Postoperative Analgesia  
HJ McQuay, DM, Clinical Reader in Pain Relief
Pain Research
Nuffield Department of Anaesthetics
University of Oxford
Oxford Radcliffe Hospital
The Churchill
Oxford OX3 7LJ
UK

Tel: 01865 226161
Fax: 01865 226160
Email: henry.mcquay@pru.ox.ac.uk
 
 
Abstract  
High-risk patients undergo simple operations as well as complicated ones. Optimal analgesia for simple procedures can be deduced from systematic reviews. For simple drug treatments we know that oral NSAIDs are the most effective, and that injected opioids need to be given at higher doses to beat the oral NSAIDs. For patients who cannot take NSAIDs combinations of paracetamol and opioid are the next best choice. For high-risk surgery spinals and epidurals using combinations of local anaesthetic and opioid may enable radical change in hospital stay and morbidity from the procedure. Against this must be set any risk from the spinal or epidural itself.
 
Keywords

Pain, Postoperative/drug therapy Analgesics, Non-Narcotic/Analgesics, Opioid/Randomized Controlled Trials

 
Writing about postoperative analgesia for the high-risk patient realistically one faces the same problems as for all patients, only more so. High-risk patients undergo simple operations as well as complicated ones. We need to provide optimal analgesia for these simple operations. For complicated procedures on high-risk patients we need evidence that complicated analgesic interventions are more successful than simple ones. We have little evidence that pre-emptive strategies are of any clinical relevance.

Analgesia is part of the package of care, and as is explained below the package of care can be manipulated to reduce hospital stay, even for the high-risk patient. This manipulation does however involve more invasive analgesic interventions, and the risks these carry, perhaps particularly in the high-risk patient, are often poorly documented. There is good evidence that the risk of adverse events is increased when high-tech approaches are used for drug administration [1], so that implementing high-tech packages because low-tech is working poorly should not be done without thinking about the risks. Perhaps the low-tech package could be delivered better, improving its efficacy, and maintaining lower levels of risk.

Postoperative analgesia is no different from other areas of medicine, in that we all have strong opinions, and often the stronger the opinion the weaker is the underlying evidence. This Chapter will summarise efforts to gather evidence for various analgesic interventions. Wherever possible recommendations are based on systematic reviews of randomised trials, because these provide the highest level of evidence of the efficacy of our interventions. We are fortunate that there is now a steady supply of systematic reviews in the pain world [2], and more general clinical recommendations have been covered elsewhere [3].
 

Amassing useful evidence

 

Efficacy

 

We know that if we want credible estimates of efficacy then these need to be taken from trials which are themselves credible and valid. The simplest starting point to assure such credibility and validity is that for efficacy we should look only to trials which are randomised to control for selection bias and double-blind to control for observer bias. If we stray from this quality standard then we are likely to overestimate treatment efficacy substantially [4].

Although we do not have a plentiful supply of randomised trials comparing treatment x with treatment y, we do have randomised trials which compare the different drugs with placebo. An analogy here is with a one hundred metre track. We can ask people to race against each other on the track, the head-to-head comparison. Alternatively we can ask each to run against the clock, rather than against each other. This would give us a listing of times from which we can produce a ranking, from fastest through to slowest. For this to be fair we need to ensure that the conditions were the same for each competitor and that we had the same timing method. Given such caveats we would have useful information about who was quickest and who was slowest, even if we could not manage to have them all race head-to-head against each other.

We have used the run-against-the-clock method to develop league tables, or rank orderings, for which analgesic works best after surgery. We used existing randomised trials which compare the different drugs with placebo so that we could compare their relative performance. To do this we developed new methods and extended existing ones, which are documented elsewhere [5]. Although the information we all want is the league-table, the credibility of that league table rests on the credibility of the methods used to derive it. The intention is to provide clinically useful information. In an ideal world one would have head-to-head comparisons of all the interventions in which one was interested. In reality these do not exist and randomised trials to detect small differences in efficacy between two analgesics would need to be massive to be able to detect differences in direction, let alone in the magnitude of the difference [6]. The league table is legitimate only because it uses information on similar patients with valid inclusion criteria (pain of moderate or severe intensity), similar measurement methods, similar outcomes, and a common comparator, placebo. While the relative ranking method is theoretically inferior, because the comparisons are not made within the same randomisation so that conditions might not have been the same for each competitor, we contend that the utility of the relative ranking far outweighs the theoretical (and acknowledged) disadvantage. We have to treat now to the best of our ability, not wait until something better comes along, which of course might never happen.
 

