Chapter  35:  Management of Cancer Pain Volume 1: Evidence Report/Technology Assessment Number 35

Internal validity

Internal validity addresses the design, conduct, and reporting of the clinical trial. Some of the items belonging to this entity have been widely used in various "quality" scales and usually include items such as concealment of random allocation, treatment blinding, and handling of dropouts. In this evidence report, we define a four-category internal validity scale: A (least bias), B (susceptible to some bias), C (likely to have large bias), I (unable to assess due to lack of reported information). Further details of each category are as follows:

A. Double-blinded, well-concealed randomization, few dropouts, and no (or only minor) reporting problem of the trial that is likely to cause significant bias.

B. Single-blinded only, unclear concealment of randomization, or some inconsistency in the reporting of the trial but is unlikely to result in major bias.

C. Unblinded study, inadequate concealment of random allocation, high dropout rate, or substantial inconsistencies in the reporting of the trial such that it may result in large bias.

I. Inadequately reported (very often trials do not report certain data; this may occur by intent or due to oversight).

Applicability

Applicability, also known as generalizability or external validity, addresses the issue of whether the evidence from the study population is broad enough to generalize to the population at large. Individual studies are often unable to achieve broad applicability due to restricted study population characteristics and small number of study subjects (Lau, Ioannidis, and Schmid, 1997). We define the applicability grade as below:

A. Patients enrolled in the trial represent a broad spectrum of the population (high degree of applicability). Typically this would be a large study, although a large study in itself does not guarantee a high degree of generalizability.

B. Patients enrolled included only a narrow/restricted population, but the result is relevant to similar types of patient population (restricted applicability). Typically this would be a small study, but may also be a large study of a very homogeneous population.

C. Patients enrolled were part of an outlier population that is not immediately relevant to the study question (very limited direct applicability or not applicable), or the study reported only limited information.

I. Not reported or insufficient information to assess external validity issues (uncertain applicability).

Because the efficacy of pain treatments may depend on the baseline level of pain, we also extracted data on baseline pain intensity of the study population to assist in the interpretation of results. We report in the evidence-grading tables, along with the applicability rating, the baseline pain intensity expressed as visual analog scale (VAS) of 0-10 cm (or 0-100 mm) when these data are reported in the study. Studies that did not provide 0-10 cm VAS data but reported qualitative descriptions or other scales are so noted in the tables.

Study size

The study size is used as a measure of the weight of the evidence. Some studies have a high dropout rate due to deaths from the underlying cancer; we provide both the enrolled and evaluable number of patients when these data are reported. A large study provides a more precise estimation of the treatment effect but does not automatically confer broad applicability unless the study included a broad spectrum of patients. Very small studies, taken individually, cannot achieve broad applicability, but several small studies that enrolled diverse populations, taken together, may have broad applicability. The study size is also included as a separate dimension used to assist the assessment of applicability.

Magnitude of the treatment effect

In each of the result tables, "effect size" reflects the difference between outcomes in the treatment arms of the study, not pretreatment versus posttreatment comparisons in the experimental group.

The following effect size scale is employed for studies that provide consistent reporting of a pain-related outcome:

For example, if an experimental opioid were compared with morphine, and both treatments were found to have a large effect on pain scores, then the effect size assigned to this study would be a "±".

It should be noted that large difference in effect does not necessarily imply a statistically significant difference.

The outcomes reported by available studies on some of the questions were heterogeneous and not amenable to categorizing the effect size on the same scale. This group of heterogeneous outcomes includes drug consumption, pain relief, and quality-of-life related indices. Further, pain intensity may not have been reported using a VAS. Pain management experts evaluated these studies and assigned a qualitative score for the effect size as follows:

Internal validity

Each nonrandomized study was assigned a grade that comprised the study design and adequacy of reporting. The grades for study design are defined as follows:

• A

  Prospective controlled trial

• B

  Cohort

• C

  Case series

We also provided subscripts that correspond to a separate assessment of the adequacy of reporting of study design, patient population characteristics, disease (pain and cancer), treatment(s), and outcomes:

1. Completely or mostly described

2. Fair reporting but some important data are missing

3. Interpretation is difficult because many data are missing

For example, a cohort study with excellent reporting would be graded as B₁.

Applicability

Three criteria were used to assess the applicability of a nonrandomized study:

• The study directly addressed the question in the evidence report

• The study had a well-defined study population

• The study did not use a restricted population

A three-category grade was assigned to each, as follows:

• A

  All three criteria were met

• B

  Two of the three criteria were met

• C

  Only one criterion was met

Magnitude of the treatment effect

Most nonrandomized studies we examined were cohorts or case series, and these studies typically provide only a response rate for a single treatment arm (e.g., pain intensity score at the end of treatment, or the difference before and after treatment). We reported these rates or scores as the treatment effects in the summary tables.

Summary of the evidence from randomized controlled trials comparing an NSAID with another NSAID

Table 4. Summary of grading of randomized controlled (more...)

Table 4. Summary of grading of randomized controlled trials comparing an NSAID with another NSAID or with placebo

Number of studies	Patients enrolled/evaluated	Internal validity	Applicability
17	1164/1050 (90.2% evaluable)	A = 2 B = 11 C = 4	A = 1 B = 8 C = 8

Table 5. Grading of individual randomized controlled (more...)

Table 5. Grading of individual randomized controlled trials comparing an NSAID with another NSAID or with placebo

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Fuccella, 1975 75129886	58 (36)	~2.5 (on a 0 to 3 descriptive scale)	++	C	C
Ventafridda, 1975 75129887	24	~3.5 (on a 0 to 4 descriptive scale)	±	B	B
Martino, 1978 79255834	18	3.3 (on a 0 to 4 descriptive scale)	±	B	C
Sacchetti, 1984 84207536	36	3.4 - 3.7 (on a 0 to 4 descriptive scale)	+	B	B
Weingart, 1985 85125751	14 (10)	44.4	++	B	C
Turnbull, 1986 88230041	28 (23)	NR	±	B	B
Epstein, 1986 86312300	29	Mean 1.2 (VAS scale or unit not provided)	±	B	B
Levick, 1988 89144615	145 (100)	Baseline pain intensity >4	±	B	C
Staquet, 1989 90094723	126 (118)	"moderate to severe pain"	±	A	C
Ventafridda, 1990a 91151427	100	NR	±	C	C
Ventafridda, 1990b 90243070	65	Mean baseline score 4.1 to 6.8 -	±	B	A
Wool, 1991 Cur Ther Res**	60	2.5 (0-3 descriptive scale)	+	B	C
Gallucci, 1992 93038986	68 (40)	4 and 5 -in the two groups (according to graph)	±	B	B
Corli, 1993 94102070	64	Integrated pain score 30 to 35	±	C	B
Toscani, 1994 94280461	100	Integrated pain score 40 to 50	±	B	B
Minotti, 1998a 98281411	180	"acute, moderate or severe pain"	±	A	C
Yalcin, 1998 98196499	49 (47)	8.5 ± 1.3 (mean ± SD)	+	C	B

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

**No UI number is available.

Individual comparisons between NSAIDs did not demonstrate significant differences in efficacy with one exception. Dipyrone was found to be more effective and better tolerated than diflunisal (Yalcin, Gullu, Tekuzman, et al., 1998). These studies could not be analyzed in combination due to heterogeneity in the methodology of outcome assessment, different drugs compared, and study duration. Adverse reactions commonly reported in these studies were nausea, vomiting, sweating, flushing, epigastric pain, loss of appetite, vertigo, sleepiness, hemorrhage, and sedation. In general, adverse effects were considered to be minor.

Fuccella, Conti, Corvi, et al. (1975) compared indoprofen (100 or 200 mg) with placebo in 36 patients with primary and metastatic cancer and neuralgia. Pain intensity and pain relief scores were the primary outcomes. Indoprofen at both doses and placebo all reduced pain intensity measured hourly for 6 hours; however, pain relief scores were unchanged with placebo. Indoprofen at both doses was significantly more effective than placebo. Indoprofen 200 mg was more effective than 100 mg, although it is not clear if there was a significant difference.

Ventafridda, Martino, Mandelli, et al. (1975) compared indoprofen (100 or 200 mg) with aspirin (600 or 1,000 mg) and with placebo in 24 patients with cancer pain. Pain intensity and side effects were the primary outcomes. Indoprofen and aspirin were not significantly different in terms of total pain relief, although both drugs were more effective than placebo. No significant side effects were reported with any of the drugs.

Martino, Emanueli, Mandelli, et al. (1978) compared indoprofen (200 mg once) with naproxen (250 mg once) in a double-blind crossover study with 18 patients with cancer pain. Primary outcomes were pain intensity and pain relief after single doses. Both drugs were equally effective at relieving pain. Both drugs relieved pain for approximately 90 minutes. Three patients had headache after indoprofen; two had headache, two had sleepiness, one had vomiting, and one had sweating after naproxen.

Sacchetti, Camera, Rossi, et al. (1984) compared single doses of intravenous ketoprofen (100 mg and 400 mg) with injectable acetylsalicylic acid (1 g) in 36 patients with severe bone pain caused by cancer. Each patient received two of the three possible treatments. Primary outcomes were pain relief and patient preference. All treatments reduced the degree of pain by approximately one-half by the second hour after treatment. Ketoprofen 400 mg was more effective and maintained pain relief longer than the other two treatments. Ketoprofen 400 mg was also preferred to ketoprofen 100 mg and acetylsalicylic acid, which were equally preferred. No adverse reactions were recorded during the 6-hour observation period.

Weingart, Sorkness, and Earhart (1985) compared ibuprofen 400 mg qid to a dummy placebo in a double-blind, crossover 9-day trial of 10 patients. The primary outcome was pain severity. All patients were also on scheduled oral narcotics as prescribed by their own physicians. Ibuprofen (with scheduled narcotics) was more effective than placebo (with scheduled narcotics). Adverse events were no more common on ibuprofen than at baseline or on placebo.

Turnbull and Hills (1986) compared naproxen (500 mg bid) with aspirin (300 mg q 4 hours) in a double-blind, 2-week trial with crossover after the first week. Twenty-eight patients with cancer pain enrolled, and primary outcomes were pain severity and pain reduction. Naproxen and aspirin were equally effective at reducing pain. No adverse events occurred during the study period.

Epstein and Stevenson-Moore (1986) compared rinsing of the mouth with benzydamine or placebo in 29 patients with oropharyngeal cancer and mucosal pain caused by radiation therapy. The authors found significantly lower pain ratings and systemic analgesic intake in the benzydamine group compared with the placebo group.

Levick, Jacobs, Loukas, et al. (1988) compared two doses of oral naproxen and placebo in 145 patients with metastatic cancer. Pain intensity, rated by patients (for outpatients) and by physicians (for inpatients), was the primary outcome. Among responders to naproxen, pain relief following the high-dose regimen was significantly greater than the low-dose regimen. Differences in adverse effects between regimens were not significant.

Staquet (1989) compared ketorolac tromethamine (10, 30, and 90 mg IM) with placebo in 126 patients with moderate to severe cancer pain. Assessed outcomes were pain intensity and pain relief, and global impression. Ketorolac was found to be more effective than placebo.

Ventafridda, Toscani, Tamburini, et al. (1990a) compared oral naproxen (550 mg bid) with diclofenac (100 mg bid) in 100 patients with cancer pain. Patients suffering from neuropathic pain were excluded. Assessed outcomes were pain severity on an integrated scale comprising pain intensity and duration. Both drugs had similar efficacy. Progression to the second step of WHO analgesic ladder was required to improve analgesia in 77 percent of cases and to decrease side effects in 23 percent.

Ventafridda, De Conno, Panerai, et al. (1990b) compared naproxen (250 mg tid) with diclofenac (100 mg bid), indomethacin (50 mg tid), ibuprofen (600 mg tid), suprofen (200 mg tid), pirprofen (400 mg tid), acetylsalicylic acid (ASA) (600 mg tid), sulindac (300 mg bid), and paracetamol (acetaminophen, 500 mg tid) in 65 patients with moderate to severe cancer pain. Naproxen, diclofenac, and indomethacin were highly effective in relieving pain and relatively well tolerated. Ten of 65 patients withdrew from the study due to gastric discomfort. This study employed a crossover design and was therefore susceptible to carryover effects.

Wool, Prandoni, Polistena, et al. (1991) compared single doses of ketorolac (30 mg) suppositories with diclofenac (100 mg) suppositories in 60 patients with cancer pain. Primary outcomes were pain severity and relief. Both treatments were effective at reducing pain for 12 hours. While ketorolac achieved significantly better pain relief than diclofenac after hour 8, the actual difference in pain scores was small. Physicians and patients were more likely to rate pain relief as excellent for ketorolac than for diclofenac. No laboratory or vital sign measurement changes occurred in the trial.

The studies by Gallucci, Toscani, Mapelli, et al. (1992) compared nimesulide (200 mg bid) with naproxen (500 mg bid) in 68 patients with cancer pain. Patients with neuropathic pain were excluded. The authors assessed severity of pain and adverse effects. Analgesic efficacy and tolerability of the two drugs were the same. Four patients in the nimesulide group and six in the naproxen group suspended treatment due to side effects such as gastric pain, hemorrhage, or vomiting. The study by Toscani, Gallucci, and Scaricabarozzi (1993) is a duplicate publication.

Corli, Cozzolini, and Scaricabarozzi (1993) compared nimesulide (200 mg po bid), diclofenac (150 mg po bid), rectal nimesulide (400 mg qd), and rectal diclofenac (200 mg qd) in 64 patients with cancer pain. Assessed outcomes were pain, sleep duration, and adverse effects. Oral and rectal preparations of both drugs provided similar analgesia when used as a first step for the treatment of cancer pain, but nimesulide caused fewer gastrointestinal (GI) side effects.

Toscani, Piva, Corli, et al. (1994) compared oral ketorolac (10 mg q6h) with diclofenac (50 mg q8h) in 100 patients with cancer pain. Pain severity was assessed using an integrated score of intensity and duration. Ketorolac and diclofenac proved equally effective with respect to pain relief. Sedation was more frequent in the diclofenac group.

Minotti, Betti, Ciccarese, et al. (1998a) compared a single low or high dose (10 mg or 30 mg) of ketorolac with 75 mg diclofenac (both intramuscular [IM]) in 180 patients with nociceptive (71%) and neuropathic (28.8%) pain due to cancer. Assessed outcomes were pain relief and pain intensity. The authors found that all three regimens were equivalent.

Yalcin, Gullu, Tekuzman, et al. (1998) compared diflunisal with dipyrone in 47 cancer patients, of whom 29.8 percent had bone metastases. They assessed pain and side effects. Dipyrone reduced pain significantly more than diflunisal. In subgroup analysis (according to metastatic, nonmetastatic, and bone metastatic cancer) dipyrone was more effective in all three groups than diflunisal. No differences were found with respect to side effects, and their incidence in both groups was low.

Summary of evidence on the side effects of NSAIDs from uncontrolled trials

[See Evidence Table 18]

We identified 12 studies of NSAIDs for treatment of cancer pain that did not meet the inclusion criteria for the randomized controlled trial section. We reviewed the seven studies with the largest sample sizes (n >= 20) that reported on adverse events of NSAIDs. One study was a randomized controlled trial that compared two different doses of the same medication; one other was a case series; and the rest were prospective cohort studies. Two were published as letters. All had pain relief or quality of life as primary outcomes. Only three described collection of adverse events information in a prospective manner.

The studies reported on a variety of NSAIDs, including oral and intravenous naproxen, oral and subcutaneous ketorolac, oral dipyrone, oral diclofenac, and oral piroxicam. Cohn, Machado, Bier, et al. (1988) studied the combination of piroxicam and doxepin. Three trials were very short (from one dose to 3 days); in three, the mean duration of treatment was 2 to 3 weeks; and one had a mean duration of treatment of 6 months.

Reporting of adverse events was very sparse. Three studies simply reported "no adverse effects." The remainder reported on primarily gastrointestinal adverse effects.

Only two studies reported on dyspepsia. On low-dose oral naproxen, 7 percent of subjects had dyspepsia (within 3 days), while 16 percent had dyspepsia on high-dose naproxen. On piroxicam, 13 percent of subjects had dyspepsia, all of which resolved with increased doses of sucralfate.

Three studies reported on gastrointestinal bleeding. During subcutaneous ketorolac (and misoprostol or omeprazole) treatment, gastrointestinal bleeding occurred in four of 36 subjects (all over 65 years old) in one study and two of 25 in a second study. During piroxicam therapy, two of 30 subjects had gastrointestinal bleeding.

Gastrointestinal perforations were reported in two studies. One of 36 subjects given subcutaneous ketorolac had a colonic perforation, and one of 30 subjects taking piroxicam had a gastric perforation.

Three studies reported discontinuation rates of 5 percent to 20 percent due to adverse events. Other reported adverse events occurred in only one subject per study.

Summary of the evidence from randomized controlled trials comparing an NSAID with another NSAID combined with a weak opioid or comparing an NSAID with a strong opioid

Table 6. Summary of the grading of randomized controlled (more...)

Table 6. Summary of the grading of randomized controlled trials comparing an NSAID with another NSAID combined with a weak opioid or comparing an NSAID with an opioid

Number of studies	Patients enrolled/evaluated	Internal validity	Applicability
25	1563/1499 (96% evaluable)	A = 6 B = 14 C = 5	A = 1 B = 10 C = 14

Table 7. Grading of individual randomized controlled (more...)

