Evidence Table 7Original Report through Update 5: Data abstraction and quality assessment of observational studies

Author, YearType of study, SettingMedications evaluated (dose, duration)Eligibility criteriaExclusion criteriaOther pain medications used or allowedNumber screened
Number eligible
Number enrolled
Number withdrawn or lost to follow-up
Number analyzed
Population characteristicsMethod of adverse event assessment and adverse events assessedQuality rating (number of criteria out of seven met)Funding sources and role of funderRate and number of adverse eventsComments
Ackerman, 2004Retrospective cohort
U.S.
Population-based (California Medicaid)
  1. Transdermal fentanyl
  2. Long-acting oxycodone
California Medicaid patients prescribed transdermal fentanyl or long-acting oxycodone during 3 consecutive monthsCalifornia Medicaid ineligible, <18 years old, prescribed other long-acting opioid, prescribed codeine, prescribed transdermal fentanyl or long- acting oxycodone after start date, or prescribed both medicationsShort-acting opioids and tricyclics controlled in analysesNR
NR
2106
Not applicableTransdermal fentanyl vs. long- acting oxycodone
Age: 67 vs. 54 years
Female: 74% vs. 65%
Non-white race: 31% vs. 26%
Cancer: 10% vs. 3.16%
Low daily dose: 41% vs. 28%
First episode of constipation event (ICD-9 code) using inpatient and outpatient claims dataFAIR. Inception cohort and number unable to be assessed NR. Not clear if assessors blinded. Adequate duration of follow-up, 90 days. (5)Janssen (transdermal fentanyl)
One author employed by funder, NR if data held by funder
Long-acting oxycodone versus transdermal fentanyl: adjusted odds ratio 2.55 (95% CI 1.33– 4.89) for constipation; 7.33 (1.98–27.13) in persons >65 years oldMany significant baseline differences between groups; analysis adjusted for dose, concomitant medications, comorbidities including cancer. Data appears to overlap with Staats 2004.
Arkinstall, 1995Prospective cohort (open-label extension of randomized trial)
Canada
Multicenter Pain clinics
Long-acting codeine, titrated to adequate pain control

Mean dose at end of trial 264 mg

Average duration 132 days
Patients completing trial by Arkinstall 1996 requesting continued long-term treatment with controlled- release codeineSame as trial by Arkinstall 1996Acetaminophen + codeine (short-acting)30 screened
30 eligible
28 enrolled
13/28 (46%) withdrawn or lost to follow-up
Not clear how many patients included in analysis
Age, gender, race NR; Diagnosis, duration of pain NR recruited from trial by Arkinstall 1996Any adverse event spontaneously reported or investigator-observed, timing not clearPOOR. Not clear if selection of patients biased; number eligible in randomized trial not clear. High overall loss to follow-up (13/28). Adverse events not specified or defined. Ascertainment techniques inadequately described (timing not clear). Assessors do not appear to have been blinded. No statistical analysis of potential confounders. Adequate duration of follow-up, 132 days. (1)Purdue (controlled release codeine)
One author (corresponding author) employed by funder, not clear if data held by funder
Long-acting codeine:
Adverse events “similar to rates reported in trial”.
Long-term use: 15/28 (54%), not clear how many discontinued medication due to adverse events.
Did not report rates of specific adverse events in long-term follow-up. Reasons for discontinuation of medication in long-term follow-up NR.
Bach, 1991Retrospective cohort
Denmark
Single center Pain clinic
  1. Long-acting morphine
  2. Buprenorphine (short-acting)
Mean dose at end of intervention 1.2 mg buprenorphine and 80 mg morphine

Average duration 58 days
Patients with chronic pain being treated with either sublingual buprenorphine or oral sustained release morphineNot specifiedAnti-inflammatory agents, tricyclic antidepressants, or anticonvulsantsUnable to assess, no inception cohortUnable to assess number withdrawn or lost to follow- up, no inception cohort 264 analyzedavg. 70 years
Gender and race NR