Harm

 

Collecting evidence about harm in postoperative pain receives much less attention than evidence about efficacy, and the rules are much less clear. With common adverse effects reasonable estimates of incidence may be detected in randomised trials. Rare (serious) adverse effects are not likely to be detected in small randomised trials, and evidence from study architectures technically weaker than randomised trials may not only be admissible but crucial. As a simple example if we have not seen a serious adverse effect in 1500 patients exposed to the treatment, the average number ‘exposed’ when a drug is registered, then we can be 95% confident that the worst possible incidence of a serious adverse effect is 1500 divided by 3, or 1 in 500 patients [7]. If you decided that 1 in 5000 was an acceptable level of risk of a serious adverse effect then you would have to study 15000 patients and not see such a problem to be 95% confident that the risk was indeed 1 in 5000.
 

The drugs we use

 

Non-opioids:- paracetamol, combinations and non-steroidal anti-inflammatory drugs

 

Effective relief can be achieved with oral non-opioids and non-steroidal anti-inflammatory drugs. Figure 1 shows part of the evolving league table for analgesic efficacy compiled from randomised trials after all kinds of surgery. Analgesic efficacy is expressed as the number-needed-to-treat (NNT), the number of patients who need to receive the active drug for one to achieve at least 50% relief of pain compared with placebo over a six hour treatment period. The most effective drugs have a low number-needed-to-treat of just over 2, meaning that for every two patients who receive the drug one patient will get at least 50% relief because of the treatment (the other patient may obtain relief but it does not reach the 50% level).
 

Figure 1: League table of number needed to treat (NNT) for at least 50% pain relief over 4-6 hours in patients with moderate to severe pain, all oral analgesics except IM morphine

 

 
For paracetamol 1 g the number-needed-to-treat is nearly 5. Combination of paracetamol 650 mg with dextropropoxyphene 65 mg improves the number-needed-to-treat slightly. Ibuprofen 400 mg is better at 3 and diclofenac 50 mg at about 2.5.

The NNT is treatment-specific, which is useful for comparison of relative efficacy. But because these NNT comparisons are against placebo, the best NNT of 2 means that while 50 of 100 patients will get at least 50% relief because of the treatment, another twenty will have a placebo response which gives them at least 50% relief, so that with ibuprofen 400 mg 70 of 100 will have effective pain relief. For comparison, with 10 mg intramuscular morphine about 53% of patients get more than 50% pain relief. Because the effect of placebo is added in, the comparisons between analgesics are not as stark as with NNT.

It is clear from Figure 1 that the oral NSAIDs do extremely well in this single-dose postoperative comparison. At these doses they all have NNT values of between 2 and 3, and the point estimate of the mean is below that of (i.e. better than) 10 mg of intramuscular morphine, even though the confidence intervals overlap. The simple analgesics, aspirin and paracetamol are significantly less effective than 10 mg intramuscular morphine. The point estimates of the NNT are higher, and there is no overlap of the confidence intervals. Analgesic efficacy of the simple analgesics is improved by combination with weak opioids, which do poorly in single doses on their own.

 

Injected analgesics — opioid and NSAID

 

Using the same methods we also have information on intramuscular pethidine and ketorolac [8]. Taken with the intramuscular morphine data shown above we can then start to think about the optimal choice within and between drug class for intramuscular injection, and indeed about the optimal choice of route of administration.

Table 1 shows comparable intramuscular analgesic efficacy of pethidine 100 mg and ketorolac 30 mg when given as a single postoperative dose for moderate to severe pain and similar to intramuscular morphine 10 mg. Adverse effects for intramuscular pethidine 100 mg were similar in type and frequency to those found with morphine 10 mg. In contrast, adverse effects with intramuscular ketorolac 30 mg were not significantly commoner than with placebo.
 

Pulling all this efficacy evidence together

 

Some selected results from the league tables for oral and intramuscular analgesics are shown in Figure 2. While the data for ketorolac 10 mg i.m. looks anomalous it is important to say that there were few patients in that analysis (Table 1). What is obvious from Figure 2 is that oral NSAIDs perform as well as 10 mg morphine or 100 mg pethidine, and that there is little difference apparent in the efficacy of oral and injected ketorolac. These two facts may seem counter-intuitive but in reality have been known for many years from head-to-head comparisons, of oral NSAID and injected opioid, and of injected versus oral NSAID.

