Clinical importance of directional statistics for electrocardiographic differentiation of smoking habits.

Statistical methods for the analysis of directional data are of relatively recent origin and have not been generally applied in clinical investigations where they are appropriate. This study compares spherical means and standard deviations for directional polarcardiographic variables with the linear means and standard deviations of the same variables for a typical industrial cohort of male white collar workers. Normal subjects were classified by smoking status and by independent clinical evidence of the presence of CHD, and both directional and linear statistics are applied to compare the PCG variables in these groups. Directional statistics more accurately define the spherical means and variances of heart vectors and thereby permit more reliable differentiation of directions, or angles to aid diagnosis.