Table 7. Grading of individual randomized controlled trials comparing an NSAID with another NSAID combined with a weak opioid or comparing an NSAID with a strong opioid

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Moertel, 1974 74251095	100	"mild to moderate pain at baseline"	+	C	C
Martino, 1976 77208209	36 (three studies)	3.5 ± 0.1 (200 mg indoprofen group) 3.3 ± 0.1 (600 mg ASA group) ~3.3 (placebo group, from graph) (all above scores are mean ± SD on a 0 to 4 scale)	±	C	B
Stambaugh, 1980a 80204844	40 (37) [SDS] 170 [RDS]**	Mean baseline scores 2.8, 2.9, and 3.1 in the three study arms (0-4 scale) [SDS] 2.07 [RDS] **	±	B	B
Stambaugh, 1980b Cur Ther Res	31	2.6 to 2.8 (0-4 descriptive scale)	+++	A	B
Stambaugh, 1981 82143186	40 (37)	"moderate to extremely severe chronic pain"	±	C	B
Stambaugh, 1982a Cur Ther Res, No.3	29 (20)	2.3 to 2.5 (0-4 descriptive scale)	++	B	C
Stambaugh, 1982b Cur Ther Res, No.6	45 (40)	2.75 to 2.80 (0-4 descriptive scale)	+	B	C
Stambaugh, 1983 84129666	30	3.4 to 3.7 (scale NR)	++	C	C
Ferrer-Brechner, 1984 84277729	30 (28)	"moderate to severe"	±	B	C
Tonachella, 1985 Cur Ther Res	20 (16)	"moderate to severe"	+	B	C
Stambaugh, 1987a 88059744	43	moderate 26/60 severe 34/60	±	B	C
Stambaugh, 1988a 89214771	160	Baseline pain intensity scores were either 2 (moderate) or 3 (severe) for all patients (0 to 3 scale)	±	B	C
Sunshine, 1988 89214773	123	Baseline pain intensity scores were moderate or severe (postpartum, postoperative, and chronic cancer pain studies)	±	C	C
Stambaugh, 1988b 89063812	30	"moderate to severe"	±	B	C
Minotti, 1989 89144649	99	"At least 40 mm of a 100-mm VAS in two separate evaluations"	±	B	B
Puglisi, 1989 Cur Ther Res	45(40)	> 4 -	±	B	C
Estape, 1990 91032539	40	Mean pain score at baseline ~3 (0-4 scale)	±	B	B
Carlson, 1990 95161866	75	"Moderate to severe at study entry"	±	A	B
Bjorkman, 1993 93170354	16 (15)	Average pain score at baseline 36 mm (0-100 VAS)	±	B	B
Staquet, 1993 Cur Ther Res	90 (88)	60% "moderate" 40% "severe"	±	B	C
Bosek, 1994 95161866	92 (70)	Not applicable	±	A	C
Dellemijn, 1994 95092369	20 (16)	Mean baseline pain score of 8.2	±	A	B
Rodriguez, 1994 94361852	121	Mean baseline pain score of 8.3	±	A	A
Chary, 1994 95052963	24	"Moderate"	±	B	C
Minotti, 1998	184 (180)	Moderate to severe (>4)	±	A	B

*Number of evaluable patients shown in parentheses if different from enrolled.

**SDS = Single-dose studies, RDS = Repeated-dose studies, NR = Not reported.

The results of this meta-analysis are in agreement with a meta-analysis reported by Eisenberg, Berkey, Carr, et al. (1994). It should be noted that questions regarding other potentially beneficial outcomes such as opioid dose-sparing by NSAIDs and lessening of side effects by drug coadministration could not be evaluated in this analysis due to the heterogeneity in the studies evaluated with respect to these endpoints (i.e., side effects, dose reduction).

Moertel, Ahmann, Taylor, et al. (1974) compared aspirin plus codeine, aspirin plus oxycodone, aspirin plus pentazocine, aspirin plus caffeine, aspirin plus pentobarbital, aspirin plus promazine hydrochloride, aspirin plus ethohepatazine, aspirin alone, and placebo in an acute study in 100 patients with mild to moderate cancer pain at baseline. Aspirin plus codeine, aspirin plus pentazocine, and aspirin plus oxycodone were superior to aspirin alone, which in turn was superior to placebo. The remaining combinations were as effective as or less effective than aspirin.

Martino, Ventafridda, Parini, et al. (1976) reported the outcomes of three trials: indoprofen versus ASA or placebo, indoprofen versus aspirin (higher dose) or placebo, and two doses of pentazocine in a total of 36 patients, 12 per trial. Indoprofen 600 mg was significantly superior to aspirin 600 mg, and both were significantly more effective than placebo. Pentazocine was equipotent to 100 mg of indoprofen. This is a poorly reported study comparing outcomes from three different acute trials.

Stambaugh, Tejada, and Trudnowski (1980a) reported the outcomes of two studies (a single-dose and a repeated-dose study) comparing zomepirac with oxycodone plus aspirin 224mg, phenacetin 162 mg, caffeine 32mg and with placebo. In the single-dose study there was no significant difference between the two treatments, while in the repeated-dose study oxycodone plus APC was superior to both doses of zomepirac in pain relief and acceptability.

Stambaugh (1980b) compared Tylox® (acetaminophen 500 mg, oxycodone hydrochloride 4.5 mg, and oxycodone terephthalate 0.38 mg), Percodan® (aspirin 224 mg, phenacetin 160 mg, caffeine 32 mg, and oxycodone terephthalate 0.38 mg), and placebo. Each treatment was given to each subject once per day on 3 successive days. The study included 31 patients with moderate or severe cancer pain. The two active treatments were equally effective and significantly more effective than placebo.

Stambaugh and Sarajian (1981) compared zomepirac, oxycodone plus aspirin 224 mg, phenacetin 162 mg and caffeine 32 mg (APC), and placebo. Forty patients with advanced malignancy and moderate to extremely severe chronic pain were included. All assessments demonstrated that zomepirac provided greater analgesia than oxycodone plus APC but the differences were not significant.

Stambaugh (1982a) compared single doses of oral butorphanol (4 mg), acetaminophen (650 mg), combination butorphanol/acetaminophen (4 mg/650 mg), and placebo. Each of 20 patients with moderate to severe cancer pain received each of the treatments in succession. Butorphanol alone and acetaminophen alone both provided better pain relief than placebo, although the difference was not significant. Combination butorphanol and acetaminophen was significantly better than placebo or butorphanol alone for at least 6 hours. More patients were sedated after taking butorphanol alone or in combination than after taking acetaminophen alone or placebo. Other side effects of all treatments included nausea, rash, and vertigo.

Stambaugh (1982b) compared single doses of oral zomepirac (100 mg and 200 mg) and parenteral morphine (4 mg and 8 mg) in 40 patients with moderate to extremely severe cancer pain. All patients received each of the four treatments. Zomepirac was equally effective at low and high dose and was as effective as morphine 8 mg. Zomepirac (at both doses) and morphine 8 mg were significantly more effective than morphine 4 mg. Pain relief from morphine 4 mg also had a shorter duration than that from the other treatments. Morphine caused more lethargy than zomepirac.

Stambaugh and Lane (1983) compared single doses of intramuscular meperidine (50 mg), hydroxyzine (100 mg), combination meperidine (50 mg) and hydroxyzine (100 mg), and saline placebo in 30 patients with moderate to severe cancer pain. Each patient received each of the four treatments on 4 separate days. Meperidine and hydroxyzine were equally effective during the first 2 hours. However, hydroxyzine provided pain relief for 6 hours, while the effectiveness of meperidine began to wane after 2 hours. Hydroxyzine was significantly more effective than placebo over the 6 hours studied, but meperidine was more effective than placebo only for the first 3 to 5 hours. Although not discussed in the text, there appeared to be no benefit of combination meperidine and hydroxyzine over hydroxyzine alone. Adverse events were not reported.

Ferrer-Brechner and Ganz (1984) compared methadone plus ibuprofen with methadone alone (plus placebo) in 30 patients with moderate to severe cancer pain. They found that addition of ibuprofen increased analgesia after a single dose without increasing side effects or changing mood levels. The use of a 1-day crossover period weakens the validity of the results.

Tonachella, Curcio, and Grossi (1985) compared diclofenac with pentazocine in 20 patients with moderate to severe cancer pain. Bone pain incidence was 20 percent and 50 percent in the two treatment sequence groups, respectively. Pain intensity (with duration), pain relief by the patient, and global evaluation of analgesic efficacy by the physician were the outcomes of this study. There was a significantly greater reduction of pain with diclofenac than with pentazocine.

Stambaugh and McAdams (1987a) compared oral ciramadol (30 mg or 90 mg) with codeine (60 mg) or placebo in a four-way crossover study of 43 patients. Ciramadol 30 mg and codeine 60 mg were equally analgesic, while ciramadol 90 mg was superior to both.

Stambaugh and Drew (1988a) compared oral ketoprofen (75 or 225 mg) with aspirin plus codeine 60 mg and with placebo in 160 cancer patients. Measures of analgesia derived from pain scores (PID and summed time-weighted pain intensity difference, SPID) demonstrated no significant differences among active treatments.

Sunshine and Olson (1988) compared ketoprofen with morphine in 23 patients with moderate to severe cancer pain. Calculated measures such as PID and SPID demonstrated no significant differences between morphine and ketoprofen.

Stambaugh and Drew (1988b) compared the oxycodone/acetaminophen-sparing effects of ibuprofen with placebo in 30 patients with moderate to severe bone pain due to metastatic cancer. The authors assessed pain relief, side effects, and additional analgesics used. Pain outcome measures are described in a very limited fashion, without variances.

Minotti, Patoia, Roila, et al. (1989) compared diclofenac, nefopam, and aspirin plus codeine in 60 patients with metastatic cancer and moderate to severe pain. The three regimens had similar analgesic efficacy, but diclofenac had a slightly better safety profile. Preference by physicians did not differ between treatments. Only 26 percent of patients completed the planned treatment period.

Puglisi and Garagiola (1989) compared pirprofen (rectally) with pentazocine and placebo in 45 patients with bone and neuropathic cancer pain. Pain relief, side effects, and treatment efficacy rated by the investigator were the assessed outcomes. No difference was observed between the two active treatments.

Estape, Vinolas, Gonzalez, et al. (1990) compared ketorolac with pentazocine in 40 cancer patients with moderate to extreme pain. Sixty percent of this population suffered from bone pain. Pain intensity, pain relief, and overall rating of the treatments demonstrated no significant differences between the two groups with respect to pain severity, pain relief, or additional analgesic medication. Significantly more patients withdrew from the pentazocine group, mainly due to nausea and vomiting. In both groups patients withdrew due to lack of efficacy, suggesting bias.

Carlson, Borrison, Sher, et al. (1990) compared ketorolac with acetaminophen plus codeine and also with placebo in an acute study (6 hours) and the two active treatments in a chronic study (7 days) in 75 patients with mixed types of cancer and moderate to severe pain. They found that acetaminophen plus codeine provided a small but significant advantage in mean daily pain relief compared with ketorolac, and produced slightly fewer side effects.

Bjorkman, Ullman, and Hedler (1993) compared diclofenac (by rectal suppository) with placebo as an adjuvant to intravenous patient-controlled analgesia (IV-PCA) morphine in 16 patients with severe cancer pain who were already receiving morphine. The authors found no significant difference in average VAS for pain between the two groups. However, diclofenac decreased morphine consumption, suggesting an opioid-sparing (but not necessarily a synergistic analgesic) effect.

Staquet and Renaud (1993) compared single-dose piroxicam (40 mg), codeine (60 mg), and combination piroxicam and codeine (20 mg/30 mg) in 88 patients with cancer pain. All three treatments were equally effective, with peak effectiveness at 3 hours after administration. Side effects included nausea, vomiting, drowsiness, and "others," with no significant difference among the treatments.

Bosek and Miguel (1994) compared the efficacies of ketorolac and morphine administered via IV-PCA for postoperative pain in 92 patients with cancer. They assessed pain relief, side effects, the amount of study medication used, and the amount of supplemental morphine. Pain relief was similar in both groups. Total morphine and the incidence of opioid-related side effects was higher in the morphine group. The authors suggest that ketorolac supplemented with a small amount of morphine is associated with lower incidence of nausea, vomiting, and pruritus than morphine alone.

Dellemijn, Verbiest, van Vliet, et al. (1994) compared naproxen with MS Contin in 20 patients with cancer-related neuropathic pain. They compared pain relief, use of rescue medications, side effects, and overall evaluation. The two treatments had equal analgesic efficacy and neither had superior patient preference. (This study is included in a meta-analysis.)

Rodriguez, Barutell, Rull, et al. (1994) compared dipyrone in a high and a lower dose with oral morphine (10 mg) in 121 patients with cancer and somatic and/or visceral pain. Assessing pain relief and side effects, the authors found dipyrone (in the higher dose) to have comparable efficacy with morphine. Dipyrone at both doses tended to be better tolerated (i.e., to produce fewer side effects) but this difference was not statistically significant. (This study is included in a meta-analysis.)

Chary, Goughnour, Moulin, et al. (1994) investigated the dose-response relationship of Codeine Contin (100 mg, 200 mg, or 300 mg q12h) in comparison to acetaminophen plus codeine (600 mg/60 mg) q12h. The outcomes assessed were pain intensity and pain relief. The doses of 200 mg and 300 mg Codeine Contin were most effective while a dose of 150 mg Codeine Contin was judged to be equipotent to acetaminophen plus codeine (600 mg/60 mg) in terms of efficacy and side effects. Analgesia and adverse effects were dose-dependent. Constipation, dizziness, fatigue, headache, itching, lightheadedness, nausea, sleepiness, and vomiting were observed in all treatment groups.

Minotti, De Angelis, Righetti, et al. (1998b) compared diclofenac alone with diclofenac plus codeine and also with diclofenac plus imipramine in 184 patients with moderate to severe pain. They assessed pain intensity, depression, efficacy (global evaluation by investigators), and spontaneously reported adverse effects. The combinations of diclofenac with codeine or imipramine were not different from diclofenac alone. On the basis of these results the authors suggested that the WHO second analgesic step is not an optimal treatment strategy. (This study is included in a meta-analysis.)

Supplemental analysis: Comparison of results from meta-analysis of randomized controlled trials and meta-analysis of nonrandomized studies

Comparison of results from meta-analysis of randomized controlled trials with results from meta-analysis of nonrandomized studies was one of the aims of the supplemental analysis of nonrandomized evidence. We planned to perform this comparison on the question of the efficacy of NSAIDs versus a combination of NSAIDs plus weak or strong opioids. However, we were not able to perform the meta-analysis of nonrandomized studies due to the extreme heterogeneity of these data and the inadequacy of reporting of the studies, thus precluding a comparison with results obtained from a meta-analysis of randomized controlled trials.

We searched extensively for nonrandomized studies that compared the administration of NSAIDs with a combination of NSAIDs plus weak or strong opioids in separate groups with identifiable data on changes in pain intensity scales. The same literature search strategy was used as in the main report, with the exception that it was extended to include all study designs, not only randomized trials.

A total of 25 studies were retrieved and scrutinized but none of the data could be compared with the results of the meta-analysis of randomized controlled trials. The reasons were as follows:

1. Lack of an arm receiving NSAIDs plus opioids (Grond, Zech, Schug, et al., 1991b [NSAIDs vs. other nonopioid analgesics]; Hughes, Wilcock, and Corcoran, 1997 [all patients were switched from opioid plus NSAIDs to NSAID]; Martino, Emanueli, Mandelli, et al., 1978 [NSAIDs vs. NSAIDs]; Pellegrini, Massidda, Pellegrini, et al., 1983 [NSAIDs vs. morphine vs. placebo]; Yalcin, Gullu, Tekuzman, et al., 1997).

2. Lack of an NSAID-only arm (Blackwell, Bangham, Hughes, et al., 1993; Joishy and Walsh, 1998; Mercadante, Sapio, Caligara, et al., 1997; Myers and Trotman, 1994).

3. Case reports (two patients in Lauretti, Reis, Mattos, et al., 1998).

4. No data given for the pertinent comparison. Data were reported for other comparisons such as per day/week of treatment, or per type of pain, typically for all patients combined regardless of treatment received (Cohn, Machado, Bier, et al., 1988; Goisis, Gorini, Ratti, et al., 1989; Grond, Zech, Schug, et al., 1991b; Grond, Radbruch, Meuser, et al., 1999; Mercadante, 1999b; Mercadante, Casuccio, Agnello, et al., 1999; Schug, Zech, and Dorr, 1990; Toscani, Barosi, Scazzina, et al., 1989; Wenk, Diaz, Echeverria, et al., 1991; Zech, Grond, Lynch, et al., 1995).

5. Data given for pertinent arm, but pertaining to outcomes other than pain intensity scores (typically dichotomous variables such as pain relief or effective pain control) (Mercadante, Maddaloni, Roccella, et al., 1992; Takeda, 1990; Ventafridda, Caraceni, and Gamba, 1990).

6. No actual pain scores were given, only reference to statistically significant difference (Gottlieb, 1990). Scores given per arm, but no data on standard deviation or standard error to allow statistical comparison Ventafridda, Tamburini, Caraceni, et al., 1987).

2.1. What is the efficacy of NSAIDs in the management of bone pain?

We found only one study addressing the question of efficacy of ibuprofen compared with placebo in a population suffering only from bone pain due to metastatic cancer (Stambaugh and Drew, 1988b). Pain outcome measures are poorly described. No other studies addressing the question of efficacy of NSAIDs in cancer patients suffering specifically from pain attributed to skeletal metastases were found. Studies that addressed the same question in patient populations with mixed types of pain (i.e., including other types of pain besides bone pain) were evaluated under main question 2.

2.2. What are the side effects of the different opioid analgesics?

Summary of evidence on the side effects of opioids or NSAIDs from uncontrolled trials evaluating the implementation of the WHO ladder

[See Evidence Table 19]

We identified 14 nonrandomized observational studies examining the implementation of WHO guidelines (or ladder) for management of cancer pain. We reviewed the seven studies with the largest sample sizes (at least 50 patients) that reported on adverse events associated with the WHO guidelines. Five were prospective cohort studies. All had pain relief or quality of life as primary outcomes. Five studies described collection of adverse events information in a prospective manner.

All but one study reported on the three steps in the WHO guidelines (Step 1, nonopioid analgesics; Step 2, adding a weak opioid; Step 3, substituting a strong opioid). Grond, Zech, Schug, et al. (1991) compared use of NSAIDs with "other non-opioid analgesics... such as dipyrone or [acetaminophen]." The trials all had a long duration with means from 1 to 2 months and ranges for individual subjects from 1 to 5 years.

Most studies reported on adverse events related to NSAIDs and opioids. Of note, adverse event rates, in general, were lower than those in studies of individual drugs or classes of drugs that we reviewed.

Nausea and vomiting were reported in 6 percent to 22 percent of subjects, more commonly in subjects in Step 3 than in Step 2, and in Step 2 than in Step 1. Constipation occurred among 3 percent to 36 percent of subjects; it occurred with increasing frequency in higher steps in the one study that reported on all three steps. Sedation occurred in 14 percent to 46 percent of subjects, again with increasing frequency in higher steps in the one study that reported on all three steps. Dry mouth was reported in 8 percent of subjects in one study and from 35 percent to 51 percent of subjects in another (with increasing frequency with higher steps). Dyspepsia occurred in 3 percent to 16 percent of patients with no difference in frequency in the one study that reported on all three steps. Bleeding was rare, occurring in less than 6 percent of subjects. Pruritus was also rare, occurring in less than 8 percent of subjects.

Other rare reported adverse events included sweating, anorexia, urinary disorders, diarrhea, restlessness, tremor, vertigo, unsteadiness, allergic reaction, confusion, and gastric hemorrhage. In one study, agitation occurred in 18 percent to 26 percent of subjects.

2.3. What is the efficacy of opioids, antidepressants, and anticonvulsants in treating neuropathic pain?

[See Evidence Table 6]

Summary of the evidence from randomized controlled trials on the efficacy of treatments for neuropathic pain

Table 8. Grading of individual randomized controlled (more...)

