PURPOSE: The increased desire and need to quantify preferences for health and healthcare has led to the development of various elicitation methods, recently noted and catalogued by the IMI-PREFER consortium. Among these methods, the threshold technique (TT) has gained attention arguably due to its simplicity and suitability in small samples. However, there are multiple ways to design and implement TT exercise in preference elicitation. We aimed to systematically identify all TT studies in health and healthcare to provide a description of current practice by summarizing empirical applications in terms of methodological approach and key study findings.
METHODS: A systematic review was conducted searching health-related TT publications on PubMed, Science Direct via Elsevier, and Embase, Medline, PsycINFO, Econlit via OVID. We also used ‘pearl growing’ to ensure that all relevant literature was included even where it was not identified by the database search. Two authors independently determined eligibility of studies. Data were extracted relating to publication characteristics, thresholding attributes and endpoints, type of respondent, sample demographics, question format, reported heterogeneity, and validity assessments. Extracted data were analyzed using descriptive statistics.
RESULTS: Overall, 75 (20%) of the 380 identified articles met the inclusion criteria for extraction (Figure 1). The first study was published in 1991. Although we found slightly more TT studies executed since 2020, we did not observe any notable publishing trends in the literature over time. Most of the studies elicited thresholds across multiple endpoints (risks, benefits, time, cost, or others), and explored heterogeneity through either split-sample comparisons or multivariate regressions. A few studies (n = 11, 15%) included multiple samples and compared preferences across respondents. The term “multiple thresholding” only appeared recently, and no formal introduction or definition was identified. The choice question format was similar across studies (i.e., a table); however, the presentation and discussion varied greatly, and a handful used more creative presentations to engage respondents. The review found few tests of validity with rationality assessment included in the minority of studies (n = 16, 21%).
CONCLUSIONS: This systematic review reveals use of the TT is steadily growing in health. However, the review identified heterogeneous study designs and lack of standard definitions, suggesting a need for guidance on conducting and reporting TT preference research.
