BACKGROUND: The increased use of patient preference data in healthcare decision making has raised concerns about the reliability and consistency of the estimates generated by patient preference studies. However, literature reviews to assess the consistency of preferences are confounded by heterogeneity in study designs. METHODS: This paper adopted a novel approach to evaluating preference consistency: comparing estimates of a single trade-off-the marginal rate of substitution (MRS) between survival improvements and risks of adverse events-across multiple patient groups and using meta-regression to assess whether MRS varied systematically between patients. A log-linear, random effects regression was run, weighted for the sample sizes of studies from which estimates were extracted. RESULTS: Using studies identified in published reviews of patient preference data, 42 estimates of MRS were generated from the 12 studies. On average, patients obtained the same utility from a 2.3% reduction in the risk of an adverse event as from a 1-month increase in survival, with a range of 0.002-13.5%. The regression model had an R2 of over 90% and suggests that MRS depended on patients' expected survival and the type of adverse event being traded. CONCLUSION: These results suggest that although preferences vary between patients, they may do so in systematic and predictable ways. Further, they do so in ways consistent with societal preferences and decision maker priorities toward end-of-life settings. Further work is required to replicate this result in other patient groups and to explore the consistency of preferences for other treatment attributes.
BACKGROUND: The increased use of patient preference data in healthcare decision making has raised concerns about the reliability and consistency of the estimates generated by patient preference studies. However, literature reviews to assess the consistency of preferences are confounded by heterogeneity in study designs. METHODS: This paper adopted a novel approach to evaluating preference consistency: comparing estimates of a single trade-off-the marginal rate of substitution (MRS) between survival improvements and risks of adverse events-across multiple patient groups and using meta-regression to assess whether MRS varied systematically between patients. A log-linear, random effects regression was run, weighted for the sample sizes of studies from which estimates were extracted. RESULTS: Using studies identified in published reviews of patient preference data, 42 estimates of MRS were generated from the 12 studies. On average, patients obtained the same utility from a 2.3% reduction in the risk of an adverse event as from a 1-month increase in survival, with a range of 0.002-13.5%. The regression model had an R2 of over 90% and suggests that MRS depended on patients' expected survival and the type of adverse event being traded. CONCLUSION: These results suggest that although preferences vary between patients, they may do so in systematic and predictable ways. Further, they do so in ways consistent with societal preferences and decision maker priorities toward end-of-life settings. Further work is required to replicate this result in other patient groups and to explore the consistency of preferences for other treatment attributes.