Objectives:

Pain associated with cancer is an important problem for large numbers of patients and their families. This report summarizes published evidence on the prevalence of cancer-related pain and the efficacy of drug and nondrug therapies for its treatment.

Search Strategy:

We identified English language human studies by searching Medline, CancerLit, and the Cochrane Controlled Trials Registry (1966 to December 1998). These searches, supplemented by bibliographies of meta-analyses, selected review articles, and suggestions from our science partners and technical experts, yielded approximately 19,000 titles.

Selection Criteria:

Cancer-related pain was defined as pain caused by cancer, by cancer treatment such as surgery, radiation, or chemotherapy, or by the side effects of treatment. We reviewed observational studies on the epidemiology of cancer pain, randomized controlled trials, and selected nonrandomized studies that assessed the effect of treatments and that met methodological criteria. Our search strategy was not restricted by age, gender, ethnicity, or type of cancer. We did not include studies of acute postoperative pain.

Data Collection and Analysis:

We summarized 24 epidemiological surveys of cancer pain and abstracted results from 188 randomized controlled trials of cancer pain treatment into evidence tables. Each trial was assessed according to its methodological quality and applicability. Meta-analysis was performed when there were sufficient data to address a specific question. We also examined data from 100 nonrandomized studies.

Main Results:

The median number of patients enrolled in randomized trials of primary analgesics (NSAIDs, opioids, and adjuvants) was 70 or fewer. Information about the location, nature, and mechanism of pain before and after treatment was minimal for all interventions examined. Heterogeneous reporting of outcomes, nonuniformity of pain measurements, and incomplete reporting of relevant data precluded all but three meta-analyses.

Nonsteroidal anti-inflammatory drugs (NSAIDs) or opioids independently reduce cancer-related pain, as do adjuvants such as antidepressants or anticonvulsants. Few studies evaluate the safety and efficacy of NSAIDs for cancer pain beyond a few days; many are single-dose studies. The studies we examined neither separated the analgesic efficacy of NSAIDs and opioids nor indicated that NSAIDs are specifically effective for bone pain nor disclosed incremental efficacy of adding a "weak" opioid to an NSAID. Comparisons between dosages and delivery forms of systemically administered opioids are limited. Radionuclides and biphosphonates reduce pain from bone involvement by tumor, as does external beam radiation, although studies of the latter lack no-radiation controls. Neurolytic celiac block is effective. The analgesic efficacy of palliative chemotherapy and hormonal interventions is difficult to estimate because of inadequate data. Physical or psychological treatments appear efficacious, but the number of relevant studies is small. Multidrug (or drug plus nondrug) therapy, spinal drug infusion, and ablative neurosurgery require better-quality evidence.

Conclusions:

Randomized controlled trials establish that many current treatment modalities can individually reduce cancer pain. These trials constitute about 1 percent of the published literature on cancer pain, enroll 1 in 10,000 patients at risk for cancer pain in developed countries, are often heterogeneous, and are often of poor methodologic quality. Many clinical questions remain unanswered and preclinical insights untranslated because of a lack of high-quality evidence. Age, gender, genetics, psychosocial context, and culture affect pain and analgesic efficacy. Multiple mechanisms of cancer pain exist. Despite the importance of pediatric cancer pain control, very few studies focus on children. High-quality trials are needed to advance progress in cancer pain relief.