Harris et al1 did not evaluate or comment on what appears to be the strongest statistical association in their study: 3.7% (18/484) of those in the prayer group were discharged within 24 hours compared with only 0.9% (5/529) of those in the usual care group (P<.005 by χ2 test if observations are independent). Since these discharges occurred before the intervention began (mean±SE,1.6 ± 0.16 days after admission), we are concerned that the statistical methods used by Harris et al,1 which assume independence of the observations, may not be appropriate for their data. While their article states that "new patients were randomly assigned," it is not clear whether the same person who was readmitted for a new episode would have constituted a new patient; Figure 1 of their article does not indicate that readmissions of the same patient were excluded.1 Since patient assignment was based on an (odd or even) identification number that never changed, readmitted patients would remain in the same treatment group. For example, a patient with an even identification number who was admitted several times during the study period and tended to stay in the coronary care unit less than 24 hours (or to have a low Mid America Heart Institute–Cardiac Care Unit [MAHI-CCU] score) would always be in the prayer group. Statistical methods that require independent observations would not treat the data from this individual correctly.2 Thus, we believe that the authors should comment on the following: Were multiple admissions from the same individuals included in their analysis? What were the potential reasons for the significantly higher (preintervention) 24-hour discharge rate of patients in the prayer group?
Donald R. Hoover, Joseph B. Margolick. Questions on the Design and Findings of a Randomized, Controlled Trial of the Effects of Remote, Intercessory Prayer on Outcomes in Patients Admitted to the Coronary Care Unit. Arch Intern Med. 2000;160(12):1875–1876. doi: