Chapter 5. Disease Outbreaks

• A disease outbreak is an epidemic that occurs suddenly and within a relatively confined geographic area.
• The occurrence of a disease outbreak requires a pathogen in sufficient quantities, a mode of transmission, and a pool of susceptible persons.
• The two primary modes of transmission of pathogens in disease outbreaks are person-to-person spread and common source of exposure.
• Criteria for determining whether a disease outbreak should be investigated include, among others: the number and severity of persons affected, uncertainty about cause, and level of public concern.
• The attack rate is a measure of the number of persons affected by the disease outbreak among persons at risk.
• Food-borne pathogens can give rise to disease outbreaks through a common source exposure.
•  Emerging infectious disease refers to an infection that has newly appeared in a population, or has existed but is rapidly increasing in incidence or geographic spread.
• Certain emerging pathogens, such as severe acute respiratory syndrome (SARS)–related coronavirus, West Nile virus, and Ebola virus are responsible for highly publicized recent outbreaks of disease with high case-fatalities.
• In 2001, the threat of bioterrorism was realized when Bacillus anthracis was mailed to at least five locations, resulting in 22 cases of anthrax and five deaths.

In many outbreak situations, distinctive clinical features of the affected individuals may suggest the underlying cause (sometimes termed pathogen). A working hypothesis can lead to prompt identification of the causal agent and implementation of control measures. Ideally, the choice of control strategy is predicated on knowledge of the source of the causal agent and how it is spread.

In other circumstances, however, the clinical features of affected individuals do not suggest a particular pathogen. An urgent response is required, although the investigator does not yet have a specific working hypothesis about the cause. Consequently, the first phase of investigation involves the collection of basic descriptive information to better characterize the illness and its pattern of occurrence. With this background descriptive data in hand, investigators can generate hypotheses and design specific analytical studies to identify the causal factor of the illness.

• 1. The presence of a pathogen in sufficient quantities to affect multiple persons
• 2. An appropriate mode of transmitting the pathogen to susceptible persons
• 3. An adequate pool of susceptible persons who are exposed to the pathogen.

These three features are presented schematically in Figure 5–1.

Some outbreaks are self-limited and terminate without any intervention. In other situations, however, the outbreak will continue unless action is taken to prevent further spread. An effective control strategy should address one or more of the three conditions necessary for an epidemic. Specifically, the following interventions could terminate an outbreak:

• 1. Removal or elimination of the source of the pathogen
• 2. Blockage of the transmission process
• 3. Elimination of susceptibility (eg, through vaccination or medication).

As the name implies, person-to-person spread occurs when the causal agent is transmitted from one individual to another. As described in Chapter 3: Patterns of Occurrence, tuberculosis is propagated in this manner, with the pathogen conveyed from an infectious person to a susceptible individual via airborne particles. The investigation of an outbreak of tuberculosis is likely to reveal an initial (or index) case, and a number of subsequent cases that develop among the close contacts of the infectious person. An effective control strategy would involve isolation of the index case and treatment of the index case with antibiotics until the individual involved is no longer infectious, and antibiotic therapy for infected close contacts.

A common-source exposure occurs when the causal agent is transmitted to affected individuals by some shared feature of the environment. For example, contaminated food can be the source of pathogenic bacteria for persons who ingest the food. Appropriate control measures for this type of outbreak would involve removal of the contaminated food and review of food preparation practices to prevent future outbreaks. Common-source outbreaks are not limited to infectious pathogens. For example, chemical contamination of shared food, air, or water can result in an outbreak of disease.

• 1. Apparent number of persons affected
• 2. Presence of unusual or severe clinical symptoms
• 3. Lack of an obvious explanation for disease occurrence
• 4. Perceived need to implement control measures
• 5. Level of public concern
• 6. Potential for contributing to medical knowledge.

The situation depicted in the Patient Profile did not involve a life-threatening condition, and the symptoms were rather typical of an acute gastrointestinal illness. On the basis of these criteria alone, an investigation of this outbreak would not be justified. On the other hand, a relatively large number of individuals were affected, the cause was uncertain, and members of this college community were concerned about further spread of the epidemic. For these reasons, an investigation of the outbreak was undertaken.

This epidemic of apparent gastroenteritis occurred on the campus of a liberal arts college in the northeastern United States. The temporal association of cases led to the working hypothesis that the cause of this epidemic was a microbial pathogen from a common source. It was suspected that there was one vehicle for transmission of the disease agent, and as the epidemic quickly reached a temporal peak, that the vehicle was quickly exhausted or removed. The epidemic was investigated by public health authorities.

Note that the AR is a measure of risk, as defined in Chapter 2: Epidemiologic Measures. Accordingly, a period of time must be specified in the estimation of an AR.

Knowledge that the college enrollment was 1164 allows the AR for the outbreak of the gastrointestinal disease to be calculated. The AR for the period from 10:30 pm January 17 to 8:00 pm January 18 was

It was readily apparent, however, that the population at risk could be defined more narrowly because all students who reported to the infirmary lived in dormitories on campus, but only about two thirds of all enrolled students lived in those dormitories. In other words, the one third of students who lived outside of dormitories did not appear to be at risk of disease. Because these students were not entered into the numerator of the equation, their inclusion in the denominator makes the calculation potentially misleading. A more precise estimate of the AR, based on the 756 students at risk in the dormitories, was

By defining the population at risk more precisely, the estimated AR increased by more than 50%.

