Storrrick, Elizabeth M.

151 Background: Decision aids (DAs) for prostate cancer treatment can improve knowledge and reduce decisional conflict, but the relative effect of pre-visit and within-visit DAs is not known, and effect sizes for minority populations has not been estimated. Methods: We conducted a 3-arm, patient-level-RCT in specialty urology and radiation oncology practices in Ohio, South Dakota, and Alaska, test the effect of pre-consultation and with-in consultation decision aids on patient knowledge immediately after specialty consultation compared to usual care. We used linear regression to estimate effects of each intervention arm, including the respective standard error, two-sided 95% confidence interval, and two-sided P value for testing the study’s hypotheses. Results: 103 patients were recruited and randomized to receive either the pre-visit decision aid, within-visit decision aid, or neither decision aids (usual care). In 2017 and 20018, we accrued similar numbers of men to pre-consultation aid (n = 37), during-consultation aid (n = 33) and usual care arms, respectively (n = 33). The median (range) age in years was 64 [49, 81]; 67.6% were White, 15.7% were Black or African American, 16.7% were American Indian or Alaska Native, and 1% were not reported. 47.6%, 45.6%, and 6.8% had a baseline clinical stage of T1, T2, and T3, respectively. The median [range] prostate specific antigen (PSA) was 8.0 [2.4, 53.7]. There were no clinically notable imbalances. We obtained usable data on 102 of the 103 patient-participants. The pre-visit decision aid arm showed a mean knowledge effect of 0.694 (0.636, 0.753). The within-visit decision arm showed a very similar mean knowledge effect of 0.686 (0.625, 0.748). The usual care arm showed a mean knowledge effect of 0.644 (0.582,0.705). The linear regression model showed, that, compared with usual care, neither intervention effect achieved statistical significance in the primary analysis (p = 0.24, 0.330, for pre-visit and within-visit, respectively). Conclusions: Modest knowledge gains of pre-visit and during-visit decision aids for prostate cancer treatment were not statistically significant. Clinical trial information: NCT03182998 .

e24176 Background: Decision aids for prostate cancer treatment can improve knowledge, but the relative effect of pre-consultation and within-consultation DAs is not known. Methods: We conducted a parallel design, 2-by-2 factorial, 4-arm, cluster-RCT in urology practices affiliated with the NCI’s Alliance research base. We compared patient knowledge immediately after consultation (%12-item% correct) intervention arms compared to usual care. We used mixed effects regression modeling to assess effect of interventions on knowledge adjusting for site-clustering, race/ethnicity (non-Hispanic White; Other), age (years), clinical stage (T1-3), PSA, and Gleason score. Results: 15 of 21 randomized centers accrued 158 pts from Nov 2017 to June 2019. The arm delivering both DAs (n = 3 sites) recruited 25 pts. The pre-consultation DA arm (n = 4 sites) recruited 39 pts. The within-consultation DA arm (n = 4 sites) recruited 44 pts. The no DA arm (n = 4 sites) accrued 50 pts. Overall, mean age was 64; 54% were AA; 73% had T1 stage; median Gleason = 7; and median PSA = 7. The combined DAs group, followed by pre-consultation reported higher median knowledge (75% and 67%, respectively). The within-consultation and no DA arms both reported median knowledge scores (58%). Within arm, inter-site standard deviations in mean knowledge scores ranged from 0.164 in in the pre-consultation arm to 0.232 in the usual care arm. Neither pre-consultation nor within-consultation DAs had statistically significant effects (p = .15 and p = .99, respectively). Pre-specified adjusted difference [97.5% CIs], comparing the pre-consultation DA and the within-consultation DA with the control was 9.4% [-7.3%, 26.2%] and -0.4% [-16.9%, 16.1%], respectively. Intra-cluster correlation (ICC) coefficient was high (.24). No interaction effects were identified (p = .98). A post-hoc analysis comparing any DA vs usual care found a mean adjusted difference [95% CI] of 5.6% [-8.5%, 19.8%] (p = 0.40), 63.9% for any DA and 58.3% for usual care. Conclusions: Possible knowledge gains associated with some DAs for early prostate cancer were obscured by larger-than-expected intra-cluster correlation effects associated with the primary outcome. How interventions were used was not measured. Support:UG1CA189823; https://acknowledgments.alliancefound.org ; R01 MD008934-06. Clinical trial information: NCT03103321 .