Because mean values during sustained activity may not adequately reflect periods of desaturation, two additional determinations were made. The lowest saturation recorded during pre-eating and eating phases was recorded for each subject, and mean values once again obtained for each time period (mean lowest SpO2). Finally, in every subject, the percentage of time with saturation less than 90% during control and meal periods was taken. As with other measurements, mean data were used for comparisons (percentage of time with SpO2 less than 90%). This threshold was chosen because it was felt that values below that level could be significant clinically.
Statistical analysis: Data analysis was performed using the StatView statistical package (Abacus Concepts, California). ANOVA was used to evaluate the changes in oxygen saturation and dyspnea before and during the meal. The grouping factors employed were method of eating and severity of obstruction. Parametric and nonparametric tests were applied to assess the differences in continuous variables only when the ANOVA results were significant. P