A total of 159 documents were obtained through the preliminary examination of the database. And 21 potential literatures were considered eligible by reading the title and abstract. After analyzing the full-text articles, 13 studies which did not meet the inclusion criteria were excluded, and 816-23 were found eligible for inclusion according to our criteria for being ultimately included in the Meta-analysis. The selection procedure was described in Fig. 1 which illustrated how the 8 studies were obtained. The 8 selected studies, comprising a total of 600 surgical patients (Oxycodone group: 297 patients; Fentanyl group: 303 patients), all were RCTs investigating the use of oxycodone compared with fentanyl alone for a postoperative IV-PCA strategy. Across the all studies, the mean age of patients ranged from 45.6 to 57.8 years for the IV- PCA, with the majority (> 60%) being female from the three available studies. Of the 8 studies, one study19 compared the oxycodone and fentanyl for IV-PCA up to postoperative 24 hours, and other seven trails16-18,20-23 were 48 hours. The dose of analgesics and surgery type varied among trials, and the main characteristics of included trials are listed in Table 1. The risk assessment of the included studies are shown in Fig. 2 and Fig. 3. Four trials17,18,20,22 detailed the methods of randomization, 5 trials18-22detailed the methods of double-blinding, and 2 trials19,22 clearly reported allocation concealment.
Meta-analysis of NRS scores
The WMDs in patient-reported NRS scores at rest were comparable with signiﬁcant differences at postoperative 8 and 12h between oxycodone and fentanyl for IV-PCA(Fig. 4). For NRS scores at movement, the WMD decreased from -12.76 (95% CI: -25.15 to -0.37, P = 0.04) at postoperative 4h to -1.04(95% CI: -1.42 to -0.66), P < 0.00001) at postoperative 12h, and increased to -3.30(95% CI: -5.66 to -0.94), P = 0.006) at postoperative 24h (Fig. 5). The results for the NRS scores of IV-PCA at all time-points in the included trials have been shown in Fig. 4 and Fig. 5.
Meta-analysis of cumulative PCA dose over 48hours
Data reporting median consumption of infused PCA over 48 hour are described in 6 trials16-19,21,22 with 475 patients. Overall, the meta-analysis showed that the amount of PCA overall the 48 hours postoperative after surgery was signiﬁcantly less in group oxycodone than in group fentanyl (WMD= -12.11, 95%CI -18.42 ~ -5.80, Z=3.76, P = 0.0002, Fig. 6), using the random-effects model (heterogeneity test, Chi2 =113.38, df =5 [P<0.0001], I2 = 96%).
Meta-analysis of incidence of PONV
Data on comparisons of the incidence of PONV of oxycodone versus fentanyl were reported by 6 trials17,18,20-23with 855 patients. The results of our fixed-effects (Chi2 =10.31, df =10 [P=0.41], I2 =3%) meta-analysis in terms of PONV were summarized in Fig. 7. There was a higher incidence of PONV (OR = 2.41, 95% CI: 1.60 ~ 3.63, P < 0.0001) in patients receiving IV-PCA with oxycodone compared with patients receiving fentanyl. Subgroup analysis showed that oxycodone versus fentanyl IV-PCA resulted a significant difference in nauase (OR=2.78, 95%CI: 1.75 ~ 4.43, P < 0.0001), while no significant difference in vomiting was observed between in the two groups (OR=1.43, 95%CI: 0.60 ~ 3.44, P=0.42).
Meta-analysis of patient satisfaction for IV-PCA
The analysis of overall satisfaction in patients for pain management at 48 hours after the surgery was carried in 4 studies17,20-22 with 324 patients. Obvious heterogeneity (Chi2 =9.36, df = 3 [P=0.03], I2 = 68%) was found in the all current included studies. The estimated OR of patient satisfaction for oxycodone versus fentanyl IV-PCA during the 48 hours after surgery was 0.73 (95% CI: 0.11~5.04) using fixed-effects model, but there was no significant difference (P=0.75, Fig. 8).
Meta-analysis of main adverse events
For the safety analysis, we selected several of the most frequent AEs. All the meta-analysis results regarding AEs showed that the no obvious difference was observed between oxycodone and fentanyl group in main adverse events (headache: OR=1.40, 95%CI [0.69, 2.83], P=0.35; pruritis: OR=1.60, 95%CI[0.67, 3.81], P=0.29; sedation: OR=1.52, 95%CI [0.76, 3.04], P=0.24, but the significant increase of dizziness with oxycodone versus fentanyl was found (OR=3.69, 95%CI [2.17, 6.26], P<0.00001)(Fig. 9). There was no report of other adverse events including rash, diarrhea, or bradycardia in either group.
Risk of Bias in Included Studies
Review Manager 5.3 software was used to assess the publication bias by funnel plot, and the details of funnel plot was presented in Fig. 10. As a result, we can observe that there was a certain asymmetry in the funnel plot, indicating that there is some degree of publication bias in the literature. However, the number of studies included was only ten, and the funnel plots may not be very reliable. More over, Egger’s test revealed that there was no significant difference in the ORs of AEs in our study (Z = 1.58; P = 0.46).