THE concept of preemptive analgesia to reduce the magnitude and duration of postoperative pain was paved in 1983 by Woolf, 1who showed evidence for a central component of postinjury pain hypersensitivity in experimental studies. Subsequently, an overwhelming amount of experimental data demonstrated that various antinociceptive techniques applied before injury were more effective in reducing the postinjury central sensitization phenomena as compared with administration after injury. 2Finally, these promising experimental findings were taken into clinical testing of the hypothesis. Although early reviews of clinical findings were mostly negative, 3–5there is still a widespread belief of the efficacy of preemptive analgesia among clinicians.
The definition of preemptive analgesia has varied, thereby causing confusion and misunderstanding of the concept. 6Because the original observations in experimental studies suggested that timing of analgesic treatment was important to obtain efficient reduction of postinjury pain hypersensitivity phenomena, we performed an updated review of studies to compare the role of timing of analgesia i.e. , preoperative versus intraoperative or postoperative initiation of analgesia. In this review we are not considering studies designed to compare preemptive analgesia versus no treatment. We have only included double-blind, randomized, controlled trials of identical or very similar analgesic regimens, where the only difference between study groups was timing of analgesia.
Reports of randomized controlled trials of preemptive analgesia for acute or chronic postoperative pain relief were systematically sought using the Cochrane Library 2000 §(www.cochrane.org) and the MEDLINE ∥(www.ncbi.nlm.nih.gov/PubMed/; 1966–2000) databases without language restriction. We used different search strategies with free text combinations, including the following search terms: preemptive analgesia, preemptive analgesia, prophylactic pain treatment, preoperative treatment, postoperative pain, postoperative analgesia, chronic pain, and long-term pain. The last search was performed on December 30, 2000. Reference lists of retrieved reports and review articles were hand-searched for additional papers. No abstracts, correspondences, or unpublished observations were included. Authors were not contacted for original data.
Inclusion and Exclusion Criteria and Data Extraction
Reports that were included consisted of double-blind randomized comparisons of identical or nearly identical analgesic regimens initiated before versus after surgical incision for postoperative pain relief with or without the use of a double dummy.
Reports that were excluded included trials of comparisons of preoperative treatment with placebo or no treatment, and trials of comparisons of preoperative with preoperative plus postoperative treatment. Such studies provide no evidence for a preemptive effect, i.e. , if timing of the initiation of the pain treatment is of importance. 5
We developed standard data collection sheets to record details of trial design, interventions, and outcome measures for every trial. Each report meeting the inclusion criteria was read independently by two of the authors and scored using a three-item, 1–5 quality scale. 7Consensus was subsequently achieved. If the reports were described as randomized, one point was given, and an additional point was given if the method of randomization was described and adequate (computer-generated, table of random numbers, etc. ), but one point was deducted if randomization was inappropriate (alternate randomization, randomization according to weekday, etc. ). If studies were described as double-blind, one point was given, and an additional point was given if blinding was described and appropriate (use of double-dummy, blinded pharmacy manufactured ampoules, etc. ), but one point was deducted if blinding was inappropriate. Finally, reports that described the numbers and reasons for withdrawals were given one point. By definition, studies without randomization and blinding were excluded. Thus, the minimum score of an included clinical trial was 2, and the maximum score was 5.
Each trial was assessed for different measures of internal sensitivity. First, trials were checked for magnitude of pain intensity. Because it is difficult to detect an improvement with low or no pain, it was noted if pain scores were less than 30 mm on a visual analog scale (VAS) or less than moderate pain on a verbal rating scale or similar score. 8Second, it was noted if a power calculation of the statistical tests was performed. Trials with sample sizes less than 10 patients per treatment group were not considered. 9
Use of other intraoperative analgesic treatment (which in theory may preempt pain in the postsurgery treatment group) was noted but not regarded to invalidate the clinical relevance of trials, since common anesthetic practice often includes analgesic use (e.g. , intraoperative fentanyl).
