Intervention Review

You have free access to this content

Opioids compared to placebo or other treatments for chronic low-back pain

  1. Luis Enrique Chaparro1,*,
  2. Andrea D Furlan2,
  3. Amol Deshpande3,
  4. Angela Mailis-Gagnon4,
  5. Steven Atlas5,
  6. Dennis C Turk6

Editorial Group: Cochrane Back and Neck Group

Published Online: 27 AUG 2013

Assessed as up-to-date: 1 APR 2013

DOI: 10.1002/14651858.CD004959.pub4


How to Cite

Chaparro LE, Furlan AD, Deshpande A, Mailis-Gagnon A, Atlas S, Turk DC. Opioids compared to placebo or other treatments for chronic low-back pain. Cochrane Database of Systematic Reviews 2013, Issue 8. Art. No.: CD004959. DOI: 10.1002/14651858.CD004959.pub4.

Author Information

  1. 1

    Hospital Pablo Tobon Uribe, Anesthesiology Department, Medellin, Colombia

  2. 2

    Institute for Work & Health, Toronto, ON, Canada

  3. 3

    University Health Network, TWH-Comprehensive Pain Unit, Toronto, ON, Canada

  4. 4

    Toronto Western Hospital Comprehensive Pain Program, Department of Medicine, Toronto, Ontario, Canada

  5. 5

    Massachusetts General Hospital, Medical Practices Evaluation Center, Boston, MA, USA

  6. 6

    University of Washington, Department of Anesthesiology and Pain Medicine, Seattle, Washington, USA

*Luis Enrique Chaparro, Anesthesiology Department, Hospital Pablo Tobon Uribe, Medellin, Colombia. luisdr74@yahoo.com.

Publication History

  1. Publication Status: New search for studies and content updated (no change to conclusions)
  2. Published Online: 27 AUG 2013

SEARCH

 
Characteristics of included studies [ordered by study ID]
Buynak 2010

MethodsMulticentre, randomized, double-blind, placebo-controlled, parallel, three arms design for 15 weeks (3 week titration and 12 week maintenance period).


ParticipantsAdults with LBP for ≥ 3 months were included. The study required participants to be dissatisfied with their current treatment, and to have a baseline pain intensity of ≥ 5/10. Daily doses of opioids had to be equivalent to ≤ 160 mg of oral morphine in the opioid users.


InterventionsTapentadol ER Group: during the titration period (3 weeks) started at 50 mg BID and 3 days later 100 mg BID. Dosing was adjusted each 3 days as required up to a maximum dose of 250 mg BID.

Oxycodone HCL CR: during the titration period (3 weeks) started at 10 mg BID and 3 days later 20 mg BID. Dosing was adjusted at a minimum of 3 day intervals as required up to a maximum dose of 50 mg BID.

Placebo capsules and tablets were administered to maintain blinding to the intervention.


OutcomesTwo different primary efficacy endpoints: change from baseline in mean pain intensity at week 12 of the maintenance period (week 15 of the study; US primary endpoint) or change from baseline in mean pain intensity over the entire 12 week maintenance period (European Union and other regions’ primary endpoint).

Secondary outcomes: Brief Pain Inventory (BPI), Short Form-36 health survey (SF-36), euroQol-5 Dimension health questionnaire, sleep questionnaire, patient's global impression of change and % patients who responded with 30% or 50% reduction in pain intensity.


NotesA significant proportion of patients from all 3 groups dropped out of the study. We received additional (unpublished) information from the authors that we used in the meta-analysis.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk"Randomization of patients to treatment was based on a computer-generated randomization list, balanced by randomly permuted blocks, and stratified by study site".

Allocation concealment (selection bias)Low risk"Randomization was implemented through an interactive voice response system (IVRS) that assigned patients to blinded study medication".

Blinding (performance bias and detection bias)
All outcomes - patients
Low riskThe study authors used placebo and capsules (one for each active treatment) to maintain the blind in this double-blind, double-dummy design.

Blinding (performance bias and detection bias)
All outcomes - providers
Low risk"Investigators were not provided with the randomization codes and the schedule was maintained with the Interactive Voice Response System".

Blinding (performance bias and detection bias)
All outcomes - outcome assessors
Unclear riskIt is not clear if the outcomes assessors for efficacy were the same ones for safety. The side effect profile of the medication could induce bias.

Incomplete outcome data (attrition bias)
All outcomes - drop-outs
High risk% drop-outs exceeded 20% in each group: 167/319 (52%) in the placebo group; 155/318 (48.7%) in the tapentadol group; 195/328 (59.4%) in the oxycodone group.

Incomplete outcome data (attrition bias)
All outcomes - ITT analysis?
High riskThe study authors performed ITT analysis using last observation carried forward, which can increase the risk of bias.

Selective reporting (reporting bias)Low riskThe trial was registered at clinicaltrials.gov (NCT00449176); the primary and the secondary outcomes were consistent in the protocol compared with the publication.

Group similarity at baselineLow riskPatients did not differ in the baseline characteristics based on the reported table 1.

Influence of co-interventionsLow riskOnly acetaminophen was allowed across the groups.

Compliance with interventions      Unclear riskThe authors reported a high rate of non-compliance (based on the flow chart).

Timing of outcome assessmentsLow riskTiming of outcomes assessment was identical in both groups.

Chu 2012

MethodsSingle-centre, randomized, double blind, placebo-controlled, parallel design for 1 month.


ParticipantsAdults with moderate to severe CLBP. Eligible patients were between ages 18 and 70 years; diagnosed with chronic nonmalignant, nonradicular LBP of at least 6 months duration. Participants were not currently taking opioid pain medication in excess of 30 mg oral morphine equivalents per day, which the research group defined as low-dose opioid therapy.


InterventionsTitrated oral morphine starting at 15 mg twice per day, followed every 2 days by a dose increase of 1 capsule per day, if tolerated, until (1) adequate analgesia (as determined by the subjects) had been achieved, (2) side effects (severe sedation, nausea or vomiting, constipation, sleep disturbances) limited further titration, or (3) a total of 8 capsules (120 mg/d of oral morphine if on active treatment) had been reached.


OutcomesThe primary outcome measure was opioid-induced hyperalgesia using cold pain and heat pain tolerance. The study also measured analgesic tolerance, pain scores, RMDQ, BDI, as well as symptoms and signs of opioid withdrawal.


NotesThe study was performed to distinguish between two different long-term effects of opioids: hyperalgesia versus tolerance.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear risk"Patients were randomly assigned to receive either sustained acting morphine (15 mg MS-Contin; Purdue Pharma, Stamford, CT) or weight-matched placebo capsules". There was no description of the method of randomization.

Allocation concealment (selection bias)Unclear riskNot clearly stated: "Of the 139 randomized patients, 69 were allocated to the morphine group and 70 were allocated to the placebo group".

Blinding (performance bias and detection bias)
All outcomes - patients
Low risk"Study drugs were encapsulated in an opaque blinding capsule (DBCaps; Capsugel, Peapack, NJ) to ensure adequate blinding of the study medications".

Blinding (performance bias and detection bias)
All outcomes - providers
Unclear riskUnclear whether the researchers took any approach to blind the clinicians who monitored the opioid titration.

Blinding (performance bias and detection bias)
All outcomes - outcome assessors
Unclear riskUnclear whether the researchers took any approach to blind the outcomes assessors.

Incomplete outcome data (attrition bias)
All outcomes - drop-outs
High risk% drop-outs exceeded 20% in each group: 21/69 (30.4%) in the morphine group; 15/70 (21.4%) in the placebo group.

Incomplete outcome data (attrition bias)
All outcomes - ITT analysis?
Unclear riskUnclear from the text if the pain scores reported came from only those who completed the trial or all participants.

Selective reporting (reporting bias)Low riskThe trial was registered at clinicaltrials.gov (NCT00246532); the primary and the secondary outcomes were consistent in the protocol compared with the publication.

Group similarity at baselineLow riskGroups were comparable regarding the most important demographic characteristics.

Influence of co-interventionsLow risk"no patients enrolled in this study were using anticonvulsant or antidepressant drugs"

Compliance with interventions      Unclear riskA higher number of participants in the morphine group discontinued the treatment compare to the patients assigned to the placebo group.

Timing of outcome assessmentsLow riskTiming of outcomes assessment was identical in both groups: 1 month.

Gordon 2010

MethodsMulticentre, randomized, double-blind, placebo-controlled, cross-over study; 4 week period. Patients who completed the trial were eligible for a 6-month open-label phase.


ParticipantsAdults reporting LBP, at least moderate in intensity, for more than 3 months and requiring more than one tablet of opioids. Patients with pain refractory to opioids were excluded as well as those with previous surgical/invasive interventions, narcotics or alcohol abusers, or those with a significant cardiovascular, pulmonary, liver or gastrointestinal disease.


InterventionsPatients underwent a 2 to 7 day washout of opioid analgesia before receiving patches of buprenorphine 10 μg/h or matching placebo patches. All patches were to be worn for 6 to 8 days. The initial dose was titrated weekly to 20 mcg/h and a maximum of 40 mcg/h using 10- and 20-mcg/h patches based on pain relief and adverse events.


OutcomesPain diaries (unmarked 0-100 mm VAS) two times/day and a 5-point ordinal scale; sleep questionnaires; PDI; Quebec Back Pain Disability Scale (QBPDI); SF-36 health survey; 0 to 3 treatment effectiveness; period 1, 2, none, preference; 1 to 4 ordinal scale for benefit; Subjective Opioid Withdrawal Scale; 1 to 3 ordinal scale for side effects.


NotesSome important demographics were not described such as work status, disability or low-back diagnosis. We received additional (unpublished) information from the authors that we used in the meta-analysis.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk"The randomization code was generated using PROC PLAN in SAS version 6.12 (SAS Institute Inc., Cary, North Carolina)".

