Web-based data system clean, fast and efficient

Recruitment was on time; results were quick

Sometimes it takes someone from outside the research field to come up with a clinical trial (CT) solution that is logical, effective, and successful.

A new web-based data management system serving clinical research has improved data collection and accuracy through an ongoing data collection and analysis process.1

In a study of Amyotrophic Lateral Sclerosis (ALS) patients, the new approach helped produce on-time recruitment, quick results and analysis, and low error rates and missing data, says Richard Buchsbaum, senior data manager at the statistical analysis center, biostatistics department of Mailman School of Public Health, Columbia University of New York, NY.

The process resulted in less than 0.7% errors on submitted forms, and data were available within 48 hours of when an event occurred. Also, an analysis data set was produced nine days after the final patient visit.1

These striking outcomes were only possible from thinking outside of the box.

Buchsbaum had been working in advertising and finance before he began working in health research about 11 years ago.

After years of working on a 24-hour basis in posting information and numbers, Buchsbaum found it astonishing to move to a field where people were expected to wait for months or years before analyzing data.

"I remember 12 to 13 years ago doing work for Bear Stearns, and when you're processing trades you have at most 24 hours to post trades and various transactions," Buchsbaum says.

"The idea that you can wait to look at information until you're done is not something I came to health research thinking was okay," Buchsbaum adds.

Looking at CT data on an ongoing basis requires more upfront thought, but it pays off with benefits down the road, he notes.

"First of all, you can address problems before they arise — both in terms of recruitment trends, loss to follow-up, increasing medication compliance, and all kinds of things that if you wait until the end is too late," Buchsbaum says. "It's much easier to approach something in real time."

Buchsbaum puts together data systems for CT projects and epidemiological trials.

"It's not all about data collection," he notes. "It's about the collection of data to answer the research question and integrating that with logistical data and using an integrated data system to uphold all of the processes of the trial."

Whether the research team is monitoring patient compliance, scheduling patient appointments, or tracking shipment of biospecimens back and forth, it's all done under one database, Buchsbaum explains.

"And investigators believe that contributed to the success of the trial," he says.

The web-based data management is not a novel approach technologically, but it is slightly unusual in how the systems are integrated, Buchsbaum says.

"You'll have a lot of systems devoted to capturing forms and others devoted to other logical aspects of the trial," he explains. "And either they don't talk to one another or they don't reinforce each other."

So often even small data collection, such the data captured for one particular patient visit, had to be completed before the work could move on to subsequent visits, Buchsbaum says.

"I'm stunned by how many data capture systems rely on coordinators or clinical trial people to make sure things are done completely and in the right order," he says. "And the status is good to know, but often is not known."

With the Web-based data management system, CT staff easily can see the status for each patient in terms of visits and data entry, Buchsbaum says.

"It is simple to see what is the latest status for each patient," he explains. "So that makes it very easy for coordinating centers in a trial to monitor where everybody is."

In the typical system, the data often are segregated, so there are many questions about data that are unanswered, he adds.

"If something is missing, you have some affirmative notion that this task wasn't done," Buchsbaum says. "But you don't know whether it's because the information wasn't entered, the task was not performed, or if the patient refused to participate or give consent."

There are a variety of possible reasons for the omission, but the answer isn't recorded because the data are segregated, he says.

This problem doesn't exist with Web-based data management system.

"We didn't consider these things separate issues, and so everything was part of the clinical trials process, and everything was fair game for the data management," Buchsbaum says.

The Web-based system also was easier for CT coordinators to use, according to feedback, he notes.

Since CT staff work on a half dozen different trials at one time, it could have been difficult for coordinators to jump in and out of a trial, trying to remember where they were with this patient or that patient, Buchsbaum explains.

"It was easy to do that with the system we designed," he says. "The system gave clear instructions of what happened at the previous visit and what comes next, so the burden was easier for coordinators."

The system also makes it easier and timelier for investigators and CR administrators to monitor trends in enrollment and retention.

For the ALS trial, there were only two patients out of 185 enrolled who were lost to follow-up, Buchsbaum says.

"That's a very low loss to follow-up rate for this population," he says.

The reason for retention success partly was due to how easily investigators could monitor patients' visits and communicate this information to CT sites, Buchsbaum says.

Sites received a list of participants who were due to have their final outcomes visit, and the sites tracked how many telephone calls were made to these people, he adds.

"We published a bunch of Web pages and real-time pages so the statistics and data center monitors and clinical monitors could keep track of each patient and what their risk was to loss of follow-up," Buchsbaum explains.

The items listed included the number of tasks performed in previous visits, the number of visits skipped, and telephone calls made to patients, he adds.

These measures served as warning signs to potential loss to follow-up.

One key to successfully implementing a Web-based management system is to have the clinical trial team meet to discuss the study's data requirements and to create an infrastructure to meet those requirements, Buchsbaum says.

"You can't shoehorn the requirements," he notes. "Some studies may not need real-time access; some require PDAs and onsite data entry, but the point is that the focus should be on the needs of each trial."

Also, CR institutions should focus on all of the processes needed to conduct the trial and the data systems needed to assist with these, Buchsbaum adds.

Despite the obvious benefits of such a Web-based management system, research institutes have not embraced the model due to the economic reality of health research, Buchsbaum says.

"In academia, it's hard to convince either sponsors or investigators that these things are worth spending money on," Buchsbaum says. "It's hard to spend money on infrastructure because it doesn't always seem to be directly revenue-generating, although it can save you money in the back end."

Reference

  1. Buchsbaum R, Kaufmann P, Barsdorf AI, et al. Web-based data management for a phase II clinical trial in ALS. Amyotroph Lateral Scler. 2008;9:1-16.