Data on postoperative pain and analgesic consumption were extracted for each report. Finally, information about type of anesthesia (general, regional) and number of patients enrolled was taken from each report.
Qualitative analysis of postoperative effectiveness was evaluated by significant difference (P < 0.05 as reported in the original investigation) in pain relief using pain scores, time to first analgesic request, and consumption of supplementary analgesics between the presurgical and postsurgical treatment groups, and by assessment of the clinical importance of observed differences. The plot of L'Abbéet al. 9of VAS pain scores with preemptive versus postsurgical regimens was used as a graphic means of exploring the consistency of efficacy and the homogeneity of the data whenever possible.
Quantitative analysis of combined data were intended by calculation of the weighted mean difference (WMD) of VAS pain scores between treatment groups (using the Review Manager software, version 4.0, the Cochrane Collaboration; The Nordic Cochrane Center, Copenhagen, Denmark). The weight given to each study in this analysis (i.e. , how much influence each study had on the overall results) was determined by the precision of its estimate by taking into account study size and SDs of the VAS scores in the individual trials. For the current use, a mean VAS for each treatment group was calculated in every trial from all available recordings performed within 24 h after surgery. Verbal rating pain scores and similar scores were converted to VAS pain scores (e.g. , a four-point verbal rating score including no, light, moderate, and severe pain was converted to 0, 25, 50, and 75 mm VAS, respectively). The possibility was recognized that data only would allow a qualitative analysis. Finally, the trials were stratified according to the type of drug (opioid, local anesthetic, N -methyl-d-aspartate [NMDA] receptor antagonist, nonsteroidal antiinflammatory drug [NSAID]), mode of administration (systemic, neuraxial, peripheral nerve block, or wound infiltration), and, if possible, to surgical procedure.
Ninety-three randomized clinical trials of preincisional versus postincisional analgesic regimens for postoperative pain control were identified. Of these, 11 studies were excluded because of lack of appropriate blinding or randomization 10–18or use of different analgesic doses preoperatively and postoperatively. 19,20Two articles were not available through the Danish University Library (Copenhagen, Denmark) or the British Library 21,22(London, United Kingdom), leaving 80 reports for analysis. Studies excluded are summarized in the Appendix.
The remaining studies could be divided into 20 trials of systemic NSAIDs, 8 trials of systemic opioids, 8 trials of systemic NMDA receptor antagonists, 24 trials of epidural, caudal, or intrathecal analgesia, and 20 trials of peripheral local anesthetic use (wound infiltration or nerve block) or combinations of treatment.
A total of 3,761 patients, of which 1,964received preincisional treatment, were studied. The range ofthe number of patients included in the studies was 10(in a crossover trial) to 128. The median quality scorewas 4 (range, 2–5) in trials with significantdifferences in pain relief between the treatment groups and 4(range, 2–5) in trials with no significant differences.The percentage of trials with a significant finding in favor of preemptive analgesia did not differ between trials of high quality (score, 4–5) and trials of lower quality (score, 2–3) (P = 0.67, Fisher test). Details of included studies are shown in tables 1–7and figures 1–4.
Quantitative analysis was performed on the mean of VAS pain scores recorded within 24 h after surgery for each treatment modality. In five trials, verbal rating scores were converted to VAS scores (two trials of NSAID and three trials of local infiltration). Data on analgesic consumption and time to first analgesic request only allowed a qualitative analysis because of the variety of analgesics, doses, and outcome reporting used. Instead, any statistical difference between treatments regarding these measures was extracted from the original reports and documented in table format as performed previously for other qualitative systematic reviews. 23–25
Acute Postoperative Pain
Nonsteroidal Antiinflammatory Drugs.