Table 8. Grading of individual randomized controlled trials on miscellaneous interventions for the treatment of neuropathic pain

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Pud, 1998 98243129	15 (13)	4.6 ± 0.6 (for amantadine group) 5.4 ± 0.6 (for placebo group)	++	A	C
Kalso, 1995 96303779	20 (15)	"at least moderate" 3.3 (range 1.4-6.2) in breast scar 5 (range 1.7-7.1) in ipsilateral arm	++	A	B
Ellison, 1997 97398217	103 (99)	6 (in this study, median=mean)	+++	A	A

*Number of evaluable patients shown in parentheses if different from enrolled.

In the study by Pud, Eisenberg, Spitzer, et al. (1988), patients with cancer and surgical neuropathic pain lasting more than 3 months were given amantadine (200 mg intravenous infusion) or placebo in a crossover trial conducted in a pain-relief unit. Amantadine reduced mean spontaneous pain intensity significantly more than placebo and also reduced "wind-up" pain in response to repeated pin pricks in the four patients in whom this was present. Amantadine had no effect on thermally or mechanically evoked pain.

In an outpatient crossover study of women with neuropathic pain after mastectomy, Kalso, Tasmuth, and Neuvonen (1995) compared oral amitriptyline with placebo. Amitriptyline (weekly escalating doses of 100 mg daily or as tolerated) significantly reduced pain intensity scores compared with placebo. However, four women dropped out as a result of adverse effects, a fifth did not adhere to the protocol, and only three of the remaining 15 chose to continue on the drug after the study.

Ellison, Loprinzi, Kugler, et al. (1997) administered capsaicin cream or a placebo cream in a crossover trial involving 99 patients with cancer and postsurgical neuropathic pain. Capsaicin cream significantly reduced pain scores over the 8-week test period, although it also caused substantial side effects such as skin burning and redness during the initial application in significantly more patients than placebo. Despite these side effects, among patients expressing a preference, capsaicin was preferred over placebo by a three-to-one margin.

2.4. What is the efficacy of complementary therapies (e.g., herbs, vitamins) in the management of cancer pain?

In a university pain clinic, Li, Cao, Xie, et al. (1994) studied the effects of Chinese herbs, ear-acupuncture, and epidural morphine on postoperative pain in 16 men with liver cancer. Any combination that included at least one of the three treatments provided better pain relief than placebo. Pain reduction and meperidine use did not differ significantly between any combination of these three therapies during the first 5 postoperative days. Each treatment block consisted of two patients.

We did not identify any other studies investigating complementary therapies such as herbs or vitamins. Although we excluded studies on postoperative pain, this study is discussed because it is the only one that employed a randomized control trial (RCT) design in this area and also enrolled patients with cancer at any stage of the disease.

2.5. What are the patient preferences, efficacy, costs, and side effects of different opioid analgesics (e.g., morphine versus hydromorphone)

Summary of evidence on relative opioid potency on cancer pain determined from single-dose, randomized controlled trials

Table 11. Grading of individual randomized controlled (more...)

Table 11. Grading of individual randomized controlled trials on opioid potency

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size 1	Internal validity	Applicability
Houde, 1960 Clin. Pharmacol. & Therap.	67 morphine vs. placebo 28 morphine vs. aspirin vs. morphine and aspirin combination	NR	+	B	B
Beaver, 1966a 67015217	38	NR	0.16	B	B
Beaver, 1966b 66154982	40	NR	0.53	B	B
Beaver, 1969 69239166	28 (23)	NR	0.25	B	B
Beaver, 1977 77141298	34 (26) oxymorphone vs. morphine vs. placebo; 33(28) oxymorphone study (po vs. IM)	NR	8.7 oxymor-phone/ morphine 0.16 oxymor-phone oral /intramu-scular	B	B
Beaver, 1978a 79029266	43(37) oral vs. intramuscular codeine; 17(13) oral vs. intramuscular oxycodone	NR	0.57 oral/intramuscular codeine; 0.5 oral/intramuscular oxycodone	B	B
Beaver, 1978b 79029237	34 (28) oxycodone vs. morphine; 30 (26) codeine vs. oxycodone vs. morphine	NR	0.68 to 0.78 oxyco-done vs. morphine study 8.44 to 11 codeine vs. morphine	B	B

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

1

All effect sizes are for relative potencies of drugs compared, based on total pain relief.

In the 1960s and 1970s a group of investigators (Beaver, Wallenstein, Houde, and Rogers, 1960-1978) evaluated the relative analgesic potency of a variety of opioid analgesics (morphine, profadol, oxymorphone, codeine, methotrimeprazine, and oxycodone) in patients with cancer. In all of these studies morphine was the standard and in some placebo was used as well. Pain intensity and pain relief were measured using categorical scales (pain intensity: 0=none, 1=slight, 2=moderate, 3=severe; and pain relief: 0=none, 1=slight, 2=moderate, 3=lots, 4=complete). Side effects were also recorded but without active questioning of patients. The analgesic assays (with few exceptions) were performed by administering a low and a high dose of the "test" and "standard" opioid medication to the same patient on different days using a double-blind crossover design. Measured outcomes include total pain relief (the area under the time-effect curve for each patient), hour-to-hour change in pain intensity from baseline (pain intensity difference from baseline pain), peak effect (based on individual reports of greatest effect within 3 hours of drug administration), and the corresponding ridit transformations of raw values, to account for distributions other than the normal distribution. The ridit transformation is similar to a probit, except that an observed rather than a theoretical distribution is the basis for the transformation. Dose-effect curves were constructed for all outcomes and compared for parallelism. Statistical confirmation of parallelism (observed in all cases) was considered a criterion for the validity of the assay and the resulting derived relative potency. In addition, the sensitivity of each assay was monitored.

This innovative design employed cancer pain as a model and produced reliable and reproducible estimates of the relative potency of test medications in this context. This experimental design complemented other techniques for assessment of experimental pain in volunteers developed around the same time (Smith, Egbert, Markowitz, et al., 1966). The use of ridit transformations of raw data (pain relief and pain intensity) often increases assay sensitivity. The crossover design allows each patient to be his or her own control, thus eliminating between-patient variability in pain intensity as a potential source of bias. Baseline pain intensity or information about the pathophysiologic substrate of pain is not reported in any of these studies. Thus, relative potency ratios of opioid analgesics are assumed to apply in the whole range of baseline pain (mild, moderate, severe) and pathophysiologic mechanisms (nociceptive, neuropathic). The majority of patients in these studies had been exposed to opioid analgesics prior to enrollment, suggesting potential tolerance to opioid test drugs. However, the existence or precise influence of tolerance on the results cannot be estimated because the duration of previous exposure and type of opioids used are not reported.

Disadvantages of this design in relation to generalizability include the mode of administration (single injection versus repeated administration in clinical practice) and exposure to other analgesics during the nonstudy hours.

Houde, Wallenstein, and Rogers (1960) described two randomized double-blind studies: the effects of morphine versus placebo, and the effects of morphine versus aspirin. Patients with advanced cancer were asked about the severity of their pain. When they required medication for pain, coded test medications were administered according to a randomized dosage schedule. Observations were made for a 6-hour period after administration of the test drug or until pain returned to the premedication level and an additional analgesic was required. Patients were also asked to report whether or not the test medication relieved at least 50 percent of their pain. Pain intensity was categorized on a 4-point scale (none, slight, moderate, and severe), and relief on a 5-point scale (none, slight, moderate, a lot, and complete). Total pain relief for the 6-hour period, calculated as the sum of the products of pain relief and its duration in hours, ranged from 0 (no relief) to 24 points (complete relief for 6 hours). The first study compared morphine 10 mg and saline placebo, both administered intramuscularly under double-blind conditions in a random order. The second study compared the analgesic effects of intramuscular morphine, oral aspirin 200 mg, and the combination of morphine and aspirin. Morphine produced considerably more relief than placebo in these patients for at least 5 hours after administration. The peak effect was noted at the first hour after morphine and saline administration when the scores were 1.5 and 0.9 respectively. Placebo was about one-third less effective than morphine. In the second study aspirin proved superior to the placebo, and aspirin plus morphine was superior to morphine alone. Morphine was significantly more effective than aspirin, but there was no significant difference between the combination of morphine and aspirin and morphine alone.

Beaver, Wallenstein, Houde, et al. (1966a) performed a single-dose, double-blind crossover evaluation of the relative analgesic potency of graded intramuscular doses of pentazocine and morphine in 38 patients with chronic pain due to cancer. The analgesic potency of pentazocine, based on "total" effect, is estimated to be approximately one-sixth that of morphine. An analysis of the analgesic response of patients in relation to their prior narcotic experience suggests that cross-tolerance between pentazocine and other opioids is low. Doses from 10 to 80 mg of pentazocine precipitated acute opioid abstinence reactions in some patients with substantial prior opioid experience. Other adverse effects such as sedation, dizziness, lightheadedness, headache, nausea, vomiting, sweating, dry mouth, pruritus, and pain on injection produced by pentazocine were qualitatively similar to those of morphine.

Beaver, Wallenstein, Houde, et al. (1966b) performed a double-blind crossover evaluation of the relative analgesic potency of methotrimeprazine and morphine in patients with cancer. The time-effect curves of the two drugs appeared to be similar, suggesting to the authors that one drug acts as a "diluent" of the other and that relative potency estimates should be approximately the same whether derived from measures of peak or total effect on pain relief. Considering either "total" or "peak" effects, the analgesic potency of methotrimeprazine was estimated to be approximately one-half (0.53) that of morphine. A significantly larger number of patients had side effects with methotrimeprazine than with morphine, largely due to the pronounced sedative effect of methotrimeprazine. Nausea and vomiting were much less frequent following administration of methotrimeprazine than morphine.

Beaver, Wallenstein, Houde, et al. (1969) evaluated the relative potency of profadol (CI-572) and morphine in a double-blind crossover comparison of graded single intramuscular doses in 23 patients with chronic pain due to cancer. Profadol was found to be about one-fourth as potent as morphine. The time-effect curves of the two drugs were similar. Profadol produced more injection-site pain than morphine, but other effects of the two drugs were similar in type and incidence.

Beaver, Wallenstein, Houde, et al. (1977) evaluated the relative analgesic potency of oral and intramuscular oxymorphone using a double-blind crossover comparison of graded single doses in patients with chronic pain due to cancer. When both duration and intensity of analgesia were considered (i.e., total effect), oral oxymorphone was one-sixth as potent as the intramuscular form. In terms of peak effect, however, oral oxymorphone was only one-fourteenth as potent. These ratios are almost identical to those obtained in a previous study comparing oral with intramuscular morphine. The analgesic effect of oral oxymorphone reached its peak later, and had a longer duration, than that of intramuscular oxymorphone. Intramuscular oxymorphone and morphine were also compared in a similar patient group. Intramuscular oxymorphone proved to be 8.7 times as potent as morphine in terms of total analgesic effect and 13 times as potent in terms of peak effect. In roughly equianalgesic doses, the occurrence of side effects was qualitatively and quantitatively similar for oral and intramuscular oxymorphone, and for intramuscular oxymorphone and intramuscular morphine.

Beaver, Wallenstein, Rogers, et al. (1978a) compared the analgesic potency of oral and intramuscular codeine using a double-blind crossover comparison of graded single doses in patients with chronic pain due to cancer. When both duration and intensity of analgesia were considered (i.e., total effect), oral codeine was six-tenths as potent as the intramuscular form. This is a high oral/parenteral analgesic relative potency ratio compared with morphine, metopon, and oxymorphone and correlates well with the results of recent studies, which have determined the oral versus intramuscular bioavailability of codeine in humans. Oral and intramuscular oxycodone were also compared in a similar patient group. Like codeine, oxycodone retained at least half of its analgesic activity when administered orally. The authors hypothesized that the high oral/parenteral relative potency ratios of codeine and oxycodone relative to morphine and its congeners are not due to more efficient absorption after oral administration, but rather because methylation at position 3 in codeine and oxycodone protects these drugs from rapid first-pass metabolism.

Beaver, Wallenstein, Rogers, et al. (1978b) compared relative analgesic potency of single graded intramuscular doses of oxycodone and morphine in a double-blind, within-patient crossover study in patients with chronic pain due to cancer. In a second part of this study, they determined the relative potency of codeine, oxycodone, and morphine. Important information on the methodology (design, pain assessment) is not explicitly reported but is implied to be the same as in previous work of the same group of investigators. When both intensity and duration of analgesia are considered (total analgesic effect), oxycodone was two-thirds to three-fourths as potent as morphine, while in terms of peak analgesia, it was eight-tenths as potent or equipotent. In doses producing equivalent peak effect, oxycodone had a shorter duration of action than morphine. Intramuscular oxycodone was also compared with intramuscular codeine for cancer pain. In terms of total analgesic effect, oxycodone was 10 times as potent as codeine, while in terms of peak analgesia it was 12 times as potent. These relative potency relationships of oxycodone, taken in conjunction with the oral/parenteral potency ratios of codeine and oxycodone established in a previous paper and several previous relative potency assays involving morphine, oxymorphone, and codeine, demonstrate a highly consistent pattern of analgesic structure-activity relationships encompassing morphine, oxymorphone, codeine, and oxycodone. The investigators point out that the results of these studies do not support the hypothesis that, in humans, the analgesic activity of codeine is due to its O-demethylation to morphine. It is not clear if patients in the first part of the study participated in the second part, or vice versa.

Summary of evidence on relative cost efficacy and preference of opioids in cancer pain

Table 51. Drug prices for dose equivalents for opioid-naive (more...)

Table 51. Drug prices for dose equivalents for opioid-naive adults and children > 50 kg body weight²

Drug	Approximate Equianalgesic Dose		A = Average 1999 wholesale Price $/Unit mg or ml B = Drug price per daily dose 3
	Oral	Parenteral	Oral A B		Parenteral A B
Opioid agonist
Morphine	30 mg q3-4h (repeat around- the-clock dosing) [60 mg q3-4h (single dose or Intermittent Dosing)]	10 mg q 3-4h	1.70/30 mg tab [note: 60 mg dose not computed for table]	$10.17 - 13.56	0.83/10mg	$4.97 - 6.64
Morphine, Controlled-release (MS Contin, Oramorph) Kadian	90-120 mg q12h 20-100 mg q12 or 24h	N/A N/A	1.74/30 mg tab $1.27/20 mg tab	$10.45 - 13.93 $1.28 - 12.75	N/A N/A	N/A N/A
Hydromorphone (Dilaudid)	8 m q 3-4h	1.5 mg q4h	1.41/8 mg tab	$8.46 - 11.28	0.23/1 mg	$0.91
Levorphanol (Levo-Dromoran)	4 mg q6-8h	2 mg q6-8h	0.56/2 mg tab	$3.36 - 4.51	3.96/20 mg/ml	$1.19 - 1.58
Meperidine (Demerol)	300 mg q2-3h	100 mg q3h	1.60/100 mg tab	$38.70 - 58.05	0.63/100 mg/ml	$5.03
Methadone (Dolophine, other)	20 mg q6-8h	10 mg q6-8h	0.19/10 mg tab	$1.11 - 1.48	0.75/10 mg/ml	$2.25 - 3.00
Oxymorphone (Numorphan)	N/A	1 mg q3-4h	N/A	N/A	2.86/1mg.ml	$17.16 - 22.89
Oxycodone (Roxicodone)	30 mg q4-6h	N/A	0.31/5 mg tab	$7.44 - 11.16	N/A	N/A
Oxycodone Controlled-release (OxyContin)	60 mg q12h	N/A	1.17/20 mg tab	$7.02	N/A	N/A
Combination opioid/NSAID preparations
Codeine with aspirin/Acetaminophen (as/ac)	180-200/mg q3-4h	130 mg q3-4h	0.61/30 mg tab	$21.98 - 34.16	0.96/30 mg/ml	$34.56 - 53.76
Hydrocodone with as/ac (in Lortab, Vicodin, etc)	30 mg q3-4h	N/A	0.48/30 mg tab	$2.91 - 3.87	N/A	N/A
Oxycodone with as/ac (in Percocet, Percodan)	30 mg q3-4h	N/A	0.90/30 mg tab	$5.43 - 7.23	N/A	N/A

2

Drug and dosing data from Jacox, Carr, Payne, et al., 1994

3

Price does not include price of injection device, skilled nursing care, etc.

Table 52. Cost of Fentanyl Patch (Transdermal)

Table 52. Cost of Fentanyl Patch (Transdermal)

Dosage Strength	Average Wholesale Price $ / Unit Patch (duration of 48-72 hours)
25 mcg per hour patch	$ 11.76
50 mcg per hour patch	$ 18.50
75 mcg per hour patch	$ 29.63
100 mcg per hour patch	$ 36.91

Table 53. Randomized comparisons of opioids for the (more...)

Table 53. Randomized comparisons of opioids for the treatment of cancer pain

Morphine

Hydromorphone

Codeine

Oxycodone

Heroin

Tramadol

Fentanyl

Methadone

Pentazocine

Butorphanol

Buprenorphine

Alfentanil

Ciramadol

Diamorphine

Dextropropoxyphene

Placebo

Morphine

20

1

2

1

1

2

2

1

1

1

1

1

Hydromorphone

3

1

Codeine

1

2

Oxycodone

Heroin

Tramadol

Fentanyl

Methadone

Pentazocine

1

1

1

Butorphanol

Buprenorphine

Alfentanil

Ciramadol

Diamorphine

Dextropropoxyphene

Placebo

1. Studies comparing the same opioid, by the same route, using different formulations, modes of administration, or dosing schedules.

2. Studies comparing an opioid with placebo.

3. Studies comparing different opioids administered by the same route.

4. Studies comparing the same or different opioid(s) administered by different routes.

Table 12. Randomized controlled trials reporting on (more...)

Table 12. Randomized controlled trials reporting on efficacy, comparing an opioid with another opioid or placebo, administered through the same route

Primary author, year, unique identifier	Opioid	Control	Route
Opioids vs. Placebo
Stambaugh, 1987b 87274551	Dezocine, Butorphanol	Placebo	Intramuscular
Dhaliwal, 1995 96135506	Codeine	Placebo	Oral
Farrar, 1998 98213107	Fentanyl	Placebo	Oral, transmucosal
Comparing different opioids through the same route
Twycross, 1976 77185960	Diamorphine	Morphine	Oral
Ventafridda, 1983 83256803	Buprenorphine	Pentazocine	Oral
Ventafridda, 1986 87059284	Methadone	Morphine	Oral
Pasqualucci, 1987 87288547	Buprenorphine	Morphine	Epidural
Grochow, 1989 89385650	Methadone	Morphine	Intravenous
Hill, 1992 93026543	Alfentanil (PCA)	Morphine (PCA)	Intravenous
Wilder-Smith, 1994 94242672	Tramadol	Morphine	Oral
Heiskanen, 1997 98074866	Oxycodone (CR)	Morphine (CR)	Oral
Mercadante, 1998 98155409	Dextropropoxyphene	Morphine (CR)	Oral

Table 13. Grading of individual randomized controlled (more...)

