Reply to “Advancing trial recruitment science through enhanced study design”

Abstract

We deeply appreciate the interest that Drs Lewicki and Stensland have taken to reflect on the operational tactics introduced in our recent Cancer article¹ and the demonstrable association between our centralized email, CISTOquestion, and patient accrual in the CISTO study (ClinicalTrials.gov identifier NCT03933826). Patient accrual remains the primary reason for failure or premature termination of clinical trials within urology and oncology at large, and reliable predictors of trial success have yet to be discretely identified.^2-4 Our goal is to share the strategies that we have found to maximize enrollment in the CISTO study so that these methods can be strategically used to optimize the success of future multi-institutional trials that strengthen evidence-based practices derived from these studies.

Among the engagement approaches used by the CISTO study team, our data suggest that CISTOquestion effectively contributed to increased and sustained enrollment and was unequivocally favored by CISTO site research coordinators, allowing for continual engagement. We agree that, from our analysis, causality between CISTOquestion and enrollment remains uncertain because of the nonrandomized rollout of CISTO study site participation and CISTOquestion. As Lewick and Stensland suggested, randomized, controlled methods for grouped site involvement and time-regulated CISTOquestion access would allow for determining a definitive effect of CISTOquestion on patient enrollment among different sites. Alternatively, it is important to consider that controlled access to CISTOquestion itself may limit site engagement, thereby counteracting enrollment for sites that rely on CISTOquestion for patient screening. This would be a tradeoff between site engagement/enrollment and determination of a true association between CISTOquestion and accrual. Such a compromise may be mitigated by allocating one half of the sites access to CISTOquestion for a brief period of time, as thoughtfully proposed.

Upon review of the feedback survey data administered to CISTO site research coordinators, CISTOquestion was regarded as a more beneficial resource versus all sites meetings, which aimed to facilitate structured engagement with CISTO study sites. Staffing and logistical coordination were substantial investments in the execution of all sites meetings, yet CISTOquestion was rated as more beneficial to CISTO site research coordinators given that principal investigators and CISTO study staff who responded to inquiries provided personalized, patient-specific responses. This was a testament to how CISTOquestion provided the most value given the resources used to initiate and maintain it. Considering the advancements in technology being integrated into trial design and clinical practice, the impact of CISTOquestion can be enhanced through the use of artificial intelligence with its capability of learning on a case-by-case basis how principal investigators respond to patient-specific inquiries. This would further minimize resource constraints that impede the success of large clinical trials.

Through the lessons conveyed in our article,¹ we hope to normalize the process of critical reflection and learning among the many moving parts and challenges of managing a large, pragmatic, multi-institutional trial. An embedded prospective investigation on trial design within a larger study such as the CISTO study would require considerable preplanning and is itself a lesson learned that would further enrich clinical trials to come.

Angela B. Smith reports grants/contracts from Genentech and Merck outside the submitted work. Krupa K. Nathan disclosed no conflicts of interest.