We would like to thank Dr Meyers for his careful attention to our article1 and his analysis of our data. We used a χ2 goodness of fit analysis of the data. We compared the two groups using data from the earlier group that did not receive corticosteroids as the "expected" data. Since "no steroids" was the standard of care at the time, we thought that this was a reasonable approach. We now recognize that this statistical method was not the optimal approach.
Dr Meyers uses χ2 of independent proportion with Yates' correction, with which he directly compares the two groups. With this method, the difference in mortality rates remains significant. However, he does not find significant differences between the groups in need for mechanical ventilation, drug reaction, frequency of PCP being the first manifestation of disease, and mortality rates between the corticosteroid-treated group and the