PMC

Upper panel: Estimated incremental clockwise rotation of the QRS axis based on FEV₁ in univariate regression analysis (see Additional file : Table E1) for mild or severe airway obstruction (FEV₁ 60 or 30% predicted, GLI). Lower panel: Estimated incremental clockwise rotation of the QRS axis based on bivariate regression analysis taking into account both FEV₁ and FRC (see Additional file : Table E1). The circle segments show the estimated effects of lung function on the electrical rightward rotation for four combinations of mild or severe obstruction (FEV₁ 60 or 30% predicted, GLI) with mild or severe hyperinflation (FRC 140 or 200% predicted, ECSC)

Mean values of the orientations of P wave (a), QRS (b) and T wave axes (c) using the Cabrera format are shown for spirometric GOLD grades 1–4 (left panel). GOLD grade 0 axes did not differ significantly from GOLD 1 and were thus omitted in the illustration to prevent an overlay. To show the additional dependence of the axes on FRC, plots of mean values versus mean values of FRC % predicted and the standard error of mean (bidirectional) for each GOLD grade 0–1 is shown (right panel). Post hoc comparisons revealed multiple significant differences of the axis orientation among GOLD grades as indicated by the means and error bars. In particular, significant differences were observed for all axes between GOLD grade 1 and 3 (p < 0.001), GOLD 1 and 4 (p < 0.001; except QRS: p = 0.008), GOLD grade 2 and 3 (p < 0.001), GOLD 2 and 4 (p < 0.001; except QRS: p = 0.015)

Structural equation model (SEM) providing a comprehensive description of the multiple relationships between influencing factors (top) and dependent variables (below). All measured (manifest) variables are indicated by rectangles. A latent variable (indicated by an oval) named “ECG axes” with the indicator variables P wave, QRS and T wave axes could be constructed in order to summarize the axes orientation and their fixed relationship to each other into a single variable. The lines with one arrow describe unidirectional effects, standardized regression coefficients are given; those with two arrows indicate mutual dependences in terms of correlations, correlation coefficients are given. The error terms needed for mathematical reasons for all dependent variables (i.e. all at which a unidirectional arrow ends) have been omitted for the sake of clarity. The numerical values of the respective unstandardized regression coefficients and covariances coefficients as well as measures of statistical significance are given in Additional file : Table E3