OUP user menu

Evidence For and Against the Use of Opioid Analgesics for Chronic Nonmalignant Low Back Pain: A Review

J. D. Bartleson MD
DOI: http://dx.doi.org/10.1046/j.1526-4637.2002.02043.x 260-271 First published online: 1 September 2002


Introduction. Opioid analgesics are very effective for treating pain, but their chronic use in nonmalignant conditions is controversial. Low back pain is a common condition, and chronic low back pain (CLBP) is the most frequent regional pain syndrome in the United States. This article reviews the evidence for and against the use of chronic opioid analgesic therapy (COAT) for patients with CLBP unrelated to cancer.

Methods. A literature review was conducted looking for reports of oral or transdermal opioid analgesic therapy for CLBP.

Results. There are very few randomized controlled trials of COAT for CLBP. The scant evidence that is available suggests that over the short-term, COAT is helpful with patients with CLBP. In the published reports, most of which are brief in duration, COAT is associated with moderate side effects but a low risk of abuse or drug addiction. COAT was not associated with adverse long-term sequelae. Longer-acting opioid analgesics may be preferable to shorter-acting agents. Patient selection and close follow-up are critical to good outcomes.

Conclusions. There is a place for the use of chronic oral or transdermal opioid analgesics in the treatment of some patients with CLBP.

  • Opioid Analgesics
  • Low Back Pain
  • Chronic Pain
  • Drug Therapy
  • Review


While opioid analgesics are the most effective pharmacologic agents available for reducing most types of pain, the chronic use of these drugs for noncancer pain is controversial. There are several reasons for the widespread reluctance to treat chronic nonmalignant pain with opioids. The first is the risk of side effects, including cognitive impairment, sedation, respiratory depression, nausea, constipation, edema, and hypogonadism. The second problem relates to the development of tolerance and the potential need to increase the medication dosage over time. A third problem is the risk of drug dependence and withdrawal phenomena if the drug is stopped or the dose reduced.

The “war on drugs” waged by society against all manner of habit-forming medications, both legal and illegal, adds to the pressure on health care providers to avoid prescribing opioid analgesics. We worry about drug diversion for illicit purposes. Our concerns are heightened by state and federal laws that scrutinize provider prescribing practices and carry with them the threat of prosecution and loss of licensure [1-4].

Another reason we avoid opioid analgesics is our desire to conform to patient and family preference to avoid strong medications that are habit forming. We hope that we can provide effective pain control with the use of lesser medications and nonpharmacologic therapies. Tertiary care center providers are appropriately reluctant to recommend long-term treatment with opioid analgesics in the course of a one-time evaluation, which could burden the patient's local provider with an obligation to prescribe opioid pain relievers on an ongoing basis. Opioid analgesics should not be prescribed without regular follow-up.

On the other hand, health care providers have an obligation to relieve pain and suffering. Some have called pain a fifth vital sign. In the case of chronic pain due to malignancy, providers are strongly encouraged and, in fact, legally bound to prescribe sufficient opioid analgesics to afford effective pain relief [3,4]. In most jurisdictions, the perception of regulatory review and punishment for prescribing opioids is far greater than the reality [4]. In recent years, states and multiple organizations have published recommendations, policies, consensus statements, and clinical practice guidelines that recommend more effective treatment of chronic nonmalignant pain and specifically include the use of opioid analgesics [2,4-7]. Because of these published recommendations, there is some risk of legal action against the provider if they do not adequately treat their patients' chronic nonmalignant pain.

Low back pain is a common condition, and chronic low back pain (CLBP) is the most common and costly persistent pain syndrome in the United States [8]. Should opioid analgesics be used to treat CLBP? This review will examine the evidence. While opioids have been shown to be helpful in the treatment of acute low back pain [9,10], this review focuses on their use in chronic, nonmalignant low back pain. While there is some anecdotal evidence regarding the use of intrathecal pumps, which administer opioid analgesics [11,12], this review is confined to the use of oral and transdermal opioid analgesics for CLBP.


