At this year’s meeting of the Society for Clinical Data Management, “standing room only” doesn’t begin to cover it. Many sessions were devoid of a place to sit, crouch or stand.

That’s a testament to the quality of the presentations at the meeting, which were clear-eyed discussions of the actual state of clinical data management—not some nirvana to be reached in the future.

As an example, Michael Goedde of Human Genome Science discussed how to create a massive but manageable data management plan in a single day. The idea: put reusable documents in a central repository, and link to them using HTML. In the event of an inspection, the underlying documents are assured to be current and easily located. His approach generates reports that show missing data that slowly declines over time, allowing auditors to chart the progress of issues being resolved.

Goedde joked about data management plans at other companies. Such documents, he said, can be too specific, even to the point of naming the employee who will do a particular task on a particular day of the week. “Their standard operating procedures (SOP) are on such a detailed level that it is almost impossible to comply. They have names of colleagues! If Maria is not in, or Monday is a holiday, then God forbid you are not compliant any more.”

Lower Your Voice

Often as not, the speakers at SCDM were describing work under adverse conditions. “Don’t yell,” admonished Michelle Harrison, manager of clinical data management at Coley Pharmaceutical. “A colleague of mine who works for a contract research organization was thrilled that she was not working for a particular sponsor, because they yell.”

Harrison said she doubted that one common recipient of abuse, the technology vendor, was necessarily such a deserving one. “Are you sending them all the information?” she asked the SCDM audience in Chicago. “Are you being clear and concise? There are very few incompetent vendors.”

Building Relationships

Her advice was not to chat about the weather but something more meaningful. Building a relationship with suppliers of all kinds, she suggested, could allow favors to be called in later. “Chances are you are going to have to ask that person on a Friday afternoon at 2:00 o’clock for some deliverable. If you have a relationship, chances are you’ll get it.”

Two presenters from PharmaNet didn’t actually say they had been yelled at, at least in so many words. But they also had advice for their fellow data managers in a trying situation: “We avoided laying blame or being judgmental. We were always open and willing to speak up,” said Marina Kelly, manager of data management.

Embracing Difficulty

Kelly described a expanded access protocol (EAP) study in which the sponsor ill-advisedly expected the data to be clean to the specifications of a Phase III study. Trouble was, the case report form was Kafkaesque, by Kelly’s account, and many of the sites slipped into a catatonic, non-responsive state.

“We received a lot of incomplete and missing data. We received multiple duplicate pages. There were a lot of issues dealing with the sites. Some of them were unresponsive. They pretty much got tired of the study,” says Kelly.

That project is ongoing. Eventually, the percentage of critical data to be cleaned was revised downward from 100 percent to 20 percent. Quality control of the adverse event data was adjusted less sharply, from 100 percent to 80 percent. Says Kelly: “The sponsor was understanding and aware of the challenges. They knew the data would never be declared clean in a traditional sense.”

A PharmaNet colleague of Kelly’s, data manager Martha Johnson, also spoke at the meeting. “No one thought to ask the data management department if they had standards for working with late-phase studies or EAP,” said Johnson. Some patients were transferred between sites, said Johnson, but without any notification to the sponsor. “There were still some we had to track down like CSI or FBI agents,” she reports.

A Global Headache

Lilly’s Kathy Haag, associate clinical data management consultant, described a post-marketing surveillance program running with 2 countries using electronic data capture, and 15 countries using paper forms. The project was the merger of two post-marketing programs. All told, there were 5,000 patients and 400 investigators.

An unidentified clinical data management system was to be retired during the project. “The best option would be to move all the data in the paper system into the electronic data capture system we were already using. Sounds easy, right?” Haag said.

In the post-approval world, she noted, it’s not so common to have a detailed data management plan. But Lilly did, and the company stuck to the plan religiously. “We used the work plan to our meetings. We reviewed it. We updated it frequently. We got new information and things would need to change,” Haag said.

The key elements of the Lilly plan: communication; documentation; risk mitigation; realignment of data management; and identification of uploading issues. Different workflows for paper and electronic queries had to be managed.

Document Everything

The study is expected to run for a decade. So every decision had to be recorded. “We also made a very concerted effort to document all the decisions we made, all the processes, and all of the rationale for what we did,” said Haag. “We developed a central repository for our study information and a collaborative site with our CRO.”

It does not sound as if the technology, never identified, was a dream come true. “The system blew up a lot. I know that’s not a technical term,” said Haag.

The folks at Rho, meanwhile, had something equally unexpected. According to Jamie Hahn, clinical data manager, the contract research organization (CRO) took on a post-approval trial of some 59,000 patients and a goal of 2,000-3,000 clinical sites. The project was mandated by the FDA, had real-time reporting of adverse events about a medical device.

One wrinkle: Because of FDA insistence on 65 percent retention of patients over the 10 years of the study, the operational plan included gathering (and securing) a large amount of personal data like subjects’ social security numbers that could be used to re-establish contact if they went missing.

Calibrating Expectations

Project manager Alison Fowlkes picked up the story, and noted that Rho develop a patient-reported outcome application that could have enrolled subjects using the internet. Except that physicians, for reasons that remain murky, greatly preferred paper—possibly because they could just hand out a booklet of forms for the patients to fill in.

As it turned out, 77 percent of the subjects used paper forms to enroll, not the 5 percent that Rho and the sponsor had anticipated. That made real-time reporting difficult, to say the least, and required intensive shipping of study binders, case report forms and other materials.

Noted Fowlkes: “We quickly saw that everyone was using paper as opposed to using the electronic application. It had major impacts on the way the study would go. We had to hire full time employees to manage that paper.”

A Hybrid Situation

How much paper? Well, it could amount to as many as 9 million paper case report forms, since there are 200 pages per subject. The sponsor had a novel idea for the monitors on the trial: six traditional monitors were augmented by the sponsor’s sales representatives.

Rho elected not to continue participating in the project. Fowlkes’ advice: “You really have to start over and take a look at your SOPs. The processes you normally have in place may not cut it.”

d9A2t49mkex