Practical Issues When Using Neuraxial Infusion

May 1, 1999

The pharmacologic tailoring guidelines of the World Health Organization represent the accepted treatment algorithm for the management of cancer-related pain syndromes. Unfortunately, the guidelines are only effective

ABSTRACT: The pharmacologic tailoring guidelines of the World Health Organization represent the accepted treatment algorithm for the management of cancer-related pain syndromes. Unfortunately, the guidelines are only effective in 70% to 90% of patients, leaving a substantial population with intractable pain. In fact, recent surveys have shown that, in the United States, only 50% of hospitalized terminally ill patients die comfortably. When patients do not respond to the WHO guidelines, practitioners should abandon the therapy and not the patient. Interventional pain management approaches including intraspinal delivery of analgesics may be an effective alternative. Before a more permanent system for intraspinal therapy is implemented, a trial must be performed to assure efficacy, rule out toxicity, and establish that the positive response to the intrathecal agent trialed is not due to placebo effects. An external drug delivery system is appropriate if the patient has less than 3 months to live. If the patient is expected to survive more than 3 months, a totally implanted system is appropriate and cost- effective. Although morphine remains the gold standard for intrathecal pain therapy, other opioids such as fentanyl, hydromorphone, sufentanil, and meperidine are now being used in patients who do not tolerate morphine. Nonopioid agents for intrathecal use in patients who have pain syndromes that are poorly responsive to opioids include local anesthetics, such as bupivacaine, and clonidine. [ONCOLOGY 13(Suppl 2):37-44, 1999]


In 1986, the World Health Organization (WHO) disseminated an algorithm for the use of analgesic medications in cancer patients with pain.[1,2] This simple pharmacologic tailoring approach could be used by physicians in both developed and developing countries (Figure 1). When used appropriately, this three-step analgesic ladder has been effective in approximately 70% to 90% of patients.[3-7]

Unfortunately, these guidelines, while adhering to the KISS principle (keep it short and simple), did not provide guidelines for the appropriate use of adjuvant medications such as tricyclic antidepressants, anticonvulsants, steroids, membrane stabilizing agents, etc., and did not specify the optimal management of the patients who do not respond. Additionally, many patients throughout the world do not, in fact, benefit from this approach because opioid medications are unavailable to them, or because barriers to the successful implementation of the WHO ladder exist.

These barriers include actual and perceived governmental regulations on opioid prescribing practices, deficits in the knowledge of health care providers, caregivers, and pharmacists on how and when to use opioid agents, religious beliefs, the fear that opioids are harmful to patients and may hasten their demise, and the pervasive misunderstanding that the use of opioids inevitably leads to tolerance and addiction.[8] The problem is probably greater than previously perceived; it has recently been estimated that only 50% of patients actually benefit from the WHO ladder because of the barriers described previously.[9-12]

Physicians should, above all, do no harm and learn to use the least invasive and least costly therapies. We should do what works for our patients. For cancer patients, single therapies like the analgesic ladder approach should be optimally administered whenever pain is reported. When the WHO fails, alternative therapies should be available.

Choose Therapies That Work; Abandon Those That Don’t

Untreated cancer pain remains an enormous and truly unfortunate problem. It has been estimated that more than 70% of all patients dying from cancer experience pain in the end stage of their disease.[13] In 1995, in the United States, approximately 547,000 patients died from their cancer. If the 70% estimate is correct, 382,000 of these patients suffered from intractable pain. Furthermore, if the survey data showing a less than 50% response rate to the WHO guidelines is correct, then approximately 190,000 patients experiencing severe pain from their disease would have had their pain well controlled if their caregivers used the simple guidelines of the analgesic ladder of the WHO. It also means that more than 190,000 patients did not.

If the WHO guidelines for the treatment of cancer pain were the only option, a very significant proportion of cancer patients in the United States would die suffering from intractable pain. Clearly, there is a place for other therapies. Recognizing the failure of the WHO guidelines to effectively treat all patients with cancer pain, many physicians have incorporated a “fourth rung to the WHO ladder” for patients who fail systemic opioid tailoring (Figure 2).

