By Deborah Borfitz

March 12, 2021 | Thanks to COVID-19, asset leads at GlaxoSmithKline (GSK) can now tap a Clinical Control Tower for answers to five critical questions about how studies are progressing and if they are veering off course. The platform was developed last year to enable trial leaders to maintain business continuity during the pandemic, says Stephen Gardiner, head of analytics and reporting. 

“The impact of COVID was seemingly difficult to predict,” says Eddie Dukes, tech product owner at GSK. “It was evolving daily and different country to country.”

Understanding current conditions affecting trial delivery, and when development plans need to be amended, requires access not just to data spread across various internal systems but also to the local know-how of on-the-ground representatives, says Dukes. The target initially was a COVID Control Tower, as he and Gardiner discussed in depth at last week’s Summit for Clinical Ops Executives (SCOPE).

The platform was designed to be a single source of truth across the clinical operations portfolio, says Gardiner, and enable action by its focus on five key risk indicators: What is the current country and payer mix? Where is COVID-19 having an impact? How is enrollment progressing against plan? What is the current projected delay? What are the local conditions on a country level?

For senior executives, the dashboard displays real-time data on multiple levels, including by portfolio, therapy, asset, trial, and country, says Dukes. Use of interactive infographics and dashboard filtering allows users to see answers to all five questions in one view providing an overall picture of risk in different countries and regions as infection rates were skyrocketing.   

As a result of the insights brought to teams in terms of increased visibility and risk oversight, Dukes adds, the pandemic had “little to no impact” on recruitment across the company’s high-priority trials. 

Internal and external data sources were leveraged to build the Control Tower, starting with trial recruitment plans that provided a baseline for delivery, says Dukes. “Actual recruitment showed us how trials were performing in the real world.”

Local conditions were assessed with the introduction of country-level surveys that served as a critical data source for the Control Tower, Dukes continues. These surveys were conducted weekly and looked at specific institutional review board guidelines such as whether CRAs could make on-site visits, home nurses could be implemented without triggering a protocol amendment, patients faced restrictions accessing sites, or eSignatures were considered an acceptable form of eConsent.     

Separately, internal teams were surveyed weekly about trial-related holdups, he adds.

COVID-related data available within the platform includes publicly accessible data from Johns Hopkins University on the prevalence of COVID-19 across different geographies around the world, says Dukes. Through collaboration with an internal biostatistics group, a predictive model for forecasting the pandemic’s impact was introduced.

The platform includes geomaps that can be used to visualize portfolio- and therapy-level data by country, he notes, and line graphs for comparing various recruitment metrics. In addition, there are tables showing the expected delay per country across the portfolio and bar graphs depicting delays by therapy area. 

Tower Build 

Platform functionality has evolved in the way it answers the five key questions, says Dukes. Users can see the current country and patient mix at the asset or trial level, providing a picture of the current position relative to the overall therapy area. They can also “slice and dice” data to understand the impact of patient enrollment on study initiation by country. 

The Control Tower can display the country-by-country impact of COVID-19 and overlay asset-, trial- or therapy area-level data to identify geographies where the pandemic is having the greatest effects, he continues. The country-level surveys enable a read on the “state of affairs regarding trial delivery across our footprint and across the country as well as providing supportive commentary from our in-country teams.” 

Views on projected delays is made available at various levels—by country and trial for those involved in day-to-day operations and by portfolio and therapy area, highlighting countries with the most delays, for senior leaders needing a more holistic view, says Dukes. Progress versus plan is likewise offered in all four views within the dashboard. Users can also use the predictive model to see projected enrollment based on the COVID-19 situation in specific countries, allowing teams to respond if the forecast deviates significantly from the plan. 

Creation of the COVID Control Tower was a multi-departmental effort with individuals “contributing what they could alongside their day job to deliver value early and often with regular releases [increasing platform functionality] from May through October,” Dukes says. The project kicked off midway through April and in just three weeks they had a “minimal viable product” providing a single view of all priority trials across the portfolio, plus all insourced phase 2 and phase 3 trials. 

The second release came within one month, expanding the Control Tower to cover GSK’s entire pharma portfolio as well as ingesting CRO data for all fully outsourced trials, says Dukes. This version of the product is when the views by portfolio, therapy area, asset, trial, and country were built out, he adds. 

By July, ingestion of the country- and trial-level survey data was automated to reduce the burden of physically uploading those data points into the system, Dukes continues. The platform was also now producing executive one-pagers providing senior executives with the headline numbers regarding the impact of COVID-19 across priority trials.

For the third release in September, vaccine data was being inserted to provide a fuller picture of conditions. This was a particularly challenging, Dukes says, since the team’s vaccine colleagues have historically worked from a separate system and data landscape. 

September also marked a major milestone in terms of ingesting the internal predictive model for COVID-19 forecasting, allowing the user base to expand via regular demos across GSK’s asset and trial lead communities, says Dukes. The product reached maturity by October when ownership of the Control Tower was transitioned to tech colleagues.

Improved R&D Infrastructure

Among the standout learnings from the Control Tower build was the “ability to work at pace,” says Dukes, noting liberal use of agile tools and “ceremonies” (aka meetings). “We were able to identify data sources, establish data flows, and build analytics to spec within three weeks.”

Problem-solving via inter-departmental collaboration and creativity was “essential to overcoming blockages and enabling us to deliver at pace,” he continues. “This project came about when we had only recently transitioned to working from home. That forced us to leverage the virtual tools made available to us and the team thrived when it came to coming up with innovative solutions to business and technical problems.”

But the greatest learning, Duke says, was understanding the data landscape and its limitations. “We were aware from the offset that we had a complex data landscape, but I don’t think we fully comprehended the complexity associated with our systems and processes until we really got our hands dirty and began to explore… how our data was made available to us.” 

That realization led to the ambitious new enterprise solution called the Clinical Control Tower to meet the needs of an integrated pharma and vaccines organization, as well as the need for an improved R&D infrastructure, says Gardiner. GSK is newly invested in an integrated data fabric to “untangle” the complex data landscape, and in data governance solutions, so the company is better able to bring data sources together in the future to support new technology initiatives.