In: Statistics and Probability
EPIDEMIOLOGY
5 c) Suppose that an investigator is concerned with adverse pregnancy outcomes including gestation diabetes,pre- eclampsia and pre-term labor. The study involved 8032 women who provided demographic and clinical data in the study sample,22 (2.7%) women developed pre-eclampsia ,35(4.2%) develop gestational diabetes and 40 (4.8%) develop pre-term labor.Suppose we wish to assess whether there are differences in each of these adverse pregnancy outcomes by race/ethnicity,adjusted for maternal age .Three separate logistic regression analyses were conducted relating each outcome,considered separately,to the three dummy or indicators variables reflecting mothers race and mothers`s age in years.The results are below.
Outcome Pre-eclampsia | Regression coefficient | Chi-square | p-value | Odds ratio (95% CI) |
Intercept | -3.066 | 4.518 | 0.0335 | |
Black race | 2.191 | 12.640 | 0.0004 | 8.948 (2.673,29.949) |
Hispanic race | -0.1053 | 0.0325 | 0.8570 | 0.9 ( 0.286,2.829) |
Other race | 0.0586 | 0.0021 | 0.9046 | 1.060(0.104,3.698) |
Mother`s age | -0.0252 | 0.3574 | 0.5500 | 0.975(0.898,1.059) |
Outcome Pre-eclampsia | Regression coefficient | Chi-square | p-value | Odds ratio (95% CI) |
Intercept | -5.823 | 22.968 | 0.0001 | |
Black race | 1.621 | 6.660 | 0.0099 | 5.056(1.477,17.312) |
Hispanic race | 0.581 | 1.766 | 0.1839 | 1.787 (0.759,4.207) |
Other race | 1.34 | 5.197 | 0.015 | 3.848 (1.299,11.395) |
Mother`s age | 0.071 | 4.314 | 0.0378 | 1.073(1.004,1.147) |
Outcome Pre-eclampsia | Regression coefficient | Chi-square | p-value | Odds ratio (95% CI) |
Intercept | -1.443 | 1.602 | 0.2056 | |
Black race | -0.082 | 0.015 | 0.9039 | 0.921(0.2443,3.483) |
Hispanic race | -1.564 | 9.497 | 0.0021 | 0.209 (0.77,0.566) |
Other race | 0.548 | 1.124 | 0.2890 | 1.730 (0.628,4.767) |
Mother`s age | -0.037 | 1.198 | 0.2737 | 0.963(0.901,1.030) |
Interpret the results of the three separate logistic regression analysis
Model 1:
We comment all the model using the significant p-value
In the first model Back race has significant contribution in the prediction of the response
If we increase 1 unit in the black race then the probability of occurring the response is increased by the 2.19%.
Model 2:
We comment the entire model using the significant p-value
In the first model Back race, other race and mothers age have significant contribution in the prediction of the response
If we increase 1 unit in the black race then the probability of occurring the response is increased by the 1.62%.
If we increase 1 unit in the other race then the probability of occurring the response is increased by the 1.34%.
If we increase 1 unit in the mother age then the probability of occurring the response is increased by the 0.071%.
Model 3:
We comment all the model using the significant p-value
In the first model Back race has significant contribution in the prediction of the response
If we increase 1 unit in the black race then the probability of occurring the response is decreses by the 0.08%.
Thanks