After the failure of opioid tailoring, it is imperative that physicians do not abandon their patients, either by walking away or by resorting to terminal sedation. Treating physicians must choose therapies that work and abandon those therapies that don’t work, and not the patients.

Therapies not included in the WHO guidelines include psychological and complementary approaches such as relaxation, massage, distraction, meditation, prayer, etc., and more invasive analgesic approaches such as temporary nerve blocks; spinal infusions of opioids, local anesthetics and/or alpha-2 agonists; neuromodulatory techniques; and various somatic and sympathetic chemical, surgical, and thermal neurodestructive procedures. Table 1 lists possible interventions that can be used in appropriately indicated patients.

Many of the techniques with which pain specialists treat pain in noncancer patients today we learned from our experience with treating the pain of cancer patients. We have learned to use the least invasive and less costly therapies before resorting to more invasive and more costly therapies. We have learned that very few patients become addicted to opioids. We have learned to do what works and abandon what does not work.

These lessons are the foundation for treatment strategies now used for the noncancer pain patients. Indeed, a pain treatment continuum, though recommended for noncancer patients, also might be used for those patients who are suffering from cancer pain (Figure 3).

Spinally Administered Opioids for Cancer Pain

When the systemic delivery of opioids is not working because of dose-limiting side effects or because pain appears to be poorly responsive to opioids, a trial of intraspinal opioids, alone or in combination with nonopioids such as local anesthetics or clonidine (Duraclon), may be indicated. Before a trial of intraspinal opioids, patients should undergo a trial of sequential oral or transdermally administered opioids (Table 2). Because most cancer pain syndromes cause constant pain in patients, it is recommended that long-acting opioids be prescribed on a timed, around-the-clock basis, and not on an as-needed basis.[7]

There should be medication allowance for pain that “breaks-through” the baseline level of the long-acting medication.[7] “Rescue” medication for breakthrough pain should be a short-acting opioid. If pain is uncontrolled, the dose of the long-acting opioid is adjusted upwards.

Physicians treating cancer patients with pain should be educated about the principles of opioid therapy, including the role of the oral route, equianalgesic doses, methods of dose titration, and side-effect management. Administration of opioids via the intravenous or subcutaneous route can be more costly than intraspinal delivery of opioids and should not be automatically recommended as an alternative to intraspinal opioid delivery.[14]

Not all pain is the same[15] and certainly not all pain syndromes react to opioids in the same way. Pain syndromes are usually categorized as nociceptive (either visceral or somatic), neuropathic, or a combination of both (so-called mixed pain). Nociceptive pain is believed to be sustained by activation of primary afferent nociceptors. It appears to be usually responsive to opioid therapy.

Neuropathic pain, on the other hand, is not mediated by nociceptors and may be responsive to opioids at doses equivalent to those for nociceptive pain syndromes, or at higher doses, or it may not respond to opioids at all.[16-18]

Neuropathic pain syndromes in cancer patients that may not respond to usual doses of opioids or may not respond at all include neural involvement by tumors; radiation fibrosis of the head and neck, the brachial plexus, intercostal nerves, the lumbar or lumbosacral plexus, the superior hypogastric plexus, or the celiac plexus; neurologic sequelae of chemotherapeutic agents; the peripheral neuropathies; and damage to neural structures by surgical interventions.

Trial of Intraspinal Analgesics

Once it has been established that a cancer patient does not respond adequately to systemically administered opioids, the patient may be a candidate for a trial of intraspinal analgesia. The term intraspinal analgesia is preferred over intraspinal opioid in this context because nonopioid agents are now available for intraspinal use.

 These nonopioid agents include the local anesthetics and the alpha-2 agonist, clonidine. Other agents that might be used, but are considered experimental, include somatostatin (Zecmil)[19] or the somatostatin analogue, octreotide (Sandostatin)[20]; SNX-111, an N-channel calcium blocker[21]; and midazolam (Versed).[22,23]

Contraindications to the Use of Intraspinal Analgesics

Table 3 outlines the indications for intraspinal analgesia for patients suffering from intractable cancer-related pain syndromes. As noted, all patients who are candidates for intraspinal delivery of analgesics should have failed more conservative therapies. There should be no contraindications to implanting a drug delivery system and intrathecal catheter. These contraindications include allergy, localized infection in the areas where surgery is necessary, sepsis, and coagulopathy. When confronted with these possible contraindications in the dying patient in pain, physicians must weigh the possible risks of the procedure against the benefits and discuss these risks and benefits with the patient and family so that they can participate in the decision.

