BLAKELY, Ga. -- When Food and Drug Administration inspectors visited Peanut Corp. of America's plant here in late 2001, they noticed peanut-processing equipment had been improperly repaired with duct or cellophane tape.
The "widespread" use of tape — some torn — concerned the inspectors because it could harbor insects, was hard to sanitize and could lead to adulterated food.
PCA President Stewart Parnell, who had taken over the plant in February, said it was a good thing they hadn't come a couple months earlier, because they would have seen even more tape then, according to the inspectors' report. Parnell promised to fix the equipment.
But last year — months before a salmonella outbreak was linked to PCA's products from the plant — a private audit found tape still used on equipment and to cover wall seams.
Details matter in food safety, and the story of how PCA came to be held responsible for one of the nation's largest and most costly salmonella outbreaks is all about details — lots of them — unseen or unreported or not acted upon until it was too late.
Federal authorities have begun a criminal investigation of PCA, and the company is bankrupt. Records produced in the FDA's investigation of PCA and in congressional hearings on the outbreak portray a company that not only failed to heed warnings about its deficiencies, but allegedly shipped products that had tested positive for salmonella after retests were negative.
More important, the case reveals a food-safety system in which every key link in the chain of protection failed, food-safety officials and lawmakers say. The outbreak "is a poster child for everything that went wrong" with the USA's food-safety system, says William Hubbard, a former FDA associate commissioner. "Down the line, you can find flaws and failures."
The U.S. food-safety net relies heavily on companies to be good operators. Yet PCA repeatedly failed to fix problems that were brought to its attention, according to regulatory records and documents made public in congressional hearings. Nestlé, for example, twice inspected PCA plants and chose not to take on PCA as a supplier because it didn't meet Nestlé's food-safety standards, according to Nestlé's audit reports in 2002 and 2006.
Regulators never found anything major wrong with PCA's Blakely plant until after the outbreak. Then, the FDA found major problems in sanitation, manufacturing and even plant design.
Unlike Nestlé, other PCA customers, including Kellogg, never audited the Blakely plant themselves. Instead, they selected PCA as a supplier based in part on an inspection by an auditing firm that was paid by PCA and that rates almost every client "excellent" or "superior," said Rep. Bart Stupak, D-Mich., citing his committee's investigation.
The outbreak resulted in 700 reported illnesses and may have contributed to nine deaths. More than 3,600 products were recalled, costing the food industry hundreds of millions of dollars and signaling to parents that many of their children's favorites — peanut crackers, cookies, ice cream — could be dangerous.
The outbreak also caused heartbreak. One family alleges it stole a mother and a Christmas from them. Shirley Almer, a 72-year-old cancer survivor, died Dec. 21 after eating salmonella-tainted PCA peanut butter in an elder-care facility, her son Jeffrey testified to Congress.
"Shirley Almer loved this country but was terribly let down by a broken and ineffective food system," he said at the congressional hearing.
The industry watchers
PCA got more than one warning from other companies that its Blakely plant had problems.
•Deibel Labs, which ran more than 1,600 salmonella tests for PCA's Blakely plant from 2004 through 2008, found almost 6% positive. It was so many that Deibel sent PCA's samples to a separate part of its Chicago lab to lessen chances that they'd contaminate other products, Charles Deibel, the firm's president, said in an interview.
For roasted products such as peanuts, a positive rate above 1 in 10,000 would be high, Deibel said. Proper roasting kills salmonella with heat. PCA never asked Deibel to look into the issue, Deibel said.
•Another lab hired by PCA, JLA, based in Georgia, told PCA in 2006 that the Blakely plant hadn't adequately documented that its roasting killed salmonella, according to a letter from JLA to PCA that congressional investigators released. After the outbreak, the FDA noted the same deficiency in its 2009 report.
•Nestlé audited the Blakely plant in 2002 and rejected it as a supplier. Nestlé's audit report said the plant needed a "better understanding of the concept of deep cleaning" and failed to adequately separate unroasted raw peanuts from roasted ones. Having them in the same area could allow bacteria on raw nuts to contaminate roasted ones, a risk known as cross-contamination.
The plant wasn't even close to Nestlé's standards, auditor Richard Hutson said in an interview. Hutson, who now heads quality assurance for several Nestlé divisions, said he shared his concerns with PCA officials at the time, but "they didn't pursue it" further with Nestlé, he says.