Twenty trials comparing preincisional with postincisional NSAID or paracetamol 26using a parallel or crossover design 26–29were identified. Various odontologic, abdominal, and orthopedic procedures were studied. The NSAIDs were diclofenac, 27,30–33naproxen, 28,34flurbiprofen, 35ketorolac, 36–42ketoprofen, 43,44diflunisal, 29and ibuprofen 45used in clinically relevant doses (table 1). Fentanyl, 30,31,33,36,38,39,42–44alfentanil, 34,37local anesthetics, 26–29,33,35,45or nitrous oxide 30,31,36–44were, as a part of the anesthesia, coadministered intraoperatively in all trials.
In two trials, pain scores were significantly improved immediately after surgery by preemptive compared with postoperative treatment. 36,41In none of the other trials were improvements observed (fig. 1A). Quantitative analysis with the calculation of the WMD of VAS scores between treatment groups using a fixed-effect model (as test for heterogeneity was nonsignificant, P = 0.78) was not significant (WMD, 0 mm; 95% confidence interval [CI], −2 to 2 mm;fig. 2A) with 14 trials. In the remaining six trials, one of which showed reduced pain scores, 41there was a lack of dispersion measures for the calculation. 27,31,35,37,41–42
In one trial, 31the number of patients needing rescue analgesics and time to first request was improved by 28% and 1.5 h, respectively. In two other studies, patient-controlled analgesia–morphine and time to first analgesic request were statistically improved by 6 mg over 6 h 36and 49 min, 44respectively. In none of the other trials was demand for supplementary analgesic different between treatment groups.
Power analysis of the statistical tests was only available in five trials, 30,34,36,40,44with a power of 75–95% of detecting a difference of 15–25 mm VAS at the 5% significance level. Furthermore, intensity of pain scores was low in eight trials (< 30 mm VAS), 27–29,33,35–36,40,45which might have impaired internal sensitivity.
In conclusion, some aspects of postoperative pain control were improved by preemptive treatment in 4 of the 20 trials. Overall, the data demonstrated preemptive NSAIDs to be of no analgesic benefit when compared with postincisional administration of these drugs.
Eight trials with nine treatment arms were identified comparing preincisional with postincisional administration of morphine (10 mg or 0.15–0.3 mg/kg), 46–48fentanyl (10 μg/kg), 49alfentanil (40–70 μg/kg), 50,51sufentanil (1 μg/kg), 49,52or pentazocine (30–60 mg) 53(table 2). In all trials, the surgical procedure was abdominal hysterectomy. In none of the trials was other intraoperative analgesics (beside the test drugs) administered except for nitrous oxide in all studies.
In no study were pain scores significantly reduced in the preemptive group (fig. 1B). In contrast, quantitative analysis of pain scores using a fixed-effect model (P = 0.75 in test for heterogeneity) revealed that the WMD in VAS scores between study groups was statistically significant in favor of the postoperative groups (5 mm; 95% CI, 1–9 mm;fig. 2B).
Supplementary analgesic consumption was significantly reduced in two studies in the preemptive group, averaging 10 mg morphine over 24 h 46and 12 mg morphine 50from 48 to 72 h, but not from 0 to 6, 6 to 12, 12 to 24, or 24 to 48 h postoperatively, rendering interpretation difficult. Time to first analgesic request was evaluated in only one trial 50and was not different between study groups.
Intensity of pain scores was considered adequate (> 30 mm VAS) in all trials. However, in only three trials was power analysis of the statistical tests performed, 48–50revealing an at least 80% power to detect a reduction in VAS scores of 20 mm 49,50or decrease in opioid consumption of 30%48at the 5% significance level.
In conclusion, no improvement in postoperative pain control was observed after preemptive administration of systemic opioids.
Intravenous or Intramuscular N -methyl-d-aspartate Receptor Antagonists.
