multiple variables measured, congestive heart failure, cardiopulmonary arrest, pneumonia, diuretics, antibiotics, and intubation/ventilation were seen less frequently in the prayer group.   Multivariate analysis of the data using the variables listed in Table 2 revealed a significant difference (P < .0001) between the two groups based on events that occurred after entry into the study. Fewer patients in the prayer group required ventilatory support, antibiotics, or diuretics.
   The hospital course after entry was graded good, intermediate, or bad, based on the following criteria. The course was considered to be good if no new diagnoses, problems, or therapies were recorded for the patient or if events occurred that only minimally increased the patient's morbidity or risk of death. The course was considered intermediate if there were higher levels of morbidity and a moderate risk of death. The course of patients who had the highest morbidity and risk of death or who died during the study was graded as bad. The grades were assigned on the basis of the hospital course alone, and no correlation was made as to the condition of the patient at the time of entry. That is, even a patient whose condition was severely critical at the time of entry received a grade of good if no new problems or diagnoses developed after entry, and if the patient recovered without new therapeutic interventions and was discharged home. In patients who had minor problems on entry but subsequently had severe life-threatening complications and prolonged hospitalization, the hospital course received a grade of bad.
   The scoring used for the three levels is summarized in Table 3. In the prayer group 85% were considered to have a good hospital course after en-

DISCUSSION
   In reviewing the social and scientific literature on the efficacy of prayer to the Judeo-Christian God there seems to be no end to articles discussing it but very few articles that actually test for the effects of prayer. The Bible records examples of the effectiveness of prayer in healing in the book of Genesis 20:17,18; Numbers 12:13; and Acts 28:8.
   Roland¹⁰ believed that a work on the effectiveness of prayer by Galton 2 in 1872 represents one of the first applications of statistics to science and one of the first objective studies of prayer. Galton,³ on reporting the effects of prayer in the clergy, found no salutary effects. He cited previous work by Guy from which he concluded that prayer for sovereigns in England did not make them live longer than other prominent people of the time.² Though perhaps a unique approach for his time, the study suffered greatly in design, as retrospective studies are prone to do. Galton also believed that prayer seemed to be a perfectly reasonable subject for research. But the literature remained silent after this, probably as a result of the furor his comments created at the time.
   In 1965 a double-blind clinical trial of the effect of prayer on rheumatic patients was reported by Joyce and Welldon,⁴ who studied 19 matched pairs of patients over 18 months, with a crossover between the control group and the prayer group at six months. During the first half of the study, the prayer group did better but in the second half the control group did better. Their results showed no significant differences as a result of prayer Subsequently, in 1969 Cohipp5 reported the result of a triple-blind study of the efficacy of prayer on 18 leukemic children. In a randomized trial, his data suggested that prayer had a beneficial effect but it did not reach significance because the number of patients was small and the initial randomization did not produce matching groups, thus nullifying any suggested benefit for the prayer group.
   I approached the study of the efficacy of prayer in the following manner suggested by Galton²:

There are two lines of research, by either of which we may pursue this inquiry. The one that promises the most trust-worthy results is to examine large classes of cases, and to be guided by broad averages; the other, which I will not employ in these pages, is to deal with isolated instances.

828 July 1988 • SOUTHERN MEDICAL JOURNAL • Vol. 81, No. 7