17 studies examining the effect of In both studies the effect of anaesthetic technique on postoperative epidural analgesia on postoperative cardiac cardiac morbidity was not investigated. morbidity showed a significant reduction in postoperative myocardial infarction in the epidural group continued for The major drawbacks of early prospective studies are more than 24 hours [20]. Patients underwent abdominal methodological flaws, including different regimens of and vascular surgery and had a history of preoperative epidural anaesthesia, different protocols of postoperative cardiac morbidity. Subgroup analysis demonstrated that pain treatment, inclusion of small numbers of patients thoracic epidural analgesia was superior to lumbar epidural and possible investigator bias.

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 618 Regional anaesthesia

In order to avoid those deficiencies of design and meth- 7 Steinbrook RA. Epidural anesthesia and gastrointestinal motility. Anesth Analg 1998; 86:837–844. odology Norris et al. [28] performed a double-masked 8 Zugel N, Bruer C, Breitschaft K, Angster R. Effect of thoracic epidural randomized controlled trial of 168 patients undergoing analgesia on the early postoperative phase after interventions on the gastro- abdominal aortic surgery. In this study strict intraopera- intestinal tract. Chirurgia 2002; 73:262–268. tive and postoperative protocols were used to guide and 9 Lee TH, Marcanonio ER, Mangione CM, et al. Derivation and prospective validation of a simple index for prediction of cardiac risk of major noncardiac optimize perioperative management and postoperative surgery. Circulation 1999; 100:1043–1049. analgesia. Participants were allocated into four treatment 10 Elkouri S, Gloviczki P, McKusick MA, et al. Perioperative complications and groups, two groups received TEA combined with a light early outcome after endovascular and open surgical repair of abdominal aortic aneurysms. J Vasc Surg 2004; 39:497–505. GA followed by either PCA or epidural patient-controlled 11 Gelman S. The pathophysiology of aortic cross-clamping and unclamping. analgesia (EPCA) and two groups received GA alone Anesthesiology 1995; 82:1026–1057. followed by either PCA or EPCA. 12 Mason RA, Newton GB, Cassel W, et al. Combined epidural and general anaesthesia in aortic surgery. J Cardiovasc Surg 1990; 31:442–447. 13 Garnett RL, Lindsay P, Cole CW, et al. Perioperative ischaemia in aortic The primary outcome was length of stay; secondary surgery: combined epidural/general anaesthesia and epidural analgesia vs outcomes included selected postoperative morbidities general anaesthesia and iv analgesia. Can J Anaesth 1996; 43:769–777. like pulmonary complications and cardiovascular events. 14 Dodds TM, Burns AK, DeRoo DB, et al. Effect of anesthetic technique on myocardial wall motion abnormalities during abdominal surgery. J Cardiovasc Cardiac morbidity was defined as myocardial infarction, Anesth 1997; 11:129–136. angina, congestive heart failure and ventricular tachyar- 15 Tuman KJ, McCarthy RJ, March RJ, et al. Effects of epidural anesthesia and rhythmia. Postoperative outcomes were not influenced analgesia on coagulation and outcome after major vascular surgery. Anesth by intraoperative and postoperative anaesthetic regi- Analg 1991; 73:696–704. 16 Park WY, Thompson JS, Lee KL. Effect of epidural anesthesia and analgesia men. This implies probably that perioperative care of on perioperative outcome: a randomized, controlled veterans affairs coop- patients undergoing vascular procedures seems to be erative study. Ann Surg 2001; 234:560–571. important regardless of the anaesthetic techniques used 17 Baron JF, Bertrand M, Barre E, et al. Combined epidural and general anesthesia versus general anesthesia for abdominal aortic surgery. Anesthe- [29]. siology 1991; 75:611–618. 18 Davies MJ, Silbert BS, Mooney PJ, et al. Combined epidural and general anaesthesia versus general anaesthesia for abdominal aortic surgery: a Conclusion prospective randomized trial. Anaesth Intensive Care 1993; 21:790–794. Existing studies fail to demonstrate improved clinical 19 Boylan JF, Katz J, Kavanagh BP, et al. Epidural bupivacaine-morphine analgesia outcome and reduced mortality for epidural anaesthesia versus patient-controlled analgesia following abdominal aortic surgery: analge- sic, respiratory and myocardial effects. Anesthesiology 1998; 89:585–593. or combined epidural/general techniques compared with 20 Beattie WS, Badner NH, Choi P. Epidural anesthesia and analgesia reduces general anaesthesia. Large, well designed randomized postoperative myocardial infarction: a meta-analysis. Anesth Analg 2001; clinical trials on a long-term basis are necessary to draw 93:853–858. 21 Ballantyne JC, Carr DB, Deferranti S, et al. The comparative effects of solid conclusions on the effect of anaesthetic techniques postoperative analgesic therapies on pulmonary outcome: cumulative meta- on eventual outcome after vascular surgery. analysis of randomized controlled trials. Anesth Analg 1998; 86:598–612. 22 Nishimori M, Ballantyne JC, Low JH. Epidural pain versus systemic opioid- based pain relief for abdominal aortic surgery. Cochrane Database Syst Rev However it seems more reasonable that the real aim 2006:CD005059. should be to improve the perioperative care of vascular Important systemic review that shows that epidural analgesia provides better pain relief and reduces the duration of tracheal intubation after open abdominal aortic surgery patients rather than to focus on the effect of surgery compared with systemic opioid-based relief. the particular anaesthetic technique on patient outcome 23 Christopherson R, Beattie C, Gottlieb S, et al. Perioperative morbidity in following vascular surgery. patients randomized to epidural or general anesthesia for lower extremity vascular surgery. Anesthesiology 1993; 79:422–434. 24 Bode RH, Lewis KP, Zarich SW, et al. Cardiac outcome after peripheral References and recommended reading vascular surgery: comparison of general and regional anesthesia. Anesthe- Papers of particular interest, published within the annual period of review, have siology 1996; 84:3–13. been highlighted as: 25 Singh N, Sidawy AN, Dezee K, et al. The effects of the type of anesthesia of special interest on outcomes of lower extremity infrainguinal bypass. J Vasc Surg 2006; of outstanding interest 44:964–970. 26 Krupski WC, Layug EL, Reilly LM, et al. Comparison of cardiac morbidity rates 1 Veering BT, Cousins MJ. Cardiovascular and pulmonary effects of epidural between aortic and infrainguinal operations. J Vasc Surg 1992; 15:354–365. anaesthesia. Anaesth Intensive Care 2000; 28:620–635. 27 Krupski WC, Layug EL, Reilly LM, et al. Comparison of cardiac morbidity rates 2 Kehlet H. Manipulation of the metabolic response in clinical practice. World J between aortic and infrainguinal operations: two-year follow up. J Vasc Surg Surg 2000; 24:690–695. 1993; 18:609–617. 3 Modig J, Borg T, Bagge L, et al. Role of extradural and of general anaesthesia 28 Norris EJ, Beattie C, Perler BA, et al. Double-masked randomized trial on fibrinolysis and coagulation after total hip replacement. Br J Anaesth 1983; comparing alternate combinations of intraoperative anesthesia and post- 55:625–629. operative analgesia in abdominal aortic surgery. Anesthesiology 2001; 4 Rosenfeld BA, Beattie C, Christopherson R, et al. The effect of different 95:1054–1067. anesthetic regimens on fibronolysis and the developments of postoperative An important study demonstrating that no differences in perioperative outcome in arterial thrombosis. Anesthesiology 1993; 79:435–443. patients undergoing surgery of the abdominal aorta occurred when comparing 5 Adolphs J, Schmidt DK, Mousa SA, et al. Thoracic epidural anesthesia TEA þ general anaesthesia followed by either intravenous or epidural patient- attenuates hemorrhage-induced impairment of intestinal perfusion in rats. controlled analgesia with general anaesthesia alone followed by either intravenous Anesthesiology 2003; 99:685–692. or epidural patient-controlled analgesia. Strict intraoperative and postoperative 6 Sielenkamper AW, Eicker K, Van Aken H. Thoracic epidural anesthesia protocols were used to guide and optimize perioperative management. increases mucosal perfusion in ileum of rats. Anesthesiology 2000; 93: 29 Rock P. Regional vs GA for vascular surgery patients. ASA annual meeting 844–851. refresher course lectures. 2007; 209.

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. British Journal of Anaesthesia 102 (6): 739–48 (2009) doi:10.1093/bja/aep096 Advance Access publication May 6, 2009

REVIEWARTICLE Failed spinal anaesthesia: mechanisms, management, and prevention

P. D. W. Fettes1*, J.-R. Jansson2 and J. A. W. Wildsmith1

1University Department of Anaesthesia, Ninewells Hospital & Medical School, Dundee DD1 9SY, UK. 2AstraZeneca R&D SE-151 85, Sodertalje, Sweden *Corresponding author. E-mail: [email protected]

Although spinal (subarachnoid or intrathecal) anaesthesia is generally regarded as one of the most reliable types of regional block methods, the possibility of failure has long been recognized. Dealing with a spinal anaesthetic which is in some way inadequate can be very difficult; so, the technique must be performed in a way which minimizes the risk of regional block. Thus, prac- titioners must be aware of all the possible mechanisms of failure so that, where possible, these mechanisms can be avoided. This review has considered the mechanisms in a sequential way: pro- blems with ; errors in the preparation and injection of solutions; inadequate spreading of drugs through cerebrospinal fluid; failure of drug action on nervous tissue; and diffi- culties more related to patient management than the actual block. Techniques for minimizing the possibility of failure are discussed, all of them requiring, in essence, close attention to detail. Options for managing an inadequate block include repeating the injection, manipulation of the patient’s posture to encourage wider spread of the injected solution, supplementation with local anaesthetic infiltration by the surgeon, use of systemic sedation or analgesic drugs, and recourse to general anaesthesia. Follow-up procedures must include full documentation of what happened, the provision of an explanation to the patient and, if indicated by events, detailed investigation. Br J Anaesth 2009; 102: 739–48 Keywords: anaesthetic techniques, regional; anaesthetic techniques, spinal; complications

Two conditions are, therefore, absolutely necessary to commoner outcome is that a block results, but is inadequate produce spinal anesthesia: puncture of the for the proposed surgery. Such inadequacy may relate to and subarachnoid injection of an anesthetic agent. three components of the block: the extent, quality, or dur- Gaston Labat, 1922 ation of local anaesthetic action, often with more than one of these being inadequate. This review has considered all Spinal (intrathecal) anaesthesia is generally regarded as one three eventualities within the definition of ‘failure’. of the most reliable of regional block methods: the needle Most experienced practitioners would consider the inci- insertion technique is relatively straightforward, with cere- dence of failure with spinal anaesthesia to be extremely low, brospinal fluid (CSF) providing both a clear indication of perhaps less than 1%. However, a figure as high as 17% has successful needle placement and a medium through which been quoted from an American teaching hospital, yet most local anaesthetic solution usually spreads readily. However, of the failures were judged to be ‘avoidable’.28 Asurveyat the possibility of failure has long been recognized, the another such institution considered that this high rate was above quote being taken from the work of Gaston Labat,24 ‘unacceptable’, and recorded the much lower, but still sig- the ‘father’ of modern regional anaesthesia. His two con- nificant, figure of 4%, with ‘errors of judgement’ as the ditions for success, although perhaps a little simplistic major factor.32 The clear implication is that careful attention when related to current knowledge, still indicate the to detail is vital, and it has been shown that a failure rate of essence of the method and provide a starting point for the ,1% is attainable in everyday practice.17 Minimizing the consideration of failure, although it may be helpful to incidence of failure is obviously a pre-requisite for gaining define exactly what this means first. Literally, the word the benefits of spinal anaesthesia, and prevention must start failure implies that a spinal anaesthetic was attempted, but with full recognition of the potential pitfalls so that clinical that no block resulted; this happens, but perhaps a practice can be tailored to their avoidance.

# The Author [2009]. Published by Oxford University Press on behalf of The Board of Directors of the British Journal of Anaesthesia. All rights reserved. For Permissions, please email: [email protected] Fettes et al.

In general terms, block failure is usually ascribed to one 160 The Americas of three aspects: clinical technique, inexperience (of the Europe unsupervised trainee especially), and failure to appreciate 140 Asia the need for a meticulous approach.10 However, such Australasia 120 broad categories reveal little about the many detailed ways in which an intrathecal injection can go astray within each 100 of the five phases of an individual spinal anaesthetic, these being, in sequence, lumbar puncture, solution injection, 80 spreading of drug through CSF, drug action on the spinal 60 nerve roots and cord, and subsequent patient management. Number of reports All of the problems involved are well described in the lit- 40 erature, but usually long ago, and many practitioners seem unaware of the issues involved. For instance, the neuro- 20 science division of AstraZeneca received 562 ‘Product 0 Defect Notification’ reports in the 6 yr to December 31, 1995 2000 2005 2007, all ascribing failed spinal anaesthetics to ineffective Calendar year bupivacaine solution (Fig. 1). Nearly one-third of reports Fig 1 Annual numbers of reports of failed spinal anaesthesia with (179) were from the UK, but virtually every country bupivacaine received by AstraZeneca between January 1, 1993 and where the drug is marketed was represented. However, December 31, 2008 plotted according to region of the world. analysis showed that the returned material was within the product’s specification in every case so a formal review, tissue or blood clot can easily obstruct the fine bore based on a literature search, was thought to be worthwhile. needles used now. Otherwise, a failed lumbar puncture is virtually always because of either poor positioning of the Search strategy patient or incorrect needle insertion, both factors being For this review ‘PubMed’ and ‘Google’ databases were within the control of the anaesthetist. Abnormalities of the searched using the terms ‘failed regional anaesthesia’, spine (kyphosis, scoliosis, calcification of ligaments, con- ‘failed regional anesthesia’, ‘failed spinal anaesthesia’, and sequences of osteoporosis), , and patient anxiety ‘failed spinal anesthesia’. Relevant articles were retrieved make both positioning the patient and needle insertion as were any possibly relevant papers in their reference lists. more difficult, especially in the elderly. Texts of regional Supporting searches were performed on subjects that may anaesthesia give more extensive instruction than can be not have been otherwise identified, specific examples being provided here, and good clinical training is the key to CSF volume, dural ectasia, and the chemical compatibility success, but most difficulties are attributable to lack of of local anaesthetics with adjuncts. adherence to the basic rules. In addition, searches were made using ‘Planet’ (an AstraZeneca internal database), ‘Biosis’, ‘Current Positioning Contents’, ‘Embase’, ‘PsycINFO’, ‘Medline’, and ‘Medline The patient is placed on a firm surface; the lumbar laminae Daily update’, using the terms ‘Failed Spinal Anaesthesia’ and spines are ‘separated’ maximally by flexing the whole and ‘Failed Spinal Anesthesia’ as sole search terms and spine (including the neck), the hips, and knees; rotation ‘spinal anesthesia’ or ‘spinal anaesthesia’ or ‘spinal anes- and lateral curvature of the spine are avoided; these points thetic’ or ‘spinal cord anesthesia’ or ‘spinal cord anaesthe- apply to lumbar puncture in both sitting and lateral hori- sia’ or ‘anesthesia, spinal’ or ‘anaesthesia, spinal’ and zontal positions; the former is usually an easier option in ‘treatment failure’ or ‘therapy failure’, and ‘Intrathecal’. ‘difficult’ patients, but sometimes the reverse is true. The All papers identified as relevant are included in this review. role of the assistant in achieving and maintaining the patient in the correct position cannot be underestimated.35 Mechanisms and their prevention Needle insertion Although its accurate identification can be difficult using Failed lumbar puncture clinical land-marks, what is judged to be the third lumbar Inability to obtain CSF, sometimes referred to as a ‘dry inter-space is used usually, but examination may indicate tap’, is the only cause of failure which is immediately that another is preferable. However, care should be taken obvious. A needle with a lumen blocked at the outset is a not to venture too cephalad and risk damage to the spinal theoretical possibility, but is most unlikely with modern cord.33 With the midline approach, insertion should start equipment. However, both needle and stylet must be precisely in the mid-line, mid-way between the posterior checked for correctness of fit before use, and the needle spines, with the needle shaft at right angles to the back in should not be advanced without the stylet in place because both planes. Small, incremental changes in needle angle

740 Failed spinal anaesthesia should be made only if there is resistance to advancement; Solution injection errors if resistance is met, cephalad angulation should be tried The appearance of CSF in the needle hub is an essential first, and such angulation may be appropriate from the pre-requisite for spinal anaesthesia, but it does not guaran- start if the patient is unable to flex fully (e.g. the obstetric tee success, which also requires that a fully effective dose patient at term). A degree of caudad angulation is some- is both chosen and actually deposited in the CSF. times needed, with a slight lateral direction being required very rarely. All authorities recommend that the anaesthetist Dose selection should have a good knowledge of spinal anatomy and Studies of many factors influencing intrathecal drug spread relate these to changes in tissue resistance as the needle is have shown that the dose injected, within the range nor- advanced so that a mental ‘picture’ of where the needle tip mally used, has only a small effect on the extent of a lies is appreciated. spinal anaesthetic, but is far more important in determin- The above points apply specifically to the midline ing the quality and duration of block.20 Overall, the actual approach; lateral or paramedian approaches are preferred dose chosen will depend on the specific local anaesthetic by some,27 especially if the mid-line ligaments are heavily used, the baricity of that solution, the patient’s subsequent calcified, but they are inherently more complex techniques. posture, the type of block intended, and the anticipated However, in the face of difficulty, the same basic rules duration of surgery. Thus, knowledge of the factors influ- apply: make sure that the patient is in the correct position encing intrathecal drug spread and clinical experience with and that the correct angles and insertion technique are any particular local anaesthetic preparation are important used. guides to choosing an effective dose. However, the need to guarantee an adequate effect means that the doses of drugs injected in standard ‘single Adjuncts shot’ techniques are larger than is strictly necessary, A calm, relaxed patient is more likely to assume and experience with dose titration during continuous spinal maintain the correct position, so explanation (before and anaesthesia showing clearly that lower doses are often during the procedure) and gentle, unhurried patient hand- effective.21 In attempts to either minimize hypotension, ling are vital; light anxiolytic premedication contributes for example by attempting to produce a unilateral block, or much to relaxing the patient; local anaesthetic infiltration speed postoperative mobilization, by decreasing duration, at the puncture site must be effective without obscuring some practitioners use lower doses than is traditional3 (e.g. the landmarks, but must include both intradermal and s.c. 5–10 rather than 15 mg of hyperbaric bupivacaine). Used injection. Achieving the correct position is a particular correctly, and in appropriate situations, such doses can be challenge in the patient in pain (e.g. from a fractured hip) reliable, but they do mean that the margin for error is and systemic analgesia (i.v. or ) helps consider- reduced and that the consequences of other problems (e.g. ably. The aim of such adjuncts is to optimize the patient’s Loss of injectate—see below) will be exaggerated and so position and to prevent any movement. As will be dis- risk an inadequate block. It becomes even more important cussed later, it takes only slight movement to displace the to ensure that the whole of that lower dose reaches the needle from its target. CSF and then spreads properly, remembering that the Advances in ultrasound technology are reaching the ‘dead space’ of the needle will contain a significant pro- stage where it can be used to overcome difficulties with portion of what is a small volume to start with. lumbar puncture, but clinicians will still need to be aware of the problems and how they should be overcome. Loss of injectate The Luer connection between syringe and needle provides Pseudo-successful lumbar puncture a ready opportunity for leakage of solution. A particular The appearance of clear fluid at the needle hub is usually variant of this problem being a leak through a defect at 6 the final confirmation that the subarachnoid space has been the junction of needle hub and shaft. Given the small entered. Rarely, however, the clear fluid is not CSF, but volumes involved, the loss of even a few drops may cause local anaesthetic injected as a ‘top-up’ for an epidural a significant decrease in the mass of drug reaching the which then proved inadequate for a Caesarean section, or CSF, and thus in its effectiveness. To avoid this, it has even spreading there from the lumbar plexus.26 long been conventional teaching that the syringe contain- Unfortunately, a positive test for glucose in the fluid does ing the injectate must be inserted very firmly into the hub not confirm that this fluid is definitely CSF because extra- of the needle, and that a subsequent check is made that no cellular fluid constituents diffuse rapidly into fluids leakage occurs. injected into the epidural space. Another, even rarer, suggested cause of clear fluid appearing at the needle hub, Misplaced injection but not confirming successful lumbar puncture, is a conge- Needle and syringe must be connected firmly, but great nital arachnoid cyst.39 care should be taken to avoid either anterior or posterior

741 Fettes et al. displacement of the needle tip from subarachnoid to epi- successful (Fig. 3A), but subsequent injection pushes the dural space, where deposition of a spinal dose of local dura forward and the solution is misplaced (Fig. 3B). anaesthetic will have little or no effect. Fluid aspiration, A variant is that the needle tip penetrates the dura, but it after attachment of the syringe, should confirm free flow is the that acts as the flap valve so that a of CSF and, thus, that the needle tip is still in the correct subdural injection results (Fig. 3C). This misplacement is space, but such aspiration may displace the tip unless per- usually thought of as leading to excessive spread during formed carefully, as may the force of the injection of the epidural block, but the equivalent phenomenon has been syringe contents. To prevent displacement at any stage, it described after intended subarachnoid injection,37 and is has been advocated that the dorsum of one hand should be a recognized complication of myelography.22 Subdural anchored firmly against the patient’s back and the fingers injection has also been identified as the cause of a failed used to immobilize the needle, while the other hand is block when either epidural11 or subarachnoid15 injection used to manipulate the syringe.40 Most practitioners would was intended. recommend aspiration for CSF after the injection to These eventualities, being subtle abnormalities of pla- confirm that correct placement is maintained, and some cement, are impossible to identify at the actual time, but advocate that this is done half way through as well rotation of the needle through 3608 after the initial although neither of these practices has been shown to appearance of CSF, and before check aspiration, has influence the outcome of the block.32 40 been advocated as a way of minimizing the possibility Tip displacement must be guarded against with any type of them occurring, the theory being that the rotation of spinal needle, but it is a particular issue with the reduces the risk of the membrane edges catching on the ‘pencil point’ needles now used widely to minimize the opening. incidence of post-dural puncture . The opening at the end of these needles is proximal to the tip, so only a minor degree of ‘backward’ movement during syringe Inadequate intrathecal spread attachment may result in epidural injection as was recog- The intrathecal spread of a local anaesthetic solution, even nized at an early stage in the widespread use of such when correctly placed, has truly been described as capri- needles.12 The distances involved are of the order of a cious.7 The factors that affect it are many, but the focus millimetre or two, but (as with leakage) misplacement of here will be on those that may result in inadequate spread. only a small amount of solution can have significant effects. An additional issue with pencil-point needles is Anatomical abnormality that the opening, being much longer than the bevel of a Intrathecal spread is governed by interplay between sol- Quincke needle, may ‘straddle’ the dura so that some sol- ution physical characteristics, gravity, and the configur- ution reaches the CSF, and some the epidural space ation of the vertebral canal. Anatomical abnormalities that (Fig. 2).41 This may be exaggerated by the dura acting as lead to problems with spread can be both overt and covert. a ‘flap’ valve across the needle opening. Initially, CSF The curves of the vertebral column are integral to solution pressure pushes the dura outwards so that aspiration is

Epidural Dura and Subarachnoid Cauda space arachnoid space equina

Epidural Dura and Subarachnoid Cauda A CSF space arachnoid space equina Aspiration

LA B LA

Correct Epidural

C LA LA LA Subdural Incorrect

Fig 3 To show how the dura or arachnoid mater may act as a ‘flap’ valve Fig 2 Possible positions of the tip of a pencil-point needle. If it is across the opening of a pencil point needle. During aspiration (A) the correctly placed (upper picture) all of the local anaesthetic solution will dura/arachnoid are pulled back allowing CSF to enter the needle. During reach the subarachnoid space, but if the opening ‘straddles’ the dura injection the dura (B) or arachnoid (C) is pushed forward and the local (lower picture) some solution will be deposited in the epidural space. anaesthetic enters the epidural or subdural space.

