OpenEpi, Version of December 2003

Test Results for DoseResponse Module

 

 

Dataset #

Left Side

Data

Compared with

Result

DoseResp1

a

Exposure

0,1,2

Stratum 1

1,131

1,104

4,51

Stratum 2

0,188

6,152

15,83

Stratum 3

3,161

12,130

22,65

Stratum 4

11,169

21,134

39,68

Stratum 5

23,157

42,97

34,52

Schlesselman, JJ Case-Control Studies, 1982, 205.  Cigarette smoking among Female Cases of Myocardial Infarction and Controls, stratified in 5 age groups.

Mantel-Haenszel extended test for trend agreed perfectly until continuity correction was added per Rosner (see below)  MH odds ratios at two levels of smokers vs. baseline non-smokers also agree.

Results will now differ slightly because of the correction.

DoseResp1

b

Exposure

0,1,2

1,131,1,104

0,188,6,152

3,161,12,130

11,169,21,134

23,157,42,97

Epi Info, Statcalc, the DOS program, 1993

Mantel-Haenszel extended test for trend agreed perfectly until continuity correction was added per Rosner (see below)  MH odds ratios at two levels of smokers vs. baseline non-smokers also agree. 

DoseResp2

Exposure 1,2,3

Women

Cases, Controls

196,603,

223,486

103,232

Men

Cases, Controls

188,348

313,383

232,206

 

 

Rosner, B. Fundamentals of Biostatistics, Duxbury, 2000.

After adding the continuity correction employed by Rosner, the Mantel extended test (with cc) agrees perfectly. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Testing by Andy Dean, using materials listed. 

Conclusion: Module is ready for use, although users are advised to confirm important results with at least one other software module.