- Email: [email protected]
Clonidine as an Adjuvant to Intrathecal Local Anesthetics for Surgery: Systematic Review of Randomized Trials
Clonidine as an Adjuvant to Intrathecal Local Anesthetics for Surgery: Systematic Review of Randomized Trials Nadia Elia, M.D., Xavier Culebras, M.D., Christian Mazza, Ph.D., Eduardo Schiffer, M.D., and Martin R. Tramèr, M.D., D.Phil. Background and Objectives: Clonidine is added to intrathecal local anesthetics to improve intraoperative analgesia and to increase the duration of sensory and motor block. The aim of this systematic review is to quantify beneficial and harmful effects of clonidine when used as an adjuvant to intrathecal local anesthetics for surgery. Methods: We included data from 22 randomized trials (1,445 patients) testing a large variety of doses of clonidine, added to intrathecal bupivacaine, mepivacaine, prilocaine, or tetracaine. Results: Clonidine 15 to 150 g prolonged in a linear, dose-dependent manner, the time to 2 segment regression (range of means, 14 to 75 minutes) and the time to regression to L2 (range of means, 11 to 128 minutes). The time to first analgesic request (median 101 minutes, range 35 to 310) and of motor block (median 47 minutes, range 6 to 131) was prolonged without evidence of dose-responsiveness. Time to achieve complete sensory or motor block, and extent of cephalic spread remained unchanged. There were fewer episodes of intraoperative pain with clonidine (relative risk, 0.24; 95% confidence interval [CI], 0.09-0.64; number needed to treat, 13) but more episodes of arterial hypotension (relative risk, 1.81; 95% CI 1.44-2.28; number needed to harm, 8) without evidence of dose-responsiveness. The risk of bradycardia was unchanged. Conclusions: This study may serve as a rational basis to help clinicians decide whether or not to combine clonidine with an intrathecal local anesthetic for surgery. The optimal dose of clonidine, however, remains unknown. Reg Anesth Pain Med 2008;33:159-167. Key Words:
Surgery, Regional anesthesia, Analgesia, Dose-response, Alpha2 adrenoreceptor agonist.
V
arious drugs may be added to intrathecal local anesthetics for surgery. For instance, clonidine, an ␣2 adrenoreceptor agonist, is thought to prolong the duration of surgical anesthesia. Previous studies have tested many different doses of clonidine, combined with a large variety of local anesthetics; therefore, the best regimen remains unknown. From the Division of Anesthesiology (N.E., E.S., M.R.T.), University Hospitals of Geneva, Geneva; Clinique de Genolier (X.C.), Genolier; and Division of Mathematics (C.M.), University of Geneva, Geneva, Switzerland. Accepted for publication October 2, 2007. Dr. Elia’s salary was provided by the EBCAP foundation (http://www.hcuge.ch/anesthesie/ebcap.htm). The project was funded by institutional funds. Part of this work was presented at the annual scientific meeting of the SSAR (Swiss Society of Anesthesia and Reanimation) in Interlaken, Switzerland, November 2-4, 2006. Reprint requests: Nadia Elia, M.D., Division of Anesthesiology, University Hospitals of Geneva, 24, rue Micheli-du-Crest, CH1211 Geneva 14, Switzerland. E-mail: [email protected] © 2008 by the American Society of Regional Anesthesia and Pain Medicine. 1098-7339/08/3302-0011$34.00/0 doi:10.1016/j.rapm.2007.10.008
Whether or not to add clonidine to intrathecal local anesthetics for surgery is a clinically relevant question. To answer that question, valid data on beneficial effects and potential for harm are needed. More than 10 years ago, the authors of a narrative review were unable to quantify the effect of clonidine added to regional anesthesia.1 Although there remains little doubt that clonidine has some beneficial effects, some authors have suggested that the interest of using intrathecal clonidine was limited, due to an increased risk of arterial hypotension, or sedation.2,3 Relevant randomized trials have been published recently, illustrating a persistent interest in the combination of clonidine and intrathecal local anesthetics. Dose-responsiveness of intrathecal clonidine alone has been evaluated in healthy volunteers4 and for the treatment of pain after cesarean delivery.5 Yet, we do not know how these data can be extrapolated to clonidine when added to local anesthetics, and to patients undergoing surgery.
Regional Anesthesia and Pain Medicine, Vol 33, No 2 (March–April), 2008: pp 159 –167
159
160
Regional Anesthesia and Pain Medicine Vol. 33 No. 2 March–April 2008
The aim of this systematic review is to provide a description of beneficial and harmful effects that can be expected when adding clonidine to intrathecal local anesthetics in patients undergoing surgery, to quantify these effects, and to test for doseresponsiveness.
Methods A wide search strategy was used to retrieve all randomized trials that compared intrathecal local anesthetic and clonidine, with the same local anesthetic regimen and a placebo (or no treatment). We searched the Medline, Embase, Cinahl, Biosis, Indmed, and Central databases for reports from 1966 to October 2006 related to intrathecal analgesia (spinal, intrathecal) and pain during surgery and anesthesia (postoperative, pain, surg*, anesthesia, anaesthesia) using the Boolean terms, “or” and “and.” Our search was limited to adults, without language restriction. We checked bibliographies of retrieved reports. We contacted investigators for translations, or for additional information. We excluded trials that were not adequately controlled (for instance, they used different local anesthetic regimens in experimental and control groups), those testing intrathecal opioids or continuous intrathecal anesthesia or combined intrathecal-epidural anesthesia or intrathecal clonidine alone (without a local anesthetic). Trials including healthy volunteers, less than 10 patients per group, or those testing intrathecal clonidine for labor were not considered. One author (X.C.) screened the abstracts of all retrieved reports and excluded articles that did not meet our inclusion criteria. Three authors (X.C., N.E., E.S.) independently read all included reports and assessed their methodological quality using a modified Oxford scale taking into account randomization, concealment of treatment allocation, blinding, and description of withdrawals.6 Because we included randomized trials only, the minimum score was 1, and the maximum was 7. One author (X.C.) extracted information on the number of patients analyzed, regimens of local anesthetics (type, dose, baricity), surgery, and dose of clonidine. Outcomes, beneficial or harmful, were extracted as reported in the original trials. Data were entered into standard spread sheet and were checked by 2 authors (N.E., E.S.) independently. Discrepancies were resolved by discussion with a fourth author (M.R.T.). Statistical Analyses Because small trials are likely to report on data that are due to random chance, and since such data
may threaten the validity of meta-analysis, we made the arbitrary decision to compute summary estimates only when an outcome was reported in at least 5 trials, or in at least 100 patients receiving clonidine. For continuous outcomes, effect size with 95% confidence interval (CI) was computed as the difference in the mean effects reported with clonidine and placebo. When mean differences were homogenous across trials (i.e., test for heterogeneity, Phetero ⬎ .1), we assumed that all trials were measuring the same true effect, and that differences were due to sampling variations. In this case, we pooled the data and computed a weighted mean difference with 95% CI using a fixed effect model. For binary outcomes, we calculated relative risks (RR) with 95% CI. When the RRs were homogenous across trials (P ⬎ .1), we pooled the data and computed a summary RR with 95% CI using a fixed effect model. As an estimate of the clinical relevance of a treatment effect, we computed numbers needed to treat (NNT) for beneficial and numbers needed to harm (NNH) for harmful binary outcomes, using RR and control event rate. We computed 95% CIs around the NNT/NNH point estimates when the RR was statistically significant.7 When a treatment effect was heterogeneous across trials (Phetero ⬍ .1), we examined whether heterogeneity was due to differences in clonidine doses. We used a linear regression model to assess whether an increase in dose was associated with an increase in treatment effect. Statistical significance was assumed when the 95% CI of the regression coefficient for clonidine dose did not include 0 (mean differences) or 1 (relative risks). Because some trials were dose-finding studies comparing more than one clonidine dose with the same control group, we included a covariance matrix of the error process taking into account the correlated nature of the data (Supplementary Appendix). We performed sensitivity analyses to test for the robustness of the results; estimations of coefficients were adjusted for various covariates (type and baricity of local anesthetics, definitions of outcomes).8 Additionally, we performed subgroup analyses using exclusively data from trials that tested clonidine-bupivacaine combinations (i.e., the most frequently tested combination). We further adjusted the coefficient estimates for bupivacaine doses. Analyses were repeated using weight-adjusted doses of clonidine, taking into account average body weights as reported in the original trials.
Intrathecal Clonidine for Surgery
Fig 1. Flow chart of retrieved, excluded, and analyzed trials. RCT, randomized controlled trial.