Table 13. Grading of individual randomized controlled trials comparing the effects of an opioid with placebo, administered through the same route

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Stambaugh, 1987b 87274551	60	7.2 (range 4.7-9.2) dezocine group (D) 6.8(range 4.9-10) butorphanol group (B) 6.9 (range 4.8-100) placebo group (P)	P<0.001 (D vs. P) 1 P=0.004 (B vs. P) 1	A	A
Dhaliwal, 1995 96135506	35 (30)	NR	+++	B	B
Farrar, 1998 98213107	92 (72)	NR	+++	A	A

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

1

P values represent statistical comparisons of 6-hour cumulative scores of total pain relief between pentazocine and placebo and butorphanol and placebo groups, respectively. Data on other pain outcomes are available in Evidence Table 4.

Table 14. Grading of individual randomized controlled (more...)

Table 14. Grading of individual randomized controlled trials reporting on the effects of an opioid compared with another opioid, administered through the same route

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Twycross, 1976 77185960	699 (89)	NR (in male patients:18.8 mm in favor of morphine, in female patients: 2.8 mm in favor of morphine)	++ (male patients) ± (female patients) in favor of morphine	I	C
Ventafridda, 1983 83256803	86 (60)	7.2 ± 0.5 (buprenorphine group, B) 8.1 ± 0.4 (pentazocine group, P)	P < 0.01 in favor of B 1	C	A
Ventafridda, 1986 87059284	66 (54)	NR	± 2	C	B
Pasqualucci, 1987 87288547	12	> 5 (did not respond to NSAIDs)	NR	C	C
Grochow, 1989 89385650	23 (18)	NR	±	B	C
Wilder-Smith, 1994 94242672	25 (20)	3.2 ± 0.8 in tramadol-first group; 3.1 ± 0.8 in morphine-first group (5-point verbal scale, VRS)	±	C	C
Heiskanen, 1997 98074866	45 (27)	NR	++	B	B
Mercadante, 1998 98155409	32	>4 (patients not responding to nonopioid drugs)	±	C	B

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

1

Comparison between the two groups was performed on an integrated pain score. The average daily pain score was obtained by the following calculations: a) "computing the correct number of hours of pain" [in each 24-hour interval]; b) multiplying the number of hours by the corresponding values (on a 5 point descriptive scale: 0 = no pain, 1 = slight, 2.5 = moderate, 5 = severe, 7.5 = terrible, and 10 = excruciating); c) adding the products obtained for each of the seven days of observation; and d) dividing the total by seven, i.e., the number of days in a week. The scores ranged from 0 to 240. The comparison reflects the first week of study (before the crossover). Note that the VAS scale that was used to report baseline pain intensity was not used for the integrated pain scores.

2

Integrated pain scores (0-240 scale) were not different between methadone and morphine groups. A 63% increase of morphine dose during the 14-day study period (initial average = 72.7 ± 39.2, final average = 119.4 ± 79.1) was significant as compared to a nonsignificant (p = 0.69) change of methadone dose.

Table 15. Grading of individual randomized controlled (more...)

Table 15. Grading of individual randomized controlled trials addressing miscellaneous questions related to the use of opioids for cancer pain

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Hoskin, 1989 90087279	20 (19)	NR	±	A	C
Cherny, 1994 94247657	168	"relatively few had severe/excruciating pain"	NA	B	B

*Number of evaluable patients shown in parentheses if different from enrolled.

NA = Not applicable;

NR = NR = Not reported.

Summary of evidence from randomized controlled trials comparing the efficacy of opioids with placebo

Stambaugh and McAdams (1987b) compared single and multiple doses of IM dezocine (10 mg), IM butorphanol (2 mg), and placebo in 60 inpatients with cancer pain. The authors assessed pain intensity and toxicity. In the single-dose study, peak analgesia with dezocine and butorphanol was similar and both were superior to placebo. In the multiple-dose study, dezocine had a longer duration of analgesia than butorphanol. Dezocine had fewer side effects than butorphanol after both single and multiple doses.

Dhaliwal, Sloan, Arkinstall, et al. (1995) compared oral controlled-release codeine (CR) with placebo in a randomized, double-blind, crossover study of 35 patients with cancer pain. Controlled-release codeine treatment resulted in significantly lower overall VAS pain intensity, categorical pain intensity, and pain scores than placebo. Consumption of "rescue" analgesic was significantly lower with controlled-release codeine than placebo. Side effects were more frequently reported in the group that received codeine. Nausea and vomiting were the most prevalent side effects. The codeine/placebo side effect ratio was 1.16. There was a significant superiority in patient and investigator preference for codeine over placebo. Cost was not evaluated. No other randomized trials comparing codeine and placebo for breakthrough pain were identified.

Farrar, Cleary, Rauck, et al. (1998) compared oral transmucosal fentanyl citrate (OTFC) with placebo for the treatment of breakthrough pain in 92 patients with a wide range of cancer type and pain type. Rescue medication in the form of immediate-release morphine was provided. All patients had relatively stable patterns of breakthrough pain before entering the study. The assessed outcomes were pain intensity (VAS), pain relief (4-point scale), and a global performance evaluation. The active treatment produced significantly larger changes in pain intensity and better pain relief than placebo at all time points. The placebo group required significantly more rescue medication than the OTFC group. The most frequent adverse effects in the active treatment arm were dizziness (17%), nausea (14%), somnolence (8%), constipation (5%), asthenia (5%), confusion (4%), vomiting (3%), and pruritus (3%). Patient preference and cost were not reported. No other randomized trials comparing fentanyl with placebo for breakthrough pain were identified.

Summary of evidence from randomized controlled trials comparing the efficacy and preference/acceptability of opioids with other opioids

A heterogeneous group of nine trials compared the efficacy and side effects of different opioids with those of morphine (six studies) or placebo (two studies), administered by the same route within each study. While the applicability of these studies scored individually is generally low, they offer the only existing evidence with respect to the subquestion. The relative efficacy of morphine and other opioids was compared by means of analgesic consumption, pain intensity, pain relief scores, or a combination of these. Side effects were reported from either structured observations or spontaneous report by patients. In a few studies, side effects were scored for intensity. Only one study in this group reported on patient preference. None of these studies compared opioids with respect to cost or reported on cost-related outcomes. The comparative doses of agents employed are not in each case demonstrated within the same trial to be equianalgesic. The results of opioid drugs compared with morphine are described below.

Summary of evidence on the side effects of orally administered opioids from uncontrolled trials

[See Evidence Table 14]

We identified 68 nonrandomized observational studies of oral opioids. We reviewed the nine studies with the largest sample sizes (n = at least 180) that reported on adverse events of oral opioids. Seven were prospective cohort studies, each examining one to five different oral opioids used for treatment of cancer pain. Two of these studies primarily examined adverse events: Sykes (1998) examined laxative use; Campora, Merlini, Pace, et al. (1991) examined emesis. The remainder had pain relief or quality of life as primary outcomes. All of these except Vijayaram, Ramamani, Chandrashekhar, et al. (1990) described collection of adverse events information in a prospective manner.

Eight studies examined a total of seven opioids (buprenorphine, methadone, morphine, oxycodone, dextropropoxyphene, pentazocine, and codeine). The ninth study reported on "strong" and "weak" opioids without further description. The studies reported a wide range of average daily opioid dosages (for example, from approximately 19 mg/day to 60 mg/day of oxycodone, and from approximately 80 mg/day to 380 mg/day of morphine). Subjects were followed for about 1 month or less, with a range of 3 days to 4 months.

Reporting of adverse events was varied across studies. Payne, Mathias, Pasta, et al. (1998) reported only a percentage of "no" or "not bothersome" adverse events. Sykes (1998) reported only on constipation requiring laxative use. Campora, Merlini, Pace, et al. (1991) reported only on nausea and vomiting. More important, the definition of adverse events varied across studies. Few studies explicitly defined symptoms. Some studies apparently reported all nausea and vomiting, while others limited reporting to "moderate to severe" nausea and two or more episodes of vomiting. Camporo, Merlini, Pace, et al. (1991) apparently defined nausea and vomiting as mutually exclusive events. No definitions of constipation or sedation (drowsiness) were reported. In addition, adverse event rates were reported as either percentage of patients with symptoms or percentage of patient-days with symptoms.

Seven studies reported on nausea and/or vomiting. Nausea (including vomiting) occurred in 7 percent to 50 percent of subjects; vomiting in 5 percent to 40 percent of subjects. In these studies, nausea and vomiting do not appear to be related to opioid dose. Constipation occurred in 24 percent to 37 percent of North American and European subjects and from 11 percent to 73 percent of patient-days. Vijayaram, Ramamani, Chandrashekhar, et al. (1991) reported that only 11 percent of subjects experienced constipation, and ascribed this low rate to the high-fiber Indian diet. Across studies, the rate of constipation was not related to the dosage of opioids. Sykes (1998), however, reported 57 percent of subjects had constipation while on "strong" opioids, while 37 percent had constipation on "weak" opioids. De Conno, Ripamonti, Sbanotto, et al. (1991) reported about the same rates of constipation with five different opioids.

Four studies reported rates of sedation. Sedation occurred in 23 percent to 54 percent of subjects and on 2 percent to 48 percent of patient-days. Three studies reported on dry mouth, occurring in 29 percent to 38 percent of subjects and on 49 percent of patient-days. Four studies reported on pruritus, which occurred in 8 percent to 11 percent of subjects and on 4 percent to 12 percent of patient-days. De Conno, Ripamonti, Sbanotto, et al. (1991) reported that trembling occurred in 8 percent to 21 percent of subjects, vertigo in 9 percent to 15 percent of subjects, agitation in 13 percent to 23 percent of subjects, and sweating in 10 percent to 27 percent of subjects. In general, no associations were found between strength of opioid and the above adverse events, except that pentazocine had higher rates of neuropsychological adverse events than other opioids.

No episodes of respiratory depression, hypotension, or coma were reported, although only De Conno, Groff, Brunelli, et al. (1996) explicitly reported on respiratory depression and coma. Only two studies reported on discontinuation of oral opioid due to adverse events. De Conno, Groff, Brunelli, et al. (1996) reported that of 196 subjects, 10 discontinued methadone because of sedation and three because of constipation. Vijayaram, Ramamani, Chandrashekhar, et al. (1991) reported that of 223 subjects, three discontinued morphine because of severe vomiting.

Sykes (1998), in a subsample of 298 subjects, reported that increased opioid dosage was associated with increased number of doses of laxatives, but not with a change in stool frequency. Ventafridda, Oliveri, Caraceni, et al. (1987), in a study of oral morphine, reported that the frequency of nausea decreased with increased dosage, while the frequency of vomiting remained stable.

Summary of evidence on the side effects of parenteral opioids from uncontrolled trials

[See Evidence Table 15]

We identified 46 articles presenting nonrandomized observational studies of parenteral opioids for treatment of cancer pain. We reviewed the seven studies with the largest sample sizes (n >= 50) that reported on adverse events of parenteral opioids. Five were prospective cohort studies. One studied subcutaneous oxycodone (which is not available in the United States); the rest studied morphine and/or hydromorphone given subcutaneously or intravenously. All of the studies examined pain relief or quality of life as primary outcomes. Only three studies described collection of information on adverse events in a prospective manner.

The studies reported a wide range of average opioid dosages, which varied in part due to different drugs and different routes. Most studies followed subjects for about 1 month.

Reporting of adverse events varied across studies; most reported on a wide range of adverse events. The definition of adverse events varied across studies. Few studies explicitly defined symptoms. The studies that included both morphine and hydromorphone, or subcutaneous or intravenous injections, did not report different adverse event rates for the different drugs or routes.

Six studies reported on nausea and/or vomiting. Nausea (including vomiting) occurred in 0 percent to 15 percent of subjects; vomiting when reported separately from nausea occurred in 0 percent to 1 percent of subjects. Constipation occurred in 0 percent to 70 percent of subjects in five studies. The large range of rates of constipation is likely due to unreported differences in definition of constipation and different laxative regimens.

All studies reported rates of sedation; however, sedation was variably described as fatigue, mild sedation or drowsiness, sedation, or severe sedation. Fatigue occurred in 17 percent of subjects, mild sedation in 51 percent of subjects, undefined sedation in 0 percent to 12 percent of subjects, and severe sedation in 4 percent to 6 percent of subjects. Five studies reported on local skin irritations or bleeding, which occurred in 3 percent to 9 percent of subjects; four studies report on local skin infections with rates of 0 percent to 8 percent. In various studies, myoclonus, confusion, dizziness, and seizures occurred in less than 10 percent of subjects. In different studies, hallucinations, mental clouding, dry mouth, and sweating were either very rare (0%-6%) or common (15%-32%).

Respiratory depression occurred in 0 percent to 2 percent of the subjects in studies evaluating subcutaneous opioids and 18 percent of the subjects receiving intravenous morphine. Hypotension was not reported. Only three studies reported on discontinuation of oral opioid due to adverse events, ranging from 0 percent to 4 percent. One patient discontinued subcutaneous oxycodone due to sedation. The rest of the subjects who discontinued parenteral opioids did so due to local irritations.

No study correlated opioid dose to adverse events.

Summary of evidence from randomized controlled trials addressing miscellaneous questions related to opioid use for cancer pain

Hoskin, Poulain, and Hanks (1989) studied the effect of a loading dose at the outset of switching from aqueous morphine to controlled-release morphine. They enrolled 20 patients with various types of cancer (breast, prostate, lung, colorectal). Placebo or aqueous morphine was administered at the same time as the first dose of controlled-release morphine. Pain intensity and side effects were the assessed outcomes. The authors demonstrated that a loading dose is not necessary when changing from aqueous to controlled-release tablet formulations of morphine.

Cherny, Thaler, Friedlander-Klar, et al. (1994) performed a combined retrospective analysis of four controlled, single-graded-dose analgesic studies to assess the relationship between inferred pain mechanism and response to an opioid drug. They compared IM injections of morphine or heroin, a single oral dose of heroin, and a single dose of another study drug. One-hundred sixty-eight patients with various types of cancer and both nociceptive and neuropathic pain were studied. Analgesic outcome was assessed by total pain relief (TOTPAR) (summary score = percentage of maximal possible pain relief represented in the area under the curve of changes in the VAS pain relief versus time). TOTPAR scores of patients with neuropathic pain were significantly lower than that of patients with nociceptive pain. No numerical pain data were reported.

2.6. What are the efficacy and side effects of the following adjuvant analgesics in the management of cancer pain: steroids, anticonvulsants (e.g., gabapentin), antidepressants (e.g., selective serotonin reuptake inhibitors), local anesthetics, hydroxyzine, psychostimulants (e.g., methylphenidate, cocaine), diphenhydramine, clonidine, and NMDA blockers (e.g., ketamine, dextromethorphan)? What is their efficacy alone and as co-analgesics with opioids?

[See Evidence Tables 5 and 6]

Summary of evidence from randomized controlled trials reporting on the efficacy and side effects of drugs used as adjuvant analgesics

Table 16. Grading of individual randomized controlled (more...)

Table 16. Grading of individual randomized controlled trials reporting on the efficacy and side effects of drugs used as adjuvant analgesics

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Antidepressants (amitriptyline vs. trazodone)
Ventafridda, 1987 88114560	45	NR 1	±	B	C
Psychostimulants (methylphenidate vs. placebo)
Bruera, 1987a 87078087	32 (28)	6.4 ± 3.1	++	B	B
Bruera, 1992 92270272	20 (19)	NR	±	B	B
Wilwerding, 1995 95291643	43 (34)	<4	±	B	C
Psychostimulants (cocaine vs. placebo)
Kaiko, 1987 88095975	34	NR	±	I	C
Anticonvulsants (phenytoin as adjuvant to buprenorphine)
Yajnik, 1992 92388770	75	range 6-10	+	C	C
Somatostatin and related analogs (octreotide vs. placebo)
De Conno, 1994 94223136	9	NR (range 5-7 from data in figure)	±	B	B
Alpha-2 adrenergic agonists (clonidine vs. placebo)
Eisenach, 1995 96058975	85	3.8 ± 2.6 (mean ± SD, only patients with neuropathic pain)	++	A	B
Local anesthetics (xylocaine)
Ellemann, 1989 92216257	10	NR (patients had "cutaneous allodynia")	±	A	B
Calcium channel antagonists (nimodipine)
Roca, 1996 97170164	42 (32)	mean 6.08	±	B	B
Santillan, 1998 98359644	54 (30)	range 0-2	+	B	C
Cholecystokinin antagonists (proglumide)
Bernstein, 1998 98318906	60 (43)	mean 3.44	±	C	B

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

1

An integrated scale for pain intensity and duration of pain was used, and baseline pain intensity was not reported.

Summary of evidence from randomized controlled trials comparing different dosing schedules, modes of administration, or formulations of the same opioid

Table 17. Grading of individual randomized controlled (more...)

Table 17. Grading of individual randomized controlled trials comparing different dosing schedules of oral controlled-release morphine

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Portenoy, 1989 89248821	51 (49)	"moderate"	±	A	B
Mignault, 1995 96045260	27 (19)	"moderate or severe"	±	B	C

*Number of evaluable patients shown in parentheses if different from enrolled.

Table 18. Grading of individual randomized controlled (more...)

Table 18. Grading of individual randomized controlled trials comparing two modes of epidural administration of morphine

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Gourlay, 1991 92107537	29 (28)	"optimized oral therapy with opioids and other adjuvant drugs could no longer provide effective analgesia without unacceptable side effects"	±	C	B

*Number of evaluable patients shown in parentheses if different from enrolled.

Table 19. Grading of individual randomized controlled (more...)

Table 19. Grading of individual randomized controlled trials comparing two modes of subcutaneous administration of hydromorphone

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Bruera, 1988a 88317010	25	NR	±	B	C

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

Table 20. Summary of grading of randomized controlled (more...)

Table 20. Summary of grading of randomized controlled trials comparing different formulations of morphine (controlled-release tablets and aqueous solution) administered orally

Number of studies	Patients enrolled/evaluated	Internal validity	Applicability
8	317/244 (77% evaluable)	A = 1 B = 5 C = 2	A = 0 B = 7 C = 1

Table 21. Grading of individual randomized controlled (more...)

Table 21. Grading of individual randomized controlled trials comparing different morphine formulations (controlled-release tablets and aqueous solution) administered orally

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Hanks, 1987 88021688	27 (18)	NR	±	C	B
Goughnour, 1989 89248823	29 (17)	"chronic severe pain"	±	A	B
Thirlwell, 1989 89248820	28 (23)	NR	±	B	B
Ventafridda, 1989 89381485	70 (64)	NR (50/100 on graph)	±	C	B
Walsh, 1992 98358029	33 (27)	NR	±	B	C
Deschamps, 1992 93132420	20 (12)	"sufficient severity to warrant the use of opioids"	±	B	B
Panisch, 1993 95096729	73 (49)	5.9 ± 1.3 (0-10 VAS) 2.4 ± 0.5 (0-3 scale)	±	B	B
Finn, 1993 93253444	37 (34)	24.5 ± 2.7 in nurse rating scale on Day 1 (dose stabilization)	±	B	B

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

Table 22. Grading of individual randomized controlled (more...)

