By Deborah Borfitz 

July 13, 2022 | Relative to clinical studies conducted at traditional bricks-and-mortar sites, decentralized clinical trials (DCTs) can deliver a higher “expected net present value” (eNPV) that equates to a roughly $10 million return on a $2 million investment for phase 2 trials and $39 million return on a $3 million investment for phase 3 studies, according to Ken Getz, professor and executive director of the Tufts Center for the Study of Drug Development (CSDD). Getz was co-presenting an Innovation Theater session at the recent DIA 2022 Global Annual Meeting on the financial return on investment (ROI) of DCTs with Pamela Tenaerts, MD, chief scientific officer at Medable.  

The economic savings, as revealed by a newly completed Tufts CSDD study, come from reduced cycle times, improved patient screening and enrollment, and fewer protocol amendments over the life of a trial. The same eNPV phenomenon was seen with patient engagement in an earlier analysis by the Clinical Trials Transformation Initiative at Duke University, says Tenaerts, previous executive director of the public-private partnership. 

A common financial ROI measure, eNPV integrates the key business drivers of cost, time, revenue, and risk into one summary metric for project strategy and portfolio decision-making by pharmaceutical companies, Tenaerts says. It's a technique widely used by portfolio managers and financial analysts across most industry sectors.   

Medable commissioned Tufts CSDD to do the study employing the technique in 2020, and provided the necessary data, producing the “first hard metrics” around DCTs, says Tenaerts. Long story short, the study shows that DCTs require a little more spending on the front end, but the returns appear sooner than in a conventional trial. Given inflation, “earlier money-making is better money-making.” Overall revenues, and ROI, are also higher.  

DCT Switchover  

The rationale for looking at the ROI of DCTs at this juncture is related to the near-instantaneous adoption of the model when the global pandemic struck with “no real piloting effort,” says Getz. On that point, he notes that “the typical adoption of a technology supporting clinical trial execution takes about six years,” based on studies of nearly 400 companies conducted by the Tufts CSDD. 

The adoption process, he says, can be broken into four primary stages—planning and initiation, taking a little over one year; evaluation, close to 16 months; adoption decision, nearly 17 months; and full implementation, about 23 months. The switchover to DCTs was so swift that “we did not have the opportunity to follow many of these traditional stages for some organizations.”  

Survey results show the same 54 companies at two different time periods just before the pandemic (February 2020) and then two years later. Across the board, the companies self-reported that they had considerably ramped up routine use of each of seven different decentralized solutions: remote monitoring, local labs, eConsent, home drug delivery, telemedicine, wearables, and home visits. In some cases, a tripling of usage level was reported, Getz says. 

Growing familiarity with DCT innovations has stirred interest in better understanding the associated investment and resource allocation “similar to what [organizations] used to get through the typical adoption cycle,” he continues. Specifically, companies are starting to seek ROI insights in terms of the when, where, and how of apportioning different solutions. “We felt that this was the right time to inform one of the primary questions: Can we get a read on the return on investment for DCT deployment?”  

The eNPV modeling approach as described here measures cash inflows (expected commercialization revenues) and R&D outflows (direct drug development operating and commercialization expenses) over the lifecycle of a single drug and “discounts it back to net present value, the value at a single point in time,” explains Getz. Future revenues generated get discounted back as the “weighted average cost of capital... essentially, a hurdle rate that companies look for to finance any investment in an asset or in capital.” 

The model calculates eNPV as the average aggregate NPV for multiple scenarios, factoring in the probability of marketplace success or failure, he says. Failure rates account for operational as well as commercial risk. For some technologies assessed in the past, for example, a certain operating model might create a longer lead time to initiate or result in lower-quality data. 

“This is only our first model, [and it’s] a really good model” but will improve further over time as it gets populated with more data, says Getz. In looking at the eNPV of DCTs, a limited amount of data is currently available for the modeling exercise. 

When a company is leading adoption of a technology, it is forced to rely on assumptions and anecdotal evidence, he notes. “We are so fortunate to have some actual data to populate our model.” 

Primary Parameters 

The modeling approach, as Getz describes it, looks at three primary parameters: development and commercialization, performance, and deployment and implementation. “We use benchmarks to give us all of the financial parameters for the typical drug that goes through R&D into the marketplace.” 

Two parameters are defined by actual data, he says, including comparative performance data based on an examination of protocols that were and weren’t supported by decentralized solutions, and the cost to deploy those solutions on a given clinical trial. 

Development benchmarks (initialization to approval, program probability of achieving approval, and mean overall direct clinical phase cost) and commercialization benchmarks (mean peak sales achieved, time to achieve peak sales, and mean exclusivity post-peak sales) come primarily from prior published work of the study’s principal investigator (Joseph DiMasi, Ph.D., associate research professor at Tufts University School of Medicine), says Getz. This is the data customarily used by the Tufts CSDD to calculate the cost to develop a single successful drug. The academic research center also publishes a lot of data on the overall development cycle, the regulatory review cycle, and success rates, based on thousands of drugs that have moved through R&D and into commercialization, he adds.  