Analyses were performed using STATA (Version 9, STATA Corp, College Station, TX), Maple 9 (University of Geneva, Geneva, Switzerland), and Microsoft Excel 11.37 for Mac (Microsoft Corporation, Redmond, WA).
Results We retrieved 37 randomized trials but subsequently excluded 15, leaving 22 (1,445 adults) for analyses (Fig 1).9-30 One author provided supplemental data from 2 trials that were included in our analyses.22,26 Fourteen studies tested a single dose of clonidine; 8 were dose-finding studies (6 tested 2 doses, 2 tested 3 doses; Supplementary Table 1). Clonidine doses ranged from 15 to 150 g. Clonidine was combined with hyperbaric bupivacaine (11 trials), isobaric bupivacaine (6 trials), hyperbaric tetracaine (3 trials), isobaric prilocaine (1 trial), or hyperbaric mepivacaine (1 trial). Patients underwent orthopedic (10 trials), urologic (5 trials), gynecologic (1 trial), abdominal (1 trial), both orthopedic and lower abdominal (2 trials) procedures, or cesarean delivery (2 trials); in 1 trial, the type of surgery was not specified. The median quality score was 2.5 (range, 1 to 7). Fourteen studies (64%) were double-blinded; the procedure of randomization was adequately described in 10 (45%), and concealment of treatment allocation was described in 4 (18%). Sensory Block Time to 2 segment regression was reported in 9 studies9,10,12,14,18,19,22,26,29 (12 comparisons). Ten comparisons (83%) showed a significant increase in time to 2 segment regression with clonidine (Fig 2). Treatment effects across trials were heterogeneous, ranging from 14 to 75 minutes increase in 2 segment regression (Phetero ⬍ .001; Table 1). There was
•
Elia et al.
161
evidence of a significant dose-response (crude regression coefficient: 1.6; 95% CI 0.5-2.8; P ⫽ .006). For each increase in 10 g of intrathecal clonidine, the time to 2 segment regression was prolonged by a little less than 2 minutes. This result remained significant after adjustment for type, and baricity of local anesthetics, and after adjustment for the dose of local anesthetic when only trials using bupivacaine were considered (Table 2). Time to regression to L2 was reported in 7 studies9-12,14,19,24 (9 comparisons). Seven comparisons (78%) showed a significant increase in time to regression to L2 with clonidine. Treatment effects were heterogeneous (Phetero ⬍ .001), and there was evidence of a dose-response (crude regression coefficient: 4.0; 95% CI 2.1-5.8; P ⬍ .001), even after adjustment for the different covariates (Tables 1 and 2). Ten studies9-11,14,18,19,24,26,28,29 (13 comparisons) reported on the time to achieve complete sensory block. One comparison only (7.7%) suggested a statistically significant increase in time with clonidine. Treatment effects were heterogeneous, with weak evidence of dose-responsiveness (Table 2). Nine studies9,11-13,19,20,22,24,26 (14 comparisons) reported on the extent of cephalic spread of the sensory block. Seven comparisons (50%) showed a significant difference in the extent of the cephalic spread with clonidine. In 5 studies, the number of blocked dermatomes was increased with clonidine; in 2, it was decreased. Treatment effects were heterogeneous (Phetero ⬍ .001), but without evidence of dose-responsiveness (Tables 1 and 2).
Fig 2. Time to 2 segments regression of the sensory block. Data are from 3 dose-finding (circles) and 6 singledose studies (squares). The horizontal line through zero represents equality. Vertical lines are 95% confidence intervals. Symbols above the line of equality indicate that the time to 2 segments regression lasts longer with clonidine compared with placebo; when the 95% confidence interval does not cross the line of equality, the difference is statistically significant. Overlapping symbols are separated for graphical purposes.
162
Number of Trials Outcome Sensory block Time to 2 segments regression (min) Time to regression to L2 (min) Time to complete sensory block (min) Extent of cephalic spread (dermatomes) Motor block Time to complete motor block (min) Duration of complete motor block (min) Analgesia Time to first request of analgesia (min)
Number of Patients Analyzed
Outcome, Median of Means (Range)
Treatment Effect*
Figure
Single Dose
Multiple Dose
Control/Clonidine
Clonidine Doses, Range (g)
2
6
3
166/208
15-150
85 (62-133)
110 (101-178)
36 (14-75)
⬍ .001
Supplementary 1
5
2
122/153
75-150
154 (82-178)
197 (92-306)
51 (11-128)
⬍ .001
Supplementary 2
7
3
219/262
15-150
15 (8-26)
19 (10-46)
2.3 (⫺6-28)
.010
Supplementary 3
5
4
141/223
15-150
14 (10-20.5)
14 (9.5-20.5)
0.1 (⫺2.8-2.9)
⬍ .001
Supplementary 4
4
3
122/183
15-150
8 (5-19)
10 (5.7-19.6)
0.8 (⫺1.0-3.7)
.730
3
7
3
178/224
15-150
145 (57-205)
210 (115-268)
47 (6-131)
⬍ .001
4
6
3
203/269
15-150
171 (55-295)
278 (129-498)
101 (35-310)
⬍ .001
Control
Clonidine
Median (Range)
Phetero†
NOTE. Outcomes are shown when they were reported in more than 5 trials or in more than 100 patients receiving intrathecal clonidine. Abbreviation: Hetero, heterogeneity. *Treatment effect is mean difference between clonidine and control. †Phetero ⬍ .05 indicates that differences in treatment effects across studies are due to factors other than random sampling variation.
Regional Anesthesia and Pain Medicine Vol. 33 No. 2 March–April 2008
Table 1. Sensory Block, Motor Block, and Analgesia
Intrathecal Clonidine for Surgery
•
Elia et al.
163
Table 2. Crude and Adjusted Estimates of the Linear Association Between Clonidine Doses and Treatment Effects Regression Coefficient (95% CI) Outcome A. Sensory block Time to 2 segments regression (min) Time to regression to L2 (min) Time to complete sensory block (min) Extent of cephalic spread (dermatomes) B. Motor block Time to complete motor block (min) Duration of complete motor block (min) C. Analgesia Time to first request of analgesic (min) D. Adverse effects Intraoperative hypotension Intraoperative bradycardia
Crude
P
Adjusted*
P
Bupivacaine†
P
1.6 (0.5-2.8)
.006
1.9 (0.8-3.1)
.001
3.7 (0.1-7.4)
.048
4.0 (2.1-5.8)
⬍ .001
4.9 (0.6-9.3)
.026
2.4 (0.3-4.5)
.029
0.4 (⫺0.2-1.0)
.173
0.8 (0.3-1.3)
.002
0.3 (⫺0.1-0.8)
.161
0.01 (⫺0.12-0.14)
.883
0.02 (⫺0.11 to 0.15)
.740
0.69 (⫺0.63-2.03)
.306
⫺0.1 (⫺0.2-0.1)
.141
⫺0.2 (⫺0.5 to 0.1)
.147
0.01 (⫺0.8-0.8)
.976
1.4 (⫺2.7-5.4)
.513
3.3 (⫺0.5 to 7.0)
.088
1.0 (⫺0.6-2.6)
.223
0.2 (⫺5.5-5.8)
.954
3.2 (⫺1.4 to 7.8)
.172
0.6 (⫺5.0-6.2)
.824
1.00 (0.93-1.08) 1.03 (0.89-1.18)
.990 .723
.494 .644
0.99 (0.88-1.11) 1.02 (0.86-1.20)
.852 .824
1.03 (0.95-1.11) 1.04 (0.88-1.22)
NOTE. Regression coefficients of continuous outcomes (A-C) represent the expected average increase in treatment effects for each increase of 10 g of clonidine, assuming a linear dose-response. Regression coefficients of binary outcomes (D) are exponentiated, and represent the average increase in relative risks for each increase of 10 g of clonidine; the model assumes a linear relationship between clonidine dose and log (relative risk). Abbreviation: CI ⫽ confidence interval. *Adjusted for type and baricity of local anesthetics (A-C) and for definitions of outcomes (D). †Restricted to trials using bupivacaine: adjusted for baricity and dose of bupivacaine (A-C) and for definitions of outcomes (in D). P ⬍ .05 suggests that there is a significant difference in treatment effect due to the doses of clonidine.