Allocation concealment (selection bias)Low risk"A block-randomization procedure was used to generate the treatment allocations: for every 4 successive patients, 2 received BTDS in the first phase and 2 received BTDS in the second phase. Study monitors, investigators, coordinators, pharmacists, patients, and sponsor clinical research personnel remained blinded to treatment allocation throughout the conduct of the study".

Blinding (performance bias and detection bias)
All outcomes - patients
Low riskThe study used matching placebo patches.

Blinding (performance bias and detection bias)
All outcomes - providers
Low riskClinical personnel remained blinded to treatment allocation.

Blinding (performance bias and detection bias)
All outcomes - outcome assessors
Unclear riskResearch personnel remained blinded to treatment allocation. Opioid withdrawal symptoms might induce bias, but investigators reported no symptoms. However, a significant difference in nausea was reported in the adverse effects table that could induce bias.

Incomplete outcome data (attrition bias)
All outcomes - drop-outs
High risk% drop-outs exceeded 20%: 29/78 (37.1%) participants did not complete the 8 weeks of treatment.

Incomplete outcome data (attrition bias)
All outcomes - ITT analysis?
Low riskThe authors performed ITT and per-protocol analyses.

Selective reporting (reporting bias)Low riskThe trial was registered at Current Controlled Trials: ISRCTN 06013881; the primary and the secondary outcomes were consistent in the protocol compared with the publication.

Group similarity at baselineLow riskNot applicable (cross-over design).

Influence of co-interventionsLow riskNot applicable (cross-over design).

Compliance with interventions      Low riskThe authors reported that only one patient was non-compliant with the study medication.

Timing of outcome assessmentsLow riskTiming of outcomes assessment was identical in both groups.

Hale 2007

MethodsMulticentre, randomized, double-blind, placebo-controlled, parallel, enrichment design clinical trial.


Participants251 patients were screened and included for the titration stage; 47/101 discontinued the intervention due to adverse events and 10 due to lack of efficacy. 143 patients were randomized; 49 completed in the active group compared to 18 in the placebo group.

Inclusion criteria: 18 years old, moderate/severe CLBP for at least 3 hours/day for minimum of 3 months, receiving opioids (60mg/d morphine equivalent) for 2 weeks before screening. Patients excluded: pregnant or lactating women, secondary source of pain such as infection or tumour, back surgery within 6 months, suspected neoplasm, dysphagia, hypersensitivity to opioids, seizure history, colostomy. Average age: 46, 49.1 (placebo, open-label). Work status was not documented. Pain diagnosis: degenerative disc disease (DDD) (approx 40%), herniated disc (20%), osteoarthritis (OA) (20%), spinal stenosis (2-5%), trauma (17%), other (25%), NB assumed not mutually exclusive groups. Duration of LBP was not documented. Previous non-opioid treatments were not documented. Previous back surgeries were not documented.


InterventionsOxymorphone PO BID, range of opioid dose: 20-260 mg after open-label phase. Median opioid dose: 60 mg. Duration of treatment: 12 weeks during the double-blind period. Placebos were administered in the same fashion; no description of placebo type was provided.


OutcomesVAS Pain Scores. The increase from baseline (at randomization) to final visit was 31.6 mm for placebo versus 8.7 mm with OPANA ER (P <0.0001). During double-blind treatment, placebo patients were approximately 8-fold more likely than OPANA ER patients to discontinue because of lack of efficacy (P < 0.001).


NotesPrevious surgeries, employment status, non-opioid medications were not documented. Enrichment design that could underestimate the side effects profile of the interventions.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear riskThe authors did not report the method for sequence generation.

Allocation concealment (selection bias)Unclear riskThe authors did not document the method for concealment of allocation.

Blinding (performance bias and detection bias)
All outcomes - patients
Unclear riskThe authors did not report the physical characteristics of the placebos.

Blinding (performance bias and detection bias)
All outcomes - providers
Unclear riskOther than participants, it was unclear who else was blinded.

Blinding (performance bias and detection bias)
All outcomes - outcome assessors
Unclear riskDescribed as 'double-blind' but not clearly stated.

Incomplete outcome data (attrition bias)
All outcomes - drop-outs
High risk% drop-outs exceeded 20% in each group: 21/70 (30%) in the oxymorphone group and 55/73 (75.3%) in the placebo group.

Incomplete outcome data (attrition bias)
All outcomes - ITT analysis?
Low riskThe authors used conservative methods for the measurement of efficacy and safety in patients who discontinued the treatment.

Selective reporting (reporting bias)Unclear riskWe could not identify the protocol of this study in the databases for registration of clinical trials.

Group similarity at baselineHigh risk52.8 % women in the open label; however, 40% were randomized to active treatment and 24% to placebo. The study authors observed higher pain ratings and higher mean daily dosage of analgesic in the placebo group compared with the opioid arm.

Influence of co-interventionsLow risk"To prevent confounding of the study through the use of other analgesics, short-acting nonsteroidal anti-inflammatory drugs or other adjuvant analgesics were not permitted".

Compliance with interventions      High risk49/70 participants in the active group and 18/73 participants in the placebo group completed the study.

Timing of outcome assessmentsLow riskTwelve week follow-up in both groups.

Hale 2010

MethodsEnrichment design, placebo-controlled, clinical trial (2 to 4 weeks open label + 12 weeks double-blinded) developed in 66 centres in USA.


Participants459 patients met the inclusion criteria but only 268 were randomized. 134 patients were analysed for the primary outcome. Male and female - 18 to 75 years old, with moderate-to-severe CLBP, at least 3 hours per day, 20 days per month, for 6 months, with non-neuropathic or neuropathic characteristics based on the Quebec Task Force Classification of Spinal Disorders (QTFCSD). Opioid treatment with > 60 mg oral morphine equivalent (12 mg hydromorphone), but < 320 mg morphine (64 mg hydromorphone) per day within 2 months prior to the screening visits, and on stable doses of all prior analgesics for at least 2 weeks prior to the screening visit. Exclusion criteria: Any other chronic pain condition that would have interfered with the assessment of LBP, surgical procedure for back pain within 6 months, fibromyalgia, CRPS, acute spinal cord compression, lower extremity weakness or numbness, cauda equina, diabetic amyotrophy, diskitis, back neoplasm GI dysfunction, psychiatric condition, MAOIs within 14 days, bowel obstruction within 60 days.


Interventions12-64 mg of hydromorphone (HM) CR once a day plus HM IR as rescue medication versus matching placebos.


OutcomesThe primary efficacy assessment was the mean change from baseline to week 12 or final visit of the double-blind phase in weekly pain intensity 0 to 10 NRS. Secondary measures: mean change from baseline to week 12 of the double-blind phase in weighted mean pain intensity NRS score (i.e., area under the pain intensity NRS score versus time curve [area under the curve, AUC]), based on pain intensity NRS scores recorded in the patient diaries; mean change from baseline to each visit in pain intensity during the 12 week double-blind phase based on the pain intensity NRS scores.


NotesAll patients were opioid users. Duration of LBP was not documented. We received additional (unpublished) information from the authors that we used in the meta-analysis.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk"This randomization schedule was generated by ICON Clinical Research".

Allocation concealment (selection bias)Low riskA randomization number was assigned on day 1 of the double-blind phase via an interactive voice response system (IVRS) to encode the patient’s assignment to one of the two treatment groups, according to the computer-generated randomization schedule.

Blinding (performance bias and detection bias)
All outcomes - patients
Low riskAfter randomization, the central laboratory blinded hydromorphone results to help protect the overall blind, even though the patients may or may not have been using the rescue medication provided.

Blinding (performance bias and detection bias)
All outcomes - providers
Low riskAfter randomization, the central laboratory blinded hydromorphone results to help protect the overall blind, even though the patients may or may not have been using the rescue medication provided.

Blinding (performance bias and detection bias)
All outcomes - outcome assessors
Unclear riskAll personnel involved with the double-blind phase of the study, including the sponsor and relevant investigational staff, were blinded to the medication codes. The side effect profile of the medication could induce bias.

Incomplete outcome data (attrition bias)
All outcomes - drop-outs
High risk% drop-outs exceeded 20% in each group: 68/134 (50%) in the Hydromorphone group and 90/134 (67%) in the placebo group.

Incomplete outcome data (attrition bias)
All outcomes - ITT analysis?
Low riskThe study authors performed primary and secondary efficacy analyses on the ITT population, defined as all patients randomized to the double-blind phase who received at least one dose of study medication after randomization.

Selective reporting (reporting bias)Low riskThe trial was registered at clinicaltrials.gov (NCT00549042); the primary and the secondary outcomes were consistent in the protocol compared with the publication.

Group similarity at baselineLow riskAll baseline characteristics were similar between groups

Influence of co-interventionsLow riskThe study authors planned primary and rescue analgesia with the study medication.

Compliance with interventions      Low riskEleven patients per group dropped out from the study due to noncompliance to the medications.

Timing of outcome assessmentsLow riskTwelve week follow-up in both groups

Katz 2007

MethodsMulticentre, randomized, double-blind, placebo-controlled, parallel, enrichment design, clinical trial.


Participants326 patients were screened and included in the open-label titration period; 59/120 discontinued the intervention due to adverse events compared to 4/120 secondary to lack of efficacy. 205 patients were randomized; 71/105 completed the study in the oxymorphone group versus 47/100 in the placebo group.

Inclusion criteria: adult opioid-naive patients, defined as those taking < 5 mg/day of oxycodone or equivalent in the 2 weeks before screening, experiencing at least moderate low-back pain (> 50/100 mm) on an everyday basis and lasting for > 3 months. The study excluded patients with reflex sympathetic distrophy, acute spinal cord compression, cauda equina compression, nerve root compression, meningitis and discitis.


