Introduction:

Phenomena prediction through prognostic scales is a valuable tool in medical sciences nowadays and it should be included in the decision-making process. Predicting postoperative mediastinitis allows to count on resources for its prevention.

Objective:

To build a prognostic scale to stratify the risk of suffering from postoperative mediastinitis.

Method:

A case-control study for the risk factors of postoperative mediastinitis was carried out at the Cardiocentro Ernesto Guevara from Santa Clara, Cuba. After the logistic regression, the model was obtained and from it, the predictors to obtain the Cuban prognostic scale of postoperative mediastinitis PREDICMED were included and weighted, which was validated through several methods.

Results:

This scale was obtained, counting on six predictors and two risk strata. Its performance was analyzed through adjustment, calibration and determination of its discriminating capacity, showing good results. Internal validation was carried out through the data division method and its capacity was compared in both subsets (development and validation) showing no differences. Its good construct validity was demonstrated, since there were no differences between the predicted and the observed probabilities. Its contents validity was also analyzed by experts. Finally, its criteria validity was determined when compared with another similar scale (Medscore). PREDICMED showed a very good discriminatory capacity (area under the curve 0.962) as well as high values of sensitivity (84.62%) and specificity (92.31%).

Conclusions:

The Cuban prognostic scale PREDICMED, to stratify the risk of postoperative mediastinitis showed good validation parameters and it was able to stratify the risk in not high and high.

        · |     · |     · ( pdf ) | ( pdf )