Predictors of participants’ retention—socioeconomic factors or nonadherence: insights from a urological clinical prospective study

Abstract

Abstract Background To investigate various patient-level variables, specifically socioeconomic status, as risk factors for withdrawal in a recently completed clinical study. We specifically investigated a non-interventional prospective study assessing the role of novel imaging as a biomarker for cancer upgradation in prostate cancer for this objective. Methods In this retrospective analysis, we assessed the association between various patient-level factors including clinic-demographic factors, socioeconomic status, and the number of non-adherences with the participants’ retention or withdrawal from the study. For socioeconomic status (SES), we used the zip code–based Economic Innovation Group Distressed Community Index (DCI) which classifies into five even distress tiers: prosperous, comfortable, mid-tier, at-risk, or distressed. Low SES was defined as those with a DCI Distress tier of at-risk or distressed. We compared values between the two retention and withdrawal groups using t-test, chi-square test, and logistic regression analysis. Results Of 273 men screened, 123 men were enrolled. Among them, 86.2% (106/123) retained through the study whereas 13.8% (17/123) withdrew from the study. The mean (SD) age was 64 (6.4) years. Overall, 31.7% (39/123) were Hispanics and 24.3% (30/123) were African Americans. The median (IQR) DCI score was 34 (10.3, 68.1) and 30.8% (38/123) of patients belonged to low SES. The median DCI score in participants who retained in the study was statistically similar to those who withdrew from the study (p=0.4). Neither the DCI tiers (p=0.7) nor the low SES (p=0.9) were associated with participants’ retention or withdrawal of the study. In terms of non-adherence, all participants in the withdrawn group had at least one non-adherent event compared to 48.1% in the retained group (p<0.001). Repetitive non-adherence was significantly higher in participants who withdrew from the study vs those who retained in the study [88.2% vs 16.9%, p <0.001]. On multivariate logistic regression analysis, the number of non-adherences (OR=12.5, p<0.001) and not DCI (OR=0.99, p=0.7) appeared to be an independent predictor for participants’ retention or withdrawal from the study. Conclusions Expanding diverse inclusion and limiting withdrawal with real-time non-adherence monitoring will lead to more efficient clinical research and greater generalizability of results.