MEDLINE (1966 through January, 2002) and EMBASE (1988 through January, 2002) databases were searched for relevant publications using the words “opioid analgesics,” “back pain,” “low back pain,” and “chronic pain.” The search was limited to human subjects and the English language. References of all reviewed articles were screened for additional relevant publications. Oral and transdermal therapies were considered, while intravenous and intrathecal therapies were not. Studies were included if the majority or a plurality of the patients reported had chronic nonmalignant low back pain. Only treatment trials of greater than seven days in duration were included. Simple retrospective surveys of patient experiences were excluded. Differentiation of presumed pain mechanism was not required for inclusion. Most of the reports did not discriminate between nociceptive, neuropathic, and idiopathic low back pain. In the few reports where an attempt was made to differentiate among pain mechanisms, low back pain was usually reported as nociceptive or “mixed.” While there is some evidence that tramadol is helpful for patients with CLBP [13-15], this medication is a weak μ opioid agonist with other actions, has a relatively low risk of producing dependence, and is not a controlled substance. For these reasons, tramadol was not included in this review.


Thirteen studies met the inclusion criteria, and these reports could be divided into the following categories: One randomized, double-blind trial [16]; one randomized, open-label trial [17]; three prospective, open-label, nonrandomized trials [18-20]; three retrospective reviews [21-25]; and five related reports, including the use of opioids in patients with osteoarthritis [26,27], many of whom had low back pain, and reports comparing different opioid analgesics in patients with low back pain [28-30]. The results are shown in Table 1 and are described in further detail below.

View this table:
Table 1

Evidence regarding opioid analgesic therapy for low back pain (LBP)

Author, yearNumber of patientsNumber of patients with LBPDrug(s)/doseStudy durationBenefitAdverse effectsType of study
Moulin et al., 1996 [16]6127 (44%)SR morphine (mean dose 83.5 mg/day) vs benztropine (active placebo)9 weeksPain better; psychologic and functional status unchangedVomiting, dizziness, constipation, anorexia/nausea, abdominal painRandomized, double-blind crossover of morphine vs benztropine (active placebo)
Jamison et al., 1998 [17]3636 (100%)Naproxen vs set-dose oxycodone vs titrated-dose oxycodone plus SR morphine (up to combined equivalent morphine dose of 200 mg/day)16-32 weeksBoth oxycodone groups had less pain and emotional distress; varied dosage and younger patients fared betterDry mouth, drowsiness, headache, constipation, nauseaRandomized, open-label comparison, then opioids for all
Schofferman, 1999 [18]3333 (100%)Methadone, levorphanol, SR morphine6- to 12-week trial followed by long-term use (mean 32 months)28 of 33 completed 6- to 12-week trial; 21 of 28 “responded” with less pain and improved functionNo toxicity or addictive behaviorOpen-label trial of several sequential opioids followed by long-term use in responders
Zenz et al., 1992 [19]10024 (24%)Buprenorphine (mean of 1.46 mg/ day), SR morphine (mean of 255 mg/ day), SR dihydrocodeine (mean of 170 mg/day)Mean duration 224 days (range 14-1,472)When measured 14 days after start of therapy, about half had ≥50% relief, 25% had 25-50% relief, and 25% had <25% relief of painConstipation and nausea were most common; responders had fewer adverse effectsOpen-label trial of several opioids
