Post-surgical patients referred for pain management may be those whose pain is proving difficult to manage, and those whose pain has resisted conventional treatment or for whom surgery is inappropriate. In both circumstances certain features of the history and examination may be vital to improve management.

There are few objective signs that the doctor can use to judge the severity of reported pain. Pain is necessarily subjective. Many patients have no obvious (visible) handicap. The most important principle is that both patients and doctor are best served if the doctor believes the patient's report. Their problems may be ill-understood, even disbelieved, at work and at home. Chronic pain changes people, affecting their personal and working lives, and ultimately their personalities. Such changes are often reversible with successful treatment. Much time and energy is wasted on procedures designed to ‘catch the patient out’. Labelling patients as malingerers, or calling the pain psychogenic may be easier than admitting that there is no successful treatment. The differential diagnosis for many pain conditions includes both malignant and non-malignant causes. Many pain treatments have no controlled evidence for either efficacy or morbidity. The management suggested here is conservative.

Continued painful stimuli alter the nervous system. Pain which had a peripheral origin may become ‘central’. An example is phantom limb pain; the peripheral neural basis for the pain no longer exists, yet the pain is felt. This concept of nervous system plasticity affects management in several ways. Conventional analgesics (Table 1) 702 may be relatively ineffective in the management of established chronic pain, because the neural basis of the pain has changed and it no longer responds to those remedies. Plasticity also implies that attacking the cables which carry the (original) pain message may be ineffective in the short term if those cables are no longer the source of the message, and ineffective in the long term if the system can ‘re-wire’. These ideas are supported by both experimental work and clinical observation. Pre-emptive strategies, such as blocking pain before amputation to reduce phantom limb pain incidence, and the idea of modulating the painful input, for instance by sustained local anaesthetic block as an alternative to irreversible attack on the cables, are being used increasingly.

The extra emphasis of a pain history compared with a ‘normal’ medical history is summarized in the series of questions shown in Table 2 703. Asking whether sensation is normal in the painful area and whether the pain is shooting or stabbing in character may help to identify the pains variously known as dysaesthetic, deafferentation, or neuropathic. It is important to distinguish these because they are unlikely to respond to conventional analgesics. These pains often have little pattern, but are less troublesome when the patient is distracted.

It is important to enquire specifically about the efficacy of each particular drug class (major and minor analgesics, anticonvulsants etc., Table 1 702). This information prevents the inept prescription of drugs which had failed previously, and may give important clues as to the kind of pain and its sensitivity to different classes of drug.

It is also important to know the dose size, frequency, duration of prescription, and the adverse effect problems for each drug which has been prescribed. Dose - response relationships apply to analgesic management, and therapeutic failure should not be presumed if the dosage was inadequate; a good example is the use of carbamazepine in trigeminal neuralgia. Is the patient taking drugs other than analgesics? Anticoagulation, for instance, is not only an (almost) absolute contraindication to pain management by injection procedures, but also interacts with some anti-inflammatory drugs.

It is important to be sure whether nerve blocks used previously were technically effective (e.g. did the patient have any numbness after extradural which included local anaesthetic?), before dismissing them as of no help to this patient. Other measures, such as transcutaneous nerve stimulation, may not have been used correctly, and may succeed if the patient receives proper instruction in their use.

Some index of function is necessary as a baseline so that improvement or deterioration may be monitored. Useful clinical outcome measures are notoriously difficult; using simple pain charts and indices of activity can work well.

The major options are drug treatment, nerve blocks, and alternatives such as transcutaneous nerve stimulation and acupuncture.

Drugs are the mainstay of chronic pain treatment. The simplest classification is conventional analgesics, from aspirin through to morphine, and unconventional analgesics, antidepressants and anticonvulsants (Table 1) 702. Those used to treating acute pain rightly assume that most such pains can be managed with the conventional analgesics. One of the many differences between acute and chronic pain is that chronic pain may not be so straightforward. About one-third of patients with both malignant and non-malignant disease have pains which respond poorly, if at all, to such conventional analgesics, major or minor. There is nothing novel in this; carbamazepine is a classic ‘unconventional’ analgesic used successfully in the management of trigeminal neuralgia. A simple rule of thumb is that pains in numb areas (deafferentation e.g. phantom limb pain, nerve compression, postherpetic neuralgia, and many abdominal pains) are unlikely to respond well to opiates. They may make the patient feel better, but analgesic effect will be poor. This distinction must be sought in the history, from the character of the pain and from previous response to conventional analgesics. Is this a pain in a numb area? How has it responded to adequate doses of conventional analgesics?