742 Failed spinal anaesthesia spread and any obvious abnormality, kyphosis, or scolio- preparation at the mid-lumbar level, the solution will sis, may interfere with the process. Examination of the spread ‘down’ the slope under the effect of gravity to pool patient should reveal whether this might occur, but it is at the ‘lowest’ point of the thoracic curve, so exposing all not possible to predict whether the effect will be excessive nerve roots up to that level to an effective concentration of spread or failure. local anaesthetic. However, if lumbar puncture is per- A very rare possibility, which is not apparent on exam- formed at the fourth lumbar or the lumbo-sacral interspace ination, is that the ligaments that support the spinal cord the local anaesthetic may be ‘trapped’ below the lumbar within the theca form complete septae and act as longi- curve, especially if the patient is in the sitting position tudinal or transverse barriers to local anaesthetic spread. during injection and maintained in that position for a This can result in a block that is entirely unilateral2 or of period thereafter. insufficient cephalad spread. Spinal or other This results in a block that is restricted to the sacral seg- pathological lesions might also limit spread, effectiveness, ments, just as has been described with a spinal catheter or both, one such case being attributed to the conse- that passes caudally.31 Prevention relies on avoiding too quences of previous intrathecal chemotherapy.142 low an injection level unless, of course, a deliberate Similarly, previous surgery within the vertebral canal may ‘saddle’ block is intended. result in adhesions that interfere with spread. An interesting ‘abnormality’ considered to have caused Ineffective drug action restricted spread in a single patient was a larger than usual volume of CSF in the lumbar theca.18 Subsequent sys- The last possible explanation for a failed spinal is that the tematic study has shown that lumbar CSF volume is the solution actually injected reaches the target nerves, but is most important factor influencing the variability seen inactive or ineffective, with a variety of explanations between individuals in the spread of an intrathecal injec- being possible. tion.9 A negative correlation was found between lumbar CSF volume and the peak sensory level achieved with Identification errors hyperbaric bupivacaine when the injection was performed Spinal anaesthetics are supplied in aqueous solution ready in both supine and sitting positions. A variation of this for injection and there is no opportunity for confusion in factor is dural ectasia, which is a pathological enlargement the preparation of the solution itself. However, other opti- of the dura seen in the majority of patients with Marfan’s cally clear solutions, such as a separate local anaesthetic syndrome and in some other connective tissue disorders.25 for skin infiltration or analgesic adjuvants, are often used from the same sterile preparation area and the possibility Solution density that confusing them may lead to an ineffective block must Consistently effective spinal anaesthesia requires that the be considered. Recognition of the possibility of such injec- practitioner has a good understanding of the factors that tion errors has led to the widespread use of syringe label- affect intrathecal spread, particularly, but not only, sol- ling in anaesthesia, but this is not as easy within a sterile ution density. A solution with a density within the normal field as it is on an anaesthetic work station. Attention to range of that of CSF (‘isobaric’) will virtually guarantee detail is essential, but minimizing the number of ampoules block of the lower limbs with little risk of thoracic nerve on the block tray (such as using the same local anaesthetic block and thus hypotension.43 However, plain solutions of for both skin infiltration and spinal) and consistent use bupivacaine, although often referred to as isobaric, are of different sizes of syringe for each component of the actually of sufficiently lower density to be hypobaric, procedure help considerably. especially at body temperature. As a result their range of spread is much less predictable than that of a truly isobaric Chemical incompatibility preparation, and occasionally the block may be no higher The mixing of two different pharmaceutical preparations than the first, or even second, lumbar dermatome when also raises the possibility of ineffectiveness as a result of administered to the non-pregnant supine patient.29 interaction between local anaesthetic and adjuvant. Local Although the impact of variations in CSF volume has yet anaesthetics seem to be compatible with most of the to be studied with these solutions, it seems likely that this common opioids, but there has been little formal study of factor may well be a factor in their variability. the effects of mixing them, and the situation is even less Solutions with a density greater than that of CSF definitive with other adjuvants such as clonidine, midazo- (hyperbaric) move very definitively under the combined lam, and other more extreme substances. Certainly, there influence of gravity and the curves of the vertebral canal. are no studies of the stability of three or more substances Over a hundred years ago, Barker, one of the early pio- when mixed together for intrathecal use, a not unknown neers of spinal anaesthesia in the UK, observed that the practice today. Chemical reaction might generate an addition of glucose to the solution made for a more obvious precipitate, but another possibility is that the pH reliable effect.4 In the standard scenario, that of a patient of the local anaesthetic solution becomes even lower than placed supine after the injection of a hyperbaric it was to start with. This would decrease the concentration

743 Fettes et al. of the un-ionized fraction that is what diffuses into nerve individuals, and anxiety alone can cause much difficulty. tissue and, unless the solution mixes well with CSF, a Further, operating tables are designed for surgical access, decreased effect could result. There is at least one report not patient comfort; and intra-abdominal stimuli may indicating that the incidence of failure is greater after the result in afferent impulses in unblocked parasympathetic addition of a vasoconstrictor solution and this could rep- and phrenic nerve fibres. The more anxious the patient, the resent an example of this effect.30 greater will be the impact of these factors and the more likely will it be that the patient will fail to cope with the Inactive local anaesthetic solution situation and claim that the anaesthetic has not worked The older, ester-type local anaesthetics are chemically properly. Expectation plays a part, and good preoperative labile so that heat sterilization and prolonged storage, par- patient counselling followed by a supportive approach ticularly in aqueous solution, can make them ineffective from the anaesthetist during the operation is important in because of hydrolysis and hence they need very careful avoiding such problems, but so is the judicious, and handling. Although the more modern amide-linked drugs pro-active use of systemic sedative and analgesic drugs. (e.g. lidocaine, bupivacaine, etc.) are much more stable Sufficient sedation to produce drowsiness, or even sleep and can be heat sterilized in solution and then stored for (with appropriate monitoring), is rarely contra-indicated several years without loss of potency, there have been a other than in the obstetric situation, and even there small number of reports attributing failure of spinal anaesthetics doses may occasionally be useful. to inactive drug.8163844 Testing the block Local anaesthetic resistance In recent years, it has become almost mandatory, certainly Very rarely a failed spinal anaesthetic has been attributed in the obstetric setting, to test the level of block formally to physiological ‘resistance’ to the actions of local before surgery commences. This apparently sensible pre- anaesthetic drugs, although the reports tend to the caution may be difficult or impossible to undertake in anecdotal.5233645A history of repeated failure of dental some patients (for example the demented patient with a or other local anaesthetic techniques is accompanied by fractured neck of femur). Excessive focus on testing can speculation that the problem is because of a sodium- also have a negative impact. Most patients will have some channel mutation that renders the drugs ineffective. anxiety about the effectiveness of the injection, and this However, no such mutation has ever been described, and will be increased if testing is started too soon. the clinical reports are incomplete, specifically failing to Conventional practice is to check motor block by testing consider not only the recognized causes of failure, but also the ability to lift the legs, followed by testing of sensory the behaviour of an anxious patient preferring general block to stimuli such as soft touch, cold, or pin prick, all anaesthesia as an explanation for the ‘resistance’. Very of which have their proponents. It is advisable to start detailed investigation would be required for ‘resistance’ to testing in the lower segments, where onset will be fastest, be accepted as an explanation. As an aside, any patient and work upwards. Proving early on that there is some giving a history of repeated failures with local anaesthesia effect encourages patient confidence; testing too soon does should be managed by an experienced clinician. the opposite. Even if there is no formal assessment of the level of block, the clinician must be confident that an adequate Failure of subsequent management block has been produced. Establishing that the level Not all of a patient’s claims of discomfort, or even pain, of block is appropriate for the projected surgery is often during a spinal anaesthetic are the result of an inadequate taken to demonstrate that the quality of block is adequate injection. A properly performed spinal anaesthetic will also, but this is not always the case if cold or pin-prick produce complete somatic, and a major degree of auto- stimuli are used. The observation that the upper block nomic, nerve block in the lower half of the body unless a level is a few dermatomes above which innervate the sur- specifically restricted method is used. However, ensuring gical field (not forgetting the deeper structures) is a good that this block occurs is only part of the process because start, but it does not guarantee that the quality of block is the unaffected components of the nervous system require sufficient. A covert pinch of the site of the proposed surgi- consideration and management. Specifically, this relates to cal incision may be a better indicator of skin analgesia, conscious awareness of the clinical setting and of ‘sen- and can be reassuring if the block has been slow in onset. sations’ transmitted through unblocked nerves, with both Indeed, there is much to be said, particularly when the factors possibly making the patient claim that the block patient is conscious, for asking the surgeon to do the same has failed. This may not actually be the case, but patient with a toothed surgical forceps before incising the skin,10 management certainly has failed if such a claim is made but surreptitiously and without asking a loaded question when the block is actually as good as it could be. such as ‘Does that hurt?’! The patient is distracted by con- Lying supine and wide awake while undergoing surgery versation and an exchange of glances between surgeon is not a pleasant experience, even for the most sanguine of and anaesthetist is all that is needed for surgery to begin.

744 Failed spinal anaesthesia

Catheter and combined techniques block, the more likely is the block to be inadequate, so the The great majority of spinal anaesthetics involve a single, more detailed this assessment should be. While the onset through needle injection and, as has been noted, this of spinal anaesthesia is rapid in most patients, it can be requires some certainty about its effectiveness for surgery. slow in some; so, ‘ of time’ should always be To take advantage of the rapid onset and profound block allowed.10 However, if most of the expected block has not of spinal anaesthesia, both continuous and combined developed within 15 min, some additional manoeuvre is spinal–epidural techniques have been introduced to almost certainly going to be needed. The possibilities, increase flexibility. If the catheters are correctly placed, their explanations, and suggested immediate responses are problems of inadequate spread, quality, and duration of as follows: effect can be dealt with although many of the potential technical problems outlined above can still apply. (1) No block: the wrong solution has been injected, it has However, both methods require a greater level of skill and been deposited in the wrong place, or it is ineffective. experience to use, the insertion of an intrathecal catheter Repeating the procedure or conversion to general can be surprisingly difficult to achieve in some patients anaesthesia are the only option. If, after operation, the and, as has been mentioned already, can result in the mis- patient has significant pruritus, it is likely that only an direction of the local anaesthetic solution, with the risk of opioid was injected. neurotoxicity.34 It is vital to leave no more than 2–3 cm (2) Good block of inadequate cephalad spread: the level of catheter within the dura to avoid this. In the combined of injection was too low, anatomical abnormality has technique, it is common to inject a relatively small volume restricted spread, or some injectate has been mis- for the spinal component, so the problems that can result placed. If a hyperbaric solution was used, flex the in a proportion not reaching the subarachnoid space are patient’s hips and knees and tilt the table head down. very relevant, but at least the epidural catheter can be used This straightens out the lumbar curve, but maintains a in attempts to rescue the situation. cephalad ‘slope’ and allows any solution ‘trapped’ in the to spread further. A variation with the same aim, but perhaps better suited to the obstetric Management of failure situation, is to turn the patient to the full lateral position with a head down tilt, reversing the side after Failure of a spinal anaesthetic is an event of significant 2–3 min. If a plain (and usually slightly hypobaric) concern for both patient and anaesthetist even when it is solution has been used, it may help to sit the patient immediately apparent, but it can have serious conse- up, but beware of peripheral pooling of blood. quences (clinical and medico-legal) if the problem only If a spinal catheter injection results in inadequate becomes evident once surgery has started. If there is any spread, the response should not be to inject more of doubt about the nature or duration of the proposed the same solution because dose has minimal effect on surgery, a method other than a standard spinal anaesthetic intrathecal spread.20 Either posture should be manipu- should be used. The trainee anaesthetist should avoid lated as above, or a different baricity of solution over-selling the technique, especially in the early days of should be tried, or the catheter should be withdrawn unsupervised practice. Promising that all will be achieved before the injection is repeated.31 by one injection leaves no room for manoeuvre, but (3) Good, but unilateral block: this is most likely because offering one injection to reduce pain and a second to of positioning, but it is possible that longitudinal liga- ensure unconsciousness does. If a spinal anaesthetic does ments supporting the cord have blocked spread. If the fail in some way, the management options are limited; so, operation is to be on the anaesthetized limb, then the the first rule is to expend every effort in prevention. surgeon should know that the other leg has sensation, and the patient should be reassured and closely moni- Prevention is better than cure tored. Otherwise, turning the patient onto the Having made the decision to use a spinal anaesthetic, the unblocked side if a hyperbaric solution was used (or block should be performed with meticulous attention to the reverse for plain solutions) may facilitate spread. detail as has been indicated above. It is impossible to over- (4) Patchy block (This term is used to describe a block emphasize this point. that appears adequate in extent, but the sensory and motor effects are incomplete.): causes of inadequate block are numerous and include all those discussed The failed block above, but the most likely explanation is that the local The precise management of the failed block will depend anaesthetic was at least partially misplaced, or that the on the nature of the inadequacy and the time at which it dose given was inadequate. If this becomes apparent becomes apparent. Thus, some monitoring of the onset of before surgery starts, the options are to repeat the the block and correct interpretation of the observations are spinal injection or to use a greater degree of systemic both vital. The slower the onset of either motor or sensory supplementation than was planned, the latter being the

745 Fettes et al.

only option after skin incision. It may not be necessary Only some of these problems have actually been to recourse to general anaesthesia, sedation, or analge- described, most being in the category of theoretical possi- sic drugs often being sufficient especially when bilities, but such concerns do reinforce the view that every patient anxiety is a major factor. Infiltration of the effort must be made to ensure that the first injection is wound and other tissues with local anaesthetic by the fully effective. surgeon may also be useful in such situations. (5) Inadequate duration: the most likely explanation is Recourse to general anaesthesia that for one of several reasons an inadequate dose of There are many ways in which an inadequate block might local anaesthetic was delivered to the CSF. be ‘rescued’, but there is a limit to how much discomfort Alternatively, lidocaine (intended for skin infiltration) or distress an individual patient can tolerate, so general was confused for bupivacaine, or the operation has anaesthesia must be considered if one or two simple taken longer than expected. Systemic supplementation measures have not rectified matters. Common sense and or infiltration of local anaesthetic may tide matters clinical experience are usually the best indicators of over, but often the only option is to convert to general exactly when to convert to general anaesthesia, so the anaesthesia. unsupervised trainee can be at a disadvantage. However, it is far better to make the decision sooner rather than later and have to deal with a seriously distressed patient. Of Repeating the block course, explaining later why the anticipated technique had not been provided can be difficult. It is another reason Where no effect at all has followed the injection it seems why prevention (getting the block right to start with) is the reasonable to repeat the procedure, paying close attention best approach, but it is also a reason for not ‘over-selling’ to avoiding the potential pitfalls. In all other situations the regional approach before operation. besides total failure, there must be some local anaesthetic If general anaesthesia is induced to supplement a par- in the CSF already, and anxieties relating to several issues tially effective spinal anaesthetic, any degree of sympath- have to be taken into account: etic nerve block will make hypotension more likely. (1) A restricted block may be because of some factor, probably anatomical, impeding the physical spread of Follow-up initiatives the solution, and it may have exactly the same impact on a second injection, resulting in a high concentration Clinical follow-up of local anaesthetic at or close to the site of injection. As with any anaesthetic complication, the details should Cauda equina lesions were described after continuous be documented fully in the notes, and the patient provided spinal anaesthesia when very restricted spread with an apology and a full explanation after operation. prompted repeat injections rather than the manipu- Giving the patient a written summary of events for presen- lation of other factors,34 and a similar problem has tation to a future anaesthetist can be very helpful, although been described after repeated needle injection.14 19 care should be taken to prevent medico-legal recourse. (2) Repeat injection, especially in response to a poor Rarely, inadequate spread has been the first indication of quality block, may lead to excessive spread,13 so it pathology within the vertebral canal. Therefore, it may be may be argued that a lower dose should be used to appropriate to look for symptoms and signs of neurologi- reduce the risk of this possibility. cal disease, and involve a neurologist if there is any suspi- (3) A good quality, but unilateral block, might lead to an cion of these being present. attempt to place a second injection into the ‘other’ It is during the follow-up of a patient in whom no block side of the theca, but the risk of placing the second was obtained, the possibility of local anaesthetic ‘resist- dose in the same side must be significant. ance’ may seem an attractive explanation. As has already (4) Barriers to spread within the subarachnoid space may been noted, much wider consideration of the possibilities, also affect epidural spread (and vice versa), so an supported by very detailed investigation, is needed than attempt at epidural block may not succeed either. has been the case in previous reports. (5) A block of inadequate cephalad spread might be over- come by repeating the injection at a higher level, but Investigating local anaesthetic effectiveness should perhaps only be attempted when the indication Spinal anaesthesia is usually a simple and effective tech- for a regional technique is considerable. nique, but ‘failure’ can occur at any time and in the hands (6) The final concern, particularly applicable to the last of any clinician, no matter how experienced. However, if mentioned, but relevant to nearly all situations where the procedure has, apparently, been routine and straightfor- a repeat block might be considered, is that the adjacent ward concerns can arise that the current supply of local nerve tissue is already affected by local anaesthetic anaesthetic is defective, especially if two or more such action so that the risk of direct needle trauma is failures occur in the same hospital within a short period of increased. time. The preparations which have been most implicated

746 Failed spinal anaesthesia are those of hyperbaric bupivacaine (probably because it determinant of sensory block extent and duration during spinal is the drug used most commonly at present), with drug anesthesia. Anesthesiology 1998; 89:24–9 from both major suppliers, Abbott38 and AstraZeneca,816 10 Charlton JE. Managing the block. In: Wildsmith JAW, Armitage EN, McClure JH, eds. Principles and Practice of Regional 44 being involved. In fact, the chemical stability of the Anaesthesia, 3rd Edn. Edinburgh: Churchill Livingstone, 2003 amide drugs and modern standards of pharmaceutical 11 Collier CB. Accidental subdural injection during attempted lumbar manufacture mean that drug inactivity is a most unlikely epidural block may present as a failed or inadequate block: radio- cause of a failed spinal anaesthetic, but it remains a possi- logical evidence. Reg Anesth Pain Med 2004; 29:45–51 bility which at least has to be eliminated. 12 Crone W. Failed spinal anesthesia with the Sprotte needle. As has been suggested, performing skin infiltration with Anesthesiology 1991; 75: 717–8 some of the solution intended for the spinal injection 13 Deshpande S, Idriz R. Repeat dose after an inadequate spinal block. Anaesthesia 1996; 51: 892 should demonstrate that it is effective. If the concern con- 14 Drasner K, Rigler ML. Repeat injection after a ‘failed spinal’: at tinues the operating theatre, pharmacy and anaesthetic times, a potentially unsafe practice. Anesthesiology 1991; 75: department records should be cross-checked to see 713–4 whether other practitioners in the hospital have experi- 15 Gershon RY. for cesarian section with a subdural enced any problems. Similarly, distributors should be able catheter. Can J Anaesth 1996; 43: 1068–71 to check whether other hospitals which have been supplied 16 Harris RW, McDonald P. Inadequate spinal anaesthesia with 0.5% with material from the same batch have reported difficulty. Marcaine Heavy (batch DK-1961). Int J Obstet Anaesth 2004; 13: 130–1 If such enquiries reveal that others are using the same 17 Harten JM, Boyne I, Hannah P, Varveris D, Brown A. Effects of material to good effect, the clinician should consider the height and weight adjusted dose of local anaesthetic for spinal advice of those two great authorities, Lee and Atkinson, anaesthesia for elective Caesarean section. Anaesthesia 2005; 60: on ‘The spinal that does not take’: 348–53 18 Hirabayashi Y, Fukuda H, Saitoh K, Inoue S, Mitsuhata H, Shimizu R. All experienced workers have encountered this Failed spinal anaesthesia: cause identified by MRI. Can J Anaesth occasionally even though accepted procedure has 1996; 43:1072–5 apparently been followed. Reflection, however, 19 Hirabayashi Y, Konishi R, Shimizu R. Neurologic symptom associ- usually discloses some flaw in technique. In 1907 ated with a repeated injection after failed spinal anesthesia. Alfred E. Barker wrote that for successful spinal Anesthesiology 1998; 89: 1294–5 analgesia it is necessary ‘to enter the lumbar dural 20 Hocking G, Wildsmith JAW. Intrathecal drug spread. Br J Anaesth 2004; 93: 568–78 sac effectually with the point of the needle, and to 21 Hurley RJ, Lambert DH. Continuous spinal anesthesia with a discharge through this, all the contemplated dose of microcatheter technique: preliminary experience. Anesth Analg the drug, directly and freely into the cerebrospinal 1990; 70: 97–102 fluid, below the termination of the cord’ (Barker, 22 Jones MD, Newton TH. Inadvertent extra-arachnoid injection in 1907). Failure to follow the details of this advice is myelography. Radiology 1963; 80: 818–22 the commonest cause of a poor result.27 23 Kavlock R, Ting PH. Local anesthetic resistance in a pregnant patient with lumbosacral plexopathy. http://www.biomedcentral. com/1471-2253/4/1 24 Labat G. Regional Anesthesia: Its Technic and Clinical Application. References Philadelphia, PA: WB Saunders Company, 1922 1 Adler R, Lenz G. Neurological complaints after unsuccessful 25 Lacassie HJ, Millar S, Leithe LG, et al. Dural ectasia: a likely cause spinal anaesthesia as a manifestation of incipient syringomyelia. of inadequate spinal anaesthesia in two parturients with Marfan’s Eur J Anaesthesiol 1998; 15: 103–5 syndrome. Br J Anaesth 2005; 94: 500–4 2 Armstrong PJ. Unilateral subarachnoid anaesthesia. Anaesthesia 26 Lang SA, Prusinkiewicz C, Tsui BCH. Failed spinal anesthesia 1989; 44: 918–9 after a psoas compartment block. Can J Anesth 2005; 52:74–8 3 Atallah MM, Shorrab AA, Abdel Mageed YM, Demian AD. 27 Lee JA, Atkinson RS. Sir Robert Macintosh’s Lumbar Puncture and Low-dose bupivacaine spinal anaesthesia for percutaneous Spinal Analgesia. Edinburgh: Churchill Livingstone, 1978 nephrolithotomy: the suitability and impact of adding intrathecal 28 Levy JH, Islas JA, Ghia JN, Turnbull C. A retrospective study of fentanyl. Acta Anaesthesiol Scand 2006; 50: 798–803 the incidence and caues of failed spinal anesthetics in a university 4 Barker AE. A report on clinical experiences with spinal analgesia hospital. Anesth Analg 1985; 64: 705–10 in 100 cases. Br Med J 1907; i: 665–74 29 Logan ML, McClure JH, Wildsmith JAW. Plain bupivacaine—an 5 Batas D, et al. Resistance to local anaesthetics: a case report. unpredictable spinal anaesthetic agent. Br J Anaesth 1986; 58:292–6 http://bja.oxfordjournals.org/cgi/qa-display/short/brjana_el;1576 30 Manchikanti L, Hadley C, MArkwell SJ, Colliver JA. A retrospec- 6 Ben-David B, Levin H, Tarhi D. An unusual explanation for a tive analysis of failed spinal anesthetic attempts in a community failed spinal. Can J Anaesth 1995; 45: 448–9 hospital. Anesth Analg 1987; 66: 363–6 7 Bier A. Versuche ueber Cocainiserung des Rueckenmarkes. 31 Morrison LMM, McClure JH, Wildsmith JAW. Clinical evaluation Dtsche Z Chir 1899; 51: 361–9 of a spinal catheter technique in femoro-popliteal graft surgery. 8 Calthorpe N. Inadequate spinal anaesthesia with 0.5% Marcaine Anaesthesia 1991; 46: 576–8 Heavy (batch DK 2016). Int J Obstet Anaesth 2004; 13: 131 32 Munhall RJ, Sukhani R, Winnie AP. Incidence and etiology of failed 9 Carpenter RL, Hogan QH, Liu SS, Crane B, Moore J. spinal anesthetics in a university hospital. Anesth Analg 1988; 67: Lumbosacral cerebrospinal fluid volume is the primary 843–8