It is essential to establish opioid responsiveness to intraspinally administered opioids. There really is no “magic” to the concept of intraspinal delivery. If patients do not obtain any analgesia from systemic opioids, the use of intrathecal opioids is unlikely to work. In this situation, analgesia during an intraspinal opioid trial is most probably due to the nonspecific (placebo) effects of the drug.

The intraspinal route may still be valuable in such patients, however, because of the availability of nonopioid analgesics. In these cases, it may be best to consider epidural analgesia with opioids and/or bupivacaine and/or clonidine. Epidural analgesia provides the sodium channel blockade of the peripheral and central nerves that may be necessary to achieve pain control. The addition of the opioid and/or clonidine delivers synergistic pain relief that allows for a reduction in the dose of all of the agents being used.[24]

An intraspinal trial should evaluate analgesic and functional outcomes, rule out toxicity of the agent trialed, and mitigate against any nonspecific effects or placebo effects. Trials of intraspinal opioids and nonopioids alike have been performed by single-shot epidural administration or intrathecal placement of the drug or drugs, repeat single-shot epidural or intrathecal placement of the drug or drugs, and continuous drug delivery via the epidural or intrathecal route using an external pump.

Mitigating the ‘Placebo Response’

Because intraspinal analgesia may have very strong placebo effects, trials for efficacy must also mitigate against these responses. The only way to accomplish this practically in any single patient is to extend the trial as long as logistically possible. Giving a patient a single-blinded or double-blinded placebo is not a reasonable approach in this setting. To withhold therapy from a “placebo responder” because that patient responded to the inactive placebo, or to implant an expensive drug delivery system in a patient who responded positively to a single-shot active agent, also makes no sense at all.

Because all patients can react to both the specific effects and the nonspecific effects (placebo responses) of the agent given, one cannot tell from any one patient’s response to a single encounter with a single agent the degree to which the resultant effect is specific or nonspecific. A placebo response during the trial is heralded by a decreasing analgesic response over time. Therefore, the only way to mitigate against the placebo response is to design a trial that mimics, as closely as possible, the final intended delivery system and deliver the intended agent as long as logistically possible.

A continuous intrathecal trial, extended for as long as possible, is the one trial that allows for both sequential trials of intrathecal agents and an extended trial that may mitigate against potent placebo responses. Although not perfect because placebo effects can, in some cases, be very long-lived, this approach is likely to be the best indicator of future effects. Our trials are usually performed on an outpatient basis (with the patient at home). Temporary catheters (24 gauge, pediatric, epidural) are placed intrathecally via a 20-gauge Tuohy needle under fluoroscopic guidance to allow tip placement as near as possible to the spinal cord segment processing the patient’s pain. All of our patients are given preoperative intravenous antibiotics to prevent infections such as Staphylococcal aureus or Staphylococcal epidermis and are sent home with oral antibiotics to cover these same bacteria for the length of the trial.

Catheters Are Not Tunneled

These catheters are not tunneled. They are covered with a clear dressing to allow observation of the catheter exit site, and are connected to external tubing that is linked to an external pump. A 0.22-µ antibacterial filter is placed in-line (Figure 4 and Figure 5), and enough medication is prepared to allow for a 1- to 2-week trial.

If the patient tolerates the drug during the trial, dosage changes are made by reprogramming the pump. If the patient does not tolerate the drug, the drug reservoir and tubing are changed proximal to the 0.22-µ filter, which is outside of the clear dressing. Then, and only then, is the tubing changed during the trial. Because the rate of infected systems is directly related to the number of times that the system is manipulated, our protocol calls for a “hands-off” policy to tubing, drug, and dressing changes.[25] The only time that the dressing is changed is within the first 48 hours; all dressing changes are performed by the physicians who implant the catheters or by the nurses who are quite familiar with the system.