56% of non-cancer pain patients had ischemic leg pain
44% other non-cancer pain

Pain duration NR
Any adverse event as assessed weekly at follow-up visits or telephone calls by pain clinic nursesPOOR. Not clear if selection of patients biased, not clear if consecutive series. Unable to assess loss to follow-up, no inception cohort. Adverse events not specified or defined. Ascertainment techniques inadequately described. Assessors do not appear to have been blinded. No statistical analysis of confounders. Duration of follow-up NR. (0)NROral long-acting morphine vs. sublingual buprenorphine:
Any adverse event: 33/114 (28.9%) vs. 19.3% (29/150)
Individual adverse events NR according to indication for treatment
Tabulated results exclude 189 patients with cancer pain. Individual side effects NR for non- cancer pain patients. Not clear if mean doses of medications equipotent between long-acting morphine and buprenorphine.
Caldwell, 2002Prospective cohort
US
Multicenter Pain clinics
Once-daily morphine titrated to adequate pain relief

Mean daily dose at end of intervention 49 mg morphine (max 120 mg/day)

26 weeks of treatment
Adults with clinical and radiographic evidence of osteoarthritis who had failed course of non-opioids for pain and completed a randomized double-blind trial of once- daily morphine, twice-daily morphine, or placebo.Patients with serious comorbid conditions or conditions that might affect assessment of pain, weight <100 lbs, steroids within 1 month, intra-articular injections within six months, opioids therapy for >3 weeks prior to baseline, substance abuse, unable to tolerate opioid during randomized trialAcetaminophen, topical analgesics, and non- steroidal anti- inflammatory agents184 screened
184 eligible
181 enrolled
52% (86/181) discontinued or withdrew prematurely
181 analyzed for adverse events
Age, gender, race NR

Characteristics and duration of osteoarthritis pain NR for patients enrolling in open-label extension
Any adverse event, assessment methods not clearPOOR. Not clear if selection of patients biased, number eligible NR. High overall loss to follow-up. Adverse events not specified or defined. Ascertainment techniques inadequately described. Patients and assessors blinded to intervention. No statistical analysis of potential confounders. Duration of follow-up appears adequate, 4 weeks. (2)Funding source not clear; one author employed by drug manufacturer of once-daily morphine (Elan Pharmaceutical)Adverse events reported in >5% of patients taking once-daily morphine either in a.m. or p.m., n =181
Constipation: 35%
Nausea: 16%
Diarrhea: 13%
Somnolence: 13%
Dizziness: 9%
Abdominal pain: 8%
Pain: 8%
Headache: 8%
Infection: 7%
Insomnia: 6%
Peripheral edema: 6%
Vomiting: 6%
Dry mouth: 4%
Accidental injury: 4%
High withdrawal and loss to follow- up rate, not clear how withdrawn patients accounted for in adverse event rates.
Dellemijn, 1998Prospective cohort
Netherlands
Single center Pain clinic
Transdermal fentanyl titrated to adequate pain relief (max 100 micrograms/hr)

Maximum tolerated dose at end of treatment 75 micrograms/hour (7 patients)

12 weeks of treatment, followed by tapering off transdermal fentanyl and substitution with fixed dose long-acting morphine (60 mg bid)
Adults with non-cancer neuropathic pain who had completed a randomized double-blind trial with intravenous fentanyl plus diazepam or salineUse of opioids or modified pain regimens during the 2 weeks before starting the study, contraindications to opioids, presence of multiple sites or other types of pain, intermittent neuropathic pain, and uncertainty about origin of painContinued other entry medications at baseline level.50 screened
50 eligible
48 enrolled
33% (16/48) discontinued or withdrew prematurely
4% (2/48) lost to follow-up
44 analyzed for adverse events
avg. 49 years
77% female
Race NR

