| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
*Department of Anaesthesiology and Intensive Care and Unit of Biostatistics and Epidemiology, Örebro Medical Center Hospital, Örebro, Sweden; and Department of Statistics, Örebro University, Örebro, Sweden
Address correspondence and reprint requests to Anil Gupta, Department of Anaesthesiology and Intensive Care, Örebro Medical Center Hospital, S-701 85 Örebro, Sweden.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
2 test. Forty-five articles could be identified in which the effects of morphine were studied in a prospective, randomized manner, and 32 of these studies included a placebo control. Pooled analyses of data from 19 studies suitable for metaanalysis showed an improvement in analgesia after morphine compared with placebo in the order of 12–17 mm on the VAS during all three phases of treatment. Studies with high quality scores showed somewhat smaller improvements. Total analgesic consumption could not be analyzed statistically, but the number of studies showing decreased analgesic consumption or no differences between groups was identical (six and six). No clear dose-response effect was seen when VAS was used as a measure of pain, but it was seen when area under the curve was used as a measure of pain. A systemic effect of peripherally-injected morphine was not possible to exclude because of the very limited data available. We conclude from this metaanalysis that intraarticularly administered morphine has a definite but mild analgesic effect. It may be dose dependent, and a systemic effect cannot be completely excluded. IMPLICATIONS: A metaanalysis was performed to study the effects of morphine, injected intraarticularly, on postoperative pain during 24 h. Nineteen studies could be identified in which data were presented in such a way that metaanalysis could be performed. The results indicate a mild analgesic effect that could be dose dependent, but a systemic effect could not be ruled out.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Kalso et al. (3) published a qualitative systematic review of the literature on the intraarticular effects of morphine in 1997. It is surprising that they found only four studies that scored more than four points on a five-point qualitative scale described by McQuay and Moore (4). Accordingly, no metaanalysis was performed because of a lack of an adequate number of high-quality studies. They concluded that morphine probably has mild analgesic effects when injected intraarticularly in humans but also recommended further randomized controlled trials (RCTs) to clarify the issue. The primary aim of this systematic review was to establish whether morphine injected intraarticularly has an analgesic effect when compared with placebo. The secondary aims were to assess whether this is a dose-dependent effect, and if so, whether it is a systemic effect or occurs via peripheral receptors. This systematic review was consequently performed in four steps: 1) We reviewed all studies published in the literature in which a direct comparison was made between morphine and placebo injected intraarticularly in humans; 2) we performed a qualitative analysis of these studies on the basis of the criteria recommended by McQuay and Moore (4); 3) we then performed a quantitative analysis of the data from all randomized, placebo-controlled trials and also a quantitative analysis of studies with a high quality; and 4) as a last step in the analysis, those studies in which different doses of morphine were given were also analyzed to see whether there was a dose-dependent effect of morphine. If a dose-dependent effect was seen, we were interested in determining whether this was a systemic effect or an effect via peripheral morphine receptors.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
In the assessment of postoperative pain, three phases were identified.
Two measures of pain relief were analyzed:
In view of the large number of studies published and the uncertainty of the conclusions drawn, our review was limited to answering three main questions: 1) Does morphine have any analgesic effect when injected intraarticularly compared with a control group receiving only placebo? 2) If there is an analgesic effect, is it dose dependent? 3) Is the possible analgesic effect systemic or local (peripheral)? Thus, studies in which morphine was not compared directly with placebo were excluded from the review.
Means and 95% confidence intervals (CI) for the difference (treatment - control) in pain score on the VAS were calculated from the studies identified. The formulas for two independent samples with unequal variances were used according to Welch (15). The CIs were plotted with VAS in millimeters on the horizontal axis, and different individual studies were plotted on the vertical axis. Means located to the left of the reference line 0 indicate that treatment is superior to placebo, whereas mean values to the right indicate the opposite.