Table 22. Grading of individual randomized controlled trials comparing two oral controlled-release formulations of morphine

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
O'Brien, 1997 98074866	85	NR	±	A	B
Broomhead, 1997 97405392	169 (152)	NR	±	B	C

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

Table 23. Grading of individual randomized controlled (more...)

Table 23. Grading of individual randomized controlled trials comparing controlled-release with immediate-release hydromorphone

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Hays, 1994 94363641	48 (44)	NR	±	C	B
Bruera, 1996 96208853	95 (75)	NR	±	A	B

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

Portenoy, Maldonado, Fitzmartin, et al. (1989) compared two dosage strengths of MS-Contin -- one 100 mg tablet with three 30 mg MS-Contin tablets -- q 12h x 3 days using a randomized, parallel, double-blind, repeated-dose design in 51 cancer patients with moderate to severe pain. The authors assessed pain intensity using a 5-point categorical scale and side effects and found comparable results for both study arms.

Mignault, Latreille, Viguie, et al. (1995) compared two treatment schedules of controlled-release morphine (q8h and q12h) in 27 patients with cancer pain. Assessed outcomes were pain intensity, pain relief, global outcome, rescue analgesics, and side effects. Treatment schedules did not differ significantly with respect to pain intensity, pain relief, and global efficacy scores. The need for supplemental medication did not differ between the two schedules.

Gourlay, Plummer, Cherry, et al. (1991) compared efficacy and side effects of intermittent bolus morphine with continuous morphine infusion, both administered epidurally, in 29 patients in whom "optimized oral therapy with opioids and other adjuvant drugs could no longer provide effective analgesia without unacceptable side effects." Outcomes (pain score, pain relief, and side effects) were assessed every 2 weeks until withdrawal from the study or death. No significant difference on VAS for pain or side effects was found between the two regimens. Patients receiving infusions showed a trend of borderline significance to improved performance in the symbol/digit test (attention and processing) over time, but this trend was not significantly greater than that seen in the group receiving bolus doses. There was no significant difference between the two groups in tests that assessed memory and vigilance. The study duration was approximately 140 days.

Bruera, Brenneis, Michaud, Macmillan, et al., (1988a) -- using a randomized, crossover, open-label design -- compared patient-controlled subcutaneous hydromorphone with continuous subcutaneous hydromorphone infusion in 25 patients with various types of cancer pain. They assessed pain intensity and side effects (both using 0-100 mm VAS). The authors found that both delivery methods were similar in regard to effectiveness and side effects during short-term hospital use. Serious side effects that developed during the study were sepsis while on self-injection (one patient), bowel obstruction while on continuous infusion (one patient), and organic brain syndrome and death while on continuous infusion (one patient).

Eight studies comparing controlled-release morphine with oral morphine solution did not report any significant difference between the two treatments with respect to analgesic efficacy (reduction of pain intensity or increased pain relief). A population of 317 patients with a wide range of cancer types as well as pain types was enrolled, from which 244 were evaluated (78.7%). Although the majority of these trials were designed to be double-blind, the overall internal validity of these studies is scored as B (median) because of the high dropout rate of 10 to 40 percent. Overall applicability is scored as A.

The above studies all addressed the same study question, and a meta-analysis was performed using pain intensity as the outcome of interest. All eight studies provided numerical data on mean pain intensities and standard errors or confidence intervals. Differences in average pain intensity (over 4 to 14 days), measured on a continuous VAS scale (0-100mm), between the two study arms were combined using a random effects model. No difference in average pain intensity was found between controlled-release morphine and morphine sulfate solution, 1.18 mm [95%CI, -1.62 mm to 3.98 mm].

The studies also found no difference between the two formulations with respect to side effects or other outcomes. It is noted that studies of controlled- versus immediate-release products that employ a double-dummy design may not detect the advantage of less frequent administration because both are being consumed. The benefit of fewer doses (and potentially higher compliance rates) with controlled-release tablets (q12h vs. q4h) is a practical advantage of this formulation.

Hanks, Twycross, and Bliss (1987) compared MS-Contin (q12h) with MS oral solution (q4h) in 27 patients with uncontrolled pain from various types of cancer (breast=8, others=9). The authors assessed pain intensity, side effects, sleep, and appetite and found no significant differences between the two regimens in terms of efficacy or adverse effects. The authors noted that MS-Contin did provide a simpler and more convenient treatment regimen, once stabilized.

Goughnour, Arkinstall, and Stewart (1989) compared controlled-release morphine (MS-Contin q12h) with MS oral solution (q4h). Twenty-nine patients with chronic severe pain due to cancer were enrolled in the study and evaluated for pain intensity and supplemental morphine requirements. The authors observed no significant differences between MS-Contin (q12h) and MS oral solution (q4h) on control of chronic severe cancer pain. Tiredness, nausea, and sedation severity scores were not significantly different between the two formulations.

Thirlwell, Sloan, Maroun, et al. (1989) compared controlled-release morphine (MS-Contin q12h) with oral morphine sulfate solution (q4h) in 28 cancer patients. They compared pain intensity, side effects, and additional morphine solution requirements. The authors found no significant differences between MS-Contin and morphine solution in pain scores or side effects.

Ventafridda, Saita, Barletta, et al. (1989) compared MS-Contin tablets with MS oral solution using a randomized, open-label, parallel design in 70 patients with various types of cancer. Pain intensity on a 5-point integrated scale, drug dosage, and side effects were the assessed outcomes. The authors found no significant difference between the two regimens in terms of efficacy. The difference between the two group means was 4.1 (SE = 9.4). The frequency of daily side effects was lower in patients treated with MS-Contin than in those treated with morphine solution.

Walsh, MacDonald, Bruera, et al. (1992) compared oral controlled-release morphine sulfate tablets with oral immediate-release morphine sulfate solution in 33 patients with cancer pain. The authors evaluated the analgesic efficacy, anxiety, sedation, depression, nausea, constipation, confusion, and patient preference. There were no significant differences with respect to pain, breakthrough pain, or side effects between the two formulations.

Deschamps, Band, Hislop, et al. (1992) compared immediate-release with controlled-release morphine (MS-Contin) in 20 patients with various types of metastatic cancer suffering from somatic or visceral pain of "sufficient severity to warrant the use of opioids." The authors evaluated pain intensity (VAS), supplemental morphine for breakthrough pain (as a percentage of daily dose of test drug), side effects (scale 0-3), and patient preference. Eight patients dropped out of the study due to inadequate pain relief. Differences in pain scores, side effects, and supplemental morphine requirement between the two groups were not significant.

Panich and Charnvej (1993) compared oral controlled-release morphine tablets with oral morphine sulfate solution in 73 patients with various types of cancer and cancer pain. Assessed outcomes included pain intensity (VAS by a nurse) and duration of sleep. The two treatments did not differ significantly with respect to pain and duration of sleep.

The study by Finn, Walsh, MacDonald, et al. (1993) compared oral morphine sulfate controlled-release tablets with oral immediate-release morphine sulfate solution in 37 patients with various types of cancer. The authors assessed pain intensity (VAS by patient), incidence of sedation, nausea, anxiety, depression, and breakthrough pain. The two treatments did not differ with respect to pain and side effects.

O'Brien, Mortimer, McDonald, et al. (1997) compared single daily doses of a novel, multiparticulate, controlled-release morphine capsule (60 mg) with twice-daily oral controlled-release morphine tablets (MS-Contin 30 mg) in 85 patients with various types of cancer. Outcomes assessed were pain intensity (VAS) and adverse events (3-point scale) using an instrument, BS-11, that consisted of horizontally arrayed boxes containing the numbers 0 to 10; patients were asked to place a cross on the number corresponding to current pain intensity. There were no significant differences between the two preparations in terms of expressed treatment preference or adverse effects.

Broomhead, Kerr, Tester, et al. (1997) compared Kadian/Kapanol capsules q24hr or q12hr with MS-Contin q12hr or placebo in a population of 162 patients with various types of cancer. Primary measures of efficacy were elapsed time to remedication and total amount of rescue medication, while secondary measures of efficacy were daily VAS for pain intensity in the prior 24 hours, sleep quality, pain intensity on the final study day, and global assessment by the patient. The two treatments were similar in efficacy and safety. There were no significant differences among the treatments for any morphine-related side effects when adjusted for baseline.

Hays, Hagen, Thirwell, et al. (1994) compared immediate-release hydromorphone (q4h) with controlled-release hydromorphone (q12h) in 48 patients with various types of cancer. Assessed outcomes were pain intensity, nausea and sedation, and recorded adverse events. There were no significant differences between treatments in pain intensity, sedation, or vomiting and no differences in proportions between patients and investigators for treatment preference.

Bruera, Sloan, Mount, et al. (1996) compared oral immediate-release hydromorphone (IRH) with oral slow-release hydromorphone (SRH) in 95 cancer patients. Assessed outcomes were pain intensity (VAS and 0-3 scale), analgesic consumption, global assessment, and adverse effects. The total number of rescue doses of opioids, global rating, side effects, and final blinded choice by both patients and investigators did not differ significantly between IRH and SRH.

3.1. What are the patient preferences, efficacy, costs, and side effects of different routes of opioid administration (e.g., sustained release opioids vs. transdermal delivery)?

We identified randomized controlled trials that compared different routes of administration for the same opioid agent and examined whether these studies evaluated the outcomes of interest listed in the subquestion. We found 10 studies comparing opioids administered by different routes. Of these, seven compared the same opioid agent administered by different routes, and three compared different opioids administered by different routes. A separate analysis of each group of studies comparing the same routes for the same opioid follows.

Summary of evidence from randomized controlled trials comparing the same opioid by different routes

Table 24. Summary of grading of randomized controlled (more...)

Table 24. Summary of grading of randomized controlled trials comparing orally with rectally administered morphine

Number of studies	Patients enrolled/evaluated	Internal validity	Applicability
3	72/66 (91.6%evaluable)	A = 0 B = 2 C = 1	A = 0 B = 2 C = 1

Table 25. Grading of individual randomized controlled (more...)

Table 25. Grading of individual randomized controlled trials comparing orally with rectally administered morphine

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Wilkinson, 1992 93099680	11 (10)	NR	±	B	C
DeConne, 1995 95222298	34(34)	>3	++ (in favor of rectal route)	C	B
Babul, 1998 98259851	27 (22)	NR	±	B	B

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

These three studies addressed the same study question, and a meta-analysis was performed using the difference of average pain intensity throughout each study's duration between treatment arms as the outcome of interest. Only two of the three studies provided numerical data of pain scores and standard errors or confidence intervals that could be combined (Babul, Provencher, Laberge, et al., 1998; Wilkinson, Robinson, Begg, et al., 1992). The difference of the average pain intensity (4 to 14 days) was measured on a continuous VAS scale (0-100mm) between the two study arms -- oral controlled-release morphine and rectal controlled-release morphine. The difference in pretreatment versus posttreatment changes in pain intensity between oral and rectal routes for controlled-release morphine, 2.28 mm [95%CI, -4.28 mm to 8.85 mm], was not significant. Although the efficacy of controlled-release morphine is similar between rectal and oral routes, in patients with dysphagia, mucositis, or upper gastrointestinal obstruction there may be obvious benefit to using the former route. It is also noted that the rectal route is not influenced by first-pass metabolism, so a given oral dose is not expected to be equianalgesic with the same dose administered rectally. In fact, all studies of various routes of administration are subject to variation in dose rather than the effect of the route per se.

Wilkinson, Robinson, Begg, et al. (1992) in a pharmacokinetic and efficacy study compared rectal with oral forms of sustained-released morphine using a randomized, open label, crossover design in 11 cancer patients. The concentration-time profiles of morphine, M3G, and M6G were compared between the two routes. Other outcomes included pain intensity and side effects (both on a 10cm VAS). No significant differences in pain or side effects were noted between oral and rectal routes.

DeConno, Ripamonti, Saita, et al. (1995) compared the same dose of morphine hydrochloride (10 mg) administered by the oral or the rectal route ("microenema") in 34 opioid-naive outpatients with various types of cancer and types of pain. Pain intensity on enrollment was >30 mm on VAS. Pain intensity and adverse effects were the assessed outcomes. The authors observed a significant difference in the onset and duration of pain relief in favor of the rectal route. There was no difference in sedation, nausea, or number of vomiting episodes between the two treatments.

Babul, Provencher, Laberge, et al. (1998) compared morphine sulfate controlled-release suppositories (MSC-R q12h) with morphine sulfate controlled-release tablets (MSC-T q12h) in a population of 27 patients with a wide range of cancer types and pain types. The outcomes assessed were pain intensity, amount of rescue analgesics, sedation using VAS scales, and present pain intensity index of the McGill Pain questionnaire. There were no significant differences between MSC-R and MSC-T in overall scores for pain intensity VAS, ordinal pain intensity, and sedation. There was a small but significant difference in overall nausea in favor of the morphine suppository formulation (MSC-R).

Table 26. Grading of individual randomized controlled (more...)

Table 26. Grading of individual randomized controlled trials comparing subcutaneously with rectally administered morphine

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Bruera, 1995 95271279	30 (23)	NR	±	B	C

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

In this study, Bruera, Fainsinger, Spachynski, et al. (1995) compared controlled-release morphine suppositories with subcutaneous morphine in 30 cancer patients with various types of cancer and pain. Outcomes included pain intensity, supplemental analgesics, nausea and sedation, and adverse events. The authors found a small but significant difference in overall ordinal pain intensity scores in favor of the morphine suppository, MS-CRS, but there were no significant differences in overall VAS pain, sedation, and nausea. The use of rescue analgesics did not differ.

This is the only study that compares morphine administration by the subcutaneous and rectal route for cancer pain.

Table 27. Grading of individual randomized controlled (more...)

Table 27. Grading of individual randomized controlled trials comparing subcutaneously with epidurally administered morphine

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Kalso, 1996 97109660	10	3.5 at rest, 7.0 at movement (estimated from figure)	±	A	C

*Number of evaluable patients shown in parentheses if different from enrolled.

Kalso, Heiskanen, Rantio, et al. (1996) compared PCA epidural morphine with PCA subcutaneous morphine in 10 patients. Assessed outcomes were pain intensity (VAS) and adverse effects. There were no significant differences between epidural and subcutaneous administration with respect to pain relief and adverse effects. During both experimental treatments, analgesia was superior and created fewer side effects compared with baseline oral morphine treatment. However, the study was not designed to compare either route with oral administration, nor is it clear what was done to optimize oral therapy. This is the only randomized trial that compares morphine by the epidural and subcutaneous routes, and its reliance on PCA may influence its results.

Table 28. Grading of individual randomized controlled (more...)

Table 28. Grading of individual randomized controlled trials comparing subcutaneously with intravenously administered hydromorphone

Primary author, Year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Moulin, 1991 91124974	20 (15)	"poorly controlled pain requiring subcutaneous opioid infusions"	±	C	B

*Number of evaluable patients shown in parentheses if different from enrolled.

Moulin, Kreeft, Murray-Parsons, et al. (1991) compared efficacy and side effects of continuous subcutaneous and continuous intravenous hydromorphone infusions for 48 hours in patients with various types of cancer and poorly controlled pain requiring subcutaneous opioid infusions. The assessed outcomes were pain intensity, pain relief, mood, and sedation. The authors observed no clinical or statistical differences between the two routes with respect to the assessed outcomes. This is the only retrieved trial that compares hydromorphone by the subcutaneous and intravenous routes.

Summary of evidence from randomized controlled trials comparing different opioids by different routes

Table 29. Grading of individual randomized controlled (more...)

Table 29. Grading of individual randomized controlled trials comparing transdermal fentanyl with orally administered morphine

Primary author, Year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Wong, 1997 97355997	47 (40)	3.9 ± 0.05 3.8± 0.08 in the two study arms	±	C	B
Ahmedzai, 1997 9732840	202	NR	±	C	B

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

The study by Wong, Chiu, Tsao, et al. (1997) compared a transdermal fentanyl patch (1 patch/3 days) with controlled-release oral morphine using an open-label, randomized design in a population of patients with a wide range of cancer types. The outcomes were pain intensity, pain frequency, degree of pain improvement, profile of mood as affected by pain, quality of sleep, and activity status, all assessed using a 5-point scale. Pain relief was observed treatment in both groups, with no significant differences between oral controlled-release morphine and transdermal fentanyl with respect to analgesic efficacy or adverse effects. Drowsiness was encountered in 5 of 20 patients in oral morphine group and in 6 of 20 in the fentanyl patch group. Other side effects were insomnia, anorexia, nausea/vomiting, and constipation, which were comparable in both groups.

Ahmedzai and Brooks (1997) in this British multicenter trial compared the transdermal fentanyl patch (1 patch/3 days) with oral controlled-release morphine (MST, q12h) in an open-label, randomized, crossover design in 202 patients with cancer. Assessed outcomes were quality of life using the WHO scale and the European Organization for Research and Treatment of Cancer (EORTC) QLQ-C30 questionnaire, and the Memorial Pain scale for pain and mood. The WHO and EORTC global assessments showed no significant differences between the two treatments. More patients expressed preference for the fentanyl patch. Fentanyl was associated with less constipation and daytime drowsiness but greater sleep disturbance and shorter sleep duration.

Summary of evidence on the side effects of transdermal fentanyl from uncontrolled trials

[See Evidence Table 16]

We identified nonrandomized observational studies of transdermal fentanyl for treatment of cancer pain and reviewed the eight studies with the largest sample sizes (n >= 30) that reported on adverse events of transdermal fentanyl. All were prospective cohort studies with pain relief or quality of life as primary outcomes. All described collection of adverse events information in a prospective manner.

The studies reported a range of average hourly fentanyl dosages from approximately 100 mcg to 200 mcg, with maximal dosages ranging from 200 mcg to 1000 mcg. Subjects were followed from 7 days to a mean of 158 days.

Reporting of adverse events varied across studies. Although most reported on a wide range of adverse events, Payne, Mathias, Pasta, et al. (1998) described only on "no" or "not bothersome" adverse events. Few studies explicitly defined symptoms. Some studies limited reporting to "moderate to severe" nausea, constipation, and sedation. No definitions of most symptoms were reported. In addition, adverse event rates were reported either as percentage of patients with symptoms or percentage of patient-days during which symptoms occurred.