Eight trials were identified comparing preincisional with postincisional ketamine 54–59or dextromethorphan 60,61in a variety of surgical procedures (table 3). Ketamine was administered in doses of 0.15–1 mg/kg and in two trials continued with intraoperative infusion of 10 μg · kg−1· min−1in the preemptive group, of which one was negative and one positive. 54,56Dextromethorphan was given in doses of 40 mg to 5 mg/kg (mean, 275 mg). Coadministered analgesic drugs included intraoperative fentanyl, alfentanil, or sufentanil in seven trials 55–61and nitrous oxide in five trials. 54–58
The worst pain score was significantly reduced by 20 mm VAS in one trial of dextromethorphan. 61In the seven other trials, no effect on pain scores was observed (fig. 1C). The WMD calculated by use of a random-effect model (P < 0.05 in test for heterogeneity) was not significant (WMD, −2 mm; 95% CI, −8 to 4 mm;fig. 2C).
Supplementary analgesic consumption was significantly reduced by preemptive analgesia in three trials (one ketamine study 54and the two dextromethorphan trials 60,61) by 40–70%, corresponding to 15–25 mg morphine 54,60and 57 mg pethidine 61over a 24–48-h observation period. In the five other trials (of ketamine), no effect 56–59or increased analgesic consumption 55was observed compared with the postincisional groups. Time to first analgesic request was evaluated in only one trial and was prolonged by 11 h by preemptive treatment. 61
Power analysis was performed in three trials 55,57,60and showed an 80% power to detect a difference of 30% or 5 mg/24 h of morphine at the 5% significance level. Intensity of pain scores were greater than 30 mm VAS in all except for one trial. 59
In conclusion, no improvement in postoperative pain control was observed from preemptive systemic ketamine. Both studies on dextromethorphan were positive, but the data are too sparse to reach a definitive conclusion.
Epidural, Caudal or Spinal Regimens.
Eighteen trials of presurgically versus postsurgically initiated epidural analgesic regimens were identified. These could be divided into trials of single-dose analgesic regimens 62–71and trials of continuous analgesic regimens extending 24–72 h into the postoperative period. 72–79Furthermore, five trials of caudal analgesia in children 80–84and one trial of intrathecal anesthesia–analgesia 85were found eligible for analysis.
Single-dose Epidural Analgesia.
Ten trials with 11 treatment arms were identified comparing different preemptive versus postincisional single-dose epidural analgesic regimens. In four trials, epidural fentanyl (4 μg/kg) 63and morphine (2–4 mg or 0.05 mg/kg) 65,68,71were evaluated. In three trials, 64,66–67epidural bupivacaine (0.5%, 15–20 ml) was studied, in three trials, combined epidural opioid (fentanyl or morphine) and local anesthetic (bupivacaine or mepivacaine), 62,69,71and in one trial, epidural morphine (2 mg) plus ketamine (60 mg). 70As a part of a balanced analgesic regimen, systemic NSAID was administered in two studies. 62,67Coadministered analgesics included intraoperative alfentanil, fentanyl, or morphine in four trials 62,65,58,71and nitrous oxide in eight trials. 62–66,69–71Surgical procedures were major thoracic and abdominal (table 4).
Epidural Opioid Regimens.
Pain scores were significantly reduced over 24 h by preemptive analgesia in one trial, 65but only at six 63and 18 h, 68respectively, and not at 2, 4, 8, 10, 24, or 48 h postoperatively in two 63,68of a total of four trials–treatment arms. 63,65,68,71Analgesic demand was significantly reduced between 12 and 50%65,68and by 14 mg of patient-controlled analgesia–morphine from 12–24 h 63in the preemptive groups in three trials.
Epidural Local Anesthetic Regimens.
Combined Epidural Regimens.
