Chinese Journal of Tissue Engineering Research ›› 2011, Vol. 15 ›› Issue (9): 1659-1662.doi: 10.3969/j.issn.1673-8225.2011.09.034

Previous Articles     Next Articles

Estimation of effect-size after dichotomized continuous variables in Meta-analysis

Zhang Hong, Wang Zhen   

  1. School of Medicine, Huzhou Normal University, Huzhou  313000, Zhejiang Province, China
  • Received:2010-08-27 Revised:2010-10-19 Online:2011-02-26 Published:2011-02-26
  • Contact: Wang Zhen, Lecturer, School of Medicine, Huzhou Normal University, Huzhou 313000, Zhejiang Province, China wangzhen@hutc.zj. cn
  • About author:Zhang Hong, Professor, School of Medicine, Huzhou Normal University, Huzhou 313000, Zhejiang Province, China zhanghong@hutc.zj. cn

PDF

689

Abstract:

BACKGROUND: Both in the field of medical research and behavioral science or psychology in research, even if the type of design is a comparative study of the two samples; for the same outcome variable, there are dichotomized variables and continuous variables, especially for the continuous variables in the dichotomized treatment analysis. Therefore, it also faces enormous challenges in the meta-analysis.
OBJECTIVE: To explore 7 different effect-size indices for estimating the population standardized mean difference (δ) based on a 2 × 2 contingency table in terms of bias and sampling variance.
METHODS: The literatures related to the outcome variable of intragroup comparative study for the continuous, dichotomous, and continuous combined with dichotomous meta-analysis or systematic review methodology in CNKI database, VIP full text database, Wanfang Chinese Doctoral database (1990/2009), Pubmed database (1979/2009) were retrieved, with key words of “group comparison study, continuous outcomes, dichotomous outcomes, combining continuous and dichotomous outcomes, meta-analysis, effect size” in Chinese, and the same key words in English. Non-comparative studies and relative meta-analysis and systematic evaluation literatures were excluded.
RESULTS AND CONCLUSION: A total of 23 documents were included, and which focus on the problems of outcome variables are the continuous variable’s effect-size selection and 7 effect-size selections of dichotomized variables in comparative studies of two groups. In the field of sociology and ethology, essentially as a continuous outcome variables in order to explain and easy to use in the analysis, the dichotomized treatment of continuous variables was widely applied, but it brought challenge for systemic evaluation and meta-analysis. For the dichotomized treatment of continuous variables, 7 kinds of commonly used effect-size was performed comparative analysis. The results showed good performance for 2 indices, one based on the probit transformation and the other based on the logistic distribution.

CLC Number: