by The Pandemic Century- One Hundred Years of Panic, Hysteria
The first thing epidemiologists are taught in the event of an outbreak is to draw up a working case definition in order to verify the diagnosis. The second is to look at the frequency of exposure among people with illness and those of comparable groups who are not ill (so-called controls). Only then is it possible to say whether the identified cases constitute an epidemic. As Fraser left for Harrisburg on August 3, he knew there were one hundred suspected cases and that there had been nineteen deaths. He also knew that all the cases involved Legionnaires who had attended the state convention in Philadelphia. However, that could merely be an artifact of the reporting: the American Legion was a close-knit group with efficient communication networks, so it was only natural that cases occurring among Legionnaires would come to attention first. Moreover, the press was already showing a keen interest in the outbreak, something that was likely to further skew the reporting. To know if there really was an epidemic underway in Pennsylvania, Fraser would need to establish whether any other groups or individuals had also fallen ill with pneumonia in the relevant time period and whether they had also been in Philadelphia or somewhere else. He would also need to establish how many Legionnaires and their families had attended the convention so as to obtain an accurate denominator with which to gauge the attack rate. Ideally, he would also need line lists detailing the name, age, and address of each patient, and, in the case of Legionnaires, the dates they had attended the convention and the hotels they had stayed at. These charts would also need to include key medical and pathological information, such as the date of onset of illness and, in the case of the deceased, the cause of death. Clearly it was a big job, and before it had ended thirty EIS officers had fanned out across the state, interviewing the families of victims or, in the case of those who had been hospitalized, the institutions where they had received treatment. In anticipation of this effort, on August 2 the CDC had dispatched Craven and Graitcer to Pittsburgh and Philadelphia respectively, and a recently qualified EIS officer, Theodore Tsai, to Harrisburg. In addition, Fraser would be joined in Harrisburg by two other newly qualified EIS officers, David Heymann, a future director of Emerging and Communicable Diseases at the World Health Organization, and Stephen Thacker, who would go on to become an assistant surgeon general in the US Public Health Service.
The other priority was to establish whether or not the outbreak was due to swine flu. This task would largely fall to Graitcer, whose job it was to liaise with state laboratories and forward throat washings and sera to the CDC’s laboratory in Atlanta. There, a team of specialists was on hand to see if the sera cross-reacted with the H1N1 swine flu, dubbed A/New Jersey/76, that had been isolated at Fort Dix in February. At the same time, CDC technicians would test for antigens to the most prevalent strain of flu then circulating in the northern hemisphere, an H3N2 virus known as A/Victoria/ 75, as well as other common infectious agents associated with pneumonia.
Within forty-eight hours of arriving in Harrisburg, Fraser had the answer to the first question: it was not swine flu. And within seventy-two hours technicians confirmed it was not the A/New Jersey or A/Victoria strains either. That left several other possibilities. At the top of the list was Chlamydia psittaci, the bacterium of psittacosis, and Coxiella burnetii, the bacterium of Q fever, a disease of cattle, sheep, and goats which was also known to cause pneumonia in humans. Another more remote possibility was Histoplasma, a fungal infection transmitted by birds and bats. Testing for these pathogens, Fraser knew, would take weeks and possibly months, and would have to be combined with the calm and careful collection of other evidence, such as dust and water samples from the Bellevue, and the examination of pathology specimens from deceased Legionnaires. But arriving at the offices of Leonard Bachmann, the secretary of health for Pennsylvania, and his chief epidemiologist, William Parkin, Fraser found the atmosphere anything but calm. Already, the phones were ringing off the hook with panicked callers, while in the press room next door newspapermen were demanding to know whether the outbreak might be something more sinister, a deliberate act of poisoning perhaps done by antiwar radicals intent on sabotaging the Bicentennial celebrations or sending a message to Gerald Ford, who two years earlier had controversially pardoned Richard Nixon for his alleged crimes in connection with the Watergate break-in. The press could be forgiven for asking such questions; in the run-up to the convention, Philadelphia’s mayor, Frank Rizzo, a tough-talking former policeman and close friend of Nixon’s, had deliberately stoked fears of a terrorist attack by posting undercover officers in and around the downtown area. Following the outbreak, Rizzo’s official spokesman, Albert Gaudiosi, had raised even more bizarre conspiracy theories, including the possibility of a covert operation by the CIA using chemical and biological weapons. Gaudiosi’s statement struck many as a blatant attempt to divert attention from the mayor’s failure to resolve a long-running garbage collection dispute—a dispute that had gone on for three weeks and had seen mounds of refuse collect on city streets. Those garbage mounds were a magnet for rats and other vermin. Might those rats be infested with plague fleas? wondered journalists. Could plague be the source of the Legionnaires’ peculiar pneumonic symptoms?
