内镜窄带成像技术诊断结直肠腺瘤的系统性评价

目的系统性评价内镜窄带成像技术（narrow band imaging，NBI）诊断结直肠腺瘤的有效性。方法检索美国医学索引数据库（Medline，1966年1月至2008年10月）、OVID数据库（1996年1月至2008年10月）、荷兰医学文摘数据库（EMBASE，1980年1月至2008年10月）、考科蓝实证医学资料库（Cochrane Library，2008年第3期）以及中国生物医学文献数据库（CBM，1997年1月至2008年10月），并按Jadad质量评分评定随机对照研究（RCT）的质量。用Rev Man4．2软件进行荟萃分析。结果7项随机临床试验中共2838例患者满足纳入标准，窄带内镜与常规内镜相比腺瘤检出率差异无统计学意义（OR=1．18，95％CI=1．00～1．39，P=0．06）。NBI可显著提高扁平腺瘤检出总数加权均数差（WMD）0．14，95％CI=0．02～0．26，P=0．02，但NBI系统退镜时间明显延长（合并WMD=1．05，95％CI=0．08～1．22，P〈0．01）。结论NBI可提高结直肠平坦型病变的检出率，并不能提高腺瘤的检出率，而且窄带内镜耗时长。NBI技术需进一步完善后，方能真正用于临床诊断。

NBI system versus conventional endoscopy in diagnosis of colorectal adenomas: a systematic review

Objective To evaluate the validity of narrow-band imaging (NBI) system in detection of colorectal adenoma, as compared to that of endoscopy of the colon and rectum, in a systematic review. Methods Relevant literatures were retrieved from Medline (January 1966 to October 2008), OVID (January 1996 to October 2008), EMBASE (January 1980 to October 2008), Coehrane Library (Issue 3, 2008) and Chinese Biological Medicine Disk (CBM disk, January 1997 to October 2008). Quality of the literatures retrieved was assessed based on the Cochrane Reviewers' Handbook and Jadad's score. RevMan version 4. 2 software was used for meta-analysis. Results Seven randomized clinical trials (2838 patients) were included in the study. Compared with white-light colonoscopy, no significant difference was observed in terms of adenoma detection rate (OR 1.18, 95% CI 1.00-1.39, P=0.06) by NBI system, which could significantly improve total number of detection for fiat lesions of the colon and rectum (pooled WMD 0.14, 95% CI 0.02-0.26, P=0.02), but with a longer withdrawn time (pooled WMD 1.05, 95% CI 0.08-1.22, P<0.01). Conclusions Detection rate for flat lesions of the colon and rectum, not for adenoma, can be improved by NBI system and meanwhile its withdrawn time is prolonged, indicating that routine use of NBI system for detecting colorectal adenomas may be recommended only with its further refined technique.