Because the patients’ dormitories were recorded on infirmary records, the ARs could be calculated by dormitory and by gender (dormitories were separated by gender). These data are presented in Table 5–1. The striking difference in the ARs clearly suggested that the residents of two dormitories (1 and 12) were at greater risk than residents of the other 12 dormitories. A combined AR for the two dormitories at greater risk can be contrasted with the other 12 as follows:

The AR for the remaining 12 dormitories was

A ratio of these attack rates may be calculated as follows:

This risk ratio means that the AR in dormitories 1 and 12 was 9.4 times greater than in the remaining 12 dormitories.

A different ratio could be constructed using only the number of cases. Such a ratio would be 32 cases (dormitories 1 and 12) divided by the 15 cases in the remaining 12 dormitories. This ratio is 32/15, or 2.1. It should be clear, however, that this latter ratio is not an appropriate comparison since it does not take into account the differing sizes of the populations at risk in the dormitories. There were only one fourth as many students at risk in dormitories 1 and 12, and so the same number of cases would not be expected to occur in these two dormitories as occurred in the other 12. If the residents of dormitories 1 and 12 experienced the same risk as other dormitory residents, then the expected number of cases in dormitories 1 and 12 would be

The frequency data in Table 5–1 can also be used to calculate rates by gender. For males, the AR was

A similar calculation for females yielded an AR of 6.6%. The ratio of attack rates for females to males was 1.2, indicating that there was not much difference in risk of acquiring this disease by gender.

Visits to some of the campus dormitories by the investigators soon revealed that not all students who became ill had visited the infirmary. Thus it became important to obtain additional data regarding the nature and extent of the outbreak, which it was hoped would be less biased by differences in care-seeking behavior. Questionnaires were prepared and distributed by hand to all students living in seven dormitories chosen to provide a representative sample of the student population. The results from this survey are presented in Table 5–2. A different picture of this epidemic emerged from these results. The overall AR could now be calculated as

Note that the denominator for this AR is 304, which was the number of returned questionnaires, and not 411, which was the total number of eligible students. Because nonrespondents could not be classified according to disease status (in the numerator of the AR calculation), their inclusion in the denominator makes the AR calculation invalid. In the initial infirmary data, the ARs for dormitories 6 and 12 were 0% and 21%, respectively. From the survey responses, however, these dormitories did not appear to have different ARs:

In other words, the infirmary records and the questionnaire data gave two very different perspectives on the distribution of the outbreak by place of residence. The explanation for this discrepancy was not immediately apparent, but could have reflected differences in approaches to data collection. The infirmary data were useful in the initial phase of the investigation because they were readily available. On the other hand, these data could have been influenced by various factors, such as variation in the severity of illness and care-seeking behavior. The student survey tended to avoid these problems.

The true AR of gastroenteritis on campus was not known. The best estimate was 36.2%, as determined from the student survey, but this AR could have been incorrect as only 74% of the survey questionnaires were returned (see Table 5–2). If the illness experience of the nonrespondents differed appreciably from the three fourths of students who responded, then the estimated AR could have been incorrect. It is possible, for example, that students who were not affected by the illness were less motivated to respond to the survey. Under these circumstances, the calculated AR of 36.2% would be higher than the true AR for all students. The extent to which the estimated and true ARs differ reflects bias, or lack of validity. Another potential source of bias could have arisen in the selection of the dormitories to be surveyed. To the extent that these dormitories were systematically different from the remaining nonsurveyed dormitories, bias could have been introduced into the estimation of the true AR. Furthermore, if students misreported their illness experience, either intentionally or unintentionally, a distorted apparent pattern of the outbreak could have emerged.

The concept of bias will be discussed at length in Chapter 10: Variability & Bias. Suffice it to say that minimizing the amount of bias in any study is desirable. To reduce the potential for bias in the present context, the amount of missing information must be minimized. Because the investigators could not force students to respond, it would be unrealistic to expect complete participation. Nevertheless, various strategies could be used to increase the response level (eg, the use of reminders and peer support). The pattern of response rates shown in Table 5–2 was clearly nonrandom, with particularly low response levels in dormitories 7, 8, and 9. Focused efforts to increase participation among residents of those dormitories might have helped to increase confidence in the validity of the findings.

Several factors could explain why the ARs estimated from infirmary records were low. Some students may have experienced mild illness that did not require medical attention, and others may have sought care elsewhere. Also, it was discovered that dormitories 1 and 12, which initially appeared to have the highest ARs, were located adjacent to the infirmary, and therefore access to this facility was more convenient for these residents.

The survey data indicated a much higher AR and more widespread nature of the disease than originally suspected; in fact, more than one third of all students appeared to be involved. In addition, the abrupt onset of disease and the clustering of cases in time suggested a common-source exposure. Data collected during the survey indicated that no large gatherings of students, such as parties or sports events, had recently occurred. Attention then was directed at meals, as most students ate at the college cafeteria. Included in the survey were questions concerning the source of meals eaten on January 16 and 17. Information from the survey is summarized in Table 5–3. Data are presented in a format that is typical for food histories. ARs are presented for those who ate and those who did not eat each meal. For example, 152 students ate breakfast on January 16, and of these 52 reported that they became ill. Thus the AR was (52/152) × 100 = 34.2%. ARs were calculated in a similar manner for all meals on these days.

