A Case Study on Blinded Monitoring of a Clinical Trial with a Web-based Application
Large and/or lengthy clinical trials need to be monitored periodically to ensure recruitment is as planned, plan for interim and final analyses, identify and fix potential bottlenecks. Blinded monitoring is a popular approach to tracking the progress of clinical trials in terms of enrollment, loss to follow-up, protocol compliance, safety and accrual of efficacy data. A recent case we worked on shows how this can be done using custom web-based tools.
The sponsor needed blinded, yet comprehensive tracking of enrollment, safety, compliance and event accrual for a lengthy and large-scale vaccine trial. The trial is a 3-look group sequential design with time-to-event endpoints. Predicting the timing of interim and final analyses is difficult because of variation in accrual rates, incidence rates, adjudication of cases etc.
Uncertainty in disease incidence leads to uncertainty in accrual of the required number of events needed for interim and final analyses. This, in turn, affects the planning of DSMB meetings and regulatory submission timelines.
Variability in site-specific recruitment rates risks delay if enrollment lags behind assumptions from the design phase. Adjudication of cases is another point at which delays are common.
The sponsor needed to ensure that scheduled visits including those for immunogenicity and reactogenicity data collection were adhered to. They also had to manage loss to follow-up at the site level to prevent extended study duration and reduction in power.
The trial is expected to run for a few years. During this time, the sponsor has to monitor progress at regular intervals.
At each such period they will need to re-run programs as new data arrives. They will need data programmers and statisticians to generate updated results. This will make each such monitoring, time-consuming and error-prone.
To address these challenges, we proposed building a custom, web-based software application that can be used for periodic monitoring of the trial. The key features of the application included:
The app provides real-time blinded monitoring of recruitment, event accrual, pending adjudication, loss to follow-up and protocol compliance (missed visits).
The app has built-in data checks that support early identification of issues and facilitate data queries.
The app uses Bayesian models to predict timing of interim and final analyses, timings of site-specific recruitment milestones and landmark event times. The predictions account for the uncertainties around recruitment and event rates.
The app allows users to track scheduled visits, immunogenicity sample collecton and lost to follow-up by site. Pending adjudication of cases by site is another important feature. This helps the sponsor identify an important bottleneck that delays event accrual.
The application supports multiple data formats including .xlsx, .csv, .sas7bdat, .RData, .rds and .json. A user-friendly GUI makes it accessible to non-statisticians such as operations and medical affairs personnel.
The app provides summaries and graphical representation of the data that are easy to follow. It also generates a downloadable PDF report that can be used for internal strategy meetings.
The app is hosted on a secure server with 24/7 availability. This ensures that confidential trial data is hosted securely, adhering to data privacy regulations.
We deployed the app in an in-house server with secure access to users.
The app has a separate module for prediction of time to enrollment and event milestones. This helped the sponsor improve strategic planning for DSMB meetings.