EVALUATION OF ANTISYPHILITIC THERAPY

M EDICAL research has long been faced with the problem of patients lapsing from observation during the evaluation of treatment. To date methods of analysis of results of therapy have assumed that the patients who lapse from observation would have had the same experience as those who remained under observation. To the extent that this assumption is not correct the calculation of the results of therapy is in error. It has been claimed that the "observed" is weighted in favor of failures as relapses would return for more treatment while those who are getting along all right would not bother to return for posttreatment observation. On the other hand it has been stated that the failures go elsewhere for retreatment instead of to the source of the original treatment and thus bias a study in favor of "good" results. In an effort to test the hypothesis underlying current evaluation of treatment (viz., the "not observed" would have had the same experience as the observed) a special study in the evaluation of therapy for syphilis, the Blue Star Research Study, was initiated by the Division of Venereal Disease, U. S. Public Health Service, in which an attempt was made to hold a group of patients to 100 per cent follow-up and compare the results of therapy with those obtained when no intensive follow-up effort was made [1]. These patients included 560 persons with secondary syphilis confirmed by darkfield examination who had had no previous antisyphilitic therapy of any kind. The follow-up effort was over 90 per cent effective over a period of two years, largely due to the work of the physicians in the cooperating treatment facilities and specially trained research investigators whose sole functions * Presented before the session on Statistical Evaluation of Clinical Data. American Statistical