| ObjectiveTo evaluate diagnosis value of CT colonography(CTC)for colorectal cancer(CRC)and to provide high quality evidence-based medical evidence for choosing diagnostic method of CRC.MethodsArticles reporting the diagnosis value of CTC for patients at a high risk of CRC in CNK1,WANGFANG,PubMed,Excerpt Medica Database(EMBase),Cochrane Library databases were searched systematically from January 1,2008 to December 31,2017.After screening literatures strictly according to inclusion and exclusion criteria,extracting basic information and data,QUADAS-2 of RevMan5.3 software was applied to evaluate the quality of included studies.MetaDiSc 1.4 was used to conducted meta?analysis.Chi-square test was used to conduct heterogeneity test and the value of l2 was used to judge which model to use to combine studies.If I2<50%,fixed effect model was used,otherwise,randomized effect model was used.Based on the model above,combined sensitivity,specificity,negative likelihood ratio and positive likelihood ratio were calculated,and curve of summary receiver operating characteristic was drawn.Stata 14.0 was used to draw Deek funnel plot.Results?A total of 3533 articles were detected initially according to retrieval strategy,including 3213 English literatures and 320 Chinese literatures.After screening,a total of 17 articles,including 12 English literatures and 5 Chinese literatures,were analyzed in meta-analysis.The quality of included studies were high generally.?Meta-analysis results of all the included subjects:11 articles reported diagnostic accuracy of CTC for CRC among all the subjuets,including 4290 patients(2177 males and 2113 females).The threshold analysis showed there was no threshold effect((t=-0.009,P=0.979)and I2 of the heterogeneity test result was 79.3%which indicated there was heterogeneity.The combined sensitivity(Sen)was 0.80(0.78?0.82),the combined specificity(Spe)was 0.88(0.87?0.90),the combined positive likelihood ratio(PLR)was 5.73(4.00?8.21),the combined negative likelihood ratio(NLR)was 0.19(0.13?0.29),the combined diagnostic odds ratio(DOR)was 29,55(16.09?54.25)and the combined area under summary ROC curve(AUC)was 0.92(0.83?0.96).? Meta-analysis results of subjects with polyps>6mm:11 articles reported diagnostic accuracy of CTC for CRC among subjects with polyps>6mm,including 6387 patients(3365 males and 3022 females).The threshold analysis showed there was no threshold effect(t=0.400,P=0.223)and I2 of the heterogeneity test result was 80.1%which indicated there was heterogeneity.The combined Sen was 0.87(0.85?0.89),the combined Spe was 0.89(0.88?0.90).the combined PLR was 8.06(5.90?11.02),the combined NLR was 0.15(0.12?0.19).the combined DOR was 56.60(34.65?92.46)and the combined AUC was 0.94(0.92?0.96).? Meta-analysis results of subjects with polyps ? 10mm:10 articles reported diagnostic accuracy of CTC for CRC among subjects with polyps?10mm,including 6087 patients(3165 males and 2922 females).The threshold analysis showed there was no threshold effect(t=0.115,P=0.751)and I2 of the heterogeneity test result was 88.3%which indicated there was heterogeneity.The combined Sen was 0.89(0.87?0.92),the combined Spe was 0.90(0.89?0.91),the combined PLR was 17.26(9.63?30.93),the combined NLR was 0.12(0.08?0.19).the combined DOR was 198.05(7.18?551.04)and the combined AUC was 0.97(0.94-1.00).?Publication bias analysis was conducted among studies including all the subjects,subjects with polyps>6mm and subjects with polyps ? 10mm.P values of the Deek funnel plots were 0.16,0.87 and 0.63,respectively,which indicated that there was no publication bias.ConclusionsThe diagnosis values of CTC among all the subjects,subjects with polyps>6mm and subjects with polyps ? 10mm were very good.With the increasing of polyps size,diagnostic accuracy of CTC for CRC also increased and diagnosis value was the highest among patients with polyps ? 10mm.However,larger sample sizes and higher quality diagnosis trials among different people are needed to provide high quality evidenced-based medical evidence for clinical doctors. |
PDF Full Text Request