Pain and its control

Henry McQuay, DM
Clinical Reader in Pain Relief
email
Pain sensation & transmission
Figure 1.
Some definitions
Psychological factors
Methods of measuring pain
Research
Figure 2.
Clinical practice
Problem areas
Treatment
Figure 3.
Figure 4.
Acute pain
Figure 5.
Low-tech: Intermittent opioid injection
Figure 6.
High-tech: Patient-Controlled Analgesia (PCA)
High-tech: Epidural Infusion
Minor surgery (and later after major surgery)
Figure 7.
Local anaesthetic blocks
Acute Pain Services
Chronic pain
Non-opioid analgesics
Oral NSAIDs, combinations and others
Comparisons
Topical NSAIDs
Opioids
Unconventional analgesics
Antidepressants
Table1. Number-needed-to-treat (NNT) for some analgesic interventions
Anticonvulsants
Others
Block Nerve Transmission
Reversible
Local Anaesthetics
Table 2. Common nerve blocks
Fibromyalgia
Intravenous regional sympathectomy
Epidural steroids and facet joint blocks
Irreversible
Surgery
Alternatives
TENS, acupuncture
Physio and variants
Behavioural Management
References
This chapter should act as a signpost. The interested reader is pointed to the best evidence available, systematic reviews where available, and the efficacy of drugs and other interventions is presented as number-needed-to-treat.

Pain sensation & transmission

The easiest way to think of pain in the nervous system is the idea of pain receptors and nerve cables dedicated to the transmission of pain signals (Figure 1: Descartes' L'Homme, 1664), a hard-wired, line-labelled system. This view has always had obvious flaws. The return of pain after an initially successful cordotomy, and the phenomenon of phantom limb pain are two examples. In a hard-wired line-labelled system the pain should not recur after the cordotomy and patients should not feel pain in a limb as they did before the accident or amputation. Such flaws mean that the simple view of a 'passive' nervous system does not explain all that we see.
Figure 1.

In the peripheral nervous system most of the nociceptive signalling of thermal and mechanical stimuli comes from activation of polymodal nociceptors which are innervated by C-fibres. If tissue is damaged these fibres also respond to local chemical stimulation and become sensitised to chemical, thermal and mechanical stimuli. Tissue damage can stimulate the synthesis of arachidonic acid metabolites from adjacent membranes, cleavage of the bradykinin precursor to release active peptide and, via the axon reflex, the release of peptides such as substance P and CGRP from the C-fibres. This inflammatory soup, which also contains 5HT, K + ions and H + ions, activates and sensitises peripheral nerve endings, and causes vasodilatation and plasma extravasation. The net result is swelling, pain and tenderness, like the contact between warm shower water or clothing on sunburnt skin.

The pharmacology of the dorsal sensory horn of the spinal cord is rich. Transmitters come from either the afferent fibres, intrinsic neurones or descending fibres. Most are concentrated in the substantia gelatinosa, one of the densest neuronal areas in the CNS, and crucial for the reception and modulation of nociceptive messages from the peripheral fibres. C-fibres terminate in the outer lamina 1 and the underlying substantia gelatinosa, whereas the large tactile fibres terminate in deeper laminae. However, as well as the lamina 1 cells sending long ascending axons to the brain, deep dorsal horn cells also give rise to ascending axons and respond to C-fibre stimulation. The C-fibre input to these deep cells may be relayed via interneurones, or arrive on the dendrites of the cells which pass vertically into the gelatinosa. This, and projection of inhibitory gelatinosa cells towards the deep cells, enables convergence and modulation of the deep cells' responses. Finally the reason for the good correspondence between the electrophysiology of these cells, (behaviour, psychophysical studies and reflex responses to pain), may be that these deep sensory cells both drive the withdrawal reflex and also, via ascending projections, are involved in the perception of pain.

At each of these stages the signal can be amplified or damped by endogenous influences such as mood or endorphins, or exogenous factors, drugs given or the circumstances of the injury. The classic damping-down scenario is the injured soldier who continues, despite a shattered leg, to get himself out of battle. Conversely a stubbed toe when you are tired and miserable is immeasurably more painful than the same injury on a cheerful morning.

For most acute pain the idea of specific cables whose transmission can be blocked is reinforced by the fact that you can operate (painlessly) on an injured foot by using a local anaesthetic block of the foot or a local anaesthetic spinal or epidural. If or when the operation site becomes the site of chronic pain, the inadequacy of the simple explanation is exposed. Phantom pain needs a more complicated explanation, because the painful foot or hand, the receptors and the cables, are no longer there. The concept of 'pain memory' in the spinal cord and brain has to be invoked.

The idea that the nervous system can change, or be 'plastic', has led to the use of the word plasticity, and this concept has had a major impact in both acute and chronic pain (McQuay & Dickenson, 1990). The simplest idea is that a memory of a pain is made and stored in the nervous system. Preventing such a memory being laid down led to the idea of pre-empting postoperative pain; any pain after the operation would be easier to treat if the memory had been minimised (McQuay, 1995). Long-term sequelae, such as the phantom pain which can occur after amputation, could also be prevented. One issue of importance for treatment is at what level of the nervous system such memories might be stored. If the memory is stored at a 'central' level, for instance in the brain, then attacking the pain in the leg or arm where it originated might not do any good. The memory might be held centrally but require continued input from the periphery to sustain it. Attacking the pain in the leg or arm would then have some logic.

Many treatments, or interventions, are used to treat both acute pain and chronic pain. Chronic pain, not surprisingly, has more twists and turns, because its origins may be more complicated and because the nervous system if damaged or bombarded continuously by pain messages can behave strangely. The concept of plasticity has led to some interventions coming into fashion, and to some going out of fashion. Long-term measures, such as cutting the nerves thought to carry a particular pain message, are going out of fashion. The reasoning is that the nervous system will 'rewire', the pain will re-emerge, and may well be more difficult to manage than it was in the first place. Better drug control of difficult pains has also reduced the necessity for destructive procedures.

Some definitions

Pain is an unpleasant sensory or emotional experience associated with actual or potential tissue damage, or described in terms of such damage. Chronic pain is pain still present after three months despite sensible treatment. Radicular pain is pain in the distribution of a nerve root, so that pain is felt in an area corresponding to one of the dermatomes down the arm or leg or round the trunk. It is typically caused by compression. Referred pain , or transferred pain, is pain felt superficially in the dermatome of an affected viscus or other deep structure innervated by that root. An example is pain felt in the left arm because of cardiac disease. Central pain is spontaneous pain caused by damage to the central nervous system, often accompanied by dysaesthesia. An example is the pain which can occur on the side of the body affected by a stroke, the thalamic or post-stroke syndrome. Phantom pain is pain felt in the missing limb or stump.

