Abstract

An investigation of 129 in-patient depressives treated with E.C.T. is presented. All were followed up to 3 months, and 108 to 6 months. Initially all were scored for the presence or absence of 35 features considered to discriminate between endogenous and neurotic depressions. Diagnoses were allocated before or early during treatment. Patients were rated at the termination of E.C.T. at 3 months and at 6 months, on a four-point scale. At 3 months, only 12 out of 63 neurotic depressives were found to have responded well to E.C.T., whereas out of 53 endogenous depressives had done so; the corresponding figures for 6 months were 8 out of 53 and 34 out of 44.

A factor analysis of the features produced three significant factors, a bipolar factor corresponding to the distinction between endogenous and neurotic depression, a general factor with high loadings for many features common to all the depressive cases studied and, probably, a paranoid psychotic factor. The bipolar factor closely resembles that extracted in a previous investigation (Kiloh and Garside, 1963) and supports the hypothesis that these are two distinct depressive populations. Among features with high positive loadings on the first factor, and thus corresponding to a diagnosis of endogenous depression, were adequate premorbid personality, absence of adequate psychogenic factors in relation to illness, a distinct quality to the depression, weight loss, pyknic body build, occurrence of previous depressive episode, early morning awakening, depressive psychomotor activity, nihilistic, somatic and paranoid delusions, and ideas of guilt. Among features with a negative loading corresponding to a diagnosis of neurotic depression were anxiety, aggravation of symptoms in the evening, self-pity, a tendency to blame others, and hysterical features.

By means of multiple regression analysis, three series of 18 weighted coefficients for the differential diagnosis between the two varieties of depression and for the prediction of E.C.T. response at 3 and 6 months were calculated. The multiple correlations between the summed features on the one hand and diagnosis and outcome at 3 and 6 months on the other were 0.91, 0.72 and 0.74 respectively. It was thought that E.C.T. response could be better predicted by the direct use of the weights for E.C.T. response than from the diagnostic weights alone. The weights based on the 18 features were complex, and therefore a further table was constructed giving simplified weights based on ten features of diagnosis and ten features for prediction of outcome of treatment with E.C.T. after 6 months.