Contaminated green onions used in mild salsa are the apparent source of the nation's worst-ever hepatitis A outbreak, a public health crisis that has sickened more than 500 and killed three who ate at a Beaver County restaurant.

That's the conclusion of investigators who interviewed hundreds of people -- both sick and well -- who ate at the Beaver Valley Mall Chi-Chi's in early October. After learning what foods the sick people ate, investigators computed which ingredients were most commonly associated with developing hepatitis A.

The conclusion: Infected people were 20 times more likely to have eaten items with green onions than people who didn't get sick.

"Green onions are the likely source of the hepatitis A outbreak in Beaver County," said Dr. Calvin Johnson, state secretary of health, during a news conference at the Beaver County State Health Center in Beaver.

"We've answered the question of the food source, but we have not yet answered the question of how that food source became contaminated."

The results were published yesterday by the Centers for Disease Control and Prevention, and they fit with a pattern of green onion contamination that sickened restaurant patrons and workers earlier this fall in Tennessee, Georgia and North Carolina.

The Food and Drug Administration has traced back scallions consumed in two of the states to three Mexican companies. Green onions from those companies are currently being stopped at the border and held for examination.

While the source of Chi-Chi's green onions is not clear, the viral strain sickening people here is common in Mexico and very close to the strains that circulated in the Southern outbreaks. CDC investigators say all four strains could possibly be circulating in one farming community.

"This confirms our belief that our hygiene and food handling processes played no role in this outbreak and our restaurants are safe to eat in," said Bill Zavertnik, chief operating officer at Chi-Chi's.

Pinning the blame on green onions took investigators 20 days.

Legions of nurses with the state Department of Health have been interviewing hundreds of sick people to determine what they ate at Chi-Chi's and whether they could have been exposed to the virus by other means. Hepatitis A is not common in Pennsylvania, but is endemic in many countries around the world. Certain groups in this country -- drug users and men who have sex with men -- are at greater risk for hepatitis A.

Public health physicians then retrieved menus and recipes from Chi-Chi's to determine the ingredients of all foods served. Based on that information, they sent follow-up questions back to patients through the nurses -- trying to clarify whether a patron ate a particular side dish, for example. Investigators also interviewed people who ate at the restaurant but didn't get sick.

As of yesterday, 575 people have been sickened in the outbreak.

But the study published yesterday compared the experience of only a subset of those who got sick -- 183 people -- with that of 83 people who didn't. All ate at Chi-Chi's Oct. 3-6.

The result was that 98 percent of people who got sick ate a menu item containing green onions. Of those who didn't get sick, only 69 percent had eaten green onions.

Even more impressive was the fact that 94 percent of the sick consumed the mild salsa provided free with chips at every table, compared with only 39 percent of the well, said Joel Hersh, director of epidemiology at the state Department of Health.

It might seem confounding that some who ate green onions stayed well, Hersh acknowledged. But it's likely that the virus wasn't uniformly distributed in the condiment, so not every chip dipped came back with hepatitis virus.

Unlike Chi-Chi's hot salsa, which is prepared off-site, mild salsa at Chi-Chi's is made in the restaurant.

Green onions were shipped to the Beaver County restaurant in 81/2-pound boxes that contained multiple small bundles of scallions. Bundles were unpacked and stored upright in a bucket that was refrigerated for about five days before processing. Green onions would then be chopped by a machine to yield approximately eight quarts, which were then refrigerated for another two days.

Throughout storage, onions were packed in ice that could have allowed the water-borne virus to spread as it melted.

Every 1 to 3 days, salsas were prepared in 40- to 80-quart batches. The salsa would then be ladled into bowls and provided free with tortilla chips to patrons. That sequence fits with the concentration of sick people reported in the study -- the vast majority were exposed over just three days.

It's not clear to investigators exactly how many scallions were used per batch of salsa. But given how few virus particles a person must consume in order to get sick, it's possible that the outbreak could have started with even less than one box of contaminated scallions, Hersh said.

Hepatitis A is spread by the consuming virus that is shed in the fecal matter of an infected person. Green onions could become contaminated during a number of points, the investigators wrote.

Adult field workers in Mexico are unlikely to have hepatitis A themselves, but their children might. Workers caring for children in the field could inadvertently contaminate scallions with dirty hands, or the onions could be contaminated if an infected person defecated in the field itself.

Alternately, the virus could have spread through contaminated water used to irrigate, rinse, process, cool or ice the green onions.

Green onions might be especially vulnerable to contamination because they are multi-layered and difficult to clean. The virus also is unusually sticky and difficult to dislodge from produce.

While the overall number of hepatitis A cases increased by 35 yesterday -- the most sizable jump in a week -- state officials weren't expressing worry. Hersh, the epidemiologist, said some people who had very mild cases might now be getting blood tests that confirm their illness.

Among the 575 cases are 75 residents of other states including West Virginia, Ohio, South Carolina, Florida, Wisconsin and Wyoming.

"People travel," Hersh explained.

The epidemiology work isn't done. There are still some results that don't fit with the pattern -- one infected person reported eating at Chi-Chi's as early as Sept. 14 and another as late as Oct. 18, Hersh said.

"The power of the data tells us that the chance of this being a random occurrence or just being wrong is slim to none," he said.