Psychological factors

The influence that exogenous factors such as the circumstances of the trauma can have on pain has been mentioned. In chronic pain it is sometimes very hard to disentangle depression from pain. Pain makes depression worse and depression makes pain worse. This pattern is all too familiar to those who manage back pain. The thinking clinician needs to deal with both the pain and the depression.

Methods of measuring pain

Because pain measurement is subjective it is often thought to be of little value. This is incorrect. By sticking to some (simple) rules, pain measurement can be made to work well in research settings, and perhaps even more important, pain can be recorded along with vital signs as part of the normal clinical course.

Research

The same pain scales tend to be used in acute and chronic pain. Proof of their validity dates from thirty years ago (Houde et al, 1960). Especially in chronic pain it is sensible to use them in conjunction with disability and quality of life measures. It makes no sense to invent your own scale. It is far better to use proven tools.

Pain intensity and pain relief can be measured by either word categories, or by visual analogue (100 mm line) scales (Figure 2).

Figure 2.

For rigorous work we use all the scales (Jadad & McQuay, 1994). They are clearly not independent, but if a patient gives a silly answer on one scale, out of step with their answers on the others, then that is an indication to requestion, so reducing 'noise'. For chronic work, when patients are recording over long periods, we tend to use diaries with the word categories, and ask about both pain intensity and pain relief. Another useful scale is the patient's 'global report', sometimes used at the end of the trial period, sometimes serially during the trial. Using the same categories as the pain relief scale, this is a composite view encompassing both pain per se and any adverse effects of treatment (Gøtzsche, 1990). The clinician's global view is a notorious overestimate and should not be used (Onghena & Van Houdenhove, 1992).

Most important and simplest is the idea of including a binary scale because of its clinical relevance. The question is phrased in various forms, around the idea of "Is your pain at least 50% relieved?", to which the patient answers yes or no.

Clinical practice

The argument for using pain charts as part of normal practice is to improve the quality of care. It is the fact of a chart rather than the form of the chart which is most important. Examples are the Burford (Burford Nursing Development Unit, 1984) or the London Hospital chart. These should be done together with vital signs, and can then be used for both clinical care and for audit.

Problem areas

Two important groups of patients, children and the unconscious, present particular problems for pain measurement. With unconscious patients there is no alternative to using variations in vital signs, such as blood pressure rise, as a proxy for a report of increased pain. We do not know how well these proxies work. For children over three years there are special scales (McGrath et al, 1993). Below three years again experienced staff interpreting crying and other behaviour is the best guide we have.

Treatment

This chapter uses systematic reviews when possible to provide the best available evidence for the various interventions used to relieve pain. The number-needed-to-treat (NNT) is used as the measure of clinical significance from quantitative systematic reviews. It shows the effort required to achieve a particular therapeutic target. NNT is treatment specific. It describes the difference between active treatment and control. The NNT is given by the equation

where:

IMPact = number of patients given active treatment achieving the target

TOTact = total number of patients given the active treatment

IMPcon = number of patients given a control treatment achieving the target

TOTcon = total number of patients given the control treatment

An NNT of 1 describes an event which occurs in every patient given the treatment but in no patient in a comparator group. This could be described as the "perfect" result in, say, a therapeutic trial of an antibiotic compared with placebo. For therapeutic benefit the NNT should be as close as possible to 1; there are few circumstances in which a treatment is close to 100% effective and the control or placebo completely ineffective, so NNTs of 2 or 3 often indicate an effective intervention. For unwanted effects, NNT becomes the NNH (number-needed-to-harm), which should be as large as possible.

Figure 3.

By far the majority of acute pain is managed with analgesics alone. Blocking nerve transmission with continuous epidural local anaesthetic is an option after major surgery. Most chronic pain is also managed initially with analgesics, but, by contrast with acute pain, commonly also involves nerve transmission block and alternative methods. Figure 4 shows a simple plan. As acute pain wanes weaker analgesics are used. If chronic pain increases stronger ones are necessary.

Figure 4.

Acute pain

The simple scheme for managing pain after trauma or surgery is shown in Figure 5.

Figure 5.

Low-tech: Intermittent opioid injection

Intermittent opioid injection can provide effective relief of acute pain (Gould et al, 1992). The main reason that adequate doses are withheld by doctors or nurses is the fear of respiratory depression. Opioids used for people who are not in pain, or in doses larger than necessary to control the pain, can slow and indeed stop breathing. The principle is that the dose has to be titrated to the effect (Figure 6). The effect is pain relief. If the dose given has not produced pain relief (the patient is still complaining of pain), and it has all been delivered and absorbed, then it is safe to give another dose. This subsequent dose may be smaller than the first. If it too doesn't succeed then the process can be repeated.

Figure 6.

There is no compelling evidence that one opioid is better than another, but there is good evidence that pethidine has a specific disadvantage (Szeto et al, 1977). Given in multiple doses its metabolite norpethidine can accumulate and act as a CNS irritant, ultimately causing convulsions. This is more likely when there is renal dysfunction. This toxic metabolite should preclude using pethidine when multiple injections are likely to be needed. The old idea that pethidine is better than other opioids at dealing with colicky pain is no longer tenable (Nagle & McQuay, 1990).

Morphine (and its family including heroin and codeine), has an active rather than a toxic metabolite, morphine-6-glucuronide. In renal dysfunction this metabolite can accumulate and result, because it is more active than morphine, in greater effect from a given dose. If you are, as you should be, titrating dose against effect, this will not matter. Less morphine will be needed. It can be a problem with unconscious patients being dosed by the clock, and whose renal function is compromised (Ball et al, 1985).

We think there are powerful risk management reasons for using one opioid only, so that everyone is familiar with dosage, effect and problems. We think that opioid should be morphine. The importance of good staff education was emphasised by the Cardiff audit, where implementing an algorithm for intermittent opioid dosing, together with staff education, had a powerful impact on pain relief and patient satisfaction (Gould et al, 1992). Also in favour of low-tech approaches is the increased risk of adverse events with high-tech approaches (Bates et al, 1995).

High-tech: Patient-Controlled Analgesia (PCA)

Intermittent opioid injection requires good staffing levels if delays between need and injection are to be minimised. One way round this logistic problem is to use PCA. A syringe driver containing opioid is connected to a button and to an intravenous line. The patient presses the button and an intravenous dose of opioid is given. This delivers opioid to the same opioid receptors as the intermittent injection, but circumvents any delays. Perhaps not surprisingly there is little difference in outcome between intermittent injection and PCA (Ballantyne et al, 1993). Good risk management with PCA should emphasise the same drug and the same machinery across the clinical service.

