Forty-five RCTs (2,400 patients) were included.
The following were inadequately described in the primary studies: the randomisation method, concealment of treatment allocation, blinding, and description of withdrawals and drop-outs. The method of randomisation was described in 7 RCTs. In many RCTS it was unclear who was blinded. Patients were instructed in the use of visual analogue scales in a minority of the studies, and most patients were sent home with a questionnaire 2 to 6 hours after surgery. Non-standardised measures were used to record the consumption of supplementary analgesia.
Morphine versus saline in studies where bupivacaine was used as an active control (9 RCTs).
Six RCTs were sensitive, as defined by a significant analgesia effect of bupivacaine compared with saline. All 6 RCTs found a significant effect on pain for intra-articular morphine versus placebo at either early or late times.
Early effect (4 RCTs): one of the 4 RCTs using 1 mg intra-articular morphine found an early effect on pain. Other RCTs that found an early effect on pain used 5 or 10 mg intra-articular morphine.
Late effect (5 sensitive RCTs): all 5 RCTs found a significant effect on pain for intra-articular morphine versus saline.
Analgesic consumption at 24 to 48 hours (4 sensitive RCTs): 3 of the 4 RCTs found significantly lower analgesic consumption for morphine versus saline.
Morphine versus saline with no active bupivacaine control (8 RCTs). Four of the 8 RCTs found significantly lower visual analogue pain scores for intra-articular morphine versus saline in the early period.
Intra-articular morphine versus systemic morphine control (6 RCTs: 3 RCTs used intravenous morphine, 2 RCTs used intramuscular morphine, and one RCT used subcutaneous morphine). Late period: no RCT found any significant difference between intra-articular and systemic morphine for pain.
Total analgesic consumption (4 RCTs): one RCT found significantly reduced analgesic consumption for 1 mg intra-articular morphine versus 1 mg intravenous morphine. Combination of intra-articular morphine and bupivacaine versus intra-articular saline (10 studies of which 8 also included a control group with a combination of intra-articular morphine plus bupivacaine).
Four of the 6 studies that were sensitive to bupivacaine alone, and showed a positive effect for morphine, also showed a significant effect for the combined treatments in the early period. All 5 studies analysed for the late period showed a positive effect for the combination.
Dose-response (5 RCTs using intra-articular morphine alone).
None of the studies had evidence of internal sensitivity. The RCTs found no consistent dose-response effect.
Adverse effects.
Only one adverse effect attributed to intra-articular morphine (1 mg) was reported (pruritus in 1 patient). The effect of intra-articular injections on the knee was not mentioned.