157

Introduction: Inability to procure a priori informed consent from patients or substitute decision makers (SDMs) limits recruitment in ICU studies. The objective of this study is to describe the experience of deferred consent as an adjunct to a priori consent in a low risk, observational, critical care biomarker study. Hypothesis: We hypothesized that a deferred consent model would optimize recruitment in a low-risk observational study. Methods: We conducted a prospective observational study to validate the prognostic utility of cell-free DNA predicting mortality. Enrolment occurred from Nov 2010-July 2012 in 2 ICUs in Hamilton, Ontario. Patients had to be recruited within 1 day of ICU admission. We employed deferred consent when a priori consent was not feasible within the recruitment window. Data and blood samples were collected daily during the first week, then weekly thereafter. Results: Over 26 months, 320 patients were approached for enrolment. Of 61 a priori consent encounters, consent was obtained for 57 (93.4%) patients; 4 declined. Of 259 deferred consent encounters, consent was obtained for 257 (99.2%) patients; 2 declined. Of 257 patients enrolled using deferred consent, full permission was granted for data retention thus far as well as future specimen acquisition for 201 (78.2%) patients; partial permission was granted for data retention thus far but no additional specimens for 21 (8.2%) patients; neither patient or SDM was available for 35 (13.6%) consents. Given the mixed consent model, 314 patients were enrolled of 320 eligible (98.1% consent rate overall). Of 57 successful a priori consent encounters, 53 (92.9%) of consents were obtained from a SDM and 4 (7.0%) from the patient. Of 222 successful deferred consent encounters, 199 (89.6%) were obtained from the SDM and 23 (10.4%) were obtained from the patient. Conclusions: In this low risk study, deferred consent was the dominant consent model (257 patients enrolled with a 99% consent rate). A priori consent was used less often (57 patients enrolled with a 93% consent rate). The mixed consent model yielded a 98% overall consent rate, which optimized patient recruitment, shortened study duration and will enhance the generalizability of our findings.