Fixed-interval v as-needed analgesics in cancer outpatients

Regular dosing compared with as-needed dosing of opioids for management of chronic cancer pain: systematic review and meta-analysis

“…However, each paper reported no significant difference between the two populations and so we have taken the results to be representative of cancer pain. The stage, or severity of malignant disease was not clearly documented within each study, five studies [33,42,47,55,56] recorded the presence of metastasis and this ranged from 41-82% . The mean trial duration and follow up was 12 days and 872 (92%) patients completed the trial period.…”

“…In only two studies [12,42] was the primary aim specifically to compare the efficacy of regular dosing with as needed dosing of opioids. The remaining 7 studies aimed to compare the efficacy of a variety of prolonged release or continuously administered opioid preparations in comparison to placebo where as needed immediate release opioid was available for use by both groups.…”

“…A wide range of opioids were used across the studies. McGuire et al [42] used a range of opioids and it was not possible to establish standardised doses for each arm; this study was therefore not included in the opioid dose meta-analysis. In the remaining studies, the opioids studied were: morphine (n=238), buprenorphine (n=634), hydromorphone (n=44) and codeine (n=60).…”

“…In the remaining studies, the opioids studied were: morphine (n=238), buprenorphine (n=634), hydromorphone (n=44) and codeine (n=60). The approach to co-analgesics was variable across the studies with three studies allowing the use of non-opioids [9,20,42], four studies allowed adjuvant analgesia provided it was already established at steady state [7,33,47,55], one study did not allow any non-trial analgesia [56] and another did not control or comment on other medication [12]. In five studies patients were excluded if they had a history or drug or alcohol abuse [11,20,33,35,56].…”

“…Overall, for the primary outcome measure there was an unclear risk of bias associated with random sequence generation and allocation concealment. The risk of blinding was variable across the studies, in particular the older papers [12,20,42] showed high or unclear risk of bias in blinding of participants and personnel. The remaining papers showed low risk in one or both domains.…”