Haythornthwaite et al., 1998 [20]1913 (68%)Methadone or SR morphine (mean morphine equivalent dose of 111 mg/day)5.8 months in opioid group, 3 months in usual care groupLong-acting opioid group had less pain, anxiety, and hostility and no cognitive impairmentConstipation, drowsiness, nausea, pruritus, hypogonadismOpen-label trial of two long-acting opioids vs sample group of usual care
France et al., 1984 [21]1615 (94%)Codeine, oxycodone, methadoneMean follow-up was 13 months (range 6-22 months)Sustained pain relief in three fourths of patientsModest dose escalation in five patients; “minimal” side effectsRetrospective review of a subset of 16 of 150 pain management program patients
Portenoy, 1986 [22], 1989 [23], 1990 [24]5820 (34%)Methadone, oxycodone, othersMinimum 9 months; median therapy was 2-3 years“Adequate” pain relief in 16 of 39, partial relief in 16 of 39Side effects in 4 of 39, abuse behaviors in 4 of 39Retrospective review of patients in two New York City pain programs; outcome information available on 39 of 58 patients
Quang-Cantagrel et al., 2000 [25]8631 (36%)SR morphine, SR oxycodone, methadone, TD fentanyl, levorphanol, oxycodone, hydrocodoneUnclear; mean length of follow-up was 8.8 months≥50% pain relief in 81%Nausea/vomiting, sedation, pruritus most common; one with addictionRetrospective review
Roth et al., 2000 [26]133 (osteo-arthritis)61 (46%)Placebo vs 20 mg SR oxycodone/day vs 40 mg SR oxycodone/day2-week trial followed by 6-18 months open-label study40 mg/day oxycodone improved pain at 2 weeks; benefit maintained in open-label study87 of 133 experienced an adverse effect including nausea, constipation, somnolence, dizzinessRandomized, double-blind, 2-week trial, then open-label
Caldwell et al., 1999 [27]167 (osteo-arthritis)82 (49%)Open-label IR oxycodone 20-60 mg/day followed by placebo vs SR oxycodone vs IR oxycodone with APAP (mean oxycodone dose about 40 mg/day)30-day open-label, then 30-day study107 patients entered randomized trial; both oxycodone groups had improved pain and sleepSide effects included somnolence, constipation, nausea, pruritus, dizziness, dry mouthOpen-label, immediate-release opioid trial followed by randomized, double-blind, placebo-controlled 30-day study
Simpson et al., 1997 [28]6868 (100%)Various short-acting opioids compared with TD fentanyl (usual dose 25 μg/hour)30 days50 of 68 patients completed study; pain and disability were significantly improved with TD fentanylNausea, constipation, skin reaction, dizziness, sleepinessOpen-label, crossover comparison of TD fentanyl with usual short-acting opioid therapy
Allan et al., 2001 [29]256102 (40%)At end of study, mean fentanyl dose was 57.3 μg/hour and mean morphine dose was 133.1 mg/day4 weeks of each opioidTD fentanyl preferred over SR morphine41% had skin problems with fentanyl; more had constipation with morphine; 10% withdrew due to adverse event with fentanyl vs 5% with morphineRandomized, cross-over comparison of TD fentanyl and SR morphine
Milligan et al., 2001 [30]532 (includes 103 of Allan's patients)237 (44.5%)Mean fentanyl dose increased from 48 to 90 μg/hour during studyUp to 12 months67% reported very good, good, or moderate pain control231 patients (43% of 532) stopped treatment, 130 (56%) due to adverse events. Nausea, constipation, and somnolence most common; 3% withdrawal syndrome, 1% abuseOpen-label trial of TD fentanyl
  • SR = sustained-release or controlled-release