While CDC scientists were testing sputa and examining lung tissue and other pathology specimens, EIS officers were extending their investigations across Pennsylvania. Each officer drove an average of 450 miles, interviewing ten patients in over six hospitals. By now, a clear clinical picture was emerging. Typically, a case of Legionnaires’ disease began with a feeling of malaise, muscle aches, and a slight headache. Within twenty-four hours, patients would exhibit a rapidly rising fever, chills, and a dry cough, as well as, on occasion, abdominal pains and gastrointestinal symptoms. Two or three days later, the patients would have a raging fever of 102–105°F, and a chest X-ray would show patchy pneumonia. Accordingly, a case was defined clinically as any person with a cough or fever of 102°F or higher, or any fever and chest X-ray evidence of pneumonia. In addition, investigators included an epidemiological criterion (a case must have attended the American Legion convention or been inside the Bellevue-Stratford between July 1 and August 18). At this point, the cases listed at the State Department of Health consisted entirely of people who had attended the convention or who had been at the Bellevue, so this clinico-epidemiological definition made sense. However, it was also possible that the line lists had been skewed by the publicity surrounding the outbreak at the Bellevue and that people had not thought to report other cases that might warrant inclusion, so the Department of Health also set up a hotline and invited members of the public to report possible epidemic cases regardless of association with the convention or the Bellevue.
By the first week of August it was clear that the epidemic had peaked and the disease was not contagious, there being no secondary cases. Tracing the epidemic curve back in time, it was evident there had been a rapid upswing in cases from July 22 to 25, followed by a plateau through July 28 and a somewhat slow decline through August 3. Moreover, there had been no cases prior to the convention, suggesting that, whatever the agent, the incubation period was two to ten days. In all, in a four-week period up to August 10, there had been 182 cases and 29 deaths, giving a case fatality rate of 16 percent. The infection had proved particularly dangerous to cigarette smokers and older age groups, with those aged sixty or more twice as likely to suffer fatal outcomes. Almost all were Legionnaires who had either resided at the Bellevue or had attended events in its lobby and hospitality suites. However, there were also a few clinically compatible cases among non-Legionnaires. These included a Bellevue air conditioner repair man, a bus driver, and several pedestrians who had merely passed by the hotel’s imposing frontage on Broad Street. Were these Broad Street pneumonias part of the same epidemiological event? And why was it that, with the exception of the air conditioner repair man, hardly any of the Bellevue’s employees appeared to have fallen ill?
Though epidemiology aspires to be an exact science, it also contains a large element of induction. As Wa
de Hampton Frost, a former professor of epidemiology at Johns Hopkins and one of the pioneers of the field, once put it: “Epidemiology at any given time is something more than the total of its established facts. It includes their orderly arrangement into chains of inference which extend beyond the bounds of direct observation.” In other words, the raw data can only be parsed so far. To get a feel for Legionnaires’ disease, Fraser realized he needed to go to the focus of the outbreak. Obtaining rooms at the Bellevue was not a problem: by now, most guests had canceled their bookings for fear of contracting the disease, and on August 10 Fraser and ten of his officers moved into the hotel and began exploring the lobby area and hospitality suites. Was there a pattern, he wondered, some sort of clue in the way that the Legionnaires had used the hotel’s facilities?