High-tech: Epidural Infusion

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. These two drug classes have a synergistic rather than simply additive effect when used together, so that lower doses of each are required for equivalent analgesia with fewer adverse effects (McQuay, 1994).

The risks are those of an epidural, (infection, haematoma, abscess), those of the local anaesthetic (cardiac and CNS) and those of the opioid, ('late' respiratory depression). Wrong doses do get given (Bates et al, 1995). There is an increased need for urinary catheterisation. The risk of persistent neurological sequelae after an epidural is about 1 in 5000 (Kane, 1981). Debate continues about whether patients with epidural infusions need to be nursed in a high dependency environment.

Minor surgery (and later after major surgery)

There is an old adage that if patients can swallow it is best to take drugs by mouth. This is very pertinent for pain relief after surgery or trauma. Effective relief can be achieved with oral non-steroidal anti-inflammatory drugs (NSAIDs), and there is no sufficient evidence that NSAIDs by injection perform better than the same drug at the same dose given by mouth (Williams et al, 1996).

The Figure below is the current view of an evolving league table for analgesic performance after surgery. The vertical axis is the number-needed-to-treat (NNT) for patients to achieve at least 50% relief of pain.

Figure 7.

Ibuprofen 400 mg (and other NSAIDs analysed) will produce at least 50% relief of pain for one out of two postoperative patients, paracetamol 1 g for one out of four patients. As more data is added the league table should guide our prescribing for the postoperative period.

The clear guideline is that the first choice is an oral NSAID. Gastric bleeding is more a problem of chronic use. Acute renal failure can be precipitated in those with pre-existing heart or kidney disease, those on loop diuretics or those who have lost more than 10% of blood volume.

Local anaesthetic blocks

There is a necessary distinction between blocks done to permit surgery, and blocks done together with a general anaesthetic to provide postoperative pain relief. There is clear evidence that blocks can indeed provide good relief in the initial postoperative period (McQuay et al, 1988), and no evidence to suggest that patients with blocks then experience rebound, and need more postoperative pain relief. Details of the blocks relevant to orthopaedics and trauma can be found elsewhere (Edmonds-Seal et al, 1980; Cousins, & Bridenbaugh, 1988; Loach, 1994). There is strong interest in whether or not there are opioid receptors lurking in the periphery, and whether or not injecting opioid, for instance into the knee joint, can produce effective relief (Stein et al, 1991; Kalso et al, 1996). The risk of neurological damage is the major drawback (Bridenbaugh, 1988). To minimise this risk blocks should not be done on anaesthetised patients.

Acute Pain Services

One remedy for poor management is the provision of an acute pain service (Report of RCS, 1990). There is dispute about what range of services such services should provide, from a full 'menu' including the high-tech options above (Ready et al, 1988) to a service which aims to educate and implement good guidelines for low-tech approaches (Gould et al, 1992; Rawal & Berggren, 1994). The service can provide a focus for education and a referral service.

Chronic pain

The same analgesics, from NSAID through to opioid, are used in chronic as in acute pain. If analgesics relieve the pain to an adequate extent, and with tolerable or controllable adverse effects, then there is little reason to use other interventions. If analgesics are ineffective other methods have to be considered. If analgesics are effective but cause intolerable or uncontrollable adverse effects then again other methods should be considered. The effectiveness and the adverse effects of the analgesics are critical.

We know from work with cancer pain that using analgesics according to the WHO ladder (Figure 4) can relieve pain for 80% of patients. For most of the 80% the relief will be good, for a minority it will only be moderate. This presumes that the pain is managed optimally, and we know from audit is often not the case. Optimal management requires that the correct drugs are available, and that they are given in the correct dose by the correct route and at the correct time. This needs staff who are well versed in the problems, and who are available to care for the patient. The second problem is the 20% of patients whose pain is not well managed by intelligent use of analgesic guidelines. The other treatment methods on Figure 3 are necessary to manage those for whom analgesics fail.

Non-opioid analgesics

Oral NSAIDs, combinations and others

Choosing the best analgesic for long term use involves the same decisions as in acute pain. Most comparisons are done using single doses, whereas patients with chronic pain take multiple doses. Historically the efficacy league table (Figure 7) for acute pain has also proved valid in chronic pain. Despite the fact that the drugs in this category are by far the most widely used, there is remarkably little good evidence about their relative efficacy and adverse effects.

No single-dose trial has shown any efficacy advantage of one NSAID over another (Gøtzsche, 1989). This does not fit well with patients' reports on multiple dosing of increased efficacy from NSAIDs with greater anti-inflammatory action. The adverse effect profile may also be different with chronic dosing. The risk of NSAID induced gastric bleeding is lowest with ibuprofen, and increases with increasing age (Henry et al, 1996). Prophylactic misoprostol should be considered for preventing NSAID associated gastrointestinal complications when age is greater than 75 years, cardiovascular disease, history of peptic ulcer or of gastrointestinal bleeding (NNTs to prevent one serious GI complication in one year 105, 58, 11 and 7 respectively) (Silverstein et al, 1995; Shield and Morant, 1996).

The efficacy dose-response curve for NSAIDs is flat compared with the dose-response for adverse effects such as gastrointestinal symptoms, dizziness, and drowsiness (Eisenberg et al, 1994). Increasing the dose to improve analgesia is therefore more likely to increase adverse effects than to improve analgesia.

Centrally-acting non-opioids include dipyrone and nefopam. Dipyrone is widely used in certain countries, and is an effective analgesic which can produce blood dyscrasias. The lack of comparative evidence makes it hard to rank its risk:benefit profile against the other analgesics.

Comparisons

NSAIDs alone produced as good analgesia as single or multiple doses of weak opioids alone or in combination with nonopioid analgesics (Eisenberg et al, 1994). Adverse effect incidence and patient dropout rates were the same for multiple doses of NSAIDs or weak opioids in combination with nonopioid analgesics (Eisenberg et al, 1994). Conversely two studies suggest that there is little advantage in osteoarthritis of either NSAIDs over paracetamol (March et al, 1994) or weak opioids in combination with paracetamol over paracetamol alone (Kjærsgaard-Andersen et al, 1990).

Topical NSAIDs

Many doctors and some pundits (Anon, 1994) are sceptical about the efficacy of topical NSAIDs. This may not be correct.

Published RCTs on chronic pain conditions (mainly knee osteoarthritis) studied over 800 subjects treated with topical NSAIDs and 322 subjects who received placebo. The analgesic response for combined placebo treatment was 30%, and for combined topical non-steroidal anti-inflammatory preparations it was 63%. For analgesic effects the odds ratio was 3.6 (2.6 - 4.8) and the number-needed-to-treat was 3.2 (2.6 - 4.1) (Moore et al, 1996a).