  • TD = transdermal

  • IR = immediate-release

  • APAP = acetaminophen

Randomized, Double-Blind Control Study

Moulin et al. [16] compared oral sustained-release morphine with benztropine (an active placebo) in the treatment of patients with non-neuropathic, chronic, regional pain of soft tissue or musculoskeletal origin who did not respond to anti-inflammatory agents and at least one tricyclic antidepressant. Patients were excluded if they had previously used a “major” opioid analgesic for their chronic pain. Forty-four percent of the patients had CLBP. Sixty-one patients entered the study; 15 dropped out due to inadequate pain relief and/or unacceptable side effects. Morphine doses ranged from 30 to 120 mg/day (mean daily dose was 83.5 mg/day). In this randomized, double-blind, crossover study, nine weeks of oral morphine conferred analgesic benefit with a low risk of addiction but did not yield psychological or functional improvement. Side effects seen more frequently in the morphine-treated group included vomiting, dizziness, constipation, anorexia/nausea, and abdominal pain.

Randomized, Open-Label Trial

Jamison et al. [17] treated 36 patients with chronic noncancer low back pain in a randomized, open-label trial that compared naproxen alone (up to 1,000 mg/day) with a set dose of oxycodone (up to 20 mg/day) and with a titrated dose of oxycodone plus sustained-release morphine with the amount of medication adjusted according to the intensity of pain (up to the combined equivalent of 200 mg of morphine per day). After a one-month washout period of no opioid medication, patients were treated with one of the above regimens for 16 weeks, at which point all patients were offered the combination of oxycodone and sustained-release morphine at a dose titrated according to their pain intensity and ability to tolerate the medication for an additional 16 weeks. The average daily opioid dose during this phase was 61.6 mg of morphine equivalent. Thereafter, the opioid dose was gradually tapered over 12 weeks, and patients were monitored for an additional four weeks while taking no narcotic analgesics. The titrated-dose group had significantly less pain and less emotional distress (anxiety, depression, and irritability) than the other two groups. Both opioid groups had significantly less pain and emotional distress than the naproxen-only group. The authors found no differences in activity level or hours of sleep. Patients receiving opioid analgesics had more dry mouth, drowsiness, headache, constipation, and nausea. In three cases, opioids were discontinued because of intolerance to the medication. Patients who varied their medication dose from week to week reported less pain and better mood than those who did not. Younger patients seemed to prefer and better tolerate opioids when compared with older persons.

Nonrandomized, Open-Label Trial

Schofferman [18] treated 33 CLBP patients with 6 to 12 weeks of a long-acting oral opioid supplemented with an immediate-release opioid for breakthrough pain. If the first opioid was not successful, up to three additional opioids were tried before determining the trial was a failure. Those who improved were treated long term. Of the 33, five (15%) had intolerable side effects, and 28 (85%) completed the 6- to 12-week trial. There was significant overall improvement in the 28 patients who completed the 6- to 12-week trial; 21 patients responded to the opioids and seven did not. The 21 “responders” were treated with opioid analgesics long term (mean of 32 months). Improvement was sustained in this group, and two patients were able to return to work. Compared with the dropouts and failures, the 21 patients who were treated had improved pain and function without toxicity or addictive behavior. Doses remained stable over time. Analgesics used included methadone, levorphanol, sustained-release morphine, and other opioids. The author concluded that long-term opioid analgesic therapy was reasonable for some well-selected patients with refractory low back pain who had failed all other forms of care.

Zenz et al. [19] reported on 100 patients chronically given oral opioids for nonmalignant pain. Patients received buprenorphine, sustained-release morphine, or sustained-release dihydrocodeine. Twenty-four of the 100 patients had back pain; this is one of the few articles that separated response according to site and type of pain. When pain was measured two weeks after initiation of opioid therapy, the results were similar for those with low back pain and the group as a whole; about half of patients had ≥50% (good) relief of pain, 25% had between 25% and 50% (partial) relief of pain, and 25% had less than 25% pain relief (nonresponders). In the 23 patients (of the original 100) assessed after one year of therapy, the results were even better, presumably because nonresponders and partial responders had dropped out. Interestingly, in this study, there was no difference in response between neuropathic pain, back pain, and osteoporosis pain patients, whereas patients with head and face pain were less likely to respond. The most common side effects were constipation and nausea. There were no cases of respiratory depression or addiction to opioids.

Haythornthwaite et al. [20] prospectively studied a group of 19 patients with chronic pain who were partially self-selected and treated with long-acting opioids (methadone or sustained-release morphine) and compared them with a group of nonrandomized, “usual care” patients who could be taking short-acting but not long-acting opioid analgesics. Sixty-eight percent of the treatment group had back pain. In the treatment group, 5% had nociceptive-only pain, 16% had neuropathic-only pain, and 79% had a mixed type of pain. Eighty percent of patients in both groups were using short-acting opioid medications at baseline. Follow-up evaluations were conducted after nearly six months in the treatment group and three months in the usual care group. Despite having greater baseline pain, patients treated with long-acting opioid analgesics improved substantially more than patients in the usual care group. In addition, the active treatment group experienced less anxiety and hostility compared with the usual treatment group, and there was no adverse effect on cognitive function in the long-acting opioid treatment group, even though they were taking a mean morphine equivalent dose of 111 mg per day. Patients whose pain improved did so at a lower morphine equivalent dose than did nonresponders (87.5 mg/day versus 151.4 mg/day). The most common side effects were constipation, drowsiness, nausea, pruritus, and hypogonadism. The authors concluded that long-acting opioids can improve mood and do not impair cognitive functioning, while reducing the level of pain in patients with chronic noncancer pain.

