The maximum quality score was 6 point [40, 41]. The quality scores were showed in additional file 1. Statistical Analysis 3-MA in vitro The primary end points variables were defined as dichotomous data (e.g., remission rate of pain used variables as follows: the effective or the AZD1152 ineffective after treatment). We standardized the therapeutic results by obtaining the relative risk (RR). RR is defined as a ratio of risk of uncontrolled pain or adverse effects occurring in transdermal

fentanyl group versus sustained-release oral morphine group. To test for heterogeneity among the trials, Cochran’s χ2 test was used. P-value of more than 0.05 for the χ2-test indicated a lack of heterogeneity across the studies, so pooled estimation of the RRs of each study was calculated by the fixed effects model. Otherwise, the random effects model was used. An estimate of the potential publication bias was carried out by funnel plot, in PS-341 purchase which the

standard error (SE) of log RR of each study was plotted against its log RR. An asymmetric plot suggested a possible publication bias. All analyses were performed strictly with RevMan software (version 4.2.8, Cochrane). P value less than 0.05 was considered as significant in difference. Results Characteristics of selected trials 578 trials were examined in the preliminary review; 32 of them were considered eligible and included in the analysis. The data extracted from 32 trials were shown in additional file 1[8–39]. A total of 2651 cancer pain patients were treated in all selected trials, 1296 with transdermal fentanyl, and 1355 with sustained-release oral morphine. 30 of selected trials were included in the analysis of clinical efficacy; and 31, 31 and 28 of selected trials were included in the analysis

of constipation, nausea/vomiting and vertigo/somnolence. Only 6 trials supplied data about QOL evaluated in Baf-A1 cell line different criteria [9, 14, 17, 32–34]. Sustained-release oral morphine was Morphine Hydrochloride-Southwest Pharm in 8 of selected trials [8, 16, 19, 25, 27, 29, 32, 33]. Trials were excluded from the analysis for one or more of the following reasons: uncorrelated, review, case report, no valid data, no followed-up time, and non-cancer pain. Trials applied either numerical rating scale or visual analogue scales for assessing cancer pain. The criterion of remission of cancer pain was described as follow. Five categories of pain relief: category 0, no remission (pain didn’t release); category 1, mild remission (pain released one quarter); category 2, moderate remission (pain released a half); category 3, obvious remission (pain released three quarters); category 4, complete remission (pain disappeared). Pain can be controlled denotes that patients gain category 2 or above of pain relief.