Within the conventional analgesic group, non-steroidal anti-inflammatory drugs at maximum recommended dose produce about 30 per cent more analgesia than 600 mg aspirin or paracetamol. The combination of opiate and non-opiate produces additive analgesia; the two mechanisms may combine to give better pain control in contexts such as bony pain. Within the class of non-steroidal anti-inflammatory agents it is necessary to find the best drug and the best analgesia with minimal adverse effects for each individual patient.

The use of major opiates for the treatment of non-malignant pain is controversial, because of fears of patients becoming addicts. This is extremely unlikely, but there is no point prescribing drugs which the patient or others involved in their care do not appreciate. A simple rule is that where the pain is sensitive to opiates, and there is no other effective remedy, opiate prescription may be tried, but only with the agreement of all concerned. The opiate would be administered by a route other than injection, unless impracticable.

For the treatment of pain insensitive to opiates, the choice between antidepressants and anticonvulsants is empirical. For burning pains low-dose antidepressants are the first choice, preferably a non-selective tricyclic such as amitriptyline, at a dose of 25 mg at night titrated over weeks to 75 mg at night. For shooting pains anticonvulsants work best, and first choice is sodium valproate (200 mg twice daily or 500 mg at night).

Nerve blocks can be thought of in two categories, reversible, as with local anaesthetic or cryoanalgesia, and irreversible, as with ablative nerve blocks and surgical or radiofrequency procedures. When the effect of reversible procedures wanes, the pain is unchanged. If pain recurs after an irreversible procedure it may have altered character, as happens in anaesthesia dolorosa.

The common nerve block procedures are summarized in Table 3 704. The major indication for nerve blocking techniques is lack of response to pharmacological management and/or unacceptable adverse effects. Many of these blocks may be done ‘diagnostically’ with local anaesthetic, as a preliminary to making a more permanent block as with cryoanalgesia, radiofrequency lesions, phenol, or surgery. Use of steroids may convert the diagnostic block to a therapeutic block, as with extradurals for the relief of low back pain. Repeated blocks with local anaesthetic may in themselves be therapeutic, as is seen when repeated stellate ganglion blocks are administered for causalgia.

Many of these procedures are performed on an out patient basis: the time between the block and when the patient goes home should be sufficient for any complications (most of which are ‘early’, such as pneumothorax developing after intercostal or stellate blocks) to be apparent before the patient leaves. Two complications present particular problems. Hypotension occurs during the first few hours after coeliac plexus block, so that patients may need to be admitted for the procedure. Chemical lumbar sympathectomy may cause a troublesome neuralgic pain, often in the groin or on the anterior aspect of the thigh. Although this may be severe, occurring in up to 10 per cent of patients, it is self-limiting (6 - 8 weeks), and transcutaneous nerve stimulation may give very effective relief.

The place of many of these blocks in pain management is ill-defined because, despite widespread use of the techniques, careful studies of quality and duration of relief and morbidity compared with other methods have not been done. Those expert in the use of these techniques have no doubt of their value. Improvement in pain control generally, with better use of drugs, has reduced the numbers of blocks performed.

Using intercostal neuralgia as an example, diagnostic intercostal blocks with local anaesthetic would confirm that the pain could be controlled by blocking the relevant nerves. Extended duration of relief could then be achieved by cryoanalgesia to those intercostal nerves. If intercostal block did not work then extradural block with steroid would follow. In general, neurolytic blocks are not recommended for the relief of non-malignant pain, because they do not last forever, recurrent pain may be more difficult to manage, and because of the morbidity. In cancer patients with pain and a poor prognosis, using the intercostal example if the pain did not respond to intercostal block, an intrathecal neurolytic block may be used. These can be particularly effective for severe pains (particularly pelvic or perineal); the limitation is the potential for motor and sphincter damage. This risk is higher with bilateral and repeat procedures, and with blocks at lower cord levels. Extradural neurolytics have limited efficacy. Although claims have been made that the paravertebral approach is preferable, patchy results may be attributed to unpredictable injectate spread.

Similar distinction between malignant and non-malignant pain holds for the use of coeliac plexus block in the management of pancreatic pain. Pain associated with pancreatic cancer responds well to coeliac plexus block, and it may also help those with abdominal or perineal pain from a tumour in the pelvis. Results are much less convincing in patients with chronic pancreatitis.

Chronic back pain accounts for about 25 per cent of non-malignant pain referrals. The major patient groups have pain which has not responded to conservative measures (no previous surgery), in whom the pathology may be due to either disc or facet joint problems, and those in whom pain recurs despite previous surgery, including arachnoiditis.

