A Comparision of Nalbuphine with Morphine for Analgesic Effects and Safety : Meta-Analysis of Randomized Controlled Trials

Although morphine is the standard opioid analgesic for pain control and has been widely used, certain drug-induced adverse effects have been reported as intolerable and need to be addressed. Nalbuphine may have a few advantages over morphine in this respect. We aimed to describe the effect of nalbuphine as well as its saftey compared to morphine by analyzing published randomized controlled trials (RCTs) with meta-analysis approach. We analysed 15 trials (820 patients). Overall, there was no evidence to show that the effect of pain relief had any difference between nalbuphine and morphine (pooled relative risks [RRs], 1.01; 95% CI, 0.91 to 1.11; P = 0.90). On the other hand, the incidences of pruritus, nausea, vomiting, respiratory depression were significantly lower in nalbuphine group compared with morphine group, and the pooled RRs were 0.78(95%CI, 0.602–0.997; P = 0.048) for nausea, 0.65(95%CI, 0.50–0.85; P = 0.001) for vomiting, 0.17(95%CI, 0.09–0.34; P < 0.0001) for pruritus, and 0.27(95%CI, 0.12–0.57; P = 0.0007) for respiratory depression. The analgesic efficacy of nalbuphine is comparable to morphine, but nalbuphine provides a better safety profile than morphine in the aspect of certain side-effects, especially related to pruritus and respiratory depression.

We consider probability distribution and checking the presence of scaling laws, using Bayesian methods (Bayesian meta-analyses). Instead of producing confidence intervals, Bayesian analyses produce credible intervals (sometimes called probability intervals). A 95% credible interval from a Bayesian analysis is a summary of the posterior distribution, such that the probability is equal to 95% that the true quantity is within the interval. This is a particularly intuitive way to express uncertainty, and is one of the most appealing aspects of a Bayesian analysis. WinBUGS software is now available for performing Bayesian analyses. Relative Risk of pain relief comparing nalbuphine and morphine using random effects model was 1.102(95% credible interval: 0.6697-1.627) based on 1000 simulated values by WinBUGS, which was similar with the outcome of conventional Meta-Analysis (RR: 1.01; 95% CI, 0.91 to 1.11). Thus our study was credible and stable with statistical methods.  Table 2). The heterogeneity between the study estimates was not significant (pruritus :I 2 = 25%, P = 0.63; nausea: I 2 = 32%, P = 0.14; vomiting: I 2 = 0.0%, P = 0.63; respiratory depression: I 2 = 2%, P = 0.36). The results drawn from analyses suggested an advantage of nalbuphine over morphine regarding pruritus, nausea, vomiting and respiratory depression. Overall, from Table 2 we can see that the incidences of all the adverse events of nalbuphine and morphine, and the respective incidences of nausea, vomiting, pruritus and respiratory depression were 0.199, 0.16, 0.047 and 0.075 for nalbuphine, and 0.307, 0.284, 0.206 and 0.197 for morphine.

Sensitivity Analysis and Meta Regression Analysis.
To evaluate the influence of each study, sensitivity analysis was performed. On the one hand, when evaluating effect, a series of pooled RRs with 95% CIs produced similarly before and after eliminating each study at a time, suggesting that our results were robust and conservative. (Table 3) We can see that the largest portion of variance was explained when the study of Minai et al. 9 was removed, I-squared decreased from 40% to 21% (R 2 = 30%), and the value of Tau-squared was 0.00, which indicated that the goodness of fit of the model was good. We conducted meta regression analysis in order to explore the source of heterogeneity in these respects of publication year, country, route of drug, disease of patients, study samplesize, and the Jadad score. First, we tested the influence of only one single attribute to the model (Table 1), but found no parameters was statistically significant. Second, we discussed the factor interactions and brought these covariates into models. None of these factors could have related to estimations of effect indeed.
On the other hand, when evaluating safety, we could make comparisons with side-effects between the two opioids in different drug route. In intravenous and intramuscular administration, we found that Pooled risk ratios (RRs) for the incidence of adverse effects of nalbuphine versus morphine were 0.12(95%CI: 0.02-0.97) for Pruritus, 0.83(95%CI: 0.64-1.09) for Nausea, 0.72(95%CI: 0.55-0.94) for Vomiting. In intrathecal and epidural administration, we can see that pooled risk ratios (RRs) for the incidence of adverse effects of nalbuphine versus morphine were 0. 22 Table 2. Pooled risk ratios (RRs) and 95% confidence intervals (CIs) for analgesic effects and safety comparing nalbuphine to morphine. * P values (two-sided) were based on the Q test of heterogeneity. ﹠ When there was heterogeneity, the random effects model was used. # stratified analysis was conducted, in which the routes by intravenously and intramuscularly were combined into one group, the routes by intrathecally and by epidural were combined into the other group.  Fields]) AND (Randomized Controlled Trial[ptyp] AND "humans"[MeSH Terms]). We got 176 hits in the Cochrane Library, and 177 hits for PubMed. We also reviewed citations listed in retrieved articles to identify additional studies.