747 Fettes et al.

33 Reynolds F. Case report: damage to the follow- 40 Tarkkila PJ. Incidence and causes of failed spinal anesthetics in a ing spinal anaesthesia. Anaesthesia. 2001; 56: 235–47 university hospital: a prospective study. Reg Anesth 1991; 16: 34 Rigler ML, Drasner K, Krejce TC, Yelich SJ, Scholnick FT, 48–51 DeFontes JA. Cauda equina syndrome after continuous spinal 41 Thomson GE, McMahon D. Spinal needle manufacture, design anesthesia. Anesth Analg 1991; 72: 275–81 and use. Bailliere’s Clin Anaesthesiol 1993; 7: 817–30 35 Rubin AP. Spinal anaesthesia. In: Wildsmith JAW, Armitage EN, 42 Westphal M, Gotz T, Booke M. Failed spinal anaesthesia McClure JH, eds. Principles and Practice of Regional Anaesthesia, 3rd after intrathecal chemotherapy. Eur J Anaesthesiol 2005; 22: Edn. Edinburgh: Churchill Livingstone, 2003 235–6 36 Schmidt SI, Moorthy SS, Dierdorf SF, Anagnostou JM. A series of 43 Wildsmith JAW, McClure JH, Brown DT, Scott DB. Effects of truly failed spinal anesthetics. J Clin Anesth 1990; 2: 336–8 posture on the spread of isobaric and hyperbaric amethocaine. Br 37 Singh B, Sharma P. Subdural block complicating spinal anesthesia. J Anaesth 1981; 53: 273–8 Anesth Analg 2002; 94: 1007–9 44 Wood M, Ismail F. Inadequate spinal anaesthesia with 0.5% 38 Smiley RM, Redai I. More failed spinal anesthetics with hyperbaric Marcaine Heavy (batch 1961). Int J Obstet Anaesth 2003; 12: 310–1 bupivacaine. Int J Obstet Anaesth 2004; 13: 131–4 45 Woolard A. Causes of local resistance. https://www.biomedcentral. 39 Stace JD, Gaylard DG. Failed spinal anaesthesia. Anaesthesia 1996; com/1471-2253/4/1/comments/comments 51: 892–3

748  Original Articles Disclosure of Risks Associated With Regional Anesthesia: A Survey of Academic Regional Anesthesiologists Richard Brull, M.D., F.R.C.P.C., Colin J.L. McCartney, M.B.Ch.B., F.R.C.A., F.C.A.R.C.S.I., F.R.C.P.C., Vincent W.S. Chan, M.D., F.R.C.P.C., Gregory A. Liguori, M.D., Mary J. Hargett, B.Sc., Daquan Xu, M.B., M.Sc., Sherif Abbas, M.D., and Hossam El-Beheiry, M.B.B.Ch., Ph.D., F.R.C.P.C. Background and Objectives: In view of the relatively few large studies available to estimate the rates of complications following regional anesthesia, we aimed to identify and quantify the risks that academic regional anesthesiologists and regional anesthesia fellows disclose to their patients before performing central and peripheral nerve blockade. Methods: We asked 23 North American regional anesthesia fellowship program directors to distribute a questionnaire to the regional anesthesiologists and regional anesthesia fellows at their institutions. The ques- tionnaire was designed to capture the risks and corresponding incidences that are routinely disclosed to patients before performing the most common central and peripheral nerve block techniques. Results: The total number of respondents was 79 from 12 different institutions. Fifty-eight (74%) respon- dents disclose risks of regional anesthesia in order to allow their patients to make an informed choice, whereas 20 (26%) disclose risks for medicolegal reasons. For central neural blockade, the most commonly disclosed risks are headache, local pain/discomfort, and infection. For peripheral nerve blockade, the most commonly disclosed risks are transient neuropathy, local pain/discomfort, and infection. For both central and peripheral nerve blockade, the risks most commonly disclosed are also those with the highest-reported incidences. Conclusions: The risks of regional anesthesia most commonly disclosed to patients by academic regional anesthesiologists and regional anesthesia fellows are benign in nature and occur frequently. Severe complica- tions of regional anesthesia are far less commonly disclosed. The incidences of severe complications disclosed by academic regional anesthesiologists and their fellows can be inconsistent with those cited in the contemporary literature. Reg Anesth Pain Med 2007;32:7-11. Key Words: Adverse effects, Anesthesia, Epidural, Spinal, Nerve block, Postoperative complications.

omplications after regional anesthesia (RA) are to capture the true incidences of such complica- Cuncommon. Unfortunately, prohibitively large tions.1 The American Society of Anesthesiologists numbers of patients are required for study in order (ASA) Closed Claims Project provides the largest collection of adverse events associated with modern RA practice in the United States;2 however, the lack of a denominator prevents the calculation of inci- See Editorial page 1 dence. The objective of this study is to identify and quantify the risks of RA that are routinely disclosed From the Department of Anesthesia and Pain Management, by academic regional anesthesiologists and their RA Toronto Western Hospital, University Health Network, Univer- fellows to patients in North American teaching hos- sity of Toronto, Toronto, Ontario, Canada; and Department of pitals. The information gathered may complement Anesthesia, The Hospital for Special Surgery, Weill Medical Col- lege of Cornell University, New York, NY. the relatively limited data available in the contem- Accepted for publication July 25, 2006. porary literature to produce a more accurate repre- Reprint requests: Richard Brull, M.D., F.R.C.P.C, Department of Anesthesia, Toronto Western Hospital, 399 Bathurst Street, Toronto, sentation of the risks associated with RA and allow Ontario, Canada M5T 2S8. E-mail: [email protected] other anesthesia practitioners to draw on the expe- © 2007 by the American Society of Regional Anesthesia and rience of experts during preoperative discussions Pain Medicine. 1098-7339/07/3201-0001$32.00/0 with their patients. Our hypothesis was that aca- doi:10.1016/j.rapm.2006.07.005 demic regional anesthesiologists and RA fellows

Regional Anesthesia and Pain Medicine, Vol 32, No 1 (January–February), 2007: pp 7–11 7 8 Regional Anesthesia and Pain Medicine Vol. 32 No. 1 January–February 2007 routinely disclose all significant risks and corre- Whitney U test with the Bonferroni correction for sponding incidences to their patients before per- multiple comparisons. forming central (CNB) or peripheral nerve blockade (PNB). Results The program directors from 12 institutions (9 Methods American and 3 Canadian) replied and agreed to participate in this study. Seven program directors After institutional review board approval (Uni- replied and agreed after the initial e-mail request, versity Health Network, Toronto, Ontario, Canada), whereas 5 replied and agreed after the reminder an information letter and questionnaire were sent e-mail request. No replies were received from the by electronic mail to all identifiable regional anes- program directors of the remaining 11 institutions. thesia fellowship program directors in North Amer- The total number of respondents (questionnaires ica on November 7, 2005. Twenty-three program returned) was 79 (70 attending anesthesiologists directors were identified from the Regional Anes- and 9 RA fellows). Fifty-eight (74%) respondents thesia Fellowship Program listings on the American answered that the primary reason for explaining Society of Regional Anesthesia and Pain Medicine regional anesthetic risks was to allow their patients website (www.ASRA.com) as well as the recently to make an informed choice regarding anesthetic published guidelines for RA Fellowship training.3 technique, whereas 20 (26%) answered for medi- The program directors were asked to distribute the colegal reasons. Among the 12 participating institu- questionnaire to “all practicing regional anesthesi- tions, 8 require a written consent form for anesthe- ologists and RA fellows” at their home institutions sia, whereas the remaining 4 institutions do not. and then return the completed questionnaires by For all 8 institutions that require a written consent mail or facsimile. After 8 weeks, a reminder mes- form for anesthesia, both general anesthesia (GA) sage was sent by electronic mail to those program and RA are addressed in a single form. The risks and directors who had not yet responded to the initial corresponding incidences routinely disclosed for request. spinal and epidural anesthesia are remarkably sim- The questionnaire was primarily designed to cap- ilar. For both spinal and epidural anesthesia, the ture the risks and corresponding incidences that are most commonly disclosed risks are headache, local routinely disclosed by the respondents to their pa- pain/discomfort, and infection (Table 1). Severe tients prior to performing the most common CNB complications of CNB, such as paralysis, cardiac and PNB techniques. From a list of complications arrest, and death, are far less frequently disclosed. for each CNB and PNB technique, the respondents For PNB, the most commonly disclosed risks are were instructed to select which risks they routinely transient neuropathy, local pain/discomfort, and disclose to their patients and indicate the corre- infection (Table 2). The 2 exceptions are axillary sponding incidence that is disclosed along with each block, where bruising is often disclosed (possibly risk according to a 6-point scale: (1) “greater than reflecting the transarterial technique), and ankle 1:10,” (2) “approximately 1:100,” (3) “approxi- block, where the risk of neuropathy is arguably mately 1:1,000,” (4) “approximately 1:10,000,” (5) rare. For both CNB and PNB, the risks most com- “approximately 1:100,000,” or (6) “less than monly disclosed are also those with the highest 1:1,000,000.” Respondents were encouraged to add likelihood of occurrence among all incidences rou- any risks (and corresponding incidences) that did tinely disclosed by our respondents (Tables 1 and not appear in the list. Additionally, respondents 2). When analyzed according to institutional coun- were asked to select the “primary reason” for dis- try of origin (United States v Canada), there were closing risks associated with RA from 2 options: (1) no significant differences for any of the responses in “to allow the patient to make an informed choice” the questionnaire. or (2) “for medicolegal reasons.” Finally, respon- dents were asked to select whether their institution Discussion required a “written consent form” for (1) “general Neurological complications of RA can be severe anesthesia,” (2) “regional anesthesia,” (3) “com- and potentially devastating to patients and their bined (general/regional anesthesia),” or (4) none of families. Candid disclosure and accurate quantifica- the above. tion of risks associated with RA are imperative to Data analysis was undertaken using SAS Version protecting both patients and anesthesiologists alike. 8.0 (SAS Institute Inc., Cary, NC). Categorical data Surprisingly, however, the results of our question- were analyzed by using the chi-square test. Non- naire suggest that relatively few regional anesthe- parametric data were analyzed by using the Mann- siologists disclose the severe risks of RA. For exam- Risk Disclosure for Regional Anesthesia • Brull et al. 9

ple, only 58% and 43% of academic regional 5 5 anesthesiologists routinely disclose the risks of per- 1:10 1:10 manent neuropathy and paralysis, respectively, to their patients undergoing CNB. Our survey did not 5 4 enable us to determine why some anesthesiologists 1:10 1:10 Arrest Death

Cardiac failed to disclose these risks. One possible explana- tion for this finding is the potential for discussions regarding anesthetic risk in the immediate preoper- 4 4 ative period to exacerbate the patients’ preoperative 1:10 1:10 Failure anxiety.4 Additionally, previous studies that have Respiratory

incidence of specified risk routinely examined which anesthetic risks patients would like to know about are conflicting.5,6 Many patients 4 3 prefer simple explanations about the main risks and 1:10 1:10 benefits, although a considerable number of pa- tients wish full-risk disclosure. Nonetheless, anes-

6 6 thesiologists have a duty to accurately disclose the significant risks of the proposed anesthetic to their 1:10 1:10 patients, including those that happen relatively fre- quently (e.g., local pain/discomfort) as well as those that happen rarely but are severe in nature (e.g., 5 5 permanent neuropathy and paralysis). 1:10 1:10 The complications of RA and their likelihood pre- Peripheral Neuropathy (permanent) Paralysis sented to the patient by the anesthesiologist likely influence the patients’ choice of anesthetic tech-

4 3 nique for their surgery. Disclosing an inflated rate

1:10 1:10 of complications may cause patients to opt for GA (transient) Peripheral

Neuropathy and forfeit the potential benefits of RA. Alterna- tively, failing to mention certain complications or deflating the rate of complications associated with 2 RA may lead the patient to choose an RA technique 1:10 when the patient would have chosen a GA had the Transient Symptoms Neurological risks been accurately disclosed. Importantly, among the anesthesiologists who do disclose the severe

2 2 risks of CNB, specifically, permanent neuropathy,

1:10 1:10 paralysis, respiratory failure, seizures, cardiac ar- rest, and death, the incidences disclosed are gener- ally in keeping with those cited in the contempo- 5 5 rary literature.7-10 The single exception is seizures Risks of Neuraxial Blockade Disclosed by Regional Anesthesiologists after CNB, which reportedly occur far less often (0.12-1.32:10,000)7-9,11 than what is disclosed by our respondents. For PNB, the contemporary liter-

Table 1. ature suggests that the incidence of severe compli- cations is considerably less common than that dis- closed by our respondents. For example, in 2 recent comprehensive prospective studies of complications after PNB, yielding 5,412 blocks in aggregate, there

Local were no cases of permanent neuropathy, 1 case of (1.8:10,000), and no cases of cardiac arrest Pain/Discomfort Bruising Infection Headache or death.12,13 Similarly, in the large and widely cited investigation of 43,552 PNBs (excluding lumbar plexus block) by Auroy and colleagues,9 there were 5 cases of seizure (1.1:10,000) and no cases of car- diac arrest or death. Unfortunately, Auroy’s study did not provide sufficient detail to determine the N%Incidence 1:10 55 73 1:10 36 1:10 48 55 74 69 90 45 59 37 49 44 58 32 43 5 7 10 14 13 18 23 32 N%Incidence 1:10 61 82 1:10 38 1:10 51 54 73 69 91 N/A 41 56 42 57 31 43 14 20 5 7 10 14 20 29 rate of permanent neuropathy after PNB. There are N, number of respondents who routinely disclose specified risk; %, percentage of total respondents who routinely disclose specified risk; incidence, Spinal anesthesia Epidural anesthesia

disclosed by respondents. The incidence is expressed as the mode calculated from all responses in aggregate; N/A, not applicable. at least 2 possible explanations for the discrepancy 0Rgoa nshsaadPi eiieVl 2N.1JnayFbur 2007 January–February 1 No. 32 Vol. Medicine Pain and Anesthesia Regional 10

Table 2. Risks of Peripheral Nerve Blockade Disclosed by Regional Anesthesiologists

Peripheral Peripheral Local Horner’s Neuropathy Neuropathy Respiratory Cardiac Pain/Discomfort Bruising Infection Syndrome (transient) (permanent) Paralysis Seizures Pneumothorax Failure Arrest Death

Interscalene block N544254555759152927231321 %715572717377213937321828 Incidence 1:10 1:102 1:104 1:10 1:102 1:104 1:104 1:103 1:103 1:103 1:105 1:105 Infraclavicular block N 45 39 45 N/A 47 43 N/A 17 29 N/A 11 18 %696070726628461830 Incidence 1:10 1:102 1:104 1:104 1:105 1:103 1:103 1:105 1:106 Axillary block N 55 55 51 N/A 54 51 N/A 21 N/A N/A 12 18 %7576707370301826 Incidence 1:10 1:10 1:104 1:104 1:104 1:104 1:104 1:105 Femoral block N 51 46 54 N/A 54 48 N/A 18 N/A N/A 12 19 %6862737163251727 Incidence 1:10 1:10 1:104 1:104 1:104 1:104 1:104 1:105 Popliteal block N 50 43 53 N/A 53 50 N/A 18 N/A N/A 11 18 %7061747369261626 Incidence 1:10 1:10 1:105 1:104 1:104 1:104 1:104 1:105 Ankle block N 55 43 43 N/A 34 32 N/A 8 N/A N/A 7 13 %7961624845121119 Incidence 1:10 1:10 1:104 1:104 1:104 1:103 1:105 1:105

N, number of respondents who routinely disclose specified risk; %, percentage of total respondents who routinely disclose specified risk; incidence, Incidence of specified risk routinely disclosed by respondents. The incidence is expressed as the mode calculated from all responses in aggregate; N/A, not applicable. Risk Disclosure for Regional Anesthesia • Brull et al. 11 between the incidence of severe complications dis- References closed by our respondents and that reported in the 1. Auroy Y, Benhamou D, Amaberti R. Risk assessment literature. The first may be that regional anesthesi- and control require analysis of both outcomes and ologists are less familiar with the contemporary lit- process of care. Anesthesiology 2004;101:815-817. erature than they should be. The second, perhaps 2. Lee LA, Posner KL, Domino KB, Caplan RA, Cheney more palatable explanation, is that much of the FW. Injuries associated with regional anesthesia in available literature is flawed, and the respondents the 1980s and 1990s: A closed claims analysis. Anes- are drawing on their own clinical experience to thesiology 2004;101:143-152. estimate and disclose the incidences of severe com- 3. Hargett MJ, Beckman JD, Liguori GA, Neal JM. plications associated with RA. Indeed, the question- Guidelines for regional anesthesia fellowship train- able validity of the available literature limits its role ing. Reg Anesth Pain Med 2005;30:218-225. in guiding discussions of risk with patients. The 4. White SM, Baldwin TJ. Consent for anaesthesia. An- aesthesia 2003;58:760-774. largest contemporary studies of risk associated with 5. Garden AL, Merry AF, Holland RL, Petrie KJ. Anaes- 2,7,14 RA are restricted to reviews of insurance claims thesia information—What patients want to know. and self-reporting by anesthesiologists,8-10 both of Anaesth Intensive Care 1996;24:594-598. which can result in misrepresentation of risk.1,15 6. Lonsdale M, Hutchison GL. Patients’ desire for infor- There are several important limitations of our mation about anaesthesia. Scottish and Canadian at- study. First, because the distribution of our sur- titudes. Anaesthesia 1991;46:410-412. vey was left to the discretion of the program 7. Aromaa U, Lahdensuu M, Cozanitis DA. Severe com- directors, we did not determine the total number plications associated with epidural and spinal anaesthe- of questionnaires distributed and therefore could sias in Finland 1987-1993. A study based on patient insurance claims. Acta Anaesthesiol Scand 1997;41:445- not calculate a true “response rate.” Moreover, 452. the rate of reply by the program directors was 8. Auroy Y, Narchi P, Messiah A, Litt L, Rouvier B, only 12 out of the 23 programs identified; how- Samii K. Serious complications related to regional ever, it is likely that some of the e-mail contact anesthesia: Results of a prospective survey in France. information gathered was outdated and/or the Anesthesiology 1997;87:479-486. programs inactive or discontinued. Furthermore, 9. Auroy Y, Benhamou D, Bargues L, Ecoffey C, Falis- although we recognize that the inclusion of RA sard B, Mercier FJ, Bouaziz H, Samii K. Major com- fellows may have skewed our results, any such plications of regional anesthesia in France: The SOS bias is likely minimal. Indeed, the number of RA Regional Anesthesia Hotline Service. Anesthesiology 2002;97:1274-1280. fellow respondents was very low compared to the 10. Moen V, Dahlgren N, Irestedt L. Severe neurological number of attending anesthesiologist respon- complications after central neuraxial blockades in Swe- dents. We, nonetheless, believe that including RA den 1990-1999. Anesthesiology 2004;101:950-959. fellows is important because fellows are often the 11. Brown DL, Ransom DM, Hall JA, Leicht CH, Schroe- ones charged with conducting the preoperative der DR, Offord KP. Regional anesthesia and local assessment and entering into discussions of risk anesthetic-induced systemic toxicity: Seizure fre- with their patients; such discussions should faith- quency and accompanying cardiovascular changes. fully reflect the practice of their expert instruc- Anesth Analg 1995;81:321-328. tors, namely the attending anesthesiologists. Fi- 12. Capdevila X, Pirat P, Bringuier S, Gaertner E, Singe- nally, the incidence of some complications lyn F, Bernard N, Choquet O, Bouaziz H, Bonnet F. Continuous peripheral nerve blocks in hospital wards disclosed by our respondents may not be gener- after orthopedic surgery: A multicenter prospective alizable beyond teaching centers. For example, analysis of the quality of postoperative analgesia and the incidence of headache routinely disclosed for complications in 1,416 patients. Anesthesiology 2005; either spinal or epidural anesthesia was curiously 103:1035-1045. similar and questionably high. 13. Fanelli G, Casati A, Garancini P, Torri G. Nerve stimu- In summary, our survey of risk-disclosure prac- lator and multiple injection technique for upper and tices among academic regional anesthesiologists lower limb blockade: Failure rate, patient acceptance, and RA fellows revealed that the most commonly and neurologic complications. Study Group on Re- disclosed risks of RA are benign in nature and gional Anesthesia. Anesth Analg 1999;88:847-852. occur frequently. Severe complications of RA are 14. Peng PW, Smedstad KG. Litigation in Canada against anesthesiologists practicing regional anesthesia. A re- far less commonly disclosed. The incidences of view of closed claims. Can J Anaesth 2000;47:105-112. severe complications disclosed by academic re- 15. Cullen DJ, Bates DW, Small SD, Cooper JB, Nemeskal gional anesthesiologists and their fellows can be AR, Leape LL. The incident reporting system does not inconsistent with those cited in the contemporary detect adverse drug events: A problem for quality im- literature. provement. Jt Comm J Qual Improv 1995;21:541-548. ƒ Editorials Informed Consent for Regional Anesthesia: What Is Necessary?

n this issue of Regional Anesthesia and Pain Medicine, Brull et al.1 described what Irisks were disclosed to patients by academic anesthesiologists and anesthesia fellows in North American regional-anesthesia fellowship programs in obtaining informed consent for regional anesthesia. Brull et al.1 found the most commonly revealed risks were benign in nature and occurred frequently. For central neural block, the most commonly disclosed risks were headache, local pain/discomfort, and infection, whereas for peripheral-nerve block, transient neuropathy, local pain/discomfort, and infection were mentioned. Severe complications of regional anesthesia, including permanent neurologic injury, paralysis, cardiac arrest, sei- zures, and death were only infrequently disclosed in the process of informed consent. For instance, a minority of anesthesiologists described risks of paralysis (43%), seizures (20%), cardiac arrest (14%), and death (29%) to patients un- dergoing epidural anesthesia.1 In addition, the authors found that the incidences of severe complications cited by the anesthesiologists were often inconsistent with the literature. Although the results of Brull et al.1 are probably typical of the practice of regional anesthesia, the results are concerning because of the ethical and legal requirement for informed consent. Informed consent requires an active commu- nication between the physician and patient, in which the physician explains the nature and purpose of the proposed procedure and the alternative techniques available, as well as a description of the risks and benefits of the procedures and alternatives. The desired outcome is that the patient will have sufficient knowl- edge to make an educated choice about whether to undergo the proposed pro- cedure or treatment. What types of risks need to be disclosed during the process of informed consent? Although the exact information to be transmitted varies from state to state, most states have adopted a “reasonable patient” standard.2-4 This standard requires the physician to disclose information that a reasonable patient under similar circumstances would want to know to make an informed decision. In- formed consent requires that a patient have a full understanding of that to which he or she has consented. The risks that should be disclosed are those that would be important in deciding whether to undertake the proposed therapy. For re- gional anesthesia, the disclosure should include both the common, but not severe, risks (e.g., local pain/discomfort, infection, headache, transient neuropathy) and the rare, but of major consequences, risks (e.g., seizure, cardiac arrest, permanent neuropathy, paralysis, and death). Certainly, a patient would want to know a risk of permanent neuropathy exists after an interscalene block performed for post- operative pain control, especially if he or she were an artist, pianist, or a surgeon. Interestingly, 3 out of 4 of the programs utilized a written informed consent for anesthesia, with most addressing general and regional anesthesia on a single form. Although the Joint Commission on Accreditation of Healthcare Organizations requires documentation of informed consent, it can be done by handwritten note, on the surgical consent, or on a separate written anesthesia consent form.2 Many attorneys believe that a written anesthesia-specific consent form, detailing se-

doi:10.1016/j.rapm.2006.10.001 See Brull et al. page 7

Regional Anesthesia and Pain Medicine, Vol 32, No 1 (January–February), 2007: pp 1–2 1 2 Regional Anesthesia and Pain Medicine Vol. 32 No. 1 January–February 2007

lected risks specific to the procedure will help in the defense of a claim for an injury.4,5 Notations to tailor the informed-consent discussion to a specific patient help to support the defense of the underlying medical issues.5 Although informed consent is seldom the major issue of liability in a claim,5,6 in a significant number of cases, the adequacy of informed consent is included as an additional allegation, thereby influencing the evaluation, defense, and resolution of a malpractice claim.5 However, patients can still allege lack of understanding of risks in the presence of a written consent form.4 Likewise, no scientific evidence indicates that separate anesthesia written informed consent is better than a handwritten note plus the written surgical consent.7 In summary, the article by Brull et al.1 brings to our consciousness the need to further educate our patients of important risks/benefits of all types of anesthesia, especially regional anesthesia. To make an informed decision, patients require accurate portrayal of rare, serious complications, as well as the more common minor ones.

