Cyclicity and premenstrual symptomatology
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
This thesis reviews the major aetiological and treatment literature on premenstrual syndrome and concludes that methodological inadequacies, particularly subject selection and retrospective measurement, leaves the area in considerable disarry. Thus the first aim of the experiment, was to examine the relationship between a single interview selection procedure and the symptom data collected over time subsequent to this interview The second aim of the experiment was to use treatment expectations to oppose negative expectancy and so further examine the adequacy of subject selection procedures. The third aim of the experiment was to collect sufficient prospective data to be able to use frequency domain time series analysis (spectral analysis) as an alternative to visual and non-probabilistic methods of determining cyclicity Forty two subjects who reported symptoms of premenstrual syndrome were recruited. The first interview, during the late luteal phase, was based upon the rating scales of Steiner et al. (1979). Daily mood and symptom data were collected for the next three or four menstrual cycles, at which time a second interview was held. Thirty subjects agreed to continue recording for additional treated cycles, at the conclusion of which a third interview was held. Groups constituted on the basis of self-reported severity do vary significantly with respect to prospectively recorded mood symptoms but not with respect to physical symptoms nor incidence. The overlapping variance between the retrospective interview ratings of severity and prospectively recorded symptoms is small but best predicted by a simple Visual Analogue Scale rating of the previous months symptom severity. Placebo induced treatment expectations significantly reduced incidence, premenstrual aversive mood and physical symptoms. Repeated use of the interview questionnaires do not show major overlapping variance and post-treatment ratings of symptom severity do not show a significant relationship with the symptom records collected during treatment. Again the rating of the previous month on a Visual Analogue Scale was the best predictor. The use of spectral analysis was successful in identifying both menstrual and non-menstrual cyclicity. Its use suggested three criteria for selecting subjects. They should have a menstrual length peak in both mood and physical symptom spectral density functions. These two series should show significant coherence over the range of significant menstrual period spectral density peaks. Finally these features should not be lost when placebo treatment is given. It was concluded that the use of the above criteria and the use of both conventional and spectral density methods would be likely to reduce the confusion and uncertainty within the area and be an appropriate means of evaluating potential treatments.