Outbreak Characterization

We use the term outbreak characterization to refer to processes that elucidate the causative biological agent, source, route of transmission and other characteristics of an outbreak. These characteristics guide the treatment of victims and the application of control measures to prevent additional cases (e.g., by removing or isolating the source). Table 3.1 includes the complete list of outbreak characteristics and methods for their elucidation that we discuss.

As mentioned in Chapter 1, some outbreak characteristics may already be known at the time that an outbreak is detected. For example, if a biosurveillance organization detects an outbreak from analysis of notifiable disease data (which is largely organism-based reporting), it will already know the causative biological agent. If a participant in a church picnic reports an outbreak to a health department, that person may also report the source as macaroni salad, having "interviewed'' most of the picnickers by phone before calling the health department. We expect the number of outbreak characteristics that are known at the time of outbreak detection to increase as biosurveillance systems collect increasing amounts of surveillance data on a continuous basis. The distinction between outbreak detection and characterization will continue to blur.

Nevertheless, a relatively crisp demarcation between the processes of outbreak detection and characterization exists. Health departments conduct disease surveillance to detect outbreaks, and they conduct investigations using different methods to characterize them. At present, the feedback loop in Figure 1.1 (Chapter 1) becomes quite active only after an investigation commences.

4.1. Outbreak Investigations

Outbreak investigations range in size from a small inquiry conducted by a single investigator to a major multinational investigation. A seasoned investigator may need only a 10-minute phone call to determine that a suspected outbreak is small, self-limited, and not worthy of additional investigation. An outbreak that is spreading rapidly and killing many individuals (such as the SARS outbreak in 2003) may warrant deployment of thousands of investigators and researchers.

When a health department suspects an outbreak based on any of the methods described in the previous sections, its staff typically initiates a preliminary inquiry to verify the available information and estimate the severity and scope of the event. The staff reviews notifiable disease records and available medical records and/or conducts open-ended interviews of a small number of individuals, asking questions and listening, quickly obtaining important information on signs/symptoms, source, and those who might have contracted the disease through contact with known cases. At this point, the staff decides if the problem is severe enough to launch a field investigation, a decision that is based on "The severity of the illness, the potential for spread, political considerations, public relations, available resources, and other factors" (CDC, 2002a). The staff also must decide whether to inform superiors and/ or request extra help, resources, or consultation. Extra investigators can divide and complete individual case investigations much more quickly than can one person.

The investigation team (or single investigator) then begins the process of interviewing all available patients and contacts. The investigators review other sources of information such as emergency department logs, pathology specimens, medical examiner records, entomological (insect) data, and animal health data (if they suspect the cause to be exposure to a sick animal). They might issue a health alert to physicians or the public requesting that similar cases be reported by healthcare providers or institutions. The investigators obtain blood, stool, urine or other specimens from affected individuals; collect materials that they suspect may be contaminated (e.g., food, water); and send samples to laboratories to be tested for organisms that may be involved based on the epidemiological information collected to that point.

The initial round of interviews and tests may yield a fairly complete characterization of the outbreak. The investigators may know the causative organism from tests done on the first infected individual, the source of the outbreak from commonalities identified among the cases identified to date, and even the complete set of affected individuals when the outbreak is geographically localized. If they do not, the outbreak investigation will continue to use many, if not all, of the analytical techniques that we will be discussing.

Throughout this process, investigators continuously formulate and refine hypotheses about outbreak characteristics that are not yet known (e.g., biological agent, source, and route of transmission), and seek to resolve differential diagnoses for the unknown characteristics by collecting additional information. As physicians do in clinical diagnosis, the investigators apply their knowledge of epidemiology to generate hypotheses and decide what additional information to collect. They may apply control measures suggested by the most likely and/or the most serious of the possible causes of the outbreak.

Outbreak investigations are labor intensive. Outbreak investigations are sometimes referred to as shoe leather epidemiology because investigators must visit numerous hospitals, homes, stores, and morgues during the course of an investigation. There are many opportunities to use information technology to improve the speed of this process and to extend the life of investigators' shoes. Significant portions of the case data that investigators assemble by hand are available electronically in clinical information systems (see Chapter 6). Opportunities also exist to provide cognitive support to investigators with their process of generating and efficiently resolving differential diagnoses of the biological agent as well as other outbreak characteristics.

4.2. General Analytic Techniques

We here provide a brief overview of general analytic techniques that investigators use to analyze case data collected during an investigation. Investigators use these techniques (e.g., spatial analysis) to elucidate outbreak characteristics.

4.2.1. Spatial Distribution of Cases

Investigators examine the spatial (geographic) distribution of cases as soon as possible. The spatial distribution of cases often provides a strong clue about the source of an outbreak. Because of the importance of spatial analysis, one of the first stories told to epidemiologists in training is the John Snow cholera story (Snow, 1855). Dr. John Snow, a London anesthesiologist and pioneer of the science of epidemiology, decided to test his hypothesis that cholera outbreaks were a result of contamination of the water supply, a view contrary to the medical beliefs of the time. He plotted the home address of people who died of cholera on a map of London; he also marked the location of neighborhood water pumps, which were the source of drinking water at the time. The striking cluster he found of cholera deaths centered on water pumps has become legendary. The concentration of cholera deaths around the Broad Street pump was twice the number of deaths in the rest of the city of London, with approximately 500 deaths in the neighborhood in 10 days. Figure 3.7 shows Snow's map with bars denoting people who died from cholera in buildings in the immediate vicinity of the Broad Street pump.