Karen B. Domino, M.D., M.P.H. Chair, ASA Committee on Professional Liability Professor, Department of Anesthesiology University of Washington School of Medicine Seattle, WA

References 1. Brull R, McCartney CJL, Chan VWS, et al. Disclosure of risks associated with regional anesthesia: A survey of academic regional anesthesiologists. Reg Anesth Pain Med 2007; 32:7-11. 2. O’Leary CE. Informed consent for anesthesia: Has the time come for a separate written consent document? ASA Newsl 2006;70:11-12. 3. Manual for Anesthesia Department Organization and Management. Park Ridge, IL: American Society of Anesthesiologists; 2005:1-18. 4. Bierstein K. Informed consent and the Medicare Interpretive Guidelines. ASA Newsl 2006;70:13-14. 5. Sanford SR. Informed consent: The verdict is in. ASA Newsl 2006;70:15-16. 6. Caplan RA. Informed consent: Patterns of liability from the ASA Closed Claims Project. ASA Newsl 2000;64:7-9. 7. Cheney FW. A separate written consent document for anesthesia: What is the indica- tion? ASA Newsl 2000;70:17-18. The Role of Intrathecal Drugs in the Treatment of Acute Pain

James P. Rathmell, MD, Timothy R. Lair, MD, and Bushra Nauman, MD Department of Anesthesiology, University of Vermont College of Medicine, Burlington, Vermont

Intrathecal opioids are widely used as useful adjuncts for specific surgical procedures and the incidence and in the treatment of acute and chronic pain, and a num- treatment of side effects associated with these drugs. ber of non-opioid drugs show promise as analgesic Available data on the use of non-opioid drugs that have drugs with spinal selectivity. In this review we examine been tested intrathecally for use as analgesics are also the historical development and current use of intrathe- reviewed. Evidence-based guidelines for dosing of in- cal opioids and other drugs that show promise for treat- trathecal drugs for specific surgical procedures and for ing pain in the perioperative period. The pharmacology the treatment of the most common side effects associ- and clinical use of intrathecal morphine and other opi- ated with these drugs are presented. oids is reviewed in detail, including dosing guidelines (Anesth Analg 2005;101:S30–S43)

pioid analgesics have long been recognized as advanced cancer; they reported the first observations among the most effective treatments for pain. In of profound and prolonged pain relief with spinal O his treatise on gout and rheumatism (1), the 17th opioids. Since these first, bold clinical experiments, we century English physician Thomas Sydenham wrote: have witnessed a rapid transition from the laboratory “Among the remedies which it has pleased Almighty to clinical practice. Intrathecal opioids are now widely God to give man to relieve his suffering, none is so used as useful adjuncts in the treatment of acute and universal and so efficacious as opium.” chronic pain, and a number of non-opioid drugs show Despite their nearly universal ability to alleviate promise as analgesics with spinal selectivity. In this pain, opioids have a number of unpleasant, even life- review, we examine the current use of intrathecal threatening, side effects: nausea and vomiting, toler- opioids and other drugs that show promise for treat- ance, pruritus, urinary retention, and respiratory de- ing pain in the perioperative period. pression. For thousands of years, these medications were used without a known mechanism of action until 1971, when a class of highly specific opioid receptors Search Strategy was identified (2). Soon thereafter, opioid receptors were localized within the brain (3) and spinal cord (4). We performed a MEDLINE search for all indexed Evidence that direct application of morphine at the articles published between 1966 and March 2004 using spinal cord level could produce spinally mediated the search terms “postoperative pain,” “spinal or in- analgesia soon followed (5). Based on this limited trathecal,” “spinal analgesia,” “intrathecal analgesia,” experimental evidence, Wang et al. (6) administered “acute pain and spinal,” and “postoperative pain and bolus intrathecal morphine and Onofrio et al. (7) re- spinal,” resulting in 1436 articles. The abstracts of all ported chronic intrathecal morphine infusions, both articles were reviewed to select articles on the devel- working in patients with severe pain associated with opment and use of intrathecal drugs that are repre- sentative of the current state of our knowledge. Those articles representing the best available evidence were selected (e.g., randomized, controlled trials were used Financial support for preparation of this manuscript was pro- vided by an unrestricted educational grant from Endo Pharmaceu- in preference to observational studies when available, ticals, Chadds Ford, Pennsylvania. case reports or case series were used only when no Accepted for publication June 22, 2005. other information was available, and retrospective Address correspondence and reprint requests to James P. Rath- analyses were generally omitted); however, there was mell, M.D., Center for Pain Medicine, Fletcher Allen Health Care, 62 Tilley Drive, South Burlington, VT 05403. Address electronic mail to no attempt to assign qualitative scores to each article [email protected]. or to combine the available data quantitatively.

©2005 by the International Anesthesia Research Society S30 Anesth Analg 2005;101:S30–S43 0003-2999/05 ANESTH ANALG REVIEW ARTICLE RATHMELL ET AL. S31 2005;101:S30–S43 INTRATHECAL DRUGS FOR ACUTE PAIN

Pharmacology Drug disposition after intrathecal administration var- ies depending on the lipid solubility of the individual drug, and the most closely studied drugs are the opi- oid analgesics. After spinal administration, typically within the (CSF) in the lumbar cistern, the drug is distributed within CSF. Ummen- hofer et al. (8) reported a series of elegant experiments in which the concentration of morphine, fentanyl, sufentanil, and alfentanil were measured within CSF, spinal cord, epidural fat, and plasma in anesthetized pigs after lumbar intrathecal administration. From their data, they developed a multiple-compartment pharmacokinetic model that closely simulates the ob- Figure 1. Disposition of opioid after intrathecal administration. served pharmacology and explains many of the clini- After intrathecal administration, the disposition of opioids is com- cal characteristics of the opioids used for intrathecal plex and multicompartmental. Simultaneously, intrathecal opioids 1) travel cephalad within the cerebrospinal fluid (CSF), 2) enter the analgesia. The fate of opioids after intrathecal admin- spinal cord, where they may bind to nonspecific sites within the istration is complex (Fig. 1). Intrathecal opioids pene- white matter or specific opioid receptors within the dorsal horn, and trate the spinal cord and the dura mater to enter the 3) traverse the dura mater to enter the epidural space where they bind to epidural fat. From the spinal cord and epidural space, epidural space. Within the spinal cord, they bind to opioids enter the plasma compartment through vascular uptake. nonspecific sites within the white matter as well as to The clinical properties of each opioid (speed of onset, duration of specific receptors within the dorsal horn. Drug within action) and degree of rostral spread result from the sum of effects for each route. Lipophilic opioids (fentanyl/sufentanil) rapidly the spinal cord eventually reaches the plasma com- cross the dura, where they are sequestered in fat and gain rapid partment though venous uptake. In the epidural access to plasma; they also enter the spinal cord, where they may space, the opioid is sequestered in fat and enters the bind to nonspecific sites within the white matter as well as specific receptors within the dorsal horn and then enter the plasma. This plasma compartment via venous uptake. The sum of results in rapid onset, limited and brief rostral spread resulting in these multiple avenues for drug disposition results in early respiratory depression and a narrow band of analgesia sur- the clinical characteristics observed. Any drug given rounding the site of injection, and a relatively short duration of action. In contrast, the hydrophilic opioid morphine traverses the intrathecally rapidly redistributes within the CSF; opi- dura slowly to the epidural space where it binds little within epi- oid is detectable in the cisterna magna after lumbar dural fat and only slowly enters the plasma. Morphine also enters intrathecal administration within 30 min, even with the spinal cord and binds little to nonspecific receptors but largely to specific receptors within the dorsal horn, where uptake into the lipophilic drugs like sufentanil (9). Indeed, all opioids plasma occurs slowly. As a result of this limited and slow transfer move within the CSF and this rapid distribution from the CSF, morphine remains in relatively large concentration within the CSF likely accounts for the small, but sig- within the CSF; this results in slow onset, extensive and prolonged rostral spread resulting in delayed respiratory depression and a nificant, incidence of respiratory depression that is broad band of analgesia surrounding the site of injection, and a observed immediately after lumbar intrathecal injec- relatively long duration of action. tion (Fig. 2) (10). The lipophilic opioids rapidly traverse the dura where they are sequestered in the epidural fat and the epidural space; however, there is limited binding enter the systemic circulation; they also rapidly pene- to fat within the epidural space and limited binding to trate the spinal cord where they bind to both nonspe- nonspecific receptors within the spinal cord white cific sites within the white matter as well as dorsal matter. Transfer to the systemic circulation is likewise horn receptors and eventually enter the systemic cir- slower than the lipophilic drugs. Concentrations culation as they are cleared from the spinal cord. This within the CSF decline more slowly than similar doses rapid transfer from the CSF to both spinal cord and of lipophilic drugs, accounting for the greater degree the epidural fat accounts for the rapid onset and the of rostral spread, delayed respiratory depression (Fig. prompt decline in CSF levels of opioid, accounting for 2), and extensive dermatomal analgesia during the minimal rostral spread, lack of delayed respiratory chronic administration. This complex pharmacoki- depression, and relatively small dermatomal band of netic behavior explains why the location of injection analgesia seen during chronic administration; vascu- for intrathecal administration remains an important lar uptake accounts for the limited duration of anal- determinant in the pattern of analgesia observed (11). gesia of lipophilic opioids (Fig. 3). Hydrophilic drugs result in slow onset and a wide Morphine, the prototypic hydrophilic opioid, un- band of analgesia surrounding the site of injection. In dergoes a similar transfer to both the spinal cord and contrast, highly lipophilic drugs transfer rapidly from S32 REVIEW ARTICLE RATHMELL ET AL. ANESTH ANALG INTRATHECAL DRUGS FOR ACUTE PAIN 2005;101:S30–S43

Figure 2. The average onset and duration of respiratory depression after intrathecal opioid administration. Respiratory depression after intrathecal administration of lipophilic opioids (fentanyl and sufen- tanil) results from rostral spread immediately after injection, occur- ring within the first 20–30 min after injection. In contrast, the hydrophilic opioid morphine traverses from the cerebrospinal fluid (CSF) to the spinal cord and epidural space slowly. There is a small degree of respiratory depression immediately after injection that is similar to that seen with the lipophilic drugs and a later and more prolonged respiratory depression resulting from rostral migration of morphine within the CSF that peaks at approximately 6 h after injection and can persist for 18–24 h.

Figure 4. Spread of analgesia after intrathecal administration in the lumbar cistern. Values shown in parentheses are octanol:water par- tition coefficients; a larger value indicates that a drug is more lipid soluble. Lipophilic opioids result in a narrow band of analgesia whereas hydrophilic opioids produce a broader band of analgesia. In contrast, intrathecal administration of a lipophilic drug in the lumbar cistern results in a broad, but short-lived, band of analgesia limited to the lumbar dermatomes similar to that seen with chronic infusion. Despite these differing patterns of analgesia, all opioids move within the cerebrospinal fluid (CSF) and are detectable in the CSF adjacent to the brainstem soon after injection. Indeed, life- threatening respiratory depression has been reported within 20 min after intrathecal injection of fentanyl, sufentanil, and meperidine (10).

Figure 3. The average onset and duration of analgesia after intra- (12), benzyl alcohol and the parabens have been im- thecal opioid administration. Lipophilic opioids (fentanyl and plicated as the cause of neurotoxicity after intrathecal sufentanil) produce rapid onset of analgesia lasting from 2–4 h, administration (13). whereas hydrophilic opioids (morphine) are slower in onset with a duration of analgesia extending 18–24 h. the CSF and result in a broad band of analgesia that is Morphine much shorter in duration and the clinical observation In early trials with intrathecal morphine, doses ranged that analgesia during prolonged infusion is limited to from 500–1000 ␮g and profound sedation and respi- a narrower band of analgesia surrounding the ana- ratory depression were not uncommon (14). In a care- tomic site of administration during epidural adminis- fully controlled study in healthy volunteers, profound tration (Fig. 4). The clinical dose range, onset, and and prolonged respiratory depression was observed duration of activity for various intrathecal opioids are in all subjects who received 600 ␮g of intrathecal mor- shown in Table 1. phine (15). In more recent years, several trials have Internationally, opioids and adjuvant analgesics are examined doses as small as 40 ␮g, typically, extending supplied in varying concentrations and in prepara- only as large as 300 ␮g. Indeed, it appears that the tions that include preservatives. Although the toxicity efficacy of doses above this range is often limited by side of the most common preservatives appears to be small effects. ANESTH ANALG REVIEW ARTICLE RATHMELL ET AL. S33 2005;101:S30–S43 INTRATHECAL DRUGS FOR ACUTE PAIN

Table 1. Pharmacologic Properties of Common Opioids used for Intrathecal Analgesia Usual dose range Onset Duration Opioid (␮g) (min) (h) IT:IV potency ratio Morphine 100–500 45–75 18–24 1:200 Fentanyl 5–25 5–10 1–4 1:10 Sufentanil 2.5–10 5–10 2–6 1:10

In recent years, clinical investigation surrounding Other Opioids use of intrathecal morphine in the perioperative pe- riod has centered on establishing the optimal dose for Fentanyl and Sufentanil specific surgical procedures. Rathmell et al. (16) com- Fentanyl and sufentanil are the best studied and most pared the need for supplemental IV morphine via commonly used lipophilic drugs for intrathecal deliv- patient-controlled analgesia after doses of intrathecal ery. Two major trends in use of these drugs as anal- morphine ranging from 0–300 ␮g after total hip and gesics have evolved in recent years: the first is a closer knee arthroplasty. After total hip replacement, intra- definition of their role as analgesics for labor, delivery, thecal morphine (200 ␮g) provided excellent analge- sia. In contrast, the degree of pain experienced after and postcesarean delivery; the second is the recogni- total knee arthroplasty exceeded the analgesia af- tion that addition of small doses of lipophilic opioids forded by even the largest dose of intrathecal mor- during spinal anesthesia for ambulatory procedures phine (300 ␮g), yet patients who received this dose can produce more rapid onset and better quality sur- nearly universally reported nausea, vomiting, and gical block and lead to more rapid recovery of motor pruritus. Intrathecal morphine has been studied as an function and allow for earlier discharge after surgery. analgesic for pain after cesarean delivery (17), various Both drugs can be summarized as having a rapid orthopedic procedures (16), and a range of other sur- onset of analgesia (10–15 min) with a short duration of gical procedures (18) from spinal (19) to cardiac sur- action (2–5 h). gery (20). Two generalizations emerge. The first and Bucklin et al. (22) performed a meta-analysis of 7 most important is that the optimal dose of intrathecal randomized, controlled trials and concluded that in- morphine depends on the specific surgical procedure, trathecal opioids (including morphine 200 ␮g, sufen- with doses Ͻ100 ␮g often sufficing for pain control tanil 2–10 ␮g, and fentanyl 25 ␮g) provide comparable after cesarean delivery, whereas doses in the area of analgesia 15–20 min after injection in early labor with 500 ␮g may be required for more extensive surgery, a more frequent incidence of pruritus compared with such as or open abdominal aortic aneu- epidural local anesthetics. Duration of analgesia de- rysm repair. The best available evidence relating to pends on the stage of labor. Viscomi et al. (23) found optimal dosing for specific procedures is presented in that 10 ␮g intrathecal sufentanil, combined with Table 2. The second generalization is that the inci- 2.5 mg of bupivacaine, provided analgesia that was dence of side effects increases in proportion to the significantly shorter in duration in advanced labor ␮ dose administered, with doses in excess of 300 g (120 Ϯ 26 min) compared with early labor (163 Ϯ producing nausea, vomiting, pruritus, and urinary re- 57 min). Intrathecal sufentanil is approximately 4.5 tention in most recipients. Indeed, there appears to be times more potent than intrathecal fentanyl in labor- a ceiling analgesic effect for intrathecal morphine ing parturients (24). Single-shot techniques provide above which the risk of side effects outweighs the predictable analgesia but of limited duration. In ob- benefits of improved analgesia. Despite the excellent stetrics, where prolonged labor and operative delivery analgesia afforded by use of intrathecal morphine, often occur, many practitioners now take advantage of carefully controlled studies in high-risk patients have failed to demonstrate any beneficial effects on out- the best of both spinal and epidural techniques using come in terms of reduced perioperative renal, pulmo- combined spinal-epidural analgesia (CSE). In a pro- nary, and cardiac complications or mortality (21). The spective, randomized comparison of CSE versus epi- prolonged duration of action along with the risk of dural analgesia for labor, Norris et al. (25) found that late respiratory depression associated with intrathecal progress and outcome of labor were similar in women ␮ morphine point to the need for hospitalization and receiving10 g intrathecal sufentanil with 2.0 mg bu- monitoring after administration. This drug is not suit- pivacaine or 10 ␮g epidural sufentanil and 12.5– able for ambulatory procedures. Preservative-free 25.0 mg bupivacaine followed by continuous epidural morphine is currently the only drug discussed in this infusion of 0.083% bupivacaine plus 0.3 ␮g/mL sufen- review that is approved by the United States Food and tanil. Intrathecal opioids also provide effective anal- Drug Administration for intrathecal use in the treat- gesia after cesarean delivery. In a review of the use of ment of acute pain. intrathecal lipophilic opioids as adjuncts for surgical S34 REVIEW ARTICLE RATHMELL ET AL. ANESTH ANALG INTRATHECAL DRUGS FOR ACUTE PAIN 2005;101:S30–S43

Table 2. Optimal Intrathecal (IT) Opioid Dose for Specific Surgical Procedures

Procedure Optimal IT opioid and dose Comments

Labor analgesia Sufentanil 2.5–5 ␮g Larger doses (Ͼ7.5 ␮g) are associated with an increased incidence of fetal bradycardia (129). Cesarean delivery Morphine 100 ␮g Fentanyl (10–40 ␮g) and sufentanil (10–15 ␮g) improve intraoperative analgesia but do not produce significant postoperative analgesia (15). Outpatient procedures under spinal Fentanyl 10–25 ␮g These intrathecal lipophilic opioids speed onset of block and improve anesthesia (e.g., knee arthroscopy) Sufentanil 5–12.5 ␮g both intraoperative and immediate postoperative analgesia without prolonging motor block (18). Transurethral resection of the prostate Morphine 50 ␮g This small dose of intrathecal morphine was equivalent to 100 ␮g after (TURP) TURP (130). Major orthopedic surgery (e.g., total Morphine 200–300 ␮g Although these doses of intrathecal morphine provide excellent joint arthroplasty) analgesia after total hip arthroplasty they are inadequate for pain relief after total knee arthroplasty, reflecting the greater degree of pain reported by patients undergoing knee replacement (16). Thoracotomy Morphine 500 ␮g Lumbar intrathecal morphine improves pain relief and reduces but does not eliminate the need for supplemental IV opioid analgesics (131). Fast-track cardiac surgery Morphine 500–600 ␮g(8␮g/kg) Lumbar intrathecal morphine administered prior to elective cardiac surgery combined with remifentanil/desflurane anesthesia provided superior analgesia when compared to a sufentanil/desflurane anesthetic (132). Major abdominal/vascular surgery (e.g., Morphine 500–600 ␮g Lumbar intrathecal morphine provided more intense analgesia than IV open abdominal aortic aneurysm patient-controlled analgesia with morphine (21). repair) spinal anesthesia, Hamber and Viscomi (18) recom- analgesia without prolonging motor block or delaying mended using 20–30 ␮g fentanyl or 5–7.5 ␮g of sufen- discharge. tanil to supplement bupivacaine spinal anesthesia for cesarean delivery. The addition of these doses of in- Meperidine trathecal opioid led to faster onset of block, improved intraoperative and postoperative analgesia that lasted Meperidine produces a local anesthetic effect in addi- 2-5 h, and decreased nausea and vomiting during tion to its properties as an opioid analgesic (28). It has cesarean delivery. been reported to provide similar surgical anesthesia Much recent work has focused on the addition of for perineal and lower extremity surgery as the sole intrathecal lipophilic opioids to local anesthetics dur- anesthetic compared to bupivacaine (29). The duration ing spinal anesthesia for brief outpatient surgery. In of sensory block after intrathecal administration of this setting, adding fentanyl or sufentanil to bupiva- meperidine was 112 Ϯ 19 min in those receiving caine or lidocaine results in more rapid onset of block 1.5 mg/kg compared with 79 Ϯ 27 min in those re- and improved analgesia, both intraoperatively and for ceiving 1.2 mg/kg; respiratory depression within the first several hours after surgery (18). In efforts to 5–50 min of administration, hypotension (Ͼ30% de- take advantage of the improved analgesia without cline in systolic arterial blood pressure), nausea, and motor blockade, a series of investigations have exam- vomiting were all common side effects (30). The addi- ined combining lipophilic opioids with small doses of tion of meperidine 10 mg to intrathecal bupivacaine local anesthetic. Ben-David et al. (26) reported that for cesarean delivery was associated with prolonged patients receiving 20 mg of 0.5% lidocaine in dextrose postoperative analgesia but with more frequent intra- with 20 ␮g fentanyl for knee arthroscopy were ready operative nausea and vomiting (31). Another random- for discharge an average of 45 min after placement of ized trial examining use of intrathecal meperidine (15 the spinal drugs (range, 28–180 min ). Likewise, ad- or 25 mg) for CSE during labor was halted because of dition of fentanyl (10 ␮g) to small doses of hyperbaric the frequent incidence of the nausea and vomiting bupivacaine (5 mg) enhanced the quality and duration associated with this drug (32). Adding small doses of of sensory block without prolonging the intensity or meperidine (0.3 mg/kg) to spinal lidocaine also pro- duration of motor block in patients undergoing knee longs postoperative analgesia after transurethral re- arthroscopy (27). The addition of intrathecal fentanyl section of the prostate (33). Despite the theoretic ap- (10–25 ␮g) to surgical spinal anesthetics hastens the peal of meperidine as a drug that can provide both onset of surgical anesthesia, enhances intraoperative opioid and local anesthetic effects, it has gained lim- analgesia, and provides several hours of postoperative ited popularity owing to the frequent association with ANESTH ANALG REVIEW ARTICLE RATHMELL ET AL. S35 2005;101:S30–S43 INTRATHECAL DRUGS FOR ACUTE PAIN

nausea and vomiting. Detailed dose-response studies Non-Opioids as Adjuvants and Analgesics examining analgesic effects at smaller intrathecal for Intrathecal Use doses are not available. Clonidine ␣ ␣ Clonidine is a specific 2-receptor agonist; 2- receptors are present within both presynaptic and Long-term, continuous infusion of hydromorphone postsynaptic terminals of primary afferent nociceptive was not associated with spinal cord toxicity in an neurons within the dorsal horn of the spinal cord. animal model (34) and has gained popularity and ␣ Spinal 2-receptor agonists alter pain transmission by acceptance as an alternative to morphine as an anal- binding presynaptically to nociceptive A␦ and C fiber gesic drug for treatment of chronic pain using contin- terminals and reducing neurotransmitter release, and uous intrathecal systems (35). Hydro- postsynaptically by hyperpolarizing second order morphone has also been used effectively as a neurons within the dorsal horn (44). Clonidine pro- neuraxial drug for continuous epidural analgesia after ␣ duces analgesia by activating 2-receptors; direct in- thoracotomy (36), prostatectomy (37), and spinal fu- trathecal administration produces selective spinally- sion (38). Liu et al. (37) randomized 16 patients who mediated analgesia (45,46). Spinal clonidine produces had undergone prostatectomy to receive hydromor- dose-dependent analgesia with hypotension, brady- phone via patient-controlled analgesia either via an cardia and sedation; it is not associated with respira- epidural or an IV route and measured the effective- tory depression or pruritus (47). ness and dose requirements. Patients receiving hydro- Clonidine has demonstrated effects on reducing morphone via the IV route required twice as much both nociceptive and neuropathic pain in experimen- hydromorphone as those receiving the same drug via tal models and in clinical use. Eisenach et al. (48) used Ϫ Ϫ an epidural route (approximately 2 ␮g·kg 1 ·h 1 intradermal capsaicin-induced hyperalgesia to pro- Ϫ Ϫ versus 4 ␮g·kg 1 ·h 1 in the IV versus epidural duce central hypersensitivity and allodynia and nox- groups, respectively, during hours 3–24 after surgery). ious heat to produce acute pain in a series of healthy There are only limited data on intrathecal dosing of volunteers. They found that 150 ␮g of clonidine ad- hydromorphone in the acute setting. Drakeford et al. ministered intrathecally, but not IV, relieved pain and (39) randomized 60 patients undergoing total joint allodynia, supporting a spinal selective mechanism for arthroplasty to receive either , morphine 500 ␮g, analgesia. In a placebo-controlled, randomized study, or hydromorphone 2 ␮g/kg intrathecally. Both mor- patients with severe despite large doses of ␮ phine and hydromorphone produced significantly opioids received epidural clonidine (30 g/h) or pla- better postoperative pain relief with similar incidence cebo (49). Epidural clonidine provided significant pain of side effects when compared with saline. Like mor- relief, particularly in patients with neuropathic pain. phine, systemic doses of hydromorphone required to Reports on the use of intrathecal clonidine in the produce similar analgesia after IV administration are perioperative period are numerous and collectively several hundred times the doses required after intra- point toward synergistic action with spinal local an- thecal administration (40) and produce similar dura- esthetics (50) with less urinary retention than spinal morphine (51). Clonidine has been demonstrated in tion of analgesia and side effects. The limited available numerous studies to prolong sensory and motor data suggest that intrathecal hydromorphone 50–100 block. Eisenach et al. (47) reviewed the use of ␮g produces analgesia and side effects similar to 100– clonidine for regional anesthesia in 1996. Their sum- 200 ␮g of intrathecal morphine with similar side ef- mary analysis of numerous studies found that fects and duration of action. clonidine 75–225 ␮g (average, 146 ␮g) added to spinal bupivacaine 13.75–15 mg prolonged sensory block Methadone from 2.5 to 3.7 h and motor block from 2.4 to 3.3 h (47). Clonidine intensifies the degree of sensory block, re- The d-enantiomer of methadone is a weak opioid with duces tourniquet pain (52), and appears to prolong low affinity to the N-methyl-d-aspartate (NMDA) re- and intensify the effects of spinal local anesthetic by ceptor, a class of receptors with potential analgesic altering systemic absorption (53). The degree of sym- effects. However, in the antinociceptive dose range, patholysis and hypotension after typical spinal local the NMDA antagonism does not appear to contribute anesthetic doses (e.g., 15 mg bupivacaine) is near max- to the mechanism of d-methadone analgesia (41). imal; thus, adding intrathecal clonidine results in no There are only limited, uncontrolled reports of use of increased hypotension (47). Hemodynamic effects of intrathecal methadone for intrathecal use via continu- clonidine after neuraxial administration begin within ous infusion for chronic pain (42) and acute postoper- 30 min, reach maximum effect within 1–2 h, and last ative pain (43). approximately 6–8 h after a single injection (47). S36 REVIEW ARTICLE RATHMELL ET AL. ANESTH ANALG INTRATHECAL DRUGS FOR ACUTE PAIN 2005;101:S30–S43