Neuropathic pain:
58% radiculopathy
19% post-traumatic neuralgia
6% post-herpetic neuralgia
4% phantom pain
6% central pain
6% post-rhizotomy pain

Pain duration NR
Any adverse event, assessment methods not clear, severity graded on 0–100 VASPOOR. Not clear if selection biased; number eligible in prior trial NR. High overall loss to follow-up (18/48). Adverse events not specified or defined. Ascertainment techniques not described. Patients and assessors not blinded to treatment. Adequate duration of follow-up appears adequate, 12 weeks. (1)Janssen (transdermal fentanyl)
Author not employed by funder, NR if data held by funder
Side effects on transdermal fentanyl occurring at any time (estimated from graph), n=44:
Nausea: 92%
Sweating: 68%
Headache: 68%
Fatigue: 58%
Vomiting: 54%
Dizziness: 53%
Constipation: 36%
Dyspnea: 36%
Pruritus: 33%
Dry mouth: 31%
Insomnia: 28%
Anorexia: 25%
Anxiety: 18%
Skin irritation: 18%
Other adverse events reported in <20%
Long-term use: 9/48 (19%) continued >2 years
High withdrawal and loss to follow- up rate, not clear how withdrawn patients accounted for in adverse event rates.
Dunbar, 1996Retrospective cohort
US
Single Center Pain clinic
6/20 (30%) oxycodone alone
6/20 (30%) methadone alone
5/20 (25%) methadone and oxycodone
1/20 (5%) morphine SR + oxycodone
1/20 (5%) hydromorphone + oxycodone
1/20 (5%) morphine SR alone

Doses NR

Pain duration NR
Patients with chronic non- cancer pain and a prior history of substance abuse who were managed on opioids for any period of timeNoneNRUnable to assess, no inception cohortUnable to assess number withdrawn or lost to follow- up, no inception cohort 20 analyzed35% peripheral neuropathy
20% chronic pancreatitis
10% failed back surgery
20% arachnoiditis
15% other

Duration NR
Prescription drug abuse assigned by physician reviewing dataPOOR. Not clear if selection of patients biased, not clear if consecutive series. Unable to assess loss to follow-up, no inception cohort. Adverse events not specified or defined. Ascertainment technique not described. Assessors do not appear to have been blinded. No statistical analysis of confounders. Duration of follow-up NR. (0)NRAbuse:
Oxycodone alone 1/6 (16.7%); methadone alone 3/6 (50%); methadone + oxycodone 3/5(60%); long-acting morphine + oxycodone 0/1 (0%); hydromorphone + oxycodone 1/1 (100%); long-acting morphine 1/1 (100%)
Only study addressing risk of abuse in higher-risk population. Diagnosis of abuse not specified or defined and assigned by physician not blinded to patient’s prior condition or current treatment. Inadequate detail regarding length of opioid treatment, dose, and severity of underlying pain. No inception cohort.
Franco, 2002Prospective cohortTransdermal fentanyl

Mean dose 42 mg/day

6 months
Patients of either gender aged 18 years or over presenting with chronic non- cancer pain susceptible to be treated with opioids and a mental status sufficient to be able to complete effectiveness tests; unsuccessful pain relief under current treatment with weak opioids at maximal doses (WHO) analgesic ladder to step 3 or previous treatment with morphine (in particular, when > 120 mg/day was required)Previous treatment with fentanyl; history of alcohol abuse, drug dependence, or severe personality disorders according DSM-III-R criteriaAnalgesicsNR
NR
236 enrolled
110(46.6%) withdrawn
236 analyzed
avg. 66.2 years
31% female
Race NR