Retrospective Reviews

France et al. [21] reported on 16 patients out of 150 treated in a pain management program who were found to need continued opioid analgesics on a long-term basis. Fifteen of the 16 had low back pain. Pain relief from the opioid analgesics was sustained in three fourths of the patients, which allowed them to “pursue meaningful activities at home or work.” Analgesics used included codeine, oxycodone, and methadone. Modest escalation of dose occurred in five patients, and side effects were “minimal.” The mean follow-up was 13 months (range 6-22 months).

Portenoy, in three different publications [22-24], described 58 patients who were treated with chronic opioid therapy at two different New York City pain services. Twenty of the 58 (34%) had low back pain. Methadone and oxycodone were the most commonly used opioids. Adequate outcome information was available for 39 of the 58 patients, and presumably, CLBP was still the most common diagnosis. Methadone and oxycodone were most commonly used, and treatment duration ranged from nine months to>10 years. Adequate relief of pain was achieved in 16 (41%) of the 39 patients, and partial relief in an additional 16 (41%). Improvements in activity, employment, sociability, and mood could not be determined. Four of the 39 experienced persistent, distressing opioid side effects (personality change in two and myoclonus in two) and four other patients developed abuse behaviors. The author concluded that there is a subpopulation of pain patients who may be able to obtain at least partial analgesia from chronic opioid therapy.

Quang-Cantagrel et al. [25] retrospectively reviewed 86 patients treated with a variety of opioid analgesics for various types of noncancer pain. Back pain was the single largest contributor, accounting for 31 (36%) of the patients. During a mean follow-up of 8.8 months, the number of different opioids used by each patient averaged 2.3. Effectiveness was defined as ≥50% pain relief. The first opioid prescribed was effective for 31 (36%) of the 86 patients and was stopped because of side effects in 25 (29%) and because of ineffectiveness in 29 (34%). Fifty-two of the 54 patients who stopped the first opioid for intolerable side effects or ineffectiveness received a different opioid, and of this group, 16 (31%) improved, 14 (27%) stopped because of intolerable side effects, and 20 (38%) stopped because of ineffectiveness. Of 30 patients who did not respond or reacted adversely to one of the first two opioids, 12 (40%) improved when given a third different opioid. Among the 18 remaining patients who had not responded to the first three opioids, a fourth was effective in 10 (56%). A fifth opioid, prescribed to seven patients, was effective for 1 (14%). Thus, 70 (81%) out of 86 patients obtained effective pain relief with one or another opioid analgesic. The comparable improvement numbers were 27 (87%) of the 31 patients with CLBP. If a patient had significant side effects from one opioid analgesic, they did not necessarily have the same side effects with another opioid. The authors concluded that if a patient with chronic pain had intolerable side effects or failed to respond to one opioid analgesic, trying a different opioid could result in lessening or elimination of the side effect or improvement in the patient's pain.

Placebo-Controlled Trials in Patients with Osteoarthritis

Two recent placebo-controlled trials of oxycodone for patients with osteoarthritis pain have some bearing on the treatment of patients with CLBP [26,27]. One was conducted without [26] and one was conducted after an open-label trial to determine if oxycodone provided benefit without significant side effects [27].

Roth et al. [26] treated 133 patients with osteoarthritis-related pain (61 or 46% with “spine or back pain”) in a randomized, double-blind, two-week trial with either: 1) Placebo; 2) 10 mg of controlled-release oxycodone every 12 hours; or 3) 20 mg of controlled-release oxycodone every 12 hours. The higher dose of controlled-release oxycodone was superior to the lower dose or placebo in improving pain and reducing the adverse effects of pain on mood, sleep, and enjoyment of life. Seventy patients (53%) discontinued study participation prematurely—more because of lack of benefit in the placebo group and more because of side effects in the active treatment groups. Side effects occurred in 87 (65%) of the 133 patients and included nausea, constipation, somnolence, and dizziness. After the two-week study, 106 patients participated in a long-term, open-label trial of at least six months' duration. During the long-term trial, patients' controlled-released oxycodone doses remained stable at about 40 mg/day, and patients continued to obtain benefit. Of the 106 patients who entered the open-label portion of the study, 60 (57%) discontinued treatment, and the most frequent reason was an adverse event (32 patients). The most common adverse events causing discontinuation of therapy were constipation, nausea, pruritus, somnolence, and nervousness.