 

Figure 2: Forest plot for selected oral and intramuscular analgesics: number needed to treat (NNT) for at least 50% pain relief over 4-6 hours in patients with moderate to severe pain

 

Number-need-to-treat (NNT) for at least 50% pain relief in postoperative pain. All doses oral unless labelled IM (intramuscular). Data for pethidine and ketorolac from [8], for morphine from [9], and for other drugs see [5]

 

Even if patients can swallow is it best to give drugs by injection or suppository?

Most postoperative pain is managed solely with medication. Perhaps because anaesthetists work with injected drugs there is a natural belief that drugs which are injected are more powerful than drugs taken by mouth.

There is however no evidence that NSAIDs given rectally or by injection perform better than the same drug at the same dose given by mouth [8;10] (Figure 2). These other non-oral routes become appropriate when patients cannot swallow.

The patient can’t swallow analgesics

The emerging information on the relative efficacy of injected opioids or NSAIDs suggests (Figure 2) that there is probably little difference in efficacy between say 10 mg morphine and 30 mg ketorolac. The non-opioid advantage of the NSAID may however be ruled out if there is concern over adverse effects. Adverse effect data on non-steroidal anti-inflammatory drugs from long term oral dosing, where gastric bleeding is the main worry, rates ibuprofen the safest [11]. In postoperative pain the main concerns are renal and coagulation problems. Acute renal failure can be precipitated in patients with pre-existing heart or kidney disease, those on loop diuretics or those who have lost more than 10% of blood volume. Non-steroidal anti-inflammatory drugs cause significant lengthening (~30%) of bleeding time, usually still within the normal range. This can last for days with aspirin, hours with non-aspirin non-steroidal anti-inflammatory drugs. Whether or not non-steroidal anti-inflammatory drugs cause significant increase in blood loss remains contentious. Importantly, increasing the dose of opioid will increase analgesia. Injecting 20 mg of morphine gave greater analgesia than 10 mg [12]. Increasing the dose of NSAID may not produce as steep an increase in the analgesic effect as occurs with opioids.
 

Opioids

 

For severe postoperative pain opioids are the first line treatment, and to date we have just two systematic reviews, morphine and pethidine [8;9]. Intermittent opioid injection can provide effective relief of postoperative pain [13], but unfortunately, adequate doses are withheld. Addiction is not a problem with opioid use in acute pain [14].

Irrespective of the route, opioids used for people who are not in pain, or in doses larger than necessary to control the pain, can slow or indeed stop breathing. The key principle is titration of dose against desired effect - pain relief - and minimising unwanted effects. If the patient is still complaining of pain and you are sure that the drug has all been delivered and absorbed, then it is safe to give another, usually smaller, dose (5 minutes after intravenous, 1 hour after intramuscular or subcutaneous, 90 minutes after oral). If the second dose is also ineffective, then repeat the process or change the route of administration to achieve faster control. Delayed release formulations, oral or transdermal, should not be used in acute pain, because delayed onset and offset are dangerous in this context.

There is no compelling evidence that one opioid is better than another, but there is good evidence that pethidine has a specific disadvantage [15] and no specific advantage. Given in multiple doses the metabolite norpethidine can accumulate and act as a central nervous system irritant, ultimately causing convulsions, especially in renal dysfunction. Pethidine should not be used when multiple injections are needed. The old idea that pethidine is better than other opioids at dealing with colicky pain is no longer tenable [16].

Morphine and its relatives have an active rather than a toxic metabolite, morphine-6-glucuronide. In renal dysfunction this metabolite can accumulate and result in greater effect from a given dose, because it is more active than morphine. If you are, as you should be, titrating dose against effect, this will not matter. Less morphine will be needed. Accumulation can be a problem with unconscious intensive care patients on fixed dose schedules when renal function is compromised.

There is no good evidence that opioid adverse effect incidence is different with different opioids at the same level of analgesia. There are strong arguments, based on minimising risk, for using one opioid only, so that everyone is familiar with dosage, effects and problems. Whichever drug you choose, simple changes to the way opioids are used, good staff education and implementation of an algorithm for intermittent opioid dosing, can have a powerful impact on pain relief and patient satisfaction [13].