Opioids

In chronic pain there are two particular problems with opioids (McQuay, 1989). The first is that adequate doses are often not available or are not given, primarily because of fears of addiction. The second is that some (rarer) chronic pain states, particularly when the nervous system is damaged, may not respond fully to opiates.

Opioids used for people who are not in pain can induce physical and psychological dependence. This does not happen to patients who receive them for pain relief, for instance after an operation or for severe pain from osteoporotic vertebral collapse. Some governments restrict medical availability on the grounds that if the drugs are available medically this will worsen the street addiction problem. There is no evidence for this. The casualties are patients who are deprived of adequate pain relief.

In chronic pain opioids are usually given by mouth. The dose is worked out by titration over a period of days, and then the drug is given regularly, not waiting for the pain to come back. Initial problems with nausea or dizziness commonly settle. If constipation is likely laxatives are given.

Patients who cannot swallow can try sublingual, transdermal or suppository dosing. Subcutaneous infusion, usually from a small (external) pump is used for terminal patients who cannot manage these other routes. Rarely the epidural route is used for combination infusion of opioid and local anaesthetic.

If patients' pain starts to increase the dose is increased. If sensible dose increases do not produce pain relief, or if increasing the opioid dose provokes intolerable or unmanageable adverse effects, then other methods have to be considered, either as well as the opioid or instead of it. A working rule is that if the pain is in a numb area, which is a marker for a damaged nervous system, we would be less confident that opiates would necessarily produce pain relief (Jadad et al, 1992), and our threshold for using other strategies would be lower.

Unconventional analgesics

Unconventional analgesics (McQuay, 1988) are drugs which have other indications in other medical settings, and are not normally thought of as analgesics. Treating chronic pain in a tertiary hospital setting we use these drugs for about one third of our patients. The hall-mark is pain in a numb area, neuropathic pain.

When the patient has symptoms and signs of nervous system damage in the area of their pain we expect the response to conventional analgesics to be reduced. Conventional analgesics have often failed already, which is why the patient has been referred. If not, we try them, before we embark on empirical testing to see if any of the unconventional analgesics can provide relief.

Antidepressants

Antidepressants work on the nervous system to relieve depression. We use them in much lower dosage (about half) to relieve pain. Classically they were used to relieve pain that was burning rather than shooting in character, and anticonvulsants were used for shooting pains. Now we tend to use antidepressants as first line for both types of pain, because we have greater success and because we believe (but see Table 1) the antidepressants cause fewer adverse effects.

We use low doses (median 75 mg amitriptyline nocte, maximum 150 mg) compared with those used to control depression. The pain relieving effect happens, if it is going to happen, well within a week, whereas 10 days is the minimum often quoted for an antidepressant effect. The older (tricyclic) antidepressants seem to be better than the selective serotonin reuptake inhibitors as analgesics. The simplest analogy is that these older drugs are like shotguns, acting on multiple transmitter pathways, whereas the newer ones are more like rifles, designed as they are to be more selective and affect only one pathway.

Table1. Number-needed-to-treat (NNT) for some analgesic interventions
Condition       Intervention        Outcome             NNT    Reference         
postoperative   (good) ibuprofen    >50% pain relief   2       (Moore et al,     
pain            400 mg                                         1996b)            
                paracetamol 1 g     >50% pain relief   4       (Moore et al,     
                                                               1996b)            
                (poor) codeine 60   >50% pain relief   >10     (Moore et al,     
                mg oral                                        1996c)            
back pain       epidural steroid    >75% relief at     >6      (Watts &        
 
                                    60 days                    Silagy, 1995;     
                                                               McQuay & Moore,   
                                                               1996)             
acute sprains   topical NSAID       >50% pain relief   2+      (Moore et al,     
etc.            (good)                                         1996a)            
trigeminal      anticonvulsants     >50% pain relief   2.5     (McQuay et al,    
neuralgia                                                      1995)             
diabetic        anticonvulsants     >50% pain relief   2.5     (McQuay et al,    
neuropathy                                                     1995)             
diabetic        topical capsaicin   >50% pain relief   4.2     (Zhang et al,     
neuropathy                                                     1994)             
neuropathic     antidepressants     >50% pain relief   2.5     (McQuay et al,    
pain                                                           1996)             

Anticonvulsants

Anticonvulsants have been used for many years to treat the shooting pains of trigeminal neuralgia and of diabetic neuropathy. How they work has never been clear. The catch-all explanation was that they stabilised nerve membranes, preventing them carrying spurious messages. The current fashionable explanation is that these drugs acting as antagonists on the N-methyl D-aspartate (NMDA) mechanism.

Anticonvulsants can provide good relief in neuropathic pain (NNT of 2-3, Table 1) (McQuay et al, 1995). Doses required for analgesic effect are close to the anticonvulsant dosing range, and carry a perceived burden of adverse effects. The systematic reviews suggest that there is little difference between the antidepressant and anticonvulsant adverse effects. Our main use is the traditional role in trigeminal neuralgia and diabetic neuropathy. We also use anticonvulsants in shooting pain which does not respond to antidepressants. Two examples are phantom limb pain and pain in the head and neck due to tumour.

Others

Clonidine and other alpha-2 adrenergic agonists have analgesic effects, both in conventional pain and in neuropathic pain (McQuay, 1992). They extend the duration of local anaesthetic effect and have a synergistic effect with opioids. Their clinical utility is limited by the adverse effects of sedation and hypotension. In neuropathic pain single doses of clonidine were effective in postherpetic neuralgia (Max et al, 1988), and in cancer pain (Eisenach et al, 1995) . Baclofen is used by intrathecal pump to treat the painful spasms of cerebral palsy. Ketamine and dextromethorphan, both drugs with NMDA antagonist action, are being used in severe neuropathic pain.

Block Nerve Transmission


Reversible
Local Anaesthetics
Local anaesthetics block nerve conduction reversibly. When the local anaesthetic wears off the pain returns. That is the pharmacologically correct statement, but another old saying, that a series of local anaesthetic blocks can be used to "break the cycle" of pain and effect a cure, now has some empirical support (Arner et al, 1990), even if we do not understand the mechanism. Arner and colleagues showed that the duration of pain relief could far outlast the duration of local anaesthetic action, and that prolonged relief could result from a series of blocks (Arner et al, 1990). Local anaesthetic blocks can thus be diagnostic and therapeutic. Diagnosis of pain for instance from a 'trapped' lateral cutaneous nerve of thigh can be confirmed by local anaesthetic block, and a series of blocks may prevent pain recurring.