During the trial, a 50% reduction in pain intensity, an improvement in function, and a concomitant significant reduction in oral or systemic analgesics is usually indicative of analgesic effectiveness. Although one would like to show improvement in the physical functioning, the disease itself may prevent this. An improvement in the quality of life may be the most that we can expect in this population. The final evaluation of the efficacy of a trial of intraspinal opioids must be individualized; the benefits must be weighed carefully against the risks of the procedure, changes to the lifestyle of the patient, and/or burden of care on the family. Patient and family input is essential for deciding whether a trial is positive or not.

Because the equianalgesic dose of spinally administered opioid is significantly less than the systemic dose, it is important to prevent acute withdrawal. To prevent acute withdrawal from systemic administration of opioids, we suggest that 50% of the orally administered opioid dose be given as an intrathecal equivalent during the first day of the trial and that the patient be allowed to continue receiving the remaining 50% of the oral dose. Each subsequent day, the oral dose should be decreased by 20% and the intrathecal dose increased by 20%.

Intraspinal Delivery Systems in the Cancer Patient

Cancer patients who respond positively to a trial of intraspinal analgesia should have a permanent delivery system implanted. There are multiple choices for intraspinal delivery systems, including totally implantable pump systems that are factory preset, rate-specific pumps; the more expensive, but more versatile, totally programmable pumps; or external, nonimplanted pump systems that are connected to either implanted catheters or implanted ports. The choice of delivery system may be influenced by the patient’s reimbursement system, but if this is not the case, there are a few recognized guidelines based on a cost analysis of the two systems for intraspinal delivery.

Lanning et al[26] and Bedder et al[27] have shown that if a patient has more than a 3-month survival time, an implanted delivery system is more cost-effective than an external system that comprises an implanted catheter or port and an external pump (Figure 6). If the patient is going to live less than 1 month, a percutaneously placed intrathecal or epidural catheter is safe, cost-effective, and efficacious.

Intraspinal Agents for the Management of Cancer Pain

Morphine remains the “gold standard” of spinally administered opioid therapy because of its long duration of action and relative ease of use. Morphine, the only analgesic approved for intraspinal use, also has been historically favored. More is known about the use of intraspinal morphine than any other opioid available. Other opioids, such as hydromorphone (Dilaudid), meperidine (Demerol), methadone (Dolophine), fentanyl (Sublimaze), and sufentanil (Sufenta), have all been used intraspinally, but less is known about these drugs. Although physicians can use these drugs intraspinally, they are not labeled for this use. Time to onset of action, duration of action, uptake and distribution, availability to supraspinal centers, and central nervous system side effects are all governed by the lipid solubility and receptor affinity of the drug (Table 4).[28]

Agents with high lipid solubility cross the dura quite rapidly and are quickly absorbed by the lipid membranes of the spinal cord. These agents tend to stay within a relatively small area surrounding the catheter tip within the intrathecal space.

Opioids, such as morphine, that have low lipid solubility and high receptor affinity cross the dura and enter the lipid substance of the spinal cord slowly, but remain bound for prolonged periods of time. Hence, the onset of analgesic action for hydrophilic opioids is slow, but generally prolonged, for approximately 12 to 24 hours after a single bolus. Because of its low lipid solubility, or high hydrophilicity, more drug remains in the cerebrospinal fluid (CSF) and is available to ascend to supraspinal centers through bulk flow of CSF.

As a result, placement of a catheter for intrathecal infusion of the drug anywhere in the thecal sac ensures analgesia anywhere in the body. Risks of cerebrospinal fluid side effects such as sedation, nausea and vomiting, and respiratory depression are greater with this hydrophilic group of opioids than in those with higher lipid solubility and higher receptor affinity, like fentanyl or sufentanil.

As expected from their physiochemical properties, lipophilic drugs such as fentanyl and sufentanil have a rapid onset of action and a prolonged duration of action. Once receptors are saturated with sufentanil, drug becomes available for redistribution through spinal vessel uptake and CSF bulk flow. Because lipophilic agents enter the substance of the lipid-containing spinal cord rapidly and are quickly eliminated from the CSF, careful catheter tip placement is essential for optimal analgesia.