Nurse administered intermittent opioid injection requires good staffing levels to minimise delay between need and injection. Staffing shortage, ward distractions and controlled drug regulations all increase the delay. Patient controlled analgesia overcomes these logistical problems. The patient presses a button and receives a pre-set dose of opioid, from a syringe driver connected to an intravenous or subcutaneous cannula. This delivers opioid to the same opioid receptors as an intermittent injection, but allows the patient to circumvent delays. Not surprisingly there is little difference in outcome between efficient intermittent injection and patient-controlled analgesia [17]. Good risk management with patient-controlled analgesia should emphasise the same drug, protocols and equipment throughout the hospital.
 

Regional Techniques

 

Asking radical questions

 

Asking radical questions about acute postoperative management such as "Why are all operations not ambulatory, pain-free and risk-free?", or, more familiar, "Why is this patient still in hospital?", is forcing a reconsideration of the role of combined (local or regional plus general anaesthesia) approaches. Instead of asking questions such as "Is regional better than general anaesthesia" we need to look at the whole episode, before, during and after surgery, not just the operative period [18;19]. Costs of each care episode should fall if the hospital stay is reduced, and a healthy patient returned home will cost the community less than a sick patient requiring considerable input from the primary care team. Our old question was whether regional or regional supplemented general anaesthesia could produce major reductions in morbidity and mortality - the general versus regional anaesthesia question. An example is vascular surgery. The Yeager study [20] did suggest improvement in patients with regional anaesthesia in patients undergoing abdominal aortic aneurysm or lower extremity vascular surgery. A feature of subsequent studies which showed no difference between regional and general anaesthesia was an increasing extent of control over all aspects of postoperative care [21;22]. The effect of the detailed protocols was that bad outcomes were reduced in all groups [23]. The implication is that even if there was a difference it would take a really huge study to show it using the morbidity and mortality outcomes [24]. But the suggestion is that it is only if the postoperative protocols allow any advantage to be expressed, such as advantage in time to feeding or time to walking, that we will see a difference between regional and general anaesthesia. Epidural local anaesthetic may well allow bowel function to return earlier [25], but only if protocol allows it will we see patients going home two days after major surgery [26]. The point is that this radical change is only possible if the procedure is done under an epidural, because the epidural makes it possible for the patient to mobilise early and for bowel function to return earlier.

For some operations there is proven advantage of regional over general anaesthesia. For hip [27] and knee [28] replacements "solid" epidural anaesthesia with sedation during surgery followed by postoperative epidural can produce reduced blood loss, faster surgery, reduced morbidity and faster rehabilitation. In this context change has been gradual rather than radical, but again the key is the epidural, both for the operation and afterwards.

Returning to the old question, general versus regional anaesthesia, a recent set of meta-analyses looked at randomized, controlled trials (RCTs) to assess the effects of seven different interventions on postoperative pulmonary function after a variety of procedures [29]. The seven were epidural opioid, epidural local anaesthetic, epidural opioid with local anaesthetic, thoracic versus lumbar epidural opioid, intercostal nerve block, wound infiltration with local anaesthetic, and intrapleural local anaesthetic. Compared with systemic opioids, epidural opioids decreased the incidence of atelectasis significantly. Epidural local anaesthetics compared with systemic opioids increased PaO 2 significantly and decreased the incidence of pulmonary infections and pulmonary complications overall. Intercostal nerve blockade did not produce significant improvement in pulmonary outcome measures.

Interestingly on the surrogate measures of pulmonary function (FEV1, FVC, and PEFR), there were no clinically or statistically significant differences, showing again the importance of choice of outcome measure. The results do confirm that postoperative epidural pain control can significantly decrease the incidence of pulmonary morbidity [29].

Epidural Analgesia

Epidural infusion via a catheter can offer continuous relief after trauma or surgery, for lower limb, spine, abdominal or chest. The current optimal infusate is an opioid/local anaesthetic mixture. Opioids and local anaesthetics have a synergistic effect, so that lower doses of each are required for equivalent analgesia with fewer adverse effects [30]. The difference between the pain severity of different pain states should be emphasised, because it also means that categoric prescriptions for the doses to be used in combination infusions are likely to be valid only for a particular pain state or set of circumstances. Indeed the dynamic nature of postoperative pain means that the dosage required on the day of surgery may be much higher than the dosage required on subsequent days.