Pain scores were not different between study groups in any of four trials–treatment arms. 62,69–71Analgesic demand or number of patients requesting analgesics was significantly reduced between 33 and 48% in the preemptive groups in two trials 70,71but only by 16 mg over 96 h in another trial. 62
Quantitative analysis, which was only possible with seven trials (eight treatment arms) because of lack of dispersion measures, revealed a nonsignificant WMD of mean VAS pain scores recorded over 24 h of −4 mm (95% CI, −9 to 2 mm; random effect mode;P = 0.04 in test for heterogeneity;figs. 3A and 4A). In two of the three trials not included in the WMD calculation, 67–68,70no significant difference in VAS was observed at any time during the postoperative course supporting the quantitative estimate.
Power analysis of the statistical tests revealing a 90% power was available in two trials, 63,67although without information of the minimal relevant difference (e.g. , number of millimeters VAS), not to be overlooked. Furthermore, in one negative 62and one positive 68trial, low pain scores may have impaired internal sensitivity.
In conclusion, the quantitative analysis of mean VAS pain scores showed no significant reduction by preemptive single-dose epidural analgesia with opioid, local anesthetic, or a mixture. However, significant reductions in analgesic demand were demonstrated in 7 of 11 treatment arms.
Continuous Epidural Analgesia.
Eight trials were identified 72–79comparing different preemptive versus postincisional initiated continuous epidural regimens that extended 24–72 h into the postoperative period.
The regimens investigated included bolus epidural bupivacaine (0.5–0.75%, 8–18 ml) 73–75plus morphine (2 mg), 74,75epidural mepivacaine (1.5–2%, 4–15 ml) 76–79plus morphine (4 mg) 78or buprenorphine (0.1 mg), 76and bolus epidural morphine (1.5 mg) plus ketamine (20 mg). 72These were in the postoperative course followed by continuous epidural bupivacaine (5–10 mg/h) plus morphine (0.2–0.5 mg/h) or fentanyl (10–12 μg/h), 73–75,78,79epidural mepivacaine (17–60 mg/h) 76,77plus buprenorphine, and by combined epidural morphine (1 mg), ketamine (10 mg), and lidocaine (32 mg) every 12 h. 72Coadministered analgesics consisted of fentanyl or alfentanil 73–75,78and nitrous oxide 72,74–79in four and seven trials, respectively. The surgical procedures were thoracotomy, major abdominal, and total knee replacement. For details, see table 5.
Visual analog scale pain scores were significantly reduced at certain time points in three trials within the first 72 h, 72,77,79ranging between 8 and 17 mm on a VAS scale. No differences between groups were observed in the other trials (fig. 3B). Quantitative analysis of WMD of mean VAS scores recorded within 24 h was not significant (WMD, −3 mm; 95% CI, −10 to 5 mm; calculated using a random-effect model as P = 0.0002;fig. 4B). Supplemental patient-controlled analgesia–morphine consumption was significantly reduced by 3 mg over 24 h in only one trial. 72
Intensity of pain scores was considered adequate (> 30 mm VAS) in negative trials 73–76,78and not a cause of possible insensitivity (although low [< 30 mm VAS] in two of the positive trials). 72,79Power analysis was performed in only four trials, 73,75,77–78revealing an 80% power to detect a 12–23-mm difference in VAS at the 5% significance level.
In conclusion, the results showed no overall improvement in postoperative pain relief with preemptive versus postincisional continuous epidural analgesia.
Caudal and Intratheal Analgesia.
Five trials comparing preemptive with postincisional caudal block were identified. 80–84The analgesics–anesthetics investigated were bupivacaine (0.25%, 0.5–0.8 ml/kg) 80,82–84plus morphine (0.02 mg/kg), 83and lidocaine (1%, 0.5 ml/kg). 81In none of the trials was other intraoperative analgesics administered except for nitrous oxide in all studies. The surgical procedures consisted of hernia repair, orhidopexy, circumcision, and operation for club foot deformities (table 6).
