The COVID-19 pandemic promoted disruptions in the conduction of clinical trials, as on-site monitoring visits were adjourned. In this context, the transition to RBM by all actors involved in clinical trials has been encouraged. In order to ensure the highest quality of data within a COVID-19 clinical trial, a centralized monitoring tool alongside Case Report Forms (CRFs) and synchronous automated routines were developed at the clinical research platform, Fiocruz, Brazilian Ministry of Health…. ## Introduction
| | | | |
|–|–|–|–|
|**Statement of significance**| | | |
|Problem|Risk-based monitoring (RBM) approaches for clinical trials are less widespread than expected.| | |
|What is Already Known|Although recommended by regulatory agencies and ICH guidelines, the lack of technologies for RBM is one of the reasons for the slow adoption of RBM practices| | |… |What This Paper Adds|This study aims to describe a tool developed for central monitoring of a COVID-19 clinical trial alongside Case Report Forms and synchronous automated routines. Additionally, the code for the tool is shared, which is worthwhile to contribute to open science and boost access to technologies for easier implementation of RBM practices in clinical trials.| | |… This centralized monitoring tool remotely reviews the quality of electronic clinical data to identify protocol deviations, pharmacovigilance alert signs, study performance metrics and data completeness through automated assessments. Using an R script to assure the reproducibility of this data quality assessments, the tool might enhance subjects’ safety, as the control of AE is improved; accelerate data cleaning before statistical analysis; evaluate performance metrics of trial sites; and decrease the number of on-site monitoring visits, which has an impact on trial costs, timelines, and subjects’ safety…. Additionally, to contribute to data quality assurance, synchronous automated routines were also built in PHP programming language to automatically transfer data between the different databases. Thereby, core data entry is improved, ensuring 100% of precision and consistency in critical data, such as randomization.
These open-science solutions are powerful tools for data management, pharmacovigilance, and clinical teams involved in clinical trials. Together with other open science initiatives, these tools may help push toward the adoption of centralized monitoring and RBM. Centralized monitoring is a layer within data quality assurance that might be regarded as a mandatory quality standard in clinical trials in a near future…. The objective of this paper is to describe and share these open science solutions. Fiocruz is reaffirming its commitment to Open-Science practices by encouraging the availability of data and information at each stage of the research process. The codes for these tools are available at ARCA Dados (doi:10.35078/QCXI6N), an important element of the Fiocruz-Brazilian Ministry of Health policy for the management and sharing of research data…. ## Methods
### Study and study site
These open-science solutions for a COVID-19 clinical trial were developed at the clinical research platform, Fiocruz, Brazilian Ministry of Health. Fiocruz clinical research platform funds and provides support to clinical research that evaluates new technologies to the Brazilian National Health System (SUS). Support includes clinical research crosscutting activities, such as monitoring, data management, ethical and regulatory affairs, and pharmacovigilance…. The CRFs were designed to allow automatic statistical analysis. Whenever possible, open text fields were replaced by closed questions reducing misunderstandings and time to answer. These closed questions were presented in three different multiple-choice field types: checkboxes, to select multiple answers at the same time; dropdown lists; and radio buttons, to select a single answer within mutually exclusive options…. Further on, these questions were analysed as categorical variables. Moreover, calculated fields were used to perform the automatic computation of two or more fields, such as age calculation or Body Mass Index. Text Boxes were designed with field validation to ensure quality. Branching logic guaranteed that conditional fields just pop up depending on the previous answer, for instance only females could have a positive pregnancy response. When relevant, some of these CRFs were developed according to CDISC (Clinical Data Interchange Standards Consortium) standards, including Concomitant Medications and Adverse Events forms…. ### The tools
To implement a centralized monitoring routine, an application was developed using the R (version 4.0.2) statistical software to write the code and the Shiny R package, used to build webpages based on R codes. This provided web-based data interactive monitoring dashboards through a web server with the Shiny server (version 1.5.13.995)…. In addition to the R script, two scripts in PHP programming language were developed to carry out the automatic export and import of data between the specified projects. The aim of these PHP scripts was to avoid errors in the distribution of medication/placebo bottles and mitigate the risk of including patients in the non-assigned arm causing unbalanced randomization and biased results…. These scripts linked the Recruitment/Inclusion and the bottles accountability project selecting the numbers of the labels to be used according to the patient ID at the randomization list and blocking the future use at the drug accountability project. This project