A clear demonstration of the advantage of the combination of local anaesthetic and opioid was seen in a comparison of 0.125% bupivacaine in saline, diamorphine 0.5 mg in 15 ml and diamorphine mixed with 0.125% bupivacaine (0.5 mg in 15 ml) infused at a rate of 15 ml/h for pain after major gynaecological surgery. The combination produced significantly superior analgesia to either of its components alone, without major adverse effects [31]. Giving the diamorphine intravenously with epidural bupivacaine was significantly less effective than giving the same dose epidurally in combination with epidural bupivacaine [32].

Many important questions are still to be answered. One practical issue is the ability of combination infusions to control pain remote from the catheter site, as with thoracic pain and a lumbar catheter. Another is whether there is any difference in efficacy or adverse effects if the drugs are given continuously rather than intermittently. For local anaesthetic alone there was little difference [33].

Combination Dosage

Three strategies in dosage are discernible, the low [31;34;35], the intermediate [36;37], and the high [38-40]. High doses (bupivacaine 0.5% 25 mg/h and morphine 0.5 mg/h) were used to produce analgesia immediately after upper abdominal surgery but at some risk [40]. The stress response was not blocked. Lower doses (bupivacaine 0.1% 4 mg/h and morphine 0.4 mg/h) did not provide total pain relief after thoracotomy [35].The issue of the minimum effective dose is of great importance, and unfortunately may have to be defined for particular circumstances. It is too early (and too circumstance-dependent) for consensus to emerge, but an intermediate dose, 0.25% bupivacaine 10 mg/h with morphine 0.2 mg/h, has its advocates for use in pain after major surgery.

The risks are those of an epidural, (dural puncture, infection, haematoma, nerve damage), those of the local anaesthetic (hypotension, motor block, toxicity) and those of the opioid, (nausea, sedation, urinary retention, respiratory depression, pruritus) (Table 3). Wrong doses do get given [1], so increased surveillance is mandatory. The risk of persistent neurological sequelae after an epidural is about 1 in 5000 [41]. Debate continues about whether patients with epidural infusions can be nursed on general wards. These techniques are only appropriate for major trauma or surgery when the potential benefits outweigh the risks.

 

Comparing blocks

 

Many of the recent comparisons, for instance in thoracic anaesthesia of one kind of block against another, are designed as A versus B comparisons. Unless the trials are very big, which most are not, one ends up concluding that the trial showed no difference, and we do not know if the trial was capable of revealing a difference if in fact there was one [42]. Choice of block also involves comparing the morbidity of the contenders. Again size of trial is crucial. Rare events will not be picked up in small trials.
 

Problems with blocks

 

One important question is whether using nerve blocks or powerful epidural techniques has deleterious consequences. For most nerve blocks, for instance axillary brachial plexus block, we just do not know the incidence of long term nerve problems. In all likelihood it is vanishingly small. In a one- year prospective survey of the French-Language Society of Pediatric Anesthesiologists Giaufre et al reported on "central blocks" (15,013), most of which were caudals. Their complication rate for these central blocks works out at 15 per 10,000 (25 incidents involving 24 patients). These were rated as minor, and did not result in any sequelae or medicolegal action [43].

In adults a report from Finland [44] reviewed all claims (1987-1993) about severe complications associated with epidural and spinal anaesthesia. Eighty-six claims were associated with spinal and/or epidural anaesthesia. There were 550,000 spinals and 170,000 epidurals. With spinals there were 25 serious complications: cardiac arrests (2), paraplegia (5), permanent cauda equina syndrome (1), peroneal nerve paresis (6), neurological deficits (7), and bacterial infections (4). With epidurals there were nine serious complications: paraparesis (1), permanent cauda equina syndrome (1), peroneal nerve paresis (1), neurological deficit (1), bacterial infections (2), acute toxic reactions related to the anaesthetic solution (2), and overdose of epidural opioid (1). This gives an overall incidence of serious complications of 0.45 per 10,000 for spinal and 0.52 per 10,000 for epidural.