Motor Block Seven studies (11 comparisons) reported on the time to achieve complete motor block; none were statistically significant. Treatment effects were homogeneous and therefore not evocative for dose-responsiveness (Tables 1 and 2). The weighted mean difference was not significant (0.72 minutes; 95% CI ⫺0.04-1.49). Ten studies9-12,14,18,19,21,22,26 (13 comparisons) reported on the duration of complete motor block (Fig 3). Eleven comparisons (85%) showed a significant increase in the duration with clonidine. Treatment effects were heterogeneous (Phetero ⬍ .001), but without evidence of dose-responsiveness (Tables 1 and 2). 9-11,13,14,24,26
cluded from the combined analysis, there was still a lack of evidence of dose-responsiveness. Seven studies9,20,22,24,26,29,30 (8 comparisons) reported on intraoperative pain. At least 1 episode of
Analgesia Nine studies15,19-22,24,26,27,30 (13 comparisons) reported on the time to first request of analgesia. Twelve comparisons (92%) showed a statistically significant increase in time with clonidine (Fig 4). Treatment effects were heterogeneous, but without evidence of dose-responsiveness, even after adjustments (Tables 1 and 2). The median increase in time was about 100 minutes. One study tested clonidine 75 g and reported on an extraordinary treatment effect (310 minutes).24 When that trial was ex-
Fig 3. Duration of complete motor block. Data are from 3 dose-finding (circles) and 7 single-dose studies (squares). The horizontal line through zero represents equality. Vertical lines are 95% confidence intervals. Symbols above the line of equality indicate that the duration of complete motor block lasts longer with clonidine compared with placebo; when the 95% confidence interval does not cross the line of equality, the difference is statistically significant. Overlapping symbols are separated for graphical purposes.
Abbreviations: CI, confidence interval; hetero, heterogeneity; n/a, not applicable. *RR, relative risk comparing clonidine with control. †NNT/H, number needed to treat/harm (a negative number needed to treat is a number needed to harm); 95% CI is shown only for statistically significant differences.
⫺8 (⫺16 to ⫺6) ⫺20 (n/a) 1.81 (1.44-2.28) 1.61 (0.87-2.99) .19 .93 15-150 15-150 152/486 (31.3) 27/330 (8.2) 72/372 (19.4) 12/231 (5.2) 5 8 Supplementary 5 Supplementary 6 Intraoperative hypotension Intraoperative bradycardia
12 5
NNT/H (95% CI)† Clonidine Control Multiple Dose
Number of Patients with Outcome/Total Number of Patients (%) Number of Trials
Single Dose Figure
Seventeen studies9-11,13-18,20,22,24-27,29,30 (23 comparisons) reported on arterial hypotension. Hypotension was defined as a decrease in systolic pressure ⬎20% (2 studies) or ⬎30% (10 studies) compared with baseline, a decrease in mean arterial pressure ⬎20% (1 study), mean arterial pressure ⬍70 mm Hg (2 studies), or need for intravenous ephedrine (2 studies). At least 1 episode of arterial hypotension was reported in 19.4% (72/372) of controls and 31.3% (152/486) of patients receiving clonidine; RR 1.81; 95% CI 1.44-2.28 (NNH 8). Relative risks were homogeneous (Phetero ⫽ .19), suggesting lack of dose-responsiveness. Thirteen studies11-14,17-19,21,22,24-26,29 (19 comparisons) reported on bradycardia. Bradycardia was defined as ⬍50 (10 studies), ⬍45 (1 study) or ⬍40 (2 studies) heartbeats per minute. At least 1 episode of bradycardia was reported in 5.2% (12/231) of controls and 8.2% (27/330) of patients receiving clonidine. Although the pooled RR suggested an increased risk of bradycardia with clonidine (RR 1.61), the difference did not reach statistical significance. Relative risks were homogeneous (P ⫽ .93), and crude and adjusted regression coefficients were nonsignificant (Tables 2 and 3).
Table 3. Adverse Effects
Adverse Effects
Clonidine Dose Range, g
Fig 4. Time to first request of analgesia. Data are from 3 dose-finding (circles) and 6 single-dose studies (squares). The horizontal line through zero represents equality. Vertical lines are 95% confidence intervals (CIs). Symbols above the line of equality indicate that the time to first request of analgesia is prolonged with clonidine compared with placebo; when the 95% CI does not cross the line of equality, the difference is statistically significant. Overlapping symbols are separated for graphical purposes.
intraoperative pain or the need for additional analgesia during surgery was reported in 9.4% (15/160) of controls and in 2.3% (4/173) of patients receiving clonidine; RR 0.24; 95% CI 0.09-0.64 (NNT 13). Relative risks were homogeneous (P ⫽ .51), suggesting lack of dose-responsiveness.
RR (95% CI)* Phetero
Treatment Effect
Regional Anesthesia and Pain Medicine Vol. 33 No. 2 March–April 2008
Outcome
164
Intrathecal Clonidine for Surgery
Further Outcomes Pain intensity after 24 hours, 24-hour morphine consumption, the need for general anesthesia due to unsuccessful intrathecal analgesia, incidence of tourniquet pain, surgeons’ evaluation of surgical conditions, sedation, time to walking or bladder voiding, postoperative hypotension, orthostatic hypotension at mobilization, or postoperative bradycardia were inconsistently reported; further analyses were deemed inappropriate. All results of analyses using weight-adjusted doses of clonidine were very similar.
Discussion Four main results emerge from these analyses. First, intrathecal clonidine prolongs the regression of the sensory block in a linear dose-dependent way. Second, clonidine prolongs the time to the first request of an analgesic, and the duration of complete motor block, with weak evidence of dose-responsiveness. Third, clonidine decreases the risk of intraoperative pain and increases the risk of arterial hypotension, without evidence of dose-responsiveness. Finally, clonidine has no relevant impact on the time to achieve complete sensory or motor block, on the extent of the cephalad spread of the sensory block, or on the risk of bradycardia. The original trials tested many different doses of clonidine and local anesthetic regimens. Unsurprisingly, treatment effects were heterogeneous. Understanding the causes of heterogeneity increases the scientific value and clinical relevance of a meta-analysis.8 We focused on the role of the dose of clonidine as a possible explanation for this heterogeneity. Testing for dose-responsiveness, by using data from independent trials remains a methodological challenge. Some may prefer to consider exclusively data from dosefinding studies.31,32 The methodological approach then consists in extracting the slopes of the linear associations that are reported in the original trials, and to compute a weighted-pooled estimate of the slopes across all trials. Our model is different because it combines data from both single and multiple dose studies. Because dose-finding studies compare 2 or more active groups with 1 single inactive control group, independence of observations can no longer be assumed and the regression model needs to be adapted to the use of such correlated data. Our model allowed controlling for differences in local anesthetic regimens and in outcome definitions. We found a significant linear relationship between clonidine doses and duration of the regres-
•
Elia et al.
165
sion of the sensory block. For the increase in the time to the first request of analgesia and of duration of motor block, the evidence of such a relationship was weak. Concerning the increase in the time to achieve complete motor block, the data were more homogenous; we may then be confident that the dose of clonidine had a marginal impact only on this outcome. To appraise the clinical relevance of our findings, we need to clarify what the ideal adjuvant to intrathecal local anesthetics for surgery would be. Intrathecal local anesthetics alone provide a rapid onset of sensory and motor block. Further shortening of these onsets with clonidine is probably of minor clinical relevance. Increasing the duration of surgical anesthesia allows for unexpected prolongation of surgical procedures, without the need for general anesthesia. Clonidine increases that duration by about 1 hour; this may be considered clinically relevant, although it may also be responsible for a delayed discharge from the recovery room, if discharge criteria include the level of sensory block. Patients who received intrathecal clonidine were less likely to experience intraoperative pain. This is clearly a beneficial effect. However, 9% of controls had at least 1 period of intraoperative pain. This high proportion was mainly due to data from 2 trials that were designed to study the effect of clonidine added to a suboptimal local anesthetic dose.20,26 Finally, prolongation of the postoperative pain-free period is clinically relevant, and there was evidence that this period could be extended by about 100 minutes when clonidine was added to a local anesthetic. In the absence of convincing data on doseresponsiveness, the choice of the dose may be driven by the fear of drug-related adverse effects. Not unexpectedly, the risk of intraoperative hypotension was increased with clonidine. We were, however, unable to demonstrate an increased risk of hypotension with increasing doses of clonidine. Almost 20% of controls experienced at least 1 episode of intraoperative hypotension, probably due to the intrathecal local anesthetics per se and the subsequent sympathetic block; the additional hypotensive effect of clonidine may have been partially masked. The evidence for an increased risk of bradycardia was poor. Both hypotension and bradycardia did not seem to pose clinical problems. Our analyses suffer mainly from 4 limitations. First, in the original trials, follow-up rarely exceeded 12 hours postoperatively; the long term analgesic profile of intrathecal clonidine remains unknown. Second, although a large variety of clonidine/ local anesthetic combinations were tested in these trials, there were no valid data on clonidine/ropiva-
166
Regional Anesthesia and Pain Medicine Vol. 33 No. 2 March–April 2008
caine or clonidine/levobupivacaine combinations. Third, it cannot be ruled out that we missed some reports, for instance, unpublished trials. And finally, most retrieved trials were of small size, and the quality of data reporting was limited. Both publication bias and an analysis that is mainly based on small and low quality trials are likely to lead to an overestimation of the effect of a treatment. In conclusion, adding clonidine to intrathecal local anesthetics for surgery increases the duration of the motor block, improves intraoperative analgesia, and delays the regression of the sensory block and the time to first analgesic request. For some of the reported effects, there is evidence of dose-responsiveness, for others, there is none. None of these outcomes is likely to last longer than 2 hours; whether they are considered clinically relevant and worthwhile depends on the context. The most prominent adverse effect is intraoperative hypotension. This study may serve as a rational basis to help clinicians decide whether or not to combine clonidine with an intrathecal local anesthetic for surgery. Whether adding clonidine to a low dose of an intrathecal local anesthetic offers advantages over a higher dose of the same local anesthetic alone remains unsolved. Finally, the optimal dose of clonidine when used as an adjuvant to an intrathecal local anesthetic remains unknown. Future trials should address these issues.