InterventionsOpen-label period: titration of Oxymorphone ER starting 5 mg PO q12h for 2 days and increasing 5 to 10 mg every 3 to 7 days until dose stabilization was achieved. This dose provided tolerability and efficacy (pain ≤ 40/100) for 5 consecutive days. Subsequently, patients were randomized into a 12 week double-blind treatment period in which they received their stabilized dose of oxymorphone ER or placebo every 12 hours. All patients were allowed oxymorphone immediate release (IR) as rescue medication for breakthrough pain. Rescue medication (NSAIDs) was restricted to a maximum of two doses each day.


OutcomesPatients kept a daily diary record of the total oxymorphone ER (or placebo) and IR doses and provided safety and efficacy assessments at the site on days 0, 4, 7, 14, 21, 28, 42, 56, 70 and 84 (±1 to 3 days).

Primary outcome: VAS pain score. Secondary outcomes: early discontinuation due to lack of efficacy; patient and physician global rating of the medication. Adverse effects profile.

Placebo patients discontinued significantly sooner from lack of efficacy than those receiving oxymorphone ER (P < 0.0001). Pain intensity increased significantly more in the placebo group (least squares [LS] mean
change 26.9 ± 2.4 [median 28.0]) than in the oxymorphone ER group (LS mean change 10.0 ± 2.4 [median 2.0]; P < 0.0001).


NotesEnrichment design that could underestimate the side effects profile of the interventions.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear riskThe study authors did not report the method of sequence generation. "A randomization code was generated by the sponsor to ensure the appropriate number of patients was allocated to each treatment group at random".

Allocation concealment (selection bias)Low risk"Patient medication kits were assigned unique 4-digit treatment numbers according to the randomization code". Central randomization by the sponsor.

Blinding (performance bias and detection bias)
All outcomes - patients
Low risk"The tablets of oxymorphone ER and placebo were over-encapsulated with gelatin to ensure that patients, investigator/study staff, and sponsor staff remained blind to study treatment"

Blinding (performance bias and detection bias)
All outcomes - providers
Low risk"The tablets of oxymorphone ER and placebo were over-encapsulated with gelatin to ensure that patients, investigator/study staff, and sponsor staff remained blind to study treatment"

Blinding (performance bias and detection bias)
All outcomes - outcome assessors
Unclear riskThe side effect profile of the medication could induce bias.

Incomplete outcome data (attrition bias)
All outcomes - drop-outs
High risk% drop-outs exceeded 20% in each group: 34/105 (32%) in the oxymorphone group and 53/100 (53%) in the placebo group.

Incomplete outcome data (attrition bias)
All outcomes - ITT analysis?
High riskThe study authors performed Last Observation Carried Forward (LOCF) for non-completers and this could favour the active treatment arm.

Selective reporting (reporting bias)Low riskThe trial was registered at clinicaltrials.gov (NCT00225797); the primary and the secondary outcomes were consistent in the protocol compared with the publication.

Group similarity at baselineLow riskPatients were similar at baseline for age, race, sex, etiology and average pain intensity.

Influence of co-interventionsLow riskThe study authors restricted all co-analgesics.

Compliance with interventions      Unclear riskThe study authors did not reported the rate of non-compliance with the interventions.

Timing of outcome assessmentsLow riskThe authors recorded the primary outcome on a daily basis and the study lasted for 12 weeks.

Khoromi 2007

MethodsSingle-centre, randomized, double-blind, active placebo-controlled, four period cross-over design for nine weeks each period.


ParticipantsPain Model: chronic sciatica (>4/10 pain intensity). Median age: 52.5 (range: 30 to 64) and duration of pain 5 years (range:0.3 to 37). 61 patients were screened, 55 randomized and 28 completed all four treatment periods. 14/28 of the completers were female. Baseline pain score (average leg) was: 4.9 ± 2.43.


InterventionsDuring 5 weeks of dose escalation and 2 weeks of maintenance at the highest tolerated dose patients received BID ER morphine (15 to 90 mg; mean 62 mg), nortriptyline (25 to 100 mg; mean 84 mg), their combination (morphine 49 mg and NT 55 mg) or benztropine-active placebo (0.25 to 1 mg); subsequently, two weeks of tapering; next period started one pain score reached > 4/10. Opioids and antidepressants were not allowed. NSAIDs and acetaminophen were used as rescue medications.


OutcomesMean scores for average leg pain during the maintenance weeks. Pain diaries consigned 0 to 10 pain score at bedtime, average back, leg and overall pain, worst back, leg and overall. Secondary: Global pain relief scores, ODI, BDI, SF-36 and general health status instrument.


NotesA carry-over effect was not noticed between treatments. 27/55 (49%) participants did not complete the four periods of treatment.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk"Patients were assigned by random numbers within blocks of four to one of four treatment sequences specified by a Latin square".

Allocation concealment (selection bias)Low risk"Randomization was performed by the NIH Pharmaceutical Development Service".

Blinding (performance bias and detection bias)
All outcomes - patients
Low risk"During the MS Contin treatment period, each blue pill contained MS Contin 15 mg and each pink pill contained inert placebo"

Blinding (performance bias and detection bias)
All outcomes - providers
Unclear riskProviders, apparently, were also blinded of the allocation of treatments.

Blinding (performance bias and detection bias)
All outcomes - outcome assessors
Low riskThe rate of guessing by the nurses was above the rate for chance only (> 25%), but did not reach a high percentage. "Patients and research staff were blinded to the randomization order".

Incomplete outcome data (attrition bias)
All outcomes - drop-outs
High risk% drop-outs exceeded 20% in each group: 27/55 (49%) participants did not complete the 4 periods of treatment.

Incomplete outcome data (attrition bias)
All outcomes - ITT analysis?
Low riskThe study authors included patients with at least two periods of treatment in the ITT analysis.

Selective reporting (reporting bias)Low riskThe trial was registered at clinicaltrials.gov (NCT00009672); the primary and the secondary outcomes were consistent in the protocol compared with the publication.

Group similarity at baselineUnclear riskNot applicable (cross-over study).

Influence of co-interventionsLow risk"None of the patients who opted to participate in the study were on opioids during the two months prior to study entry".

Compliance with interventions      Low riskOnly one patient dropped out due to non-compliance with the treatments.

Timing of outcome assessmentsLow riskTiming of outcomes assessment was identical in both groups.

O'Donnell 2009

MethodsTwo identical trials published in a single paper that were multicentre, randomized, double-blind, parallel group and active-controlled (celecoxib) studies. One study was conducted in 56 centres and the other one in 59 centres located in the USA.


ParticipantsAdults > 18 years old, with a diagnosis of LBP > 12 weeks of duration, requiring regular use of analgesics, who experienced moderate to severe LBP at baseline visit.

Exclusion criteria: CLBP that was either neurological in aetiology, due to recent major trauma, or was due to a visceral disorder; rheumatoid arthritis; spondyloarthropathy; spinal stenosis; malignancy; fibromyalgia; a herniated disc associated with neurological impairment within the past 2 years; psoriasis; seizure disorder; alcohol/analgesic/narcotic or other substance abuse within the past 2 years; asthma; urticaria or allergic-type reactions after taking aspirin or NSAIDs; gastrointestinal (GI) perforations, obstructions or bleeding; failed back surgery pain; active/suspected oesophageal,gastric, pyloric channel or duodenal ulceration or bleeding within 90 days prior to the first dose of study medication; unstable cardiovascular (CV) disease.


InterventionsEligible subjects were randomized in a 1:1 ratio to receive either celecoxib 200 mg twice a day (bid) or tramadol HCl 50 mg (no titration) four times a day (qid) for 6 weeks.


OutcomesEfficacy assessments were performed at baseline, screening and weeks 1, 3 and 6. The primary efficacy evaluation was based on the 0 to 10 NRS pain scale. The primary efficacy endpoint was the proportion of subjects responding successfully to their respective treatments at week 6. Successful responders were defined as subjects completing 6 weeks of treatment and having a ≥ 30% improvement from baseline to week 6 on the NRS-pain scale.


NotesImportant clinical factors as previous surgeries, previous analgesics, and non pharmacological treatments were not described in the baseline assessment.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk"Randomization was carried out using a computer-generated schedule".

Allocation concealment (selection bias)Unclear riskWe had insufficient information to permit judgement.

Blinding (performance bias and detection bias)
All outcomes - patients
Low riskThe participants were blinded but it was not clear who else was blinded in the trial.

Blinding (performance bias and detection bias)
All outcomes - providers
Unclear riskWe had insufficient information to permit judgement.

Blinding (performance bias and detection bias)
All outcomes - outcome assessors
Unclear riskThe study authors did not test the success of blinding of the interventions.

Incomplete outcome data (attrition bias)
All outcomes - drop-outs
High riskThe rate of drop-out was higher and exceeded 20% in the tramadol group compared with the celecoxib group.

Incomplete outcome data (attrition bias)
All outcomes - ITT analysis?
Low riskThe study authors performed tests for superiority on the ITT population, defined as randomized subjects who received at least one dose of study medication, and a sensitivity analysis on the evaluated population.

Selective reporting (reporting bias)Low riskThe trials were registered at clinicaltrials.gov (NCT00290901 and NCT00662558); the primary and the secondary outcomes were consistent in the protocols compared with the publication.

Group similarity at baselineUnclear riskWe had insufficient information to permit judgement as the authors did not report some important baseline characteristics.

Influence of co-interventionsLow riskCointerventions were similar between groups.

Compliance with interventions      High riskThe study authors reported a higher dropout rate in the intervention group (tramadol) versus the celecoxib group.

Timing of outcome assessmentsLow riskTiming of outcomes assessment was identical in both groups.

Peloso 2004

MethodsMulticentre, randomized, double-blind, and placebo-controlled trial.


