Assessment of the significance and severity of premenstrual tension--II. Comparison of methods.

Seven methods for the detection of premenstrual tension (PMT) have been compared, using daily records of bipolar mood scores kept by 44 women with the premenstrual syndrome during 133 menstrual cycles. Four of the methods were simple procedures based on contrasting the premenstrual scores with those elsewhere in the cycle and using either a ratio or difference to detect PMT. In the other three methods, the mood scores were fitted to a model. The comparison suggest that methods which (1) use bipolar scores, (2) optimise the thresholds separating PMT+ from PMT- cycles and (3) are based on known PMT symptom patterns, are likely to be satisfactory whatever their level of sophistication. Methods based on unipolar scores on the other hand, succeed only when PMT severity is measured as the difference between two quantities, and fail when ratios are used. In this study with its bipolar mood score data base, six of the methods gave similar results. Discrepancies between methods (9-23%) were associated with double symptom peaks and with the appearance of symptoms early in the luteal phase. For such cycles only the method based on a 5-term Fourier series provided useful information.