Acknowledgments Special thanks go to D. Haake, library of the Medical Faculty of Geneva University, for his help in searching electronic databases, and to the authors who responded to our inquiries. Individual trial data and additional figures and tables are freely accessible on the authors’ web site (http://www. hcuge.ch/anesthesie/data.htm).
Appendix Supplementary Material Supplementary Table 1, Figures 1-6, and Appendix can be found in the online version of this article at www.rapm.org (doi:10.1016/j.rapm.2008.10.008).
References 1. Eisenach JC, De Kock M, Klimscha W. Alpha(2)adrenergic agonists for regional anesthesia. A clinical review of clonidine (1984-1995). Anesthesiology 1996; 85:655-674. 2. Gabriel JS, Gordin V. Alpha 2 agonists in regional anesthesia and analgesia. Curr Opin Anaesthesiol 2001; 14:751-753.
3. Förster JG, Rosenberg PH. Clinically useful adjuvants in regional anaesthesia. Curr Opin Anaesthesiol 2003; 16:477-486. 4. Eisenach JC, Hood DD, Curry R. Relative potency of epidural to intrathecal clonidine differs between acute thermal pain and capsaicin-induced allodynia. Pain 2000;84:57-64. 5. Filos KS, Goudas LC, Patroni O, Polyzou V. Hemodynamic and analgesic profile after intrathecal clonidine in humans. A dose-response study. Anesthesiology 1994;81:591-601, 27A-28A. 6. Elia N, Tramer MR. Ketamine and postoperative pain–a quantitative systematic review of randomised trials. Pain 2005;113:61-70. 7. Tramèr MR, Walder B. Number needed to treat (or harm). World J Surg 2005;29:576-581. 8. Thompson S, Sharp S. Explaining heterogeneity in meta-analysis: a comparison of methods. Stat Med 1999;18:2693-2708. 9. Racle JP, Benkhadra A, Poy JY, Gleizal B. Prolongation of isobaric bupivacaine spinal anesthesia with epinephrine and clonidine for hip surgery in the elderly. Anesth Analg 1987;66:442-446. 10. Racle JP, Poy JY, Benkhadra A, Jourdren L, Fockenier F. Prolongation of spinal anesthesia with hyperbaric bupivacaine by adrenaline and clonidine in the elderly [in French]. Ann Fr Anesth Reanim 1988; 7:139-144. 11. Bonnet F, Diallo A, Saada M, Belon M, Guilbaud M, Boico O. Prevention of tourniquet pain by spinal isobaric bupivacaine with clonidine. Br J Anaesth 1989;63:93-96. 12. Bonnet F, Brun-Buisson V, Saada M, Boico O, Rostaing S, Touboul C. Dose-related prolongation of hyperbaric tetracaine spinal anesthesia by clonidine in humans. Anesth Analg 1989;68:619-622. 13. Wu CI, Wang JJ, Ning FS, Chao CH, Liu RJ, Chen CH, Mok MS. Effect of increasing amounts of clonidine during hyperbaric tetracaine spinal anesthesia [in Chinese]. Ma Zui Xue Za Zhi 1989;27:117123. 14. Seah YS, Chen C, Chung KD, Wong CH, Tan PP. Prolongation of hyperbaric bupivacaine spinal anesthesia with clonidine [in Chinese]. Ma Zui Xue Za Zhi 1991;29:533-537. 15. Fogarty DJ, Carabine UA, Milligan KR. Comparison of the analgesic effects of intrathecal clonidine and intrathecal morphine after spinal anaesthesia in patients undergoing total hip replacement. Br J Anaesth 1993;71:661-664. 16. Fukuda T, Dohi S, Naito H. Comparisons of tetracaine spinal anesthesia with clonidine or phenylephrine in normotensive and hypertensive humans. Anesth Analg 1994;78:106-111. 17. Niemi L. Effects of intrathecal clonidine on duration of bupivacaine spinal anaesthesia, haemodynamics, and postoperative analgesia in patients undergoing knee arthroscopy. Acta Anaesthesiol Scand 1994;38: 724-728. 18. De Negri P, Borrelli F, Salvatore R, Visconti C, De Vivo P, Mastronardi P. Spinal anesthesia with clonidine and
Intrathecal Clonidine for Surgery
19.
20.
21.
22.
23.
24.
25.
bupivacaine in young humans: interactions and effects on the cardiovascular system. Minerva Anestesiol 1997; 63:119-125. Larsen B, Dorscheid E, Macher-Hanselmann F, Büch U. Does intrathecal clonidine prolong the effect of spinal anesthesia with hyperbaric mepivacaine? A randomized double-blind study [in German]. Anaesthesist 1998;47:741-746. Pan PM, Huang CT, Wei TT, Mok MS. Enhancement of analgesic effect of intrathecal neostigmine and clonidine on bupivacaine spinal anesthesia. Reg Anesth Pain Med 1998;23:49-56. Juliao MC, Lauretti GR. Low-dose intrathecal clonidine combined with sufentanil as analgesic drugs in abdominal gynecological surgery. J Clin Anesth 2000;12:357362. Dobrydnjov I, Axelsson K, Samarutel J, Holmstrom B. Postoperative pain relief following intrathecal bupivacaine combined with intrathecal or oral clonidine. Acta Anaesthesiol Scand 2002;46:806-814. Kirdemir P, Gögüs¸ N. Intratekal Klonidin Unilateral Spinal Blogu Ve Analjeziyi Uzatiyor Mu [in Turkish]? Gülhane Tip Dergisi 2002;44:436-441. Santiveri X, Arxer A, Plaja I, Metje MT, Martinez B, Villalonga A, Lopez M. Anaesthetic and postoperative analgesic effects of spinal clonidine as an additive to prilocaine in the transurethral resection of urinary bladder tumours. Eur J Anaesthesiol 2002;19:589-593. Braz J, Koguti E, Braz L, Croitor L, Navarro L. Effects of clonidine associated to hyperbaric bupivacaine during high level spinal anesthesia [in Portugese]. Rev Bras Anesthesiol 2003;53:561-572.
•
Elia et al.