Again in adults a French survey [45] reported on 40,640 spinals and 30,413 epidurals. Of the 98 severe complications 89 were attributed fully or partially to the regional anaesthesia. There were 26 cardiac arrests with spinals (6.4 ± 1.2 per 10,000 patients), and six were fatal. There were three arrests with epidurals. Of 34 neurological complications (radiculopathy, cauda equina syndrome, paraplegia), 21 were associated either with paraesthesia during puncture (n = 19) or with pain during injection (n = 2), suggesting nerve trauma or intraneural injection. Neurological sequelae were significantly commoner after spinal anaesthesia (6 ± 1 per 10,000) than after each of the other types of regional procedures (1.6 ± 0.5 per 10,000).

Yuen et al [46] described 12 patients (out of an estimated 13,000) with complications after lumbar epidurals. Eleven patients had lumbosacral radiculopathy or polyradiculopathy, ten after epidural and one after subarachnoid injection of medication during intended epidural. One patient had a thoracic myelopathy after an unintended spinal. From their data one may estimate the incidence of short-term (persisting less than one year) neurological sequelae after an epidural as 9 per 10,000, and for longer term (persisting more than one year) 2 per 10,000. As the authors point out these incidences appear to have changed little from previous case series [46].

An important RCT looked at the question of whether there was any difference in long-term cognitive dysfunction after total knee replacement surgery in 262 older adults (median age 69 years; 70% women) after epidural or general anaesthesia [47]. The importance is that it is the largest trial to date of the effects of general versus regional anaesthesia on cerebral function, with more than 99% power to detect a clinically significant difference on any of the neuropsychological tests. Preoperative neuropsychological assessment was repeated postoperatively at 1 week and 6 months. Cognitive outcome was assessed by within-patient change on 10 tests of memory, psychomotor, and language skills. There were no significant differences between the epidural and general anaesthesia groups on any of the 10 cognitive tests at either 1 week or 6 months. Overall, 5% of patients showed a long-term clinically significant deterioration in cognitive function.
 

Summary

 

For simple procedures on high-risk patients the evidence for drug treatments suggests a clear strategy of using injected opioid, oral NSAID or oral paracetamol/opioid combinations. The advent of COX-2 NSAIDs will require us to compare existing regimes with the new drugs. For complicated procedures there is at last evidence that spinals and epidurals using combinations of local anaesthetic and opioid may enable radical change in hospital stay and morbidity from the procedure. Against this must be set any risk from the spinal or epidural itself. We await evidence which balances the benefit and the risk.

 

 

 
References

 