Metaanalysis was performed with the weighted-analysis technique based on that of Grizzle et al. (16) and as described by Greenland (17). The study weights used in combining the results from the different studies were the inverse variance computed from the estimated standard error, 1/SE (2), where the formula for SE was given by Welch (15). We tested the essential homogeneity assumption by the 2 test described by Greenland (17), and when homogeneity was rejected, we calculated the final CI for the summary effect with a correction based on random effect modification, also described by Greenland (17). All computations could be implemented by standard statistical software, SPSS (18), with additional commands using the internal syntax of the package. The final 95% CIs were visualized graphically in the same way as the CIs of the basic studies. The analysis was performed for all 19 studies suitable for metaanalysis as well as for those studies that had a higher quality score.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Twenty-seven studies could be identified in which morphine was compared directly with placebo (normal saline) in patients undergoing arthroscopic knee surgery (Table 1). A total of 1748 patients were studied. Thirteen of these studies found a beneficial effect of morphine (19–31), whereas 14 others did not find any beneficial effect (11–13,32–42). In one study (42), no beneficial effect of morphine was found when all patients were included, but significantly better analgesia was provided when a subgroup of patients with VAS pain intensity >10 mm was analyzed. General anesthesia was used in 12 of the 13 studies that showed a beneficial effect of morphine, and in one study, local anesthesia was used for the operative procedure. Ten of the 13 studies used intraoperative tourniquet, and in 8 studies, intraoperative narcotics were given as a part of the anesthetic technique. In nine studies in which 4–10 mg morphine was injected intraarticularly, better analgesia was reported in the morphine groups compared with the placebo group. In six studies, a dose of 1 mg morphine was injected intraarticularly (in two studies, two different doses of morphine were given). Of these, two studies found no benefit of 1 mg morphine compared with placebo (34,38), and of the four studies in which VAS was less in the morphine group, one study found this to be less only between 0 and 2 h (Phase I) (20), whereas the other three found it to be less between 6 and 24 h (Phase III) (24,28,31).
|
|
|
|
|
3 and 4) to see whether the quality had an effect on the heterogeneity and on the difference between placebo and treatment groups. Homogeneity was present only when quality scores 4 were analyzed and only during the early (0–2 h) and intermediate (2–6 h) phase (Table 2). In this latter group with a high quality score, the treatment group continued to show better analgesia than the placebo group. However, the differences between the treatment and placebo groups were less than when all studies were included. Seven studies could be identified in which the authors studied a dose-response relationship for the effects of morphine injected intraarticularly (Table 3). A total of 613 patients were studied in these seven studies. In two studies, 5 mg morphine was found to be better than 1 mg during the first 24 h after surgery (24,29). In the other four studies, there were no advantages in giving 1 mg compared with 0.5 mg of morphine (2), 1 mg compared with 2 mg of morphine (37,42), 2 mg morphine compared with 5 mg (39), or 1 mg versus 3 mg of morphine (43). In one study, varying doses of morphine (1, 2, and 4 mg) were compared with placebo (44). The authors found a dose-response reduction in pain intensity as measured by the area under the curve, but not as measured by the VAS.
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
We used more liberal inclusion criteria in our metaanalysis; i.e., all RCTs in which the results could be statistically analyzed were included and not just those with high quality scores. To ensure that an RCT is of high quality, McQuay and Moore (4) suggest that the study should be 1) randomized, and that the method of randomization should be described and appropriate; 2) double-blinded, and the method of blinding should be described and appropriate; and 3) any withdrawals from the trials should be described. Many of the studies described in the literature may have followed these criteria but may not have stated it in the study report. Should these studies then be excluded from a metaanalysis? A similar argument was put forward by Fisher (45), who stated that well designed studies might be rejected simply because the authors failed to specify certain criteria demanded by the metaanalyst (and such criteria might have been standard practice at the time the study was conducted but were not routinely reported). By using more liberal inclusion criteria, we were able to draw conclusions from a larger number of studies (and patients). The conclusions should, however, carry a warning that we do not know whether the authors have strictly followed the methodology of high-quality trials, but we have no reason to believe otherwise.