Only in the trial of combined caudal bupivacaine and morphine 83pain scores and analgesic demand were significantly reduced by the preemptive treatment, ranging 50% and 1 mg of morphine over a 24-h observation period. In no other trials were differences between treatment groups observed. 80–82,84
In one trial, preoperative spinal bupivacaine (15 mg) was compared with an identical postsurgical treatment in patients undergoing abdominal hysterectomy with general anesthesia. 85No difference in pain scores were observed between treatment groups, but morphine consumption was significantly greater from 0 to 12 h after surgery in the preemptive compared with the postincisional group.
In conclusion, preemptive treatment was ineffective in four of five studies of caudal block and in the one study of intrathecal block.
Peripheral Local Anesthetics.
Twenty trials comparing preemptive with postincisional application of peripheral local anesthetics were found eligible for analysis. These could be divided into trials of wound infiltration, peripheral nerve block, and intraperitoneal infiltration.
Sixteen trials compared preoperative incisional local anesthetics with similar postincisional administration. 86–101Bupivacaine (0.25–0.5%), ropivacaine (0.75%), and lidocaine (1–1.5%) were administered in volumes between 4 and 45 ml depending on the extent of the surgical incision and type of procedure. Intraoperative fentanyl or alfentanil and nitrous oxide were coadministered in 10 87–90,92,93,96–98,101and 13 studies, 86–91,94,95,97–101respectively. Evaluated surgical procedures were hernia repair, appendectomy, hysterectomy, tonsillectomy, total knee replacement, laparoscopy, breast biopsy, and odontologic surgery (table 7).
Pain scores were significantly reduced 24 h after surgery in the preemptive group in one trial 101and at certain time points in the postincisional group in two other trials. 93,98In the other trials, no differences in pain scores between groups were observed (fig. 3C).
Quantitative analysis was only performed with 14 trials because of lack of dispersion measures in the last two trials. 86,92Using a fixed-effect model (P = 0.29), the WMD of VAS pain scores between treatment groups was nonsignificant (WMD, 0 mm; 95% CI, −3 to 4;fig. 4C).
Analgesic demand was significantly reduced by 50% over a 6-h observation period in one trial, 86and time to first analgesic request was prolonged by 4 h in another trial 101in the preemptive compared with the postsurgical treatment groups. In none of the other trials were significant differences observed between study groups.
A number of studies suffered from low internal sensitivity because of low pain scores in either group. 94,95,99Furthermore, statistical power analysis was only performed in seven of the trials, 86,90,93–97revealing an 80–90% power of detecting a difference of 10–15 mm VAS. In summary, there is no evidence for improved pain relief with preemptive local anesthetic wound infiltration compared with a similar postincisional administration.
Peripheral Nerve Blocks and Intraperitoneal Local Anesthetic.
Three trials investigated an ilioinguinal iliohypogastric nerve block in patients undergoing cesarean delivery, 102axillary block in hand or forearm surgery, 103and intercostal nerve block in patients undergoing thoracotomy. 104In the latter study, preincisional versus postincisional intravenous morphine and intramuscular diclofenac was coadministered using a multimodal approach 104(table 7).
No significant difference in pain relief was observed after cesarean section, but results were difficult to interpret because of technical difficulties in obtaining a sufficient block in the preemptive group and because of low pain scores in either group. 102In the trial of axillary block, postoperative pain and analgesic demand were improved in the postincisional compared with the preemptive group. 103In contrast, pain scores were reduced during a vital capacity breath test but not at rest, and analgesic demands were not improved by preemptive versus postincisional treatment in the trial of thoracotomy. 104
Finally, pain scores and demand for supplementary ketorolac were reduced by 10 mm VAS and 13 mg, respectively, in the preemptive treatment group from 8 to 24 h after surgery in one trial of topical intraperitoneal 0.5% bupivacaine. 105In conclusion, the limited data available do not allow conclusions as to a positive effect of preemptive analgesia with peripheral nerve blocks or intraperitoneal local anesthetic.
Chronic Postoperative Pain
Only one study was available comparing preemptive versus postincisional continuous epidural mepivacaine in patients undergoing thoracotomy. 77Pain scores and the percentages of pain-free patients were improved in the preemptive group at 3 and 6 months after surgery in a fashion parallel to findings on acute pain scores.