There are few studies examining clonidine in com- (61). Adenosine produced no effect on acute thermal bination with intrathecal opioids. Sites et al. (54) found or chemical pain but reduced mechanical hyperalgesia that combining intrathecal morphine (250 ␮g) with and allodynia from intradermal capsaicin for at least intrathecal clonidine (25 or 75 ␮g) reduced the need 24 h (62). Adenosine concentrations in CSF increased for supplemental analgesics and improved pain con- after intrathecal, but not IV, administration of opioids, trol after total knee arthroplasty. Side effects were suggesting a role for adenosine in spinal opioid recep- similar with the exception that patients receiving tor activation and analgesia (63). A randomized study clonidine had more hypotension during the first 6 h of 25 parturients who received 10 ␮g of intrathecal after surgery. sufentanil with or without 500 ␮g of adenosine showed no differences in the degree or duration of Neostigmine pain relief (64). Further clinical trials of intrathecal adenosine are warranted; its role as an analgesic will Spinal administration of neostigmine, an acetylcho- likely be limited to injury associated with acute hy- linesterase inhibitor, inhibits breakdown of the endog- peralgesia (e.g., surgery) and treatment of neuropathic enous neurotransmitter acetylcholine, thereby induc- pain (62). ing analgesia (50). Eisenach et al. (55) demonstrated that acetylcholine has intrinsic analgesic properties, and that the concentration of acetylcholine in CSF is Epinephrine increased during painful electrical stimulation. They Adding the vasoconstrictor epinephrine (0.1–0.6 mg) further suggested that enhanced amounts of acetyl- to spinal local anesthetics intensifies and extends the choline released from preganglionic sympathetic neu- duration of sensory and motor block in a dose- rons after spinal neostigmine administration may dependent fashion (65,66). Epinephrine may produce counteract the sympatholytic actions of local anesthet- both direct analgesia through ␣-adrenergic receptor ics or ␣ -agonists (reducing the degree of hypoten- 2 binding as well as prolongation of local anesthetic sion) and add a synergistic antinociceptive effect to effect by vasoconstriction and decreased clearance. spinal ␣ -agonists (55). 2 Adding epinephrine to intrathecal opioid (sufentanil, In preliminary dose-ranging studies in surgical pa- fentanyl) has no effect on the intensity or duration of tients and healthy volunteers, intrathecal neostigmine analgesia for labor (67,68) but may increase the inci- 10–50 ␮g provided analgesia (56). Nausea and lower dence of nausea and vomiting (67). Using epinephrine extremity weakness were common in doses exceeding is a safe and effective means of prolonging and inten- 100–150 ␮g, but sedation, pruritus, respiratory de- sifying spinal anesthesia, although at the expense of pression, and hemodynamic changes did not occur delayed return of motor function and micturition (69). (57). Adding 6.25–25 ␮g neostigmine to spinal bupiv- Thus, epinephrine is an unsuitable adjunct for use in acaine improves sensory and motor block (56), delays the ambulatory setting. resolution of block (56), and reduces postoperative analgesic requirements (58). All doses were associated with a frequent incidence of nausea and vomiting that Ketorolac was resistant to pharmacologic treatment (56,58). Cyclooxygenase-2 (COX-2) activity in the spinal cord One study (59) demonstrated that adding 1–5-␮g plays a key role in sensitization to sensory stimuli neostigmine to 100 ␮g intrathecal morphine improved during acute (70), but intrathecal ad- analgesia without increasing side effects after gyneco- ministration of COX-2 specific inhibitors has minimal logical surgery. The incidence of nausea and the pro- analgesic effects in an incisional model of postopera- longation of recovery from spinal anesthesia suggest tive pain (71). Studies in experimental animals sug- that neostigmine is not a useful adjuvant; further ex- gests that COX-1 plays an important role in spinal amination of small dose neostigmine in combination cord pain processing and sensitization after surgery with intrathecal opioids is warranted. and that spinally administered specific COX-1 inhibi- Adenosine tors may be useful to treat postoperative pain (72). In a phase I safety assessment of intrathecal ketorolac in Adenosine produces receptor-specific analgesia via volunteers, a single 0.25–2 mg intrathecal dose of ke- both peripheral and central mechanisms, and adeno- torolac did not affect sensory or motor function or sine receptors are present in high density within the deep tendon reflexes, and there were no subjective dorsal horn of the spinal cord (60). After animal tox- sensations, neurologic or otherwise, reported by the icity testing suggested safety, a phase I dose-ranging volunteers. Ketorolac did not reduce pain reported trial of 0.25–2 mg of intrathecal adenosine in healthy from heat stimuli applied to the lateral calf (73). Intra- volunteers showed no effect on arterial blood pres- thecal ketorolac lacks efficacy in normal rats subjected sure, end-tidal carbon dioxide, or neurologic function; to acute, noxious heat stimuli but enhances the antino- headache and back pain were common side effects ciceptive effects of clonidine (74). Further study is ANESTH ANALG REVIEW ARTICLE RATHMELL ET AL. S37 2005;101:S30–S43 INTRATHECAL DRUGS FOR ACUTE PAIN

needed to define the role of intrathecal ketorolac in fetal adverse effects. Current reports suggest that the use treating acute pain. of in a dose not exceeding 1–2 mg at a concentration not exceeding 1 mg/mL, delivered either Midazolam alone or as an intrathecal adjuvant, has positive effects on perioperative pain and does not increase the inci- There has been much recent attention toward the use dence of adverse events (75). Although current formula- of the benzodiazepine midazolam as an intrathecal tions are preservative-free, commercially available stock drug in the treatment of both acute and chronic pain. solutions of midazolam hydrochloride are supplied in Yaksh and Allen (75) recently reviewed the existing concentrations of 5 mg/mL at a pH 3.4–3.6 (75). The animal and human data regarding use of intrathecal solubility rapidly declines to Ͻ1 mg/mL at pH 4.5–5, midazolam. Basic science work with ␥-amino butyric and cloudiness (precipitation) has been reported when acid suggests that it may play an important role in the 5 mg/mL is diluted with higher pH saline or CSF regulating primary as well as dorsal (sensory) and (80). motor horn excitability. Preclinical studies in animal models have demonstrated significant analgesia with- out changes in sympathetic outflow. At larger doses Complications Associated with Intrathecal (typically 3 or more times the dose required to pro- duce analgesic effects), reversible degradation of mo- Opioid Use tor strength and coordination appear. Early toxicity Pruritus studies in rabbits (76) suggested that midazolam pro- duced significant spinal cord toxicity, even after Pruritus after intrathecal administration of opioids is single-shot administration in clinically relevant doses. common and occurs more often than after systemic However, recent animal studies examining the effects administration. Szarvas et al. (81) reviewed the patho- of long-term intrathecal administration of midazolam physiology and treatment of neuraxial opioid-induced showed no discernible toxicity (77). Yaksh and Allen pruritus. The incidence of pruritus after intrathecal (75) conclude that current data “. . .support the asser- administration of opioid varies from 30% to 100% tion of a degree of safety for this modality [intrathecal (82–87). The incidence among the commonly used midazolam] within the doses and concentrations intrathecal opioids (morphine, fentanyl, sufentanil) examined.” has been reported to be similarly frequent (88,89). The The use of intrathecal midazolam in humans has exact mechanism of neuraxial opioid-induced pruritus been reported in at least 18 peer-reviewed reports remains unclear (83). ’s reversibility of pru- with an estimated 797 patients (75) since 1986. The ritus supports the existence of an opioid receptor- overall clinical effects are characterized by an in- mediated central mechanism. The mechanism does creased duration of sensory and motor block when not appear to be histamine related (90). Pharmacolog- administered with spinal local anesthetic, an increase ical therapies include antihistamines, 5-HT3-receptor in time to first request for supplemental analgesia antagonists, opiate antagonists and combination postoperatively, and a decrease in postoperative anal- agonist-antagonists, propofol, and nonsteroidal anti- gesic requirements. There appears to be no increase in inflammatory drugs (83). Histamine is not released adverse effects, including hypotension, bradycardia, and does not appear to be causative (91). Antihista- micturition, or nausea/vomiting, when midazolam is mines are thus unlikely to have any role in prevention. combined with another intrathecal local anesthetic The sedative properties may be helpful in interrupting and/or opioid compared with groups not receiving the -scratch cycle but without relieving itch sensa- intrathecal midazolam. Tucker et al. (78) reported a tion (92). Despite its widespread use, diphenhydra- prospective observational study of 1100 patients who mine has little demonstrated efficacy in the treatment underwent various surgical procedures with spinal of neuraxial opioid-induced pruritus (84,90). Ondan- anesthesia with or without the addition of 2 mg of setron has demonstrated efficacy in both prevention intrathecal midazolam. Intrathecal midazolam was not and treatment of pruritus associated with neuraxial associated with an increased risk of symptoms sugges- opioids (84,93). The opioid antagonists naloxone and tive of neurological impairment, including motor or sen- naltrexone (91), as well as the agonist-antagonist nal- sory changes and bladder or bowel dysfunction. Tucker buphine (94), are the most effective drugs for preven- et al. (79) randomized 30 parturients to receive intrathe- tion and treatment of neuraxial opioid-induced pruri- cal midazolam 2 mg, fentanyl 10 ␮g, or the combination tus. When the pure antagonists are used in larger of both drugs. Labor pain was not altered by midazolam doses, they also reverse analgesia (94). Nalbuphine alone, was modestly reduced by fentanyl alone, and was appears to be the most effective drug when compared reduced most by the combination of the two drugs. They with naloxone, diphenhydramine, and ondansetron concluded that intrathecal midazolam enhanced the an- (94,95). Propofol, in subhypnotic doses, has also algesic effect of fentanyl without increasing maternal or proven effective in the prevention and treatment of S38 REVIEW ARTICLE RATHMELL ET AL. ANESTH ANALG INTRATHECAL DRUGS FOR ACUTE PAIN 2005;101:S30–S43

neuraxial opioid-induced pruritus. Propofol, 10 mg added to spinal bupivacaine (107,108). Others have bolus followed by 30 mg/24h infusion (96) or 10 mg demonstrated a dose-dependent increase in nausea without infusion (97), markedly reduced the incidence and vomiting with intrathecal morphine (109,110). To- of pruritus. However, studies in obstetric populations gether, these data suggest that intrathecal morphine have failed to show any effect of similar doses of does not increase the incidence of nausea and vomit- propofol on neuraxial opioid-induced pruritus (98,99). ing after major surgery when compared with systemic Treatment of opioid-induced pruritus remains a chal- administration of morphine, particularly if the dose is lenge. Ondansetron, propofol, and nalbuphine have Ͻ100 ␮g (111). In contrast, fentanyl and sufentanil proven efficacy in the treatment of neuraxial opioid- have been associated with little or no nausea or vom- induced pruritus (Table 3). iting after intrathecal administration of a single dose (103,112,113). Urinary Retention Nausea and vomiting induced by neuraxial opioids may be a systemic effect, particularly with lipophilic Urinary retention after administration of intrathecal opioids, or may be the result of cephalad migration of opioids is common. This side effect can be observed drug in the CSF and subsequent interaction with opi- soon after intrathecal injection of morphine and lasts oid receptors located in the area postrema (111). Sen- for 14–16 h, regardless of the dose used (100). The sitization of the vestibular system to motion and de- incidence of urinary retention appears to be most fre- creased gastric emptying produced by opioids may quent with intrathecal morphine at ϳ35% (101) and is also play a role. more common after neuraxial administration than af- For shorter or less painful procedures, use of li- ter IV or IM administration (102). The incidence does pophilic opioids will minimize the risk of nausea and not appear to be dose related but is more frequent vomiting. Both fentanyl and sufentanil were shown to after intrathecal morphine than after lipophilic opioids decrease the rate of intraoperative nausea and vomit- (103,27). Indeed, intrathecal sufentanil alone for litho- ing during cesarean delivery when compared with tripsy was associated with a shorter time to voluntary local anesthetic alone for spinal anesthesia (18). Dexa- micturition than spinal lidocaine (104). methasone and droperidol have been shown to be Opioids affect urination via several mechanisms in- effective for prevention of nausea and vomiting after cluding alteration of parasympathetic tone and central epidural morphine (114). Combinations of scopol- analgesic action, which modify the pain threshold for amine and promethazine used as preventative meas- the bladder and contribute to retention (100). Urinary ures decreased the incidence of nausea and vomiting retention secondary to neuraxial opioids is likely re- (111). lated to interaction with opioid receptors located in the sacral spinal cord. This interaction promotes inhi- bition of sacral parasympathetic nervous system out- Respiratory Depression flow, which causes marked detrusor muscle relaxation The most feared complication of opioid administra- and an increase in maximal bladder capacity (100). tion is respiratory depression. The pharmacology and Naloxone can prevent or reverse urodynamic changes time course of opioid-induced respiratory depression after neuraxial morphine; however the dose required have been discussed previously in this review (Fig. 2). may partially or completely reverse analgesia (100). The true incidence of clinically significant respiratory Nalbuphine may also reverse the urinary effects of depression is not known, but evidence from smaller neuraxial morphine (105). If patients are unable to controlled trials and large observational studies con- void for Ն6 h, should be per- firm that it is infrequent (115). The majority of pro- formed to prevent myogenic bladder damage result- spective studies of epidural morphine have not de- ing from prolonged distention. Because urinary reten- tected clinically significant respiratory depression, but tion is infrequent with use of lipophilic intrathecal they are hindered by relatively small sample sizes and opioids, they are the preferred adjuvants for outpa- are markedly underpowered to detect such a rare tient surgery with spinal anesthesia (18). event (116). Ko et al. (117) reviewed the use of the term Nausea and Vomiting “respiratory depression” and found that there is no clear definition, leading to difficulty and confusion All opioids, regardless of , can when comparing available studies. The incidence is produce nausea and vomiting. The incidence after infrequent for doses commonly used clinically but the neuraxial administration is approximately 30% (106), incidence is dose-dependent for both hydrophilic and but the incidence varies with the particular opioid lipophilic opioids (118). The incidence of respiratory used and, in some settings, also varies with the dose depression associated with continuous epidural infu- administered. Intrathecal morphine led to a dose- sions containing opioids has been estimated from dependent increase in vomiting in volunteers (15) but large observational studies, with estimates ranging caused no increase in nausea and vomiting when from 0.09% to 0.4% (119–125). The risk of respiratory NSHANALG ANESTH 2005;101:S30–S43

Table 3. Incidence, Proposed Mechanisms, and Treatment for Intrathecal Opioid-Related Side Effects

Complication Incidence Proposed mechanism Treatment Commentary Pruritus • Varies from 30%–100% Exact mechanism unclear. Postulates • Ondansetron 4–8 mg IV • Propofol may be less effective for • Increased in parturients include “itch center” in the central • Nalbuphine 4 mg IV the parturient • Varies with opioid used nervous system, medullary dorsal horn • Propofol 10 mg IV bolus ϩ small dose • Histamine release does not appear • Likely dose dependent activation, antagonism of inhibitory infusion to be causative • Increased with epinephrine transmitters. Modulation of the • Infusion of naloxone serotonergic pathway and involvement • Oral naltrexone of prostaglandins may be important as well Urinary retention • As frequent as 35% with morphine Likely related to interaction with opioid • Opioid antagonist and agonist- • Inability to void postoperatively is • More common than with IV/IM administration receptors in the sacral spinal cord, antagonist including naloxone, a multifactorial problem—risk • Morphine Ͼ fentanyl/sufentanil promoting inhibition of sacral naltrexone and nalbuphine factors include surgery type • Not dose related parasympathetic nervous system (especially hernia) preexisting outflow. This causes detrusor muscle obstructive problems, volume relaxation and an increase in maximal status, spinal anesthesia with bladder capacity. epinephrine. • This is ␮- and ␦-receptor mediated Nausea and vomiting • Different rates for different opioids (morphine Ͼ May be a systemic effect vs. cephalad • Use smallest effective dose • Intrathecal opioids appear to have fentanyl ϭ sufentanil) migration in the cerebrospinal fluid • For ambulatory procedures use a protective effect against • Likely dose dependent (CSF) and interaction with opioid lipophilic opioid instead of morphine. intraoperative nausea and • Frequent incidence with IV/IM receptors in the area postrema. Dexamethasone and droperidol have vomiting during cesarean delivery Sensitization of the vestibular system to shown efficacy as well as combinations motion and decreased gastric emptying of scopolamine and promethazine. produced by opioids may play a role. NRTEA RG O CT PAIN ACUTE FOR DRUGS INTRATHECAL

Respiratory depression • Overall estimates: 0.07%–0.49%. Secondary to rostral spread in CSF • Prevention: Establish a monitoring • Risk factors include large dose; AL. ET RATHMELL ARTICLE REVIEW • Dose dependent protocol for your institution and train concomitant use of additional • All opioids can cause personnel; a dedicated acute pain opioid and or sedatives, age Ͼ65 • Typically lipophilic early onset (Ͻ2h)and service can establish appropriate yr, opioid naı¨ve patient morphine both early and late onset (6–12 h) protocols and monitor adverse events • Late onset depression more • Opioid antagonists including naloxone, apparent with morphine naltrexone, nalmefene • There is not an opioid free from risk S39 S40 REVIEW ARTICLE RATHMELL ET AL. ANESTH ANALG INTRATHECAL DRUGS FOR ACUTE PAIN 2005;101:S30–S43

depression after epidural or intrathecal opioid is less 2. Goldstein A, Lowney LI, Pal BK. Stereospecific and nonspecific than 1%, and limited data suggest that the risk is interactions of the morphine congener levorphanol in subcel- lular fractions of mouse brain. Proc Natl Acad SciUSA similar to that of opioids delivered via the parenteral 1971;68:1742–7. route (116). In a double-blind study of healthy volun- 3. Pert CB, Snyder SH. Opiate receptor: demonstration in nervous teers randomly assigned to receive placebo, IV mor- tissue. Science 1973;179:1011–4. phine (0.14 mg/kg), or intrathecal morphine (300 ␮g), 4. LaMotte C, Pert CB, Snyder SH. Opiate receptor binding in primate spinal cord: distribution and changes after dorsal root depression of the ventilatory response to hypoxia was section. Brain 1976;112:407–12. similar in magnitude after either IV or intrathecal 5. Yaksh TL, Rudy TA. Analgesia mediated by a direct spinal morphine but longer lasting after intrathecal adminis- action of narcotics. Science 1976;192:1357–8. tration (126). 6. Wang JK, Nauss LA, Thomas JE. Pain relief by intrathecally Risk factors for the development of respiratory de- applied morphine in man. Anesthesiology 1979;5:149–51. 7. Onofrio BM, Yaksh TL, Arnold PG. Continuous low-dose in- pression include large doses, concomitant use of ad- trathecal morphine administration in the treatment of chronic ditional opioids and/or sedatives, administration in pain of malignant origin. Mayo Clin Proc 1981;56:516–20. opioid-naı¨ve patients, and age Ͼ65 yr (116,122,123). 8. Ummenhofer WC, Arends RH, Shen DD, Bernards CM. Com- Detection of respiratory depression after intrathecal parative spinal distribution and clearance kinetics of intrathe- cally administered morphine, fentanyl, alfentanil, and sufen- administration of opioid may be difficult. Respiratory tanil. Anesthesiology 2000;92:739–53. rate may or may not decrease (124), and significant 9. Swenson JD, Owen J, Lamoreaux W, et al. The effect of distance hypercapnia can occur despite a normal respiratory from injection site to the brainstem using spinal sufentanil. Reg rate (124). Pulse oximetry may be valuable (15), but Anesth Pain Med 2001;26:306–9. 10. Eisenach JE. Lipid soluble opioids do move in cerebrospinal the most reliable clinical sign of significant respiratory fluid. Reg Anesth Pain Med 2001;26:296–7. depression appears to be a depressed level of con- 11. Shafer SL, Eisenach JC. Location, location, location. Anesthesi- sciousness (116,124). Protocols for monitoring vary, ology 2000;92:641–3. but typical duration of monitoring is 18 to 24 h after 12. Hodgson PS, Neal JM, Pollock JE, Liu SS. The neurotoxicity of intrathecal morphine and 4 to 6 h after intrathecal drugs given intrathecally (spinal). Anesth Analg 1999;88: 797–809. fentanyl or sufentanil (124). Lipophilic opioids are 13. Hetherington NJ, Dooley MJ. Potential for patient harm from now used more frequently in the ambulatory setting, intrathecal administration of preserved solutions. Med J Aust where patients are discharged shortly after surgery; 2000;173:141–3. respiratory depression more than 2 h after intrathecal 14. Samii K, Chauvin M, Viars P. Postoperative spinal analgesia with morphine. Br J Anaesth 1981;53:817–20. injection of fentanyl or sufentanil has never been de- 15. Bailey PL, Rhondeau S, Schafer PG, et al. Dose-response phar- scribed. Patients can be safely managed on regular macology of intrathecal morphine in human volunteers. Anes- wards when personnel are trained, emergency guide- thesiology 1993;79:49–59. lines are available, patient dosing and selection are 16. Rathmell JP, Pino CA, Taylor R, et al. Intrathecal morphine for postoperative analgesia: a randomized, controlled, dose- appropriate, and respiratory rate and patient level of ranging study after hip and knee arthroplasty. Anesth Analg consciousness are checked hourly (124). The efficacy 2003;97:1452–7. and safety of spinal opioids in surgical wards are best 17. Dahl JB, Jeppesen IS, Jørgensen H, et al. Intraoperative and assured when these analgesic techniques are used un- postoperative analgesic efficacy and adverse effects of intra- der the supervision of organized acute pain services thecal opioids in patients undergoing cesarean section with spinal anesthesia. Anesthesiology 1991;91:1919–27. (127,128). 18. Hamber EA, Viscomi CM. Intratheal lipophilic opioids as ad- juncts to surgical spinal anesthesia. Reg Anesth Pain Med 1999;24:255–63. Conclusions 19. Urban MK, Jules-Elysee K, Urquhart B, et al. Reduction in postoperative pain after spinal fusion with instrumentation Intrathecal administration of opioids can provide ex- using intrathecal morphine. Spine 2002;27:535–7. cellent pain relief after a wide range of operative pro- 20. Casey WF, Wynands JE, Ralley FE. The role of intrathecal cedures, including procedures performed in the am- morphine in the anesthetic management of patients undergo- ing coronary bypass surgery. J Cardiothorac Anesth 1987;1: bulatory setting. Understanding of optimal doses for 510–6. specific procedures has improved, but side effects re- 21. Fle´ron MH, Weiskopf RB, Bertrand M, et al. A comparison of main common. Knowledge of spinally mediated anal- intrathecal opioid and intravenous analgesia for the incidence gesia is evolving rapidly and a number of non-opioid of cardiovascular, respiratory, and renal complications after abdominal aortic surgery. Anesth Analg 2003;97:2–12. compounds may prove useful as analgesics that will 22. Bucklin BA, Chesnut DH, Hawkins JL. Intrathecal opioids reduce side effects and improve treatment for both versus epidural local anesthetics for labor analgesia: a meta- nociceptive and neuropathic pain. analysis. Reg Anesth Pain Med 2002;27:23–30. 23. Viscomi CM, Rathmell JP, Mason SB, et al. Analgesic efficacy and side effects of subarachnoid sufentanil-bupivacaine ad- ministered to women in advanced labor. Reg Anesth 1996;21: References 424–9. 1. Sydenham T Tractatus de podagra et hydrope. Geneva, 1686. 24. D’Angelo R, Nelson K, Meister G, et al. Dose response study of In: Latham RG. The Works of Thomas Sydenham, MD, 3rd ed., spinal fentanyl in early labor. Anesthesiology 1998; SOAP: Vol. 2. Birmingham, AL: Classics of Medicine Library, 1979. A17. ANESTH ANALG REVIEW ARTICLE RATHMELL ET AL. S41 2005;101:S30–S43 INTRATHECAL DRUGS FOR ACUTE PAIN