Studies in which morphine was compared with a local anesthetic, a nonsteroidal antiinflammatory drug, clonidine, or another narcotic analgesic, but not a placebo group, were excluded. Although Kalso et al. (3) argue that the highest hierarchy of evidence is when local anesthetic is used intraarticularly because local anesthetics are known to provide reliable analgesia of predictable duration, a systematic review of intraarticular local anesthetics found only weak evidence for reduction of postoperative pain after arthroscopic knee surgery (46). Also, combining local anesthetics with morphine or a nonsteroidal antiinflammatory drug may result in interactions and possible synergistic effects, thus confounding the conclusions drawn. Thus, the primary aim of this systematic review was to quantify the effect of morphine versus placebo and not versus an active drug.
Results from many studies are presented only as graphs or diagrams, and sometimes it is difficult to identify the exact value of the studied variable. For instance, VAS was often presented as a diagram, with time on the x axis and pain intensity (cm) on the y axis. Because resolution of the y axis was sometimes poor (e.g., intervals of 2 cm on the scale, absence of marks representing digits, or three-dimensional figures) (38,24), it was sometimes impossible to identify the correct pain intensity score at a given time point. All studies in which two independent authors were in agreement on the possibility of extraction of data from figures are included in the metaanalysis.
Although McQuay and Moore (4) do not discuss statistical problems when evaluating high-quality articles, this is one of the most common problems in RCTs. One of the problems encountered was that one of the authors used two different statistical methods for the same data and showed that, depending on the statistical method used, the result could be significant (P < 0.05) or not significant (P > 0.05) (22). Some authors have presented CIs without mean values (23), others have presented mean values without SD, SEM, or CI (26,38), and still others have presented only changes from baseline values rather than actual values (13). Most of these studies could not, unfortunately, be used in any constructive metaanalysis, thus limiting the number of studies suitable for analysis.
The above-mentioned problems deal with the selection of studies suitable for the metaanalysis. As for the analysis itself, we found a significant heterogeneity between the studies when all studies were included in the analysis. This means that the studies do not estimate a common, constant difference between placebo and treatment. Instead, there seems to be a variation in the difference that is more than that of simple random variation. The variation can be caused by characteristics that differ between the studies and at the same time affect the treatment. The interpretation of the summarized mean values is in this case a bit more complex. They now represent means in the distribution of the differences rather than a simple variable. By excluding some of the studies from the metaanalysis, in particular those with effect-related study characteristics, homogeneity could possibly have been achieved. This was found to be true when only high-quality studies were included in the metaanalysis (score 4). The relatively small number of available studies speaks against this approach, and because in our case the calculated mean values were found in the region in which treatment was superior to placebo and the CIs were narrow, we believe that the heterogeneity does not in any essential way contradict the main conclusions of the metaanalysis. In addition, as stated by Greenland (17), the analysis of heterogeneity can be the most important result of a metaanalysis.
The major issue in this systematic review was to establish whether morphine injected intraarticularly has an analgesic effect when compared with placebo. Thus, we included only studies that were prospective, randomized, and double-blinded. The subsequent metaanalysis showed a reduction in pain intensity by a mean value of 12–17 mm on VAS during the three postoperative phases. This confirms the early findings of Khoury et al. (7) that morphine injected intraarticularly produces analgesia. Taking into account that morphine receptors have been described on peripheral nerves in animals (47) and humans (48), one may conclude that morphine also acts via peripheral receptors to induce analgesia in humans. However, these results lead to further questions that must be answered. Why do some studies show a good effect of morphine injected intraarticularly and others do not? Is the effect dose dependent? Is this a systemic effect or an effect on peripheral receptors?
