A statistical model to predict nonsurvival in congenital diaphragmatic hernia.

OBJECTIVE

To develop a predictive model using echocardiographic indices to identify nonsurvivors from survivors in preoperative patients with congenital diaphragmatic hernia (CDH).

STUDY

DESIGN

Eight cases of CDH, with a mortality rate of 38%, underwent echocardiographic study before surgical repair. Left ventricular mass (LVMASS) using the area-length method of Wyatt et al. and fractional shortening (FS) by M-mode measurements were determined.

RESULTS

We identified a nonlinear nonoverlapping distribution that predicted nonsurvivors from survivors, p=0.04. Multiple regression analysis demonstrated the quantity (LVMASS x FS)(1/2) to be correlated with nonsurvival with a coefficient of determination r(2)=0.55. Comparison of the means of the quantity (LVMASS x FS)(1/2) for the two groups suggested two distinct populations, p=0.04.

CONCLUSION

The mathematical quantity (LVMASS x FS)(1/2) calculated from echocardiographic measurements obtained preoperatively in babies with CDH may predict nonsurvival despite maximal intervention.