Experimenting with Drugs — Uplink Archive

Plain Dealer reporters investigated FDA inspection records spanning two decades and found that more than half of drug experiments lacked adequate informed consent from human subjects, with some cases showing no consent at all. The reporters obtained and analyzed a computerized FDA database of over 8,000 inspection records to uncover patterns of inadequate disclosure of risks and benefits to research participants.

When the Food and Drug Administration inspects drug experiments, it finds more than half the time that the human subjects have not been sufficiently informed about their possible risks and benefits. That's the conclusion in articles by Plain Dealer reporters Bill Sloat and Keith Epstein. The series ran over four days beginning Dec. 15. The project began more than a year ago after Sloat and Epstein finished stories about Norplant and how some women did not give written consent when Norplant was tested on them. Managing Editor Gary Clark suggested that the two try to find out if there were other instances where the FDA found inadequate consent.

The two reporters obtained more than 5,000 pages of federal documents along with a computer database containing the FDA inspections from 1974 through 1995 of clinical research involving experimental drugs being tested on humans.

The reporters found the database almost by accident. They had filed many FOIA requests for FDA inspection reports, but didn't get any on the grounds there were none. They started calling employees from the FDA and someone told them to ask for the 483s, which was the actual inspection report. Armed with that information, they got a printout of some 483s.

But then someone mailed them a huge printout of the actual database. With that, they knew for sure the database existed on a computer.

It took another three months, but Epstein, who works in The Plain Dealer's Washington bureau, finally got the records on floppy disks.

That database provided numbers that were the basis for much of the series as well as most of the information for the graphics that ran with the project.

The database contained 8,039 records. There was no single field for each inspection so we used FoxPro 2.6 for Windows to calculate that the FDA conducted 4,154 inspections of medical researchers.

The database had a column for type of deficiency, and we found that in more than half the inspections - 2,201 - the consent was inadequate.

The database even revealed 47 instances where there was no consent.

One of the major facts Sloat and Epstein wanted to know was how federal research locations compared with those at universities, hospitals or private facilities such as clinics.

But the database had no classification indicating what type of facility was being inspected.

In addition, it had the usual pattern of using different spellings or words to describe the same institution.

So Sloat and I spent a week in early 1996 going through the entire database (we were able to throw out some records that did not have a location).

We added a new column and gave each location a code letter such as G for government or U for university. Private facilities or those we could not identify went into the same category, which we called other. But we were able to identify most of them from the name.

Once that new database was completed, we were able to go far beyond the information available at first.

We found out very quickly that the government locations, such as Veteran's Administration hospitals, had the worst track record for providing information to the people used in experiments.

Because we now had a unique number for each location, we were able to pinpoint the government locations and universities that had the most violations over the years. All these facts were used in graphics that ran every day.

To make the story interesting, Sloat and Epstein had to find the names of people who were part of studies. The task was painstaking because identities of patients were omitted or deleted from many records. In other cases, the government used code names.

Sloat and Epstein traveled extensively throughout the country and even to Europe tracking experiments inspected by the FDA.

Sloat and Epstein even found people who died during a drug experiment, including a Pennsylvania woman who received a substance also used as a gasoline additive to try to dissolve gallstones.

That woman was only known in records as TE3. Sloat and Epstein were able to match public death records with FDA documents to find the clues that enabled them to identify the woman as Laura Michalski.

They talked to Michalski's widower and children who were unaware that Michalski had been part of a drug experiment.

Tom Gaumer can be reached at (216) 999-4500, or send e-mail to TGAUMER@plaind.com