Improved meta analysis using predicted relative risk

Khan, Shahjahan and Hasan, Md Masud (2008) Improved meta analysis using predicted relative risk. In: 2008 International Conference on Recent Development in Statistical Sciences, 26-27 Dec 2008, Dhaka, Bangladesh.

[img]
Preview
PDF
Khan_Hassan.pdf

Download (153Kb)

Abstract

[Abstract]: This paper proposes a new method of improved meta analysis to combine relative risk for both homogeneous and heterogeneous set of studies. The standard meta analyses don't give any conclusive result when the effects of heterogenous studies are combined. The proposed improved meta analysis uses the predicted relative risk, and chi-square test to check the heterogeneity of the effects. Confidence intervals for the relative risks obtained via improved method concentrate more towards the value of the pooled estimate than that of the standard meta analysis. Exclusion of identified studies with outliers from the analysis brings the results of the remaining studies closer to the pooled estimate. An illustration shows that the new method improves the results and provide conclusive estimate of the relative risk.


Statistics for USQ ePrint 4831
Statistics for this ePrint Item
Item Type: Conference or Workshop Item (Commonwealth Reporting Category E) (Paper)
Refereed: Yes
Item Status: Live Archive
Additional Information: No evidence of copyright restrictions on web site.
Depositing User: Professor Shahjahan Khan
Faculty / Department / School: Historic - Faculty of Sciences - Department of Maths and Computing
Date Deposited: 04 Feb 2009 06:14
Last Modified: 02 Jul 2013 23:12
Uncontrolled Keywords: relative risk, predicted relative risk, odds ratio, chi-square test, standard meta analysis
Fields of Research (FOR2008): 01 Mathematical Sciences > 0104 Statistics > 010401 Applied Statistics
11 Medical and Health Sciences > 1117 Public Health and Health Services > 111706 Epidemiology
URI: http://eprints.usq.edu.au/id/eprint/4831

Actions (login required)

View Item Archive Repository Staff Only