[A new concept of psychosomatic research exemplified by premature labor]

One of the most serious problems of psychosomatic research is the adequate choice of the method applied to clinical questions. In the past empirical investigations have been based upon univariate and moreover linear correlations. In this study a multivariate nonlinear concept was developed which was generated in a first trial proven and confirmed prospectively in a second one using two sorts of mathematical analyses. The first included 35 patients with premature labour and 25 healthy pregnant women as controls. The latter was conducted with 238 patients being recruited in the first trimester of pregnancy, 30 of whom developed premature labour in the course of the study. Using a questionnaire consisting of the main aetiological variables identified in the basic investigation and using the statistical procedures mentioned above the occurrence of premature labour was predicted in more than 90%. Only one patient who was not suspected to develop premature uterine activity was not classified correctly. Besides clinical usefulness the model of research presented provides evidence for the most important aetiologic conditions of the disease as far as psychosocial factors are concerned and offers a new approach to quantifying research including somatic and non-somatic variables.