Engaging Patients and Clinicians to Determine Design Features of n-of-1 Trials [Internet]

Excerpt

Background: Achieving patient-centered care and optimizing personalized treatment decisions require having information on the effects of a given treatment for a particular patient. Interindividual variability in treatments effects (known as heterogeneity of treatment effects [HTEs]) occurs when treatments that are effective for the average patient may vary in the degree of effectiveness between patients and may even be minimally effective or harmful in some patients. The n-of-1 trial is a single-patient, multiple-period crossover experiment comparing ≥2 treatments within patients. They address HTEs by providing empirical data on treatment effects for individual patients. Despite their potential to inform treatment selection, uptake of n-of-1 trials in clinical practice remains low. Previous research suggests that implementation has been hindered by a lack of understanding of use cases (eg, conditions, diseases, symptoms, and ideal circumstances) and design features that make n-of-1 trials most marketable to patients and clinicians.

Objectives: We aimed (1) to conduct in-depth focus groups with patients and clinicians to determine the circumstances under which n-of-1 trials are considered most valuable; (2) to survey a national sample of patients with multiple chronic conditions (MCCs) to elicit n-of-1 trial preferred medical conditions, symptoms, and outcomes; and (3) to survey a second national sample to determine preferences for an n-of-1 trial design. We surveyed patients with MCCs, who are often excluded from randomized controlled trials and are likely to have substantial HTEs.

Methods: Focus groups with 54 patients and 24 clinicians and a subsequent nationally representative Harris Poll Online (HPOL)-administered survey of 501 multimorbid patients elicited preference for n-of-1 trial conditions/symptoms (eg, “list conditions willing to be in an n-of-1 trial for”) and design attributes (eg, “how many times per day willing to monitor?”). These data informed a second national survey of 501 multimorbid patients with conjoint questions that elicited preference for n-of-1 trial designs. Specifically, patients were asked to choose from among 3 pairs of n-of-1 trial designs, with each design composed of 8 attributes (eg, frequency of self-monitoring) at different levels (eg, 1 vs 3 times/day). Levels were based on interquartile ranges of design preferences identified through the responses from survey 1. Regression analyses determined coefficients for relative importance of different attributes. Positive marginal utilities corresponded to participants preferring baseline (lower-level, less-intensive options; eg, 1 time/day), and negative marginal utilities corresponded to preference away from baseline. P values indicated whether attributes significantly affected choices.

Results: In the focus groups and first national survey, hypertension, hyperlipidemia, diabetes, depression, back pain, arthritis, insomnia, and respiratory issues were the most patient-preferred conditions, while blinding, treatment options, intensity of self-tracking, cost, study duration, and doctor involvement were the most important design attributes warranting further exploration. The second survey revealed that participants preferred no costs versus $100 cost (utility difference, 1.52 [SE, 0.07], P < .001) and less versus more time commitment per day (utility difference, 0.16 [SE, 0.07], P < .015) but did not have preferences for the other 6 attributes. In subgroup analyses, participants who were ≥65 years old, were White, and had income ≤$50 000 were more averse to costs than were their counterparts (P < .05 for all).

Conclusions: Patients with MCCs selected chronic conditions for which n-of-1 trials are feasible. Further, they preferred designs that minimize out-of-pocket costs and the daily time burden of self-monitoring, with no preferences for other attributes like blinding and study duration. Optimizing the dissemination of n-of-1 trials in clinical practice may benefit from aligning with these patient-preferred attributes.

Limitations: Survey methods may not capture nuances in patients' comprehension of an n-of-1 trial.