Coronary calcification and depression: a logistic regression model.

Abstract

This research has two parts. In the first one we study if subjects that present simultane-ously trait depression and state depression have a higher coronary calcium score than the subjects with state depression only. In the second part we construct a multivariable logistic regression model and propose a predictor of the presence of severe coronary calcification, based in the model. In the first study participated subjects between 40 and 60 years old, without history of coronary disease, but that underwent a cardiac angiotomography by indication of their physicians. Depression was assessed by the Spielberg´s Inventary of Trait/ State Depression. We found positive associa-tions between each one of the scores in the scales and sub-scales of the test and the calcium scores: state eutimia 0,345, state distimia 0,322, state depression 0,401, trait eutimia 0,376, trait distimia 0,325, trait depression 0,407. In the second part, using a multivariable logistic regression model we estimated the probability of a coronary calcification with score greater than 400 UH, as a function of a set of prediction variables (state depression, trait depression, arterial hypertension, dislipemia and age). The original model was progressively simplified, until a final model was obtained with a traditional cardiovascular risk factor (dislipemia), a psychological risk factor (trait depression) and an epidemiological risk factor (age). Once the parameters of the model were obtained, we explored the possibility of constructing a predictor. When the weights assigned to the sensibility and to the specificity, were taken equal to each other, the optimal value of the outage probability was comprised between 0.3 and 0.4. The area under the ROC curve was significant to level p<0.05 with value 0.74. The construction of models like the one that motivates the present research, with a psychological component not taken into account in previous models, could contribute to the task of identifying new relevant interactions from an epidemiological point of view.