ABSTRACT
OBJECTIVE: The aim of this study was to validate the performance of a new pattern recognition method for quantifying physiological hot flushes against conventional fixed criterion methods and to explore its suitability for use in ambulatory clinical trials. METHODS: This study performed a secondary analysis of baseline and outcome sternal skin conductance monitoring data from two recent randomized controlled trials of cognitive-behavioral therapy (CBT) for hot flushes in breast cancer patients (MENOS1) and healthy peri- and postmenopausal women (MENOS2) using a revised pattern recognition method (Bahr monitor; software version 1.1.6). Sensitivity and specificity were recalculated and compared with previous findings, based on conventional criteria, using monitor-defined flushes as "gold standard" and combined baseline data. Outcome data for physiologically measured flushes between the CBT group and the treatment-as-usual group were separately reexamined using the revised method for each trial. RESULTS: Pattern recognition showed higher concordance (36%), sensitivity (0.64), and specificity (0.99) than the standard method. Hot flushes recorded during the day showed slightly higher concordance (39%) and fewer false-negatives than 24-hour recordings. Based on the revised method, well women randomized to CBT (MENOS2) had significantly fewer flushes than those randomized to treatment as usual at posttreatment (P < 0.04), but CBT did not impact on physiologically measured hot flushes for breast cancer patients (MENOS1). CONCLUSIONS: Pattern recognition can identify flushes more reliably by detecting the shape of physiological signals rather than by relying solely on the amplitude of their signals (as used in conventional criteria). With application of the revised method, ambulatory sternal skin conductance monitoring detects changes after a CBT intervention in well women but not in breast cancer survivors.