Among the benchmarks for the development shape of the curve are a 77-month and 47-month time horizon for the typical project initiated in phase 2 or phase 3, respectively, through to approval, says Getz. The probability of a program making it from phase 2 to phase 3, as he has reported previously, is less than 20%. And of all projects that enter phase 3, only 56% will ultimately be approved. 

The direct cost of the phase two and phase three portions of a development program are, respectively, $65 million and $285 million, he reports. The expense portion of the equation “follows a curve that is shaped by the amount of time that those expenses are being incurred.” 

The commercialization benchmarks are associated with the curve following approval. It currently takes 10 years for a drug to reach peak sales, in an amount averaging $1.8 billion, says Getz.  

“Typically, a drug will only enjoy about one year beyond achieving peak sales before the patent expires,” he says. “And you see a rapid decline in revenue as generic drugs and biosimilars begin to take hold. So, that helps us characterize the shape of the curve for the commercialization portion.” 

Measuring Performance 

Performance parameters were drawn from recently completed protocols falling into one of two cohorts—those that did not involve any virtual or remote solution (154 studies) and those that did (33 studies). The latter group of trials used a wearable device or mobile application to support application of the trial, says Getz. 

“This all comes from a working group study that we completed at the very beginning of 2020, so all of the protocols had to have closed the database by December of 2019,” he adds, meaning they were all conducted before the pandemic. “That’s one of the limitations [of the protocol performance study]. As we get more granular data, we will have an even better read on impact during the pandemic and post-pandemic as well.” 

The study shows that cycle times (with one exception) are faster for protocols supported by a DCT solution. In terms of total trial duration, from protocol approval to database lock, DCT solutions offer a mean speed improvement of 246 days for a phase 2 trial and 360 days for a phase 3 trial.  

The coefficients of variation around the mean are high, he notes, which is why the modeling exercise used conservative assumptions. “But we went into the modeling with a lot of confidence in the conservative values because of the comparative improvements that we saw from these DCT-supported protocols... [including] much lower screen failure rates, lower dropout rates, and a smaller average number of amendments.” 

These can all be significant cost drivers, as the Tufts CSDD has repeatedly shown. “The number of substantial amendments is extremely costly, unplanned, and unbudgeted,” Getz says. “A single amendment typically adds three months of time on average, and in a phase 3 study, upwards of half a million dollars in direct unbudgeted cost.”   

Sensitivity Analysis 

Medable provided data on the investment required to deploy DCT support in a clinical trial, he says. In total, 59 actual contracts—33 for phase 2 studies and 26 for phase 3 studies—were examined to come up with a mean contract value for a single clinical trial by phase and the average number of trials in that phase supporting a development program. 

For the average number of phase 2 clinical trials per program (four), each with a mean contract value of $473,000, a total investment of $1.9 million is required, Getz reports. The figure for phase 3, based on a mean contract value is $1,042,000 for the average number of required trials (three), is roughly $3.1 million. 

Getz shared results of eNPV modeling in the context of different DCT scenarios with a “very conservative base case improvement of 10% [cycle time reduction],” or three months of time savings for both phases 2 and 3. “We get roughly five times the $1.9 million investment in terms of a return. And that translates to nearly $10 million increase [in the eNPV]. 

“As we change the shape of the DCT-supported curve relative to the typical drug or program life cycle for phase 3,” he continues, “if we look at even a 5% reduction in cycle time, we... [see] a 5.5 times multiple over that $3.1 million investment, which delivers roughly a 17-plus million dollar increase in eNPV.” 

Next Steps 

Findings from the sensitivity analysis are “very compelling,” says Getz, despite the limitations posed by the small sample size and lumping together all DCT solutions into one bucket. The plan is to look at specific solutions by individual disease condition soon, to see where the returns are the most compelling. 

The Tufts CSDD also hopes to measure and quantify impact beyond cycle time, recruitment and retention, and protocol amendments, Getz says. This would include the impact of improved automation on dose adherence and investigative site implementation costs. 

Medable intends also to continue working on the eNPV model, says Tenaerts, noting that many contracts are still in flight. “We will refine and rerun it again with more actual and more granular data.” Actual numbers are needed for “responsible adoption of DCTs.” 

The company is also attempting to “get hard numbers” on the value of DCTs to the patient experience, Tenaerts adds. A platform is being created whereby Medable can automatically send out surveys to patients, and the same sort of data will be collected from study sites. 

eNPV modeling looks purely at financial value, Getz concludes. The intangible benefits include the goodwill generated when patients have a positive experience in a clinical trial, “when you are able to demonstrate that you are making a commitment to a patient community.”