To verify how many of the 10,000 registered Pennsylvania Legion members had actually attended the convention and to reconstruct their movements, Fraser distributed questionnaires to Legionnaires across the state. As well as confirming their attendance in Philadelphia, Legionnaires were asked to provide details of which hotel they had stayed at, and how many hours they had spent inside the Bellevue or on the sidewalk outside. The two-page questionnaires also contained checklists about key convention activities and functions. Had they gone to the Keystone Go-Getter Club Breakfast on the morning of July 23 in the Rose Garden on the eighteenth floor of the Bellevue? Had they attended the ticket-only Commander’s Bicentennial Ball in the Bellevue’s lavish second floor ballroom that same evening? Fraser also quizzed them about their consumption of food, coffee, and alcohol, and whether they had added ice and mixers to their drinks or bought anything from street vendors during the Legion parade through downtown. In addition, officers interviewed other guests and nonconventioneers who had stayed at or visited the Bellevue during the same period. Finally, EIS officers interviewed Bellevue staff to establish whether they had suffered any illnesses. To help with the surveys, Rizzo even provided Fraser and Sharrar with a team of homicide detectives. According to Sharrar, the detectives “did not miss a trick” and proved particularly adept at quizzing Legionnaires about their interactions with female sex workers, many of whom had passed themselves off as hotel guests in order to gain access to the hospitality suites.
It quickly became apparent that nearly everyone had spent time in the first floor lobby area, where the registration desk was set up, talking to other delegates running for election or chatting to family and friends. And nearly everyone had ridden in the elevators, either to visit the rooftop restaurant or to visit the bars and hospitality suites. Typical cases included Jimmy Dolan and John Bryant Ralph—“J.D” and “J.B.” in the anonymous line lists. Members of the Williamstown Legion post, Dolan and Ralph were thirty-nine and forty-one respectively and had been buddies since childhood. To save money, the pair had stayed at the Holiday Inn in midtown with Jimmy’s cousin, Richard Dolan, the forty-three-year-old commander of Pennsylvania Legion Post 239. Well built and with a reputation for partying, all three had attended the Commander’s Bicentennial ball, drinking until well past midnight. The trio had also spent many hours in the lobby area, but had avoided the hotel’s bars and restaurants. Within days of returning to Williamstown, both Jimmy Dolan and Ralph were complaining of fever, headaches, and coughs, and on July 29 Jimmy Dolan was admitted to the hospital. He died three days later, the pathologist recording the cause as “bilateral consolidation lungs, bloody sputum terminal.” The day after, August 2, Ralph also succumbed to the mysterious disease, the cause of death being listed as “gross massive bilateral lobar pneumonia.” By contrast, Richard Dolan had suffered no symptoms of illness.
Three “statistically significant” factors emerged from the questionnaires. First, ill delegates had spent on average four or five more hours in the Bellevue than had healthy delegates, and considerably more time in the lobby than controls. The correlation between the amount of time spent in the lobby and illness applied particularly to those who had slept at the Bellevue, but also held for Legionnaires who had stayed at other hotels. However, this correlation did not hold for hotel staff who worked in the lobby area and had spent as much if not more time there than Legionnaires. Indeed, with the exception of an air conditioning repair man, who had developed flu-like symptoms on July 21 and had returned to work four days later, there was no evidence of illness or disease in any of the hotel’s thirty full-time employees. Second, there appeared to be a small correlation between illness and visits to hospitality suites, with delegate cases visiting on average 2.6 hospitality rooms as compared to 1.8 visited by nondelegate cases. However, no one hospitality room had been visited by more than one-third of cases. Third, while cases were more likely to have drunk water at the Bellevue than noncases, only two-thirds admitted to drinking water in any form, presumably because they preferred to quench their thirst with alcohol and/or carbonated drinks. In short, as Sharrar put it, the typical case “was most likely to be a friendly, thirsty, elderly, male delegate who hung around the hotel lobby.”
In any outbreak investigation, once the existence of the epidemic has been confirmed and the diagnosis established, the next questions are who, where, when, how, and what? Following the surveys, there could be little doubt that Legionnaires were the who, the convention was the when, and the Bellevue was the where. But that left the how and what wide open. Had Legionnaires’ disease been triggered by exposure to a fomite, such as dust or ash particles, or was it due to some kind of gas? Alternatively, could the pathogen have been water- or food-borne? Moreover, if the common denominator was the Bellevue, how did one explain the apparent immunity of the hotel’s staff? Was it possible that the conspiracy theories were right and the outbreak had been a deliberate act of espionage?