Pain clinics use such blocks commonly, for shoulder pain (suprascapular nerve block (Emery et al, 1989; van der Heijden, 1996)), for intercostal neuralgia, for rectus sheath nerve entrapment, postoperative scar pains and other peripheral neuralgias (Table 2). What is not clear is the extent to which adding steroid to the local anaesthetic makes a difference, either prolonging the duration of effect of a particular procedure or increasing the chance of success of a series of blocks.

Table 2. Common nerve blocks
BLOCK                                           COMMON INDICATIONS                     
Trigger point                                   focal pain (e.g. in muscle)            
Peripheral:              intercostal            pain in dermatomal distribution        
                         sacral nerves                                                 
                         rectus sheath                                                 
Extradural                                      Uni or bilateral pain (lumbosacral,    
                                                cervical, thoracic etc.) Midline       
                                                perineal pain                          
Intrathecal                                     Unilateral pain (neurolytic            
                                                injection for pain due to              
                                                malignancy, limbs, chest etc.)         
                                                (Midline perineal pain)                
Autonomic                                                                              
Intravenous regional                            reflex sympathetic dystrophy           
sympathectomy                                                                          
Stellate ganglion                               reflex sympathetic dystrophy           
                                                arm pain                               
                                                brachial plexus nerve compression      
Lumbar sympathetic                              reflex sympathetic dystrophy           
                                                lumbosacral plexus nerve compression   
                                                vascular insufficiency lower limb      
                                                perineal pain                          
Coeliac plexus                                  abdominal pain                         

Fibromyalgia

Similar injections are done for the trigger points of fibromyalgia, but there do not appear to be any controlled comparisons of injections with other treatments.

Intravenous regional sympathectomy

Intravenous regional sympathetic blocks (IRSBs) are used widely in patients with reflex sympathetic dystrophy (RSD). A systematic review of 7 RCTs of IRSBs found that none of the 4 guanethidine trials showed significant analgesic effect. Two reports, one using ketanserin and one bretylium with 17 patients in total, showed some advantage of IRSBs over control (Jadad et al, 1995). Adding guanethidine in IRSBs does not appear to be more effective than local anaesthetic alone.

Epidural steroids and facet joint blocks

Two other common (for back pain) pain clinic procedures are epidural steroid injection and facet nerve blocks. Epidural steroids in back pain have been studied in two systematic reviews (Watts & Silagy, 1995; Koes et al, 1995). Overall the combined data showed statistically significant (odds ratios) improvement for both short-term (1 to 60 days) and long-term (12 weeks up to 1 year). The clinical significance is that the NNT for short-term (1 to 60 days) greater than 75% pain relief from the ten trials with short-term outcomes combined, was just under 6, with 95% confidence intervals from 4 to 12 (McQuay & Moore, 1996). This means that for 6 patients treated with epidural steroid one will obtain more than 75% pain relief short-term.

The NNT for long-term (12 weeks up to 1 year) improvement from the five trials combined, was about 11, with 95% confidence intervals from 6 to 90. This means that for 11 patients treated with epidural steroid one will obtain more pain relief over this longer term period. There is still the interesting question of whether local anaesthetic alone could achieve these results (breaking the cycle), or whether the steroid is an essential component.

Better results are believed to be achieved the earlier the patient is treated (Benzon, 1986), and in patients who have not had back surgery. It may take as long as a week for benefit from the steroid to be felt, so that it is unwise to dismiss the injection as a failure after one hour. If the injection produced incomplete or short-lived relief, then it is worth repeating, and a course of three injections is recommended. No additional benefit accrued from more than three injections (Benzon, 1986).

There is considerable current controversy about the potential for epidural steroid to produce long-term neurological sequelae. Intrathecal injection of steroid can produce neurological sequelae. It is therefore important that intrathecal injection is avoided.

Classically facet joint injection with local anaesthetic and steroid is indicated when pain is worse when sitting, and pain is provoked by lateral rotation and spine extension. Recent studies suggest that whether or not the injection is actually in the facet joint makes little difference (Lilius et al, 1989), and indeed cast some doubt on long-term utility (Carette et al, 1991). Short-lived success (less than six weeks) with local anaesthetic and steroid is said to be improved by use of cryoanalgesia or radiofrequency blocks to the nerves to the joints.

Irreversible

The destructive procedures are aimed at cutting, burning or damaging (Table 2) the nerve fibres carrying the pain signals. The flaw in the logic is that the nervous system can all too often rewire, finding a way around the lesion. If that happens, and the pain returns, then it may be even more difficult to manage - severe neuropathic pain can result. In general neurolytic blocks in non-malignant pain are not recommended, because they do not last forever, and recurrent pain may be more difficult to manage, and because of the morbidity. In cancer pain these neurolytic block procedures do have a place, when there is a short (<3 m) prognosis, or where alternatives such as pains-taking drug control or long-term epidural infusion are not possible. Similar distinction between cancer and non-cancer pain holds for coeliac plexus block in pancreatic pain. Pain associated with pancreatic cancer responds well to coeliac plexus block (Eisenberg et al, 1995), and it may also help those with abdominal or perineal pain from tumour in the pelvis. In chronic pancreatitis results are much less convincing.

The limitation is the potential for motor and sphincter damage. This risk is higher with bilateral and repeat procedures, and higher the lower the cord level of the block. Extradural neurolytics have limited efficacy. While claims have been made that the paravertebral approach is preferable, patchy results may be attributed to unpredictable injectate spread. Our results of spinal infusion of a combination of local anaesthetic and opioid are superior to neurolytic blocks, providing good analgesia with minimal irreversible morbidity.

Surgery

The relevant neurosurgical interventions for orthopaedic pain include dorsal column stimulation, rhizotomy, cordotomy and dorsal root entry zone (DREZ) lesions. The indications are usually non-malignant neuropathic pain which has failed to respond to pharmacological measures. The difficulties of trials of uncommon surgical procedures are well known. These procedures are usually documented by glowing case series. Longer term outcomes may not be so good (Abram, 1993).

Alternatives

TENS, acupuncture

The rationale for transcutaneous nerve stimulation (TENS) is the gate theory (Melzack and Wall, 1965). If the spinal cord is bombarded with impulses from the TENS machine then it is distracted from transmitting the pathological pain signal. We know from systematic reviews that TENS has limited efficacy in acute (and postoperative) pain, and also in labour pain. As yet there is no systematic review which gathers together the many trials in chronic pain.

What we do know is that attention to detail makes considerable difference to TENS efficacy in chronic pain (Johnson et al, 1992). Patients need to be told that it is useless expecting success unless the machine is connected for at least an hour at a time. They need to be told where to put the electrodes, how to put them on, how to manipulate the stimulus to best advantage, and indeed to turn the machine on.