Caldwell et al. [27] treated 167 patients with moderate-to-severe osteoarthritis pain (82 or 49% with “back and neck pain”) for 30 days with open-label, immediate-release oxycodone in a dose of 20-60 mg/day. One hundred seven (64%) whose pain improved without significant side effects were randomized to a 30-day trial of either: 1) Placebo; 2) Controlled-release oxycodone; or 3) Immediate-release oxycodone with acetaminophen. The oxycodone dosage (mean of about 40 mg/day) was based on titration during the open-label trial. There was no washout period. Both forms of oxycodone, when added to the patient's usual nonsteroidal anti-inflammatory drug, were superior to placebo for reducing pain and improving quality of sleep. Side effects included somnolence, constipation, nausea, pruritus, dizziness, and dry mouth. Controlled-release oxycodone resulted in lower incidences of nausea and dry mouth.

Nonrandomized Prospective Comparison Studies

There have been three recent articles that prospectively compared one or two opioid analgesics with the patient's previous treatment, which included opioid analgesic therapy for most or all of the studied patients [28-30].

Simpson et al. [28] took 68 patients on various short-acting oral opioids for CLBP whose pain was inadequately controlled and switched them to transdermal fentanyl for one month, after which the oral opioid was resumed. The most common dose of fentanyl was 25 μg/hr. Fifty patients (74%) completed the study; 43 (86%) of the 50 reported overall benefit from transdermal fentanyl. Pain and disability were significantly less on transdermal fentanyl than on oral opioids. “Many” patients (nearly half of those remaining in the study) reported “minor” side effects, including nausea, constipation, local skin reaction, dizziness, and sleepiness.

Allan et al. [29] compared transdermal fentanyl with sustained-release oral morphine in the treatment of 256 patients with chronic noncancer pain, 102 (40%) of whom had low back pain. The patients, all of whom were already on potent opioids, including 76% on morphine, were randomized to receive four weeks of oral sustained-release morphine followed immediately by four weeks of transdermal fentanyl or vice versa. Dosage was based on pretrial opioid use and titrated upward, if needed. Although only 196 patients (77%) completed the trial, patient preference could be analyzed in 212 patients (83%). One hundred and thirty-eight (65 %) of 212 patients preferred transdermal fentanyl, while 59 (28%) preferred sustained-release oral morphine, and 15 (7%) had no preference. Patient preference for fentanyl was similar in patients with nociceptive, neuropathic, and mixed pain. Pain control was better with fentanyl than with morphine. More patients experienced constipation with morphine than with fentanyl (48% versus 29%), but 41% of patients experienced mild or moderate cutaneous problems associated with wearing the transdermal fentanyl patches. More patients withdrew because of adverse events occurring with fentanyl treatment than with morphine (10% versus 5%).

Milligan et al. [30] studied the efficacy and safety of up to 12 months of therapy with transdermal fentanyl in an international, multicenter, open-label trial of 532 patients with chronic noncancer pain severe enough to require continuous treatment with a potent opioid analgesic. Of the total, 237 (44.5%) had low back pain. One hundred three of the patients reported by Allan [29] were also recruited for this study. While the authors did divide the patients into type of pain, many patients had more than one type of pain, categorization of the patients with low back pain was not included, and results were not reported by type of pain. Immediately before entering the study, 48% of patients were taking morphine and 28% fentanyl. In essence, this study compared transdermal fentanyl with the patient's previous opioid analgesic therapy. There were 301 (57%) patients who completed the trial, and 231 (43%) who discontinued treatment prematurely. The most common reason for stopping transdermal fentanyl was an adverse event occurring in 130 of 231 (56%) patients. Of the 130 patients who discontinued fentanyl because of an adverse event, 38 (29%) did so because of nausea, 26 (20%) due to vomiting, 20 (15%) due to somnolence, 13 (10%) due to dizziness, 13 (10%) due to increased sweating, and nine (7%) due to anorexia. The mean fentanyl starting dose was 48 μg/hr, which increased to 90 μg/hr at month 12. Five hundred twenty-four (98%) of 532 total patients were included in the efficacy analysis because they had been assessed at baseline and at least once thereafter. Sixty-seven percent of the efficacy analysis group reported very good, good, or moderate pain control on fentanyl. “Global satisfaction” of very good or good was reported by 42% of patients and was stable over the 12 months of follow-up. Of those patients not previously on fentanyl, 67% preferred fentanyl to their previous opioid. There was no change overall on a quality-of-life measure. Nausea, constipation, somnolence, and increased sweating were the most common side effects. There was a 3% incidence of an opioid withdrawal syndrome and a 1% incidence of drug abuse/dependence.