167
26. Dobrydnjov I, Axelsson K, Thorn SE, Matthiesen P, Klockhoff H, Holmstrom B, Gupta A. Clonidine combined with small-dose bupivacaine during spinal anesthesia for inguinal herniorrhaphy: a randomized double-blinded study. Anesth Analg 2003;96:14961503, table of contents. 27. Strebel S, Gurzeler JA, Schneider MC, Aeschbach A, Kindler CH. Small-dose intrathecal clonidine and isobaric bupivacaine for orthopedic surgery: a doseresponse study. Anesth Analg 2004;99:1231-1238, table of contents. 28. Jeon YT, Jeon YS, Kim YC, Bahk JH, Do SH, Lim YJ. Intrathecal clonidine does not reduce post-spinal shivering. Acta Anaesthesiol Scand 2005;49:1509-1513. 29. Kanazi GE, Aouad MT, Jabbour-Khoury SI, Al Jazzar MD, Alameddine MM, Al-Yaman R, Bulbul M, Baraka AS. Effect of low-dose dexmedetomidine or clonidine on the characteristics of bupivacaine spinal block. Acta Anaesthesiol Scand 2006;50:222-227. 30. van Tuijl I, van Klei WA, van der Werff DB, Kalkman CJ. The effect of addition of intrathecal clonidine to hyperbaric bupivacaine on postoperative pain and morphine requirements after Caesarean section: a randomized controlled trial. Br J Anaesth 2006;97:365-370. 31. Holt S, Suder A, Weatherall M, Cheng S, Shirtcliffe P, Beasley R. Dose-response relation of inhaled fluticasone propionate in adolescents and adults with asthma: meta-analysis. BMJ 2001;323:253-256. 32. Lee A, Ngan Kee WD, Gin T. A dose-response metaanalysis of prophylactic intravenous ephedrine for the prevention of hypotension during spinal anesthesia for elective cesarean delivery. Anesth Analg 2004; 98:483-90, table of contents.
Surgery, Regional anesthesia, Analgesia, Dose-response, Alpha2 adrenoreceptor agonist.
V
arious drugs may be added to intrathecal local anesthetics for surgery. For instance, clonidine, an ␣2 adrenoreceptor agonist, is thought to prolong the duration of surgical anesthesia. Previous studies have tested many different doses of clonidine, combined with a large variety of local anesthetics; therefore, the best regimen remains unknown. From the Division of Anesthesiology (N.E., E.S., M.R.T.), University Hospitals of Geneva, Geneva; Clinique de Genolier (X.C.), Genolier; and Division of Mathematics (C.M.), University of Geneva, Geneva, Switzerland. Accepted for publication October 2, 2007. Dr. Elia’s salary was provided by the EBCAP foundation (http://www.hcuge.ch/anesthesie/ebcap.htm). The project was funded by institutional funds. Part of this work was presented at the annual scientific meeting of the SSAR (Swiss Society of Anesthesia and Reanimation) in Interlaken, Switzerland, November 2-4, 2006. Reprint requests: Nadia Elia, M.D., Division of Anesthesiology, University Hospitals of Geneva, 24, rue Micheli-du-Crest, CH1211 Geneva 14, Switzerland. E-mail: [email protected] © 2008 by the American Society of Regional Anesthesia and Pain Medicine. 1098-7339/08/3302-0011$34.00/0 doi:10.1016/j.rapm.2007.10.008
Whether or not to add clonidine to intrathecal local anesthetics for surgery is a clinically relevant question. To answer that question, valid data on beneficial effects and potential for harm are needed. More than 10 years ago, the authors of a narrative review were unable to quantify the effect of clonidine added to regional anesthesia.1 Although there remains little doubt that clonidine has some beneficial effects, some authors have suggested that the interest of using intrathecal clonidine was limited, due to an increased risk of arterial hypotension, or sedation.2,3 Relevant randomized trials have been published recently, illustrating a persistent interest in the combination of clonidine and intrathecal local anesthetics. Dose-responsiveness of intrathecal clonidine alone has been evaluated in healthy volunteers4 and for the treatment of pain after cesarean delivery.5 Yet, we do not know how these data can be extrapolated to clonidine when added to local anesthetics, and to patients undergoing surgery.
Regional Anesthesia and Pain Medicine, Vol 33, No 2 (March–April), 2008: pp 159 –167
159
160
Regional Anesthesia and Pain Medicine Vol. 33 No. 2 March–April 2008
The aim of this systematic review is to provide a description of beneficial and harmful effects that can be expected when adding clonidine to intrathecal local anesthetics in patients undergoing surgery, to quantify these effects, and to test for doseresponsiveness.
Methods A wide search strategy was used to retrieve all randomized trials that compared intrathecal local anesthetic and clonidine, with the same local anesthetic regimen and a placebo (or no treatment). We searched the Medline, Embase, Cinahl, Biosis, Indmed, and Central databases for reports from 1966 to October 2006 related to intrathecal analgesia (spinal, intrathecal) and pain during surgery and anesthesia (postoperative, pain, surg*, anesthesia, anaesthesia) using the Boolean terms, “or” and “and.” Our search was limited to adults, without language restriction. We checked bibliographies of retrieved reports. We contacted investigators for translations, or for additional information. We excluded trials that were not adequately controlled (for instance, they used different local anesthetic regimens in experimental and control groups), those testing intrathecal opioids or continuous intrathecal anesthesia or combined intrathecal-epidural anesthesia or intrathecal clonidine alone (without a local anesthetic). Trials including healthy volunteers, less than 10 patients per group, or those testing intrathecal clonidine for labor were not considered. One author (X.C.) screened the abstracts of all retrieved reports and excluded articles that did not meet our inclusion criteria. Three authors (X.C., N.E., E.S.) independently read all included reports and assessed their methodological quality using a modified Oxford scale taking into account randomization, concealment of treatment allocation, blinding, and description of withdrawals.6 Because we included randomized trials only, the minimum score was 1, and the maximum was 7. One author (X.C.) extracted information on the number of patients analyzed, regimens of local anesthetics (type, dose, baricity), surgery, and dose of clonidine. Outcomes, beneficial or harmful, were extracted as reported in the original trials. Data were entered into standard spread sheet and were checked by 2 authors (N.E., E.S.) independently. Discrepancies were resolved by discussion with a fourth author (M.R.T.). Statistical Analyses Because small trials are likely to report on data that are due to random chance, and since such data
may threaten the validity of meta-analysis, we made the arbitrary decision to compute summary estimates only when an outcome was reported in at least 5 trials, or in at least 100 patients receiving clonidine. For continuous outcomes, effect size with 95% confidence interval (CI) was computed as the difference in the mean effects reported with clonidine and placebo. When mean differences were homogenous across trials (i.e., test for heterogeneity, Phetero ⬎ .1), we assumed that all trials were measuring the same true effect, and that differences were due to sampling variations. In this case, we pooled the data and computed a weighted mean difference with 95% CI using a fixed effect model. For binary outcomes, we calculated relative risks (RR) with 95% CI. When the RRs were homogenous across trials (P ⬎ .1), we pooled the data and computed a summary RR with 95% CI using a fixed effect model. As an estimate of the clinical relevance of a treatment effect, we computed numbers needed to treat (NNT) for beneficial and numbers needed to harm (NNH) for harmful binary outcomes, using RR and control event rate. We computed 95% CIs around the NNT/NNH point estimates when the RR was statistically significant.7 When a treatment effect was heterogeneous across trials (Phetero ⬍ .1), we examined whether heterogeneity was due to differences in clonidine doses. We used a linear regression model to assess whether an increase in dose was associated with an increase in treatment effect. Statistical significance was assumed when the 95% CI of the regression coefficient for clonidine dose did not include 0 (mean differences) or 1 (relative risks). Because some trials were dose-finding studies comparing more than one clonidine dose with the same control group, we included a covariance matrix of the error process taking into account the correlated nature of the data (Supplementary Appendix). We performed sensitivity analyses to test for the robustness of the results; estimations of coefficients were adjusted for various covariates (type and baricity of local anesthetics, definitions of outcomes).8 Additionally, we performed subgroup analyses using exclusively data from trials that tested clonidine-bupivacaine combinations (i.e., the most frequently tested combination). We further adjusted the coefficient estimates for bupivacaine doses. Analyses were repeated using weight-adjusted doses of clonidine, taking into account average body weights as reported in the original trials.
Intrathecal Clonidine for Surgery
Fig 1. Flow chart of retrieved, excluded, and analyzed trials. RCT, randomized controlled trial.
Analyses were performed using STATA (Version 9, STATA Corp, College Station, TX), Maple 9 (University of Geneva, Geneva, Switzerland), and Microsoft Excel 11.37 for Mac (Microsoft Corporation, Redmond, WA).