ParticipantsInclusion criteria: patients with CLBP severe enough to require medications for greater than 3 months.

Exclusion criteria: use of sedative hypnotics, short-acting analgesics, topical preparation/medications and anaesthetics or muscle relaxants for a period of less than 5 half-lives of the given medication prior to the double-blind phase; use of medication that could reduce the seizure threshold; use of opioids or initiation of nutraceuticals within 6 weeks of the double blind phase; history of seizure disorder or unstable medical disease, renal or hepatic dysfunction, substance abuse, inflammatory disease and more severe pain in a location other than the lower back or other disease states that may interfere with the interpretation of pain; neurological deficit in the lower extremities, tumour or infections of the spinal cord or meninges, symptomatic disk herniation, severe spinal stenosis, spondylolisthesis or instability of lumbar vertebrae, acute vertebral fractures; back surgery (except if procedure was > 5 years prior to study enrolment); intolerant to tramadol or acetaminophen.

This study enrolled 338 participants; 167 in the opioid group and 171 in the control group.


InterventionsCombination tablets of tramadol (37.5 mg) and acetaminophen (325 mg) compared with placebo. Patients were treated for 3 months.


OutcomesPain (VAS), Function (Roland Disability Questionnaire).


Notes


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear riskWe had insufficient information to permit judgement.

Allocation concealment (selection bias)Unclear riskWe had insufficient information to permit judgement.

Blinding (performance bias and detection bias)
All outcomes - patients
Low risk"Identical appearing tablets containing either tramadol 37.5 mg/acetaminophen 325 mg or matching placebo".

Blinding (performance bias and detection bias)
All outcomes - providers
Low risk"After the initial titration phase, patients could adjust the daily dosage of study medication as needed up to a maximum of 2 tablets QID and a minimum of 3 tablets/day".

Blinding (performance bias and detection bias)
All outcomes - outcome assessors
Unclear riskThe active and the placebo group were indistinguishable for the outcomes assessors. The side effect profile of the medication could induce bias.

Incomplete outcome data (attrition bias)
All outcomes - drop-outs
High risk% drop-out exceeded 20% in each group: 81/167 (48%) in the tramadol/acetaminophen group and 110/171 (64%) in the placebo group.

Incomplete outcome data (attrition bias)
All outcomes - ITT analysis?
Low riskThe study authors performed an ITT analysis for safety and efficacy.

Selective reporting (reporting bias)Unclear riskWe could not identify the protocol of this study in the databases for registration of clinical trials.

Group similarity at baselineLow riskDemographic characteristics were similar between groups.

Influence of co-interventionsLow riskThe study authors did not allow any pain medication or treatment other than the study medication during the course of the study, except for rescue medication (acetaminophen 500 mg, up to 4 tablets daily) during the first 6 days of the double blind phase, provided the patient was taking no more than 6 tablets of study medication daily.

Compliance with interventions      Unclear riskThe study authors did not contemplate counting the tablets across the duration of the study.

Timing of outcome assessmentsLow riskTiming assessment was identical between groups.

Ruoff 2003

MethodsMulticentre randomized, double-blind, controlled trial.


ParticipantsInclusion criteria: Patients with CLBP severe enough to require medications for greater than 3 months.

Exclusion criteria: Previously discontinued tramadol therapy due to adverse effects or tramadol within 30 days before entry; antidepressants, cyclobenzaprine or antiepileptic drug for pain or TENS, chiropractic or acupuncture within 3 weeks of double-blind phase; sedative hypnotics, short-acting analgesics, topical anaesthetics or muscle relaxants for a period of < 5 half-lives of the specific medication before the double-blind phase; corticosteroids: injections or systemic within 3 months before screening phase; severe pain in location other than lower back or neurologic deficits in the lower extremities; contraindications to opioids or acetaminophen; major psychiatric disorders; history of suicide or substance abuse

This study enrolled 322 participants; 162 in the opioid group and 160 in the control group.


InterventionsCombination tablets of tramadol (37.5 mg) and acetaminophen (325 mg) compared with placebo. Patients were treated for 91 days.


OutcomesPain (VAS), function (Roland disability questionnaire) and disability (Roland disability questionnaire).


Notes


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk"Randomization was performed using SAS version 8 (SAS Institute Inc., Cary, North Carolina)".

Allocation concealment (selection bias)Unclear riskWe had insufficient information to permit judgement.

Blinding (performance bias and detection bias)
All outcomes - patients
Low risk"Patients, investigators, clinical staff, and study monitors remained blinded to treatment assignments until therapy was complete and the database was finalized".

Blinding (performance bias and detection bias)
All outcomes - providers
Low risk"Patients, investigators, clinical staff, and study monitors remained blinded to treatment assignments until therapy was complete and the database was finalized".

Blinding (performance bias and detection bias)
All outcomes - outcome assessors
Unclear riskThe side effect profile of the medication could induce bias.

Incomplete outcome data (attrition bias)
All outcomes - drop-outs
High risk% drop-out exceeded 20% in each group: 71/162 (43%) in the tramadol/APAP and 86/160 (53%) in the placebo group.

Incomplete outcome data (attrition bias)
All outcomes - ITT analysis?
Low riskThe study authors performed ITT analysis for efficacy assessment.

Selective reporting (reporting bias)Unclear riskWe could not identify the protocol of this study in the databases for registration of clinical trials.

Group similarity at baselineLow riskDemographic characteristics and pain scores were similar at baseline.

Influence of co-interventionsUnclear riskWe had insufficient information to permit judgement.

Compliance with interventions      Unclear riskThe study authors did not count the tablets across the duration of the study.

Timing of outcome assessmentsLow riskTiming assessment was identical between groups.

Schnitzer 2000

MethodsMulticentre randomized, double-blind, control trial (enrichment design).


ParticipantsInclusion criteria: patients with CLBP severe enough to require medications for greater than 3 months

Exclusion criteria: neurologic deficit in the lower extremities; tumours or infections; lesion amenable to surgery; more severe pain in a location other than the low-back, fibromyalgia, disk herniation, spondylolisthesis, spinal stenosis, instability of lumbar vertebrae; vertebral fracture; conditions such as tumour, infection, inflammatory disease, significant hepatic or renal disease, morbid obesity or borderline personality disorder; use of systemic corticosteroids or injections in the lower back within 3 months; use of TENS; history of narcotic or alcohol abuse; score at least 3 out of 5 in the Waddell's test

This study recruited 380 patients with 254 participants enrolled in the randomized group; 127 in the opioid group and 127 in the control group.


InterventionsTramadol (50 mg) with maximum 8 tablets per day (200 to 400 mg/day) or placebo.


OutcomesTime to therapeutic failure, pain (VAS) and function (RMDQ).


Notes


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk"Computer generated random numbers were used to ensure that any given patient would be assigned randomly to one of the 2 treatment groups".

Allocation concealment (selection bias)Unclear riskWe had insufficient information to permit judgement.

Blinding (performance bias and detection bias)
All outcomes - patients
Low risk"Identical appearing capsules of tramadol HCI and placebo were prepared by the R.W. Johnson Pharmaceutical Research Institute".

Blinding (performance bias and detection bias)
All outcomes - providers
Low risk"Patients were randomized either to continue treatment with tramadol or to receive placebo. Dosage adjustments were allowed but the daily dose was to be maintained within the range 200-400 mg of tramadol, or an equivalent amount of placebo capsules".

Blinding (performance bias and detection bias)
All outcomes - outcome assessors
Unclear riskOutcomes assessors could be bias based on the side effects profile of tramadol.

Incomplete outcome data (attrition bias)
All outcomes - drop-outs
High risk% drop-out exceeded 20% in each group: 36/127 (28.3%) in the tramadol group and 72/127 (56.7%) in the placebo group.

Incomplete outcome data (attrition bias)
All outcomes - ITT analysis?
Low riskThe study authors performed ITT analysis.

Selective reporting (reporting bias)Unclear riskWe could not identify the protocol of this study in the databases for registration of clinical trials.

Group similarity at baselineUnclear riskWe had insufficient information to permit judgement.

Influence of co-interventionsLow risk"Patients were told to maintain a constant level of exercise throughout the study. Physiotherapy (i.e., back exercises, therapy, hot/cold packs, and massages) started before entrance into the open label/run-in phase was continued throughout both the open label and double blind phases of the study. Physiotherapy could not be initiated during the open label or double blind phases of the study".

Compliance with interventions      High riskMore patients discontinued in the placebo groups due to inadequate pain relief. The study authors did not count the tablets across the duration of the study.

Timing of outcome assessmentsLow riskTiming assessment was identical between groups.

Steiner 2011

MethodsDouble-blind, placebo-controlled study with an enriched design. 27-day open-label with a TD, and those who tolerated and responded to treatment were randomized into a 12 week, double-blind, placebo-controlled phase.


ParticipantsOpioid naive patients. Men and women aged ≥ 18 years; moderate to severe LBP persisting for a minimum of 3 months prior to study entry. Subjects were naïve (< 5 mg of oxycodone/day in the last 14 days). Patients with spinal stenosis, spondylosis, spondylolisthesis and OA were eligible.


InterventionsAt the investigator’s discretion, patients receiving BTDS (Buprenorphine transdermal system) 20 or placebo TDS 20 who experienced unacceptable side effects were permitted to decrease the dosage of double-blind study medication to BTDS 10 or placebo TDS 10 and could remain at that level. If analgesia was deemed inadequate with BTDS 10 or placebo TDS 10, patients were allowed to retitrate their dosages up to BTDS 20 or placebo TDS 20, at the investigator’s discretion.


