The confidence limits for all parameters derived from a 2×2 contingency table are uniquely determined by the limiting matrices that arise from adding or subtracting certain values to or from the four cell entries. Various methods to determine these addends are discussed and a computer program is presented that produces reliable results. The limiting matrices thus produced bypass the more complicated expressions for the standard errors of the many epidemiological parameters that are used to interpret the 2×2 table.
Get full access to this article
View all access options for this article.
References
1.
BrownCh. C.: The validity of approximation methods for interval estimation of the odds ratio. Am J Epidemiol113: 474–480, 1981.
2.
CornfieldJ. A.: A statistical problem arising from retrospective studies. Proc. of 3rd Berkeley Symposium, University of California Press4: 135–148, 1956.
3.
FisherR. A.: Confidence limits for a cross-product ratio. Aust J Stat4: 41, 1962.
4.
GartJ. J.ThomasD. G.: The performance of three confidence limit methods for the odds ratio. Am J Epidemiol115: 453–469, 1982.
5.
MantelN.HankeyA. F.: The odds ratio of a 2×2 contingency table. Am Stat29: 143–145, 1976.
6.
ThomasD. G.: Exact confidence limits for the odds ratio in a 2×2 table. J Stat Soc20: 105–108, 1971.
7.
Waltner-ToewsD.: Nomenclature of risk assessment in 2×2 tables. Int J Epidemiol11: 411–412, 1982.