Results We retrieved 37 randomized trials but subsequently excluded 15, leaving 22 (1,445 adults) for analyses (Fig 1).9-30 One author provided supplemental data from 2 trials that were included in our analyses.22,26 Fourteen studies tested a single dose of clonidine; 8 were dose-finding studies (6 tested 2 doses, 2 tested 3 doses; Supplementary Table 1). Clonidine doses ranged from 15 to 150 g. Clonidine was combined with hyperbaric bupivacaine (11 trials), isobaric bupivacaine (6 trials), hyperbaric tetracaine (3 trials), isobaric prilocaine (1 trial), or hyperbaric mepivacaine (1 trial). Patients underwent orthopedic (10 trials), urologic (5 trials), gynecologic (1 trial), abdominal (1 trial), both orthopedic and lower abdominal (2 trials) procedures, or cesarean delivery (2 trials); in 1 trial, the type of surgery was not specified. The median quality score was 2.5 (range, 1 to 7). Fourteen studies (64%) were double-blinded; the procedure of randomization was adequately described in 10 (45%), and concealment of treatment allocation was described in 4 (18%). Sensory Block Time to 2 segment regression was reported in 9 studies9,10,12,14,18,19,22,26,29 (12 comparisons). Ten comparisons (83%) showed a significant increase in time to 2 segment regression with clonidine (Fig 2). Treatment effects across trials were heterogeneous, ranging from 14 to 75 minutes increase in 2 segment regression (Phetero ⬍ .001; Table 1). There was
•
Elia et al.
161
evidence of a significant dose-response (crude regression coefficient: 1.6; 95% CI 0.5-2.8; P ⫽ .006). For each increase in 10 g of intrathecal clonidine, the time to 2 segment regression was prolonged by a little less than 2 minutes. This result remained significant after adjustment for type, and baricity of local anesthetics, and after adjustment for the dose of local anesthetic when only trials using bupivacaine were considered (Table 2). Time to regression to L2 was reported in 7 studies9-12,14,19,24 (9 comparisons). Seven comparisons (78%) showed a significant increase in time to regression to L2 with clonidine. Treatment effects were heterogeneous (Phetero ⬍ .001), and there was evidence of a dose-response (crude regression coefficient: 4.0; 95% CI 2.1-5.8; P ⬍ .001), even after adjustment for the different covariates (Tables 1 and 2). Ten studies9-11,14,18,19,24,26,28,29 (13 comparisons) reported on the time to achieve complete sensory block. One comparison only (7.7%) suggested a statistically significant increase in time with clonidine. Treatment effects were heterogeneous, with weak evidence of dose-responsiveness (Table 2). Nine studies9,11-13,19,20,22,24,26 (14 comparisons) reported on the extent of cephalic spread of the sensory block. Seven comparisons (50%) showed a significant difference in the extent of the cephalic spread with clonidine. In 5 studies, the number of blocked dermatomes was increased with clonidine; in 2, it was decreased. Treatment effects were heterogeneous (Phetero ⬍ .001), but without evidence of dose-responsiveness (Tables 1 and 2).
Fig 2. Time to 2 segments regression of the sensory block. Data are from 3 dose-finding (circles) and 6 singledose studies (squares). The horizontal line through zero represents equality. Vertical lines are 95% confidence intervals. Symbols above the line of equality indicate that the time to 2 segments regression lasts longer with clonidine compared with placebo; when the 95% confidence interval does not cross the line of equality, the difference is statistically significant. Overlapping symbols are separated for graphical purposes.
162
Number of Trials Outcome Sensory block Time to 2 segments regression (min) Time to regression to L2 (min) Time to complete sensory block (min) Extent of cephalic spread (dermatomes) Motor block Time to complete motor block (min) Duration of complete motor block (min) Analgesia Time to first request of analgesia (min)
Number of Patients Analyzed
Outcome, Median of Means (Range)
Treatment Effect*
Figure
Single Dose
Multiple Dose
Control/Clonidine
Clonidine Doses, Range (g)
2
6
3
166/208
15-150
85 (62-133)
110 (101-178)
36 (14-75)
⬍ .001
Supplementary 1
5
2
122/153
75-150
154 (82-178)
197 (92-306)
51 (11-128)
⬍ .001
Supplementary 2
7
3
219/262
15-150
15 (8-26)
19 (10-46)
2.3 (⫺6-28)
.010
Supplementary 3
5
4
141/223
15-150
14 (10-20.5)
14 (9.5-20.5)
0.1 (⫺2.8-2.9)
⬍ .001
Supplementary 4
4
3
122/183
15-150
8 (5-19)
10 (5.7-19.6)
0.8 (⫺1.0-3.7)
.730
3
7
3
178/224
15-150
145 (57-205)
210 (115-268)
47 (6-131)
⬍ .001
4
6
3
203/269
15-150
171 (55-295)
278 (129-498)
101 (35-310)
⬍ .001
Control
Clonidine
Median (Range)
Phetero†
NOTE. Outcomes are shown when they were reported in more than 5 trials or in more than 100 patients receiving intrathecal clonidine. Abbreviation: Hetero, heterogeneity. *Treatment effect is mean difference between clonidine and control. †Phetero ⬍ .05 indicates that differences in treatment effects across studies are due to factors other than random sampling variation.
Regional Anesthesia and Pain Medicine Vol. 33 No. 2 March–April 2008
Table 1. Sensory Block, Motor Block, and Analgesia
Intrathecal Clonidine for Surgery
•
Elia et al.
163
Table 2. Crude and Adjusted Estimates of the Linear Association Between Clonidine Doses and Treatment Effects Regression Coefficient (95% CI) Outcome A. Sensory block Time to 2 segments regression (min) Time to regression to L2 (min) Time to complete sensory block (min) Extent of cephalic spread (dermatomes) B. Motor block Time to complete motor block (min) Duration of complete motor block (min) C. Analgesia Time to first request of analgesic (min) D. Adverse effects Intraoperative hypotension Intraoperative bradycardia
Crude
P
Adjusted*
P
Bupivacaine†
P
1.6 (0.5-2.8)
.006
1.9 (0.8-3.1)
.001
3.7 (0.1-7.4)
.048
4.0 (2.1-5.8)
⬍ .001
4.9 (0.6-9.3)
.026
2.4 (0.3-4.5)
.029
0.4 (⫺0.2-1.0)
.173
0.8 (0.3-1.3)
.002
0.3 (⫺0.1-0.8)
.161
0.01 (⫺0.12-0.14)
.883
0.02 (⫺0.11 to 0.15)
.740
0.69 (⫺0.63-2.03)
.306
⫺0.1 (⫺0.2-0.1)
.141
⫺0.2 (⫺0.5 to 0.1)
.147
0.01 (⫺0.8-0.8)
.976
1.4 (⫺2.7-5.4)
.513
3.3 (⫺0.5 to 7.0)
.088
1.0 (⫺0.6-2.6)
.223
0.2 (⫺5.5-5.8)
.954
3.2 (⫺1.4 to 7.8)
.172
0.6 (⫺5.0-6.2)
.824
1.00 (0.93-1.08) 1.03 (0.89-1.18)
.990 .723
.494 .644
0.99 (0.88-1.11) 1.02 (0.86-1.20)
.852 .824
1.03 (0.95-1.11) 1.04 (0.88-1.22)
NOTE. Regression coefficients of continuous outcomes (A-C) represent the expected average increase in treatment effects for each increase of 10 g of clonidine, assuming a linear dose-response. Regression coefficients of binary outcomes (D) are exponentiated, and represent the average increase in relative risks for each increase of 10 g of clonidine; the model assumes a linear relationship between clonidine dose and log (relative risk). Abbreviation: CI ⫽ confidence interval. *Adjusted for type and baricity of local anesthetics (A-C) and for definitions of outcomes (D). †Restricted to trials using bupivacaine: adjusted for baricity and dose of bupivacaine (A-C) and for definitions of outcomes (in D). P ⬍ .05 suggests that there is a significant difference in treatment effect due to the doses of clonidine.
Motor Block Seven studies (11 comparisons) reported on the time to achieve complete motor block; none were statistically significant. Treatment effects were homogeneous and therefore not evocative for dose-responsiveness (Tables 1 and 2). The weighted mean difference was not significant (0.72 minutes; 95% CI ⫺0.04-1.49). Ten studies9-12,14,18,19,21,22,26 (13 comparisons) reported on the duration of complete motor block (Fig 3). Eleven comparisons (85%) showed a significant increase in the duration with clonidine. Treatment effects were heterogeneous (Phetero ⬍ .001), but without evidence of dose-responsiveness (Tables 1 and 2). 9-11,13,14,24,26
cluded from the combined analysis, there was still a lack of evidence of dose-responsiveness. Seven studies9,20,22,24,26,29,30 (8 comparisons) reported on intraoperative pain. At least 1 episode of
Analgesia Nine studies15,19-22,24,26,27,30 (13 comparisons) reported on the time to first request of analgesia. Twelve comparisons (92%) showed a statistically significant increase in time with clonidine (Fig 4). Treatment effects were heterogeneous, but without evidence of dose-responsiveness, even after adjustments (Tables 1 and 2). The median increase in time was about 100 minutes. One study tested clonidine 75 g and reported on an extraordinary treatment effect (310 minutes).24 When that trial was ex-
Fig 3. Duration of complete motor block. Data are from 3 dose-finding (circles) and 7 single-dose studies (squares). The horizontal line through zero represents equality. Vertical lines are 95% confidence intervals. Symbols above the line of equality indicate that the duration of complete motor block lasts longer with clonidine compared with placebo; when the 95% confidence interval does not cross the line of equality, the difference is statistically significant. Overlapping symbols are separated for graphical purposes.