By now, speculation was rife, with several newspapers suggesting the Legionnaires had been poisoned with paraquat, a weed killer known to cause pulmonary edema and breathing problems. Another suggestion was phosgene gas, a pulmonary agent that had been deployed by the Germans, and later the Allies, in World War I, which causes choking and shortness of breath. Other suggestions were that the Legionnaires’ symptoms could be due to poisoning with nickel carbonyl, a highly toxic liquid that can trigger chemical pneumonitis and cardiorespiratory failure, or else cadmium poisoning from the cadmium pitchers that the bar staff had used to mix the Legionnaires’ drinks. Fraser asked CDC technicians to screen pathology specimens from the deceased Legionnaires for traces of these toxins and poisons and instructed EIS officers to examine the restaurants, bars, rooms, and hospitality suites for traces of the same chemicals. If the pathogen had been phosgene, Fraser reasoned, it could have been added to the Legionnaires’ drinks or sucked in gaseous form via the elevator shaft, from where the constant motion of the elevators would have distributed it to the upper floors of the hotel. That could be why the survey had turned up no association between the illness and Legionnaires’ presence in a particular hospitality suite. However, everyone had ridden the elevators and had gotten in and out at the lobby. Phosgene is also rapidly excreted from the body, making it an ideal poisoning agent. However, it usually causes severe kidney damage, and none of the kidneys from Legionnaires exhibited signs of trauma. Nor did any of the specimens contain paraquat. By contrast, traces of nickel were found in the lungs, liver, and kidneys of six Legionnaires and two of the Broad Street pneumonias. However, these were well within normal levels and were not elevated compared to those of controls.
As the obvious candidates were excluded, Fraser began to consider more remote possibilities, including the air conditioning system. Most modern hotels boast rooftop chiller units, as cold air settles and it is impossible to drive cold air upwards. However, the Bellevue employed an old cold water system operated via two Carrier refrigeration machines located in the subbasement. Installed in 1954, these chillers had a capacity of 800 and 600 tons respectively and used Freon 11 refrigerant to cool the water. This chilled water was then pumped up to the roof of the hotel from where it was circulated downward to
some sixty air-handling units (AHUs). Most of these used approximately 75 percent recirculated air and 25 percent outside air, but in the case of the AHU located directly above the lobby desk, all of the air was recirculated.
At the same time, a separate system, using “cooled” water from a cooling tower on the roof, was employed to condense the refrigerant. In the event of accidental leakage, the chilled water system was designed to be replenished automatically via a float valve in a nearby water expansion tank. Unfortunately, due to a fault in the valve, the water pipes at the top of the hotel had become filled with air, resulting in the failure of one of the AHUs serving the Rose Garden restaurant on the eighteenth floor. To rectify this, staff had hooked up a temporary connection using a garden hose that ran from the water tower to a pipe leading to the AHU. This makeshift system solved the problem of the faulty float in the expansion tank, but if the valves at either end of the hose were left open or leaked and various safety valves malfunctioned, it was conceivable that water from the water tower could have found its way into two steel tanks, also located in the roof, that supplied the hotel’s drinking water. Since the water in the tower had been treated with chromate to preserve the pipes, this made it a potential contamination risk. The water tower was also uncovered and exposed to the elements, meaning it would be very easy for droppings from pigeons roosting on the balconies to get into the potable water.
Another potentially worse hazard was the 800-ton basement chiller unit. The unit had been leaking F-11 coolant continuously since May, prompting the Bellevue’s management to put in repeated calls to the Carrier company to fix the problem. However, these repairs had been only partly successful, and with the summer conference season imminent, management had opted to postpone further servicing until later in the year. Unfortunately, air from the subbasement exhausted directly onto Chancellor Street on the southern side of the hotel. In theory the exhausted air could have contained F-11 coolant from the faulty chiller in a gaseous state. In addition, piped vents from the chillers also discharged air onto Chancellor Street just three feet away from the exhaust fan, meaning it was possible that some of this air could have been sucked back into the subbasement via an air shaft adjacent to the point of exhaust. Fraser was unable to determine the “ultimate fate of this air,” but as the subbasement was also served by two large fans that exhausted air via another shaft that extended up to the roof, he could not discount the possibility that contaminated air had been circulated throughout the hotel. Fueling Fraser’s suspicions about leaking chiller coolant was the fact that an air conditioner repair man had signed off sick on July 20, the day before the convention opened. As the man reported having a cough and a temperature of 102°F, his name had been included in the line list. However, he did not develop pneumonia and on July 24 was well enough to return to work. Later, it was discovered that his wife and two daughters had been sick with a respiratory illness at the same time, prompting Sharrar to argue that he should never have been included in the line list and that his illness was probably due to flu, not Legionnaires’ disease.