50.8% neuropathic pain

Pain duration NR
Incidence, nature, time of onset, duration and intensity were recorded using non-specific and specific questions related to expected adverse events. Intensity determined by patient subjective evaluation. Investigator determined relationship between the treatment and adverse events.POOR. Not clear if selection of patients biased, number eligible NR. High overall loss to follow-up. Adverse events not specified or defined. Ascertainment techniques inadequately described. Patients and assessors not blinded to intervention. No statistical analysis of potential confounders. Duration of follow-up appears adequate, 6 months. (1)NRTransdermal fentanyl (n=236)
Any adverse effect: 177(75%)
Somnolence=53(22.5%)
Nausea=51(21.6%)
Vomiting=36(15.3%)
Constipation=36(15.3%)
Dizziness=59(25%)
Irritability=12(5.1%)
Urinary retention=10(4.2%)
Sweating=22(9.3%)
Local pruritus=9(3.8%)
High withdrawal rate
Green, 1996Retrospective cohortMethadone

Mean dose NR (range 30 to 120 mg/day)

Duration NR
Patients with chronic non- cancer pain on methadoneNRNRUnable to assess, no inception cohortUnable to assess number withdrawn or lost to follow- up, no inception cohort 11 analyzedavg. 56 years
27% female
Race NR

73% chronic back pain
18% neuropathy
9% chronic headaches

Pain duration NR
Any adverse event, assessment methods not clearPOOR. Not clear if selection of patients biased, not clear if consecutive series. No inception cohort, unable to assess loss to follow- up. Adverse events not specified or defined. Ascertainment technique not described. Assessors do not appear to have been blinded. No statistical analysis of potential confounders. Duration of follow-up NR. (0)NRMethadone:
Any adverse effect: 6/11 (55%)
Abuse: 1/11 (9%)
Overdose on patient's methadone by family member or friend: 1/11 (9%)
Sudden death: 1/11 (9%)
Severe anorexia, sedation, and nausea: 1/11 (9%)
Small study, not clear how patients selected for methadone treatment or how selected for inclusion. No inception cohort.
Hartung, 2007Prospective cohort
  1. Transdermal fentanyl
  2. Sustained-release oxycodone
  3. Sustained-release morphine
Oregon fee-for-service Medicaid enrollees with an initial prescription of a long- acting opioid (at least 28 days worth of medication) from January 1, 2000 and December 31, 2004 with continuous prescriptions for opioidsNot specifiedNRNR5684 included in analyses, 2027 with non- cancer pain (338 transdermal fentanyl, 508 methadone, 447 sustained-release oxycodone, 734 sustained- release morphine)Mean age: 62 vs. 49 vs. 54 vs. 52 years (p<0.001)
Female sex: 75% vs. 64% vs. 67% vs. 64% (p=0.002)
Non-white race: 6% vs. 10% vs. 12% vs. 8% (p=0.028)
Morphine equivalent dose/day: 98 vs. 237 vs. 67 vs. 77 mg (p<0.001)
Back pain: 57% vs. 65% vs. 59% vs. 65% (p=0.016)
Fibromyalgia: 15% vs. 27% vs 20% vs. 19% (p<0.001)
Mortality
Emergency department encounter related to constipation, alteration of consciousness, malaise, fatigue, lethargy, respiratory failure, opioid poisoning
Hospitalization related to one or more of the above symptoms
Opioid poisoning
Overdose symptoms (alteration of consciousness, malaise, fatigue, lethargy, respiratory failure)
Constipation
NRTransdermal fentanyl, methadone, and sustained-release oxycodone versus sustained- release morphine (referent), hazard ratios
Emergency department encounter or hospitalization: 1.42 (0.63 to 3.21) vs. 0.70 (0.29 to 1.69) vs. 0.52 (0.22 to 1.23)
Mortality: 0.89 (0.43 to 1.84) vs. 0.78 (0.29 to 2.13) vs. 0.98 (0.45 to 2.14)
Emergency department encounter: 1.27 (1.02 to 1.59) vs. 1.13 (0.91 to 1.41) vs. 0.91 (0.76 to 1.10)
Hospitalizations: 1.16 (0.85 to 1.59) vs. 1.09 (0.78 to 1.52) vs. 0.87 (0.67 to 1.14)
Opioid poisoning: NR vs. 2.41 (0.26 to 22.59) vs. 1.16 (0.11 to 12.83)
Overdose symptoms: 1.10 (0.72 to 1.68) vs. 1.57 (1.03 to 2.40) vs. 1.07 (0.74 to 1.53)
Constipation: 0.95 (0.40 to 2.25) vs. 0.66 (0.29 to 1.53) vs. 0.72 (0.34 to 1.55)
Controlled for age, race, sex, long- term care residence, number of unique prescribers, Charlson Comorbidity Index, concomitant drugs (benzodiazepines, sedative hypnotics, muscle relaxants, short-acting opioids), history of opioid dependence, abuse, or enrollment in a substance abuse treatment program
Milligan, 2001Prospective cohort
International Multicenter Pain clinics
Transdermal fentanyl (titrated)