This review is intended to focus on patients with nonmalignant CLBP and not on patients with nonmalignant pain in general. Patients present with distinct pain syndromes, and providers need to know if a specific therapy is indicated for their patient's problem.

Low back pain may be best viewed as a regional pain syndrome without a uniform underlying pathophysiology. Does it matter that the studies reviewed do not, for the most part, differentiate between nociceptive and neuropathic pain? In the past, it was thought that opioid analgesics were not as effective for neuropathic pain as for nociceptive pain, but evidence is mounting in recent years that opioids are helpful for both kinds of pain [31-34]. Most CLBP is probably musculoskeletal and mechanical in nature, and thus, nociceptive. Many patients with low back pain have additional lower extremity pain, which can indicate nerve root involvement, and thus, they may have an additional, neuropathic component. In view of the presumed nociceptive or mixed-type pain in patients with CLBP and the mounting evidence that opioid analgesics can benefit neuropathic as well as nociceptive pain, the reported results are likely valid even though patients were not subcategorized by presumed pain mechanism.

The evidence presented is flawed. Some of the studies examined patients with low back pain alone, while some of the reports included more than one kind of pain. There is but a single randomized, double-blind controlled study, which was of nine weeks' duration, and only 44% of the patients in that study had CLBP [16]. Unfortunately, the study did not stratify patients according to site of their soft tissue or musculoskeletal pain. Presumably, CLBP patients' pain relief was similar to the group as a whole. Thus, that study clearly supports the use of opioid analgesics in patients with CLBP as do the other reports. The short durations of the prospective treatment trials lasting weeks to months are in contrast to the chronicity of the patients' low back pain [16-20,26-30]. Of note, there are two published reports of even shorter duration that show benefit from oxycodone, for patients with CLBP [35], and from controlled-release codeine, in patients with various kinds of chronic, nonmalignant pain [36]. These two studies did not meet inclusion criteria for this review because the active treatment periods were one week or less. The durability of the benefit and relative lack of late side effects and complications are supported by the retrospective reviews [21-25]. Many of the reports are either not controlled or poorly controlled. Most are unblinded. Some studies conducted a controlled phase only after response to a drug had been demonstrated. As noted above, some studies lumped patients with several different regional pain syndromes. Some of the reports grouped opioid analgesics and some examined a single medication. Some reports compared different opioid analgesics rather than comparing opioid analgesics with nonopioids. Despite these shortcomings, the weight of evidence favors the use of opioid analgesics in some patients with CLBP.

Which opioid analgesic should be chosen? The literature is unclear. Morphine, fentanyl, oxycodone, and methadone are the opioids that have been most commonly used and recommended. Most of the published studies have used long-acting opioid analgesics, and some authors are explicit in recommending round-the-clock, regular use of long-acting opioid analgesics with supplementary use of short-acting opioids on an as-needed basis [1,37]. Three of the articles reviewed [28-30] suggested a patient preference for transdermal fentanyl. However, it should be noted that, in all three studies, the patients had been taking alternative, usually oral, opioid analgesics before entering each study. Preference for fentanyl could simply demonstrate improvement when patients were given a new opioid as was found by Quang-Cantagrel et al. [25].