Five studies reported on nausea and/or vomiting. Nausea occurred in 8 percent to 43 percent of subjects and on 69 percent of patient-days in a study in which subjects were often receiving cancer treatments that could cause nausea; vomiting occurred in 5 percent to 18 percent of subjects. In five studies, constipation occurred in 0 percent to 66 percent of subjects; and in one study constipation occured in 35 percent of patient-days. However, in the study that described a absence of constipation, 42 percent of subjects were on taking laxatives.

Four studies reported rates of sedation. Sedation occurred in 3 percent to 77 percent of subjects. Two studies reported on dry mouth, occurring in 34 percent and 53 percent of subjects. Six studies reported on pruritus and mild local skin reactions, which occurred in 2 percent to 42 percent of subjects. Diarrhea, dyspnea, dizziness/vertigo, sweating, and confusion were also reported in up to half of subjects.

Respiratory depression was reported in 0 percent to 7 percent of subjects. Hypotension occurred in only one subject, who overdosed on opioids. Discontinuation of fentanyl due to adverse events occurred in 13 percent of subjects. The reported reasons for discontinuation were respiratory depression, vertigo, constipation, hallucinations, sedation, and impaired thinking.

Table 30. Grading of individual randomized controlled (more...)

Table 30. Grading of individual randomized controlled trials comparing orally or intravenously administered morphine with orally or intravenously administered oxycodone

Primary author, Year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Kalso, 1990 90263355	20	7.6 average	0	A	B

*Number of evaluable patients shown in parentheses if different from enrolled.

Kalso and Vainio (1990) compared morphine with oxycodone in IV PCA (initial 4 days) and oral formulations (4 days) using a randomized, double-blind, crossover design in 20 patients with metastatic cancer. Patients self-administered analgesics for pain during the initial 4 days of the study, and oral dosages were derived from assumed bioavailabilities of morphine and oxycodone. Outcomes were pain severity and side effects. Morphine and oxycodone effectively relieved cancer pain with both IV PCA and oral administration. An average VAS of 7.6 on a 0-10 scale before opioid titration decreased to 1.1 with both opioids. Morphine caused more nausea and hallucination. At the end of study, 5 patients preferred morphine, 5 preferred oxycodone, and 10 had no preference. However, in this study the route of drug administration was changed in the two respective groups for both opioids at the same time, from oral to IV PCA, and outcomes within the same group were not compared.

Summary of evidence on efficacy of spinal opioids from uncontrolled trials

Table 31. Grading of individual uncontrolled studies (more...)

Table 31. Grading of individual uncontrolled studies on spinal opioids

Primary author, Year, unique identifier	Series size	Baseline pain (VAS 0-10 cm)	Internal validity 1	Applicability
Intrathecal morphine or morphine plus bupivacaine
Madrid, 1988 88187426	100	"no history of strong opioid use"	B2	C
Cheng, 1993 94322423	100	>4 ("intolerable")	B1	B
Nitescu, 1995 95306964	200	No pain outcomes	C	C
Epidural morphine
Zenz, 1985 85186375	139	"severe cancer pain that could not be controlled with conventional opioids"	C2	C
Liew, 1989 89260900	252	"insufficient analgesia or unacceptable side effects"	C1	C
Du Pen, 1990 91052525	350	No pain outcomes	C3	C
Plummer, 1991 91270942	284	No pain outcomes	C3	C
Samuelsson, 1995 95248158	146	No pain outcomes	C3	C

1

See Methods (Chaper 2), "Grading of the evidence for nonrandomized studies" for a description of grading.

Two of these studies evaluated intrathecal morphine administered as intermittent boluses through chronic indwelling subarachnoid catheters. One study evaluated the combination of intrathecal morphine with bupivacaine. Five studies evaluated epidural morphine. Pain relief as an outcome was assessed in two of the three studies on chronic intrathecal therapy. Only one study (Cheng, Tang, Chu, et al., 1993) reported actual pain scores over 12 weeks as well as prestudy pain scores; no statistical analysis was performed. In that study, intrathecal morphine was given from 0.2 mg to 2.0 mg per dose as needed, at a frequency of at least twice per day to provide continuous pain control.

We identified five studies reporting on chronic epidural opioid administration. Two studies (Du Pen, Peterson, Williams, et al., 1990; Plummer, Cherry, Cousins, et al., 1991) do not report on analgesic efficacy. Samuelsson, Malmberg, Eriksson, et al. (1995) report on analgesic efficacy for a limited trial period of 10 days but not beyond that. Zenz, Piepenbrock, and Tryba (1985) in a retrospective study found that epidural opioid administration reduced pain intensity by greater than 50 percent in 87 percent of patients whose pain could not be controlled with conventional analgesic approaches. Finally, Liew and Hui (1989) reported good (57.1%) or excellent (28.2%) pain relief during an initial 3-month observation period of epidural morphine administration via percutaneous catheter. Patients with head or neck malignancy showed relatively poor responses.

Plummer, Cherry, Cousins, et al. (1991) comment that "it is not possible to document efficacy on a retrospective basis when the goal of treatment is to control a subjective symptom such as pain." They suggest that questions of side effects and efficacy can be answered adequately only by prospective studies enrolling groups that receive alternative forms of treatment, such as continuous subcutaneous infusion of morphine.

Summary of evidence on the side effects of spinal opioids from uncontrolled trials

[See Evidence Table 17]

We identified 52 articles presenting nonrandomized observational studies of spinal opioids for treatment of cancer pain. We reviewed the nine studies with the largest sample sizes (at least 100) that reported on adverse events of spinal opioids. Five were prospective cohort studies. The studies included epidural and/or intrathecal morphine or other opioids with or without bupivacaine. Six studies primarily examined pain relief or quality of life as primary outcomes. Two reported primarily on complications of spinal opioid treatment. Only three studies described collection of information about adverse events in a prospective manner.

Intrathecal opioids were given from 0.5 mg to 2.0 mg per dose or day. Mean daily dose of epidural opioids varied from 16 mg/day to 70 mg/day. Most studies followed subjects for a mean of 3 to 5 months.

Reporting of adverse events varied across studies, with some reporting only infections and reasons for discontinuation. The definition of adverse events varied across studies. Few studies explicitly defined symptoms.

Four studies reported on nausea and/or vomiting, which occurred in 9 percent to 40 percent of subjects. Three studies reported on constipation, which occurred in 17 percent to 34 percent of subjects.

Sedation was reported in only two studies, at a rate of 1 percent and 2 percent. Four studies reported on pruritus or skin inflammation, which occurred in 1 percent to 38 percent of subjects. Urinary retention occurred in 4 percent to 73 percent of subjects (four studies), headache in 3 percent to 18 percent of subjects (five studies), and skin breakdown and confusion in 2 percent of subjects (one study each).

Respiratory depression occurred in 0 percent to 1 percent of the subjects in five studies. Hypotension was not reported. Meningitis was raised as a possibility in all studies and occurred in 0 percent to 4 percent of subjects; other catheter-related infections occurred in 0 percent to 9 percent of subjects (six studies). Injection pain occurred in 1 percent to 56 percent of subjects in four studies; various other catheter problems occurred in 1 percent to 8 percent of subjects in four studies. Only four studies reported on removal of catheters and discontinuation of spinal opioids due to adverse events, ranging from 0.3 percent to 10 percent. Catheters were removed primarily because of infection or catheter-specific problems. In addition, removal occurred because of confusion, local pain, hyperesthesia, and nausea.

No study correlated opioid dose to adverse events.

3.5. Is the potential benefit of avoiding sedation and cognitive failure with spinal opioids offset by the risks of chronic spinal catheterization?

We did not identify any studies addressing this question.

Summary of evidence from randomized controlled trials reporting on the efficacy of biphosphonates in the management of cancer-related pain

Table 32. Summary of grading of randomized controlled (more...)

Table 32. Summary of grading of randomized controlled trials reporting on the efficacy of biphosphonates in the management of cancer-related pain

Number of studies	Patients enrolled/evaluated	Internal validity	Applicability
29	3309/3046 (92.1%)	A = 3 B = 12 C = 14	A = 1 B = 5 C =23

Table 33. Grading of individual randomized controlled (more...)

Table 33. Grading of individual randomized controlled trials reporting on the efficacy of biphosphonates in the management of cancer-related pain

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Etidronate
Smith, 1989 89068935	57 (51)	NR	±	B	C
Belch, 1991 91303191	173 (166)	NR	±	B	B
Salmon Calcitonin
Hindley, 1982 83077452	32	>5	++	C	C
Roth, 1986 87015715	40 (38)	NR	+++	C	C
Blomqvist, 1988 88240810	50 (49)	6 and 6.1 in two arms	±	B	C
Pamidronate
VanHolten-Verzantvoort, 1987 88037750	122	NR	+++	C	C
VanHolten-Verzantvoort, 1991 91274037	167 (144)	NR	++	B	B
Van Holten-Verzantvoort, 1993 93187683	205 (161)	NR	+++	C	C
Glover, 1994 95042148	61 (51)	>4 on 9 pain/frequency measurements	+++	C	C
Hortobagyi, 1996 97081204	382 (380)	Percentage of subjects in each arm in 9-point pain/frequency categories: 17%, 14% (0) 40%, 39% (1-3) 43%, 47% (4-9)	++	A	A
Coleman, 1997 98026760	51 (46)	NR	+	C	C
Vinholes, 1997 98157475	52 (48)	median 2.8 and 2.7 (5-point scale) in two arms	++	A	B
Cascinu, 1998 98200996	70 (64)	NR	++	C	C
Clodronate
Siris, 1980 80078029	13 (10)	Severe bone pain was a symptom in 9 of 10 subjects	+	C	C
Elomaa, 1983 83113852	34 (33)	NR	+	C	C
Adami, 1989 89346401	92	Between 10 and 20 on 20-point VAS	+++	C	C
Martoni, 1991 91149007	38 (33)	NR	+	C	C
Lahtinen, 1992 93023374	336	NR	+	B	B
Elomaa, 1992 92324804	75	NR	+	B	C
Ernst, 1992 92166465	24 (21)	NR	++	C	C
Kylmala, 1993 93249770	99 (not stated)	NR	±	C	C
Clemens, 1993 93229572	38 (26)	NR	+++	B	C
Paterson, 1993 93115782	173	NR	+	B	C
Robertson, 1995 95395501	55 (33)	Median (range) Drug: 3.2 (1.6-7.5) Placebo: 4.8 (2.1-6.9)	+	B	C
O'Rourke, 1995 95222321	84 (80)	NR	±	B	C
Strang, 1997 8155699	55 (46)	47 median (5-97 range), 100 mm VAS	±	B	C
Kylmala, 1997 97466816	57 (55)	NR	±	B	C
Ernst, 1997 97348667	60 (46)	NR	+	C	C
McCloskey, 1998 98147943	614 (536)	NR	++	A	B

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

The literature of biphosphonates is quite heterogeneous, with differing inclusion criteria, concomitant medical and radiotherapeutic treatments, disease categories, dosage regimens, choice of agent, and duration of follow-up. Differences in pain assessment are also great, whether directly or indirectly on the basis of analgesic intake or "requirement" for palliative radiation therapy. However, no study showed a negative effect of biphosphonate therapy on skeletal symptoms of metastatic disease or myeloma. In general, positive effects were harder to demonstrate in the presence of concurrent chemotherapy, such as estramustine, which itself might have a positive effect on bone symptoms. Therefore, the literature in aggregate suggests that biphosphonates are effective in reducing pain symptoms from bone involvement by tumor, although the magnitude of this benefit may be reduced when biphosphonate therapy is delivered in conjunction with other tumor-directed therapy that may also reduce such symptoms.

Summary of evidence from randomized controlled trials reporting on the efficacy of radionuclides in the management of cancer-related pain

Table 34. Summary of grading of randomized controlled (more...)

Table 34. Summary of grading of randomized controlled trials reporting on the efficacy of radionuclides in the management of cancer-related pain

Number of studies	Patients enrolled/evaluated	Internal validity	Applicability
4	569/396 (69.6%)	A = 2 B = 1 C = 1	A = 1 B = 2 C =1

Table 35. Grading of individual randomized controlled (more...)

Table 35. Grading of individual randomized controlled trials reporting on the efficacy of radionuclides in the management of cancer-related pain

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
153Samarium-EDTMP
Resche, 1997 98051358	114 (56)	NR	++	C	C
Serafini, 1998 98211795	118 (30)	NR	+++	A	B
Strontium-89
Lewington, 1991 92000881	32 (26)	NR	++	A	B
Quilty, 1994 94316817	305 (284)	NR	++	B	A

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

Summary of evidence from randomized controlled trials reporting on the efficacy of chemotherapy in the management of cancer-related pain

Table 36. Summary of grading of randomized controlled (more...)

Table 36. Summary of grading of randomized controlled trials reporting on the efficacy of chemotherapy in the management of cancer-related pain

Number of studies	Patients enrolled/evaluated	Internal validity	Applicability
7	1138/1106 (97.2% evaluable)	A = 1 B = 2 C = 4	A = 1 B = 2 C =4

Table 37. Grading of individual randomized controlled (more...)

Table 37. Grading of individual randomized controlled trials reporting on the efficacy of chemotherapy in the management of cancer-related pain

Primary author, Year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Schmidt, 1979 79132757	165	NR	±	C	C
Coates, 1987 88065744	308 (305)	NR	++	B	C
Fossa, 1990 91159052	72 (43)	"Quite a bit" or "very much" pain = 50%	NR	C	C
Labianca, 1991 92075588	182	Mean (Range) FA+5-FU: 1.2 (0-9.5) 5-FU: 1.1 (0-8)	NR	C	C
Fraser, 1993 93160019	40	NR	+	B	B
Sullivan, 1995 96117533	210	Pain = 50%	±	A	A
Tannock, 1996 96243733	161	5-point scale: 0 = 1% 1 = 33% 2 = 42% 3 = 19% 4 = 6%	+++	C	B

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

Table 38. Summary of grading of randomized controlled (more...)

Table 38. Summary of grading of randomized controlled trials reporting on the efficacy of hormonal therapy in the management of cancer-related pain

Number of studies	Patients enrolled/evaluated	Internal validity	Applicability
3	468/401 (85.7% evaluable)	A = 0 B = 0 C = 2 I = 1	A = 0 B = 0 C =2 I = 1

Table 39. Grading of individual randomized controlled (more...)

Table 39. Grading of individual randomized controlled trials reporting on the efficacy of hormonal therapy in the management of cancer-related pain

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Da Silva, 1993 94085467	76 (52)	Bone pain: Zoladex = 38% Zoladex+Flutamide = 34%	±	I	I
Rizzo, 1990 90214984	47	NR	++	C	C
Boccardo, 1990 91243718	345 (302)	Pain = 68%	I	C	C

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

The literature on the effects of chemotherapeutic agents on pain is quite heterogeneous, with differing inclusion criteria and different agents. The use of analgesic medication is reported in some of these studies, but the consumption of analgesics is not recorded in most of them. In only one study was there a significant difference in the pain outcome between the two treatment arms. In conclusion, no specific chemotherapeutic agent was demonstrated to improve pain relief over another.

Schmidt, Scott, Gibbons, et al. (1979) compared procarbazine, imidazole-carboxamide, and cyclophosphamide in relapsing patients with advanced carcinoma of the prostate. Pain relief was one of the subjective responses assessed and reported simply as not different between the treatment groups.

Coates, Gebski, Bishop, et al. (1987) compared intermittent and continuous chemotherapy for advanced breast cancer. They assessed five quality-of-life dimensions including pain and the overall-quality-of-life index and reported that continuous chemotherapy is better than intermittent chemotherapy. The pain scores for those in the continuous group were significantly better during the first three cycles of chemotherapy but this difference became nonsignificant in later comparisons.

Fossa, Aaronson, Newling, et al. (1990), in an open, randomized, Phase III trial, evaluated the effect of estramustine versus mitomycin on several morbidity factors, including pain in patients with hormone-resistant prostate cancer and bone pain. The authors report that neither of the two treatments improved pain. No pain data or statistical analyses are reported.

Labianca, Pancera, Aitini, et al. (1991), in a multicenter Phase III trial, investigated the effects of leucovorin as an adjuvant treatment to fluorouracil (FU) in patients with advanced colorectal cancer. The group evaluated bone pain as a primary outcome in two groups, one receiving FU alone and the other FU plus leucovorin. They found similar worsening of bone pain in both groups compared with baseline.

Fraser, Dobbs, Ebbs, et al. (1993) compared standard CMF (cyclophosphamide, methotrexate, and fluorouracil) treatment to epirubicin in treating 40 women with advanced breast cancer. Pain was measured as part of a quality-of-life assessment. The CMF group had significantly better pain scores at 2 months in a 6-month follow-up.

Sullivan, McKinnis, and Laufman (1995) compared scores on the Functional Living Index-Cancer (FLIC) quality-of-life questionnaire in 210 patients with metastatic colon cancer who received either 5-fluorouracil alone or fluorouracil plus leucovorin in a multicenter trial. Patients completed quality-of-life questionnaires at baseline and every 8 weeks thereafter. Symptoms, including abdominal pain and pain from any source, were assessed weekly on a 4-point scale, as were body weight and need for hospitalization. The percentage of patients experiencing pain relief was higher in the combination group, but the difference was not statistically significant. Pain relief improved when body weight remained stable or increased and worsened when the tumor did not respond to treatment.

Tannock, Osoba, Stockler, et al. (1996) treated 161 patients suffering from hormone-resistant prostate cancer with prednisone alone or with prednisone and mitoxantrone. The primary response variable was pain, which was measured on a 6-point scale completed by patients and by analgesic use. Pain declined by at least 2 points in 29 percent of patients receiving combination therapy and in 12 percent of those receiving prednisone alone (P = 0.01). An additional seven patients in each group reduced analgesic consumption by at least 50 percent without an increase in pain. The response to combination therapy lasted 15 weeks, twice as long as that to prednisone alone (P < 0.001). Possible cardiac toxicity was reported in five of 130 patients in the mitoxantrone group.

As with chemotherapy, studies on hormonal therapy having pain as an outcome did not demonstrate any benefit in the management of pain. Pain in these studies was recorded poorly or not at all (no numerical data) and served only as a means for monitoring the progress of cancer.

Da Silva (1993) reported on a subset of 76 patients from a multicenter EORTC trial of metastatic prostatic carcinoma treated either by orchiectomy or administration of flutamide (a depot form of a luteinizing hormone-releasing hormone analog). Pain was assessed with a 4-point scale as a part of a larger quality-of-life study, apparently designed to compare physicians' assessments with patients' assessments. Of the 52 patients with pain at baseline, 25 reported being pain-free at 6 months and 17 at 12 months. No other information is given about the analysis.

In an open trial without a control group, Rizzo, Mazzei, Mini, et al. (1990) administered one of four doses of subcutaneous leuprorelin acetate depot, one injection per week for 4 weeks, to 47 patients with advanced metastatic prostate disease. The authors do not report how pain was measured. Although two-thirds of all patients improved, and 8 of 16 patients with bone cancer were pain-free at 3 months, there were no significant differences among the four study groups.