After the outbreak, the FDA found problems at the Blakely plant that were similar to those found by Nestlé, including inadequate cleaning and storing of raw and roasted peanuts too close together.
Nestlé also rejected PCA's Texas plant in 2006.
Neither Deibel, JLA nor Nestlé shared their findings with anyone other than PCA, which is common industry practice. Congressional lawmakers don't fault companies for not sharing proprietary data, but some now say that foodmakers' microbiological test results should be reported to regulators.
Had the FDA seen PCA's salmonella test results, it might have detected a problem sooner, said Stephen Sundlof, the FDA's head of its food-safety center, at a congressional hearing in February.
That a lab detected salmonella in PCA products but reported it only to PCA is a practice that "we can't afford to have in our food-safety system," said Rep. Bruce Braley, D-Iowa, at the same hearing.
The issues noted by Nestlé and JLA, and the frequent salmonella positives found by Deibel, went undetected by regulators. That's part of what Stupak calls a "total systemic breakdown" of the U.S. food-safety system.
The FDA didn't inspect the Blakely plant itself, after its 2001 check, until the outbreak. That's not unusual. The agency's inspection staff is so strapped, it inspects food facilities an average of once every five to 10 years unless they're deemed high risk, which peanut processors were not.
About half the FDA's food inspections are done by state inspectors, whose departments are paid by the FDA to do that work.
After the outbreak, and a 13-day inspection of the Blakely plant in January, the FDA delivered a scathing report. It said the plant didn't clean up after finding salmonella, had poor controls to prevent contamination and had poor design to prevent roof leaks.
Most important, the FDA discovered that PCA shipped products that had tested positive for salmonella, then negative on a retest. Shipping such product is "universally condemned," the FDA's Sundlof testified to Congress, because salmonella can be missed in tests. Products should be destroyed after one positive result, regulators say.
But nine inspections of the Blakely plant — by Georgia agricultural inspectors in 2006, 2007 and 2008 — found only minor issues, many of which were quickly fixed, said Oscar Garrison, Georgia assistant Agriculture commissioner. Two of the checks were done for the FDA.
Garrison defends the state inspections as a "snapshot in time." Even rigorous inspections wouldn't always detect problems if a processor is intent on "breaking the law," he said.
But Stupak, who held two hearings on the outbreak, said the FDA's 2009 inspection report notes numerous violations of good manufacturing practices that weren't found by the state and which FDA officials later testified should have been caught.
The PCA case has cast a spotlight on the rigor of state inspections done for the FDA. Some states do a good job; some don't, Hubbard said. The FDA knows it needs to raise standards, said Michael Taylor, food-safety expert at George Washington University.
"There's a basic breakdown when an FDA-contracted inspection doesn't detect problems that seem so obvious," Taylor said.
PCA, which closed its three plants after the outbreak, has disputed some of the FDA's assertions. Parnell, who shares PCA's ownership with an investor group, worked at its Virginia headquarters. He and other former PCA managers refused to comment.
The plant in Blakely used to employ 50 but now sits deserted. Its paint is faded and chipped, as if a symbol of the deficiencies the FDA said were inside.
When the outbreak hit, PCA supplied 2.5% of the nation's peanut products, including peanut butter sold to institutions and paste and meal used in foods made by hundreds of companies.
To win customers, Parnell "extolled" the fact that an auditor, AIB International, had rated the plant as "superior," said King Nut CEO Martin Kanan at a congressional hearing. King Nut sold peanut butter under its name that was made by PCA.
That rating also satisfied Kellogg, which began buying PCA's peanut paste for sandwich crackers in 2007. Kellogg CEO David Mackay testified at a congressional hearing in March that PCA was a "dishonest supplier" and that Kellogg had done "everything we could" to ensure safety.
PCA had been audited by AIB, "the most commonly used auditor in the U.S.," Mackay said. PCA had verified that it had fixed issues raised in the first audit in 2006, Kellogg says. Kellogg visited the plant but didn't audit. Kellogg also got certificates from PCA — issued by private labs paid by PCA — saying the product was salmonella-free, Kellogg says.
But AIB's rating of PCA has since come under attack, along with the common practice of foodmakers paying for their own audits.