OutcomesThe primary efficacy outcome was the "average pain over the last 24 hours" score at Week 12 on an 11-point numerical rating scale collected at each clinic visit during the double blind phase (Weeks 1, 2, 4, 8, and 12). Secondary efficacy variables were sleep disturbance, as measured by the sleep disturbance subscale of the 12-item Medical Outcomes Study (MOS) Sleep Scale (0 to 100 questionnaire where higher scores indicate greater sleep disturbance); the mean daily number of tablets of non-opioid supplemental analgesic medications used during Weeks 2 through 12 of the double-blind phase.


NotesThere were eight exploratory variables: percent reduction in average pain score; Patient Global Impression of Change (PGIC); ODI; BPI; MOS short-form health survey (SF-36), were all collected at the beginning of the run-in period, at the randomization visit, and at each clinic visit during the double-blind phase (Weeks 1, 2, 4, 8, and 12); daily "pain right now" for those who took supplemental analgesics. The use of oxycodone for supplemental analgesia for the first six days of the double-blind phase was recorded in patient diaries. The study authors calculated the time from randomization to discontinuation because of lack of therapeutic effect.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear risk"Patients who tolerated BTDS and achieved the required analgesic response (described below) were randomized in a 1:1 ratio to BTDS 10 or matching placebo transdermal system (TDS) for the 12-week double-blind phase". The study authors did not report the method of randomization.

Allocation concealment (selection bias)Unclear riskWe had insufficient information to permit judgement.

Blinding (performance bias and detection bias)
All outcomes - patients
Low risk..."matching placebo transdermal system (TDS)"

Blinding (performance bias and detection bias)
All outcomes - providers
High riskThe provider was alert of efficacy and side effects of the medications and had the chance to increase or decrease the dosage of the drug provided.

Blinding (performance bias and detection bias)
All outcomes - outcome assessors
Unclear riskWe had insufficient information to permit judgement.

Incomplete outcome data (attrition bias)
All outcomes - drop-outs
High risk% drop-out exceeded 20% in each group: 86/256 (34%) in the buprenorphine group and 84/283 (30%) in the placebo group.

Incomplete outcome data (attrition bias)
All outcomes - ITT analysis?
Low riskThe study authors performed sensitivity analyses of the primary efficacy variable by applying different methods for imputing missing pain scores, including the retained drop-out ITT analysis. In the retained drop-out ITT analysis, the study authors included any "average pain over the last 24 hours" scores collected subsequent to the discontinuation of study drug and prior to completion or discontinuation from the study in the analysis, and attributed them to the randomized treatment.

Selective reporting (reporting bias)Low riskThe trial was registered at clinicaltrials.gov (NCT00490919); the primary and the secondary outcomes were consistent in the protocol compared with the publication.

Group similarity at baselineLow riskAll demographics reported were similar between groups.

Influence of co-interventionsLow riskAll patients were provided with IR oxycodone for supplementary analgesia during the first six days following randomization. During weeks 2 to 12 of the double-blind phase, patients were permitted to use sponsor-provided acetaminophen 500 mg every six hours up to a maximum of 2 g/day for supplemental analgesia. Alternatively, patients for whom acetaminophen was contraindicated could use ibuprofen 200 mg every six hours up to a maximum of 800 mg/day. Use of supplemental analgesic medication was to be suspended for 30 hours prior to assessing pain at study visits to reduce any confounding effect of this medication on analgesia provided by the patient’s blinded treatment.

Compliance with interventions      Unclear riskWe had insufficient information to permit judgement.

Timing of outcome assessmentsUnclear riskTiming assessment was identical between groups.

Uberall 2012

MethodsRandomized, double-blind, placebo-controlled, and active-controlled multicenter study. Patients with at least moderate LBP the treatment over 6 weeks (1 week wash out, 4 weeks treatment period and 1 week follow-up phase).


ParticipantsAdults aged 18 to 75 years, with LBP > 3 months. Patients who were taking analgesics for LBP but the treatment was not satisfactory, reporting at least moderate pain (> 3/10).

Exclusion criteria: neurological etiology, recent low-back trauma, significant medical or psychiatric disease. Other reasons for exclusion: rheumatoid arthritis, psoriasis, spondyloarthropathies, metabolic bone disease, spinal stenosis, spinal fractures, fibromyalgia, herniated disc, substance abuse, pregnancy, child-bearing potential, amongst others.


InterventionsParticipants had a wash-out phase for one week. They were assigned to receive flupirtine 400 mg OD, Tramadol 200 mg OD or matching placebos during four weeks.


OutcomesParticipants recorded lowest, average and highest pain scores. They also recorded categorical pain scores too. SF-12 and a short version of the SF-36 were also collected. QLIP inventory was also used. Other assessments included the patient's global assessment of disease status scale and the patient investigator global assessment of response to therapy. Additionally, participants and investigators used a seven step global impression of change scale after completion of the 4-week treatment period.


Notes


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk"Randomization of patients to treatment was based on a computer-generated allocation list with a block-size of six, and stratified by study site".

Allocation concealment (selection bias)Unclear riskWe had insufficient information to permit judgement.

Blinding (performance bias and detection bias)
All outcomes - patients
Low riskStudy authors described the study as double blinded with a double-dummy methodology.

Blinding (performance bias and detection bias)
All outcomes - providers
Unclear riskOutcomes were collected by pain diaries. However, further information was required to permit judgement.

Blinding (performance bias and detection bias)
All outcomes - outcome assessors
Unclear riskIt was unclear if patients were aware of the allocation based on the potential side effects of each treatment.

Incomplete outcome data (attrition bias)
All outcomes - drop-outs
High risk% drop-out exceeded 20% in each group: 33/118 (27%) in the tramadol group, 28/123 (22%) in the flupirtine group and 26/122 (21%) in the placebo group.

Incomplete outcome data (attrition bias)
All outcomes - ITT analysis?
Low riskThe study authors performed ITT analysis.

Selective reporting (reporting bias)Low riskThe study was registered (EudraCT 2009-013268-38) and the outcomes by protocol versus reported were equal.

Group similarity at baselineLow riskDemographics were similar between groups.

Influence of co-interventionsLow riskPatients had access to diclofenac as rescue therapy. They were urged to discontinue its use 24 hours before clinic visits. No other medications were allowed.

Compliance with interventions      Low risk"Rates of compliance with study medication, which were estimated by returned tablet count, were more than 95% in all three study groups".

Timing of outcome assessmentsLow riskTiming assessment was identical between groups.

Vorsanger 2008

MethodsMulticentre, 3 arms (2 of tramadol and one of placebo), randomized, enrichment design trial; 3 weeks of open trial plus 12 weeks of double-blinded phase.


Participants619 patients were included in the open-label period; 128/233 discontinued the treatment due to adverse effects and 41/233 due to lack of efficacy; 386 were randomized; 42/128 discontinued the treatment in the tramadol 300 mg group versus 42/129 in the tramadol 200 mg group and 61/129 in the placebo group.

Inclusion criteria: adults with VAS Score ≥ 40/100. CLBP> 6 months. Requiring at least 90 days of NSAIDs, COX-2, opioids or muscle relaxant.

Exclusion criteria: CRPS, inflammatory pain, fibromialgia, lumbar spine surgery, not well controlled medical condition, spinal manipulation for CLBP, under TENS therapy patients,  medications not allowed: NSAIDs, steroids, opioids, neuroleptics, SSRIs, SNRIs, carbamazepine and quinidine.


InterventionsEligible patients receiving tramadol F.R 300 mg once daily at the end of the run-in period were randomized in a 1:1:1 ratio to receive in a double-blinded fashion tramadol extended release 300 mg, tramadol extended release 200 mg or placebo.


OutcomesPain intensity VAS since the previous visit, current pain intensity VAS, global assessment of study medication. Following randomization, mean scores/or pain intensity VAS since the previous visit, averaged over the 12 week study period, increased more in the placebo group (12.2 mm) than in the tramadol ER 300 mg (5.2 mm, P = 0. 009) and 200 mg (7.8 mm, P = 0.052) groups. Secondary efficacy scores for current pain intensity, VAS, patient global assessment, Roland Disability Index, and overall sleep quality improved significantly (P ≤ 0.029 each) in the tramadol ER groups compared with placebo.


NotesEnrichment design that can underestimate the side effects profile of the interventions.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskComputer generated randomization

Allocation concealment (selection bias)Unclear riskWe had insufficient information to permit judgement.

Blinding (performance bias and detection bias)
All outcomes - patients
Low riskThe study authors used only tramadol (100 mg tablets) and placebo tablets which were identical in appearance and texture.

Blinding (performance bias and detection bias)
All outcomes - providers
Unclear riskWe had insufficient information to permit judgement.

Blinding (performance bias and detection bias)
All outcomes - outcome assessors
Unclear riskOutcomes assessor can be biased by the side effects profile of the drugs.

Incomplete outcome data (attrition bias)
All outcomes - drop-outs
High risk% drop-out exceeded 20% in each group: 42/128 (32%) in the high dose tramadol group and 61/129 (47%) in the placebo group.

Incomplete outcome data (attrition bias)
All outcomes - ITT analysis?
Low riskEfficacy analyses were performed on the ITT population, including all patients who received at least one dose of study medication and had primary efficacy information recorded at randomization.

Selective reporting (reporting bias)Unclear riskWe could not identify the protocol of this study in the databases for registration of clinical trials.

Group similarity at baselineLow riskDemographics were similar between groups.

Influence of co-interventionsLow riskStudy authors did not allow patients to use NSAID corticoesteroids, opioids, or other analgesic during the study, with the exception of low-dose aspirin or acetaminophen as described earlier. They also excluded neuroleptic, SSRIs, SNRIs, carbamazepine, or quinidine medications.

Compliance with interventions      Unclear riskThe study authors did not counting the tablets across the duration of the study.

Timing of outcome assessmentsLow riskTiming assessment was identical between groups.

Webster 2006

MethodsMulticentre, randomized, double-blind, active and placebo-controlled, four arms trial for 12 weeks (short-term).


