Given subject variability, our sample size was sufficient to detect a minimum difference of 16 percent between the groups at each time period with beta =.20 and alpha =.05 for the Bonferroni multiple comparisons. Although a repeated measures A NOVA could have been performed with missing data, the analysis would not be as powerful and could possibly have led to invalid conclusions. Of the three patients who were readmitted to the study, one admission for each patient was used in the repeated measures ANO\A because the other three admissions had missing data points during the hospital course (eg, too ill to perform pre- and post-bronchodilator spirometry at every time point early in the course).
All six admissions of these three patients were included in the admission and discharge analyses. The problem of missed spirometric measurements has been previously encountered by other investigators studying severely ill asthmatics, that is, several of our patients were unable to exert the effort necessary to perform all scheduled spirometric tests. canadian health&care mall
Both groups improved markedly over the course of hospitalization with no significant differences detected either in spirometry, duration of hospital stay, or subjective patient response.