The relationship between eating a particular meal and developing illness was assessed by contrasting ARs in those who ate the meal and those who did not. Not surprisingly, most ARs were approximately 36%, with little difference between those who did or did not eat a particular meal. The exception was found for the January 17 lunch meal. A ratio of ARs (risk ratio) for eaters and noneaters of this meal was calculated as follows:

This risk ratio indicates that those who ate this meal were more than seven times as likely to have become ill as those who did not eat this meal. Similar risk ratios were calculated for the other meals and were close to 1.0. For example, the risk ratio for dinner on January 17 was 37.5/32.6 = 1.2.

Having identified the meal at which the students most probably were exposed to the causal pathogen and knowing each student’s time of onset of symptoms, it was possible to calculate the incubation period, in this case the time between eating the meal and the onset of symptoms, for 101 ill students. The distribution of these incubation periods is shown in Table 5–4 and is presented graphically in Figures 5–2 and 5–3. In Figure 5–2, the number of cases that occurred is shown by time in hours between exposure and onset of illness. In Figure 5–3, the cumulative percentage distribution of cases is presented by incubation period. The median incubation period is the time by which 50% of the cases have occurred. The median incubation period in Figure 5–3 is 10 hours. A follow-up survey was directed at obtaining information about the noon meal of January 17. Data from the follow-up study are presented in Table 5–5.

To identify the food(s) responsible for the outbreak, dietary histories were analyzed. The questionnaires provided information about particular foods that 251 students ate at the noon meal on Friday, January 17. If students were uncertain about whether they ate the food in question, they were not included in the analysis of that particular food. As a consequence, the total of those who ate and those who did not eat each specific item did not equal 251 for all items.

The ratios of ARs yielded values greater than 1.0 for certain items, indicating that the risk of illness was greater among students who ate that item than among those who did not eat the item. For other items the values were less than 1.0, indicating that risk of illness was lower among those who ate the item in question. For example, the risk ratio was 0.7 for fish chowder, 1.0 for fruit salad, and 8.0 for lamb stew pie. It was clear from these data that the risk ratio for illness was considerably higher for those who ate lamb stew pie than for those who did not. This information suggested that lamb stew pie was the most likely source of exposure to the pathogen. Further investigation of the preparation of the lamb stew pie indicated that it was prepared the previous day (January 16), refrigerated, and warmed on the morning it was served.

Although no specific laboratory studies were performed, the etiologic agent that caused this outbreak could be inferred from available information. The illness was marked by gastrointestinal symptoms of limited duration, usually without fever or vomiting; the median incubation period was 10 hours; and a meat dish was presumed to be the most likely source of the pathogen. Based on these observations, the etiologic agent probably was Clostridium perfringens. When ingested from inadequately cooked foods, particularly stews, meats, or gravies, type A strains of this bacterium produce a toxin that causes gastrointestinal symptoms. The incubation period, which may range from 8 to 22 hours, is typically 10–12 hours. The absence of fever differentiated this particular gastrointestinal illness from shigellosis or salmonellosis, and the absence of vomiting distinguished this illness from food poisoning by staphylococcal bacteria or chemical agents.

Because food poisoning from C perfringens does not involve person-to-person transmission, it is not necessary to isolate affected patients. Supportive treatment should include oral administration of fluids and electrolytes, or intravenous therapy in severe cases. Antibiotic treatment is not required. Taking measures to reduce bacterial proliferation in food sources can prevent outbreaks of food poisoning from C perfringens. Thorough cooking of meats and stews at an adequate temperature is essential. Dividing stews into small cooking portions helps ensure uniform heating and decreases the need to store and reheat leftovers.

As shown in Figure 5–4, the number of persons reported with food-borne disease is a small percentage of the total number of persons exposed to contaminated food within the population. The area of the largest square represents the total number of persons who are exposed to contaminated food. Among those persons, some, but not all, experience illness. Of those with symptoms, only a portion will seek medical attention, and an even smaller fraction will have diagnostic specimens obtained. Some of those specimens will have diagnostic tests performed, of which a proportion will confirm a pathogen, a fraction of which will be reported officially.

Given this chain of selective factors, it is clear that national surveillance data yield an incomplete picture of the true occurrence of food-borne disease. It is estimated that there are about 76 million persons who contract food-borne illnesses in the United States each year. Of these individuals, about 325,000 are hospitalized and 5000 die.

Recognizing the limitations of surveillance data, this information still represents our best window into the population experience. In 2001, 1238 food-borne diseaseoutbreaks were reported to the CDC, involving 25,035 affected persons. Of the total reported outbreaks, only 445 (36%) had a confirmed agent, involving 14,090 affected persons.

As shown in Figure 5–5, bacterial agents accounted for over half of the reported food-borne outbreaks with known causes, followed by viruses, chemicals, parasites, and multiple agents. The mean number of affected persons varied by type of agent (Figure 5–6), with the largest average reported outbreak size for viral and bacterial agents, and the least for chemical agents.

More detailed information on the reported food-borne outbreaks involving bacterial pathogens in the United States during 2001 is shown in Table 5–6. Salmonella enteritidis was responsible for the largest numbers of reported outbreaks and affected persons. Clostridium perfringens, the pathogen responsible for the outbreak described in the Patient Profile, accounts for the next largest number of reported outbreaks and affected persons, followed by Staphylococcus aureus, Escherichia coli serotype O157:H7, Campylobacter jejuni, Salmonella typhimurium, and Shigella sonnei.