ParticipantsThe study authors screened 1061 patients for eligibility and randomized 719 patients into four groups. For each patient randomized to the placebo group, they allocated 2 patients to each of the active groups.

Inclusion criteria: patients aged 18 to 70, with CLBP for at least 6 months, requiring daily analgesics. Baseline pain intensity (PI) score 5 at the screening visit, a mean daily PI score 5 recorded in a diary over the last 3 days of a 4 to 10 day washout period while off all analgesics except acetaminophen, and a confirmatory PI score 5 at the baseline visit at the conclusion of the washout period.

Exclusion criteria: CLBP secondary to malignancy, autoimmune disease, fibromyalgia, recent fracture, or infection; positive urine drug screens for any illicit substance at baseline; history of substance abuse within 5 years, or involvement in litigation regarding their lower back condition; pregnancy; allergy to study medications; severe hepatic, pulmonary, or renal impairment; unstable cardiac disease; corticosteroid therapy; intraspinal analgesic infusion or spinal cord stimulator in the preceding month; major surgery in the preceding 3 months; percutaneous or open procedure of the lumbosacral spine in the preceding 4 months; or high doses of central nervous system depressants or phenothiazines.


InterventionsPatients were randomized to receive placebo, oxycodone qid, or oxytrex qid or oxytrex bid. Each oxytrex tablet contained 1 μg naltrexone; oxytrex bid and qid treatments provide 2 and 4 μg naltrexone/day, respectively. Following a washout, patients with pain > 5 on a 0 to 10 scale were dose-escalated weekly from 10 up to 80 mg/day until reaching adequate pain relief (< 2) or a tolerable level of side effects. Following titration, the dose was fixed for 12 weeks.


OutcomesThe primary efficacy measure used was the 11-point numerical diary Pain Intensity Scale. Patients were asked to record a numerical score at bedtime each day for the overall pain intensity during the past 24 hours (0 = no pain and 10 = severe pain). Secondary efficacy measures included the SF-12, and ODI which were collected at baseline, monthly and at the end of treatment. Other secondary efficacy assessments, conducted at each clinic visit included: the Quality of Analgesia, for which patients rated pain relief as "poor", "fair", "good", "very good" or "excellent" and the Global Assessment of Study Drug, for which patients gave an overall rating as "poor", "fair", "good", "very good" or "excellent", taking into consideration the quality of pain relief, side effects, activity level, mood and sense of well-being in this evaluation.


Notes


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskComputer generated.

Allocation concealment (selection bias)Low risk..."Patients then remained on their individual fixed doses randomized via a central call-in system to 1 of 4 treatments in a 1:2:2:2".

Blinding (performance bias and detection bias)
All outcomes - patients
Low riskAll study medications were identical in appearance, and patients, site personnel, and study monitors were blinded to treatment assignments.

Blinding (performance bias and detection bias)
All outcomes - providers
Low riskAll study medications were identical in appearance, and patients, site personnel, and study monitors were blinded to treatment assignments.

Blinding (performance bias and detection bias)
All outcomes - outcome assessors
Low riskWe had insufficient information to permit judgement. The side effects of the drugs could induce bias.

Incomplete outcome data (attrition bias)
All outcomes - drop-outs
High riskAll active groups and the placebo group had a drop-out rate > 20%.

Incomplete outcome data (attrition bias)
All outcomes - ITT analysis?
Low riskThe primary analysis population for both efficacy and safety included the ITT population consisting of all randomized patients who took at least one dose of study medication and had at least one post-baseline PI assessment.

Selective reporting (reporting bias)Unclear riskWe could not identify the protocol of this study in the databases for registration of clinical trials.

Group similarity at baselineLow riskDemographics were similar between groups.

Influence of co-interventionsLow risk"No other analgesics were allowed during the treatment period".

Compliance with interventions      Unclear riskThe study authors did not count the tablets across the duration of the study.

Timing of outcome assessmentsLow riskTiming assessment was identical between groups.

 
Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion

Adams 2006Single arm, open label, short term study (3 months only) in opioid-naive patients.

Allan 2005Open label, randomized, parallel groups, multicenter study in opioid naive patients. Opioid versus opioid comparison.

Beaulieu 2007Opioid versus opioid comparison.

Cloutier 2013All participants received an opioid as analgesic rescue.

Gaertner 2006Narrative review of 6 previously published open label studies of CR oxycodone.

Gordon 2010aThe study included patients with at least 6 weeks of LBP. The inclusion criteria of this review was restricted to studies that have evaluated at least 3 months of LBP.

Gostick 1989Opioid versus opioid comparison.

Gould 2009Secondary analysis based on pain descriptors of Hale 2007.

Hale 1997Opioid versus opioid comparison.

Hale 1999Oxycodone CR versus Oxycodone IR comparison.

Hale 2005Patients randomized to the oxymorphone versus oxycodone phase only. Opioid compared with placebo phase was not randomized.

Hale 2009Tapentadol versus oxycodone comparison. The number of patients with CLBP was unclear.

Jamison 1998The study was not blinded.

Kalso 2007This is a secondary analysis of Allan 2005 (see above).

Kuntz 1996Less than 4 week study duration.

Landau 2007The primary site of pain was back in the 49.1% of the population.

Li 2008The study was developed for 7 days only.

Likar 2007Open label and opioid versus opioid comparison.

Moulin 1996Less than 50% of study sample included diagnosis of CLBP and results were not reported by diagnostic condition.

Muller 1998Less than four week study duration.

Nicholson 2006A comparison of a single medication administered in the morning versus the evening.

Nicholson 2006aOpioid versus opioid comparison.

Pascual 2007Open label study.

Peniston 2009Retrospective analysis.

Perrot 2006Opioid versus opioid comparison.

Portenoy 2007The trial was focused in the treatment of breakthrough low-back pain.

Rauck 2006Open label study and opioid versus opioid comparison.

Rauck 2007Open label study and opioid versus opioid comparison.

Salzman 1999Opioid versus opioid comparison.

Sarbu 2008Study about pharmacokinetics properties of tramadol.

Taylor 2007Multicentre patient reported survey.

Volinn 2009Observational study.

Vondrackova 2008After opioid taper and opioid titration during pre-randomization, patients were converted to an equivalent study medication with additional OxyNorm as rescue medication in the double-blind phase. The study therefore does not represent a "true" placebo-controlled design.

Wallace 2007Open label observational study.

Weinstein 2006Open label study.

Wiesel 1980Prospective observational study.

 
Characteristics of ongoing studies [ordered by study ID]
NCT01081912

Trial name or titleA randomized double-blind, placebo-controlled Trial to evaluate the efficacy, tolerability and safety of hydrocodone bitartrate controlled-release capsules in opioid-experienced subjects with moderate to severe CLBP.

MethodsRandomized, double-blind trial.

ParticipantsAdult patients suffering from CLBP. Subjects must be classified as non-neuropathic, neuropathic, or symptomatic for more than 6 months after LBP surgery.

InterventionsThe trial will consist of a screening phase (up to 14 days), an open-label conversion and titration phase (up to 6 weeks), a 12-week placebo-controlled treatment phase, and a 2 week follow-up phone call. Participants are randomized to receive 10 to 50 mg of hydrocodone or placebo capsules.

OutcomesPrimary outcome: change in average pain intensity as measured daily by a 0 to 10 Numerical Rating Scale (NRS). Secondary outcome: the change in pain intensity as measured in the clinic by a 0 to 10 NRS.

Starting dateMarch 4, 2010.

Contact informationJohn Ning, MD.

NotesStudy completion date: October 2011.

NCT01358526

Trial name or titleA Randomized, double-blind, placebo-controlled, multicentre trial with an enriched study design to assess the efficacy and safety of oxycodone/naloxone controlled-release tablets (OXN) compared to placebo in opioid-experienced subjects with moderate to severe pain due to chronic low back pain who require around-the-clock opioid therapy

MethodsRandomized, double-blind trial.

ParticipantsAdults with moderate to severe CLBP as their predominant pain condition for at least 3 months prior to screening period; the pain must be related; sciatica must be ruled out. Study will include patients with a stable regimen of opioid.

InterventionsOxycodone/naloxone controlled-release tablets (10/5 mg, 20/10 mg, 30/15 mg, 40/20 mg) taken orally every 12 hours or placebo for 12 weeks.

OutcomesThe "average pain over the last 24 hours" at week 12 of the double-blind period; The Sleep Disturbance Subscale of the MOS Sleep Scale; Patient Global Impression of Change (PGIC).

Starting dateMay 2011.

Contact informationPurdue Pharma LP.

NotesStudy completion date: October 2012.

NCT01455519

Trial name or titleTrue functional restoration and analgesia in non-radicular low back pain: a prospective double blind, placebo-controlled study of hydromorphone ER.

MethodsRandomized, double-blind trial.

ParticipantsPatients with CLBP. Non-radiating pain (below buttocks), no frank weakness or atrophy, no sensory or reflex changes.

InterventionsTotal target dose of 32 mg/day. All subjects will have a lead in for 2 weeks; then begin a "forced" 2-week up-titration schedule as follows: 8 mg/d (1 pill, 5 days), 16 mgday (2 pills, 5 days), and 24 mg/day (3 pills, 5 days) then finally 32 mg/day (4 pills/day) for the "stable dose" phase of the study, or identical placebo pills.

OutcomesThe efficacy of hydromorphone extended release in chronic non-radicular low-back pain to improve pain, function and activity.

Starting dateOctober 2011.

Contact informationakirsling@ric.org

NotesEstimated primary completion date: November 2013.

NCT01571362

Trial name or titleA multicentre, 12 week, double-blind, placebo-controlled, randomized withdrawal study to determine the efficacy and safety of ALO-02 (oxycodone hydrochloride and naltrexone hydrochloride) extended-release capsules in subjects with moderate to severe CLBP.