Boccardo, Decensi, Guarneri, et al. (1990) compared goserelin with goserelin plus flutamide in an open-label trial of patients with prostate cancer. Bone pain was a secondary outcome in their analysis, on which this interim report was based. While no difference between the two groups was observed with respect to progression-free survival, more prompt relief of bone pain was evident in the goserelin plus flutamide group. In this report no numerical or graphic data are presented and no statistical comparisons were made.

4.4. What is the efficacy of external-beam radiation and radionuclides in treating cancer pain?

Summary of evidence from randomized controlled trials reporting on the efficacy of radiation therapy in the management of cancer-related pain

Table 40. Summary of grading of randomized controlled (more...)

Table 40. Summary of grading of randomized controlled trials reporting on the efficacy of radiation therapy in the management of cancer-related pain

Number of studies	Patients enrolled/evaluated	Internal validity	Applicability
14	3859/3571 (92.53%)	A = 1 B = 3 C = 9 I = 1	A = 3 B = 7 C =3 I= 1

Table 41. Grading of individual randomized controlled (more...)

Table 41. Grading of individual randomized controlled trials reporting on the efficacy of radiation therapy in the management of cancer-related pain

Primary author, year, unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Tong, 1982 82233408	1016 (759)	NR	±	C	A
Madsen, 1983 84111122	57	Pain present = 88%	±	C	B
Price, 1986 87042093	288	Single fraction: Mild/mod = 33% Severe = 73% Multiple fraction: Mild/mod = 29% Severe = 48%	±	C	A
Okawa, 1988 89072185	80	Pain = 100%. Baseline data provided in a table	±	C	A
Hoskin, 1992 92187922	270	No pain = 3% Mild, moderate, or severe = 97%	++	C	C
Medical Research Council, 1992 92313868	235	No pain = 41% Mild, moderate, or severe = 59%	±	I	I
Porter, 1993 93239545 N	126	Active arm: 11.3 Placebo arm 10.0 16-point pain/frequency scale	±	B	B
Macbeth, 1996 96409389	509	2-fractions group: Pain = 60% 13-fractions group: Pain = 54%	±	A	A
Teshima, 1996 97073536	38	RT arm: 5.59 RT+MP arm: 6.15 RTOG scale (0-9)	±	C	C
Niewald, 1996 97138004	100	Slight: 26% Moderate: 10% Severe: 64%	±	B	B
Rees, 1997 97281585	216 (187)	NR	±	C	B
Nielsen, 1998 98345210	241 (239)	NR	±	B	B
Bailey, 1998 98408959	356	NR	±	C	B
Jeremic 1998 98418606	327	No pain = 5% Mild = 13% Moderate = 60% Severe =21%	+	C	B

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

Fourteen trials involving a total of 3,859 patients compared various fractional dosing schedules of radiotherapy for painful bone metastases. The dosing schemes were fairly diverse. All reported no overall difference in pain relief between the single dose and fractionated doses. Meta-analyses were not possible due to the heterogeneity of the dosing schedules, the variability in the anatomic sites and fields involved, and the outcomes assessed. Short courses of treatment with moderate doses appear to yield similar results compared with longer courses, and seem preferable for convenience. Some studies suggest the possibility that a single dose may be sufficient, but this is not yet verified. In sum, single-dose (i.e., nonfractionated) radiation does appear to have a similar effect on bone pain as fractionated dosing, although the minimal dose of radiation effective for pain relief was not determined in these studies.

Tong, Gillick, and Hendrickson (1982) compared four dosing schedules in 750 patients with multiple metastases and two dosing schedules in 266 patients with solitary metastases to determine the optimal palliative fractionation schedule for bone pain due to osseous metastases. Overall, 54 percent of the patients received complete pain relief with no significant differences between the various dosing schedules.

Madsen (1983) compared 4 Gy in six fractions over 3 weeks, with two fractions per week, with 10 Gy in two fractions over 1 week in patients with painful bone metastases. Pain intensity was assessed by patient self-evaluation, and satisfactory pain control was achieved in 48 percent of the patients in both treatment groups.

Price, Hoskin, Easton, et al.(1986) compared a single fraction of 8 Gy with 30 Gy in 10 daily fractions. No difference was found in the speed of onset or duration of pain relief between the two treatment regimes. Pain relief was independent of the histology of the primary tumor. The study suggests that mechanisms unrelated to tumor cell killing may be important determinants of pain relief in the early posttreatment period. However, the optimum dose required in a single fraction to produce pain relief was not known.

Okawa, Kita, Goto, et al. (1988) compared three fractional radiotherapy dosing schemes (conventional daily fraction, large once-a-day fraction, and twice-a-day fraction) for painful bone metastases. Overall pain relief was seen in about 75 percent of the patients, and no difference was found among the three groups at the end of treatment. Quicker pain relief was observed in the two groups that received larger daily fractional dose (12-14 days) compared with conventional daily fraction (25 days).

Hoskin, Price, Easton, et al. (1992) compared a single dose of 4 Gy or 8 Gy in treating 270 patients with painful bone metastases. Pain was measured on a 4-point scale as was analgesic use at 2, 4, 8, and 12 weeks. At 4 weeks, the response rate was 69 percent in the 8-Gy group, significantly higher than the 44 percent in the 4-Gy group. At 12 weeks, only 9 percent of the 8-Gy group were retreated with radiotherapy, whereas 20 percent of the 4-Gy group were retreated. Pain scores were not statistically different at 12 weeks.

The Medical Research Council (1992) conducted a trial that compared two regimens of palliative thoracic radiotherapy in 235 patients with inoperable non-small-cell lung cancer. The study compared two fractions of 8.5 Gy 1 week apart with one fraction of 10 Gy. Chest pain, measured on a 4-point scale, was eliminated in about 40 percent of patients from both groups; between 40 percent and 60 percent of patients received at least some relief. The authors do not state whether this difference was statistically significant.

Porter, McEwan, Powe, et al. (1993) evaluated the efficacy of local field external beam radiation with or without strontium-89 as adjunct treatment in reducing pain among men with hormone-resistant prostatic cancer. Pain was assessed by analgesic use and by the Radiation Therapy Oncology Group analgesic and pain scoring system, which rates the frequency and severity of pain on 4-point scales and multiplies the two scores to create a pain score. Index pain sites were recorded at baseline and assessed with pain scores as described. At 6 months, the strontium-89 group had better pain scores, although the difference between groups was not significant. However, the strontium-89 group had significantly fewer new pain sites and did not require further radiotherapy for a mean of 35 weeks, compared with 20 weeks for the radiotherapy-alone group. The strontium-89 group also had significantly higher rates of hematological toxicity.

In a large study,Macbeth, Bolger, Hopwood, et al. (1996) studied 509 patients with nonmetastatic but inoperable non-small-cell lung cancer. One group received 17 Gy in two fractions 1 week apart, and another received 39 Gy in 13 fractions, 5 days/week. Chest pain, a secondary endpoint, was measured with the Rotterdam Symptom checklist. After 3 months, pain scores did not differ significantly between the two groups.

Teshima, Inoue, Ikeda, et al. (1996) administered radiotherapy alone or radiotherapy plus methylprednisolone to 38 patients with metastatic bone tumors. Pain was evaluated by physicians using Radiation Therapy Oncology Group pain scores, and patients completed an unnamed quality-of-life questionnaire. Pain scores did not differ significantly between treatment groups, although data are reported by subgroup, not by treatment group, so statistical power is likely to be low.

Niewald, Tkocz, Abel, et al. (1996) studied 100 patients with bone metastases from primary tumors mostly in the breast, lung, and prostate. One group received a short course of radiotherapy (20 Gy, given as 4 Gy daily over 5 to 7 days) and another received a longer course (30 Gy, given as 2 Gy in 15 daily fractions over 19 to 21 days). Pain was measured on a 4-point scale before and 1 day after treatment and every 3 months thereafter. At the end of treatment, 68 percent of patients receiving short-course radiotherapy and 85 percent of those in the longer course experienced at least partial relief from pain, although the difference was not statistically significant. Mean time between therapy and onset of at least partial pain relief was 13 days in the short-course therapy group and 9 days in the longer-course group, although again the difference was not statistically significant. Given the short life expectancy of these patients, the authors recommend short-course therapy.

Rees, Devrell, Barley, et al. (1997) compared a 17 Gy dose of radiotherapy (in two fractions 1 week apart) with a 22.5 Gy dose (given in five daily fractions) as palliative care for 216 outpatients patients with lung cancer (mean age, 70 years). Patients recorded their symptoms on questionnaires before treatment, on the last day of treatment, every week for 6 months posttreatment, then every 4 weeks thereafter. Chest pain was indicated on a 4-point scale. At least one questionnaire was received from 187 patients. At least some improvement in chest pain was reported by 88 percent of evaluable patients, but the authors note that this improvement was often only slight (one point lower on the scale), and there was no statistically significant difference between the treatment groups.

Nielsen, Bentzen, Sandberg, et al. (1998) compared the effects of a single radiotherapy dose of 8 Gy to four doses of 5 Gy each in the palliative treatment of 241 patients with bone metastases. Pain was measured on a VAS in all patients and by a 5-point scale and analgesic use in about half of the patients. There was no significant pain score difference between treatments at any time during the study.

Bailey, Parmar, and Stephens (1998) compared conventional radiotherapy (1 treatment/day, 5 days/week, for 6 weeks) to continuous hyperfractionated accelerated radiotherapy (3 treatments/day for 12 days) in 356 outpatients with inoperable, localized cancer. Pain was measured with the Rotterdam Symptom Checklist (RSCL) questionnaire, which employs a standard 4-point scale. Pain relief did not differ significantly between the two treatment groups. Results were better in the accelerated radiotherapy group at 3 months but not at 1 or 2 years.

Jeremic, Shibamoto, Acimovic, et al. (1998) administered at random one of three doses of radiotherapy (4, 6, and 8 Gy) to 327 patients with metastatic bone pain for 1 to 8 weeks. Pain was measured with a 4-point scale. Patients receiving 6 and 8 Gy had similar and better responses than patients receiving 4 Gy. The authors conclude that 8 Gy is the minimum dose but acknowledge that 6 Gy should be evaluated more fully and that 4 Gy has some therapeutic value.

Summary of evidence from randomized controlled trials reporting on the efficacy of pain education in the management of cancer-related pain

Pain education

Table 42. Summary of grading of randomized controlled (more...)

Table 42. Summary of grading of randomized controlled trials reporting on the efficacy of pain education in the management of cancer-related pain

Number of studies	Patients enrolled/evaluated	Internal validity	Applicability
5	1428/1192(75.4%)	A = 0 B = 4 C = 0	A = 2 B = 2 C = 0

Table 43. Grading of individual randomized controlled (more...)

Table 43. Grading of individual randomized controlled trials reporting on the efficacy of pain education in the management of cancer-related pain

Primary author, year unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Kravitz, 1996 97063270	97 (87)	Intervention group = 2.03 Control group = 2.96	±	B	A
De Wit, 1997 98074868	383 (313)	18.2 (0-48 range) McGill Pain Questionnaire (0-50)	++	B	B
Trowbridge, 1997 97457514	510 (320)	NR	++	B	B
Elliot, 1997 97281966	438 (320)	2.9	+	B	A

*Number of evaluable patients shown in parentheses if different from enrolled.

NR = Not reported.

The studies listed in Tables 42 and 43 evaluated different interventions in education at the level of the patient, medical staff, and even the community. Also, different types of pain seemed to be addressed though specifics were not always provided. Different approaches to education were employed as well. While the internal validity score of these studies is B (median), their applicability is also B due to the limited number of studies in combination with the diversity of interventions.

The number of studies is small, and given their disparity as to types of intervention and even type of pain, it is difficult to draw any broad conclusions. The individual studies were of acceptable quality. There is considerable room for further studies in this area.

Kravitz, Delafield, Hays, et al. (1996) randomized patients into intervention (n = 40) and control (n = 38) groups to determine if a graphic display of pain levels, similar to the manner in which vital signs are charted, would produce improvement in pain intensity levels. Patients had active cancer prior to entering the study and current pain (or worst pain over the prior 24 hours) of at least 2.5 on a 0-10 scale. There was no treatment effect demonstrated. Weaknesses of the study were that pain reports were charted only once per day and that house staff caring for patients may not have known how to use the data or may have been too busy to respond to it.

In a nonblinded study, de Wit, van Dam, Zandbelt, et al. (1997) evaluated the effect of a pain education program on 313 cancer patients with chronic pain. Two hundred and nine patients were at home without visiting nurse services and 104 were receiving nursing at home. Each of these groups was divided into two groups: a control group and a group that received specific education on cancer pain. The educational intervention consisted of 30-60 minutes of face-to-face instruction; written materials were also left with the patient. In the group not receiving home nursing care, there was a statistically significant decrease in pain intensity in the intervention group. However, there was a high dropout rate in the home nursing services group, primarily due to death; only 31 of 53 of those in the intervention group were able to complete the study.

Trowbridge, Dugan, Jay, et al. (1997) conducted two randomized control trials involving 510 oncology outpatients and 13 oncologists. Two values on a 0-3 pain management index scale were obtained for each patient. Data analysis was performed on only 320 patients who reported cancer-related pain. The intervention consisted of placing a summary sheet with patient's pain assessment in each chart prior to an office visit; physicians were to review this sheet prior to the visit. Prescribing patterns differed in the two groups on follow-up. There was improvement in the patients in the intervention group in only one measure of pain, but no difference between the groups on a pain management index interpreted as detecting undertreated pain.

Elliot, Murray, Oken, et al. (1997) performed a randomized controlled community education intervention. Community opinion-leader clinicians, physicians, nurses, clergy, pharmacists, and social workers attended a 2-day educational program and then returned to their communities to work on outreach. Three hundred twenty patients and family members, 150 nurses, and 124 physicians completed the study. The endpoint was patients' pain intensity scores, which did not show any significant improvement between the control and intervention groups. The investigators felt that the relatively brief duration (15 months) of the intervention may have contributed to its lack of effectiveness.

Hypnosis

Table 44. Summary of grading of randomized controlled (more...)

Table 44. Summary of grading of randomized controlled trials reporting on the efficacy of hypnosis in the management of cancer-related pain

Number of studies	Patients enrolled/evaluated	Internal validity	Applicability
5	315/252 (80%)	A = 3 B = 2 C = 0	A = 5 B = 0 C = 0

Table 45. Grading of individual randomized controlled (more...)

Table 45. Grading of individual randomized controlled trials reporting on the efficacy of hypnosis in the management of cancer-related pain

Primary author, year unique identifier	Study size*	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Zeltzer, 1982 83059117	45 (33)	NR	++	B	A
Wall, 1989 89148152	42 (20)	NR	++	B	A
Syrjala, 1992 92270268	67 (45)	NR	+++	A	A
Syrjala, 1995 96187377	94	NR	++	A	A
Sloman, 1995 96099376	67 (60)	NR	+++++**	A	A

*Number of evaluable patients shown in parentheses if different from enrolled.

**Three study arms. Effect sizes reflect comparisons of two active treatments with one control.

NR = Not reported.

Five randomized studies examine hypnosis in conjunction with cognitive-behavioral techniques involving acute procedure-related pain and oral mucositis pain related to bone marrow transplants (see Tables 44 and 45). They include studies in the pediatric and adult age groups. Hypnosis seems to help with both procedure-related and mucositis-related pain. Cognitive-behavioral treatments may also be helpful. More studies are needed, with larger numbers and with control groups.

Zeltzer and LeBaron (1982) compared hypnosis with nonhypnotic behavioral techniques during bone marrow aspiration and lumbar puncture in 33 patients aged 6-17. Nonhypnotic techniques included deep breathing, distraction, and encouragement of self-control but specifically not fantasy. Pain and anxiety were rated separately by the patients and by an independent observer. Of the 33 patients, 11 underwent bone marrow aspiration solely, 6 underwent lumbar puncture solely, and 16 had both procedures. Hypnosis focused on having the children become involved with fantasy and imagery. Hypnosis was more effective than nonhypnotic techniques in reducing pain associated with bone marrow aspiration and lumbar puncture.

Wall and Womack (1989) compared hypnosis and active cognitive training techniques for their effects on pain in 20 pediatric patients undergoing either lumbar puncture or bone marrow aspiration. Subjects had already undergone one prior lumbar puncture or bone marrow aspiration and got training in one of the two techniques prior to the second procedure. Unfortunately there was no control group. Both techniques seemed to reduce pain; neither was more effective than the other.

Syrjala, Cummings, and Donaldson (1992) randomly assigned 67 patients with hematological malignancies who were undergoing bone marrow transplant (BMT) to four groups treated with hypnosis, cognitive-behavioral skills, therapist contact or a control group that received treatment as usual (i.e., standard therapies for pain, nausea, and emesis but without any psychologist contact). They measured oral pain, nausea/emesis, and opioid use. Forty-five patients completed the study.The hypnosis intervention did reduce oral pain in these transplant patients compared with those in the other arms of the study, but there was no difference as to nausea/emesis or opioid use.

Syrjala, Donaldson, Davis, etal. (1995) compared oral mucositis pain in 94 BMT patients with hematological malignancies who were in one of four treatment groups: therapist support; relaxation and imagery training; cognitive-behavioral coping skills, which included relaxation and imagery; and treatment as usual for the control. Relaxation and imagery reduced pain, used alone or with cognitive-behavioral techniques; there was no additional effect of the cognitive-behavioral techniques. An additional 67 patients were randomized but did not complete the study. The training sessions in the various techniques were initiated prior to hospitalization for BMT.

Sloman (1995) studied 67 people with "intermediate and advanced" various neoplasms who were randomized to one of three groups: relaxation and imagery training by audiotape; relaxation and imagery training by a nurse; and a control group that did not get have specific training. In both treatment groups, pain intensity and severity were reduced, as was the use of nonopioid analgesia as needed. For a reduction in pain sensation, live instruction rather than audiotape was needed.

Summary of evidence from randomized controlled trials reporting on the efficacy of celiac plexus block and splanchnicectomy in the management of cancer-related pain

Table 46. Summary of grading of randomized controlled (more...)

Table 46. Summary of grading of randomized controlled trials reporting on the efficacy of celiac plexus block and splanchnicectomy in the management of cancer-related pain

Number of studies	Patients enrolled/evaluated	Internal validity	Applicability
5	263/variable*	A = 1 B = 0 C = 4	A = 2 B = 3 C = 0

*Patients assessed until death.

Table 47. Grading of individual randomized controlled (more...)