Study Selection.
RCTs were eligible for analysis if the following criteria was met: (1) all patients should be randomly divided into nalbuphine and morphine groups; (2) Studies should evaluate efficacy and safety of nalbuphine compared with morphine; RCTs were eligible if they included at least one group receiving nalbuphine and one group receiving morphine. (3) Studies should provide the value of odds ratio/relative risk and 95% confidence interval; otherwise, data could be converted into relative risk and 95% confidence interval. Two reviewers independently searched literature with the same retrieval strategy, assessed retrieved titles and abstracts, and downloaded potentially relevant articles for further assessment.
Data Extraction. Both investigators independently extracted the following information from the original articles: publication year, patient population details, patient type, interventions, number of cases and controls, number of patients that needed additional analgesic or patients with inadequate analgesia, number of patients with pain relief, the relative risk and 95% confidence interval, and the incidence of adverse effects. Disagreements on data extraction were resolved through discussion. While, our evaluation standard was as follows: (1) pain could be measured by the 100-mm visual analog scale (VAS), the 0-10 verbal rating score (VRS), the verbal category scale, or a four point score (0 = pain free; 1 = mild pain; 2 = moderate pain; 3 = severe pain), but it was patients whether needed additional analgesic during the early/late observation time after interventions that should be the basis of curative effect in our study evaluation; (2) respiratory depression was defined as a respiratory rate of < 8 breaths/min or an arterial PaCO 2 > 50 mmHg at any time postoperatively.
Quality assessment. Study quality was judged by the Jadad scale score (5 points) according to the criteria proposed by Jadad and colleagues 21 , which evaluates studies based on randomization, blinding and dropouts. A study with a Jadad score between 3 and 5 was considered a high quality study 22 .
Statistical Analysis. Data were extracted and summarized using relative risks with 95% confidence intervals (CIs) by the Review Manager 5.2. If the 95% CI included a value of 1 or P > 0.05, it was assumed that there was no statistically significant difference between nalbuphine and control 11 . We assessed heterogeneity of the study during this meta-analysis with chi-squared test and by calculating the value of I-squared, and P < 0.1 was considered statistically significant. Generally, if I 2 > 56%, it prompts a significant heterogeneity, and trials were pooled using random effects model; if I 2 < 31%, it indicates an insignificant heterogeneity, and trials were pooled using fixed effects model. Potential publication bias was assessed by Begg's test 23 and Egg's test 24 . Sensitivity analyse and subgroup analysis could be conducted. Meta regression analysis by the stata statistical software version 10.0 was used to analyze sources of heterogeneity. Tau-squared expresses remel estimate of between-study variance, and the smaller the value is, the better the goodness of fit of the model becomes. R-squared represents how much the covariate currently into the model can explain the amount of variation between the research. All P values reported are two-sided.