MethodsRandomized, double-blind trial.

ParticipantsThe patient has had moderate to severe CLBP for at least 3 months duration before screening.

Interventions20 to 160mg total daily dose of oxycodone, divided into symmetric doses and administered twice daily versus placebo for 12 weeks.

OutcomesChange in baseline in daily average pain numerical rating scale scores; % reduction in daily average pain numerical rating scale scores; changes in brief pain inventory-short form; change from baseline in brief pain inventory-short form; coanalgesia requirement; changes in patient's global assessment of low back pain; changes in Roland Morris disability questionnaire; change from baseline in healthcare resource use questionnaire; changes in EQ-5D health questionnaire.

Starting dateJune 2012.

Contact informationPfizer.

NotesOngoing trial.

NCT01789970

Trial name or titleA 12 week, randomized, double-blind, placebo-controlled, randomized-withdrawal study to evaluate the efficacy and safety of hydrocodone bitartrate extended-release tablets (CEP-33237) at 30 to 90 mg every 12 hours for relief of CLBP who require opioid treatment for an extended period of time.

MethodsRandomized, double-blind trial.

ParticipantsThe patient has had moderate to severe CLBP for at least 3 months duration before screening.

InterventionsHydrocodone bitartrate extended-release tablets or placebo tablets will be self-administered by patients at doses of 15, 30, 45, 60, or 90 mg, with each dose taken every 12 hours, during 12 weeks.

OutcomesWeekly average of daily worst pain intensity (WPI) scores; weekly average of daily average pain intensity (API) scores; adverse effects profile.

Starting dateMarch 2013.

Contact informationTeva GCO.

NotesEstimated primary completion date: January 2014.

 
Comparison 1. Tramadol compared to placebo

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Pain intensity (higher score means worse pain levels)51378Std. Mean Difference (IV, Fixed, 95% CI)-0.55 [-0.66, -0.44]

 2 Disability (higher ratings mean greater disability)51348Std. Mean Difference (IV, Fixed, 95% CI)-0.18 [-0.29, -0.07]

 3 Side effects5Risk Difference (M-H, Random, 95% CI)Subtotals only

    3.1 Nausea
51401Risk Difference (M-H, Random, 95% CI)0.09 [0.05, 0.13]

    3.2 Constipation
41147Risk Difference (M-H, Random, 95% CI)0.05 [0.02, 0.09]

    3.3 Somnolence
3911Risk Difference (M-H, Random, 95% CI)0.06 [-0.01, 0.13]

 
Comparison 2. Buprenorphine compared to placebo

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Mean pain intensity2653Std. Mean Difference (IV, Fixed, 95% CI)-2.47 [-2.69, -2.25]

 2 At least 30% of pain relief or moderate improvement2594Odds Ratio (M-H, Fixed, 95% CI)1.49 [1.08, 2.06]

 3 At least 50% of pain relief1498Odds Ratio (M-H, Fixed, 95% CI)1.39 [0.97, 1.99]

 4 Disability (higher ratings mean greater disability)1101Std. Mean Difference (IV, Fixed, 95% CI)-0.14 [-0.53, 0.25]

 
Comparison 3. Strong opioids compared to placebo

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Mean pain intensity61887Std. Mean Difference (IV, Fixed, 95% CI)-0.43 [-0.52, -0.33]

 2 At least 30% of pain relief or moderate relief3819Odds Ratio (M-H, Fixed, 95% CI)1.91 [1.41, 2.58]

 3 At least 50% of pain relief2750Odds Ratio (M-H, Fixed, 95% CI)1.89 [1.34, 2.66]

 4 Disability41375Std. Mean Difference (IV, Fixed, 95% CI)-0.26 [-0.37, -0.15]

 5 Side effects5Risk Difference (M-H, Random, 95% CI)Subtotals only

    5.1 Nausea
52346Risk Difference (M-H, Random, 95% CI)0.12 [0.05, 0.19]

    5.2 Constipation
52346Risk Difference (M-H, Random, 95% CI)0.11 [0.04, 0.19]

    5.3 Somnolence
52346Risk Difference (M-H, Random, 95% CI)0.06 [0.02, 0.10]

 
Comparison 4. Opioids (all types) compared to placebo

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Mean change in pain intensity in enriched design trials (higher changes indicate less favourable scores)3382Mean Difference (IV, Fixed, 95% CI)-21.34 [-22.77, -19.91]

 2 Side effects10Risk Difference (M-H, Random, 95% CI)Subtotals only

    2.1 Nausea
103747Risk Difference (M-H, Random, 95% CI)0.10 [0.07, 0.14]

    2.2 Somnolence
83257Risk Difference (M-H, Random, 95% CI)0.06 [0.03, 0.09]

    2.3 Constipation
93493Risk Difference (M-H, Random, 95% CI)0.07 [0.04, 0.11]

    2.4 Headaches
103747Risk Difference (M-H, Random, 95% CI)0.03 [0.01, 0.05]

    2.5 Dry mouth
61724Risk Difference (M-H, Random, 95% CI)0.06 [0.02, 0.10]

    2.6 Dizziness
93493Risk Difference (M-H, Random, 95% CI)0.08 [0.05, 0.11]

    2.7 Pruritis
62865Risk Difference (M-H, Random, 95% CI)0.04 [0.02, 0.05]

    2.8 Fatigue
61645Risk Difference (M-H, Random, 95% CI)0.03 [0.01, 0.05]

    2.9 Upper respiratory tract infection
1318Risk Difference (M-H, Random, 95% CI)-0.02 [-0.08, 0.03]

    2.10 Sinusitis
1318Risk Difference (M-H, Random, 95% CI)0.02 [-0.03, 0.06]

    2.11 Vomiting
73119Risk Difference (M-H, Random, 95% CI)0.07 [0.04, 0.09]

    2.12 Anorexia
2386Risk Difference (M-H, Random, 95% CI)0.04 [0.01, 0.07]

    2.13 Increased sweating
41350Risk Difference (M-H, Random, 95% CI)0.04 [0.02, 0.05]

    2.14 Hot flushes
2593Risk Difference (M-H, Random, 95% CI)0.03 [0.00, 0.05]

 
Comparison 5. Tramadol compared to celecoxib

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 At least 30% of pain relief or moderate improvement11583Risk Ratio (M-H, Fixed, 95% CI)0.82 [0.76, 0.90]

 
Comparison 6. Opioids (all types) compared to antidepressants

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Pain (higher score means worse pain level)2272Std. Mean Difference (IV, Fixed, 95% CI)0.21 [-0.03, 0.45]

 2 Disability (higher ratings mean greater disability)156Std. Mean Difference (IV, Fixed, 95% CI)-0.11 [-0.63, 0.42]

 
Summary of findings for the main comparison. Strong opioids compared to placebo for chronic low-back pain

Strong opioids compared to placebo for chronic low-back pain

Patient or population: people with chronic low-back pain
Settings: Outpatient pain management
Intervention: strong opioids compared to placebo

OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments

Assumed riskCorresponding risk

ControlStrong opioids compared to placebo

Mean pain intensity on a numerical scale. For example

0 (no pain) to 10 (maximum pain)
* The baseline for the most representative study (Buynak 2010) is 7.6 (SD 1.33)The mean mean pain intensity in the intervention groups was

0.43 standard deviations lower (0.52 to 0.33 lower)
Not applicable1887
(6 studies)
⊕⊕⊕⊝
moderate1,2,3
The magnitude of this difference is in the range of small to moderate.

Disability (higher ratings mean greater disability).

Various instruments were used, for example 0% (no disability) to 100% (bed-ridden)
* The baseline for the most representative study (Buynak 2010) is 6.7 (SD 1.61).The mean disability (higher ratings mean greater disability) in the intervention groups was

0.26 standard deviations lower
(0.37 to 0.15 lower)
Not applicable1375
(4 studies)
⊕⊕⊕⊝
moderate2,4,5
The magnitude of this difference is small.

At least 30% of pain relief or moderate reliefStudy populationOR 1.91
(1.41 to 2.58)
819
(3 studies)
⊕⊕⊕⊝
moderate6,7
The magnitude of this OR is large.

327 per 1000481 per 1000
(406 to 556)

At least 50% of pain reliefStudy populationOR 1.89
(1.34 to 2.66)
750
(2 studies)
⊕⊝⊝⊝
very low8,9,10
The magnitude of this OR is large.

236 per 1000369 per 1000
(293 to 451)

Side effects - SomnolenceStudy populationRD: 6%

(2% to 10%)
2346
(5 studies)
⊕⊝⊝⊝
very low10,11
This difference is not clinically important (< 10%).

25 per 100086 per 1000
(45 to 125)

Side effects - NauseaStudy populationRD: 12% (5% to 19%)2346
(5 studies)
⊕⊕⊝⊝
low11
This difference is clinically important (> 10%).

102 per 1000223 per 1000
(151 to 291)

Side effects - ConstipationStudy populationRD: 11% (4% to 19%)2346
(5 studies)
⊕⊝⊝⊝
very low10,11,12
This difference is clinically important (> 10%).