The major pitfall is to miss a treatable cause of the pain in the rush to treat the symptoms. In patients who have failed to respond to conservative management, the most difficult decision may be what changes in signs and symptoms warrant further investigation. Pain despite surgery should become less common as awareness increases that pain alone may not justify laminectomy: 40 per cent of these patients have recurrent pain after 1 year. While such patients are still being referred, it is important to remember that some may still have a surgically remediable lesion.

Prolapsed invertebral disc produces pain in the back (lumbago), with pain down the leg (sciatica) if severe. It often occurs once or twice a year, with or without a history of provocation, and often settles on its own (conservative management). The pain is thought to come from swollen nerve roots, and then the disc retracts with scarring after months. A herniated disc compressing a nerve root may produce limitation of one specific movement only, and straight leg raising of the good leg may produce crossed pain in the bad leg.

Back pain, with or without leg pain, may be caused by degeneration of the facet joints. Distinction from disc disease may be possible from the history. Facet joint degeneration may give pain on sitting: this may then be relieved by standing up and walking, the opposite of the usual story for disc problems. On examination pain from the facet joints may be elicited by lying the patient prone and extending the facet joints by lifting the legs.

Pain recurring after previous surgery can present special diagnostic problems. The causes of such recurrent pain include recurrent prolapsed invertebral disc and postoperative complications, such as arachnoiditis.

Does the patient have recurrent disc problems? The patient may have a good original history, positive myelogram, disc protrusion removed at operation, and have done well for several months or indeed years. Then history repeats itself, and the pain (and any radiation) may be the same as before the operation. The disc causing the pain may not have been removed: a plain radiograph of the lumbar spine will reveal the level of operation. If a prolapsed invertebral disc was found on myelography and removed at operation, and yet the pain persists, that disc may not have been the cause of the pain. Original investigations (myelography) must include the conus in order to exclude other causes of the pain coexisting with disc lesion. Patients in this category may respond to facet joint procedures, suggesting that this may have been the cause of the original pain.

Postoperative complications that cause pain include dural damage, arachnoid hernia, nerve root pressure, sciatica, and sterile osteitis. Adhesions may occur after a variety of intrathecal insults, resulting in the clinical syndrome of arachnoiditis, which can cause serious long-term problems. The causes include myelography, bleeding, rough or recurrent surgery, and infection: a few cases are idiopathic. Half the patients develop symptoms, signs, or both within a year; in a small number (15 per cent) 10 years may elapse before problems develop. In one series of 80 patients, 43 had undergone myelography, and in many this had been a technically difficult procedure. Spinal surgery had been performed in 51. One-quarter of the patients had progressive disease. In a clinic with a high proportion of ‘failed surgery’ back pain patients arachnoiditis may be a common problem. Classically, arachnoiditis causes a burning constant pain with sciatic distribution to one or both legs, with signs of gradual and progressive loss of leg reflexes. In reality the pain may be difficult to distinguish from that caused by other back problems and the signs may be unconvincing. The diagnosis rests on the history, examination, and myelographic evidence.

The major diagnoses that need to be excluded before symptomatic treatment is initiated are cauda equina claudication or tumour, primary or secondary malignancy in the pelvis, arterio-venous malformation of the spinal cord, and tumour and infection around the spine. Cauda equina claudication, caused by spondylosis producing narrowing of an already narrow lumbar canal, occurs mainly in those over 60 years of age, and there is often a long history. The pain is sciatic, with pins and needles on standing and walking, relieved by sitting and lying down. The pain is often made worse by bending forward (flexion presumably accentuating the narrowing). Stopping walking relieves claudication proper, but may not do so in the cauda equina condition. Some patients may be able to walk further by bending forward. Cycling often does not bring on the pain. On examination one or both ankle jerks may be absent. Myelography shows a complete block or marked stenosis. Cauda equina tumours have the reputation of being classic diagnostic pitfalls: patients are often labelled hysterics. The site of pain is in the back, a sciatic component is reported by 50 per cent of patients. There is characteristically a progressive history, the pain being worse at night: the patient has to get out of bed and sit in a chair. The pain may be made worse by jolting or jarring rather than by the twisting or bending which can bring on pain from disc problems. Micturition difficulties may develop some time before motor or sensory signs are found. The clinical diagnosis rests largely on the history, because there may be few neurological signs. Primary or secondary malignancy in the pelvis may produce back pain with or without leg pain; this may arise from the disease mass or from lumbar plexus invasion. Arterio-venous malformation of the spinal cord can produce back pain, with or without leg pain. The pain may be worse on walking. Neurological signs appear progressively, and results of myelographic examination may be reported as normal. Tumour and infection around the spine can produce back pain: causes include primary myeloma, bony secondaries, extradural tumour, tuberculosis, osteomyelitis, and ankylosing spondylitis.