Statistical improvements in postoperative pain relief by the preemptive compared with the postincisional treatment were observed in some parameters or time points in 24 of 80 (82 treatment arms) trials. Quantitative analyses of WMD of average VAS pain scores recorded within 24 h after surgery were in no case significant in favor of the preemptive treatment.
The review revealed a lack of evidence for preemptive treatment with NSAIDs, intravenous opioids, intravenous ketamine, peripheral local anesthetics, and caudal analgesia to be of any benefit with respect to postoperative pain relief compared with a similar postincisional treatment. Results from trials of single-dose epidural treatment were inhomogeneous, with more than half of the trials showing statistically significant, but in most cases small, improvements with preemptive analgesia. Results from a third of the trials of continuous epidural analgesia demonstrated, at certain time points, statistically improved pain relief or analgesic demand by preemptive treatment, but overall interpretation of all continuous epidural regimens did not support the hypothesis that preemptive analgesia is of greater benefit than analgesia administered after the onset of the surgical procedure.
We tested the clinical evidence for timing of analgesia to improve postoperative pain control in the early and long-term postoperative period in this systematic review. Only trials designed to compare similar preincisional and postincisional treatment were included, excluding a number of studies from the analysis. 19,20We chose to include a statistical combination of data from the independent trials in a quantitative analysis in addition to the qualitative systematic review. This was done to produce a single estimate of the effect of the intervention and to help resolve disparities between conflicting studies. 106However, only data on pain scores could be quantitatively analyzed. For the quantitative analysis, we chose to use recordings of average pain scores within the first 24 h postoperatively as we considered this to be a clinically relevant measure and a way to overcome difficulties if only one of several recordings were found significant in an individual study. This analysis may therefore have overlooked potential positive findings within the immediate postoperative period or during the next few postoperative days. With these assumptions, our qualitative and quantitative analysis should be viewed together to achieve an overall synthesis of the results.
A concern was the lack of internal sensitivity and power in some of the negative studies. Validity criteria for the included studies was a number of 10 or more patients per treatment group. 9Internal sensitivity was evaluated with respect to pain intensity, since it has been recognized that it is difficult to detect an improvement with low or no pain. 8Furthermore, similar pain scores in study groups receiving active treatment may reflect similar analgesic effects or no effects at all. Inclusion of a placebo group in the comparison would solve the problem with similar or low pain scores. 107Although pain intensity was low in some trials, and because only rather few trials on preemptive analgesia did include a placebo group, we did not exclude such trials from the analysis, but instead documented studies with low pain scores in the Results and in the tables.
Criticism has previously been raised against a number of negative studies in which both study groups received intraoperative opioid. 108Such treatment may have caused a similar preemptive effect in both the preoperative and postoperative treatment groups and thereby contributed to the lack of difference in postoperative pain control between groups. Furthermore, various anesthetics have been demonstrated to suppress spinal sensitization in experimental studies. 109However, such studies have not been excluded from our analysis, since the objective was to investigate if preemptive techniques combined with conventional intraoperative management, which often includes intraoperative opioids or nitrous oxide, can improve postoperative pain control. Although trials were quality assessed, potential pitfalls in individual trials, such as inadequacy of used statistics, may have remained unidentified. Finally, pooling of data from a class of analgesics (e.g. , NSAIDs) may blur a possible effect of one specific agent (e.g. , ketorolac). However, no such pattern was observed.
A total of 80 trials meeting the strict inclusion and exclusion criteria were identified. The trials were divided into those of NSAIDs, intravenous opioids, parenteral NMDA receptor antagonists, epidural analgesia (single dose or continuous), caudal analgesia, and peripheral local anesthetics. A common feature of the analysis was that timing of analgesia did not influence the quality of postoperative pain control, whatever the type of preemptive analgesia. This conclusion may have clinical relevance. It implies that NSAIDs should not routinely be given preemptively because of the lack of enhanced analgesic effects and because of potential adverse effects such as increased intraoperative bleeding with the preoperative treatment compared with postoperative treatment.