25. Norris MC, Fogel ST, Conway-Long C. Combined spinal- 47. Eisenach JC, De Kock M, Klimscha W. Alpha(2)-adrenergic epidural versus epidural labor analgesia. Anesthesiology 2001; agonists for regional anesthesia: A clinical review of clonidine 95:913–20. (1984–1995). Anesthesiology 1996;88:523–6. 26. Ben-David B, DeMeo PJ, Lucyk C, Solosko D. A comparison of 48. Eisenach JC, Hood DD, Curry R. Intrathecal, but not intrave- minidose lidocaine-fentanyl spinal anesthesia and local nous, clonidine reduces experimental thermal or capsaicin- anesthesia/propofol infusion for outpatient knee arthroscopy. induced pain and hyperalgesia in normal volunteers. Anesth Anesth Analg 2001;93:319–25. Analg 1998;87:591–6. 27. Ben-David B, Solomon E, Levin H, et al. Intrathecal fentanyl 49. Eisenach JC, DuPen S, Dubois M, et al. Epidural clonidine with small-dose dilute bupivacaine: better anesthesia without analgesia for intractable cancer pain. The Epidural Clonidine prolonging recovery Anesth Analg 1997:85;560–5. Study Group. Pain 1995;61:391–9. 28. Wagner LE 2nd, Eaton M, Sabnis SS, Gingrich KJ. Meperidine 50. Chiari A, Eisenach JC. Spinal anesthesia: Mechanisms, agents, and lidocaine block of recombinant voltage-dependent Naϩ methods, and safety. Reg Anesth Pain Med 1998;23:357–62. channels: evidence that meperidine is a local anesthetic. Anes- 51. Gentili M, Bonnet F. Spinal clonidine produces less urinary thesiology 1999;91:1481–90. retention than spinal morphine. Br J Anaesth 1996;76:872–3. 29. Ehikhametalor KO, Nelson M. Intrathecal as sole 52. Bonnet F, Diallo A, Saada M, et al. Prevention of tourniquet anaesthetic agent for operative procedures of the lower limb, pain by spinal isobaric bupivacaine with clonidine. Br J An- inguinal area and . West Indian Med J 2001;50:313–6. aesth 1989;63:93–6. 30. Hansen D, Hansen S. The effects of three graded doses of 53. Bioco O, Bonnet F, Mazoit JX. Effects of epinephrine and meperidine for spinal anesthesia in African men. Anesth Analg clonidine on plasma concentrations of spinal bupivacaine. Acta 1999;88:827–30. Anaesthesiol Scand 1992;36:684–8. 31. Yu SC, Ngan Kee WD, Kwan AS. Addition of meperidine to 54. Sites BD, Beach M, Biggs R, et al. Intrathecal clonidine added to bupivacaine for spinal anaesthesia for Caesarean section. Br J a bupivacaine-morphine spinal anesthetic improves postoper- Anaesth 2002;88:379–83. ative analgesia for total knee arthroplasty. Anesth Analg 2003; 32. Booth JV, Lindsay DR, Olufolabi AJ, et al. Subarachnoid me- 96:1083–8. peridine (Pethidine) causes significant nausea and vomiting 55. Eisenach JC, Detweiler DJ, Tong C, et al. Cerebrospinal fluid during labor. The Duke Women’s Anesthesia Research Group. norepinephrine and acetylcholine concentrations during acute Anesthesiology 2000;93:418–21. pain. Anesth Analg 1996;82:621–6. 33. Murto K, Lui AC, Cicutti N. Adding low dose meperidine to 56. Liu SS, Hodgson PS, Moore JM, et al. Dose-response effects of spinal lidocaine prolongs postoperative analgesia. Can J An- spinal neostigmine added to bupivacaine spinal anesthesia in aesth 1999;46:327–34. healthy volunteers. Anesthesiology 1999;90:710–7. 34. Johansen MJ, Satterfield WC, Baze WB, et al. Continuous in- 57. Eisenach JC, Hood DD, Curry R. Phase I human safety assess- trathecal infusion of hydromorphone: safety in the sheep ment of intrathecal neostigmine containing methyl- and pro- model and clinical implications. Pain Med 2004;5:14–25. pylparabens. Anesth Analg 1997;85:842–6. 35. Ackerman LL, Follett KA, Rosenquist RW. Long-term out- 58. Lauretti GR, Hood DD, Eisenach JC, Pfeifer BL. A multi-center comes during treatment of chronic pain with intrathecal study of intrathecal neostigmine for analgesia following vagi- clonidine or clonidine/opioid combinations. J Pain Symptom nal hysterectomy. Anesthesiology 1998;89:913–8. Manage 2003;26:668–77. 59. Almeida RA, Lauretti GR, Mattos AL. Antinociceptive effect of 36. Savage C, McQuitty C, Wang D, Zwischenberger JB. Posttho- low-dose intrathecal neostigmine combined with intrathecal racotomy pain management. Chest Surg Clin N Am 2002;12: morphine following gynecologic surgery. Anesthesiology 251–63. 2003;98:495–8. 37. Liu S, Carpenter RL, Mulroy MF, et al. Intravenous versus 60. Sawynok J. Adenosine receptor activation and nociception. Eur epidural administration of hydromorphone: effects on analge- sia and recovery after radical retropubic prostatectomy. Anes- J Pharmacol 1998;317:1–11. thesiology 1995;82:682–8. 61. Eisenach JC, Hood DD, Curry R. Phase I assessment of intra- 38. Lowry KJ, Tobias J, Kittle D, et al. Postoperative pain control thecal injection of and American formulation of adenosine in using epidural catheters after anterior spinal fusion for adoles- humans. Anesthesiology 2002;96:24–8. cent scoliosis. Spine 2001;26:1290–3. 62. Eisenach JC, Hood DD, Curry R. Preliminary efficacy assess- 39. Drakeford MK, Pettine KA, Brookshire L, Ebert F. Spinal nar- ment of intrathecal injection of an American formulation of cotics for postoperative analgesia in total joint arthroplasty: a adenosine in humans. Anesthesiology 2002;96:29–34. prospective study. J Bone Joint Surg Am 1991;73:424–8. 63. Eisenach JC, Hood DD, Curry R, et al. Intrathecal but not 40. Abram SE, Mampilly GA, Milosavljevic D. Assessment of the intravenous opioids release adenosine from the spinal cord. J potency and intrinsic activity of systemic versus intrathecal Pain 2004;5:64–8. opioids in rats. Anesthesiology 1997;87:127–34. 64. Rane K, Sollevi A, Segerdahl M. A randomized double-blinded 41. Chizh BA, Schlutz H, Scheede M, Englberger W. The evaluation of adenosine as adjunct to sufentanil in spinal la- N-methyl-d-aspartate antagonistic and opioid components of bour analgesia. Acta Anaesthesiol Scand 2003;47:601–3. d-methadone antinociception in the rat spinal cord. Neurosci 65. Kito K, Kato H, Shibata M, et al. The effect of varied doses of Lett 2000;296:117–20. epinephrine on duration of lidocaine spinal anesthesia in the 42. Miles J. Intrathecal therapy for chronic pain. Stereotact Funct thoracic and lumbosacral dermatomes. Anesth Analg 1998;86: Neurosurg 2001;77:156–8. 1018–22. 43. Fernandez-Liesa JI, Panadero A. Use of intrathecal hyperbaric 66. Moore JM, Liu SS, Pollock JE, et al. The effect of epinephrine on methadone in postoperative analgesia of thoracic surgery. Reg small-dose hyperbaric bupivacaine spinal anesthesia: clinical Anesth Pain Med 2000;25:325. implications for ambulatory surgery. Anesth Analg 1998;86: 44. Pertovaara A. Antinoceception induced by alpha 2 adrenocep- 973–7. tor agonists, with special emphasis on medetomidine studies. 67. Camann W, Minzter B, Denney R, Datta S. Intrathecal sufen- Prog Neurobiol 1993;40:691–709. tanil for labor analgesia: effects of added epinephrine. Anes- 45. Ossipov MH, Harris S, Lloyd P, Messineso E. An isobolo- thesiology 1993;78:870–4. graphic analysis of the antinociceptive effect of systemically 68. Goodman SR, Kim-Lo SH, Ciliberto CF, et al. Epinephrine is and intrathecally administered combinations of clonidine and not a useful addition to intrathecal fentanyl or fentanyl- opiates. J Pharmacol Exp Ther 1990;255:1107–16. bupivacaine for labor anesthesia. Reg Anesth Pain Med 2002; 46. De Kock M, Crochet B, Morimont C, Scholtes JL. Intravenous 27:374–9. or epidural clonidine for intra and post operative analgesia. 69. Liu SS, MsDonald SB. Current issues in spinal anesthesia. Anesthesiology 1993;79:525–31. Anesthesiology 2001;94:888–906. S42 REVIEW ARTICLE RATHMELL ET AL. ANESTH ANALG INTRATHECAL DRUGS FOR ACUTE PAIN 2005;101:S30–S43

70. Kroin JS, Ling ZD, Buvanendran A, Tuman KJ. Upregulation of 94. Wang JJ, Ho ST, Tzeng JI. Comparison of intravenous nalbu- spinal cyclooxygenase-2 in rats after surgical incision. Anes- phine infusion versus naloxone in the prevention of epidural thesiology 2004;100:364–9. morphine-related side effects. Reg Anesth Pain Med 1998;23: 71. Yamamoto T, Sakashita Y. The role of the spinal opioid recep- 479–84. tor like1 receptor, the NK-1 receptor, and cyclooxygenase-2 in 95. Charuluxananan S, Kyokong O, Somboonviboon W, et al. Nal- maintaining postoperative pain in the rat. Anesth Analg 1999; buphine versus ondansetron for prevention of intrathecal 89:1203–8. morphine-induced pruritus after cesarean delivery. Anesth 72. Zhu X, Conklin D, Eisenach JC. Cyclooxygenase-1 in the spinal Analg 2003;96:1789–93. cord plays an important role in postoperative pain. Pain 2003; 96. Torn K, Tuominen M, Tarkkila P, Lindgren L. Effects of sub- 104:15–23. hypnotic doses of propofol on the side effects of intrathecal 73. Eisenach JC, Curry R, Hood DD, Yaksh TL. Phase I safety morphine. Br J Anaesth 1994;73:411–2. assessment of intrathecal ketorolac. Pain 2002;99:599–604. 97. Borgeat A, Wilder-Smith OH, Saiah M, Rifat K. Subhypnotic 74. Conklin DR, Eisenach JC. Intrathecal ketorolac enhances an- doses of propofol relieve pruritus induced by epidural and tinociception from clonidine. Anesth Analg 2003;96:191–4. intrathecal morphine. Anesthesiology 1992;76:510–2. 75. Yaksh TL, Allen JW. The use of intrathecal midazolam in 98. Beilin Y, Bernstein HH, Zucker-Pinchoff B, et al. Subhypnotic humans: a case study in progress. Anesth Anlag 2004;98: doses of propofol do not relieve pruritus induced by intrathe- cal morphine after cesarean section. Anesth Analg 1998;86: 1536–45. 310–3. 76. Erdine S, Yucel A, Ozyalcin S, et al. Neurotoxocity of midazo- 99. Warwick JP, Kearns CF, Scott WE. The effect of subhypnotic lam in the rabbit. Pain 1999;80:419–23. doses of propofol on the incidence of pruritus after intrathecal 77. Johansen MJ, Gradert TL, Satterfield WC, et al. Safety and morphine for cesarean section. Anaesthesia 1997;52:270–5. efficacy of continuous intrathecal midazolam infusion in the 100. Rawal N, Mollefors K, Axelsson K, et al. An experimental sheep model. Anesth Analg 2004;98:1528–35. study of urodynamic effects of epidural morphine and of nal- 78. Tucker AP, Lai C, Nadeson R, Goodchild CS. Intrathecal mi- oxone reversal. Anesth Analg 1983;62:641–7. dazolam I: A cohort study investigating safety. Anesth Analg 101. Wheeler M. Adverse events associated with postoperative opi- 2004;98:1512–20. oid analgesia: a systematic review. J Pain 2002;3:159–80. 79. Tucker AP, Mezzatesta J, Nadeson R, Goodchild CS. Intrathe- 102. Peterson TK, Husted SE, Rybro L, et al. Urinary retention cal midazolam II: Combination with intrathecal fentanyl for during IM and extradural morphine analgesia. Br J Anaesth labor pain. Anesth Analg 2004;98:1521–7. 1982;54:1175–8. 80. Borner U, Muller H, Zierski J, et al. CSF compatibility of 103. Liu S, Chiu AA, Carpenter RL, et al. Fentanyl prolongs lido- antispaticity agents. In: Muller J, Zierski J, Penn RD, eds. caine spinal anesthesia without prolonging recovery. Anesth Local-spinal therapy of spasticity. Berlin: Springer Verlag, Analg 1995;80:730–4. 1986: 81–4. 104. Lau WC, Green DR. Subarachnoid sufentanil for extracorpo- 81. Szarvas S, Harmon D, Murphy D. Neuraxial opioid-induced real shock wave lithotripsy provides earlier discharge of the pruritus: a review. J Clin Anesth 2003;15:234–9. outpatient than intrathecal lidocaine. Reg Anesth 1997;84: 82. Kyriakides K, Hussain SK, Hobbs GJ. Management of opioid- 1227–31. induced pruritus: a role for 5-HT3 antagonists? Br J Anaesth 105. Malinovsky JM. Nalbuphine reverses urinary effects of epi- 1999;82:439–41. dural morphine: a case report. J Clin Anesth 2002;14:535–8. 83. Shah MK, Sia AT, Chong JL. The effect of the addition of 106. Chaney MA. Side effects of intrathecal and epidural opioids. ropivacaine or bupivacaine upon pruritus induced by intrathe- Can J Anaesth 1995;42:891–903. cal fentanyl in labour. Anaesthesia 2000;55:1008–13. 107. Kleinschmidt S, Schellhase C, Mertzlufft F. Continuous seda- 84. Yeh HM, Chen LK, Lin CJ, et al. Prophylactic intravenous tion during spinal anaesthesia: Gamma-hydroxybutyrate vs. ondansetron reduces the incidence of intrathecal morphine- propofol. Eur J Anaesthesiol 1999;16:23–30. induced pruritus in patients undergoing cesarean delivery. 108. Weber EWG, Slappendel R, Gielen MJM, Dirksen R. Intrathecal Anesth Analg 2000;91:172–5. addition of morphine to bupivacaine is not the cause of post- 85. Dahlgren G, Hultstrand C, Jakobsson J, et al. Intrathecal sufen- operative nausea and vomiting. Reg Anesth Pain Med 1998;23: tanil, fentanyl, or placebo added to bupivacaine for cesarean 81–6. section. Anesth Analg 1997;85:1288–93. 109. Jacobson L, Chabal C, Brody MC. A dose-response study of 86. Scott PV, Fischer HB. Spinal opiate analgesia and facial intrathecal morphine: efficacy, duration, optimal dose, and pruritus: a neural theory. Postgrad Med J 1982;58:531–5. side effects. Anesth Analg 1988;67:1082–8. 87. Stocks GM, Hallworth SP, Fernando R, et al. Minimum local 110. Yamaguchi H, Watanabe S, Fukuda T, et al. Minimal effective analgesic dose of intrathecal bupivacaine in labor and the effect dose of intrathecal morphine for pain relief following transab- dominal hysterectomy. Anesth Analg 1989;68:537–40. of intrathecal fentanyl. Anesthesiology 2001;94:593–8. 111. Borgeat A, Ekatodramis G, Schenker CA. Postoperative nausea 88. Nelson KE. A comparison of intrathecal fentanyl and sufen- and vomiting in regional anesthesia: a review. Anesthesiology tanil for labor analgesia. Anesthesiology 2002;96:1070–3. 2003;98:530–47. 89. Dahl JB, Jeppesen IS, Jorgensen H, et al. Intraoperative and 112. Lo WK, Chong JL, Chen LH. Combined spinal epidural for postoperative analgesic efficacy and adverse effects of intra- labour analgesia: Duration, efficacy and side effects of adding thecal opioids in patients undergoing cesarean section with sufentanil or fentanyl to bupivacaine intrathecally vs plain spinal anesthesia: qualitative and quantitative systematic re- bupivacaine. Singapore Med J 1999;40:639–43. view of randomized controlled trials. Anesthesiology 1999;91: 113. Honet JE, Arkoosh VA, Norris MC, et al. Comparison among 1919–27. intrathecal fentanyl, meperidine, and sufentanil for labor an- 90. Krause L, Shuster S. Mechanism of action of antipruritic drugs. algesia. Anesth Analg 1992;75:734–49. BMJ 1983;287:1199–200. 114. Ho ST, Wang JJ, Tzeng JI, et al. Dexamethasone for prevention 91. Kjellberg F, Tramer MR. Pharmacological control of opioid- of nausea and vomiting associated with epidural morphine: a induced pruritus: a quantitative systematic review of random- dose-ranging study. Anesth Analg 2001;92:745–8. ized trials. Eur J Anaesthesiol 2001;18:346–57. 115. Gwirtz KH, Young JV, Byers RJ, et al. The safety and efficacy 92. Krajnik M, Zylicz Z. Understanding pruritus in systemic dis- of intrathecal opioid analgesia for acute postoperative pain: ease. J Pain Symptom Manage 2001;21:151–68. seven years experience with 5969 surgical patients at Indiana 93. Borgeat A, Stirnemann HR. Ondansetron is effective to treat University Hospital. Anesth Analg 1999;88:599–604. spinal or epidural morphine-induced pruritus. Anesthesiology 116. Etches RC, Sandler AN, Daley MD. Respiratory depression 1999;90:432–6. and spinal opioids. Can J Anaesth 1989;36:165–85. ANESTH ANALG REVIEW ARTICLE RATHMELL ET AL. S43 2005;101:S30–S43 INTRATHECAL DRUGS FOR ACUTE PAIN

117. Ko S, Goldstein DH, VanDenKerkhof EG. Definitions of “re- 123. Morgan M. The rational use of intrathecal and extradural spiratory depression” with intrathecal morphine postoperative opioids. Br J Anaesth 1989;63:165–88. analgesia: a review of the literature. Can J Anaesth 2003;50: 124. Ready LB, Loper KA, Nessly M, Wild L. Postoperative epidural 679–88. morphine is safe on surgical wards. Anesthesiology 1991;75: 118. Standl TG, Horn E, Luckmann M, et al. Subarachnoid sufen- 452–6. tanil for early postoperative pain management in orthopedic 125. Zimmermann DL, Stewart J. Postoperative pain management patients: a placebo-controlled, double-blind study using spinal and acute pain service activity in Canada. Can J Anaesth microcatheters. Anesthesiology 2001;94:230–8. 1993;40:568–75. 119. Gustafsson LL, Schildt B, Jacobsen K. Adverse effects of extra- 126. Bailey PL, Ku JL, Pace NL, et al. Effects of intrathecal morphine dural and intrathecal opiates: report of a nation-wide survey in on the ventilatory response to hypoxia. N Engl J Med 2000;343: Sweden. Br J Anaesth 1982;54:479–86. 120. Stenseth R, Sellevoid O, Breivik H. Epidural morphine for 1228–34. postoperative pain: experience with 1085 patients. Acta An- 127. Brodner G, Mertes N, Buerkle H, et al. Acute pain aesthsiol Scand 1985;29:148–56. management; analysis, implications and consequences after 121. Fuller JG, Mcmorland GH, Douglas MJ, Palmer L. Epidural prospective experience with 6349 surgical patients. Eur J An- morphine for analgesia after caesarean section: a report of 4880 aesth 2000;17:566–75. patients. Can J Anaesth 1990;37:608–12. 128. Rawal N. Acute pain services revisited: good from far, far from 122. Rawal N, Allvin R, EuroPain study group on Acute Pain. good. Reg Anesth Pain Med 2002;27:117–21. Epidural and intrathecal opioids for postoperative pain man- agement in Europe: a 17-nation questionnaire study of selected hospitals. Acta Anaesthesiol Scand 1996;40:1119–26. Anaesthesia, 2009, 64, pages 27–31 doi:10.1111/j.1365-2044.2008.05655.x ......

Audit of motor weakness and premature catheter dislodgement after epidural analgesia in major abdominal surgery

I. Ko¨ nigsrainer,1 S. Bredanger,2 R. Drewel-Frohnmeyer,2 R. Vonthein,3 W. A. Krueger,2 A. Ko¨ nigsrainer,1 K. E. Unertl2 and T. H. Schroeder2

1 Department of General-, Visceral- and Transplantation Surgery, University Hospital Tu¨bingen, Tu¨bingen, Germany 2 Department of Anaesthesiology and Critical Care Medicine, University Hospital Tu¨bingen, Tu¨bingen, Germany 3 Department of Medical Biometry, University Hospital Tu¨bingen, Tu¨bingen, Germany

Summary In a quality improvement audit on epidural analgesia in 300 patients after major abdominal surgery, we identified postoperative lower leg weakness and premature catheter dislodgement as the most frequent causes of premature discontinuation of postoperative epidural infusion. Lower limb motor weakness occurred in more than half of the patients with lumbar epidural analgesia. In a second period monitoring 177 patients, lumbar catheter insertion was abandoned in favour of exclusive thoracic placement for epidural catheters. Additionally, to prevent outward movement, the catheters were inserted deeper into the epidural space (mean (SD) 5.2 (1.5) cm in Period Two vs 4.6 (1.3) cm in Period One). Lower leg motor weakness declined from 14.7% to 5.1% (odds ratio 0.35; 95% confidence interval 0.16–0.74) between the two periods. Similarly, the frequency of premature catheter dislodgement was reduced from 14.5% to 5.7% (odds ratio 0.35; 95% confidence interval 0.17–0.72). With a stepwise logistic regression model we demonstrated that the odds of premature catheter dislodgement was reduced by 43% for each centimetre of additional catheter advancement in Period Two. We conclude that careful audit of specific complications can usefully guide changes in practice that improve success of epidural analgesia regimens.