36 per 1000148 per 1000
(76 to 226)

* Of the included trials for this outcome, we chose the study that is a combination of the most representative study population and has the largest weighting in the overall result in Revman (Buynak 2010). This figure represents the baseline mean in the control group of this particular study.
CI: Confidence interval; RD: Risk difference; OR: Odds ratio; SD: Standard deviation.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

 1 Four trials had low risk of selection bias and one trial (Chu 2012) was unclear. All five trials had low risk of performance bias, and low risk of reporting bias. However, all five trials suffered from high risk of attrition bias, and some trials also had high risk of detection bias because it was unclear if the outcome assessor were blinded.
2 I² = 0%
3 See Figure 5. Funnel plot could not demonstrate bias.
4 Selection bias: three trials low risk of bias, and one trial unclear (Chu 2012). All four trials had low risk of performance bias. Detection bias was unclear in 3 trials, except Khoromi 2007. Attrition bias was judged high in all four trials. Reporting bias was not a problem in any trial.
5 See Figure 6. Funnel plot could not demonstrate bias.
6 All trials had risk of attrition bias and performance bias. One trial was unclear about randomization method.
7 Total number of events was 335.
8 Both trials had high risk of attritiion bias. Both trials are unclear about performance bias. One trial was not clear about method of randomization.
9 Heterogeneity: Chi² = 5.39, df = 1 (P = 0.02); I² = 81%.
10 Total number of events was < 300.
11 All trials had high risk of attrition bias. Most trials had a problem with performance bias, and one trial was not clear about method of randomization.
12 Heterogeneity: Tau² = 0.01; Chi² = 33.50, df = 4 (P < 0.00001); I² = 88%.
 
Table 1. GRADE Table: Tramadol compared to placebo

Quality assessmentNo of participantsEffectQuality



No of studiesDesignRisk of bias InconsistencyIndirectnessImprecisionOther considerationsWeak opioids compared to placeboControlRelative
(95% CI)
Absolute

Pain intensity (higher score means worse pain levels)

51RCTserious2serious3no serious indirectnessserious4none5689689-SMD 0.55 lower (0.66 to 0.44 lower)⊕⊕ΟΟ
LOW

Disability (higher ratings mean greater disability)

51RCTserious2no serious inconsistency6no serious indirectnessno serious imprecision7none5676672-SMD 0.18 lower (0.29 to 0.07 lower)⊕⊕⊕Ο
MODERATE

 1 Peloso 2004, Ruoff 2003, Schnitzer 2000, Uberall 2012 and Vorsanger 2008.
2 High risk of attrition bias. Difficult to judge other biases because the information was insufficiently reported. It is likely that there was also performance and detection bias due to difficulty in blinding.
3 Heterogeneity: Chi² = 28.34, df = 4 (P < 0.0001); I² = 86%.
4 SMD (effect size crosses 0.5. SMD is -0.55 [-0.66, -0.44].
5 See Figure 3 (for pain) and Figure 4 (for function).
6 Heterogeneity: Chi² = 1.24, df = 4 (P = 0.87); I² = 0%.
7 SMD (effect size) is -0.18 [-0.29, -0.07].
 
Table 2. GRADE Table: Buprenorphine compared to placebo

Quality assessmentNo of participantsEffectQuality



No of studiesDesignRisk of biasInconsistencyIndirectnessImprecisionOther considerationsBuprenorphine compared to placeboControlRelative
(95% CI)
Absolute

Mean pain intensity (higher score means worse pain levels)

2RCTvery serious1serious2no serious indirectnessno serious imprecision3none312341-MD 0.58 lower (0.61 to 0.55 lower)⊕ΟΟΟ
VERY LOW

Disability (higher ratings mean greater disability)

1RCTserious4no serious inconsistencyno serious indirectnessvery serious5none5150-MD 3 lower (11.44 lower to 5.44 higher)⊕ΟΟΟ
VERY LOW

 1 One trial had a high risk of selection, performance and detection bias. The two trials included in this comparison had high risk of attrition bias.
2 Heterogeneity: Chi² = 196.24, df = 1 (P < 0.00001); I² = 99%.
3 SMD -2.47 [-2.69, -2.25].
4 The trial included in this comparison had a high risk of attrition bias.
5 There is only one study. The SMD is -0.14 [-0.53, 0.25].
 
Table 3. GRADE Table: Strong opioids compared to placebo

Quality assessmentNo of participantsEffectQuality



No of studiesDesignRisk of biasInconsistencyIndirectnessImprecisionOther considerationsStrong opioids compared to placeboControlRelative
(95% CI)
Absolute

Mean pain intensity (higher score means worse pain levels)

6RCTserious1no serious inconsistency2no serious indirectnessno serious imprecisionnone31154733-SMD 0.43 lower (0.52 to 0.33 lower)⊕⊕⊕Ο
MODERATE

Disability (higher ratings mean greater disability)

4RCTserious4no serious inconsistency2no serious indirectnessno serious imprecisionnone5845530-SMD 0.26 lower (0.37 to 0.15 lower)⊕⊕⊕Ο
MODERATE

At least 30% of pain relief or moderate relief

3RCTserious6no serious inconsistencyno serious indirectnessno serious imprecision7none204/418
(48.8%)
131/401
(32.7%)
OR 1.91 (1.41 to 2.58)154 more per 1000 (from 80 more to 229 more)⊕⊕⊕Ο
MODERATE

29.7%150 more per 1000 (from 76 more to 225 more)

At least 50% of pain relief

2RCTserious8serious9no serious indirectnessserious10none146/386
(37.8%)
86/364
(23.6%)
OR 1.89 (1.34 to 2.66)133 more per 1000 (from 57 more to 215 more)⊕ΟΟΟ
VERY LOW

37.1%156 more per 1000 (from 70 more to 240 more)

Side effects - Somnolence

5RCTvery serious11no serious inconsistencyno serious indirectnessserious10none135/1548
(8.7%)
20/798
(2.5%)
See comment61 more per 1000 (from 20 more to 100 more)⊕ΟΟΟ
VERY LOW

2.5%61 more per 1000 (from 20 more to 100 more)

Side effects - Nausea

5RCTvery serious11no serious inconsistencyno serious indirectnessno serious imprecisionnone357/1548
(23.1%)
81/798
(10.2%)
See comment122 more per 1000 (from 50 more to 190 more)⊕⊕ΟΟ
LOW

9.1%109 more per 1000 (from 45 more to 170 more)

Side effects - Constipation

5RCTvery serious11serious12no serious indirectnessserious10none148/1548
(9.6%)
29/798
(3.6%)
See comment112 more per 1000 (from 40 more to 190 more)⊕ΟΟΟ
VERY LOW

5%154 more per 1000 (from 55 more to 262 more)

 1 Four trials had low risk of selection bias and one trial (Chu 2012) was unclear. All five trials had low risk of performance bias, and low risk of reporting bias. However, all five trials suffered from high risk of attrition bias, and some trials also had high risk of detection bias becasue it was unclear if the outcome assessor were blinded.
2 I² = 0%
3 See Figure 5.
4 Selection bias: three trials low risk of bias, and one trial unclear (Chu 2012). All four trials had low risk of performance bias. Detection bias was unclear in three trials, except Khoromi 2007. Attrition bias was judged high in all four trials. Reporting bias was not a problem in any trial.
5 See Figure 6.
6 All trials had risk of attrition bias and performance bias. One trial was unclear about randomization method.
7 Total number of events was 335.
8 Both trials had high risk of attritiion bias. Both trials were unclear about performance bias. One trial was not clear about method of randomization.
9 Heterogeneity: Chi² = 5.39, df = 1, P = 0.02; I² = 81%.
10 Total number of events < 300
11 All trials had high risk of attrition bias. Most trials had a problem with performance bias, and one trial was not clear about method of randomization.
12 Heterogeneity: Tau² = 0.01; Chi² = 33.50, df = 4, P < 0.00001; I² = 88%.
 
Table 4. GRADE Table: Tramadol compared to celecoxib

Quality assessmentNo of participantsEffectQuality



No of studiesDesignRisk of biasInconsistencyIndirectnessImprecisionOther considerationsOpioids (all types) compared to NSAIDsControlRelative
(95% CI)
Absolute

Pain (higher score means worse pain level) (Better indicated by lower values)

0No evidence availablenone0--not pooled

Disability (higher ratings mean greater disability) (Better indicated by lower values)

0No evidence availablenone0--not pooled

At least 30% of pain relief or moderate improvement

1RCTserious1no serious inconsistencyserious2serious3none412/785
(52.5%)
508/798
(63.7%)
OR 0.63 (0.52 to 0.77)112 fewer per 1000 (from 62 fewer to 160 fewer)⊕ΟΟΟ
VERY LOW

63.7%112 fewer per 1000 (from 62 fewer to 160 fewer)

 1 One study was included. There was uncertainty about allocation conceallment (selection bias), and blinding (performance and measurement bias). There was problem with drop-outs (attrition bias).
2 Indirectness in the outcome measure. This trial used "at least 30% pain relief OR moderate improvement". There is not report of mean pain scores.
3 Imprecision because there is only one study in this category.
 
Table 5. GRADE Table: Opioids compared to antidepressants

Quality assessmentNo of participantEffectQuality



No of studiesDesignRisk of biasInconsistencyIndirectnessImprecisionOther considerationsOpioids (all types) compared to antidepressantsControlRelative
(95% CI)
Absolute

Pain (higher score means worse pain level) (Better indicated by lower values)

2RCTsserious1no serious inconsistency2no serious indirectnessvery serious3none135137-SMD 0.21 higher (0.03 lower to 0.45 higher)⊕ΟΟΟ
VERY LOW

Disability (higher ratings mean greater disability)

1RCTserious4no serious inconsistencyno serious indirectnessvery serious5none2828-SMD 0.11 lower (0.63 lower to 0.42 higher)⊕ΟΟΟ
VERY LOW

 1 Two studies were included. All studies had problems with drop-outs (attrition bias), One study did not have a clear description of concealment of allocation (selection bias). Both studies might have some issues with blinding of provider and outcome assessor (performance and measurement bias).
2 Heterogeneity: Chi² = 0.00, df = 1 (P = 0.98); I² = 0%.
3 Two studies. Total population is less than 400. The 95% CI includes the no effect.
4 Only one study was included in this comparison. This study had issues with drop-outs (attrition bias), and potentially blinding of providers (performance bias).
5 There was only one study, and the 95% CI included the no effect.