Stupak said congressional investigators found that AIB gives 98% of companies a "superior" or "excellent" rating. He also said that e-mails between AIB and PCA point to a relationship that's too cozy to ensure a tough audit.
"You lucky guy. I am your AIB auditor," AIB's Eugene Hatfield wrote PCA on Dec. 22, says an e-mail released by Stupak's committee.
In 2008, PCA had more than a month's warning before its AIB audit. Former PCA employees, sanitation director Anne Bristow and Bobby Mallard, said in interviews that the plant was deep-cleaned beforehand.
"Five days later, it would be back to normal," said Mallard, who ran a peanut-roasting line. "It was dirty."
The last AIB audit, done on one day in March 2008, found few problems. "Excellent cooperation was received by the writer," wrote Hatfield in the report. "On some occasions, the items were immediately corrected."
AIB refused an interview request but defends its audits on its website. It says Hatfield had inspected 200 peanut facilities in his career and did a PCA check that was so detailed he found beetles behind duct tape.
AIB also says the Blakely plant ran for months without a manager in mid-2008, providing ample time for it to deteriorate between AIB's audit and the FDA's January 2009 inspection.
AIB also draws criticism from a former food-industry official. Its audit of PCA was "superficial," said Jim Lugg, former food-safety chief for bagged salad maker Fresh Express, who reviewed AIB's audit of PCA at USA TODAY's request.
One example of "shallow treatment of a big issue," Lugg says, is that the audit notes that PCA had a written program to evaluate suppliers and had an approved list. But AIB did no further checking of the suppliers. Years ago, Fresh Express stopped using AIB audits because it found them inadequate, he adds.
Lugg also questions why another audit firm ranked the PCA plant so high even though the auditor noted many problems.
In April 2008, NSF Cook & Thurber inspected the Blakely plant for a client, which it says wasn't PCA. The audit found so many "minor" deficiencies at the plant — including use of tape — that the plant ranked in the bottom 6% of audits done by NSF last year, NSF said in a statement, adding that it stood "100% behind" the audit.
Still, NSF gave the plant an "opportunity for improvement" rating on food safety and quality, just below the "acceptable-excellent" rating. Lugg says that rating appears too high, given the concerns noted in the audit, including criticisms of the plant's condition, sanitation and pest control.
"The whole idea (of third-party audits) isn't working," says former FDA official Hubbard. "The inspectors are either telling the client what they want to hear, or they're doing a perfunctory audit, or they're poorly trained."
Kellogg, while defending its oversight of PCA, now says it will do its own inspections of high-risk suppliers. It spent less than $20 million on PCA products. Its cracker recall will cost up to $70 million, Mackay testified.
Many companies need to do more due diligence on suppliers, food-safety experts say. "There needs to be a revolution in the supply chain," says Michael Doyle, director of the University of Georgia Center for Food Safety.
The end result
Since the recall, Parnell has been portrayed by congressional lawmakers as a man most concerned with getting product out the door.
Former employees also say too little was spent on the Blakely plant. "It was production, production, production," says Mallard. "Then clean for 15 minutes."
"I'd tell Stewart that this needs to be changed right away," Bristow says. "He'd say, 'We'll get on it.' It wasn't done."
The plant's roof leaked so badly, "It rained in the plant," says Teresa Spencer. Rainwater can carry salmonella from bird droppings. The roof leaked even after PCA fixed it, Mallard says.
PCA also left key jobs open. In addition to losing its plant manager in 2008, it lacked a quality manager for at least four months, NSF's audit says.
Parnell's side remains untold. At a congressional hearing in February, he invoked his constitutional right not to testify. His lawyer also refused comment for this story, citing the criminal probe, as did Parnell's daughter, who did PCA's books.
Jeffrey Almer and his sisters sat behind Parnell in the hearing room.
Their mother, Shirley, a woman who dragged her grown sons onto the dance floor, had entered a Minnesota rest home after Thanksgiving to recover from a urinary tract infection. The day before her expected release, her family was told she had hours to live. Her mother had lived to be 101.
When Parnell refused to testify, Jeffrey, 46, a finance employee for Best Buy, says he felt rage. It was directed at Parnell but also at the food-safety system that he says failed his mother.
"My mom should be here today," he testified.
Contributing: Tom Ankner