Methodological issues in case-control studies IV: Validity and efficiency of various analysis strategies for continuous variables using the unconditional logistic regression model.

Computer simulation has been used to evaluate the performance of the unconditional logistic regression model when used to analyse continuous data from case-control studies. The size of the bias in the odds ratio estimate introduced by small sample sizes and by the use of the incorrect analysis model has been estimated for various underlying population conditions, together with the power as a measure of efficiency. It is concluded that the model is robust over a wide range of exposures to the risk factor, in the sense that both small sample sizes and use of the incorrect model introduced relatively small biases. The power of the test is again little altered by use of the incorrect analysis model.