Discussion
Overall, 820 patients were included in the meta-analysis. This meta-analysis of randomized controlled trials provides the solid evidence to date regarding the efficacy and safety comparing nalbuphine with morphine. We discovered nalbuphine was comparable to morphine regarding analgesic efficacy. As we know, the evaluation of incidence of pain relief we extracted was not the most direct evidence, which was weakly expressed in clinical evidence for clinical effects, and thus, could not do better than the direct evidence of pain scores for evaluation. Even so, the outcomes we provided were worth considering. In addition, study quality of included studies had been considered, and in general, none of the ten studies 7-16 , which were made use of analyzing pain relief, was of low quality. There were four RCTs that got 3 scores, four RCTs that got 4 scores, and two RCTs that got 5 scores. In addition to this, allocation concealment had been evaluated, 5 of 10 studies reported allocation concealment [7][8][9][10]15 , which thus could present a more comprehensive evaluation of the possible bias in a randomized controlled study. All studies eligible for analysis used a randomized controlled design, which improved the reliability of the evidence. The study by Etches et al. 20 had a small sample size (morphine 5 mg, nalbuphine 10 mg, nalbuphine 20 mg: 6,4,5). In this study, 4 patients who received epidural nalbuphine 10 mg and all 5 who received epidural nalbuphine 20 mg got inadequate analgesia, and all 6 patients who received morphine had satisfactory analgesia (morphine vs. nalbuphine 10 mg, not significant; morphine vs. nalbuphine 20 mg, P < 0.01 ) 20 . If the study by Etches et al. 20 was included into evaluation of analgesic effects, the incidence of pain relief in pooled analyses was still no significant difference (RRs,1.00; 95% [CI], 0.89 to 1.12; P = 0.95; I 2 = 51%). The study by Baxter et al. 25 compared the analgesic efficacy and side-effects of epidural nalbuphine with epidural morphine in a randomised double-blind study in post-thoracotomy patients, revealing the pain scores were lowest in the morphine group (P < 0.01), which indicated an advantage in analgesic efficacy for morphine. Figure 1 shows the outcome of analgesic effects comparing nalbuphine with morphine, then we can find that the study by Minai et al. 9 presented a positive result, which was one of factors causing heterogeneity. However, we can not ignore this research, for that its quality was all right.
Except for pain relief, there are many aspects to evaluate the effect of drugs, such as speed of drug action, efficacy of maintaining time, and pain scores. Culebras et al. 15 conducted a study to compare the analgesic efficacy and adverse effects of intrathecal nalbuphine and intrathecal morphine for postoperative pain relief after cesarean deliveries. And it showed postoperative analgesia lasted significantly longer in the morphine group, compared with the nalbuphine groups (P < 0.0001).
As we know, pruritus is the most frequent side effect associated with spinal morphine 26 that limits its use. This adverse effect is often difficult to treat and patients responds poorly to conventional treatments 27 . And The study by Somrat et al. reported that 3 mg of nalbuphine is effective in the treatment of intrathecal morphine-induced pruritus after cesarean delivery 28 . This study demonstrated a 20.6% incidence of morphine-induced pruritus by different kinds of routes. While, another two retrieved studies 15,16 , in which drugs were given by intrathecally, demonstrated a 38% incidence of intrathecal morphine-induced pruritus, and this is consistent with previously reported values [29][30][31][32] . Mixed agonist-antagonist opioid effects of nalbuphine have been reported for prevention of pruritus after epidural morphine [33][34][35] Duration of action of intravenous nalbuphine is shorter than the duration of epidural morphine induced pruritus, and continuous intravenous infusion is needed to treat this side effect 34 .
Side effects such as pruritus, nausea, vomiting and urinary retention, are common 36 , but the most serious problem is respiratory depression 25 . Nalbuphine has a plateau effect on respiratory depression when given on its own 29 . It has been shown to reverse the respiratory depression from both intravenous 30 and epidural 31 opioids. The study by Baxter et al. reported that a 200 ug kg −I bolus followed by a 50 ug kg −I hr −l infusion of nalbuphine may be administered to post-thoracotomy patients receiving epidural morphine, to prevent respiratory depression without causing significant side-effects or cardiovascular stimulation 25 .
When comparing nalbuphine with morphine, the pooled RRs were 0.17 for pruritus, and 0.27 for respiratory depression. The values of RRs were between 0.1 and 0.3, which indicated a strong correlation. Nalbuphine had a great advantage over morphine regarding these two side-effects of pruritus and respiratory depression. Our analysis also found other clinical advantages of nalbuphine, such as less nausea and vomiting. Sedation had been reported with postoperative analgesia with nalbuphine 31,32,37 , however, it was comparable to that produced by epidural morphine in the study by Baxter et al. 17 There were only two retrieved studies 12,19 that involved sedation, and the incidence of sedation were 0.114 with nalbuphine and 0.228 with morphine (Table 3).
Therefore, nalbuphine, which has a similar analgesia effect with morphine but has an advantage over morphine in some way, is another option for pain control.
In conclusion, our current meta-analysis indicates the analgesic efficacy of nalbuphine is comparable to morphine, but nalbuphine provides a better safety profile than morphine in the aspect of certain side-effects, especially related to pruritus and respiratory depression.