Mean final dose 90 micrograms/hr

12 months
Patients >18 years old with chronic nonmalignant pain >6 weeks requiring continuous treatment with a potent opioidAllergy or hypersensitivity to opioids, life-threatening disease, skin condition precluding use of transdermal system, history of substance abuse, other significant diseaseImmediate-release morphine for breakthrough painScreened unclear
Eligible unclear
532 enrolled

(Study reports number eligible = number enrolled)
62% (231/532); 226 withdrew, 5 lost to follow- up
530 analyzed for adverse events
avg. 51 years
52% female
99% white

51% neuropathic
69% nociceptive
70% somatic
7.5% visceral

Pain duration average 8.8 years
Any adverse event possibly or definitely treatment-related, recorded monthly and at study discontinuation, assessment method not describedPOOR. Not clear if selection of patients biased, number eligible NR. High overall loss to follow-up. Adverse events not specified or defined. Ascertainment technique inadequately described. Patients and assessors not blinded. Inadequate statistical analysis (age only). Duration of follow- up appears adequate, 12 months. (1)Janssen (transdermal fentanyl)
One author employed by Janssen, NR if data held by funder.
Transdermal fentanyl:
Severe nausea: 48/530 (9%)
Severe vomiting: 42/530 (8%)
Severe diaphoresis: 37/530 (7%)
All serious adverse events: 146/530 (28%)
Serious adverse events probably or possibly treatment related: 38/530 (7%)
One or more adverse events considered possibly or definitely related to study medication: 387/530 (73%) and 170/530 (32%)
Withdrawals due to adverse events: 130/530 (25%)
103 patients had participated in trial by Allan. High overall withdrawal rate; not clear how withdrawn patients accounted for in adverse event rates. No significant difference in adverse event rates between older (>65) and younger patients, raw numbers not presented.
Ringe, 2002Prospective cohort
Germany
Multicenter
Transdermal fentanyl (titrated)

Mean dose NR
42/64(65.6%) 25 mg/h
3/64(4.6%) 50 mg/h
17/64(25.6%) required unspecified up-titration

Median observation duration=30 days
Patients with at least one osteoporotic vertebral fracture causing pain that required continuous administration of strong opioidsOsteoporotic fracture of the femoral neck or with osteoporosis caused by malignant diseasesNonopioid analgesics
Baseline=38/64(59%)
Day 15=8/64(12.5%)
Weak opioids
Baseline=17/64(26.6%)
Day 15=4/64(6.3%)
Strong opioids
Temporary=2/64(3.1%)
Screened unclear
Eligible unclear
64 enrolled
15(23%) withdrew
64 analyzed
Mean age=71 years
86% female
Race nr