Brief descriptions of key features of the common bacterial food-borne diseases are presented in Table 5–7. Some of these attributes are very characteristic of specific pathogens. For example, S enteritidis infection typically results from the consumption of raw or undercooked eggs, E coli O157:H7 infection usually arises from the eating of undercooked ground beef, and C jejuni infection is most commonly associated with dining on undercooked chicken. The incubation periods from exposure to symptom development range from a few hours for S aureus food poisoning to several days (eg, E coli O157:H7 and C jejuni infections). Bloody diarrhea is associated with E coli O157:H7 and Shigella infections, whereas the diarrhea related to the other illnesses is not bloody. Some of these outbreaks also have a characteristic seasonality to their occurrence. For example, as shown in Figure 5–7, the reported outbreaks of S enteritidis were heavily concentrated in the months of April through July.

The national information on food-borne diseases just described was collected through passive surveillance. That is to say that those responsible for collecting the information did not go out into the community to obtain it. Rather, they relied upon existing information that was reported to them voluntarily by laboratories and health care providers. The advantages of passively collected information are that it is timely and inexpensive to collect. The principal limitations of this type of information are that it may be incomplete and inaccurate.

In contrast, active surveillance refers to a system in which those responsible for collecting the information go into the community (usually defined geographically) to obtain it. Information can be collected from interviews, examinations, and special testing, as well as routine data sources. Compared with passive surveillance, the advantages of active data gathering are that it tends to be more complete and accurate. On the other hand, information obtained through active surveillance typically takes more time and expense to amass.

An active surveillance system for food-borne disease was established in the United States in 1996 and is referred to as the Food-borne Diseases Active Surveillance Network (FoodNet). By 2002, this system was expanded to nine states with an aggregate population of over 37 million persons (13% of the national population). Ten specific pathogens (eight bacteria and two parasites) were included in the most recent surveillance data. Over the period 1996 though 2002, declines in incidence were seen for Campylobacter (24%), S typhimurium (31%), Listeria (38%), and Yersinia (43%). Shigella, S enteritidis, and E coli O157:H7 did not have consistent trends over this period of time. As shown in Figure 5–8, infections with Campylobacter and Salmonella had much greater incidence in early childhood than at older ages.

Overall, about one in six persons with a positive culture were hospitalized, although this varied widely by pathogen. Listeria, for example, accounted for less than 1% of all diagnoses, but it had the most serious consequences, with 90% of cases hospitalized and a case fatality of 20%.

The epidemiology of food-borne disease has changed over time for a variety of reasons. First, there have been changes in dietary habits within the population. Over the past two decades in the United States, the consumption of whole milk and red meat has declined, whereas the intake of poultry, fresh vegetables, and fruits has increased. The availability of domestic fresh fruits and vegetables has been supplemented by importing these commodities from tropical countries. Food-borne outbreaks of disease caused by a variety of Salmonella species have been traced to imported fruits and vegetables, and many other outbreaks probably go undetected because the imports are widely distributed.

Another trend that has affected the epidemiology of food-borne disease is the rising consumption of food in commercial establishments. Pathogens such as Norwalk virus can be spread from an infected food handler to a consumer through contaminated food. Training in hygiene and sanitation often is minimal in this industry because the food handlers are low paid, poorly educated, and have a rapid turnover in employment. For many of these individuals, sick leave may not be available, so they may work while ill with infectious diseases.

Yet another factor that has influenced the epidemiology of food-borne disease is the growing presence of large-volume food production facilities. With a trend toward mass-produced foods that are distributed to consumers in remote locations, food-borne diseases may affect large numbers of persons in a variety of locations. Consequently, the classic hallmarks of food-borne disease outbreaks—high attack rates, short incubation periods, and clustering of affected persons in time and space—may not occur, making the outbreak difficult to detect. Once an outbreak is suspected, however, the search for the underlying cause typically follows the basic approach outlined in the preceding section.

Although evolutionary processes can give rise to new infectious agents, most emerging infections probably result from previously existing but obscure pathogens. A number of factors can promote the emergence of these preexisting pathogens:

• 1. Ecologic changes
• 2. Shifts in human populations
• 3. International travel and commerce
• 4. Changes in technological or industrial practices
• 5. Microbial adaptation
• 6. Lapses in the public health system.

Ecologic changes can contribute to emerging infections by placing people into close contact with pathogens not previously encountered by some or all human populations. In addition, ecologic changes can result in conditions that increase the size of a microbial population. The role of ecologic changes in the emergence of an infectious disease is well illustrated by Lyme disease, a systemic illness with characteristic skin rash, joint pain, and in advanced stages, neurologic and cardiac manifestations. This condition is caused by the spirochete Borrelia burgdorferi, which was first isolated and identified in 1982. This spirochete is borne by a particular deer tick. The first outbreak of Lyme disease was detected in the Connecticut town after which the disease is named. The emergence of Lyme disease in New England probably occurred as a consequence of reforestation of land that had been previously cleared for agricultural purposes. As the forests returned, the deer population grew, in turn allowing the expansion of the tick population and transmission of the spirochete to humans through tick bites.