In taking the pain history, potentially the most useful questions for distinguishing the causes of the pain are those relating to potentiating and relieving factors. Pain that is worse when sitting and improved by walking is suggestive of facet joint problems. Pain worse at night may be indicative of a cauda equina tumour. When pain is worse on walking cauda equina claudication or arteriovenous malformation need to be excluded before the presumptive diagnosis of disc disease is made. The serial measurement and recording of neurological signs is very important in these patients: change can then be monitored objectively. The simple reminders of sensory innervation are that L1, L2 and L3 supply the anterior aspect of the thigh, L4 the medial malleolus and toes, and S1 the lateral malleolus. More weight should be placed on numbness with tingling than on the finding of numbness alone. L2 and L3 are involved in hip flexion, L3 and L4 in knee extension (knee jerk), other hip and knee movements being L5. Ankle movements involve S1 and S2 (ankle jerk S1). Gauging limitation of straight leg raising may require common sense. Gross limitation is incompatible with the ability to sit comfortably at 90° with legs outstretched. There is disagreement as to whether myelography or scans are ‘best’ in helping to decide whether surgery is indicated. The lowest rate of false-positive findings occurs with myelography with screening. If abnormality at the conus is suspected then radiculograms will be inadequate. The combination of myelography and scanning may be the final arbiter in problem cases.

Injections into extradural (lumbar or sacral) steroid and facet joint carry a low morbidity and are usually performed on an outpatient basis. Even short-term benefit from injections may be better than poor relief with drugs, and drugs often cause more problems than the injections. Results from extradural injection of steroids may be improved the earlier the patient is treated, and when there is good evidence that the pain is due to nerve root irritation. When used as a first-line injection treatment for back pain, roughly half of patients have short-term (4 - 8 weeks) benefit, and 10 per cent have pain relief for 6 months or more. These figures improve in direct relation to the duration of the pain and are better in patients who have not undergone back surgery.

Since it may take as long as 1 week for benefit from the steroid to be felt, it is unwise to dismiss the injection as a failure after 1 h. If the injection produces incomplete or short-lived relief it is worth repeating, and a course of three injections is recommended. No additional benefit accrues from more than three injections.

Facet joint injection with local anaesthetic and steroid as a diagnostic/therapeutic procedure may be indicated by a history of pain worse when sitting, and pain on lateral rotation and spine extension. Short-lived success (less than 6 weeks) may be improved by use of cryoanalgesia or radiofrequency blocks to the nerves to the joints.

If sciatic pain is the predominant feature, sacral foramen block may be appropriate. Some patients with long histories of pain, particularly where arachnoiditis is suspected, may have symptoms and signs suggestive of deafferentation. Diagnostic lumbar sympathectomy with local anaesthetic may help such pains. Short-lived success may be prolonged with chemical sympathectomy.

Some patients with back pain, particularly those who have had back surgery, respond poorly; this is disheartening for both patient and doctor. It is important that they receive honest advice and treatment, and that patients who still have surgically remediable disease are identified (a tiny minority). Those with arachnoiditis are at risk of progressive disease. Alternative methods such as transcutaneous nerve stimulation, acupuncture, or psychological/behavioural therapy are often tried at this point. This is probably an unfair test of any treatment, and all can produce short-term pain relief in some individuals, but this is often unsustained. There is no simple way to manage these ‘failed’ patients. It may be difficult to protect them from the worst excesses which may be offered to them as they go desperately from place to place seeking help. The heterogeneity of conditions means that no single new treatment or drug is likely to be the complete answer.

