Directly Observed Therapy using REDCap Twilio and Video

Directly Observed Therapy using REDCap, Twilio, and Video Uploads from Mobile Devices John T. Clark, Erin C. Quigley, Aimee L. Mc. Rae-Clark, Pharm. D Jihad S. Obeid, MD, Leslie A. Lenert, MD, MS ABSTRACT METHODS (cont’d) DISCUSSION Documentation of patient compliance with therapy in clinical studies is always difficult. Directly observed therapy is a gold standard approach for documentation. We have created a tool kit for documentation of compliance via text message prompted video capture of medication consumption. Consented participants are instructed to document medication ingestion using small video clips captured on their mobile device when prompted via the SMS link and upload videos via the survey (Figure 2 b). Uploaded videos prompt a data entry trigger (DET) to contact a listening web service, which then downloads the videos using the REDCap API with custom PHP code. Video metadata including the timestamp is extracted using Exif. Tool (1) and inserted into the participant record (Figures 3 and 4). This allows study staff to ensure the video was not recorded at an earlier time. Study staff then review the video and validate that the medication was ingested. Overall the system has worked well. Our biggest challenge has been capturing videos when participants have poor wi-fi or cellular connections. Since this uses a web page as a transfer mechanism, it is wholly dependent on the participant connected to a strong network. INTRODUCTION Documentation of compliance with medication prescriptions is a critical aspect of pharmaceutical research. Various approach have been used to “prove” that patients have been compliant with the medication under study. These include pill counts, use of blister packs for medication, and electronic pill bottle caps that record openings. With the exception of biomarker tests, which are applicable only to a small number of pharmaceuticals, most measures of compliance stop short of certain knowledge that the prescribed medication has been taken. For example, pill bottle openings do not necessarily lead to drug ingestion. Therefore, the gold standard for measurement of adherence has been Directly Observed Therapy (Do. T). Do. T is difficult to implement and inconvenient for patients. We describe an approach for Do. T implemented within REDCap using short message service (SMS) texts for video confirmation via mobile devices. As a result, there has been interest (and potential funding) in creating a crossplatform mobile application that captures this data and queues for video upload. DET Figure 5 – Overview of workflow CONCLUSION Figure 1 - Project longitudinal setup METHODS A pharmacotherapy clinical study was setup in REDCap using the longitudinal model with daily encounters (Figure 1). REDCap Automated Survey Invitations were setup to send SMS text with a survey link to participants twice per day using Twilio service (Figure 2 a). a b Figure 3 - Video is uploaded and script processes within seconds Figure 4 - Video metadata is added to record Do. T was feasible to implement using text reminders and video uploads. Previous medication trials conducted in the same clinical population as the present study have reported adherence rates as low as less than 20% using biological markers as indicators of compliance (2) and poor concordance between objective measures of compliance and participant self-report (3). This methodology could be applied to other clinical trials or research studies where ingestion of medications or other activity needs to be observed by investigators as part of the consented research. REFERENCES RESULTS Figures 2 - Text message via Twilio (a) with a link to the mobile survey for video upload in REDCap (b). The system was piloted on 42 participants, with a total of 2431 text messages, which prompted 1776 video uploads through mobile devices, yielding a 72% overall validation rate. 1. https: //sourceforge. net/projects/exiftool/ 2. Gray KM, et al. Drug and Alcohol Dependence. (2017) 177: 249 -257. 3. Mc. Rae-Clark AL, et al. Journal of Substance Abuse Treatment. (2015) 57: 70 -74. Acknowledgements: This work was supported by NIH Grant #s UG 3 DA 043231, UL 1 TR 001450