Primary osteoporosis=70%
Secondary osteoporosis=30%

Median duration of pain=14 days
All adverse events assessed by severity (mild, moderate, severe) and relationship to treatment (none, unlikely, possible or probable)POOR. Not clear if selection of patients is biased. High overall loss to follow-up. Adverse events not specified or defined. Ascertainment technique inadequately described. Patients and assessors not blinded. No statistical analysis of confounders. Inadequate duration of treatment (30 days). (0)Janssen-Cilag GmbHTransdermal fentanyl:
Patients with at least one adverse event: 25(39%)
Withdrawal due to adverse events: 13(20.3%)
Roth, 2000Prospective cohort (open-label extension of randomized trial)
US
Multicenter Rheumatology clinics
Long-acting oxycodone (titrated)

Average dose 40 mg/day

6 month initial period with two optional 6 month extension periods
Patients completing clinical trial (Roth 2000) who wished to continue controlled-release oxycodone therapySevere organ dysfunction or history of drug or alcohol abuseNo rescue medications allowed133 screened
133 eligible
106 enrolled
60 withdrew
106 analyzed for adverse events
NR, population participated in study by Roth 2000Any adverse event Spontaneously reported or observed by investigator at each visit (weekly to once every 8 weeks)FAIR. Selection of patients does not appear biased. High overall loss to follow-up. Adverse events not specified or defined. Ascertainment technique adequately described. Patients and assessors not blinded. Inadequate statistical analysis (duration of treatment only). Duration of follow-up appears adequate, 6–18 months. (3)Purdue (sustained release oxycodone)
One author employed by funding source, NR if data held by funder
Long-acting oxycodone:
Long-term use: 46/106 (43%)
Withdrew due to adverse event: 32/106 (30%)
Constipation: 55/106 (52%)
Somnolence: 32/106 (30%)
Nausea: 25/106 (24%)
Pruritus: 21/106 (20%)
Nervousness: 16/106 (15%)
Headache: 14/106 (13%)
Insomnia: 14/106 (13%)
Hospitalization during observation period: 13/106 (12%), 5/106 (5%) possibly related to intervention
Varying periods of follow-up. Number enrolled (106) does not match numbers reported in duration of follow-up (114). Not clear how withdrawn patients accounted for in adverse event rates.
Staats, 2004Retrospective cohort
U.S.
Population-based (California Medicaid)
  1. Transdermal fentanyl
  2. Long-acting oxycodone
  3. Long-acting morphine
Random sample of California Medicaid patients, no prior constipation diagnosis, no long-acting opioid during previous 3 months, prescribed one of the included long-acting opioids during 3 consecutive monthsClaims for two or more opioids of interest, use of other opioids other than codeineNot specifiedNR
NR
1836
Not applicableTransdermal fentanyl vs. long- acting oxycodone vs. long- acting morphine
Age: 66 vs. 54 vs. 56 years
Female: 71% vs. 60% vs. 56%
Non-white race: 34% vs. 30% vs. 40%
Cancer: 38% vs. 15% vs. 38%
Dose (morphine equivalent); 116 vs. 232 vs. 208
First episode of constipation event (ICD-9 code) using inpatient and outpatient claims dataFAIR. Inception cohort and number unable to be assessed NR. Not clear if assessors blinded. Adequate duration of follow-up, 90 days. (5)Janssen (transdermal fentanyl)
One author employed by funder, NR if data held by funder
Long-acting oxycodone and long-acting morphine versus transdermal fentanyl (comparator): adjusted odds ratio 1.78 (95% CI 1.05–3.03) and 1.44 (0.80–2.60) for constipationMany significant baseline differences between groups; analysis adjusted for dose, concomitant medications, comorbidities including cancer. Data appears to overlap with Ackerman 2004.

From: Evidence Tables

Cover of Drug Class Review: Long-Acting Opioid Analgesics
Drug Class Review: Long-Acting Opioid Analgesics: Final Update 6 Report [Internet].
Carson S, Thakurta S, Low A, et al.
Portland (OR): Oregon Health & Science University; 2011 Jul.
Copyright © 2011 by Oregon Health & Science University.

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.