Three systematic reviews discuss acupuncture in chronic non-malignant pain, (Patel et al, 1989; ter Riet et al, 1990; Bhatt Sanders, 1985). These show an effect, but that effect in clinical practice is often short-lived (3 days), and is therefore expensive in time. It is difficult to know what is the real place of acupuncture, like other complementary interventions, because of the lack of trials comparing complementary with mainstream procedures. In this context see this systematic review (Puett & Griffin, 1994).

Physio and variants

Pain clinics keep a very open mind about other interventions. If patients benefit from alternatives we are only too pleased. The evidence from back pain however suggests that on rigorous outcome measures physiotherapy and other forms of manipulation have but limited success. Such analyses often did not include any measure of quality of life. If they make the patient feel better and they are cheap then it is a decision for the third party payer whether or not these physiotherapy manoeuvres should be offered. Systematic reviews include (Abenhaim & Bergeron, 1992; Anderson et al, 1992; Assendelft et al, 1992; Brunarski, 1984; Koes et al, 1991; Ottenbacher and DiFabio, 1985; Powell et al, 1993; Shekelle et al, 1992.

Behavioural Management

Back schools through to behavioural management programmes offer a range of help for patients, help to cope with their (usually back) pain problems. Making decisions about the benefits of psychologically-based treatments of medical problems is not easy, and especially difficult to compare with other treatments and to measure relative benefit and cost. Patients whose pain has proved intractable to all reasonable medical and other interventions are chronic consumers of health care - GP or hospital clinic time, analgesic and psychotropic drugs, repeated admissions and sometimes surgery. If rehabilitation treatment enables these patients to carry on more satisfying lives with minimum medical help, how can it be most effectively and economically offered?

Randomised comparison of the St Thomas' four-week inpatient treatment with eight-week half-day outpatient treatment, with fitness training, planned increases in activity, activity scheduling, drug reduction, relaxation and cognitive therapy as the pain management methods taught by the same staff team (Pither & Nicholas, 1991) showed that for every three patients treated as inpatients rather than outpatients, one patient fewer was taking analgesic or psychotropic drugs. For every four patients treated as inpatients rather than outpatients, one patient fewer sought additional medical advice in the year after treatment. For every five patients treated as inpatients rather than outpatients, one patient more had a ten-minute walking distance improved by more than 50%. For every six patients treated as inpatients rather than outpatients, one patient fewer was depressed (Bandolier, 1995).

Systematic reviews in this area include (Cohen et al, 1994; Cutler et al, 1994; Fernandez & Turk, 1989; Gebhardt, 1994; Hyman et al, 1989; Malone et al, 1988; Mullen et al, 1987; Suls and Fletcher, 1985).

References

Abenhaim, L. and Bergeron, A.M. (1992). Twenty years of randomized clinical trials of manipulative therapy for back pain: a review. Clin Invest Med, 15, 527-35.

Abram, S.E. (1993). 1992 Bonica Lecture. Advances in chronic pain management since gate control. Reg Anesth , 18 , 66-81.

Anderson, R., Meeker, W.C., Wirick, B.E., Mootz, R.D., Kirk, D.H. and Adams, A. (1992). A meta-analysis of clinical trials of spinal manipulation. J Manipulative Physiol Ther , 15 , 181-94.

Anonymous (1994). Rational use of NSAIDs for musculoskeletal disorders. Drug and Therapeutics Bulletin , 32 , 91-5.

Arner, A., Lindblom, U., Meyerson, B.A. and Molander, C. (1990). Prolonged relief of neuralgia after regional anesthetic blocks. A call for further experimental and systematic clinical studies. Pain , 43 , 287-297.

Assendelft, W.J., Koes, B.W., Van der Heijden, G.J. and Bouter, L.M. (1992). The efficacy of chiropractic manipulation for back pain: blinded review of relevant randomized clinical trials. J Manipulative Physiol Ther , 15 , 487-94.

Ball, M., McQuay, H.J., Moore, R.A., Allen, M.C., Fisher, A. and Sear, J. (1985). Renal failure and the use of morphine in intensive care. Lancet , 1 , 784-6.

Ballantyne, J.C., Carr, D.B., Chalmers, T.C., Dear, K.B.G., Angelillo, I.F. and Mosteller, F. (1993). Postoperative patient-controlled analgesia: meta-analyses of initial randomised controlled trials. J Clin Anesth , 5 , 182-193.

Bandolier (1995). http://www.ebandolier.com; issue 22.

Bates, D.W., Cullen, D.J., Laird, N., Petersen, L.A., Small, S.D., Servi, D., Laffel, G., Sweitzer, B.J., Shea, B.F., Hallisey, R., Vliet, M.V., Nemeskal, R. and Leape, L.L. (1995). Incidence of adverse drug events and potential adverse drug events. JAMA , 274 , 29-34.

Benzon, H.T. (1986). Epidural steroid injections for low back pain and lumbosacral radiculopathy. Pain , 24 , 277-295.

Bhatt Sanders, D. (1985). Acupuncture for rheumatoid arthritis: an analysis of the literature. Semin Arthritis Rheum , 14 , 225-31.

Bridenbaugh, P.O. (1988). Complications of local anesthetic neural blockade. In Neural Blockade (eds. M.J. Cousins and P.O. Bridenbaugh), pp. 695-717. Lippincott, Philadelphia.

Brunarski, D.J. (1984). Clinical trials of spinal manipulation: a critical appraisal and review of the literature. J Manipulative Physiol Ther , 7 , 243-9.

Burford Nursing Development Unit (1984). Nurses & Pain. Nursing Times , 18 , 94.

Carette, S., Marcoux, S., Truchon, R., Grondin, C., Gagnon, J., Allard, Y. and Latulippe, M. (1991). A controlled trial of corticosteroid injections into facet joints for chronic low back pain. New England Journal of Medicine , 325 , 1002-1007.

Cohen, J.E., Goel, V., Frank, J.W., Bombardier, C., Peloso, P. and Guillemin, F. (1994). Group education interventions for people with low back pain. An overview of the literature. Spine , 19 , 1214-22.

Cousins, M.J. and Bridenbaugh, P.O. (1988). Neural Blockade . Lippincott, Philadelphia.

Cutler, R.B., Fishbain, D.A., Rosomoff, H.L., Abdel-Moty, E., Khalil, T.M. and Rosomoff, R.S. (1994). Does nonsurgical pain center treatment of chronic pain return patients to work? A review and meta-analysis of the literature. Spine , 19 , 643-52.

Edmonds-Seal, J., Paterson, G.M.C. and Loach, A.B. (1980). Local nerve blocks for postoperative analgesia. Journal of the Royal Society of Medicine , 73 , 111-114.