Post-herpetic neuralgia, like other neuralgias, is a pain in the distribution of a nerve. It follows an acute herpetic (shingles) attack, the pain being due to destruction of cells in the posterior nerve root. If enough are destroyed there will be sensory loss and this is directly proportional to the degree of dysaesthesia. The incidence of post-herpetic neuralgia increases with age: 90 per cent of affected patients are over 60 years old. The condition can last for many years. There is rarely diagnostic difficulty: the site of the pain is in the distribution of the shingles, and scarring is often visible. Patients are not usually woken by their pain. There may be visceral involvement in the acute attack, the viscera involved being those supplied with afferent fibres by the posterior nerve roots corresponding to the affected skin areas. Symptoms such as constipation, indigestion, frequency, or dysuria, depending on the affected nerve, may persist. There is a strong feeling that the success of any remedy is in inverse relation to the duration of the pain: claims for putative remedies must be examined in this light. While sympathetic blocks (stellate, lumbar sympathetic, or extradural) can undoubtedly be effective in relieving the pain of acute zoster, there is as yet no controlled evidence to show whether such pain relief prevents the development of post-herpetic neuralgia. The use of antiviral agents such as topical idoxuridine 35 or 40 per cent in DMSO or acyclovir 800 mg 4-hourly orally can help ‘abort’ the acute zoster attack if given early. It is not yet known whether curtailing the acute attack in this way lowers the incidence of post-herpetic neuralgia. While sympathetic, regional, or even local nerve blocks may provide short term relief, there is little evidence that they alter the natural history of the disease process. If the affected area still has sensation, diagnostic nerve blocks with local anaesthetic may produce short-term relief, and this may be extended by measures such as cryoanalgesia. Neurolytic agents are not indicated; initial short-term relief may be followed by pain worse than that initially experienced by the patient. Conventional analgesics have a very limited role in management. The pain does not appear to be sensitive to opiates. Of the unconventional analgesics amitriptyline 25 mg at night is the drug of choice, reducing the dose to 10 mg in the elderly or infirm. Its use may be difficult in elderly men with prostrate problems. Persistent shooting or stabbing pain may respond to anticonvulsants, and valproate 200 mg twice daily or clonazepam 500 &mgr;g at night (rising to twice daily) may be helpful. Local anaesthetic creams or topical non-steroidal agents are of limited benefit, but may help hyperaesthetic pain. Transcutaneous nerve stimulation may help relieve dysaesthetic pain, as indeed may a simple vibrator applied to the junction of skin areas with normal and abnormal sensation.

The more common causes of facial pain are trigeminal neuralgia, atypical facial pain, post-herpetic neuralgia, post-traumatic neuralgia, temperomandibular dysaesthesia, and malignancy. Post-herpetic neuralgia affecting the trigeminal division is not uncommon but rarely presents difficulties with diagnosis. Distinguishing between trigeminal neuralgia and atypical facial pain is not difficult.

Trigeminal neuralgia occurs more often in the middle-aged and is twice as common in females as in males. Patients are referred because the diagnosis is in doubt, because straightforward management with carbamazepine has failed, because of tolerance, side-effects, or allergy, or because previous invasive measures have failed. The pain is, for practical purposes strictly unilateral; multiple sclerosis patients account for most of the 2 per cent of those with bilateral disease. Trigeminal neuralgia is twice as common on the right side. The most common pattern is pain in both the mandibular and maxillary divisions. The pain in the face is characteristically sharp, severe (paroxysmal) and brief, lasting no more than a few seconds. Tic doloureux describes the facial contortions which may accompany the pain. A high frequency of these attacks may lead to a persistent pain which is duller in nature. The pain may be brought on by thermal, tactile, or proprioceptive stimuli, but not by nociceptive stimuli. These intermittent attacks may last for 6 to 8 weeks. Long spontaneous remissions of months or even years may occur in the early stages, but these tend to become shorter as the disease progresses. The severity and frequency of the attacks thus increase over the years. There should be no abnormal neurological signs in trigeminal neuralgia, and pain relief with carbamazepine is taken to be diagnostic. If carbamazepine is well tolerated but ineffective, increasing the dose in divided doses to daily maximum of 1500 mg should be tried. If the drug is poorly tolerated (nausea and vomiting, ataxia, skin rash, or blood dyscrasia) phenytoin 100 mg thrice daily may be tried, alone, or in addition to the lowest tolerated carbamazepine dose. The role of nerve blocks is to provide pain relief when pharmacological management is unsuccessful, or as an adjunct to allow dose reduction. Individual divisions of the 5th nerve may be blocked with local anaesthetic where they leave the skull and enter the face and mouth (infraorbital, supraorbital, inferior dental, and meatal nerves). Besides their diagnostic value it is not uncommon for these blocks to break a cycle of pain. Relief may then be obtained for much longer than is achieved with local anaesthetic.

Short-lived success with local anaesthetic may be prolonged with cryoanalgesia. In one study of 24 patients, median duration of relief was 186 days (range 0 - 1236), contrasting with median duration of sensory loss of 67 days (range 14 - 80), which indicates the ‘reversible’ nature of the cryoanalgesia lesion. Cryoanalgesia, ‘peripheral’ radiofrequency lesions, neurolytic injections, or avulsion and/or nerve section should not be regarded as curative; they are ineffective in the long term. Relapse rates increase with time. The advantage of cryoanalgesia is that the patient is no worse off if the pain returns, and the cryoanalgesia can be repeated successfully. This is not always the case with the other measures.