With regard to NMDA receptor antagonists, trials of ketamine were uniformly negative, while the only two existing studies of dextromethorphan were positive of a preemptive effect. Further data are obviously needed to allow a final conclusion as to the clinical recommendation of preemptive treatment with dextromethorphan.
Pain control was, at certain time points, improved by preemptive analgesia in 7 of 11 treatment arms of trials of single-dose epidural analgesia. However, validity and clinical relevance was questionable in several cases and difficult to interpret. Results were therefore considered to reveal a lack of evidence for any important effect (rather than evidence for lack of effect) with preemptive analgesia. Preemptive continuous epidural treatment extending into the postoperative period might theoretically have an improved capacity to reduce nociceptive input and thereby central neuroplasticity caused not only by incision and on-going surgery but also by postsurgical inflammation. However, the results were uniformly negative. In the few studies with improved analgesia, this was only observed at certain time points and not in the overall quantitative analysis. An explanation for the negative findings of continuous epidural regimens may be that, despite continuous treatment, it was insufficient to prevent the development and maintenance of injury-induced central sensitization.
It is widely assumed that preemptive analgesia may reduce the risk of developing chronic postoperative pain. This assumption may be supported by data suggesting that patients with high intensity of acute postoperative pain scores also have a higher risk of developing a chronic pain state. 110In the only trial to compare the effect of identical preincisional versus postincisional treatment 77on long-term pain, the percentage of patients with pain at 6 months postoperatively was significantly reduced. Obviously, more data are needed, and in other trials of preemptive treatment versus no treatment in prostatectomy, 20thoracotomy, 111or amputation, 112only one demonstrated an effect on chronic postoperative pain. 20However, in this study, 20the follow-up rate was low (65%) and with a diversity between pain and activity scores at the different follow up intervals, making interpretation difficult.
It may be considered surprising and disappointing that the overall conclusion of this systematic review has been negative as to a potential beneficial effect of preemptive analgesia on postoperative pain. The issue of preemptive analgesia for postoperative pain relief has been a topic of several articles and editorials, in which terminology and definition has varied, thereby creating much of the controversy about this concept. 6,108,113The concept has been further complicated by mixing results from trials of preincisional versus postincisional treatment and trials of pretreatment versus no treatment. 6A number of suggestions have been offered to explain negative results: outcome measurement problems, too low or too high noxious stimulation induced by the surgical procedure, insufficient afferent blockade–analgesia, insufficient central inhibition, and insufficient duration of the treatment. 6,108,113,114The current analysis of clinical trials has only focused on one aspect of this discussion, namely, whether timing of conventional analgesic therapy, i.e. , preinjury versus postinjury initiation of analgesia, has a clinically significant impact on postoperative pain relief. One conservative conclusion that may be drawn from this review is that there is no need for further trials to investigate the role of timing of preemptive single-dose (short-lasting) analgesic treatment on the postoperative pain pattern. Furthermore, only three of eight trials investigating preemptive continuous epidural treatment extending into the postoperative period demonstrated improved pain relief at certain time points. Thus, overall results are also negative when timing is considered as the variable in prolonged analgesic treatment. It is important to realize, however, that these conclusions do not preclude a possible beneficial effect of an aggressive, perioperative, analgesic intervention on short- and long-term pain after surgery. We suggest that future studies redirect their focus from timing of perioperative analgesia (preemptive analgesia) to protective analgesia, aimed at the prevention of pain hypersensitivity (pathologic pain). These studies should investigate the effects of intensive and prolonged, multimodal analgesic (“protective”) interventions versus less aggressive, conventional perioperative analgesia on immediate and late postoperative pain.