Eisenach, J.C., DuPen, S., Dubois, M., Miguel, R. and Allin, D. (1995). Epidural clonidine analgesia for intractable cancer pain. The Epidural Clonidine Study Group. Pain , 61 , 391-9.

Eisenberg, E., Berkey, C.S., Carr, D.B., Mosteller, F. and Chalmers, T.C. (1994). Efficacy and safety of nonsteroidal antiinflammatory drugs for cancer pain: a meta-analysis. J Clin Oncol , 12 , 2756-65.

Eisenberg, E., Carr, D.B. and Chalmers, T.C. (1995). Neurolytic celiac plexus block for treatment of cancer pain: a meta-analysis. Anesth Analg , 80 , 290-5.

Emery, P., Bowman, S., Wedderburn, L. and Grahame, R. (1989). Suprascapular nerve block for chronic shoulder pain in rheumatoid arthritis. British Medical Journal , 299 , 1079-80.

Fernandez, E. and Turk, D.C. (1989). The utility of cognitive coping strategies for altering pain perception: a meta analysis. Pain , 38 , 123-35.

Gebhardt, W.A. (1994). Effectiveness of training to prevent job-related back pain: a meta-analysis. Br-J-Clin-Psychol , 33 , 571-4.

Gøtzsche, P.C. (1989). Patients' preference in indomethacin trials: an overview. Lancet , 1 , 88-91.

Gøtzsche, P.C. (1990). Sensitivity of Effect Variables in Rheumatoid Arthritis: A Meta-analysis of 130 Placebo Controlled NSAID Trials. Journal of Clinical Epidemiology , 43 , 1313-1318.

Gould, T.H., Crosby, D.L., Harmer, M., Lloyd, S.M., Lunn, J.N., Rees, G.A.D., Roberts, D.E. and Webster, J.A. (1992). Policy for controlling pain after surgery: effect of sequential changes in management. British Medical Journal , 305 , 1187-1193.

Henry, D., Lim, L.L-Y, Rodriguez, L.A.G., Gutthann, S.P., Carson, J.L., Griffin, M., Savage, R., Logan, R., Moride, Y., Hawkey, C., Hill, S. and Fries, J.T. (1996). Variability in risk of gastrointestinal complications with individual non-steroidal anti-inflammatory drugs: results of a collaborative meta-analysis. British Medical Journal , 312 , 1563-1566.

Houde, R.W., Wallenstein, S.L. and Rogers, A. (1960). Clinical pharmacology of analgesics: a method of assaying analgesic effect. Clinical Pharmacology and Therapeutics , 1 , 163-174.

Hyman, R.B., Feldman, H.R., Harris, R.B., Levin, R.F. and Malloy, G.B. (1989). The effects of relaxation training on clinical symptoms: a meta analysis. Nurs Res , 38 , 216-20.

Jadad, A.R., Carroll, D., Glynn, C.J., Moore, R.A. and McQuay, H.J. (1992). Morphine responsiveness of chronic pain: double-blind randomised crossover study with patient-controlled analgesia. Lancet , 339 , 1367-71.

Jadad, A.R. and McQuay, H.J. (1994). Pain measurement. In Outcome Measures in Trauma (Eds. J. Fairbank, P. Pysent and A. Carr), pp. 17-24. Butterworth Heinemann, Oxford.

Jadad, A.R., Carroll, D., Glynn, C.J. and McQuay, H.J. (1995). Intravenous regional sympathetic blockade for pain relief in reflex sympathetic dystrophy: a systematic review and a randomized, double-blind crossover study. Journal of Pain and Symptom Management , 10 , 13-20.

Johnson, M.I., Ashton, C.H. and Thompson, J.W. (1992). Long term use of transcutaneous electrical nerve stimulation at Newcastle Pain Relief Clinic. Journal of the Royal Society of Medicine , 85 , 267-268.

Kalso, E., Tramer, M., Carroll, D., McQuay, H. and Moore, R.A. (1996). Pain relief from intra-articular morphine after knee surgery: A
qualitative systematic review. submitted ,

Kane, R.E. (1981). Neurologic deficits following epidural or spinal anesthesia. Anesthesia and Analgesia , 60 , 150-161.

Kjærsgaard-Andersen, P., Nafei, A., Skov, O., Madsen, F., Andersen, H.M., Krøner, K., Gjøderum, O., Pedersen, L. and Branebjerg, P.E. (1990). Codeine plus paracetamol versus paracetamol in longer-term treatment of chronic pain due to osteoarthritis of the hip. Pain , 43 , 309-318.

Koes, B.W., Assendelft, W.J., van der Heijden, G.J., Bouter, L.M. and Knipschild, P.G. (1991). Spinal manipulation and mobilisation for back and neck pain: a blinded review. British Medical Journal , 303 , 1298-303.

Koes, B.W., Scholten, R.P.M., Mens, J.M.A. and Bouter, L.M. (1995). Efficacy of epidural steroid injections for low-back pain and sciatica: a systematic review of randomized clinical trials. Pain , 63 , 279-88.

Lilius, G., Laasonen, E.M., Myllynen, P., Harilainen, A. and Grönlund, G. (1989). Lumbar facet joint syndrome. Journal of Bone and Joint Surgery , 71 , 681-684.

Loach, A., ed. (1994). Orthopaedic anaesthesia . Arnold, London.

Malone, M.D., Strube, M.J. and Scogin, F.R. (1988). Meta analysis of non medical treatments for chronic pain. Pain , 34 , 231-44.

March, L., Irwig, L., Schwarz, J., Simpson, J., Chock, C. and Brooks, P. (1994). N of 1 trials comparing a non-steroidal anti-inflammatatory drug with paracetamol in osteoarthritis. British Medical Journal , 309 , 1041-6.

Max, M.B., Schafer, S.C., Culnane, M., Dubner, R. and Gracely, R.H. (1988). Association of pain relief with drug side effects in postherpetic neuralgia: a single dose study of clonidine, codeine, ibuprofen, and placebo. Clinical Pharmacology and Therapeutics , 43 , 363-71.

McGrath, P.J., Ritchie, J.A. and Unruh, A.M. (1993). Paediatric pain. In Pain Management and Nursing Care (Eds. D. Carroll and D. Bowsher), pp. 100-123. Butterworth Heinemann, Oxford.

McQuay, H.J. (1988). Pharmacological treatment of neuralgic and neuropathic pain. Cancer Surveys , 7 , 141-59.

McQuay, H.J. (1989). Opioids in chronic pain. British Journal of Anaesthesia , 63 , 213-26.

McQuay, H.J. (1992). Is there a place for alpha2 adrenergic agonists in the control of pain? In Toward the use of alpha2 adrenergic agonists for the treatment of pain (Eds. J.M. Besson and G. Guilbaud), pp. 219-232. Elsevier, Amsterdam.

