Mantel-Haenszel techniques and logistic regression: always examine one's data first and don't overlook the simpler...

The authors point out that, in addition to logistic regression, there are other, simpler techniques available for making an adjusted estimate of association between an outcome and a risk factor. Also, the consequence of mismodelling with regression, i.e. of missing the real relationship between an outcome and a risk factor, is illustrated with an example. The need to 'examine one's data' prior to performing multivariable techniques is emphasised.