Date Published: April 25, 2019
Publisher: Public Library of Science
Author(s): Mark Harrison, Luke Spooner, Nick Bansback, Katherine Milbers, Cheryl Koehn, Kam Shojania, Axel Finckh, Marie Hudson, Matthew Quaife.
To understand preferences for and estimate the likely uptake of preventive treatments currently being evaluated in randomized controlled trials with individuals at increased risk of developing rheumatoid arthritis (RA).
Focus groups were used to identify key attributes of potential preventive treatment for RA (reduction in risk of RA, how treatment is taken, chance of side effects, certainty in estimates, health care providers opinion). A web-based discrete choice experiment (DCE) was administered to people at-risk of developing RA, asking them to first choose their preferred of two hypothetical preventive RA treatments, and then between their preferred treatment and ‘no treatment for now.’ DCE data was analyzed using conditional logit regression to estimate the significance and relative importance of attributes in influencing preferences.
Two-hundred and eighty-eight first-degree relatives (60% female; 66% aged 18–39 years) completed all tasks in the survey. Fourteen out of fifteen attribute levels significantly influenced preferences for treatments. How treatment is taken (oral vs. infusion β0.983, p<0.001), increasing reduction in risk of RA (β0.922, p<0.001), health care professional preference (β0.900, p<0.001), and avoiding irreversible (β0.839, p<0.001) or reversible serious side effects (β0.799, p<0.001) were most influential. Predicted uptake was high for non-biologic drugs (e.g. 84% hydroxycholoroquine), but very low for atorvastatin (8%) and biologics (<6%). Decisions to take preventative treatments are complex, and uptake depends on how treatments can compromise on convenience, potential risks and benefits, and recommendations/preferences of health care professionals. This evidence contributes to understanding whether different preventative treatment strategies are likely to be acceptable to target populations.
Rheumatoid arthritis (RA) is thought to develop through “multiple hits”, with genetic and environmental risk factors, followed by antibodies such as rheumatoid factor (RF) and anti-citrullinated protein antibodies (ACPA),[3–5] that accumulate during an “at-risk” pre-clinical phase. This pre-clinical phase lasts 3–5 years before culminating in clinical disease and can be described in five phases.[6–9] Asymptomatic phases (A-C) depend on whether individuals have genetic (Phase A) or environmental (Phase B) risk factors, or systemic autoimmunity associated with RA (Phase C). In Phase D, people first develop symptoms such as joint pain but do not have clinical arthritis. In Phase E individuals are described as having unclassified arthritis and in Phase F, RA. Increasingly, it is thought these pre-clinical phases offer a window of opportunity for potential preventive treatment.
A web-based discrete choice experiment (DCE) survey was administered to a sample of self-reported first-degree relatives (FDRs) of RA patients from the USA. FDRs represent individuals at an elevated risk of RA.[18,19] DCEs, developed in market research, assume any product, including health care interventions, can be described by characteristics (attributes) and that value of products depends on the levels of these characteristics. An RA-related treatment attribute might be how treatment is taken, and levels could be oral, injection or infusion. DCEs are useful where products or services do not exist (i.e. before launch). DCEs are underpinned by random utility theory. When choosing between options, individuals are assumed to assign a perceived utility (or attractiveness) to each alternative which depends on characteristics of both the alternative and the individual, and choose the one with the highest perceived utility. Random utility theory acknowledges that the utility people assign to an alternative cannot be measured (and must be treated as a random variable) meaning it is not possible to predict with certainty which alternative people will choose. Instead, random utility theory attempts to predict the probability that alternative A will be preferred to alternative B, and that this choice is proportional to the degree to which alternative A is valued more (has a higher utility) than alternative B.
Of 525 initial respondents, 288 (55%) had an FDR with physician-confirmed RA currently taking a drug for RA, meeting our inclusion criteria as FDRs. No participants were excluded for completing the survey within 3 minutes (minimum time-to-complete 3.28 minutes; mean 9.11 (SD 8.29)) and none always chose Treatment A or treatment B. The ‘no treatment for now’ option was chosen 33% of the time (for 856 of the total 2592 choices made), and 10% of participants (n = 29) always chose no treatment. The sample was primarily female (60%), aged 18–39 years (66%), and most reported European (84%) ancestry (Table 3). A minority (12%) reported no medical insurance; those with private insurance mainly had employer plans (51%), 22% had Medicare/Medicaid coverage. Most (91%) indicated that they would be willing to pay out of pocket for preventive treatment for RA.
Results of this study suggest asymptomatic, at-risk individuals, may be willing to take preventative treatment to reduce their risk of RA, although they have different appetites for risk, reassurance, and convenience of treatments. Importantly, our findings suggest preventive treatments preferred by at-risk individuals will not necessarily be those offering the largest reduction in risk of RA; uptake is likely to be driven as much by how treatment is taken, opinions/preferences of health care professionals, and the risk and reversibility of side effects. Assuming that treatments currently being evaluated in RCTs all offer preventive efficacy, we predict highest uptake of convenient treatments offering low-to-moderate reductions in risk of developing RA and low risks of serious side-effects. This suggests that non-biologic DMARDs like hydroxychloroquine and potentially methotrexate are more likely to be acceptable preventive options than the biologic DMARDs being tested.
We provide preliminary predictions of uptake of preventive treatments for RA in currently RCTs. Our findings suggest people may take preventive treatment, but reduction in the risk of developing RA is one of multiple factors influence their preferences. Furthermore subgroups exist that prioritize safety, reassurance, and convenience. Decisions to take preventive treatment are complex. Only treatments that balance these preferences will be acceptable to at-risk populations, and as preferences are not predicted by socio-demographic characteristics they may need to be elicited on an individual basis. Preferences of potential recipients should also be considered when designing RCTs to prioritize studies of interventions likely to taken by potential recipients if they meet their primary endpoint.