Aging Data: Study Highlights Delays From Time of Clinical Trial Enrollment to Publication

The average age of data among randomized clinical trials published in six high-impact medical journals in 2015 was 3 years, according to a new analysis. Substantial delays were identified during the enrollment and publication time points, leading the researchers to argue that improvements can be made all along the research process.

“Things are moving so fast in medicine. There's a premium on being able to generate knowledge that's as close to the present time as possible,” senior study author Harlan Krumholz, MD (Yale-New Haven Hospital, CT), told TCTMD. “The thing with any of these trials is that the follow-up period is immutable—so if you follow up for a year, it's going to take you at least a year to do anything. You're not going to shorten that.” By contrast, he continued, “the amount of time it takes to enroll people and how long it takes to get the paper out [are] under our control.”

For the study, published online Friday in JAMA Network Open, Krumholz along with first author John Welsh, BA (Yale-New Haven Hospital), and colleagues looked at the 341 clinical trials published in 2015 by the following journals: Annals of Internal Medicine, BMJ, JAMA, JAMA Internal Medicine, Lancet, and New England Journal of Medicine. Almost two-thirds of the trials studied drugs (60.4%), 6.2% looked at devices, and 33.4% assessed other interventions. About half of the trials had a follow-up period of 1 month to 1 year (50.4%), and 19.9% of trials required more than 4 years to complete enrollment.

The overall median data age (primary endpoint), defined as the midpoint of data collection until publication, was 33.9 months. Of note, the studies that had a follow-up period of 1 month or less were not necessarily the fastest to complete, with a median data age of 30.6 months. The overall median publication time—from final data collection to publication—was 14.8 months, with 18.5% of trials requiring 2 or more years.

On multivariate analysis, older data age (> 235 days) was associated with inconclusive or unfavorable trial results. Also, solely industry-funded trials were associated with a shorter time to publication (< 180 days) compared with those sponsored by government organizations.

“These aren't small, preliminary trials that end up in very low-impact places where people can argue that it wasn't a big deal that there was a delay,” Krumholz said. “These are the ones in our most prominent journals and yet we're still seeing they're taking a long time.”

Leveraging Digital Resources

“How do we wipe the slate clean and . . . fundamentally transform the way in which we pursue clinical trials?” Krumholz asked. “Because this is an indication that it's just not working as well as it could. You can imagine if many of these clinical trials are identifying treatment strategies that are going to help people get better outcomes, then every day it's delayed from publication is a day that deprives people of that knowledge.”

In the current digital age, he said, researchers should be leveraging electronic records to “preregister people so that once the trial is ready, we're ready to go.”

You can imagine if many of these clinical trials are identifying treatment strategies that are going to help people get better outcomes, then every day it's delayed from publication is a day that deprives people of that knowledge. Harlan Krumholz

Similarly, Ori Ben-Yehuda, MD (Cardiovascular Research Foundation, New York, NY), told TCTMD that digital technologies should be able to aid in the recruitment of clinical trials and hence shorten the overall time to publication. On the flip side, digital tools lead to a greater volume of data, which “requires more cleaning and oversight. The more data you collect, the more potential for actual errors,” he said.

As for the data age shown here, Ben-Yehuda said, “I don't think that 3 years is really such a long time in the context of medical care. Things don't change that much over 3 years. So I'm not sure how big of a problem this is.”

Yet once the data age of a study reaches around 7 or 8 years, Krumholz argued, the generalizability of the results is questionable.

“There's always room for improvement, and we need to try and shorten the time as much as possible but never at the expense of quality,” he continued. “So you want to get the right patients. You want to have oversight. You want to have time for making sure the data is correct and clean and analyzed correctly. If anything, there is also the counter of: are things rushed at times too much? We have things being presented at meetings that may be preliminary, and is that the right approach?”

Another target for shortening overall time to publication is the start-up period even before the first patient is enrolled, Ben-Yehuda said. Institutional review board submissions, Food and Drug Administration approvals, and site contracts often take a long time to finalize, but employing standardized, accepted language across sites could help with that, he suggested.

“There are risks in clinical trials and the issue of indemnification is one that always delays things,” he added, suggesting that perhaps a national fund be set up to handle complications that occur during clinical trials—much like the one in place for vaccine-related complications.

Think Like Patients

“In a biomedical evolution, we've got to be able to have a clinical trial enterprise which can rapidly test our hypotheses and do so in rapid cycle, so that we can reload and do the next trial,” Krumholz argued. “I think that up until now, every one has just been painfully slow and expensive and also time is running. So when these trials really get extended, they end up costing more than they would otherwise.”

Because conversation about this topic has been kept relatively mum among researchers, he said he hopes this study “sparks a discussion about how do we do better in clinical trials, what are the key performance indicators that we should be taking as evidence, and how do we become transparent about for each trial how successful was it in efficiently achieving its goals.”

Lastly, Krumholz encouraged trialists to begin thinking of patients enrolled in their trials not as subjects, but partners. “The more they work with patients as partners, the more they realize that for people with disease, time moves differently—like the next year before you learn the result matters,” he said. “If you've got Parkinson's or some progressive debilitating disease in particular [and] you were the one writing the paper, you would write it the same day. You're driven to get the knowledge out there as fast as possible because your life depends on it.”