F31. WHAT ARE THE IMPLICATIONS OF RECRUITMENT SOURCE (CLINICAL OR NON-CLINICAL) IN SCHIZOPHRENIA TRIALS FOR SUCCESSFUL RECRUITMENT AND RETENTION IN TERMS OF SCREEN FAILURE, OR NON COMPLETION OF TRIAL?

Autor: Koschalka, Luke, Martinez, Mairelys, Smigorski, Krzysztof
Předmět:
Zdroj: Schizophrenia Bulletin; 2019 Supplement 2, Vol. 45, pS266-S267, 2p
Abstrakt: Background Recruitment is one of the greatest challenges in performing randomized controlled trials. The recruitment environment is a competitive one, with a large number of studies recruiting from a finite patient pool. Commercial pressures mean products are moved through the research and regulatory process as quickly as possible. This often results in ambitious recruitment targets. As a result, recruitment strategies often move beyond the traditional, clinically-driven approaches that focus on a Principal Investigator′s (PI) own patient list and that of their wider clinical network, and towards strategies such as online recruitment, community outreach approaches, self-referral and other approaches (hereafter termed collectively as ′Non-ClinicaĹ approaches). Whilst these approaches bring more potential participants to a study, it has not been established what implications there are for quality of participant in terms of recruitment and retention. This poster examines whether research participants from Clinical (C) and Non-Clinical (NC) sources differ in two aspects - levels of study randomization and of study completion. Methods Recruitment source data were gathered from four Phase 3 and one Phase 4 clinical trials in schizophrenia. 3258 patients were classified as either C referrals (i.e. PI's own patient, referral from another practitioner, or research database) or NC referrals (from advertising, internet-derived referrals, self-referrals, or outreach programs). Chi-square analyses were carried out to examine goodness-of-fit. Results For combined data from all trials, highly significant differences were found between C and NC referrals for number of patients randomized vs number not randomized (p <.001). C referrals were highly significantly more likely to randomize (also significant for each study individually). Highly significant differences were also found for patients who randomized, but completed the study vs those who randomized, yet failed to complete the study (p <.01) also significant for four out of five studies individually). Patients from C referral sources who randomized were significantly more likely to complete the study. The evidence shows that C sources of referrals produced highly-significantly lower rates of failures to randomize and early terminations than did NC sources of referrals. Discussion Whilst the division between C and NC sources of data is arbitrary, it does possess face validity. In addition, these data would suggest that it possesses some clinical validity. The significantly different rates of study randomization and completion indicate underlying qualitative differences between these sources of referrals. A number of factors might be hypothesized to underlie these differences, amongst them implicit pre-screening by PIs of C referrals, patient investment levels, or other demographic factors (e.g. differing levels of comorbidity, distance from site) etc. However, at present, available data are not of sufficient granularity to be able to answer these questions. Future data recording should allow for better encoding of relevant demographic factors. Greater consistency and accuracy of encoding of referral source would also allow for more detailed analysis of the validity of the C and NC constructs. Future studies might also examine whether NC referrals that do randomize and go on to complete studies perform in a similar way to C referrals. [ABSTRACT FROM AUTHOR]
Databáze: Complementary Index