Literature bias, which can reduce the validity of a systematic review, comes mainly from the fact that all the related studies can not be collected without exception. The following three types of literature bias are most common. (1) Publication bias. The reason is that some studies can not get published, or the publication has been delayed. (2) Literature retrieval bias. This bias is the result that not all published studies are retrieved during the course of systematic review. (3) Literature inclusion bias. While an improper literature selection strategy is adopted, some useful studies are kicked out by error, leading to such bias. A funnel plot (Y-axis is rumple size, X-axis effect) is usually useful to assess literature bias in a systematic review.
Objective To study the association between the insertion(I)/deletion(D) polymorphism of the angiotensin-converting enzyme gene intron 16 (ACE/ID) and diabetic nephropathy (DN) among Chinese population by Meta-analysis. Methods Odds ratios of ACE/ID genotype distributions were analyzed in NIDDM or IDDM patients with and without DN. All the related studies on ACE/ID polymorphism and DN were identified while poor-qualified studies were excluded, and the risk of publication bias was estiMetad. The Meta-analysis software, RevMan 3.1, was applied for investigating heterogeneity among individual studies and summarizing effects across studies by proper statistical methods. Result The pooled odds ratios (with 95%CI) of DD vs ID + II and II vs ID + DD were 2.17 (1.74-2.70) and 0.49 (0.36-0.66) in NIDDM (12 studies), 3.92 (2.05-7.47) and 0.19 (0.09-0.43) in IDDM (3 studies) respectively (Plt;0.01). Conclusion In both NIDDM and IDDM, ACE/ID polymorphism is believed to be associated with diabetic nephropathy. The number of DN patients with DD genotype has been increased while that with II genotype decreased.