Comparing two models of health behaviour to explain tamoxifen non-adherence in women with breast cancer

Abstract

Background: Tamoxifen non-adherence is apparent in up to half of breast cancer survivors and is associated with increased risk of recurrence and reduced quality of life. However, factors contributing to non-adherence are currently poorly understood. This study investigated the predictive validity of the Common Sense Model of Illness Representations (CSM) and the Theory of Planned Behaviour (TPB) in explaining intentional and unintentional non-adherence in a large sample of women prescribed tamoxifen following primary breast cancer. Methods: 817 women took part in the questionnaire study. Participants were recruited in clinic or online and completed questionnaires assessing illness perceptions, treatment beliefs, adherence, quality of life, social support, distress and the key TPB components. Patients (n=357) in their first year of treatment were sent follow-up questionnaires at three, six and twelve months. Expected results: Longitudinal data will be analysed using structural equation modelling to identify predictors of non-adherence, as well as mediators and moderators. Current stage of work: Baseline data were analysed using logistic regression to investigate if the TPB or the CSM successfully predicted non-adherence after controlling for confounders. Patients were classified as non-adherent based on Medication Adherence Rating Scale scores. Both intentional and unintentional non-adherence were best predicted by a combination of TPB and CSM variables. These health behaviour models were better able to predict intentional (R2 = 36%) than unintentional non-adherence (R2 = 17%). Discussion: Understanding modifiable predictors of tamoxifen non-adherence will facilitate the development of interventions to support women, with the aim of increasing adherence and improving clinical outcomes.