Not all ecologic changes that contribute to emerging infections are caused by human activity. Natural changes in ecology can also give rise to emerging infections, as illustrated by the hantavirus pulmonary syndrome (HPS). Hantavirus is named for a river in South Korea in which the virus was originally discovered. It is suspected that outbreaks of disease caused by the Hantaan virus occurred in Asia for many centuries, and that the virus was responsible for the condition known as Korean hemorrhagic fever, which affected thousands of soldiers involved in the Korean War during the 1950s. Until 1993, however, hantavirus was not known to cause disease in the United States or to cause an adult respiratory distress syndrome. Beginning with an influenza-like illness, HPS progresses rapidly to respiratory failure, with a high case fatality.

The first outbreak of HPS was detected in the Four Corners region of the southwestern United States during 1993. A new strain of hantavirus was identified, and the reservoir for this pathogen was shown to be the deer mouse, Peromyscus maniculatus. The population of this deer mouse in the Four Corners region rose 10-fold between 1992 and 1993, a result of the end of a 6-year drought. The mouse’s predators had decreased during the drought, and with an ample food supply after the drought, the mouse population exploded, expanding the reservoir for the hantavirus and increasing the likelihood of contact with humans.

Emerging infections can be traced to human societal changes, such as migration from high- to low-risk areas. As described in Chapter 3: Patterns of Occurrence, the resurgence of tuberculosis in the United States during the late 1980s was attributable in part to migrants arriving from high-prevalence countries, such as those in Asia and Latin America.

For centuries, international travel has been a means by which pathogens are introduced into previously unexposed populations. There are many examples, including the importation of yellow fever from Africa to the Americas during the sixteenth and seventeenth centuries. Another historical illustration is the spread of cholera during the nineteenth century from India to the Middle East and then to Europe. However, the opportunities for global spread of infections have multiplied in recent years. First, the volume of international travel has expanded, increasing the number of person-to-person contacts. Second, the speed of international travel by airplane makes it more likely that infected individuals will be contagious on arrival at their destinations. When modes of transit were considerably slower than today, there was greater opportunity for an infection to run its course before an infected traveler reached a destination. Third, as noted in the previous section, the growing globalization of food and other markets increases the number of opportunities to convey pathogens from one country to another.

There is no more impressive example of the role of international travel in the spread of an emerging infectious disease than that of SARS. This disease caused by a newly emerged coronavirus and characterized by fever and atypical pneumonia, arose in November 2002 in Guangdong Province in southern China. A physician in Guangdong became infected while caring for patients and subsequently traveled to Hong Kong. He stayed overnight at a hotel, and the next day he was admitted to a hospital where he subsequently died. While staying at the hotel, however, his infection was spread to 12 other persons.

The chain of transmission of SARS at the hotel in Hong Kong is illustrated in Figure 5–9. The affected physician (person A) spread the disease to ten other persons who stayed on the same floor of the hotel and to two others who were on other floors. After being infected, these contacts then traveled around the world to Viet Nam, Singapore, Canada, the United States, and Ireland, where they then spread the disease to others. Collectively, over 400 cases of SARS in eight countries could be traced directly or indirectly to the affected physician. When the global chain of transmission was broken in July 2003, a total of over 8000 persons were infected and 774 deaths had occurred in 29 countries. With such a rapidly fatal illness, it would have been difficult to spread SARS around the world so quickly in the absence of airplane travel.

The spread of infectious disease across wide geographic areas can occur without person-to-person transmission. For instance, technologic changes now allow food and other products to be produced in mass quantities and to be distributed widely, thereby enhancing the spread of emerging infections. A recent multistate outbreak of Escherichia coli O157:H7 infections illustrates this phenomenon. In 1993, a physician in the State of Washington first noted and reported the hemolytic uremic syndrome in a cluster of children who had presented to local emergency rooms with bloody diarrhea. Investigation by state health officials and officials from three other western states identified more than 500 affected persons. Their illnesses were linked to the consumption of hamburgers at a particular chain of fast-food restaurants. Cultures of stool from affected persons revealed E coli O157:H7, an organism that lives in the intestines of healthy cattle, can contaminate meat during slaughter, and is a pathogen to humans. This organism was isolated from 11 lots of hamburger, collectively accounting for over a million ground beef patties. The affected hamburger was traced back to five slaughterhouses in the United States and one in Canada, which were the likely sources of the contaminated meat.

Evolutionary changes in microbes also can give rise to emerging infections. Antibiotic resistance is one such mechanism of adaptation. The problem of multidrug-resistant tuberculosis was described in Chapter 3: Patterns of Occurrence. Other current examples of pathogens for which antibiotic resistance may impede disease control efforts include penicillin-resistant Streptococcus pneumoniae and nosocomial enterococci resistant to vancomycin. In addition to antibiotic resistance, evolutionary changes to established pathogens can limit their control in other ways. Many viruses mutate rapidly, yielding new strains of pathogens with surface antigens that differ from those of existing strains; this phenomenon is known as antigenic drift. Reinfection with the new viral strain can arise because its antigens are not recognized by the immune system of the host. In rare instances, a new variant of a pathogen may even result in an entirely new clinical syndrome. The manifestations of streptococcal toxic shock syndrome and necrotizing fasciitis linked with group A streptococcal infection may represent this type of phenomenon.

The failure of public health measures also can give rise to emerging infections. For instance, a nonfunctioning water filtration plant in Milwaukee, Wisconsin resulted in contamination of the municipal water supply with Cryptosporidium in 1993. This pathogen is a protozoan that infects epithelial cells of the gastrointestinal, biliary, and respiratory tracts of humans and other animals. In humans, the primary manifestations of illness are diarrhea and cramping abdominal pain. Among immunodeficient persons, including those with acquired immune deficiency syndrome (AIDS), infection can follow a fulminant course, even leading to death. The outbreak in Milwaukee produced illness in more than 400,000 individuals, 4400 of whom required hospitalization.

