Processor violations
By Brent Johnson NICAR
Hundreds of meat and poultry plants operating throughout the country are repeatedly issued citations following inspections by the U.S. Department of Agriculture. But the enforcement record against these gutting operations has most often been gutless. An investigative story by Elliot Jaspin and Scott Montgomery of Cox News Service found that few plants ones cited hundreds of times - are ever forced shut by the USDA.
Critical condition
Jaspin and Montgomery used FoxPro to analyze the massive "Feedback File" database provided by the USDA following a FOIA request. Although they requested files from 1990-1996, the USDA only released information from 1996. The reporters wanted to know how many "critical violations" - defined by the USDA as a plant condition "certain" to cause both food contamination and to reach and then harm consumers - occurred at given plants. "We were expecting 200 tops for a particular plant," Jaspin said. However, they found a poultry plant in Arkansas operated by Tyson Foods that had amassed 1,700 violations. That's when they knew they had a story.
File me up, file me down
The USDA database arrived on CD-ROM organized by a series of subdivisions. The data were divided into quarters of the year. Each quarter was divided into 26 regions. Each region was divided by week. "So there were 13 zip files for each region, and within those, when unzipped, were different 'circuits,' or strings of meat and poultry plants," Jaspin said. "We're talking in the neighborhood of 5,000-6,000 little databases."
"When you have large databases, the whole idea is to divide and conquer, to be able to work in a feasible fashion," Jaspin said.
Transferring the data from the CD-ROM to the hard drive using a batch file presented the first challenge. Then appending the zip files into databases for each quarter awaited. "We had to smash (the unzipped files) into one large database actually four. One database would've been too unwieldy - over 25 million records."
Divide and conquer
Dirty data always lurks. Jaspin and Montgomery had to work around an identification problem. They thought each plant would have a unique number but then found individual plants entered multiple times. By cleaning the unique number problem up, they could link violations from inspection records to particular plants by joining tables.
Finally ready to analyze the data, they created "subset databases" to answer particular questions, such as the frequency of critical violations occurring at plants.
"When you have large databases, the whole idea is to divide and conquer, to be able to work in a feasible fashion," Jaspin said.
Jaspin and Montgomery also analyzed the data by geographic distribution of violations and plant size. Smaller databases were created and sent to individual Cox newspapers so they could pursue local stories.
USDA puts up, but doesn't shut down
Through their database analysis of 1996 inspections, the reporters discovered 299 plants with 100 or more "critical violations" and 7 with more than 1,000. That year, the USDA shut down only 6 plants.
"Heavens," an epidemiologist at the Centers for Disease Control and Prevention told the reporters upon hearing those numbers. "The burning question is why is a plant with that many violations - why has the (federal) inspector not been pulled from that plant? And I have no idea."
NICAR IS IN THE PROCESS OF ACQUIRING THE SUMMARY DATA OF THE USDA's PROCESS DEFICIENCY REPORTS, WHICH INCLUDE A LIST OF VIOLATIONS - CRITICAL, MAJOR OR MINOR - BY PLANT.
Brent Johnson can be reached at (573) 882-0684 or by e-mail at bjohnson@nicar.org
A PROFILE OF PAUL ADRIAN, AN INVESTIGATIVE REPORTER AT WBNS-TV IN COLUMBUS, OHIO, WHO HAS ALSO USED THE SUMMARY DATA IN A STORY, WILL APPEAR IN NEXT MONTH'S ISSUE.