According to our systematic review, there seems to be wide variability in the analgesic effects when morphine is injected intraarticularly. This variability is seen not only between studies (population variability), but also within studies (patient variability). In addition, lack of consistency is seen between studies, i.e., some studies report early beneficial effects of morphine, whereas others show only late effects. Factors that might affect the results could include the type of surgery and pain intensity, preexisting inflammatory reaction, and when the tourniquet was released, as well as study design (underpowered studies may result in negative results). There is little doubt that the presence of preexisting inflammatory reaction increases the efficacy of intraarticular morphine (48). The role of the tourniquet is more controversial. Two studies that specifically addressed this issue reported contradictory results (5,6), and the mechanism for the potential positive effect of tourniquet remains unclear. The hypothesis that the presence of tourniquet would improve the binding of morphine to the intraarticular receptors is not proven (5).
Many studies with negative results seem to be too small (underpowered) to detect significant differences. Because pain intensity in many studies is modest, large sample sizes are required to reliably detect significant differences. Thus, studies in which pain intensity is more pronounced (anterior cruciate ligament repair) and in which there is preexisting inflammation (e.g., chronic arthritis) may more likely detect significant effects with smaller sample sizes. This was, in fact, the observation made by Rosseland et al. (42) when they excluded patients with minimal pain. They found that there was a significantly better effect of morphine compared with placebo.
Another issue to be addressed is whether the potential effect of intraarticular morphine is dose dependent or not. Once again, the results are equivocal, with some studies suggesting a dose-dependent effect while others do not. A well-designed study of high quality showed a dose-dependent analgesic effect after intraarticular morphine (44). Three doses of morphine (1, 2, and 4 mg) injected intraarticularly after arthroscopy performed under general anesthesia were compared with placebo. Although no statistical differences were shown between the groups in the VAS pain scores, the area under the VAS pain curve decreased with increasing doses of morphine. Similarly, although no differences were seen in analgesic consumption at specific time points, the overall cumulative analgesic consumption decreased with increasing doses of morphine. Thus, this study showed that the analgesic effect of morphine is dose dependent. It is, however, possible that the peak effect of morphine was not seen after 4 mg and that a further increase in the dose would provide better analgesia. Kalso et al. (3) argued that doses of 1 mg morphine, when diluted to 20 mL and injected into the knee joint, would produce concentrations (in the knee) of more than 1000 times that seen after the systemic injection of morphine. This suggests that the peripheral opioid receptors should be fully saturated with 1 mg morphine in 20 mL saline. Yet Likar et al. (44) found a dose-dependent analgesic effect of morphine injected intraarticularly. One explanation for these contradictory findings might be that the analgesic effect of morphine is due to its systemic absorption after intraarticular injection. In contrast to the findings of Likar et al. (44), a study compared the effects of 1 vs 2 mg morphine and found that patients receiving morphine 2 mg had more pain than those receiving 1 mg (43). Two other studies had similar results. Björnsson et al. (41) found higher pain scores than placebo during the first 2 h after the injection, although the difference did not reach statistical significance. Similarly, Gupta et al. (13) also found higher pain scores during the first 2 h in the morphine group after 3 mg morphine. Morphine causes histamine release, and this may be the mechanism for the short-lasting local hyperalgesia seen in some studies. In another study, Rosseland et al. (42) found no improved analgesia when comparing 1 vs 2 mg of morphine intraarticularly. The reasons for these conflicting results are unknown.
