Purpose: Patient-reported outomes (PRO) may facilitate prompt treatment. We describe the development and psychometric properties of the first instrument to monitor for symptoms of breast cancer (BC) recurrence.
Methods: This study is nested in the MyHealth randomized trial of nurse-led follow-up based on electronically-collected PROs. We constructed items assessing symptoms of potential recurrence through expert interviews with six BC specialists in Denmark. Semi-structured cognitive interviews were carried out with a patient panel to assess acceptability and comprehensibility. Items were subsequently tested in a population of 1170 women 1-10 years after completing BC treatment. We carried out multiple-groups confirmatory factor analysis (CFA) and Rasch analysis to test dimensionality, local dependence (LD) and differential item functioning (DIF) according to sociodemographic and treatment-related factors. Clinical data was obtained from the Danish Breast Cancer Group registry.
Results: Twenty-two items were generated for the Breast Cancer Recurrence instrument (BreastCaRe). Cognitive testing resulted in clearer items. Seven subscales based on general, bone, liver, lung, brain, locoregional and contralateral recurrence symptoms were proposed. Both CFA and Rasch models confirmed the factor structure. No DIF was identified. Five item pairs showed LD but all items were retained to avoid loss of clinical information. Rasch models taking LD into account were used to generate a standardized scoring table for each subscale.
Conclusions: The BreastCaRe has good content and structural validity, patient acceptability and measurement invariance. We are preparing to examine the predictive validity of this new instrument.
Keywords: Breast cancer follow-up; Multiple groups factor analysis; Patient-reported outcome; Rasch models; Recurrence.
© 2021. The Author(s), under exclusive licence to Springer Nature Switzerland AG.