We conducted this combined analysis of available data from studies with information on this issue to clarify the association between Trichomonas vaginalis infection and cervical neoplasia.We performed MEDLINE searches (1966-1993) using the key words and phrases 'trichomonas vaginitis' and 'neoplasms, cervix' for articles published in English, and searched citations of the articles obtained from MEDLINE: A total of 24 articles (two cohort studies and 22 case-control) were included in this data analysis. In the analysis, the studies were evaluated for heterogeneity using Breslow-Day tests for homogeneity of the odds ratios and of rate ratios. If the odds ratios from studies are heterogeneous, it is not appropriate to combine them using the Mantel-Haenszel method. Also, publication bias was evaluated by assessing the association between the observed effect size and the variance of the effect size using a rank correlation test.The combined summary relative risk for the two cohort studies was 1.93 (95

作者：Z F, Zhang；C B, Begg

来源：International journal of epidemiology 1994 年 23卷 4期