had a pre-filled database with all numbers assigned to each bottle available. These values were used to feed the Structured Query Language (SQL) fields on drugs bottle selections at Recruitment/Inclusion project. Additionally, these scripts automatically save the patient ID across the two other projects ensuring a perfect linkage. These two PHP scripts that perform synchronous automated replication had been used for other clinical trials management purposes such as scheduling patients visits…. The initial page of the application presents a brief tutorial on how to use the centralized monitoring and a description of each of the ten tabs: patient inclusion rates, recruitment failure, missing data, inclusion and exclusion criteria, adverse events, laboratorial abnormalities, data validation, follow-up visits according to the study schedule and queries…. Additionally, the overall current inclusion enrolment rate and the trial performance indicators per trial site are presented in graphs…. ### Inclusion and exclusion criteria, adverse events and laboratory results
Including subjects outside the inclusion criteria age range is a protocol violation that must be avoided in order to ensure the safety of the trial subjects. Age and pregnancy are the sort of data that needs 100% SDV, however, this code adds an additional safety layer, allowing to check remotely the difference between the date of birth and the date of inclusion enrolment, as well as pregnancy…. As per the previous code’s features, a table of the secondary ID, trial site, and start and end dates are provided for each of these AE’s evaluations. A search field allows further refinement of the AEs investigations. These are major parameters for the pharmacovigilance staff and may inform a Data Safety Monitoring Board (DSMB) report…. To avoid interruptions of the most needed therapeutic trials for COVID-19 and assure the quality of data according to GCP and regulatory requirements, the industries, contract research organizations (CRO) and regulatory agencies have been encouraged to transition to risk-based monitoring (RBM). Although RBM is not a response designed to mitigate the impact of COVID-19 on clinical trials, its adoption, which has been recommended for almost 10 years, was strongly needed in the context of the pandemic…. In June 2023, the U.S. Food and Drug Administration announced the availability of a draft guidance with updated recommendations for GCP aimed at modernizing the design and conduct of clinical trials, making them more agile without compromising data integrity or participant protections. The draft guidance is adopted from the International Council for Harmonisation’s (ICH) recently updated E6(R3) draft guideline that was developed to enable the incorporation of rapidly developing technological and methodological innovations into the clinical trial enterprise [17], currently under public consultation…. The centralized monitoring tool for the remote appraisal of electronic data was designed to address the most common protocol deviations, pharmacovigilance alert signs, study performance and data completeness through automated evaluations. A script (R software) accesses the clinical trial dataset to perform these evaluations. None of these breaks the blind or information about the outcomes. The data are presented in dashboards, but, they cannot be modified using this tool. To preserve the trial dataset integrity and audit logs for tracking data handling, all discrepancies identified by the script must be manually corrected through the electronic data capture system (REDCap) query module…. There are many approaches to evaluating the quality of a clinical trial dataset, and although they are complementary, using R script-based central monitoring has advantages. As the R script is coded as a structured language, it assures the reproducibility of these assessments across actors and trials [18]. For instance, REDCap reports are more prone to errors as unreliable variables selection…. This version of the code has limitations. The results are descriptive and, in large trials, further statistical tests may be needed [21]. There is no restricted access per role within the trial, as for most clinical data software. The query assessments use data from a REDCap audit log, meaning that a routine to download this dataset is needed to update the site performance metrics dashboard…. Still, the inclusion of trial sites alongside the trial requires adjustments to the code to reproduce the actual expected recruitment rate. Finally, as the tool was developed using the REDCap data capture system and long-format datasets, additional work is needed to run the script in other formats or systems. However, as it is an open source software, future improvements of this open-science solution through collaborations and further data sharing are expected [22]…. ## Conclusion
Sharing data between researchers worldwide is an increasingly important aspect of addressing diseases and developing new therapeutic, diagnostic, and epidemiological approaches. Faced with the perspective of ensuring sustainability in Public Institutions, it is important to consider free methodologies and tools that support greater access and execution of quality research. These open science solutions are a complementary tool to clinical trials’ data management and monitoring teams rather than an on-site monitoring substitute. Together with other open science initiatives, these powerful tools may help push towards the adoption of centralized monitoring and RBM. Centralized mon