TIME DOMAIN ANALYSIS OF SPIROGRAMS

Abstract

Spirograms are widely used simple tests of ventilatory lung function. This thesis evaluates their analysis in the time domain as a means of improving their yield of diagnostic information. The most precise method for recording spirograms was determined. Errors due to non-instantaneous cooling of gas within spirometers were found to be greater than the known non-linearity of pneumotachographs when recording simulated spirograms.

The analysis of spirograms inthe time domain involves considering the spirogram as a cumulative distribution of transit times and deriving the statistical moments of this distribution. A valid procedure for comparing the moments of truncated spirograms has been proposed and these truncated moments have been shown to be highly reproducible within individuals and retain discriminatory ability. A multi-exponential model of the spirogram has been considered in detail as a means of overcoming errors inherent in the moments of some spirograms. The most satisfactory technique for applying this model to describe spirograms has been determined and the model was found to reflect correctly abnormalities in spirograms for commonly found ventilatory disorders. A cross-sectional population survey of a male workforce was carried out to determine whether time domain analysis of spirograms detects abnormalities hitherto unappreciated. Normal values for conventional and time domain spirometric indices were determined. Smokers were found to have two types of spirogram abnormality previously not recognised. The more commonly found abnormality was evident early within the spirogram, and the other was only evident in the tail of the spirogram. These findings do not agree with current hypotheses that smoking damage should be first manifest in the terminal part of a maximal forced expiratory manoeuvre. Alternative hypotheses are proposed. Time domain analysis of spirograms has been found to demonstrate hitherto unrecognised changes in spirograms due to smoking.