Estimating Probabilities and Odds Ratios in Gestational Outcomes: A Dummy Variable Regression Illustration

Dummy variable regression model assumes a linear relationship between the categorical variables and outcome variable. Also its coefficients of regression might not be directly interpretable in terms of probability changes or odds ratios, which may potentially limit the usefulness of the model. This might not hold true if the gestation length is dichotomous. This study explores an alternative method of using dummy variable regression in estimating probabilities, odds and odds ratios in gestational outcomes if the outcome variable is continuous rather than the use of logistic regression. This involves first partitioning each of the parent independent variables into a set of mutually exclusive categories or subgroups and then use dummy variables to represent these categories in a regression model. In such a regression model, each parent independent variable is represented by one dummy variable of 1’s and 0’s less than the number of its categories. Any level of a parent independent variable that is not specifically represented by a dummy variable is referred to as the excluded level of that parent variable while the others are termed the included levels in the regression model. This study is limited as follows: there is need to ensure that the model assumptions of linearity and independence of observations, continuity of the outcome variable as well as categorical nature of the predictor variables are met. A pilot cross sectional study design was carried out at Alex Ekwueme Federal University Teaching Hospital Abakaliki where data on age, parity and sex of last births were collected from 41 anti-natal women. The overall results of analysis showed that R2 = 0.066; 95% CL = 1.234-2.765, p - value = 0.780 indicating an insignificant relationship between the outcome variable and the categorical predictor variables, signaling the end of analysis. For illustration purposes only, we estimated probabilities and used it to estimate any desired odds and odds ratios like the odds that the randomly selected mother has a male and female births with gestation length of more than 39.5 weeks gave 0.636 and 0.639 as odds for male and female respectively while odds ratio was 0.995 implying that for every 1000 female births with a gestation length of more than 39.5 weeks, there are 995 males’ births with the same gestation length of 39.5 weeks. We concluded that dummy variable regression enables one to estimate probabilities and odds ratios of continuous outcome variable and so compares favorably with logistic regression.