A multivariate model to predict the distance from the skin to the epidural space in an obstetric population

Several attempts to relate epidural space depth to individual patient parameters or details of technique have yielded modest correlations. An attempt has been made to construct a multivariate model to predict the depth from the skin of the epidural space with use of several such factors.

The depth of the epidural space from the skin was measured in 263 obstetric patients by using calibrated needles. In 53 patients, the angle of insertion of the epidural needle was also measured, with protractors. For each patient, 14 descriptors of patient habitus and anesthetic technique were recorded. Multiple regression analysis was used to construct models incorporating combinations of these variables to predict the epidural space depth; the model balancing the highest correlation coefficient and greatest statistical significance was selected. The validity of the model was tested on the 53 patients in whom the angle of the needle had been measured but whose data have not been used to construct the model.

A model containing the seven variables present-weight, height, body mass index, age, parity, interspace, and patient position-yielded r = .689, < .0001. This model predicted the depth of the epidural space to within 8 mm and predicted the depth in the validation group well. Correction for the angle of insertion of the needle did not improve the predictive power of the model.

While this model is the best predictor of epidural space depth yet published, it is probably not sufficiently accurate to be clinically useful in confirming proper epidural catheter placement. Further work in this area is probably not justified.