Haematemesis and Melaena

discipline will induce the patient to take the necessary precautions to prevent another abortion from occurring.(3) That each of the "successfully" treated pregnancies proceeded beyond the 28th week by chance.The second possibility seems to us unlikely.The last suggestion has been fully considered above.As to the first possibility, our results would indicate that progesterone is the essential hormonal factor for maintaining pregnancy, and that adequate progesterone therapy should not be dis- counted as a method of preventing the recurrent factor in abortion sequences.The results certainly do not conform with Hamblen's (1939) warning that exogenous progesterone may hinder the intrinsic metabolism of progesterone, nor do they support the contention (Hoffman, 1944) that pro- gesterone therapy should be employed only in cases in which the pregnanediol level has been shown to be sub- normal.SummaryA series of 45 cases of habitual abortion treated prophy- lactically with progesterone pellet implantation (six 25-mg.pellets) yielded successful results in 86% of the patients with a history of two previous abortions, 89% with three previous abortions, and 75% with four or more previous abortions.The series was subdivided into " primary " habitual abortion (in which all previous pregnancies had resulted in abortion) and secondary" habitual abortion (in which one or more previous pregnancies, but not those immediately preceding the treated pregnancy, had proceeded beyond the 28th week).In these groups the results for primary habitual abortion were: 90% success with two previous abortions, 88% success with three previous abortions, and 60% success with four or more previous abortions; and for secondary habitual abortion, 75% success for two consecutive abortions immediately preceding the treated pregnancy and uniform success for the four cases of three or more abortions.If the true incidence of "recurring" abortion is not much less than 0.1% these results indicate that the treatment was of value.Statistical Methods Used.-The probabilities of obtaining the observed number of successes and failures in each group have been calculated direct from the binomial theorem, using the three different sets of estimates of the expected success rates.The combined probabilities for each set of three groups have been calculated from the formula: X= -2 logep.Our thanks are due to Professor A. Bradford Hill and Mr. P. Armitage for their interest and advice concerning the statistical methods used, and to members of the staff of the Chelsea Hospital for Women, who have referred many of the cases in this series to us for implantation.