Notes on the use of OpenEpi for lesson 14:
It is suggested that you input all the two-by-two tables from lesson 14 into OpenEpi and look for the statistics presented in the lesson. Some of the tables in this lesson are the same as in lesson 12.
Please note the following before you input the tables:
- On p. 414 of the Companion Text (14.1 expo 1) the p-value for smokers should be 0.247 rather than 0.49.
- On p.415 (14.1 expo 2) you will notice that the aRR =1.25. You will not be able to find this value (or its confidence interval) from the OpenEpi output. OpenEpi does not provide a precision based adjusted estimate (e.g., aRR) as part of its output. The Directly Adjusted estimate provided by OpenEpi is very similar, but will often yield slightly different results.
- On page 429 (14.2, expo 2), the p-values for the non-smokers and smokers should be 0.0182 rather than 0.01.
- On page 453 (14.6, expo1), be careful about inputting data for the test for trend. The order in OpenEpi is different from ActivEpi. To input this data, select Dose Response / Trend from OpenEpi. In this example, you will be provided with the mRR from OpenEpi rather than the aRR and the Mantel-Haenszel chi-square presented in OpenEpi uses a correction factor not used in ActivEpi, so the results will be similar, but will differ somewhat.
- You will not find the precision based adjusted estimates (e.g., aRR, aOR) from the OpenEpi printout. OpenEpi does not provide this measure. You will find a different adjusted measure of effect e.g., Directly Adjusted RR, Directly Adjusted OR, etc. Although the methods are similar, the results will not be identical to the aRR or aOR as presented in ActivEPi. OpenEpi does provide the Mantel- Haenszel estimate which you will recall is the preferred summary measure among the weighted measures.
- Please note that the Mantel-Haenszel chi-square presented in OpenEpi uses a correction factor not used in ActivEpi, so the results will be similar, but will differ somewhat.
After inputing the tables from Lesson 14 into OpenEpi, please make sure that you can locate and interpret the Mantel-Haenszel chi-square and it’s associated p-value, the Mantel-Haenszel adjusted estimate and its corresponding confidence interval. You really don’t need to worry about calculating these by hand, but it is very important that you can interpret the results from such a printout.
You will also notice that ActivEpi emphasizes different statistics than OpenEpi. If you looked at other software, you would likely see other statistics reported. Please notice that all of these results are very similar, however, there is considerable debate in the literature about which is the best to use. In this course, I will have you focus on the statistics that David Kleinbaum recommends. I hope that going through this lesson and looking at the formulas and how many of them are derived, you will get a better sense of why some of these formulas are preferred.