1. Bates DW, Cullen DJ, Laird N, Petersen LA, Small SD, Servi D, Laffel G, Sweitzer BJ, Shea BF, Hallisey R, Vliet MV, Nemeskal R, & Leape LL. Incidence of adverse drug events and potential adverse drug events. Journal of the American Medical Association 1995;274:29-34.
2. Jadad AR & McQuay HJ. Meta-analyses to evaluate analgesic interventions: a systematic qualitative review of their methodology. Journal of Clinical Epidemiology 1996;49:235-43.
3. McQuay HJ, Justins D, & Moore RA. Treating acute pain in hospital. British Medical Journal 1997; 314:1531-5.
4. Schulz KF, Chalmers I, Hayes RJ, & Altman DG. Empirical evidence of bias: dimensions of methodological quality associated with estimates of treatment effects in controlled trials. Journal of the American Medical Association 1995;273:408-12.
5. McQuay HJ and Moore RA. An evidence-based resource for pain relief. Oxford: Oxford University Press, 1998.
6. Moore RA, Gavaghan D, Tramer MR, Collins SL, & McQuay HJ. Size is everything--large amounts of information are needed to overcome random effects in estimating direction and magnitude of treatment effects. Pain 1998;78:209-16.
7. Eypasch E, Lefering R, Kum CK, & Troidl H. Probability of adverse events that have not yet occurred: a statistical reminder. British Medical Journal 1995;311:619-20.
8. Smith LA, Carroll D, Edwards JE, Moore RA, & McQuay HJ. Single dose ketorolac and pethidine in acute postoperative pain: a systematic review. (Submitted)
9. McQuay HJ, Carroll D, & Moore RA. Injected morphine in postoperative pain: a quantitative systematic review. Journal of Pain and Symptom Management 1999;17:164-74.
10. Tramèr MR, Williams JE, Carroll D, Wiffen PJ, Moore RA, & McQuay HJ. Comparing analgesic efficacy of non-steroidal anti-inflammatory drugs given by different routes in acute and chronic pain: a qualitative systematic review. Acta Anaesthesiologica Scandanavica 1998;42:71-9.
11. Henry D, Lim LL-Y, Rodriguez LAG, Gutthann SP, Carson JL, Griffin M, Savage R, Logan R, Moride Y, Hawkey C, Hill S, & Fries JT. Variability in risk of gastrointestinal complications with individual non-steroidal anti-inflammatory drugs: results of a collaborative meta-analysis. British Medical Journal 1996;312:1563-1566.
12. Norholt SE, SindetPedersen S, Larsen U, Bang U, Ingerslev J, Nielsen O, Hansen HJ, & Ersboll AK. Pain control after dental surgery: A double-blind, randomised trial of lornoxicam versus morphine. Pain 1996;67:335-343.
13. Gould TH, Crosby DL, Harmer M, Lloyd SM, Lunn JN, Rees GAD, Roberts DE, & Webster JA. Policy for controlling pain after surgery: effect of sequential changes in management. British Medical Journal 1992;305:1187-1193.
14. Porter J & Jick H. Addiction rate in patients treated with narcotics. New England Journal of Medicine 1980;302:123.
15. Szeto HH, Inturrisi CE, Houde R, Saal S, Cheigh J, & Reidenberg M. Accumulation of norperidine, an active metabolite of meperidine, in patients with renal failure or cancer. Annals of Internal Medicine 1977;86:738-741.
16. Nagle CJ & McQuay HJ. Opiate receptors; their role in effect and side-effect. Current Anaesthesia & Critical Care 1990;1:247-252.
17. Ballantyne JC, Carr DB, Chalmers TC, Dear KB, Angelillo IF, & Mosteller F. Postoperative patient-controlled analgesia: meta-analyses of initial randomized control trials. Journal of Clinical Anesthesia 1993;5:182-93.
18. Kehlet H. Multimodal approach to control postoperative pathophysiology and rehabilitation. British Journal of Anaesthesia 1997;78:606-617.
19. Kehlet H. Postoperative pain relief-what is the issue? British Journal of Anaesthesia 1994;72:375-378.
20. Yeager M, Glass D, Neff R, & Brinck-Johnsen T. Epidural anaesthesia & analgesia in high risk surgical patients. Anesthesiology 1987;66:729-36.
21. Bode RJ, Lewis KP, Zarich SW, Pierce ET, Roberts M, Kowalchuk GJ, Satwicz PR, Gibbons GW, Hunter JA, & Espanola CC. Cardiac outcome after peripheral vascular surgery. Comparison of general and regional anesthesia. Anesthesiology 1996;84:3-13.
22. Christopherson R, Beattie C, Frank SM, Norris EJ, Meinert CL, Gottlieb SO, Yates H, Rock P, Parker SD, Perler BA et al. Perioperative morbidity in patients randomized to epidural or general anesthesia for lower extremity vascular surgery. Perioperative Ischemia Randomized Anesthesia Trial Study Group. Anesthesiology 1993;79:422-34.
23. Beattie C, Roizen MF, & Downing JW. Cardiac outcomes after regional or general anesthesia: do we know the question? Anesthesiology 1996;85:1207-9.
24. Rigg JRA & Jamrozik K. Outcome after general or regional anaesthesia in high-risk patients. Current Opinion in Anaesthesiology 1998;11:327-331.
25. Liu SS, Carpenter RL, Mackey DC, Thirlby RC, Rupp SM, Shine TS, Feinglass NG, Metzger PP, Fulmer JT, & Smith SL. Effects of perioperative analgesic technique on rate of recovery after colon surgery. Anesthesiology 1995;83:757-65.