Abbreviations: CI, confidence interval; hetero, heterogeneity; n/a, not applicable. *RR, relative risk comparing clonidine with control. †NNT/H, number needed to treat/harm (a negative number needed to treat is a number needed to harm); 95% CI is shown only for statistically significant differences.
⫺8 (⫺16 to ⫺6) ⫺20 (n/a) 1.81 (1.44-2.28) 1.61 (0.87-2.99) .19 .93 15-150 15-150 152/486 (31.3) 27/330 (8.2) 72/372 (19.4) 12/231 (5.2) 5 8 Supplementary 5 Supplementary 6 Intraoperative hypotension Intraoperative bradycardia
12 5
NNT/H (95% CI)† Clonidine Control Multiple Dose
Number of Patients with Outcome/Total Number of Patients (%) Number of Trials
Single Dose Figure
Seventeen studies9-11,13-18,20,22,24-27,29,30 (23 comparisons) reported on arterial hypotension. Hypotension was defined as a decrease in systolic pressure ⬎20% (2 studies) or ⬎30% (10 studies) compared with baseline, a decrease in mean arterial pressure ⬎20% (1 study), mean arterial pressure ⬍70 mm Hg (2 studies), or need for intravenous ephedrine (2 studies). At least 1 episode of arterial hypotension was reported in 19.4% (72/372) of controls and 31.3% (152/486) of patients receiving clonidine; RR 1.81; 95% CI 1.44-2.28 (NNH 8). Relative risks were homogeneous (Phetero ⫽ .19), suggesting lack of dose-responsiveness. Thirteen studies11-14,17-19,21,22,24-26,29 (19 comparisons) reported on bradycardia. Bradycardia was defined as ⬍50 (10 studies), ⬍45 (1 study) or ⬍40 (2 studies) heartbeats per minute. At least 1 episode of bradycardia was reported in 5.2% (12/231) of controls and 8.2% (27/330) of patients receiving clonidine. Although the pooled RR suggested an increased risk of bradycardia with clonidine (RR 1.61), the difference did not reach statistical significance. Relative risks were homogeneous (P ⫽ .93), and crude and adjusted regression coefficients were nonsignificant (Tables 2 and 3).
Table 3. Adverse Effects
Adverse Effects
Clonidine Dose Range, g
Fig 4. Time to first request of analgesia. Data are from 3 dose-finding (circles) and 6 single-dose studies (squares). The horizontal line through zero represents equality. Vertical lines are 95% confidence intervals (CIs). Symbols above the line of equality indicate that the time to first request of analgesia is prolonged with clonidine compared with placebo; when the 95% CI does not cross the line of equality, the difference is statistically significant. Overlapping symbols are separated for graphical purposes.
intraoperative pain or the need for additional analgesia during surgery was reported in 9.4% (15/160) of controls and in 2.3% (4/173) of patients receiving clonidine; RR 0.24; 95% CI 0.09-0.64 (NNT 13). Relative risks were homogeneous (P ⫽ .51), suggesting lack of dose-responsiveness.
RR (95% CI)* Phetero
Treatment Effect
Regional Anesthesia and Pain Medicine Vol. 33 No. 2 March–April 2008
Outcome
164
Intrathecal Clonidine for Surgery
Further Outcomes Pain intensity after 24 hours, 24-hour morphine consumption, the need for general anesthesia due to unsuccessful intrathecal analgesia, incidence of tourniquet pain, surgeons’ evaluation of surgical conditions, sedation, time to walking or bladder voiding, postoperative hypotension, orthostatic hypotension at mobilization, or postoperative bradycardia were inconsistently reported; further analyses were deemed inappropriate. All results of analyses using weight-adjusted doses of clonidine were very similar.
Discussion Four main results emerge from these analyses. First, intrathecal clonidine prolongs the regression of the sensory block in a linear dose-dependent way. Second, clonidine prolongs the time to the first request of an analgesic, and the duration of complete motor block, with weak evidence of dose-responsiveness. Third, clonidine decreases the risk of intraoperative pain and increases the risk of arterial hypotension, without evidence of dose-responsiveness. Finally, clonidine has no relevant impact on the time to achieve complete sensory or motor block, on the extent of the cephalad spread of the sensory block, or on the risk of bradycardia. The original trials tested many different doses of clonidine and local anesthetic regimens. Unsurprisingly, treatment effects were heterogeneous. Understanding the causes of heterogeneity increases the scientific value and clinical relevance of a meta-analysis.8 We focused on the role of the dose of clonidine as a possible explanation for this heterogeneity. Testing for dose-responsiveness, by using data from independent trials remains a methodological challenge. Some may prefer to consider exclusively data from dosefinding studies.31,32 The methodological approach then consists in extracting the slopes of the linear associations that are reported in the original trials, and to compute a weighted-pooled estimate of the slopes across all trials. Our model is different because it combines data from both single and multiple dose studies. Because dose-finding studies compare 2 or more active groups with 1 single inactive control group, independence of observations can no longer be assumed and the regression model needs to be adapted to the use of such correlated data. Our model allowed controlling for differences in local anesthetic regimens and in outcome definitions. We found a significant linear relationship between clonidine doses and duration of the regres-
•
Elia et al.
165
sion of the sensory block. For the increase in the time to the first request of analgesia and of duration of motor block, the evidence of such a relationship was weak. Concerning the increase in the time to achieve complete motor block, the data were more homogenous; we may then be confident that the dose of clonidine had a marginal impact only on this outcome. To appraise the clinical relevance of our findings, we need to clarify what the ideal adjuvant to intrathecal local anesthetics for surgery would be. Intrathecal local anesthetics alone provide a rapid onset of sensory and motor block. Further shortening of these onsets with clonidine is probably of minor clinical relevance. Increasing the duration of surgical anesthesia allows for unexpected prolongation of surgical procedures, without the need for general anesthesia. Clonidine increases that duration by about 1 hour; this may be considered clinically relevant, although it may also be responsible for a delayed discharge from the recovery room, if discharge criteria include the level of sensory block. Patients who received intrathecal clonidine were less likely to experience intraoperative pain. This is clearly a beneficial effect. However, 9% of controls had at least 1 period of intraoperative pain. This high proportion was mainly due to data from 2 trials that were designed to study the effect of clonidine added to a suboptimal local anesthetic dose.20,26 Finally, prolongation of the postoperative pain-free period is clinically relevant, and there was evidence that this period could be extended by about 100 minutes when clonidine was added to a local anesthetic. In the absence of convincing data on doseresponsiveness, the choice of the dose may be driven by the fear of drug-related adverse effects. Not unexpectedly, the risk of intraoperative hypotension was increased with clonidine. We were, however, unable to demonstrate an increased risk of hypotension with increasing doses of clonidine. Almost 20% of controls experienced at least 1 episode of intraoperative hypotension, probably due to the intrathecal local anesthetics per se and the subsequent sympathetic block; the additional hypotensive effect of clonidine may have been partially masked. The evidence for an increased risk of bradycardia was poor. Both hypotension and bradycardia did not seem to pose clinical problems. Our analyses suffer mainly from 4 limitations. First, in the original trials, follow-up rarely exceeded 12 hours postoperatively; the long term analgesic profile of intrathecal clonidine remains unknown. Second, although a large variety of clonidine/ local anesthetic combinations were tested in these trials, there were no valid data on clonidine/ropiva-
166
Regional Anesthesia and Pain Medicine Vol. 33 No. 2 March–April 2008
caine or clonidine/levobupivacaine combinations. Third, it cannot be ruled out that we missed some reports, for instance, unpublished trials. And finally, most retrieved trials were of small size, and the quality of data reporting was limited. Both publication bias and an analysis that is mainly based on small and low quality trials are likely to lead to an overestimation of the effect of a treatment. In conclusion, adding clonidine to intrathecal local anesthetics for surgery increases the duration of the motor block, improves intraoperative analgesia, and delays the regression of the sensory block and the time to first analgesic request. For some of the reported effects, there is evidence of dose-responsiveness, for others, there is none. None of these outcomes is likely to last longer than 2 hours; whether they are considered clinically relevant and worthwhile depends on the context. The most prominent adverse effect is intraoperative hypotension. This study may serve as a rational basis to help clinicians decide whether or not to combine clonidine with an intrathecal local anesthetic for surgery. Whether adding clonidine to a low dose of an intrathecal local anesthetic offers advantages over a higher dose of the same local anesthetic alone remains unsolved. Finally, the optimal dose of clonidine when used as an adjuvant to an intrathecal local anesthetic remains unknown. Future trials should address these issues.