West Nile virus is a member of the Flaviviridae family of viruses. It was first recognized in the West Nile district of Uganda (after which it was named) in 1937. Disease from this agent was not observed in the Western Hemisphere until 1999. The primary amplifying hosts for the virus are birds, with mosquitoes of the species Culex pipiens serving as vectors for transmission. Humans and horses are infected incidentally through bites from infected mosquitoes.

In humans, West Nile disease has an incubation period of 2 to 14 days, with symptoms of fever, malaise, headache, rash, and muscle weakness. About 1% of affected persons develop severe neurological disease, which may include inflammation of the brain or spinal cord and a paralysis that is similar to poliomyelitis. The case fatality rate generally is 10% or less. As shown in Figure 5–10, West Nile virus spread across the United States in successive annual outbreaks between 1999 and 2003. Initially, it began in the northeast, with subsequent expansion to the south and midwest, and eventually to the far west.

SARS is caused by SARS-associated coronavirus, and as indicated earlier, it first appeared in southern China in November 2002. It is thought to have arisen as a genetic mutation of a virus that affects nonhuman animals. It is easily spread by person-to-person transmission via inhalation of the virus. The incubation period is 2 to 7 days, and the symptoms are characterized by fever, dry cough, and respiratory distress that can progress to respiratory failure. The distribution of SARS cases by week of onset in 2002–2003 is shown in Figure 5–11. It can be seen that the initial peak of cases occurred in early February 2003, followed by successive peaks in March and April related to secondary spread of the infection.

At present, vaccines do not exist for these three cited emerging viral diseases. Therefore prevention of spread represents the first line of defense against these infections. In order for preventive measures to be effective, new cases must be recognized quickly and reported to public health authorities. In the cases of Ebola and SARS, in which transmission occurs by person-to-person contact, appropriate infection control practices are critical. Immediate isolation of infected individuals, along with standard hygienic precautions (eg, hand cleansing), and appropriate barrier protection are indicated. For West Nile virus, protection against mosquito bites is critical. Appropriate measures include draining standing water where mosquitoes breed, covering skin surfaces with clothing and insect repellant, and avoiding outdoor activities from dusk to dawn when mosquitoes tend to feed. Education of the public and health care professionals about these disease control strategies is critical to containing the spread of these emerging diseases.

All of the outbreaks described in the preceding sections were related to the natural occurrence of disease. Intentional exposure of humans to infectious agents also can result in disease, although such episodes historically have been uncommon events.

The intentional outbreak of anthrax in the United States during 2001 provides some important lessons about bioterrorism. In brief, at least five letters containing spores of B anthracis were mailed from Trenton, New Jersey to offices in Florida, New York City, and Washington, DC. Four of the envelopes were recovered, two of which were postmarked on September 18 and the other two were postmarked on October 9. The letters made reference to the terrorist attack on the World Trade Center on September 11, 2001. All of the letters contained spores of B anthracis of the Ames strain, although the specific source of the material involved was not determined. The spores were contained within a powder, and it is estimated that the material may have included hundreds of billions of spores.

As shown in Figure 5–12, a total of 22 cases of bioterrorism-related anthrax were identified. Two distinct clusters of cases were observed around the time of delivery of the two separate mailings. The first cluster occurred between September 22 and October 1 and predominantly involved cutaneous disease. The second cluster was between October 14 and 25 and primarily involved inhalational anthrax.

Together, the two clusters included 11 cases of confirmed inhalational anthrax and 11 cases of cutaneous anthrax (7 confirmed and 4 suspected). Five of the inhalational anthrax patients died (case fatality = 45%), but none of the cutaneous anthrax patients died (case fatality = 0%). The affected individuals resided in seven different states (New York, 7 cases; New Jersey, 6; Maryland, 3; Florida and Virginia, 2 each; and Connecticut and Pennsylvania, 1 each). Twelve of the affected persons were mail handlers, among whom inhalational anthrax was the predominant form of disease (67%). For the mail handlers with known times of exposure, the incubation period for inhalational anthrax was 4.5 days. The second largest occupational category of exposure was work at a media company (6 cases), where cutaneous anthrax predominated (83%).

This anthrax episode illustrates several of the reasons that biological agents may be attractive to terrorists:

• 1.Stability: The agent was sent through the mail, surviving transport over distance, handling, and time.
• 2.Infectivity: The agent was “weaponized” by using material with a high concentration of spores, uniform particle size, low electrostatic charge, and reduction of clumping. Also, the recipients were highly susceptible to contracting the disease.
• 3.Low visibility: The use of the postal service allowed the terrorist to be virtually invisible at the time of release.
• 4.Ease of delivery: The agent was mailed at low cost from a remote location, with a delay in the time of exposure making it difficult, if not impossible, to trace the perpetrator.

The anthrax attacks also demonstrate the impact that a bioterrorist event can have on the public health system. The ability to deliver antibiotics to large numbers of potentially exposed individuals was critical to preventing further infection. A total of 33,000 persons were placed on antibiotics by public health authorities, with 10,000 requested to complete a 60-day course. The number of people who took antibiotics of their own volition is unknown. The United States Army Medical Research Institute of Infectious Diseases performed 19,000 assays for anthrax during the outbreak, and state and local health departments performed countless analyses on suspicious white powders.