26. Bardram L, Funch-Jensen P, Jensen P, Crawford ME, & Kehlet H. Recovery after laparoscopic colonic surgery with epidural analgesia, and early oral nutrition and mobilisation. Lancet 1995;345:763-4.
27. Sharrock NE &Salvati EA. Hypotensive epidural anesthesia for total hip arthroplasty: a review [see comments]. Acta Orthopaedica Scandinavica 1996;67:91-107.
28. Williams-Russo P, Sharrock NE, Haas SB, Insall J, Windsor RE, Laskin RS, Ranawat CS, Go G, & Ganz SB. Randomized trial of epidural versus general anesthesia: outcomes after primary total knee replacement. Clinical Orthopaedics and Related Research 1996;199-208.
29. Ballantyne JC, Carr DB, deFerranti S, Suarez T, Lau J, Chalmers TC, Angelillo IF, & Mosteller F. The comparative effects of postoperative analgesic therapies on pulmonary outcome: cumulative meta-analyses of randomized, controlled trials. Anesthesia and Analgesia 1998;86:598-612.
30. McQuay H. Epidural Analgesics. In Wall, P. and Melzack, R. (ed) Textbook of Pain . London: Churchill Livingstone, 1994, p. 1025-1034
31. Lee A, Simpson D, Whitfield A, & Scott D. Postoperative analgesia by continuous extradural infusion of bupivacaine and diamorphine. British Journal of Anaesthesia 1988;60:845-850.
32. Lee A, McKeown D, Brockway M, Bannister J, & Wildsmith JAW. Comparison of extradural and intravenous diamorphine as a supplement to extradural bupivacaine. Anaesthesia 1991;46:447-450.
33. Duncan LA, Fried MJ, Lee A, & Wildsmith JA. Comparison of continuous and intermittent administration of extradural bupivacaine for analgesia after lower abdominal surgery. British Journal of Anaesthesia 1998;80:7-10.
34. Cullen M, Staren E, El-Ganzouri A, Logas W, Ivankovich D, & Economou S. Continuous epidural infusion for analgesia after major abdominal operations: A randomised, prospective, double-blind study. Surgery 1985;10:718-728.
35. Logas W, El BN, El GA, Cullen M, Staren E, Faber L, & Ivankovich A. Continuous thoracic epidural analgesia for postoperative pain relief following thoracotomy: a randomized prospective study. Anesthesiology 1987;67:787-91.
36. Bigler D, Dirkes W, Hansen R, Rosenberg J, & Kehlet H. Effects of thoracic paravertebral block with bupivacaine versus combined thoracic epidural block with bupivacaine and morphine on pain and pulmonary function after cholecystectomy. Acta Anaesthesiologica Scandanavica 1989;33:561-4.
37. Seeling W, Bruckmooser K, Hufner C, Kneitinger E, Rigg C, & Rockemann M. [No reduction in postoperative complications by the use of catheterized epidural analgesia following major abdominal surgery]. Anaesthesist 1990;39:33-40.
38. Hjortso NC, Neumann P, Frosig F, Andersen T, Lindhard A, Rogon E, & Kehlet H. A controlled study on the effect of epidural analgesia with local anaesthetics and morphine on morbidity after abdominal surgery. Acta Anaesthesiologica Scandinavica 1985;29:790-6.
39. Schulze S, Roikjaer O, Hasselstrom L, Jensen N, & Kehlet H. Epidural bupivacaine and morphine plus systemic indomethacin eliminates pain but not systemic response and convalescence after cholecystectomy. Surgery 1988;103:321-327.
40. Scott N, Mogensen T, Bigler D, Lund C, & Kehlet H. Continuous thoracic extradural 0.5% bupivacaine with or without morphine: effect on quality of blockade, lung function and the surgical stress response. British Journal of Anaesthesia 1989;62:253-7.
41. Kane RE. Neurologic deficits following epidural or spinal anesthesia. Anesthesia and Analgesia 1981;60:150-161.
42. McQuay H & Moore A. Placebos are essential when extent and variability of placebo response are unknown. British Medical Journal 1996;313:1008.
43. Giaufre E, Dalens B, & Gombert A. Epidemiology and morbidity of regional anesthesia in children: a one- year prospective survey of the French-Language Society of Pediatric Anesthesiologists. Anesthesia and Analgesia 1996;83:904-12.
44. Aromaa U, Lahdensuu M, & Cozanitis DA. Severe complications associated with epidural and spinal anaesthesias in Finland 1987-1993. A study based on patient insurance claims. Acta Anaesthesiologica Scandinavica 1997;41:445-52.
45. Auroy Y, Narchi P, Messiah A, Litt L, Rouvier B, & Samii K. Serious complications related to regional anesthesia: results of a prospective survey in France. Anesthesiology 1997;87:479-86.
46. Yuen EC, Layzer RB, Weitz SR, & Olney RK. Neurologic complications of lumbar epidural anesthesia and analgesia. Neurology 1995;45:1795-801.
47. Williams-Russo P, Sharrock NE, Mattis S, Szatrowski TP, & Charlson ME. Cognitive effects after epidural vs general anesthesia in older adults. A randomized trial. Journal of the American Medical Association 1995;274:44-50.


5822 words