It is usually best to plan for versatility when implanting an intraspinal system. If one wants to use lipophilic agents (and one should always plan for such an eventuality), the catheter tip should always be placed as close as possible to the spinal cord segment processing and modulating the patient’s pain.

The appropriate dose of opioid for intraspinal use varies greatly and depends on the patient’s age, pain syndrome, and the systemic dose of the drug needed for analgesia before the move to intraspinal delivery. As a general rule, patients with neuropathic pain may require higher doses than patients with nociceptive pain, and the elderly usually require less than patients who are younger. However, the dosing of all patients should be individualized. Table 5 lists generally accepted intraspinal conversion ratios for intraspinal morphine, and Table 6 provides a suggested equianalgesic dosing chart for converting morphine to other opioid agents.

Simple Guidelines: The Art of Medicine

In our practice, patients are usually started on intrathecal morphine at the lowest dose possible. If the patient continues to have pain, we increase the dose of morphine by 20% every 3 to 7 days and allow for oral rescue medications, specifically a short-acting opioid, and adjuvant medications such as membrane stabilizing agents, anticonvulsants, and tricyclic antidepressants. Our usual ceiling for intrathecal morphine (or morphine equivalent mg of another opioid) is 20 mg per 24 hours. We understand that this is an arbitrary ceiling, but we feel that if patients do not respond to 20 mg of morphine/day, then another strategy is needed.

Some patients develop the so-called opioid hyperalgesia syndrome above the 20-mg ceiling. However, to be fair, many patients will not develop this syndrome and will actually tolerate higher doses. The difficulty lies in how to determine whether the pain that the patient experiences at higher doses is due to disease or due to opioid hyperalgesia. If a patient does not respond to morphine at its ceiling, other strategies are needed, including changing opioid or adding synergistic medications such as bupivacaine or clonidine. Bupivacaine is usually added to an opioid, started, in our practice, at 5 to 7 mg/day, and titrated upwards. Bupivacaine can usually be titrated to above 30 mg/day in most patients without the development of neurologic side effects such as paresis, or urinary retention. Titration should be slow, as in titrating morphine, and should not exceed 20% increments.

Clonidine, too, is added to the opioid and started in our practice at 50 µg/day, and titrated upwards. Dose-limiting side effects of clonidine include sedation, dry mouth, bradycardia, and hypotension. At lower doses, clonidine is primarily an alpha-2 agonist and a potent antihypertensive agent, but at higher doses, it is primarily an alpha-1 agonist and a potent hypertensive agent. Also, it must be remembered that one should not suddenly withdraw clonidine for fear of rebound hypertension.

If a patient develops side effects to morphine before analgesia, we switch first to hydromorphone at half the equianalgesic dose. We then titrate the hydromorphone to the 20-mg morphine equivalent ceiling of hydromorphone, or 3 to 4 mg of hydromorphone. If the patient develops side effects with hydromorphone, we suggest using fentanyl at concentrations of 2 to 3 mg/mL. Fentanyl, a lipophilic agent, is very well tolerated and has less supraspinal effects than morphine or hydromorphone.


Even if the cancer pain guidelines of the World Health Organization for pharmacologic tailoring of analgesics were used effectively, a large population of cancer patients would still suffer from pain. Clearly, in spite of these simple guidelines, too many patients are dying with pain.

It is imperative that physicians do not abandon the patient if the WHO guidelines do not work; rather abandon the therapies that do not work!

The broad range of interventional pain management approaches should be considered if pain is refractory.

Spinal analgesia is a costly and invasive alternative to the use of systemically administered opioids and should never be used as a primary system unless the use of systemic opioids is not an option. Patients who have opioid responsive pain and who do not tolerate orally administered opioids after sequential drug trials are candidates for a trial of intraspinal opioid therapy. Patients who have pain that is poorly responsive may be candidates for the intraspinal delivery of alternative, nonopioid medications such as local anesthetics or clonidine. An intraspinal trial can assess analgesic and functional outcomes, and toxicity. A continuous intrathecal trial extended for as long as possible is best because it mimics the intended implanted drug delivery system, allows for sequential trialing of differing agents, and may mitigate against placebo effects. The ultimate choice of a delivery system should consider prognosis, reimbursement, and medical issues.