...... Correspondence to: Torsten H. Schroeder E-mail: [email protected] Accepted: 16 June 2008

Peri-operative analgesic protocols that include a com- Audits have proved generally useful in maintaining bination of general and epidural analgesia have been standards of care and might be useful in identifying introduced for a wide variety of surgical procedures problems with postoperative epidural therapy, especially to reduce stress response and allow immediate postoper- those important in the setting of a local unit. We ative rehabilitation [1, 2]. Epidural analgesia significantly conducted an audit of postoperative epidural analgesia in reduces the incidence of cardiovascular events, acute our unit by first identifying common causes for our respiratory failure, deep venous thrombosis and gastro- postoperative epidural failures, then initiating new stand- intestinal paralysis after major abdominal surgery [3, 4]. ards and finally monitoring the success of these inter- Therefore, it is important to maximise efficacy of ventions. postoperative epidural analgesia. Up to 15% of post- operative epidural analgesia failures are due to technical Methods problems leading to premature dislodgement of the catheter or disconnection of the infusion. Failures due This project was undertaken to improve the quality of to inadequate analgesia despite properly-functioning epidural analgesia in our unit. The Ethics Committee catheters account for another 8–12% [5–7]. Such failures of the University Hospital Tu¨bingen approved the cause patient discomfort and delays in the rehabilitation conduct of this audit and publication of its results. process. The audit was organised in two parts.

2008 The Authors Journal compilation 2008 The Association of Anaesthetists of Great Britain and Ireland 27 I. Ko¨ nigsrainer et al. Improving epidural analgesia after abdominal surgery Anaesthesia, 2009, 64, pages 27–31 ...... Æ ......

Period One of legs; 1 = knee flexion barely possible-foot lifting First, to identify the main reasons for premature cessation unconfined; 2 = knee flexion impossible-foot lifting of postoperative epidural analgesia, we retrospectively barely possible; 3 = no movement possible), paraesthesia analyzed the data of 300 patients in whom epidural (i.e. prickling sensations in the lower limbs or body), the analgesia had been used after major abdominal surgery in details of the prescribed epidural drug infusion, and any the Department of General, Visceral, and Transplantation free-text notes. Motor weakness of the legs was defined as Surgery over a 13-month period. a Bromage score > 0. We classed ‘insufficient analgesia’ as any state in which unilateral distribution of the epidural Conduct of epidural analgesia and anaesthesia block, or pain as reported on the tracking sheet. The Our local policy is that exclusion criteria for pre-operative catheter insertion site was inspected once per day. catheter placement are: patient refusal, coagulation dis- According to hospital protocol, the continuous infusion orders, signs of local or systemic infection, and anatomical was stopped no sooner than the fourth postoperative day abnormalities. Our standard hospital protocol invol- or after postoperative ileus had resolved (with the final ves identifying the epidural space with the patient decision left to the discretion of the responsible attending sitting upright, using a midline approach and the loss- anaesthetist). The epidural catheter was removed when of-resistance technique to either air or saline. It was our the tracking sheet confirmed the patient was pain-free. local policy that either the lumbar or thoracic vertebral interspace was selected for epidural catheter placement Period Two on the basis of the dermatomal extent of the surgical Following the review of the data of Period One, we procedure and skin incision. Thus, patients scheduled for changed certain policies and (a) limited catheter place- colorectal surgery or surgery in the lower pelvic region ment to thoracic interspaces T8-11, (b) stipulated that received lumbar epidural analgesia, while higher surgical catheters be advanced 1 cm deeper into the epidural incisions received thoracic epidural analgesia. The cath- space. Thus, practitioners who had placed their catheters eter is routinely affixed to the skin at the insertion site 4 cm into the epidural space in Period One were with a consisting of steri-stripesTM (3M Health instructed to advance them at least 5 cm deep; those Care, St Paul, MN, USA), a porous tape with a small hole who had placed their catheters 5 cm deep were instructed which was closed with an additional piece of tape. The to advance them now 6 cm (with an absolute upper limit rest of the catheter is generally taped along the midline in the new guidelines of 6 cm). The remainder of the of the patient’s back. Routinely, a test dose of 20 mg protocols for conduct of analgesia and anaesthesia were ropivacaine is applied via the catheter. General anaesthesia unchanged. During this second period 177 patients ) is generally induced with thiopentone (5 mg.kg 1) and received an epidural catheter. ) sufentanil (0.2 lg.kg 1) and maintained with sevoflurane in oxygen ⁄ air. According to local guidelines, the patients Statistical analysis received an epidural loading dose of ropivacaine 45 mg Variables are given as mean (SD), median [interquartile and sufentanil 10 lg followed by continuous epidural range] or as frequency, as appropriate. Significance of the ) infusion of ropivacaine 30 mg.h 1 and sufentanil differences between the groups was evaluated by univar- ) 3 lg.h 1. Postoperatively, continuous infusion was iate comparisons: analysis of variance for continuous ) reduced to ropivacaine 12 mg.h 1 and sufentanil variables and chi-squared tests for discrete data. Non- ) 3 lg.h 1 and then titrated according to presence of pain parametric tests (Wilcoxon, Kruksal–Wallis) were used ) and sensory block. Additionally, paracetamol 4 g.24 h 1 where distributions were extremely skewed. Odds ratios ) or metamizole up to 5 g.24 h 1 were prescribed. (OR) and estimated 95% confidence intervals (CI) for single risk factors were stratified by time period. For Postoperative care and data collection quantitative variables the OR per unit, e.g. for one Postoperatively, patients with epidural catheters were additional centimetre of catheter advancement, was assigned to the care of an acute postoperative pain service estimated by logistic regression. We used a stepwise team. A tracking sheet was initiated by the anaesthesiol- variable selection to find a model that described the main ogist inserting the epidural catheter. The sheet contained factors involved in an adjusted manner. Calculations were data on catheter placement, patient demographics, and done with JMP 7.0 [8]. prescribed epidural drug infusion. Patients were visited twice daily on the ward by a member of the acute pain Results service team. The ‘tracking’ sheet recorded the following: vital parameters, presence of absence of pain, motor Patient characteristics are listed in Table 1 and suggest weakness (modified Bromage scale: 0 = free movement similar populations between the two time periods. The

2008 The Authors 28 Journal compilation 2008 The Association of Anaesthetists of Great Britain and Ireland Anaesthesia, 2009, 64, pages 27–31 I. Ko¨ nigsrainer et al. Improving epidural analgesia after abdominal surgery ...... Æ

Table 1 Patient characteristics, duration of epidural catheter Following the change in policy to thoracic placement placement, and surgery details. Values are median [interquartile only (2.3% of epidural catheters were nonetheless inserted (IQ) range], mean (SD) or number (proportion) as indicated. into a lumbar intervertebral space) there was a significant reduction in postoperative lower limb motor weakness Period One Period Two (n = 300) (n = 177) (5.1% vs 14.7%; OR 0.35; 95% CI 0.16–0.74). The percentage of patients with inadequate postoperative Age; years 59 (13) 60 (14) analgesia remained unchanged (10.3% vs 12.2%; OR Gender (male:female) 165 : 132 106 : 71 0.82; 95% CI 0.45–1.51). BMI; kg.m)2 25.8 (6.3) 25.1 (5.7) ASA; I:II:III (%) 7 : 72 : 21 8 : 64 : 26 Following the change to deeper catheter placement, Catheter duration: we found epidural catheters inserted 5.2 (1.5) cm in Days 4 [3–5] 4 [3–4] Period Two vs 4.6 (1.3) cm in Period One. As a result, Thoraco-abdominal surgery (%) 4 3 Upper abdominal surgery (%) 52 55 the rate of catheter dislodgement was significantly Colorectal surgery (%) 31 25 reduced from 14.5% to 5.7% (OR 0.35; 95% CI 0.17 Miscellaneous surgery (%) 13 17 to 0.72). We further calculated by linear regression that the OR for catheter dislodgement was reduced by 43% BMI, Body mass index; ASA, American Society of Anesthesiology. for every cm of additional catheter advancement into the epidural space (unit OR 0.57; 95% CI 0.34–0.93) Period One Period Two within Period Two. The frequency of catheter dis- lodgement was independent of gender, body mass index, 30 age, site of insertion (thoracic or lumbar), ASA status, or 20 type of surgery. 10

Cases per month 0 Discussion 1 3 5 7 9 11 13 15 17 19 Month This quality improvement audit achieved a significant reduction of postoperative leg weakness and premature Figure 1 Monthly postoperative epidural catheter case load (black line: thoracic epidural catheters, grey line: lumbar epi- catheter dislodgement in patients by promoting the near- dural catheters) during Period One (13 months) and Period exclusive use of thoracic placement of epidural catheters Two (7 months). and by inserting the catheters deeper into the epidural space. mean duration of catheter placement and the nature of The use of epidural analgesia for intra- and postoper- the surgery performed did not differ significantly between ative pain control has become an important element of Period One and Period Two. early rehabilitation programs. The occurrence of lower In Period One, 19.7% of the epidural catheters were limb motor weakness frequently impedes patient mobi- placed in a lumbar intervertebral space and 80.3% in a lization and leads to transient discontinuation of epidural thoracic intervertebral space (Fig. 1). A review of the analgesics, increasing patient discomfort. This impairment database identified postoperative lower limb motor of motor function results from the anaesthesia of the weakness as the most frequently noted abnormality taken lumbar and sacral plexuses. It has been shown that after care of by the acute pain service team (14.7%) (Table 2): epidural application, the highest concentrations of local 52.4% of patients with epidural catheters placed in a anaesthetic agents are found in the spinal roots and pia lumbar intervertebral space developed postoperative arachnoid. Significant amounts of local anaesthetic are motor weakness of the legs, compared with only 4.8% demonstrated in the peripheral parts of the spinal cord. of patients with a thoracic epidural catheter. Additionally, diffusion of the drugs along the spinal roots

Table 2 Site and failure rate (by motor weakness and dislodgement Period One Period Two Odds ratio of catheter) of epidural. Values are (n = 300) (n = 177) [95% CI] p value mean (SD) or number (proportion). CI, confidence interval. Thoracic:lumbar site; n 241 : 59 173 : 4 0.11 [0.04–0.33] < 0.0001 Lumbar epidural; % 19.7 2.3 0.11 [0.04–0.33] < 0.0001 Motor weakness; % 14.7 5.1 0.35 [0.16–0.74] 0.0013 Catheter dislodgement; % 14.5 5.7 0.35 [0.17–0.72] < 0.0001 Epidural catheter depth; cm 4.6 (1.3) 5.2 (1.5) 0.79 [0.64–0.98] 0.004 Insufficient analgesia; % 12.2 10.3 0.82 [0.45–1.51] 0.57

2008 The Authors Journal compilation 2008 The Association of Anaesthetists of Great Britain and Ireland 29 I. Ko¨ nigsrainer et al. Improving epidural analgesia after abdominal surgery Anaesthesia, 2009, 64, pages 27–31 ...... Æ ...... into the cerebrospinal fluid and the spinal cord might space is easily filled and then a retrograde flow of epidural affect the two major descending motor pathways that run solutions could loosen the grip [21]. Moreover, 30–40% downwards superficially at lumbar levels [9–11]. In of catheters which are properly placed initially have been contrast, it is unclear whether drug penetration is deep shown to move outwards within the epidural space over enough to affect a significant proportion of ascending and time [22]. Consequently, increasing frequency of dis- descending tracts and contribute to leg weakness when lodgement over the duration of catheterization has been local analgesics are injected at thoracic levels [11]. Our shown [23, 24]. However, we examined neither the results are consistent with those of previous studies duration of catheterization before dislodgement nor the reporting a rate of leg motor impairment between 40% rate of fluid leakage so the main reasons for premature and 50% in patients after lumbar epidural analgesia [12]. dislodgement in our study remain unclear. In our Period One, lumbar epidural analgesia was In a study of epidural analgesia during labour, catheter preferred at our institution in patients scheduled for insertion to a depth of 7 cm was associated with the lower pelvic and colorectal surgery. The rationale for highest rate of insertion complications, while insertion to catheter insertion site was that the intervertebral segments a depth of 5 cm was associated with the highest incidence of the affected splanchnic nerves of the inferior mesen- of satisfactory analgesia [25]. Therefore, we limited the teric plexus (L1–L2) are stimulated by lesser pelvis, left catheter insertion depth to 5–6 cm. Subcutaneous tun- colon, sigmoid, and rectal surgery as well as the derma- nelling and suture techniques have been used to prevent tomal segments of the laparotomy incision. Additionally, outward movement of epidural catheters [22], but are it seemed reasonable to assume that the danger of direct somewhat invasive and time-consuming. Others have nerve damage would be lower with lumbar needle recommended using an ‘epidural catheter clamp’ to insertion than it would for thoracic needle insertion reduce rates of premature catheter dislodgements from (although we are not aware of published data to 16% to 4% [17] but we achieved the same effect by simply substantiate this). advancing the catheter deeper into the epidural space. We To reduce the rate of leg motor weakness in Period were, though, surprised that a relatively small change in Two we considered two possible treatment changes. First, insertion depth of < 1 cm had such a beneficial effect on we could have confined the catheter insertion site to dislodgement. thoracic levels (based on the considerations above), or Our study does have some limitations. While we could second we could have used lower doses of local demonstrate a significant reduction of leg motor weakness anaesthetics. However, the latter might have led to and unintentional catheter dislodgements, no data were insufficient analgesia [13]. Since we were already using a collected on end-points such as reduced morbidity, time reasonably low dose of ropivacaine with an opioid, which to discharge or reduction of the number of unscheduled has been recommended as an effective postoperative patient visits by the members of the postoperative pain analgesic after abdominal surgery [13–15], we chose not service team. Moreover, this is not an interventional study to reduce local anaesthetic concentrations further but but an audit. Generally, results may have been con- instead change the insertion sites to thoracic level. Our founded by seasonal variation, the fact the staff were results suggest that thoracic epidural analgesia was as aware of the audit (i.e. no blinding), staff changes, and successful as lumbar epidural analgesia [9, 16]. other aspects that we could not control as we might in a While our rate of premature catheter dislodgement formal study. The data analysis for time Period One was (14.5%) during Period One may seem high, comparable limited to 13 months. During this time span, a full data set rates are reported by others [6, 17–19]. We considered was available, and no changes in the peri-operative this unacceptably high, as it hampered postoperative management occurred. Time Period Two was much rehabilitation and was unpleasant for the patients if a new shorter and ended after 7 months because – independent catheter placement became necessary. Our intention in of this audit – there was a change of the general advancing the catheter insertion deeper was to establish a anaesthesia induction protocol. We did not think it broader margin for outward catheter mobility during correct to include this subsequent data in this audit. While patient mobilization. One reason for premature catheter we assessed occurrence of pain by it being noted on our dislodgement is fluid leakage along the outside of the tracking charts, we did not use a formal visual analogue catheter into the muscles of the back or to the skin [19]. scale so our assessment of this important variable is a The fluid can accumulate under the adhesive dressing and shortcoming. Thus, while we demonstrate improvements loosen the tape fixation to the skin [20]. Towards its tip, in quality of motor weakness and catheter dislodgement the catheter is usually gripped tightly by the ligamentum rates, we cannot show that this is translated into any flavum. Due to the presence of a midline septum or benefit in pain control. Nonetheless we can be confident postsurgical adhesions the reduced capacity of the epidural that there was unlikely to have been a significant increase

2008 The Authors 30 Journal compilation 2008 The Association of Anaesthetists of Great Britain and Ireland Anaesthesia, 2009, 64, pages 27–31 I. Ko¨ nigsrainer et al. Improving epidural analgesia after abdominal surgery ...... Æ in this measure as a result of the changes we introduced randomized, double-blind trial in patients with retropubic (Table 2). prostatectomy. Acta Anaesthesiologica Scandinavica 2007; 51: In summary, future audits or studies will extend to cover 294–8. other surgical disciplines where postoperative lumbar 13 Scott DA, Chamley DM, Mooney PH, Deam RK, Mark epidural analgesia is frequent, e.g. urology or gynaecology. AH, Hagglof B. Epidural ropivacaine infusion for post- operative analgesia after major lower abdominal surgery – We feel that our rate of 6% for premature catheter a dose finding study. Anesthesia and Analgesia 1995; 81: dislodgement remains unacceptably high and there is 982–6. room for improvement. Studies assessing combinations of 14 Niiyama Y, Kawamata T, Shimizu H, Omote K, Namiki A. methods to further reduce this rate (e.g. fixation methods, The addition of epidural morphine to ropivacaine improves epidural catheter clamp, and catheter advancement) are epidural analgesia after lower abdominal surgery. Canadian desirable. Journal of Anesthesia 2005; 52: 181–5. 15 Scott DA, Blake D, Buckland M, et al. A comparison of epidural ropivacaine infusion alone and in combination with References 1, 2, and 4 microg ⁄ mL fentanyl for seventy-two hours of 1 Rodgers A, Walker N, Schug S, et al. Reduction of post- postoperative analgesia after major abdominal surgery. operative mortality and morbidity with epidural or spinal Anesthesia and Analgesia 1999; 88: 857–64. anaesthesia: results from overview of randomised trials. 16 Lee CJ, Jeon Y, Lim YJ, et al. The influence of neck flexion British Medical Journal 2000; 321: 1493. and extension on the distribution of contrast medium in the 2 Holte K, Foss NB, Andersen J, et al. Liberal or restrictive high thoracic epidural space. Anesthesia and Analgesia 2007; fluid administration in fast-track colonic surgery: a ran- 104: 1583–6. domized, double-blind study. British Journal of Anaesthesia 17 Clark MX, O’Hare K, Gorringe J, Oh T. The effect of the 2007; 99: 500–8. Lockit epidural catheter clamp on epidural migration: 3 Jorgensen H, Wetterslev J, Moiniche S, Dahl JB. Epidural a controlled trial. Anaesthesia 2001; 56: 865–70. local anaesthetics versus opioid-based analgesic regimens on 18 Liu SS, Allen HW, Olsson GL. Patient-controlled epi- postoperative gastrointestinal paralysis, PONV and pain after dural analgesia with bupivacaine and fentanyl on hospital abdominal surgery. Cochrane Database of Systematic Reviews wards: prospective experience with 1,030 surgical patients. 2000; 4: CD001893. Anesthesiology 1998; 88: 688–95. 4 Beattie WS, Badner NH, Choi P. Epidural analgesia 19 Motamed C, Farhat F, Remerand F, Stephanazzi J, reduces postoperative myocardial infarction: a meta-analysis. Laplanche A, Jayr C. An analysis of postoperative epidural Anesthesia and Analgesia 2001; 93: 853–8. analgesia failure by computed tomography epidurography. 5 Ballantyne JC, McKenna JM, Ryder E. Epidural analgesia – Anesthesia and Analgesia 2006; 103: 1026–32. experience of 5628 patients in a large teaching hospital 20 Scott DA, Beilby DS, McClymont C. Postoperative anal- derived through audit. Acute Pain 2003; 4: 89–97. gesia using epidural infusions of fentanyl with bupivacaine. 6 Neal JM, Wilcox RT, Allen HW, Low DE. Near-total A prospective analysis of 1,014 patients. Anesthesiology 1995; esophagectomy: the influence of standardized multimodal 83: 727–37. management and intraoperative fluid restriction. Regional 21 Collier CB, Gatt SP. Dislodgement of epidural catheters in Anesthesia and Pain Medicine 2003; 28: 328–34. labour. Anaesthesia 2001; 56: 604–5. 7 Pan PH, Bogard TD, Owen MD. Incidence and charac- 22 Burstal R, Wegener F, Hayes C, Lantry G. Subcutaneous teristics of failures in obstetric neuraxial analgesia and anes- tunnelling of epidural catheters for postoperative analgesia to thesia: a retrospective analysis of 19,259 deliveries. prevent accidental dislodgement: a randomized controlled International Journal of Obstetric Anesthesia 2004; 13: 227–33. trial. Anaesthesia and Intensive Care 1998; 26: 147–51. 8 Altman DG. Practical Statistics for Medical Research, 1st edn. 23 Scott AM, Starling JR, Ruscher AE, DeLessio ST, Harms Charlotte, NC: Baker and Taylor, 2008. BA. Thoracic versus lumbar epidural anesthesia’s effect on 9 Bromage PR. Lower limb reflex changes in segmental epi- pain control and ileus resolution after restorative proctoco- dural analgesia. British Journal of Anaesthesia 1974; 46: 504–8. lectomy. Surgery 1996; 120: 688–95. 10 Cohen EN. Distribution of local anesthetic agents in the 24 Tsui BC, Gupta S, Finucane B. Confirmation of epidural neuraxis of the dog. Anesthesiology 1968; 29: 1002–5. catheter placement using nerve stimulation. Canadian Journal 11 Bromage PR, Joyal AC, Binney JC. Local anesthetic of Anesthesia 1998; 45: 640–4. drugs: penetration from the spinal extradural space into 25 Beilin Y, Bernstein HH, Zucker-Pinchoff B. The optimal the neuraxis. Science 1963; 140: 392–4. distance that a multiorifice epidural catheter should be 12 Heid F, Schmidt-Glintzer A, Piepho T, Jage J. Epidural threaded into the epidural space. Anesthesia and Analgesia ropivacaine – where are the benefits? A prospective, 1995; 81: 301–4.

2008 The Authors Journal compilation 2008 The Association of Anaesthetists of Great Britain and Ireland 31 “I would have everie man write what he knowes and no more.”–Montaigne BRITISH JOURNAL OF ANAESTHESIA Volume 102, Number 6, June 2009

British Journal of Anaesthesia 102 (6): 729–30 (2009) doi:10.1093/bja/aep085 Editorial I

Spinal anaesthesia: a century of refinement, and failure is still an option

On August 24, 1898, August Bier1 and his assistant colleagues7 dissect the spinal technique into its sequential Hildebrandt undertook ‘experiments on [their] own components, providing a framework to understand the bodies’ which were part of their historic initial investi- mechanisms that may account for failure. Such infor- gations of spinal anaesthesia. Bier’s description of these mation is a prerequisite for maximizing success rate, and experiments is notable for the manner in which he docu- for rational clinical management of a failed block. In the mented the lack of sensibility after injection of first of two related research papers, Ruppen and col- into Hildebrandt’s subarachnoid space, which included a leagues8 explore the cerebrospinal fluid (CSF) concen- burning cigar, a strong blow to the shin with an iron trations of anaesthetic associated with the development of hammer, and strong pressure and traction on the testicles, adequate spinal anaesthesia after administration of plain none of which provoked pain. Absent from this report are bupivacaine 0.5%. Their principal finding was the enor- descriptions of similar assessments being performed by mous range of concentrations (26–781 mgml21), which Hildebrandt on Bier. This was not out of deference to did not correlate with the level of block. Further, the varia- Bier, but rather the Pravaz syringe failed to fit on the bility in samples obtained at randomized time-points needle, and a significant amount of the cocaine intended between 5 and 45 min post-injection was so large as to for Bier’s subarachnoid space was lost, resulting in a prohibit any meaningful pharmacokinetic assessment. An failed block. even larger variability (3.36–1020 mgml21) was seen in Although the aetiology of this historic failure is no the second study9 where samples were obtained from mystery, a particularly perplexing and frustrating problem patients who had received the same anaesthetic but had with spinal anaesthesia is the occasional failure to achieve inadequate anaesthesia. The range of concentrations from adequate sensory block, despite an apparently orthodox these two studies overlapped, with only four of the 20 injection of an adequate dose of local anaesthetic. failed spinals having a CSF concentration below any Believing certain subjects to be ‘hyper-resistant’ to spinal associated with successful spinal anaesthesia. anaesthesia, Sebrechts2 coined the term ‘rachi-resistance’ Although these findings may appear incomprehensible, in 1934 to describe this phenomenon. He postulated that they can be understood by appreciating the limitations of this aberration reflected a ‘peculiar idiosyncrasy which single-site assessments of CSF anaesthetic concentrations, renders the nerve roots of certain individuals insensitive or a point well appreciated and considered by the authors of resistant to the action of anaesthetic solutions’. This these two research papers. Because local anaesthetic will concept garnished a modicum of support, some proponents often distribute unevenly within the subarachnoid space, proposing that this effect might be due to reduced per- the concentration at a single point cannot be used to deter- meability of the roots.3 However, most were critical, mine the true volume of distribution, nor can concen- suggesting such failures more likely reflect inactive sol- trations found below the conus reliably predict those at ution, or a variety of anatomic variations or abnormalities more rostral locations servicing clinically relevant derma- such as arachnoid adhesions, unusual arrangements of the tomes. Consequently, faced with a relatively low concen- dentate ligament, or a dilated lower end of the thecal tration, clinical correlation or imaging is needed to permit sac.4–6 Ultimately, the term ‘rachi-resistance’ disappeared any distinction between a poor injection, a large volume from the anaesthesia literature. of CSF below the conus, extensive spread, or an anoma- A review article and two clinical studies in this issue of lous sample. With the exception of technical failure, the British Journal of Anaesthesia attempt to shed light on similar considerations apply to interpretation of high this elusive subject. In their systematic review, Fettes and concentrations.

