OBJECTIVE: There have been promising developments in technologies to stratify rheumatoid arthritis (RA) patients to treatments. However, the acceptability of these approaches in clinical practice could depend on their predictive ability, the time it takes to process the results and the cost to the health service. This study sought to quantify the public’s and patients’ preferences for a new, guided, approach to prescribing biologics over conventional ‘trial and error’.
DATA AND METHOD: An online discrete choice experiment (DCE) was designed to elicit Swedish individuals’ preferences for prescribing biologics with a stratifying algorithm (described as a prescribing programme) or conventionally through trial-and-error. The survey was first piloted in qualitative ‘think aloud’ interviews (n=3). Members of the general public and diagnosed RA patients were then recruited via an internet panel provider. The approaches to prescribing were described by five attributes: delay to starting treatment; positive predictive value; negative predictive value; risk of infection; and cost saving to the healthcare system. The un-labelled DCE was blocked into four surveys. Each survey contained six choice-sets with two prescribing programmes (A or B) and a conventional approach (opt-out) generated by a D-efficient design using Ngene. Internal validity was tested through the inclusion of a dominant choice-set. Background questions included sociodemographics and questions about health status and disease activity, where applicable. DCE data were analysed using conditional logit models and heteroskedastic conditional logit models (HCLM) which allowed the scale term to vary by the RA status of the respondent.
RESULTS: Two-hundred and eighty-nine people (patients, n=133; public, n=156) completed the DCE. There were no statistically significant differences in the sociodemographic characteristics of the patient and public sample, except for the employment status as ‘long term sick leave’ which was reported more often amongst RA patients. In the public sample, all attributes were in line with a priori expectations with increasing levels of delay and risk being disliked and increasing levels of costsavings and predictive values being preferred. All attributes except risk were statistically significant (p<0.01). In the patient sample, cost saving to the healthcare system was also not statistically significant (p=0.111). The scale term in the HCLM was not statistically significant (p=0.681) suggesting heterogeneity in preferences was driven by differences in taste. The most important attributes in both samples were the positive and negative predictive values of the prescribing approaches. A negative and statistically significant alternative specific constant for the ‘conventional prescribing’ option suggested all respondents preferred the stratified approach above and beyond the attributes presented in the choice experiment. It was estimated 91.2% of RA patients and 74.9% of the public would choose the biologic calculator in a typical prescribing scenario.
CONCLUSION: The results of this study suggest that both patients and the public are receptive to stratified approaches to treatment. However, decision-makers should consider the importance of positive predictive value to consumers when appraising the effectiveness of stratifying technologies. Further research will be conducted to confirm the functional form of the preferences and the drivers of preference heterogeneity within and across the two samples.
