Abstract
Background: Breast cancer (BC), the most prevalent malignancy in women, poses a significant burden on various stakeholders, including healthcare payers, patients, caregivers, and society as a whole, both globally and within New Zealand (NZ). Advances in treatment offer diverse options, each with unique benefits, risks, side effects, and costs, complicating treatment decision-making for physicians and patients. The increased demand for high quality health services highlights the need for integrating patients preferences into treatment decision-making. Conventionally, optimal treatment strategies for BC have been largely determined by physicians following clinical guidelines or evidence-based research, with a scant understanding of patients’ treatment values and preferences. Transitioning towards a patient-centred approach can enhance health outcomes and treatment adherence by prioritizing individual needs and preferences. Therefore, this thesis aims to elicit patients preferences for BC treatment in the NZ context. Specifically, it seeks to quantify treatment preferences in terms of the benefit-risk trade-offs and marginal willingness to pay, as well as explore the preference heterogeneity.
Methods: This thesis employed a mixed-method design, encompassing a systematic six-stage approach to developing a discrete choice experiment (DCE) instrument and collecting data. Stage 1 combined a scoping review and qualitative interviews to identify treatment attributes important to BC patients. Stages 2 and 3 further refined and prioritized these attributes through triangulating perspectives from various stakeholders, establishing the final list for the DCE survey. Stage 4 involved developing the pilot online DCE survey using a Bayesian fractional experimental design. Stage 5 validated the pilot survey through pre-tests and pilot tests, including ‘think aloud’ and focus group discussions. Quantitative analysis of the pilot data provided a priori coefficients to refine the final DCE survey. In Stage 6, two versions of the final DCE surveys were created, with respondents randomly assigned to one version. The qualitative data was analysed with thematic analysis, whereas the DCE data were analysed using conditional logit models to quantify the preference weight and relative importance of treatment attributes, interaction conditional logit models to explore observed preference heterogeneity, and mixed logit models to evaluate unobserved preference heterogeneity, benefit-risk trade-offs, marginal willingness to pay (mWTP), and marginal willingness to wait (mWTW).
Results: Thematic analysis identified four overarching themes characterizing patients preferences for BC treatment: (1) Positive treatment outcomes, (2) Treatment-related side effects and their negative impact on quality of life (QoL), (3) Accessibility, availability, and timeliness of treatment, and (4) Cost of treatment. From these themes, six attributes were chosen for inclusion in the final DCE survey: overall survival (OS), risk of hair loss (HL), risk of peripheral neuropathy (PN), waiting time to access treatment, QoL, and out-of-pocket (OOP) cost. A total of 233 high-quality DCE responses were included in the final analysis. Analysis of the DCE data showed that BC patients prioritized OS and QoL more highly than the other four attributes. The interaction conditional logit model indicated that medical insurance, income level, healthcare system utilization, marital status, and education level were statistically significant predictors of patients preferences. The benefit-risk trade-off analysis demonstrated that patients were willing to accept a maximum increase in the risk of HL and PN of 60.3% and 20.6%, respectively, for a one-year increase in OS.
Additionally, patients showed a strong mWTP for improvements in OS and QoL, being willing to pay an average of $582.01 and $1,456.75 per month for a one-year increase in OS and an improvement in QoL, respectively. Unsurprisingly, they exhibited negative mWTP values for increases in the risks of HL and PN, as well as for increased waiting time to access treatment. Similarly, the mWTW analysis revealed that patients were willing to accept longer waiting times as a trade-off for better OS and QoL or to avoid increased risks of side effects or higher OOP costs.
Conclusions: This thesis fills a crucial research gap by examining how BC patients in NZ make decisions and trade-offs between treatment benefits and risks. Notably, this is the first study to assess BC patients' treatment preferences in NZ both qualitatively and quantitatively. The key contributions of this thesis include quantifying the trade-offs BC patients make during treatment decision-making. These insights can enhance treatment protocols, inform policy decisions, and support a more patient-centred and effective healthcare system in NZ, benefiting both BC population and society at large.