The search yielded 17,471 records, of which 39 were eligible for inclusion (Fig. 1). The reasons for exclusion at full-text review are reported in Appendix 5. The included studies investigated 4054 adult participants (1858 males and 2196 females) with a mean age range of 30–75 years and a mean sample size of 103 participants. All included studies were published in English, and 56.4% (n = 22 studies) were from the USA. The studies investigated a variety of orthopaedic surgeries, with the most common being knee arthroplasty (14 studies [37,38,39,40,41,42,43,44,45,46,47,48,49,50]) and hip arthroplasty (eight studies [38, 40, 51,52,53,54,55,56]). Most studies only reported short-term data (i.e. < 1 month after randomisation), and no studies reported outcomes in the long term (> 3 months after randomisation). A summary of study characteristics is provided below and in Appendix 6.

In total, 21 studies [37, 41, 42, 45,46,47,48,49,50,51, 53, 54, 57,58,59,60,61,62,63,64,65] had an overall low risk of bias (Fig. 2), and most scored well in random sequence generation. A total of 18 studies had a high risk of bias, and eight studies had not blinded participants or study personnel, six had not blinded outcome assessors, seven displayed selective outcome reporting, and five had high levels of attrition or missing data.

Summary of risk of bias. Each domain was scored as low risk of bias (+), unclear (?) or high (−) risk of bias

Included studies investigated multimodal analgesics (seven studies [39, 51, 53, 54, 57, 60, 66]), opioid analgesics (three studies [40, 51, 67]), non-opioid analgesics (eight studies [38, 41, 54, 61, 62, 68,69,70]), behavioural interventions (three studies [59, 71, 72]), acupressure or acupuncture (four studies [45, 46, 63, 64]), cryotherapy (five studies [44, 48, 49, 55, 73]), electrotherapy (six studies [42, 43, 47, 50, 56, 65]), and other therapies (five studies [37, 52, 58, 74, 75]).

In total, 28 studies [37,38,39, 42,43,44,45,46,47, 50, 52,53,54,55,56,57,58, 62,63,64,65, 68,69,70, 72,73,74,75] compared their intervention with a placebo/no intervention or usual care control, and 14 [37, 40, 41, 48, 49, 51, 54, 59,60,61, 66,67,68, 71] compared an intervention with an active control, for example, a cryotherapy treatment at a lower temperature [50] or nurse-administered versus patient-controlled analgesia [70]. Further details on the intervention and control groups for each study are outlined in Appendix 6.

Additional pooled and unpooled data (mean scores, CIs, and GRADE scores) are provided in Appendix 7, and meta-analysis results are provided in Appendix 8.

The mean daily opioid dose was reported by all 39 studies. Meta-analyses results are shown in Figs. 3 and 4.

Mean opioid dose medium-term forest plot. CBD cannabidiol, CI confidence interval, df degrees of freedom, IV inverse variance, SD standard deviation