McQuay, H. (1994). Epidural Analgesics. In Textbook of Pain (Eds. P. Wall and R. Melzack), pp. 1025-1034. Churchill Livingstone, London.

McQuay, H.J. (1995). Preemptive analgesia: a systematic review of clinical studies. Ann Med , 27 , 249-58.

McQuay, H.J., Carroll, D. and Moore, R.A. (1988). Postoperative orthopaedic pain--the effect of opiate premedication and local anaesthetic blocks. Pain , 33 , 291-5.

McQuay, H.J. and Dickenson, A.H. (1990). Implications of nervous system plasticity for pain management. Anaesthesia , 45 , 101-2.

McQuay, H., Carroll, D., Jadad, A.R., Wiffen, P. and Moore, A. (1995). Anticonvulsant drugs for management of pain: a systematic review. British Medical Journal , 311 , 1047-52.

McQuay, H. and Moore, R.A. (1996). Epidural steroids (letter). Anaesthesia and Intensive Care , 24 , 284-6.

McQuay, H., Nye, B.A., Carroll, D., Wiffen, P.J., Tramèr, M. and Moore, R.A. (1996). A systematic review of antidepressants in neuropathic pain. Pain (in press) ,

Melzack, R. and Wall, P.D. (1965). Pain mechanisms: a new theory. Science , 150 , 971-978.

Moore, R.A., Nye, B.A., Carroll, D., Wiffen, P.J., Tramèr, M. and McQuay, H.J. (1996a). A systematic review of topically-applied non-steroidal anti-inflammatory drugs. submitted ,

Moore, R.A., McQuay, H.J. and Gavaghan, D.J. (1996b). Deriving dichotomous outcome measures from continuous data in randomised controlled trials of analgesics. Pain (in press) ,

Moore, R.A. and McQuay, H.J. (1996c). Single-patient data meta-analysis of 3,453 postoperative patients: oral tramadol versus placebo, codeine and combination analgesics. Pain (submitted) ,

Mullen, P.D., Laville, E.A., Biddle, A.K. and Lorig, K. (1987). Efficacy of psychoeducational interventions on pain, depression, and disability in people with arthritis: a meta analysis. J Rheumatol , 14 Supp , 33-9.

Nagle, C.J. and McQuay, H.J. (1990). Opiate receptors; their role in effect and side-effect. Current Anaesthesia & Critical Care , 1 , 247-252.

Onghena, P. and Van Houdenhove, B. (1992). Antidepressant-induced analgesia in chronic non-malignant pain: a meta-analysis of 39 placebo-controlled studies. Pain , 49 , 205-19.

Ottenbacher, K. and DiFabio, R.P. (1985). Efficacy of spinal manipulation/mobilization therapy. A meta analysis. Spine , 10 , 833-7.

Patel, M., Gutzwiller, F., Paccaud, F. and Marazzi, A. (1989). A meta-analysis of acupuncture for chronic pain. Int J Epidemiol , 18 , 900-6.

Pither, C.E. and Nicholas, M.K. (1991). Psychological approaches in chronic pain management. British Medical Journal , 47 , 743-61.

Powell, F.C., Hanigan, W.C. and Olivero, W.C. (1993). A risk/benefit analysis of spinal manipulation therapy for relief of lumbar or cervical pain. Neurosurgery , 33 , 73-8; discussion 78-9.

Puett, D.W. and Griffin, M.R. (1994). Published trials of nonmedicinal and noninvasive therapies for hip and knee osteoarthritis. Ann-Intern-Med , 121 , 133-40.

Rawal, N. and Berggren, L. (1994). Organization of acute pain services: a low-cost model. Pain , 57 , 117-23.

Ready, L.B., Oden, R., Chadwick, H.S., Benedetti, C., Rooke, G.A., Caplan, R. and Wild, L.M. (1988). Development of an Anaesthesiology based postoperative pain management service. Anesthesiology , 68 , 100-106.

Royal College of Surgeons of England, the College of Anaesthetists. (1990). Report of the working party on pain after surgery . Royal College of Surgeons, London.

Shekelle, P.G., Adams, A.H., Chassin, M.R., Hurwitz, E.L. and Brook, R.H. (1992). Spinal manipulation for low-back pain. Ann Intern Med , 117 , 590-8.

Shield, M.J. and Morant, S.V. (1996). Misoprostol in patients taking NSAIDs. British Medical Journal , 312 , 846.

Silverstein, F.E., Graham D. Y., Senior, J.R., Davies, H.W., Struthers, B.J., Bittman, R.M. and Geis, G.S. (1995). Misoprostol reduces serious gastrointestinal complications in patients with rheumatoid arthritis receiving nonsteroidal anti-inflammatory drugs. American College of Physicians , 123 , 241-249.

Stein, C., Comisel, K., Haimerl, E., Yassouridis, A., Lehrberger, K., Herz, A. and Peter, K. (1991). Analgesic effect of intraarticular morphine after arthroscopic surgery. New England Journal of Medicine , 325 , 1123-1126.

Suls, J. and Fletcher, B. (1985). The relative efficacy of avoidant and nonavoidant coping strategies: a meta analysis. Health Psychol , 4 , 249-88.

Szeto, H.H., Inturrisi, C.E., Houde, R., Saal, S., Cheigh, J. and Reidenberg, M. (1977). Accumulation of norperidine, an active metabolite of meperidine, in patients with renal failure or cancer. Annals of Internal Medicine , 86 , 738-741.

ter Riet, G., Kleijnen, J. and Knipschild, P. (1990). Acupuncture and chronic pain: a criteria-based meta-analysis. J Clin Epidemiol , 43 , 1191-9.

van der Heijden, C.J.M.G., van der Windt, D.A.W.M., Kleijnen, J., Koes, B.W. and Bouter, L.M. (1996). Steroid injections for shoulder disorders: a systematic review of randomized clinical trials. British Journal of General Practice , 46 , 309-16.

Watts, R.W. and Silagy, C.A. (1995). A meta-analysis on the efficacy of epidural corticosteroids in the treatment of sciatica. Anaesthesia and Intensive Care , 23 , 564-569.

Williams, J., Tramèr, M., Carroll, D., Wiffen, P.J., McQuay, H.J. and Moore, R.A. (1996). Keep taking the tablets! A systematic review of direct comparisons of non-steroidal anti-inflammatory drugs given by different routes for acute pain. submitted.

Zhang, W.Y. and Li, WanPoA. (1994). The effectiveness of topically applied capsaicin. A meta-analysis. Eur J Clin Pharmacol , 46 , 517-22.