Although the durations of the reported trials of opioid analgesics for CLBP are relatively short, it is noteworthy that the side effects reported, while frequent, were all temporary and resolved after stopping the medication. In the longer-duration studies lasting 12 months to several years, “abuse behaviors” or withdrawal symptoms developed in 1-10% of patients [21-24,30], indicating that there are risks associated with COAT. Nonetheless, the long-term use of opioid analgesics is associated with very few permanent sequelae. This is in contrast to the situation with the long-term use of nonopioid analgesics, which can be associated with damage to the kidneys, in the case of nonsteroidal anti-inflammatory drugs, and the liver, in the case of acetaminophen. The lack of long-term adverse effects with opioid analgesics also differs from operative interventions, such as laminectomy, fusion, placement of a dorsal column stimulator, or even placement of a morphine pump, where associated complications can be permanent. Because of the lack of long-term, irreversible consequences, the use of COAT for CLBP should at least be considered before invasive surgical procedures are performed.

The results of this review are in keeping with previous, similar reviews [33,37-39], which recommend opioid analgesics for the treatment of chronic, nonmalignant pain, including musculoskeletal pain [38] and CLBP [37]. These articles also outline the relative contraindications to COAT including preexisting substance abuse disorders, personality disorders, certain medical conditions, and certain occupational factors. The articles also review methods to prevent and monitor for drug-related problems, including patient and family education, frequent visits, use of patient-provider treatment contracts, family interviews to assess the patient's condition, and drug testing. Patient-provider contracts are written and signed agreements that detail the risks, hoped-for and realistic benefits, and specific parameters of use of the prescribed opioid analgesic. Often included are requirements for frequent follow-up visits, accounting of all doses prescribed, and insistence on a single prescribing provider and a single dispensing pharmacy [40,41].

The use of opioid analgesics for chronic nonmalignant pain, in general, remains controversial. In recent years, numerous authors have clearly described the issues involved [42-45]. Of interest, these authors all recommend the use of opioid analgesics, if needed, for patients with chronic nonmalignant pain. Important outcomes to be monitored include pain relief, side effects, physical and psychosocial functioning, and certain aberrant drug-related behaviors [42,45].

Could opioid analgesic therapy worsen rather than improve pain? While more commonly reported in animals than in humans, allodynia and hyperalgesia have been observed with several opioids [46-48]. The increased pain may be due to high doses of the opioid, toxic levels of opioid metabolites, or opioid withdrawal. Increased pain related to COAT was not reported in any of the studies reviewed for this article.

While some patients with CLBP obtain benefit from COAT, it is clear from the articles reviewed that COAT does not eliminate discomfort in CLBP patients. Many of the patients reported in these studies were receiving additional treatments. In order to optimize patient outcomes, it is reasonable and appropriate to use opioid analgesics with other pharmacologic and nonpharmacologic therapies in patients with CLBP.

Can we do a better job of assessing the role of COAT for patients with CLBP in future studies? There is no lack of patients with CLBP to study. Double-blind controlled studies are preferred, but difficult to carry out with medications that have significant side effects. Using an “active” placebo, as Moulin et al. [16] did, makes sense. Clearly, studies of longer duration are needed to evaluate therapy for a condition that is chronic. How long is long enough?—at least a year, if not longer. Because of side effects and lack of benefit, the drop-out rates in the reported studies are very high. This makes it difficult to draw conclusions on an “intention-to-treat” basis. In part because of the high drop-out rate, large numbers of patients are needed to adequately assess the potential benefit of COAT for CLBP. In addition to measuring pain, future studies must assess the effect of treatment on functional status, quality of life, and ability to work. Ample baseline information about the patients and the cause of their CLBP must be collected in order to help determine which patients are likely and which are unlikely to benefit from COAT. With the increasing number of spine centers caring for large numbers of patients with CLBP, it should be possible to conduct additional prospective studies of longer duration with adequate numbers of enrollees.


While the objective evidence is weak, there is a place for the use of chronic oral or transdermal opioid analgesics in the treatment of patients with CLBP. While we cannot yet identify them in advance, there appears to be a subset of patients for whom these medications provide substantial pain relief, improve quality of life, and restore the ability to function on a day-to-day basis. For some, this can mean the difference between working and going on disability. Longer-acting agents may be preferable to shorter-acting drugs, although the latter can be useful for breakthrough pain. Treatment contracts can be helpful. Patients treated long term with opioid analgesics need to be closely monitored to ensure that the benefits obtained outweigh any adverse effects.


The author is most appreciative of the superb assistance of Linda A. Schmidt.


View Abstract