# The Author [2009]. Published by Oxford University Press on behalf of The Board of Directors of the British Journal of Anaesthesia. All rights reserved. For Permissions, please email: [email protected] Editorial I

Despite these limitations, there are some interesting K. Drasner insights that can be gleaned from the data. For example, the highest concentration in any sample came from a University of California patient with inadequate anaesthesia, supporting the San Francisco concept of maldistribution as an aetiology of failure, a CA 94110 point highlighted by the authors. A similar observation USA was made in a very early study of hyperbaric lidocaine administered through continuous spinal catheters.10 Of E-mail: [email protected] 16 patients receiving an initial injection of lidocaine 150–200 mg, there was only one failure requiring general anaesthesia. The anaesthetic concentration in the CSF sample from this patient was roughly 2 SD above the mean, References and an X-ray demonstrated the catheter to be curled with 1 Bier A. Versuche fiber Cocainisirung des Rfickenmarkes. Deutsche the end resting in the cul-de-sac at the level of the second Zeitschriftir Chirurgie 1899; 51: 361–9 sacral . Such findings are also consistent with data 2 Sebrechts J. Spinal anaesthesia. Br J Anaesth 1934; 12: 4–27 obtained from in vitro investigations modelling subarach- 3 Black WR, Walters GAB. ‘Rachi-resistance’ and spinal anaesthesia. noid anaesthetic distribution.11 Br Med J 1937; 1: 218–9 4 Lake NC. ‘Rachi-resistance’ and spinal anaesthesia. Br Med J 1937; That failure from maldistribution cannot be readily distin- 1: 297 guished clinically from technical failure has important impli- 5 Hewer CL. ‘Rachi-resistance’ and spinal anaesthesia. Br Med J cations for management of a failed spinal. In such cases, 1937; 1: 360 repetitive injection can distribute in the same pattern, reinfor- 6 Burke J. ‘Rachi-resistance’ and spinal anaesthesia. Br Med J 1937; 1: cing the already high concentrations, and potentially result- 584–5 ing in neurotoxic injury. Review of the closed claims 7 Fettes PDW, Jansson JR, Wildsmith JAW. Failed spinal anaesthesia: database12 and case reports13 – 15 appear to confirm these con- mechanisms, management and prevention. Br J Anaesth 2009; 102: 739–48 cerns. On the basis of these considerations, we previously 12 8 Ruppen W, Steiner LA, Drewe J, Hauenstein L, Brugger S, made suggestions for management of a failed spinal, which Seeberger MD. Bupivacaine concentrations in the lumbar cere- included evaluation of the likelihood of technical error and brospinal fluid of patients during spinal anaesthesia. Br J Anaesth adjustment of dosage for the second injection. However, 2009; 102: 832–8 these recommendations impart significant delay, as one must 9 Steiner LA, Hauenstein L, Ruppen W, Hampl KF, Seeberger MD. allow sufficient time for achievement of near-maximal block Bupivacaine concentrations in lumbar cerebrospinal fluid in before assessment of sensory anaesthesia. A more efficient— patients with failed spinal anaesthesia. Br J Anaesth 2009; 102: 839–44 and probably much safer—alternative is to simply limit the 10 Mo¨rch ET, Rosenberg MK, Truant AT. Lidocaine for spinal anaes- combined anaesthetic dosage to the maximum reasonable to thesia. A study of the concentration in the spinal fluid. Acta administer as a single intrathecal injection. Anaesthesiol Scand 1957; 1: 105–15 In four of the failed spinals, there was no detectable 11 Rigler ML, Drasner K. Distribution of catheter-injected local sensory anaesthesia, despite CSF concentrations that anesthetic in a model of the subarachnoid space. Anesthesiology exceeded the lowest value associated with a completely 1991; 75: 684–92 successful block. And, in one case, the concentration was 12 Drasner K, Rigler M. Repeat injection after a ‘failed spinal’—at times, a potentially unsafe practice. Anesthesiology 1991; 75: above the 5th percentile for successful spinal anaesthesia, 713–4 the cut-off (albeit somewhat arbitrary) that the authors 13 Waters JH, Watson TB, Ward MG. Conus medullaris injury fol- thought should be adequate for successful spinal anaesthe- lowing both tetracaine and lidocaine spinal anesthesia. J Clin sia. Although it is possible that this reflects the limitations Anesth 1996; 8: 656–8 of sampling and errant concentrations obtained in the 14 Hirabayashi Y, Konishi R, Shimizu R. Neurologic symptom associ- study of successful spinals, it is intriguing to consider ated with a repeated injection after failed spinal anesthesia. the possibility that these patients were actually resistant to Anesthesiology 1998; 89: 1294–5 15 Loo CC, Irestedt L. Cauda equina syndrome after spinal anaes- the anaesthetic. Such consideration is more than a theoreti- thesia with hyperbaric 5% lignocaine: a review of six cases of cal possibility, as there are well-described mutations of the cauda equina syndrome reported to the Swedish Pharmaceutical voltage-gated sodium channel that can profoundly affect Insurance 1993–1997. Acta Anaesthesiol Scand 1999; 43: function, including altered sensitivity to anaesthetic.16 17 371–9 Further advances in pharmacogenetics might thus identify 16 Catterall WA, Dib-Hajj S, Meisler MH, Pietrobon D. Inherited genetic diversity as a factor in spinal anaesthetic failure, neuronal ion channelopathies: new windows on complex neuro- justifying a reintroduction of the term ‘rachi-resistance’ logical diseases. J Neurosci 2008; 28: 11768–77 17 Sheets PL, Jackson JO, Waxman SG, Dib-Hajj SD, Cummins TR. A into the language of anaesthesia. Who knows, Sebrechts Nav1.7 channel mutation associated with hereditary belief that there are familial tendencies in susceptibility, erythromelalgia contributes to neuronal hyperexcitability and and that the ‘the Anglo-Saxon ...has resistance to larger displays reduced lidocaine sensitivity. J Physiol 2007; 581: doses than the Italian’ might just turn out to be correct. 1019–31

730 Current Anaesthesia & Critical Care 20 (2009) 71–73

Contents lists available at ScienceDirect

Current Anaesthesia & Critical Care

journal homepage: www.elsevier.com/locate/cacc

FOCUS ON: REGIONAL ANAESTHESIA Role of ultrasound in modern day regional anaesthesia

Narendra Babu Siddaiah, Anand Sardesai*

Addenbrooke’s Hospital, Cambridge University Hospitals NHS Trust, Cambridge CB2 0QQ, United Kingdom

summary

Keywords: The aim of any regional anaesthesia technique is to locate target nerves and deposit local anaesthetic Ultrasound around them. Nerve stimulation using anatomical landmarks has been conventional method of nerve Regional anaesthesia localisation. With the advance in technology, the use of ultrasound in regional anaesthesia has been Nerve block steadily increasing. In addition to localisation of nerves and other structures under direct visualisation, ultrasound has the potential to reduce complications and minimise the local anaesthetic toxicity by injecting right dose at the right site. The principles, advantages, disadvantages and applications of ultrasound have been reviewed in this article. Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction emits and receives high-frequency sound waves (greater than 20,000 cycles/s, 20 kHz) and thus ultrasound waves are not audible Nerve blocks have been traditionally done using anatomical to the human ear. Ultrasound frequencies useful in clinical medi- landmarks to guess approximate position of the nerve and then cine are in the megahertz (MHz) range.5 using either paraesthesia or nerve stimulation to confirm the The ultrasound probe is a transducer that contains crystals proximity of needle to the nerve. This assumes that every person’s which exhibit piezoelectric effect. Piezoelectricity is the ability of anatomy is same, yet we know that this assumption is wrong. There a substance to convert mechanical energy into electricity and vice are interpersonal variations and also variations in the same person versa. When an electrical current is applied to piezoelectric on different sides of the body.1 It is also assumed with these ‘blind crystals (quartz) within the ultrasound probe, they vibrate techniques’ that when local anaesthetic is injected it spreads resulting in ultrasound waves. As the ultrasound waves pass uniformly around the nerves which if does not happen results in through body tissues of different acoustic impedance, the signals block failure. ‘Threshold current’ i.e. the current which indicates are attenuated. The degree of attenuation of signals depends on that the position of the stimulating needle is close to the nerve, but the attenuation coefficient of the tissue. Also, it varies directly not close enough to cause nerve damage, is relatively poorly with the frequency of the ultrasound wave and the distance from identified.2 It is possible to be actually eliciting paraesthesia or be the source. Reflected waves are transformed into electrical signals intraneural with the needle without producing any motor by the transducer which are processed to generate an image on response.3,4 Thus the science of nerve stimulation is neither precise the screen.6 nor perfect. Echogenicity or how bright the object appears depends on the La Grange and his colleagues first reported the use of ultrasound amount of ultrasound reflected back to the probe; the greater the for providing nerve block in 1978. They successfully performed reflection, the brighter is the image and vice versa. Structures with supraclavicular brachial plexus blocks with the help of a Doppler high water content like blood let the ultrasound waves pass ultrasound blood-flow detector to identify the subclavial artery.5 through with minimal reflection and so appear black or dark (hypoechoic). On the other hand, waves are reflected and scattered 2. Principles by the bone and tendons, hence these appear white or bright (hyperechoic). With bone, the periosteum appears as a white Sound that we hear is between the frequency range of 20 and hyperechoic line. Structures of intermediate density such as the 20,000 Hz (20–20,000 cycles/s). An ultrasound probe (transducer) liver parenchyma or the thyroid gland reflect less. Hence they appear grey on the screen. The depth of the structure from the probe is calculated by the * Corresponding author. E-mail addresses: [email protected] (N.B. Siddaiah), sardesai1@aol. software accounting for the speed of sound in tissue (1540 m/s on com (A. Sardesai). average) and the time taken for the ultrasound to return.

0953-7112/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.cacc.2008.12.004 72 N.B. Siddaiah, A. Sardesai / Current Anaesthesia & Critical Care 20 (2009) 71–73

3. Equipment 5. Advantages of ultrasound guided nerve identification

For anaesthesia purposes we need an ultrasound machine  Nerves can be easily identified. The structures surrounding the which is ideally small in size, portable as it is likely to be shared nerve like blood vessels (e.g. saphenous vein)12 can be easily between different theatres and different anaesthetists and durable identified which in turn can help the identification of the as it has to move between theatres and by different anaesthetists so nerves. damage during transit is likely.  Needles can be visualised as a hyperechoic line. In real-time Colour Doppler is used to differentiate vascular from nonvas- navigation of ultrasound, the needle can be guided towards cular structures. Low frequency probes (3–5 MHz) are useful to the nerve. This can prevent the risk of inadvertent needle scan deeper structures like liver, gallbladder and kidneys while entry into the blood vessels, pleura13 or structures like spinal a frequency of 4–7 MHz is better for deep structures, such as the canal.14 brachial plexus in the infraclavicular region and the sciatic nerve in  Local anaesthetic spread can be seen with the ultrasound so if adults. inadequate a further injection could be made.15 Incomplete Superficial structures such as the supraclavicular brachial plexus nerve blocks can be prevented by ensuring circumferential require high-frequency probes (10–16 MHz) that provide high axial spread of the local anaesthetic. resolution.  Ultrasound can identify the point of division of the nerve.16 The hard disk should be high capacity with a CD burner in This is of help when complete nerve block before the division order to store and edit a large number of images and films 7 of nerve is needed as local anaesthetic can be injected although images and video can be stored on external hard disks proximal to the point of division of nerve identified by as well. ultrasound. As the competition between manufacturers is increasing and as  Placement of peripheral nerve catheters under direct the technology improves the machines are getting cheaper while vision.17,18 the image quality is improving. Ultrasound machines used by the  Ultrasound guided nerve blocks can be a relatively painless radiologists have multiple probes and softwares. They are useful for procedure when muscle contraction is avoided without nerve both diagnostic and therapeutic purposes. These are highly stimulation.19,20 sophisticated, bulky and expensive varying from £100,000 to  In blocks such as the 3-in-1 block, it has been demonstrated £150,000. Whereas a simple portable ultrasound machine for that the success rate is higher using ultrasound and that the regional anaesthesia, peripheral arterial and central venous can- dose of local anaesthetic required is lower.21 nulation costs about £15,000.  Faster sensory onset time and longer duration of blocks.15,19  Improved quality of block.22  Less incidence of diaphragmatic paralysis as less volume can be 4. Appearance of peripheral nerve under ultrasound used for interscalene block.23  No risk of radiation. A single nerve fibre is surrounded by endoneurium. Group of nerve fibres forms a nerve fascicle. Each nerve fascicle is sur- Patient groups where ultrasound guided blocks are particularly rounded by perineurium. Nerve fascicles together form a nerve useful in: which is surrounded by epineurium. Nerve fibres themselves do not reflect any ultrasound so they appear dark. Only the  Children, in whom nerve blocks are performed under connective tissue surrounding the nerve fibres, fascicles and the anaesthesia. nerve itself reflects ultrasound and then appears bright or white.  Patients with peripheral neuropathy where it is difficult to On transverse (short-axis, nerve in cross-section) view that is elicit nerve stimulus although nerve damage can occur even generally used, a nerve is round or oval shaped with internal without direct nerve needle contact.24 hypoechoic (dark) nerve fibres and nerve fascicles surrounded by  Use of Doppler has been described in obese patients or where the hyperechoic (bright) connective tissue.8 It has been compared the vascular landmarks can’t be easily identified.25–27 to bunch of drinking straws viewed end on. In the long-axis view, the probe is parallel to the long axis of the Ultrasound-assisted nerve block has been described for: nerve. In this view nerves appear tubular with parallel bright and dark bands corresponding to the fascicles and interfascicular  Upper limb blocks: by interscalene,23,28 connective tissue respectively. supraclavicular,13,22 infraclavicular19,29 and axillary30 Roots of the brachial plexus appear hypoechoic in the inter- approaches. scalene and supraclavicular regions more like blood vessels  Lower limb blocks: femoral nerve,15,21 lumbar plexus,31 sciatic without any flow. One possible explanation that has been sug- nerve.32 gested is that the nerves at this level have not grouped into fascicles  Other nerve blocks: coeliac plexus,33 stellate ganglion,34 yet.8 When the brachial plexus is scanned distally typical appear- pudendal nerve35 ance of the peripheral nerve is seen.  Assessment of the depth of the epidural space36 and to assess Nerves vary in shape (round, oval or triangular) and a same the lumbar epidural space during pregnancy.37 nerve can have different shapes as it passes along different structures.9 Disadvantages of ultrasound guided nerve blocks: Nerves are fairly mobile structures as demonstrated by movement of the branches of the sciatic nerve in the popliteal  It has the potential to avoid or reduce nerve damage. Whether fossa with foot movement.10 In order to identify the nerve it can eliminate damage completely remains to be seen. Patient anatomy in terms of its size, depth and location before the related factors and operator errors can still result in nerve insertion of needle, a preliminary scan should be performed. This damage. helps in the assessment of point, angle and direction of needle  The ultrasound machines are still expensive although the cost insertion.11 is coming down. N.B. Siddaiah, A. Sardesai / Current Anaesthesia & Critical Care 20 (2009) 71–73 73

6. Summary vision: a preliminary report of four cases. Reg Anesth Pain Med 2008 Sep– Oct;33(5):477–82. 18. Mariano ER, Ilfeld BM, Cheng GS, Nicodemus HF, Suresh S. Feasibility of ultra- Ultrasound has revolutionised the practise of regional anaes- sound-guided peripheral nerve block catheters for pain control on pediatric thesia. It has helped in a better understanding of the anatomic medical missions in developing countries. Paediatr Anaesth 2008;18:598–601. variability between patients. It has potential to provide a safe and 19. Marhofer P, Sitzwohl C, Greher M, Kapral S. Ultrasound guidance for infraclavicular brachial plexus anaesthesia in children. Anaesthesia reliable nerve block. It remains to be seen how it delivers this 2004;59:642–6. promise. 20. Soeding PE, Sha S, Royse CE, Marks P, Hoy G, Royse AG. A randomized trial of ultrasound-guided brachial plexus anaesthesia in upper limb surgery. Anaesth Intensive Care 2005;33:719–25. References 21. Marhofer P, Schro¨gendorfer K, Wallner T, Koinig H, Mayer N, Kapral S. Ultra- sonographic guidance reduces the amount of local anesthetic for 3-in-1 blocks. 1. Cash CJC, Sardesai AM, Berman LH, Herrick MJ, Treece GM, Prager RW, et al. Reg Anesth Pain Med 1998;23:584–8. Spatial mapping of the brachial plexus using three-dimensional ultrasound. Br J 22. Williams SR, Chovinard P, Arcand G, Harris P, Ruel M, Boudreault D, et al. Radiol 2005;78:1086–94. Ultrasound guidance speeds execution and improves the quality of supra- 2. Tsai, Vuckovic I, Dilberovic F, Obhodzas M, Kapur E, Divanovic K, et al. Intensity clavicular block. Anesth Analg 2003;97:1618–23. of the stimulating current may not be a reliable indicator of intraneural needle 23. Riazi S, Carmichael N, Awad I, Holtby RM, McCartney CJ. Effect of local anaes- placement. Reg Anesth Pain Med 2008;33:207–10. thetic volume (20 vs 5 ml) on the efficacy and respiratory consequences of 3. Urmey WF, Stanton J. Inability to consistently elicit a motor response following ultrasound-guided interscalene brachial plexus block. Br J Anaesth 2008 sensory paresthesia during interscalene block administration. Anesthesiology Oct;101(4):549–56 [Epub 2008 Aug 4]. 2002 Mar;96(3):552–4. 24. Koff MD, Cohen JA, McIntyre JJ, Carr CF, Sites BD. Severe brachial plexopathy 4. Choyce A, Chan VW, Middleton WJ, Knight PR, Peng P, McCartney CJ. What is after an ultrasound-guided single-injection nerve block for total shoulder the relationship between paresthesia and nerve stimulation for axillary arthroplasty in a patient with multiple sclerosis. Anesthesiology 2008 brachial plexus block? Reg Anesth Pain Med 2001 Mar–Apr;26(2):100–4. Feb;108(2):325–8. 5. La Grange P, Foster PA, Pretorius LK. Application of the Doppler ultrasound 25. Abramowitz HB, Cohen CH. Use of Doppler for difficult axillary block. Anes- blood flow detector in supraclavicular brachial plexus block. Br J Anaesth thesiology 1981;55:603. 1978;50:965–7. 26. Kestembaum AD, Steuer M, Marano M. Doppler guided axillary block in a burn 6. Kossoff G. Basic physics and imaging characteristics of ultrasound. World J Surg patient. Anesthesiology 1990;73:586–7. 2000;24. p. 134–42. 27. Vaghadia H, Jenkins LC. Use of a Doppler ultrasound stethoscope for intercostal 7. Marhofer P, Greher M, Kapral S. Ultrasound guidance in regional anaesthesia. Br nerve block. Can J Anaesth 1988;35:86–9. J Anaesth 2005;94:7–17. 28. Sinha SK, Abrams JH, Weller RS. Ultrasound-guided interscalene needle 8. Silvestri E, Martinoli C, Derchi LE, Bertolotto M, Chiaramondia M, Rosenberg I. placement produces successful anesthesia regardless of motor stimulation Echotexture of peripheral nerves: correlation between ultrasound and his- above or below 0.5 mA. Anesth Analg 2007;105:848–52. tologic findings and criteria to differentiate tendons. Radiology 1995;197: 29. Sandhu NS, Capan LM. Ultrasound-guided infraclavicular brachial plexus block. 291–6. Br J Anaesth; 2002 Aug. 9. Schafhalter-Zoppoth I, Gray AT. The musculocutaneous nerve: ultrasound 30. Lo N, Brull R, Perlas A, Chan VW, McCartney CJ, Sacco R, et al. Evolution of appearance for peripheral nerve block. Reg Anesth Pain Med 2005;30:385–90. ultrasound guided axillary brachial plexus blockade: retrospective analysis of 10. Schafhalter-Zoppoth I, Younger SJ, Collins AB, Gray AT. The ‘‘seesaw’’ sign: 662 blocks. Can J Anaesth 2008 Jul;55(7):408–13. improved sonographic identification of the sciatic nerve. Anesthesiology 31. Karmakar MK, Ho AM-H, Li X, Kwok WH, Tsang K, Ngan Kee WD. Ultrasound- 2004;101:808–9. guided lumbar plexus block through the acoustic window of the lumbar 11. Vincent WS, Chan MD. Ultrasound imaging for nerve block: a standard practice ultrasound trident. Br J Anaesth 2008;100(4):533–7. for the future? ASA Newsl; 2005:59. 32. Karmakar MK, Kwok WH, Ho AM, Tsang K, Chui PT, Gin T. Ultrasound-guided 12. Gray AT, Collins AB. Ultrasound-guided saphenous nerve block. Reg Anesth Pain sciatic nerve block: description of a new approach at the subgluteal space. Br J Med 2003;28:148. Anaesth;2007. 13. Kapral S, Krafft P, Eibenberger K, Fitzgerald R, Gosch M, Weinstabl C. Ultra- 33. Caratozzolo M, Lirici MM, Consalvo M, Marzano F, Fumarola E, Angelini L. sound-guided supraclavicular approach for regional anesthesia of the brachial Ultrasound-guided alcoholization of celiac plexus for pain control in oncology. plexus. Anesth Analg 1994;78:507–13. Surg Endosc 1997;11:239–44. 14. Van Geffen GJI, Rettig HC, Koornwinder T, Renes S, Gielen MJM. Ultrasound- 34. Kapral S, Krafft P, Gosch M, Fleischmann D, Weinstabl C. Ultrasound imaging for guided training in the performance of brachial plexus block by the posterior stellate ganglion block: direct visualization of puncture site and local anes- approach: an observational study. Anaesthesia 2007;62:1024–8. thetic spread. A pilot study. Reg Anesth 1995;20:323–8. 15. Marhofer P, Schro¨gendorfer K, Koinig H, Kapral S, Weinstabl C, Mayer N. 35. Rofaeel A, Peng P, Louis I, Chan V. Feasibility of real-time ultrasound for Ultrasonographic guidance improves sensory block and onset time of three-in- pudendal nerve block in patients with chronic perineal pain. Reg Anesth Pain one blocks. Anesth Analg;1997. Med 2008;33:165–9. 16. Schwemmer U, Markus CK, Greim CA, Brederlau J, Kredel M, Roewer N. Sono- 36. Currie LM. Measurement of the depth to the epidural space using ultrasound. graphic imaging of the sciatic nerve division in the popliteal fossa. Ultraschall Br J Anaesth 1984;56:345–7. Med 2005 Dec;26(6):496–500. 37. Grau T, Leipold RW, Horter J, Conradi R, Martin E, Motsch J. The lumbar epidural 17. Koscielniak-Nielsen ZJ, Rasmussen H, Hesselbjerg L. Long-axis ultrasound space in pregnancy: visualization by ultrasonography. Br J Anaesth imaging of the nerves and advancement of perineural catheters under direct 2001;86:798–804.