Table 47. Grading of individual randomized controlled trials reporting on the efficacy of celiac plexus block and splanchnicectomy in the management of cancer-related pain

Primary author, year, unique identifier	Study size	Baseline pain (VAS 0-10 cm)	Effect size	Internal validity	Applicability
Ischia, 1992 92197865	61	> 6 in 58% (in all three CPB groups n=20 per group )	±	C	B
Mercadante, 1993 93205431	20	Opioid group 6.6+/-0.7 CPB group 5.5+/-0.4	+	C	C
Lillemoe, 1993 93256637	137	Alcohol group 2.1+/-0.3 Placebo group 2.0+/-0.3	++	A	A
Kawamata, 1996 96377487	21	NSAID+morph. group 4.9 Celiac block group 5.5 (data from figure)	+	C	C
Polati, 1998 98162436	24	NCPB group median 7.5 (range 6-10) Anesthetic block + opioids group 7 (range 6-9)	++ short-term 0 long-term	C	B

Ischia, Ischia, Polati, et al. (1992) compared the efficacy and morbidity of three posterior percutaneous neurolytic celiac plexus block (NCPB) techniques in 61 patients with pain from unresectable pancreatic cancer. The authors assessed the quality of pain according to the Arner and Arner classification (Arner and Arner, 1985) and the intensity of pain before and after NCPB. No statistically significant differences were found among the three techniques in terms of either immediate or "up-to-death" pain relief or survival duration. There were significant differences between the three techniques with respect to transient side effects such as hypotension (higher in transaortic and retrocrural techniques) and diarrhea (higher with transaortic technique). Morbidity was negligible. Pain was abolished in 70-80 percent of patients immediately after the block and in 60-75 percent until death.

Mercadante (1993) compared the analgesic efficacy of celiac plexus block with that of NSAID-opioid sequence according to the WHO ladder using a unblinded, randomized design in 20 patients with pancreatic cancer (dextropropoxyphene was used as the weak opioid and morphine as the strong opioid). Pre-block opioids were continued in both groups, and the selection and dose were titrated to achieve VAS < 4 cm. Measured outcomes were pain intensity, opioid consumption, an integrated score for the two variables, and side effects. No difference was found between the two groups in pain intensity, but the integrated scores (pain intensity and opioid consumption) were significantly lower in the NCPB block group through the fourth week post-block and also on the day before death. The incidence of opioid-related side effects such as nausea and constipation did not differ significantly between the two groups, although the authors stated that most side-effects in the NCPB group were the result of ongoing opioid therapy and not NCPB. Objective criteria to determine adverse events related to the block were not presented. One case of prolonged diarrhea, two cases of orthostatic hypotension (still present 48 hours post-block), and one case of back pain at the site of injection were reported in the celiac block group. In this study celiac plexus block was related to produced "limited and easily controlled" adverse effects in the acute post-block period but had a long-term benefit in reducing opioid consumption and related side effects. The authors conclude that the celiac block reduces opioid related side-effects and that an integrated score (VAS multiplied by the sum of one plus one-tenth of the total daily opioid dose expressed as morphine equivalents) is a useful, objective measure that supports this conclusion.

Lillemoe, Cameron, Kaufman, et al. (1993) using a randomized, double-blind design compared chemical splanchnicectomy intraoperatively using 20 cc of either 50 percent alcohol versus saline placebo in 139 patients with unresectable, proven histologically, pancreatic cancer. These 139 patients were drawn from a sample of 371 patients with suspected pancreatic carcinoma, of whom the majority were excluded because of either a resectable neoplasm (N = 202) or a benign inflammatory condition (N = 30). Pain intensity, mood, and degree of pain interference with activities were the outcomes assessed preoperatively; at baseline; at 2, 4, and 6 months; and before death. Mean pain scores were significantly lower in the alcohol group at 2-, 4-, and 6-month follow-up and at the final assessment (p < 0.05). Patient subgroups with, and those without, preoperative pain both showed significant long-term differences in pain relief between alcohol and placebo. Opioid consumption declined in 70 percent of patients in the alcohol group but in none of the placebo controls (p < 0.001), implying that the incidence of opioid side effects was likely lower, too. A striking finding of this study is that of a longer survival after alcohol than placebo block (p < 0.0001) in the subgroup of patients who had pain prior to laparotomy.

Kawamata, Ishitani, Ishikawa, et al. (1996) compared the effect on pain relief and quality of life of NCPB with "the traditional NSAID-morphine combination" in patients with advanced pancreatic cancer. The authors used an unblinded design in which 21 patients were randomized to receive NCPB (n = 10) or NSAID plus morphine (n = 11). NCPB was accomplished with 15-20 ml of 80 percent alcohol "bilaterally injected" after confirming needle placement and analgesia using 8 ml of 2 percent lidocaine with contrast medium. The oral morphine dosage was increased whenever VAS (0-10 scale) was >3. The range of morphine dose at the initiation of the study was 20-60 mg/day. Subcutaneous morphine was administered when patients were unable to swallow. The study assessed pain intensity, morphine consumption, performance status, quality of life using a multiple-factor-scale and a single-scale (VAS) instrument, and side effects. Differences in VAS pain scores through the first 4 weeks and in morphine consumption for weeks 4-7, inclusive, significantly favored the NCPB group. Multiscale quality-of-life scores did not differ between groups; single-scale scores declined from baseline in the NSAID-morphine group at weeks 8 and 10 but not in the NCPB group. Differences in loss of appetite in weeks 6 and 8 and nausea in week 8 favored the NCPB group, but no other differences were found in side effects between NCPB or morphine treatments. During the interval of significantly higher morphine consumption in the morphine group compared with the NCPB group in weeks 4-7, the prevalence of other opioid side effects (nausea, tiredness, constipation) also increased, but not significantly. The authors propose that "NCPB does not directly improve the quality of life in patients with pancreatic cancer pain, but it may prevent deterioration in quality of life by the long-lasting analgesic effect, limitation of side-effects and the reduction of morphine consumption, compared to treatment only with NSAID-morphine." The number of patients in each treatment group, however, is small, and the statistical analysis of outcomes where repeated measurements were made does not appear to correct for the multiple comparisons that were made.

Polati, Finco, Gotti, et al. (1998), at an academic medical center, treated a population of pancreatic cancer patients using NCPB with a total of 14 ml of absolute alcohol (n = 12) or a total of 12 to 16 ml of 2 percent mepivacaine (n = 12) in addition to diclofenac sodium and, as tolerated or required, oral opioid therapy. Pain scores were significantly lower in the NCPB group at 24 and 48 hours but not at later times. The NCPB group required lower doses of NSAIDs for the remainder of their lives, and lower doses of opioids through half of their post-block survival times. The authors found in their own series, like those of others, that visceral pain progresses to acquire somatic and/or neuropathic features as disease progresses.

6.1 What are the morbidity and mortality of cordotomy in treating cancer pain?

We found no randomized trials addressing this subquestion. We identified the following nonrandomized studies that address this question.

Summary of evidence from case series on the efficacy, morbidity, and mortality of cordotomy in the management of cancer-related pain

Table 48. Outcomes of cordotomy from nonrandomized (more...)

Table 48. Outcomes of cordotomy from nonrandomized studies

Primary author, year, unique identifier	N	Pain relief 1		Survival (months; mean or range)	Complications				Mortality due to cordotomy 2
		Immediate 3	Long term 4		Motor	Respiratory	Urinary	Hypo-tension
Rothbard, 1972 72236751	10	100%	100%	NR	NR	NR	NR	NR	0
Meglio, 1981 82088062	52	92.3%	63%-73%	NR	9%	6.8%	4.5%	2.2%	9%
Cowie, 1982 83019163	56	95%	55%-73%	~15	3.6%	35%	11%	NR	3.5%
Ischia, 1984a5 84271886	69	71%	71%	4.4	NR	NR	7.2%	NR	10.1%
Ischia, 1984b6 84163936	36	97.2%	59.5%	NR	38.9%	0	58.33%	36.1%	19.4%
Gildenberg, 19847 84113768	20	60%	60%	6-24	NR	NR	NR	NR	NR
Nagaro, 19948 95199762	10	80% 9	NR	13.5	20%	0	0	0	0
Fenstermaker, 1994 95199762	6	83% 10	83%	4-10	NR	NR	33.3%	NR	0

NR = not reported

1

Percentage of patients with satisfactory pain outcome as defined in each repor

2

Mortality rate due to the procedure

3

Within 2 weeks after the procedure

4

Beyond 2 weeks and until death, a range is tabulated if more than one assessment was done during follow-up

5

Unilateral cordotomy

6

Bilateral cordotomy

7

Complication rates do not include those patients that could not be assessed because they were bedridden.

8

Results shown only for percutaneous cervical cordotomy (same paper also reported 13 patients treated with subarachnoid phenol block). Sixty percent of patients had general fatigue 1 week after percutaneous cervical cordotomy.

9

VAS (1-10 cm) are reported (mean ± SD). Pain outcome was considered satisfactory when VAS was less than 3. The mean VAS at baseline: 8.5 ± 0.9; at 1 week after the cordotomy: 3 2.7 (p < 0.001).

10

Five of six patients had "good" to "excellent" result.

Table 49. Grading of individual uncontrolled studies (more...)

Table 49. Grading of individual uncontrolled studies on cordotomy

Primary author, year, unique identifier	Series size	Baseline pain (VAS 0-10 cm) 1	Internal validity 2	Applicability
Rothbard, 1972 72236751	10	NR	C3	C
Meglio, 1981 82088062	52	NR	C1	C
Cowie, 1982 83019163	56	NR	C1	C
Ischia, 1984a 84271886	69	>6 (68.9%)	C1	C
Ischia, 1984b 84163936	36	NR	C1	C
Gildenberg, 1984 84113768	20	NR	C2	C
Nagaro, 1994 95199762	10	8.5 ± 0.9 (mean ± SE)	B1	C
Fenstermaker, 1994 95199762	6	NR	C3	C

NR = Not reported

1

Only if a standard VAS scale was used

2

See Chapter 2 for additional information on grading nonrandomized, uncontrolled trials

Rothbard, Kotsilimbas, Jacobson, et al. (1972) report on the outcome of high (C₂) and low (C_5-6) cervical percutaneous radiofrequency cordotomy to relieve intractable pelvic pain in 10 patients with cervical cancer "unresponsive to other forms of definitive or palliative treatment." Seven had undergone radiotherapy, three had had radical surgery, and all had inoperable tumor found on exploratory laparotomy. The authors report that all 10 patients were relieved of their pain for 6 to 29 months during which time "all forms of analgesics, previously necessary, were discontinued." It is unclear whether each patient received a high cervical lesion contralateral to a low one (as implied by the authors) because no details as to the type of lesion(s) in each patient are reported. The authors did not report any acute or chronic complications. No specifics regarding pain or analgesic management in these patients prior to the procedure were provided, nor is any mention made of mortality.

Meglio and Cioni (1981) report on the outcome of percutaneous cervical cordotomy in 52 patients suffering from chronic unilateral cancer pain due to a variety of primary tumors identified according to location rather than histological diagnosis. The mean follow-up period was 11 weeks and the longest follow-up period was 38 weeks. Seven patients were lost to follow-up. The mean duration of pain was 9.1 months, and 50 percent of patients had received or were receiving radiotherapy to the primary lesion. In this case series the mortality was 9.6 percent (5 of 52) and "only complete pain relief was considered because of the difficulties in grading partial pain relief." Results were "excellent" (i.e., free of original pain and without complications) in 73 percent of patients after 1 week and in 63 percent of patients after 15 weeks. Pain recurred in 9 of 43 patients (21%) during the first 2 weeks after the operation and no recurrence of the original pain occurred beyond 3 weeks postoperatively. Observed complications were respiratory dysfunction (6.8%), paresis (9%), bladder dysfunction (4.5%), hypotension (2.2%), and asthenia (2.2%). Respiratory insufficiency, hypotension, and weakness occurred only in patients with lung cancer. Pain (contralateral or at other sites) arose in 38.5 percent within 3 to 30 days after the operation and in nearly all (18 of 20 cases) was due to tumor progression.

Cowie and Hitchcock (1982) reported on their experience with anterolateral cordotomy in a series of 56 patients with intractable pain. Of these, 43 patients suffered from cancer and were followed up for a period of at least 3 years. Forty-nine patients underwent high cervical cordotomy (44 unilateral and five bilateral) and seven underwent thoracic cordotomy. Ninety-five percent of survivors had effective relief on discharge from hospital, 73 percent at 6 months, and 55 percent at 1 year follow-up. "Effective relief" is defined on a 4-point scale as grade I = "no pain, no analgesia required" or grade II = "infrequent and/or mild pain, weak nonnarcotic analgesics effective." The incidence of complications was urinary retention 6 subjects (11%), ataxia 1 subject (~2%), hemiparesis 2 subjects (3.6%), respiratory failure 2 subjects (3.6%), and dysaesthesia 4 subjects (7.1%). Two patients (3.5%) died due to respiratory failure.

Ischia, Luzzani, Ischia, et al. (1984a) present the results of unilateral percutaneous cervical cordotomy for the treatment of pain from vertebral metastases in a retrospective analysis of 69 patients. The primary pathology was mainly breast, lung, and prostate cancer, but other sites were also represented and seven patients had unknown primaries. All of the patients had been previously treated for pain with radiotherapy (43.5%) and/or opioids through enteral, parenteral, or spinal routes. The majority of these patients (68.1%) suffered unilateral pain of severity greater than 6 on a 0-10 scale. The majority (72.5%) had chronic pain, 20.3 percent had chronic plus incident pain, and 7.2 percent had incident pain only. Two patients died within 7 days after the procedure but not necessarily as a result of the procedure. In those patients with unilateral pain (46), cordotomy provided complete relief in 17 while in three pain was partially abolished and in six was persistent. Of the 18 patients with bilateral pain and ipsilateral residual pain after unilateral cordotomy, 14 were well controlled medically. Motor deficit and urinary retention were assessed in all patients except those who were bedridden (27.5%) or catheterized (15.9%), respectively. The authors conclude that the procedure is indicated for oncological vertebral pain, particularly with an incident component, when primary oncological fusion or surgical therapy is not possible.

Ischia, Luzzani, Ischia, et al. (1984b) report the immediate and short-term results and complications of bilateral percutaneous cervical cordotomy in 36 patients with chronic bilateral pain due to cancer. In 30 patients cordotomies were performed separately on each side at intervals ranging from 1 to 2 weeks. In six other patients bilateral cordotomies were performed in a single stage. Cordotomies were initially successful in 35 of 36 patients (97.2%). Within 7 days after the procedure five of 36 patients (14%) died from brain metastases, cachexia, coma, sepsis, and "cardiocirculatory collapse in a patient with coronary disease." Immediately after the procedure 39 percent had motor weakness and 6.9 percent were unable to walk (but 72.4 percent were already impaired such that new weakness could not be assessed). Bladder dysfunction could be assessed in 24 (66%), half of whom had permanent urinary retention (33%) while two had urinary incontinence (7%). Thirteen patients (36.1%) had arterial hypotension (mostly orthostatic). Long-term pain relief in the remaining 32 patients (survival range 2 weeks - 9 months) was complete in 15 (47%) and partial in four (12.5%), while pain was persistent or recurred in a new site in 40 percent.

Nagaro, Amakawa, Yamauchi, et al. (1994) described two case series in a single report. Because percutaneous cervical cordotomy (PCC) was performed at their institution from 1980 to 1986, and subarachnoid phenol block using fluoroscopy (SABP-F) from 1987 to 1991 for pain control in costopleural syndrome, they were able to assess efficacy and complications in 10 and 13 patients respectively. Pain scores were 7 or higher (0-10 scale) in all patients before PCC, which reduced mean pain score from 8.5 ± 0.9 to 3.0 ± 2.7. After PCC, analgesics were unnecessary in four patients. Side effects of the procedure were general fatigue (60%) and hemiparesis (20%). SABP-F reduced pain score from 7.5 ± 1.9 to 2.7 ± 2.6 and allowed discontinuation of analgesics in five patients 1 week after the block. SABP-F produced no complications. The authors concluded that PCC is effective but has a serious risk of complications, while SAPB-F is a safe and effective method for managing chest and/or neck pain from costopleural syndrome.

Fenstermaker, Sternau, and Takaoka (1995) in a technical note described the results of an improved approach for CT-assisted percutaneous anterior cervical cordotomy for the treatment of cancer-related pain in six patients with various tumors. Four of the six patients were resistant to morphine treatment, and three had undergone previous lateral cervical cordotomies on the opposite side. This procedure had "excellent" results in three of the six (50%), a "good" result in two (33%), and "fair" in one (17%). Complications of permanent and transient bladder dysfunction, respectively, were seen in two of six patients. The range of survival was 4 to 10 months.

Evidence-based Practice Center Staff

• EPC/Project Director:

• Joseph Lau, MD

• Assistant Project Director:

• Leonidas Goudas, MD, PhD

• Primary Technical Expert:

• Daniel Carr, MD

• Investigators:

• Rina Bloch, MD; Ethan Balk, MD, MPH; John P.A. Ioannidis, MD

• Statistician:

• Norma Terrin, PhD

• Project Manager:

• Deirdre DeVine, M Litt

• Research Associates:

• Priscilla Chew, MPH; Maria Gialeli-Goudas, LLM

• Technical Editor:

• Thomas A. Lang, MA

• Research Assistant:

• David Liu, BA

Technical Expert Advisory Group

• Jane Ballantyne, MD

• Department of Anesthesiology

• Massachusetts General Hospital

• Boston, Massachusetts

• Claudia Bannon

• American Cancer Society

• Atlanta, Georgia

• Representing: American Cancer Society

• Ada Jacox, PhD, RN

• College of Nursing

• Wayne State University

• Detroit, Michigan

• Representing: American Pain Society

• Leslie Jaggers, Pharm D

• Professional and Scientific Affairs

• American Society of Health System Pharmacists

• Bethesda, Maryland

• Representing: American Society of Health System Pharmacists

• Jacqueline LaPierre, RPH

• Pain Program

• H. Lee Moffitt Cancer & Research Institute

• Tampa, Florida

• Representing: American Society of Health System Pharmacists

• Douglas Merrill, MD

• Valley Pain Treatment Center

• Phoenix, Arizona

• Representing: American Society of Anesthesiologists

• Richard Payne, MD

• Department of Neurology

• Memorial-Sloan Kettering Cancer Center

• New York, New York

• Representing: American Society of Clinical Oncology

• Karen Stanley, MSN, RN

• Claremont, California

• Representing: Oncology Nursing Society

• Vincent Jay Vanston, MD

• Scranton-Temple Residency Program

• Temple University Medical School

• Scranton, Pennsylvania

• Representing: American College of Physicians

Partner Organizations

Partners

American Cancer Society

American College of Physicians

American Pain Society

American Society of Anesthesiologists

American Society of Clinical Oncology

American Society of Health-System Pharmacists

Oncology Nursing Society

Associate Partners

American Academy of Family Physicians

American Academy of Neurology

American Physical Therapy Association

Hospice Association of America

Hospice and Palliative Nurses Association