Successful long-term management of intraspinal opioid therapy requires a skilled practitioner and a responsive system for monitoring. The appropriate use of these approaches adds immeasurably to our armamentarium for the treatment of intractable cancer pain.

Questions From the Audience

Samuel Hassenbusch, MD: I would like to make two comments. One, you had mentioned using fentanyl as a third-line opioid. You should note here that you, in your practice, compound fentanyl so that you can go to a higher concentration than allowed by what is commercial available at 50 µg/mL. One might use, as an alternative, sufentanil, because sufentanil is 10 times more potent than fentanyl on a weight to weight basis and 1,000 times more potent than morphine. The other comment that I would like to make refers to midazolam. The formulation of midazolam contains preservatives that preclude intraspinal use at this time.

Russell K. Portenoy, MD: Two questions: First, what do you mean when you say that a person with pain that is unresponsive to systemic opioids will not respond to intraspinal opioids? Second, why is a placebo test on an individual patient not useful?

If you’re actually measuring pain relief, quantitating it beyond just a quantal yes or no, then can’t you interpret it if the placebo gets a response of X but you get a response to the active drug of 3X?

Elliot S. Krames, MD: Please remember that I am only advocating the use of intraspinal therapy after failure of conservative therapy.

Let me answer your first question regarding opioid nonresponsiveness. Hypothetically, an oncologist is treating a patient with neuropathic pain who has lumbosacral plexus involvement. This patient is receiving high-dose, systemically administered opioids and is not responding. The patient is still screaming out in pain, in spite of receiving 2 g of morphine per 24 hours. At this point, the family and the oncologist are getting nervous–something must be done. The oncologist calls the anesthesiologist and says: “Put in a pump.” The dutiful anesthesiologist performs a trial, usually a single-shot epidural or intrathecal trial. He or she might even do a 3-day sequential trial, which appears to be what is mostly practiced, and the patient has miraculous improvement in his or her pain. Based on this trial, the pump is implanted and a few days later or even a week or two later the patient no longer has pain control. Sadly, this is not too uncommon. I hear of these cases almost every week from anesthesiologists around the world. They cannot understand the sudden loss of “miraculous pain control” and are at a loss of what to do.

Systemic opioid therapy was not working because the patient had a poorly responsive syndrome. Intraspinal opioids do not work when systemic opioids do not work! I want to emphasize that I am not saying that neuropathic pain is opioid nonresponsive, but if a single patient has a syndrome that is not responding to opioids at very high doses given systemically, that person is not going to respond for very long if you give the same agent intraspinally. This hypothetical patient responded early on to the nonspecific effects of the therapy, the placebo effect.

Now, the second question. Suppose you do a double-blind trial and the patient responds better to the drug than the saline. Well, if you’re going to do that, you had better add something that causes some side effect from the saline because anything that causes a side effect is going to have a higher placebo effect.


1. World Health Organization: Cancer Pain Relief. Geneva, WHO, 1986.

2. World Health Organization: Cancer Pain Relief and Palliative Care: Report of a WHO Expert Committee. Geneva, WHO, 1990.

3. Ventafridda V, Tamburini M, Caraceni A, et al: A validation study of the WHO method for cancer pain relief. Cancer 59:850-856, 1987.

4. Takeda F: Results of field testing in Japan of the WHO draft interim guidelines on relief of cancer pain. Pain Clin 1:83, 1986.

5. Schug SA, Zech D, Dorr U: Cancer pain management according to WHO guidelines. J Pain Symptom Manage 1:27-32, 1990.

6. Walker VA, Hoskin PJ, Hanks GW: Evaluation of WHO analgesic guidelines for cancer pain in a hospital-based palliative care unit. J Pain Symptom Manage 3:145-149, 1988.

7. Jacox A, Carr DB, Payne R, et al: Clinical Practice Guideline Number 9: Management of Cancer Pain. US Dept of Health and Human Services, Agency for Health Care Policy and Research; 1994, AHCPR publication 94-0592.

