Epidemiologic InvestigationAlthough CDC epidemiologists had generated a long list of possible causes of the epidemic before arriving in Haiti (including infectious agents, ingestible substances, and toxins), the initial phases of the investigation yielded some important clues. Pathology reports had initially suggested a toxic exposure, and the duration of the epidemic suggested an ongoing exposure rather than a single point source. The patients were young, which suggested to the researchers either a unique exposure of some kind (for example, bottles, pacifiers, or baby food), a disease manifestation unique to children, or a dose-response relationship that would render young children most susceptible. Moreover, the presence of a fever in almost all of the presenting cases suggested an infection or other predisposing illness for which a medication or herbal remedy might have been used.

O’Brien began by creating what epidemiologists refer to as a case definition. In the Haitian epidemic, a case was defined by a diagnosis of anuria (failing to urinate) for more than 24 hours for unknown reasons in a patient younger than 18 years old occurring after November 3, 1995. Locating cases meeting this definition was significantly more challenging.

The General Hospital had no admission or discharge diagnosis log books, and no systematic method of collecting or maintaining medical records. Researchers had to depend largely on physicians’ own recall of cases. The Ministry of Health did have records stemming from an ongoing investigation of meningococcal infections at the General Hospital that had picked up some cases of renal failure. Private practitioners in Haiti as well as health care workers in the community also identified cases fitting the case definition. Soon after O’Brien arrived, 84 confirmed and 24 possible cases had been identified. In looking over their identified cases, researchers discovered that despite the fact that almost all of the children had siblings younger than age 18, there was only one set of siblings with renal failure. This clue virtually ruled out an infectious agent as a cause of the epidemic, and the researchers began to con-Chronicles
centrate in earnest on commonly ingested substances taken by the affected children.

Identifying the Chemical Culprit

Once O’Brien and her colleagues in Haiti had identified the cases com¬prising the epidemic, they set up a classic case-control study to identify common elements or risk factors in the histories of case patients. Control subjects were comparable to case subjects in age and sex and were enrolled as they were admitted as inpatients to the General Hospital. These hospital control subjects had a history of fever but no history of anuria. The case-control study had its limitations: parents often didn’t recall all of the medications taken by their children, hospital charts were often absent or not detailed, and bottles had been thrown out or were unidentifiable because they had been cleaned and reused. Nonetheless, the study was large enough to give the researchers some measure of confidence, and unfortunately, more children with renal failure were entering the hospital every day.

Researchers painstakingly collected histories and bottles of all medications used by case and control subjects and sent them to CDC for analysis. O’Brien and her colleagues noted that among 453 bottles of medication that had been either mentioned or collected in their study, only two medications, both acetaminophen preparations manufactured by Pharval, were associated with illness in the case-control study. One was labeled Afebril and the other Valodon. All medications taken by case subjects had come from three consecutive lot numbers. The researchers sent out a driver to buy as much Afebril and Valodon as he could locate to see if there were other lot numbers on the market or if CI, C2, and C3 were the only ones available. When the driver returned with preparations from many other lot numbers, the researchers began to focus on C1,C2, and C3 as potentially contaminated.

The case-control study revealed that 55 case subjects had been exposed to either Valodon or Afebril, while only eight case subjects appeared to have no exposure. Among the controls, the situation was even more striking. Only five control subjects had been exposed to Afebril or Valodon, while 47 had no such exposure. Given such a high odds ratio, the researchers began to feel confident that they had discovered the culprit.

They immediately began looking at the common ingredients in Afebril and Valodon: acetaminophen, sodium cyclamate, potassium sorbate, glycerin, propylene glycol, citric aid, sodium citrate, red coloring, strawberry and raspberry flavoring, and water. Was there too high a dose of one of the ingredients? Was a toxin present? More likely was the possibility that a toxic substance had been substituted for one or more of the usual ingredients. At that point, however, all they could do was to speculate and wait for laboratory results. By June 21, exactly a week after O’Brien’s arrival in Haiti, the presence of approximately 15% DEG had been confirmed in samples of both Valodon and Afebril by the CDC lab in Atlanta. Concerned about the security of their communications, O’Brien’s supervisor in Atlanta merely confirmed over the phone that “it was what we thought it was.”

Once the CDC had confirmed the presence of DEG, O’Brien immediately called the head of the РАНО mission, Marie-Andree Diouf, MD. In what the participants describe as “a long night,” a succession of people

were notified concerning the DEG findings. Dr. Diouf came to the hotel where the epidemiological team was assembled. The head of the Caribbean health organization CARE С was notified, as was the American embassy. Because the team had concerns about how the news would break in Haiti, the Ambassador himself was rousted from sleep. In a late-night meeting hastily convened at the РАНО office, the Haitian Minister of Health, an advisor, and a fully armed bodyguard learned of the DEG contamination. Dr. Mallebranche tried to contact Rudy Boulos at Pharval but learned that both Reginald and Rudy were in Miami. Only the youngest brother, Franz, appeared at the meeting. When he was told that Afebril and Valodon had both tested positive for DEG, his initial response was a denial that Pharval even made these drugs. He then claimed that the products involved must be counterfeit.

The rest of the night was spent in drafting a statement for the Minister of Health and in devising a strategy for initiating a drug recall. Although there were discussions about a broad recall of all acetaminophen products or all Pharval products, in the end the recall was limited to the two Pharval acetaminophen products implicated by the laboratory findings.
At 7:00 a.m. on Saturday, June 22, Dr. Mallebranche briefed the President of Haiti. At 8:30 a.m., the Minister announced to a startled Haitian population that an “unintentional poisoning” had been discovered and that no one should take Pharval’s Afebril or Valodon—that they were, in essence, being outlawed. Police officers were instructed to go around the country and confiscate all available Afebril and Valodon. A broad media campaign in the newspapers, on radio, and on television featured frequent and repetitive spots; however, many poor Haitians had limited access to these media outlets. (Many of the spots, in addi¬tion, were broadcast in the US and other countries with large Haitian immigrant populations.) Flyers were sent home with Haitian schoolchildren, and notices were issued to Haiti’s medical societies while police moved with bullhorns through the streets. The Ministry of Health sent employees to visit all 10 departments (provinces) in Haiti to conduct spot checks of pharmacies. To help speed up the governments efforts, the FDA sent its drug specialist in San Juan, Puerto Rico, Jorge Guadalupe, to assist in the Haitian drug recall. The Haitian Minister of Health’s Saturday a.m. announcement was picked up by Reuters International, and on Monday, June 24, а РАНО press release about the epidemic was posted on the Internet.

The first week following issuance of the public warnings, seven Haitian children were admitted to the hospital with DEG poisoning, all of whom had taken either Afebril or Valodon prior to the public warning. After that first week, however, only three children were diagnosed with the syndrome, and no cases were reported after July 1996.

After telling reporters that he was “100% sure” that his plant was not at fault, Rudy Boulos allowed Selanikio to examine the firms quality control records and procedures. To his surprise, Selanikio learned from production records that Afebril and Valodon were identical, named separately for marketing and promotion purposes. Selanikio found that suspect Lot С1 had been produced on September 12, 1995, Lot C2 on November 6, 1995, and C3 on December 19, 1995. These dates were all consistent with the onset of the epidemic. For suspect lots CI and С2 the firm had kept quality control samples known as “retain” samples of the finished products, and Selanikio promptly shipped them to CDC for analysis along with samples from other products made around the same time. The retain samples proved to be contaminated with anywhere from 12% to 30% DEG. Samples were also collected from the other two Haitian pharmaceutical manufacturers.