Acknowledgments Special thanks go to D. Haake, library of the Medical Faculty of Geneva University, for his help in searching electronic databases, and to the authors who responded to our inquiries. Individual trial data and additional figures and tables are freely accessible on the authors’ web site (http://www. hcuge.ch/anesthesie/data.htm).
Appendix Supplementary Material Supplementary Table 1, Figures 1-6, and Appendix can be found in the online version of this article at www.rapm.org (doi:10.1016/j.rapm.2008.10.008).
References 1. Eisenach JC, De Kock M, Klimscha W. Alpha(2)adrenergic agonists for regional anesthesia. A clinical review of clonidine (1984-1995). Anesthesiology 1996; 85:655-674. 2. Gabriel JS, Gordin V. Alpha 2 agonists in regional anesthesia and analgesia. Curr Opin Anaesthesiol 2001; 14:751-753.
3. Förster JG, Rosenberg PH. Clinically useful adjuvants in regional anaesthesia. Curr Opin Anaesthesiol 2003; 16:477-486. 4. Eisenach JC, Hood DD, Curry R. Relative potency of epidural to intrathecal clonidine differs between acute thermal pain and capsaicin-induced allodynia. Pain 2000;84:57-64. 5. Filos KS, Goudas LC, Patroni O, Polyzou V. Hemodynamic and analgesic profile after intrathecal clonidine in humans. A dose-response study. Anesthesiology 1994;81:591-601, 27A-28A. 6. Elia N, Tramer MR. Ketamine and postoperative pain–a quantitative systematic review of randomised trials. Pain 2005;113:61-70. 7. Tramèr MR, Walder B. Number needed to treat (or harm). World J Surg 2005;29:576-581. 8. Thompson S, Sharp S. Explaining heterogeneity in meta-analysis: a comparison of methods. Stat Med 1999;18:2693-2708. 9. Racle JP, Benkhadra A, Poy JY, Gleizal B. Prolongation of isobaric bupivacaine spinal anesthesia with epinephrine and clonidine for hip surgery in the elderly. Anesth Analg 1987;66:442-446. 10. Racle JP, Poy JY, Benkhadra A, Jourdren L, Fockenier F. Prolongation of spinal anesthesia with hyperbaric bupivacaine by adrenaline and clonidine in the elderly [in French]. Ann Fr Anesth Reanim 1988; 7:139-144. 11. Bonnet F, Diallo A, Saada M, Belon M, Guilbaud M, Boico O. Prevention of tourniquet pain by spinal isobaric bupivacaine with clonidine. Br J Anaesth 1989;63:93-96. 12. Bonnet F, Brun-Buisson V, Saada M, Boico O, Rostaing S, Touboul C. Dose-related prolongation of hyperbaric tetracaine spinal anesthesia by clonidine in humans. Anesth Analg 1989;68:619-622. 13. Wu CI, Wang JJ, Ning FS, Chao CH, Liu RJ, Chen CH, Mok MS. Effect of increasing amounts of clonidine during hyperbaric tetracaine spinal anesthesia [in Chinese]. Ma Zui Xue Za Zhi 1989;27:117123. 14. Seah YS, Chen C, Chung KD, Wong CH, Tan PP. Prolongation of hyperbaric bupivacaine spinal anesthesia with clonidine [in Chinese]. Ma Zui Xue Za Zhi 1991;29:533-537. 15. Fogarty DJ, Carabine UA, Milligan KR. Comparison of the analgesic effects of intrathecal clonidine and intrathecal morphine after spinal anaesthesia in patients undergoing total hip replacement. Br J Anaesth 1993;71:661-664. 16. Fukuda T, Dohi S, Naito H. Comparisons of tetracaine spinal anesthesia with clonidine or phenylephrine in normotensive and hypertensive humans. Anesth Analg 1994;78:106-111. 17. Niemi L. Effects of intrathecal clonidine on duration of bupivacaine spinal anaesthesia, haemodynamics, and postoperative analgesia in patients undergoing knee arthroscopy. Acta Anaesthesiol Scand 1994;38: 724-728. 18. De Negri P, Borrelli F, Salvatore R, Visconti C, De Vivo P, Mastronardi P. Spinal anesthesia with clonidine and
Intrathecal Clonidine for Surgery
19.
20.
21.
22.
23.
24.
25.
bupivacaine in young humans: interactions and effects on the cardiovascular system. Minerva Anestesiol 1997; 63:119-125. Larsen B, Dorscheid E, Macher-Hanselmann F, Büch U. Does intrathecal clonidine prolong the effect of spinal anesthesia with hyperbaric mepivacaine? A randomized double-blind study [in German]. Anaesthesist 1998;47:741-746. Pan PM, Huang CT, Wei TT, Mok MS. Enhancement of analgesic effect of intrathecal neostigmine and clonidine on bupivacaine spinal anesthesia. Reg Anesth Pain Med 1998;23:49-56. Juliao MC, Lauretti GR. Low-dose intrathecal clonidine combined with sufentanil as analgesic drugs in abdominal gynecological surgery. J Clin Anesth 2000;12:357362. Dobrydnjov I, Axelsson K, Samarutel J, Holmstrom B. Postoperative pain relief following intrathecal bupivacaine combined with intrathecal or oral clonidine. Acta Anaesthesiol Scand 2002;46:806-814. Kirdemir P, Gögüs¸ N. Intratekal Klonidin Unilateral Spinal Blogu Ve Analjeziyi Uzatiyor Mu [in Turkish]? Gülhane Tip Dergisi 2002;44:436-441. Santiveri X, Arxer A, Plaja I, Metje MT, Martinez B, Villalonga A, Lopez M. Anaesthetic and postoperative analgesic effects of spinal clonidine as an additive to prilocaine in the transurethral resection of urinary bladder tumours. Eur J Anaesthesiol 2002;19:589-593. Braz J, Koguti E, Braz L, Croitor L, Navarro L. Effects of clonidine associated to hyperbaric bupivacaine during high level spinal anesthesia [in Portugese]. Rev Bras Anesthesiol 2003;53:561-572.
•
Elia et al.
167
26. Dobrydnjov I, Axelsson K, Thorn SE, Matthiesen P, Klockhoff H, Holmstrom B, Gupta A. Clonidine combined with small-dose bupivacaine during spinal anesthesia for inguinal herniorrhaphy: a randomized double-blinded study. Anesth Analg 2003;96:14961503, table of contents. 27. Strebel S, Gurzeler JA, Schneider MC, Aeschbach A, Kindler CH. Small-dose intrathecal clonidine and isobaric bupivacaine for orthopedic surgery: a doseresponse study. Anesth Analg 2004;99:1231-1238, table of contents. 28. Jeon YT, Jeon YS, Kim YC, Bahk JH, Do SH, Lim YJ. Intrathecal clonidine does not reduce post-spinal shivering. Acta Anaesthesiol Scand 2005;49:1509-1513. 29. Kanazi GE, Aouad MT, Jabbour-Khoury SI, Al Jazzar MD, Alameddine MM, Al-Yaman R, Bulbul M, Baraka AS. Effect of low-dose dexmedetomidine or clonidine on the characteristics of bupivacaine spinal block. Acta Anaesthesiol Scand 2006;50:222-227. 30. van Tuijl I, van Klei WA, van der Werff DB, Kalkman CJ. The effect of addition of intrathecal clonidine to hyperbaric bupivacaine on postoperative pain and morphine requirements after Caesarean section: a randomized controlled trial. Br J Anaesth 2006;97:365-370. 31. Holt S, Suder A, Weatherall M, Cheng S, Shirtcliffe P, Beasley R. Dose-response relation of inhaled fluticasone propionate in adolescents and adults with asthma: meta-analysis. BMJ 2001;323:253-256. 32. Lee A, Ngan Kee WD, Gin T. A dose-response metaanalysis of prophylactic intravenous ephedrine for the prevention of hypotension during spinal anesthesia for elective cesarean delivery. Anesth Analg 2004; 98:483-90, table of contents.