8. Portenoy RK: Inadequate outcome of opioid therapy for cancer pain: Influences on practitioners and patients, in Patt RB (ed): Cancer Pain, pp 119-128. Philadelphia, Lippincott, 1993.

9. Portenoy RK: Cancer pain—epidemiology and syndromes. Cancer 63(11 suppl):2298-2307, 1989.

10. Portenoy RK, Miransky J, Thaler HT, et al: Pain in ambulatory patients with lung or colon cancer: Prevalence, characteristics, and impact. Cancer 70:1616-1624, 1992.

11. Bon Roenn JH, Cleeland CS, Gonin R, et al: Physician attitudes and practice in cancer pain management; A survey from the Eastern Cooperative Oncology Group. Ann Intern Med 119:121-126, 1993.

12. Cleeland CS, Gonin R, Hatfield AK, et al: Pain and pain treatment in outpatients with metastatic cancer. N Engl J Med 330:592-596, 1994.

13. Foley KM: The treatment of cancer pain. N Engl J Med 313(2):84-95, 1985.

14. Hassenbusch S: Cost modeling for alternate routes of opioid administration for cancer pain. Presented at a symposium on Managing Pain in Patients With Advanced Cancer: The Role of Neuraxial Infusion. January 11, 1999, New York, New York.

15. Woolf CJ, Bennett, GJ, Doherty M, et al: Towards a mechanism-based classification of pain. Pain 77:227-229, 1998.

16. Lee EM: Pathogenesis and mechanisms of phantom pain. Curr Rev Pain 1:310-319, 1997.

17. Willis WD (Gebhard GF, Hammond DL, and Jensen TS, eds): Central Plastic Responses to Pain: Proceedings of the 7th World Congress on Pain, Progress in Pain Research, and Management, Vol 2, pp 301-324. Seattle, IASP Press, 1994.

18. Lipman AG: Pharmacologic approaches to neuropathic pain. Curr Rev Pain 1:285-295, 1997.

19. Meynadier J, Chrubasik J, Dubar M, et al: Intrathecal somatostatin in terminally ill patients: A report of two cases. Pain 23:9-12, 1985.

20. Penn RD, Paice JA, Kroin JS: Intrathecal octreotide for cancer pain. Lancet 335(8691):738, 1990.

21. Bowersox SS, Gadbois T, Singh T, et al: Selective N-type neuronal voltage-sensitive calcium channel blocker, SNX-111, produces spinal antinociception in rat models of acute, persistent, and neuropathic pain. J Pharm Exp Ther 279:1243-1249, 1996.

22. Bahar M, Cohen ML, Grinshpon Y, et al: Spinal anesthesia with midazolam in the rat. Can J Anaesth 44:208-215, 1997.

23. Goodchild CS: Nonopioid spinal analgesics; animal experimentation and implications for clinical developments. Pain Rev 4:33-58, 1997.

24. Coombs DW, Saunders RL, Lachance D, et al: Intrathecal morphine tolerance: Use of intrathecal clonidine, DADLE, and intraventricular morphine. Anesthesiology 62:358-363, 1985.

25. Van Dongen RTM, Crul BJP, de Bock M: Long-term intrathecal infusion of morphine and morphine/bupivacaine mixtures in the treatment of cancer pain: A retrospective analysis of 51 cases. Pain 55:107-111, 1993.

26. Lanning RM, Hrushesky WJ: Outpatient time-specified infusion of fluoropyrimidines by implanted pump is less costly than flat delivery by external pump. Prog Clin Biol Res 341:397-409, 1990.

27. Bedder MD, Burchiel JK, Larson A: Cost analysis of two implantable narcotic delivery systems. J Pain Symptom Manage 6:368-373, 1991.

28. Cousins MJ, Cherry DA, Gourlay GK: Acute and chronic pain: Use of spinal opioids, in Cousins MJ, Bridenbaugh PO (eds): Neural Blockade in Clinical Anesthesia and Management of Pain, 2nd edition, pp 955-1029. Philadelphia, J.B. Lippincott, 1988.