Other consequences of the anthrax attacks included disruption of mail delivery to the US Capitol for weeks. The decontamination of the Hart Senate Office building took over 3 months to complete at a cost of over $23 million. The total cost to the nation of the anthrax attacks is beyond reckoning.

The fact that only 22 persons became ill, and of those only 5 died, is a tribute to the rapid and effective responses of the clinicians and public health workers involved. Good communication was essential between the physicians caring for the patients and the officials investigating and attempting to control the outbreak. For example, the morbidity and mortality could have been far worse if the index case was not diagnosed so quickly by an astute clinician and quickly linked by investigators to a potential exposure from a suspicious letter received at work.

Even before the anthrax episode of 2001, the threat of biological terrorism was recognized. In 1999, the United States Congress initiated an effort to enhance capabilities to respond to bioterrorism. The CDC was designated to coordinate planning, and in that context, conducted a public health assessment of potential biological terrorism agents. The following criteria were used in that assessment:

• 1.Public health impact—based upon illness and death
• 2.Delivery potential to large populations—based upon stability of the agent, the ability to mass produce and distribute, and the potential for human-to-human transmission
• 3.Public perception—based upon fear and public disruption
• 4.Special preparedness needs—based on requirements for surveillance, testing, and vaccine or antibiotic stockpiles.

Using these criteria, a set of three priority categories was established for public health preparedness, as summarized below:

• A. Agents with the greatest potential for mass casualties and that require broad-based preparedness
• B. Agents with some potential for mass casualties, but lower morbidity and mortality and less public perception
• C. Agents that do not present a high risk at present but could emerge as future threats.

While all of these agents warrant consideration from the perspective of public health preparedness, only those highest-risk agents (Category A) are presented here. The criteria used in the evaluation and the corresponding weights assigned to each of these agents are summarized in Table 5–9. The agents with the highest potential for causing death are B anthracis, Y pestis, C botulinum, and viral hemorrhagic fevers. The ability to disseminate the agent to large populations was deemed greatest for B anthracis. The ability to generate public fear was highest for smallpox, anthrax, and viral hemorrhagic fevers. Special preparations are required for all of these agents, with slightly lower ranking for viral hemorrhagic fevers because effective vaccines and antibiotics are not available for these agents.

Clinical features of the Category A agents are presented in Table 5–10. In general, very low levels of exposure are required to produce infection. The incubation periods and durations of illness are particularly short for anthrax, plague, and botulism. The initial symptoms of all of these diseases are nonspecific, and parallel those associated with influenza. The risk of death is relatively high for all of these agents, although somewhat lower for tularemia and variable for the viral hemorrhagic fevers.

Most of the agents on the Category A list are infrequent natural causes of disease in the United States. For those that are part of the regular notifiable disease reporting process, the numbers of cases observed over the 8-year period from 1992 through 1999 are presented in Figure 5–13. The most common naturally occurring disease was tularemia, with a mean annual number of new cases exceeding 100. Botulism accounted for the next largest number of reports, with fewer than 50 cases per year on average, followed by plague with about 10 cases per year. During the entire 8-year period, only one case of anthrax was observed and there were no cases of smallpox.

The relative rarity of these conditions under natural circumstances can help raise suspicion of a terrorist-related incident when an outbreak occurs, as was illustrated in the anthrax episode of 2001. Other features of the natural occurrence of these diseases may differ from those of an intentional exposure, thereby providing clues to the origin. For instance, in the United States, tularemia occurs naturally with greatest frequency in midwestern states. Tularemia in a different location, such as an urban setting on the west coast, would clearly suggest an unnatural pattern of exposure. Similarly, under normal circumstances plague tends to occur almost exclusively in western states, and food-borne botulism is particularly common in Alaska among the native population. Outbreaks of these diseases in other parts of the country might suggest the possibility of bioterrorism.

Seasonality is an important feature of endemic tularemia and plague, both of which predominate during the summer months. Outbreaks of these diseases at other times of the year might signal an intentional exposure. Other demographic patterns within the affected population also might be helpful in distinguishing naturally occurring events from bioterrorist episodes. For example, food-borne botulism occurs with the highest incidence in infants, as illustrated in Figure 5–14. In a food-borne botulism outbreak without a predilection for affecting young children, the possibility of an intentional exposure might be considered.

In considering the future likelihood of bioterrorism, most experts suggest that the issue is not whether such episodes will occur, but when, where, and how they will happen. There is ample opportunity for access to these deadly agents, given the high levels of past production around the world and the ease with which they can be produced in the future. Moreover, the delivery of these agents can be simple, efficient, inexpensive and difficult to trace, as illustrated by the anthrax attacks of 2001. As long as there are conflicts between people, one must assume that the risk of bioterrorism persists. As with all threats to the health of the public, priority must be given to preventing disease occurrence, detecting it quickly when it occurs, treating those affected, and controlling the spread of disease to other individuals.

The investigation of disease outbreaks (sudden and geographically limited epidemics) is an essential role of epidemiology. The primary goals of an outbreak investigation are the identification of the causal agent (the pathogen) and the prevention of further cases. The propagation of a disease outbreak requires a pathogen, a viable mode of transmission, and an adequate pool of susceptible persons. Elimination of one or more of these three components will terminate the outbreak. Two basic modes of transmission are person-to-person spread and common-source exposure. Infectious illnesses can be transmitted by either mode, whereas noninfectious environmental pathogens usually produce disease outbreaks through a common-source transmission.