More central procedures need to be judged on their associated operative morbidity and mortality and long-term morbidity (sensory or motor loss, anaesthesia dolorosa, damage to other cranial nerves). Unfortunately the operations with the lowest recurrence rates are likely to be those with the highest complication rates. The most predictable manoeuvre to ablate tic doloureux is sensory root rhizotomy, usually performed in the middle cranial fossa, and, although there may be some sparing of sensation, corneal ulceration, drooling, and trophic lesions of the skin may occur. The recurrence rate with this procedure is low (4 per cent at 4 years). The most serious morbidity is anaesthesia dolorosa, a constant burning pain in the numb area. The incidence of this complication is probably 5 to 10 per cent, and the only treatment is symptomatic.

Vascular decompression procedures are currently fashionable, and time will tell whether the low complication rate is balanced by a high recurrence rate. If that proves to be the case such major surgery may not be justified. Radiofrequency lesions at the gasserian ganglion, where the electrode is inserted through the foramen ovale under radiographic control, use controlled temperatures to destroy smaller pain fibres selectively. If sensation is preserved, recurrence rates for the first division may be as high as 25 per cent; there is also a risk of anaesthesia dolorosa (0.3 - 3 per cent).

The injection of glycerine into Meckel's cave appears not to have been as successful as early reports claimed. The clinical utility of alcohol block of the gasserian ganglion, less predictable than rhizotomy and carrying a higher risk of the same complications, is also limited in the light of the alternatives.

Atypical facial pain is often diagnosed by exclusion of other causes of facial pain. Classically it affects middle-aged women. The facial pain is often wrongly attributed to dental problems, and continues despite extraction(s). While the site of pain is in the face, it may not tie in with the distribution of any particular nerve, and it may cross the midline. The pain is dull, compared with the sharp pain of trigeminal neuralgia. The pain may be persistent or recurrent, uni- or bilateral, occur in the absence of any muscular or joint problems, and may be associated with no abnormal neurological signs, akin to trigeminal neuralgia. Although the pain is episodic, it is unlike that of trigeminal neuralgia in that it builds up gradually to a climax and it may last for hours and even days. It is anatomically imprecise and covers a much larger territory than that supplied by an individual cranial nerve. The pain is never ‘electric’ and is usually described as tearing or crushing. The pain may be associated with other symptom complexes, such as drastic colon, dysfunctional uterine bleeding, headaches, and low back pain. These complexes may recur sequentially or simultaneously in response to stress.

It is wise to look for a hidden cause: teeth, sinuses, and eyes may be incriminated. At least three ‘parallel’ conditions may be associated with atypical facial pain. Temporomandibular dysaesthesia (facial arthromyalgia) may be difficult to distinguish from atypical facial pain but there is usually a history of pain in muscles and/or joints, clicking, sticking, or trismus, and a feeling of buzzing or fullness in the ear. On examination there may be ridged buccal mucosa, crenated tongue, or masseteric hypertrophy, and arthritic change may be seen on plain radiographs of the condylar surface. Atypical odontalgia may be distinguished from atypical facial pain because the continuous or throbbing pain is felt in the teeth (tooth), is hypersensitive to all stimuli, and may move from tooth to tooth. Glossodynia or oral dysaesthesia is the sensation of a dry mouth, burning tongue and gums, with denture intolerance, disturbance of taste and salivation, and no organic pathology.

The pathogenesis remains mysterious, but the patients need to be reassured that it is a ‘real pain’, which is best managed medically rather than by surgery, which may make it more refractory. The association with adverse life events (80 per cent of patients) emphasizes that this condition may affect the emotionally fit as a stress response. Tricyclic antidepressants are the treatment of choice. Dothiepin is recommended: treatment (75 to 150 mg at night) should be maintained for 3 months.

Pain after trauma and pain in or near an operative scar, occurring for months and even years after surgery, can be extremely difficult to manage. Because the principles of management are similar they are discussed together.

Post-traumatic neuralgia may be reported by a patient seeking compensation for an accident. Much has been made of the intransigence of the neuralgia until the compensation is settled; unfortunately for many of our patients the pain remains long after the compensation is paid. The aetiology of postoperative wound pain is unknown: why are some patients affected and not others? As always when this is the case, doctors ascribe a psychogenic overlay. Post-traumatic neuralgia may be experienced in any part of the body, and may be associated with damage to bone, nerve, muscle, or other tissue. Specific information to be sought in the history and examination includes damage to nerves in the affected area at the time of injury and the current neurological status of the painful area. Is it numb?, is there painful pins and needles?, and is there any difference in temperature between affected and unaffected areas?

Three particular situations seem most likely to result in postoperative wound pain: thoracotomy (particularly thoraco-abdominal incisions), nephrectomy, and inguinal herniorrhaphy, but pain at or around scars from other operations is also seen. The quality of the pain is often described as burning, even if the area still has normal sensation (unusual). There is not usually hyperaesthesia. The aura of the pain (the extent of the area affected) may increase as time elapses.