In three studies, no differences were found in VAS scores between the systemic and intraarticular groups, whereas in two studies intraarticular morphine was found to be better than IV. In the latter two, no differences in VAS scores were seen during the first two to four hours after the injection. However, better analgesia was reported during the next 24 hours in the group receiving intraarticular morphine. Thus, it seems that the analgesia is prolonged when morphine is injected intraarticularly, whereas systemic morphine has a short duration of effect. It has been proposed that glucoronidation of morphine intraarticularly may produce morphine-6-glucoronide, which has a longer half-life that may account for the more prolonged effect (28). In the same study, plasma morphine-6-glucoronide and morphine-3-glucoride concentrations were measured after the injection of 5 mg of intraarticular morphine, and it was found that the latter was quite large. The plasma concentration of morphine was measured after systemic and intraarticular administration in two studies (28,29). Whereas Joshi et al. (28) found that the plasma concentrations were small during the sampling period (up to four hours), Richardson et al. (29) found a mean plasma concentration of 3.5 ng/mL two hours after intraarticular administration of 5 mg morphine, and a systemic dose of 5 mg morphine produced a concentration of 6.3 ng/mL—approximately twice as much after IV compared with intraarticular morphine. This is somewhat surprising because one would not have expected such large plasma concentrations after the intraarticular injection of morphine; it makes one wonder whether this effect is, after all, a systemic effect. No studies have measured plasma concentrations after 6–24 hours of intraarticular administration of morphine when the analgesic effect of morphine (intraarticular) appears to be maximal. Could there be an axonal flow of morphine via peripheral nerves into the spinal fluid and the analgesic effects of morphine seen are via spinal receptors (and not peripheral morphine receptors), causing the delayed effect? Future studies should focus on the role of inflammation in producing peripheral analgesia and also on whether the analgesic effect of morphine is via opioid receptors or via inhibition of prostaglandin mechanisms in inflamed tissues.
Metaanalysis of the studies available in the literature has shown that morphine injected into the intraarticular space produces analgesia up to 24 hours after the injection, and this could be a dose-dependent effect. Whether the effect is via peripheral opioid receptors or a systemic effect remains to be shown conclusively.
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  K. A. Berg, A. M. Patwardhan, T. A. Sanchez, Y. M. Silva, K. M. Hargreaves, and W. P. Clarke Rapid Modulation of {micro}-Opioid Receptor Signaling in Primary Sensory Neurons J. Pharmacol. Exp. Ther., June 1, 2007; 321(3): 839 - 847. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. L. Rittner, S. A. Mousa, D. Labuz, K. Beschmann, M. Schafer, C. Stein, and A. Brack Selective local PMN recruitment by CXCL1 or CXCL2/3 injection does not cause inflammatory pain J. Leukoc. Biol., May 1, 2006; 79(5): 1022 - 1032. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. M. Hargreaves Peripheral Opioid Regulation of Nociceptors. Focus on "Morphine Directly Inhibits Nociceptors in Inflamed Skin" J Neurophysiol, April 1, 2006; 95(4): 2031 - 2031. [Full Text] [PDF]
|
|
|
|
 |
|
 S. A. Strassels, E. Mcnicol, and R. Suleman Postoperative pain management: A practical review, part 2 Am. J. Health Syst. Pharm., October 1, 2005; 62(19): 2019 - 2025. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Vintar, N. Rawal, and M. Veselko Intraarticular Patient-Controlled Regional Anesthesia After Arthroscopically Assisted Anterior Cruciate Ligament Reconstruction: Ropivacaine/Morphine/Ketorolac Versus Ropivacaine/Morphine Anesth. Analg., August 1, 2005; 101(2): 573 - 578. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Likar, S. A. Mousa, G. Philippitsch, H. Steinkellner, W. Koppert, C. Stein, and M. Schafer Increased numbers of opioid expressing inflammatory cells do not affect intra-articular morphine analgesia Br. J. Anaesth., September 1, 2004; 93(3): 375 - 380. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G Watterson, R Howard, and A Goldman Peripheral opioids in inflammatory pain Arch. Dis. Child., July 1, 2004; 89(7): 679 - 681. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Keita, J. L. Benifla, V. Le Bouar, R. Porcher, B. Wachowska, K. Bedairia, J. Mantz, and J. M. Desmonts Prophylactic ip injection of bupivacaine and/or morphine does not improve postoperative analgesia after laparoscopic gynecologic surgery: [L'injection intraperitoneale preventive de bupivacaine et/ou de morphine n'ameliore pas l'analgesie postoperatoire apres une intervention gynecologique laparoscopique] Can J Anesth, April 1, 2003; 50(4): 362 - 367. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. B. Akinci, F. Saricaoglu, A. Atay, M. N. Doral, and M. Kanbak Analgesic effect of intra-articular tramadol compared to morphine after arthroscopic knee surgery Can J Anesth, April 1, 2003; 50(4): 423 - 424. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
