Presentation (CTI)

OBJECTIVES: As decentralized designs are increasingly used in observational treatment-outcome research, recruitment remains a key challenge despite gains in reach and efficiency. We describe recruitment methods, innovations, and operational metrics from decentralized observational studies and considerations for selecting fit-for-purpose design/recruitment models.
METHODS: Four recent prospective observational studies (three with ongoing recruitment) in rare disease, immunology, and neurology, conducted in single (US)- or multi-country (US+ Europe) settings, were reviewed for participant identification/recruitment methods, operational metrics, and design adaptations. Considerations relevant to decentralized recruitment/design methods were summarized.
RESULTS: Decentralized design suitability depended on whether key variables (medical-record data and patient-reported outcomes (PROs)) could be collected remotely, and whether country-specific ethics/regulatory requirements across studies permitted decentralized procedures. Recruitment models included fully virtual sites (n=2) and hybrid models (virtual/traditional sites) (n=2), with planned enrollment ranging from 60-450 participants. Studies used centralized call center/data-management approaches; 3 used websites and social media/online patient-network campaigns, and 1 partnered with a patient advocacy group. Prescreening questionnaires yielded up to 10-fold more respondents than ultimately eligible, highlighting the value of well-designed patient-facing pre-screeners. Enrollment conversion rates ranged from 2%-23% of screened participants. Traditional sites were added/considered when pre-enrollment loss to follow-up was ≥50% (n=2 studies), when evidence generation was better supported in-person (blood draws (n=1 study), PROs involving physical movement (n=1 study)), or when local regulations (European countries) did not allow virtual designs (n=2 studies). A key consideration was whether identification of participants is needed at treatment initiation; potential solutions include partnerships with specialty pharmacies and major treatment centers.
CONCLUSIONS: Decentralized recruitment/operations can support efficient enrollment in observational studies, especially in the US, but recruitment models should be fit-for-purpose and adaptable to population needs, participant feedback, and operational challenges. Hybrid models may improve recruitment/evidence generation when virtual approaches limit retention, time-sensitive identification, collection of site-supported assessments, or use in some countries.