Conventional analgesics may have a very limited ability to relieve both post-traumatic and postoperative neuralgias. Antidepressants and anticonvulsants are more likely to help, particularly if the pain is burning and occurring in an area of altered sensation. A starting regimen might be amitriptyline 25 mg at night with valproate 200 mg twice daily. An alternative anticonvulsant is clonazepam, starting at 500 &mgr;g at night, and increasing the dose in the absence of adverse effects to 1.5 or 2 mg thrice daily. Diagnostic blocks with local anaesthetic should be tried when appropriate. Intercostal blocks may provide relief of pain following thoracotomy or nephrectomy ilio-inguinal blocks are suitable for post-herniorrhaphy pain. If the relief is short-lived, the duration may be extended by repeating the block with cryoanalgesia. This may (especially for the ilio-inguinal nerve) be best achieved with surgical exposure under direct vision rather than with a percutaneous approach. Stellate ganglion (for head, neck, upper limbs, and upper chest wall) or lumbar sympathectomy may be the first-line of management, particularly for post-traumatic neuralgia, and the greatest chance of success is if the pain is in an area of altered sensation. Extending the duration of pain relief with successful ‘single-shot’ stellate ganglion block may be achieved by repeating the block or by infusing local anaesthetic via a catheter. If there is undoubted success with local anaesthetic blocks, but this cannot be sustained despite repeated injections, surgery may be appropriate. Short-lived relief from lumbar sympathectomy with local anaesthetic may be prolonged by using phenol.

Perineal neuralgia with no malignant cause most commonly affects postmenopausal women. Again it is a most difficult pain to treat successfully. The aetiology is unknown, but pelvic varicosities are thought to be a cause of pelvic pain. The symptoms and signs (or lack of) in perineal pain would seem to be distinct from those of the pelvic pain syndrome. The pain described variously as a ‘tugging’, ‘pulling’ or ‘crawling’ sensation, commonly in the clitoral area. Many women with this condition prefer to stand, because the pain is worse when they sit. Conventional analgesics are rarely effective, and unfortunately antidepressants and anticonvulsants do not help most patients. Women are often prescribed local hormone therapy; few seem to be helped. Sacral extradural anaesthesia with local anaesthetic may be tried diagnostically. Many of these patients will still have their pain even when the relevant area is numb from the local anaesthetic. If the local anaesthetic does help, the duration of relief may be prolonged with sacral extradural cryoanalgesia. Lumbar sympathectomy may be a logical procedure if the cause is pelvic varicosities, but it has not proved successful.

A large proportion of amputees suffer from phantom pain: 78 per cent of the 55 per cent responding to a survey of 5000 amputees had such pain, although previous reports had given a much lower incidence of pain. There seem to be few predictors of disabling phantom pain; age at amputation, years since amputation, reason for amputation, site of amputation, and pain before surgery were similar in those with and without problematic phantom pain. The aetiology of the pain is unknown, but the mechanism is presumed to be central, the lack of sensory input causing decreased tonic inhibition.

Pain in the stump may occur early after surgery due to surgical postoperative complications, or late in a well-healed stump: then the origin of the pain is attributed to neuroma development.

At least 50 different methods of treating phantom limb pain are in use, but only 1 per cent of patients derive long-term benefit from any of them. Whichever method is used, best results come with early treatment. Neither minor nor major conventional analgesics are of great benefit, but opiates may make the patient feel better. The use of anticonvulsants has been more successful, but there is no controlled evidence to support their use. Severe shooting or lancinating pains are most likely to respond, and clonazepam 500 &mgr;g at night rising to 1.5 to 2 mg thrice daily if the drug is well tolerated may be the most effective of this group. Sympathetic blocks with local anaesthetic may be used diagnostically, and short-lived relief may be prolonged with repeat local anaesthetic blocks or chemical sympathectomy. Despite claims for the efficacy of these procedures there is little evidence for long-term benefit. Surgical procedures seem to be the least successful.

Stump pain often coexists with phantom pain, and treatment of the phantom with anticonvulsants may also help the stump pain. A purely ‘local’ stump pain due to neuroma development may benefit from local anaesthetic injection to the site, repeated if required, and with either percutaneous or direct vision cryoanalgesia to extend the duration of relief.

Ischaemic limb pain may be helped by lumbar sympathectomy when surgery is inappropriate. Sympathectomy increases skin blood flow, and rest pain can be relieved in the majority of patients. The results are equivalent to surgical sympathectomy, but with lower morbidity and mortality. One of the enigmas is that the pain relief appears to last longer than the measurable haemodynamic effects of the procedure. Intermittent claudication is not helped reliably by sympathectomy, and long-term results on ulcer healing are unconvincing.