Not all disease outbreaks warrant investigation. The decision to investigate an outbreak typically is based on the severity of illness, the number of affected persons, uncertainty about the pathogen, and the perceived need to control further spread of the disease. Investigations usually are conducted by local, state, or federal public health officials.

In this chapter, the principles and methods of outbreak investigation are illustrated by the evaluation of an episode of food poisoning on a college campus. This outbreak attracted attention because almost 50 students sought medical care for acute gastrointestinal symptoms in a period of less than 24 hours.

A measure of the risk of developing an illness over a specified period of time is the attack rate (AR). Based on infirmary records, the AR among all students was estimated to be 4%, but among those residing in dormitories, the estimated AR was 6%. The AR did not differ by gender, but students residing at two dormitories had higher ARs than residents of the other 12 dormitories.

To collect more detailed, and perhaps more complete, information than could be obtained from infirmary records, investigators conducted a survey of residents based on a representative sample of dormitories. The survey results indicated a much higher AR for dormitory residents (36%) than had been estimated from infirmary records (6%). The possibility of bias (systematic error) in both estimates must be considered. Nevertheless, the results of the survey suggested that (1) the outbreak was widespread among dormitory residents and (2) the prior definition of high-risk dormitories based on infirmary records might be inaccurate.

Comparison of ARs for persons who ate and did not eat specific foods indicated a strong association between eating a particular food and the risk of gastrointestinal illness. The distribution of times from ingestion to the onset of symptoms (incubation period) revealed a median interval of about 10 hours. A follow-up survey revealed that lamb stew pie was the most suspect common-source exposure for this outbreak.

Although cultures of the foods and specimens from affected individuals were not obtained, the nature of the symptoms, the median incubation period, and the presumed source of exposure implicated C perfringens as the pathogen. Knowledge of the factors that contribute to the proliferation and transmission of this bacteria was used to control the outbreak; such knowledge could be used to prevent similar episodes in the future.

At the national level, information on food-borne outbreaks is collected through passive surveillance, involving the voluntary submittal of information by physicians and laboratories to the CDC. Although this information likely represents a small portion of the total outbreaks, it provides data quickly and at limited cost. This information suggests that bacterial agents are the most common sources of food-borne infection, followed by viruses. S enteritidis was the most commonly reported pathogen, followed by C perfringens. Typical sources of infection vary by agent, as do the symptoms.

An active surveillance system for selected food-borne diseases also exists in the United States and has been useful for tracking the incidence of these infections. The traditional epidemiology of food-borne outbreaks—high attack rates, short incubation periods, and clustering of affected individuals in time and place—is changing for a number of reasons. The composition of the typical diet has shifted, with decreased consumption of red meat and increased consumption of poultry, fresh fruits, and vegetables. Consequently, exposure to pathogens associated with the latter products is rising. The increased consumption of food from commercial establishments also has resulted in greater opportunities for transmission of infections from food handlers to consumers. Increasing numbers of facilities that produce food in massive quantities, combined with the use of distribution systems that deliver foods to geographically remote markets, tend to diminish any clustering in time and space of persons who are infected with pathogens borne by these foods.

Emerging infections can be defined as infections that are appearing in a population for the first time or that have existed in a population but are rapidly increasing in incidence or geographic range. A number of factors contribute to the emergence of certain infections:

• 1. Man-made or natural changes in the environment (eg, accounting for Lyme disease and HPS)
• 2. Demographic shifts in populations (eg, accounting for the spread of HIV)
• 3. Increased international travel and commerce (eg, the rapid global spread of severe acute respiratory syndrome [SARS])
• 4. Technologic and industrial changes (eg, accounting for the rise of hemolytic uremic syndrome in the Northwest in 1993)
• 5. Adaptation of microbes (eg, accounting for antibiotic-resistant organisms)
• 6. Lapses in the public health system (eg, accounting for infections due to water-borne Cryptosporidium).

Recent outbreaks involving several emerging infectious diseases were highlighted. Infection with Ebola virus, which has occurred in multiple African countries, causes a highly fatal hemorrhagic fever. West Nile virus has spread from its origins in Africa to the United States, where it has migrated across the country over the past few years. This seasonal disease is transmitted by infected mosquitoes and occasionally results in serious neurologic consequences. SARS is caused by a coronavirus that originated in China in late 2002 and creates a highly contagious atypical pneumonia that can lead to respiratory failure.

Bioterrorism is the intentional use of biologic agents for causing fear, as well as human illness and death. Although occasional episodes have occurred in the past, the anthrax attacks of 2001 underscored the threat of bioterrorism. This episode involved the mailing of at least five envelopes containing weapons-grade material containing B anthracis. A total of 22 persons were infected, resulting in five deaths.

The CDC has identified a series of agents that pose the greatest threats to public health from bioterrorism. These so-called Category A pathogens include those responsible for smallpox, anthrax, plague, botulism, tularemia, and viral hemorrhagic fevers. Patterns of occurrence that differ from those of the naturally occurring infections may give rise to the suspicion of an intentional exposure.

Effective responses to an attack by bioterrorists include the prompt recognition, treatment, and where appropriate, isolation of cases. If possible, the source of the infection should be identified in order to limit any further exposure. Other persons who were potentially exposed should be identified, monitored and treated with appropriate preventive or therapeutic measures.