Diagnostic block with local anaesthetic under radiographic control is followed if successful by the injection of a small (1 - 3 ml) volume of 6 per cent aqueous phenol. The major short-term risk of the procedure is injection of neurolytic solution in the wrong place. Genitofemoral neuralgia occurs in 5 to 10 per cent of patients and usually resolves over 10 weeks. Bilateral block can cause impotence.

Intravenous guanethidine blocks are an alternative if anticoagulation precludes lumbar sympathetic block. Spinal opiates may be appropriate to manage severe rest pain in the terminally-ill patient.

In both reflex sympathetic dystrophy (continuous pain associated with sympathetic hyperactivity in part of a limb after trauma), and causalgia (burning pain in hand or foot after nerve injury), the sympathetic nervous system ‘maintains’ the pain, and blocking the system can improve the pain. In both conditions the patient protects the painful limb; the changes due to disuse, up to the full-blown osteoporotic picture known as Sudeck's atrophy, may be seen. The natural history is for a degree of resolution, but this can be expedited by sympathetic block and vigorous rehabilitation, and an operative sympathectomy may be of value. Repeated intravenous blocks with local anaesthetic relieve pain sufficiently to allow rehabilitation. Oral steroids are an alternative in the management of reflex sympathetic dystrophy.

Barnard D, Lloyd J, Evans J. Cryoanalgesia in the management of chronic facial pain. J Maxillofacial Surg 1981; 9: 101 - 2.

Beard R, Reginald P, Pearce S. Pelvic pain in women. Br Med J 1986; 293: 1160 - 2.

Benzon H. Epidural steroid injections for low back pain and lumbosacral radiculopathy. Pain 1986; 24: 277 - 95.

Carette S, et al. A controlled trial of corticosteroid injections into facet joints for chronic low back pain. N Engl J Med 1991; 325: 1002 - 7.

Fearnside M, Adams C. Tumours of the cauda equina. J Neurol Neurosurg Psychiatr 1978; 41: 24 - 31.

Feinmann C, Harris M, Cawley R. Psychogenic facial pain: presentation and treatment. Br Med J 1984; 288: 436 - 8.

Foley K. The treatment of cancer pain. N Engl J Med 1985; 313: 84 - 95.

Hannington-Kiff J. Pharmacological target blocks in painful dystrophic limbs. In: Wall P, Melzack R. Textbook of Pain. London: Churchill Livingstone; 1989; 754 - 66.

Kerr R, Cadoux-Hudson T, Adams C. The value of accurate clinical assessment in the surgical management of the lumbar disc protrusion. J Neurol Neurosurg Psychiatr 1988; 51: 169 - 73.

Martin G. The management of pain following laminectomy for lumbar disc lesions. Ann R Coll Surg Engl 1981; 63: 244 - 52.

McKendrick M, McGill J, White J, Wood M. Oral acyclovir in acute herpes zoster. Br Med J 1986; 1529 - 32.

McQuay H. Pharmacological treatment of neuralgic and neuropathic pain. Cancer Surv 1988; 7: 141 - 59.

McQuay H, Dickenson A. Implications of nervous system plasticity for pain management. Anaesthesia 1990; 45: 101 - 2.

Miller H. Pain in the face. Br Med J 1968; ii: 577 - 80.

Shaw M. Russell J, Grossart K. The changing pattern of spinal arachnoiditis. J Neurol Neurosurg Psychiatr 1978; 41: 97 - 107.

Sherman R, Sherman C, Parker L. Chronic phantom and stump pain among American Veterans: results of a survey. Pain 1984; 18: 83 - 95.

Verrill P. Sympathetic ganglion lesions. In: Wall P, Melzack R. Textbook of Pain. London: Churchill Livingstone; 1989; 773 - 83.

Waddell G. An approach to backache. Br J Hosp Med 1982; 28: 187.

Wall P. Introduction. In: Wall P, Melzack R. Textbook of Pain. London: Churchill Livingstone; 1989: 1 - 18.

Watson C, Evans R, Reed K, Merskey H, Goldsmith L, Warsh J. Amitriptyline versus placebo in post-herpetic neuralgia. Neurology 1982; 54: 37 - 43.

Wood K. Peripheral nerve and root chemical lesions. In: Wall P, Melzack R. Textbook of Pain. London: Churchill Livingstone; 1989: 768 - 72.

Wyburn-Mason R. Visceral lesions in herpes zoster. Br Med J 1957; i: 678 - 681.