FIGURE 3.7 John Snow's cholera map showing 115 cholera deaths in the immediate vicinity of a pump on the corner of Broad Street and Cambridge Street. According to the legend, Snow advised unbelieving officials simply to remove the pump handle. (From http://www.ph.ucla.edu/epi/snow.html.)

FIGURE 3.7 John Snow's cholera map showing 115 cholera deaths in the immediate vicinity of a pump on the corner of Broad Street and Cambridge Street. According to the legend, Snow advised unbelieving officials simply to remove the pump handle. (From http://www.ph.ucla.edu/epi/snow.html.)

At the beginning of an investigation, the investigators may only have the home address of each reported case. Therefore, the map they plot first is typically the home address of each case. During the course of an investigation, they may create many maps as they test hypotheses that the exposures may have occurred at work, school, a restaurant, fruit stand, or events such as conventions, picnics, and sporting events. Investigators may also map the location of individuals not affected. Snow did this and demonstrated that there were no cholera fatalities among brewery workers on Broad Street; these men had an allowance of free beer every day, which they apparently preferred to the water from the Broad Street pump.

Geographic information systems are modern descendants of Snow's painstakingly developed map. These systems partly automate spatial analysis. Spatial scans (see Chapter 16) are computer algorithms that more fully automate spatial analysis; these scans construct and search maps automatically for clusters of disease like that around the Broad Street pump. They can ask and answer questions such as the following: If I were to map the people in a community who developed pneumonia in the past week by using their work addresses, would the cases cluster in particular hospitals? This type of analysis would be very useful in SARS surveillance as SARS caused many hospital-based outbreaks in 2003. This type of analysis can be done routinely (e.g., daily or more frequently) even in the absence of a known outbreak as a method of outbreak detection. Spatial scans are an example of how the distinction between outbreak detection and characterization is blurring. When used for outbreak detection, spatial scans both find and spatially characterize outbreaks in one step.

Cohort Exposure Contagious Disease

Time Time

FIGURE 3.8 Hypothetical epidemic curves that would suggest a cohort exposure and a contagious disease.

4.2.2. Temporal Distribution of Cases

Investigators also examine the temporal distribution of cases as soon as possible by plotting an epidemic curve, which is a graph of the number of cases by date of onset of illness (Figure 3.8). The epidemic curve can provide a clue to the biological agent, source, and route of transmission. If the epidemic curve, for example, shows a sudden increase in cases, the investigator might suspect that the cause of the outbreak is contamination of food, air, or water and that the causative biological agent is more likely to be an agent with a propensity or ability to be transmitted in one of these ways (Figure 3.8, left) because such contaminations can infect a large cohort of individuals in a short period, producing a steep epidemic curve. If the epidemic curve rises more gradually, an investigator would suspect a communicable disease such as measles, in which the number of cases increases in an exponential fashion owing to successive generations of infection (Figure 3.8, right). A more level epidemic curve would suggest a continuous source of exposure, such as a persistently contaminated swimming pool.

4.2.3. Disease Incidence, Mortality Rates, and Attack Rates

Disease incidence is one measure of the magnitude of an outbreak (as are maps and epidemic curves). Disease incidence is the number of new cases in a population during a defined period such as a week. If disease incidence for every day or week during an outbreak is plotted, the result is an epidemic curve.

For lethal diseases, investigators gauge the severity (virulence) of the disease by the case fatality rate, which is the probability of death among diagnosed cases. Recall that investigators observed a 30% case fatality rate for the outbreak that they initially thought was Japanese encephalitis; however, the case fatality rate was highly atypical of Japanese encephalitis and led them to suspect a different disease. Investigators also compute other mortality rates. Age-specific mortality rates, for example, can help characterize an outbreak that is poorly understood by revealing that the disease affects the elderly or young with greater frequency or severity.

If investigators suspect an environmental exposure, they will calculate the attack rate, which is the fraction of people or animals exposed to a specific factor (e.g., macaroni salad or another infected individual) who subsequently contract the disease. If the attack rate in a population that is exposed to a specific factor is higher than a comparison group that is not exposed to the factor, it suggests a possible link between the factor and illness. If the analysis includes a comparison with a carefully matched control population of individuals known not to have the disease, the analysis is called a case-control study (described below). An investigator would conduct a case-control study if the less formal measurement of attack rate did not produce a definitive answer to the outbreak characteristic in question (e.g., if it did not point to macaroni salad, then the more formal case-control study likely would have).

4.2.4. Cohort and Case-Control Studies

An investigator conducts a cohort or a case-control study to test one or more hypotheses about some characteristic (oftentimes the source) of the outbreak. A cohort study compares the rate of illness of those exposed to specific factors (e.g., macaroni salad) to the rate of illness among those not exposed. A case-control study compares the frequency of specific factors in affected individuals relative to their frequency in unaffected individuals, controlling for age and other potentially confounding factors that may be correlated with the disease in question but do not cause it.

A cohort study is technically easier to conduct than is a case-control study. It is typically used for those outbreaks in which there is a small well-defined population available for interview. Examples of suitable cohorts are everyone who attended a wedding, a school, a camp, or a business conference or who ate at a specific restaurant on a given day. An investigator designs a form with three main types of questions: (1) contact and demographic information, (2) presence and onset of illness, and (3) specific exposures. For example, for a wedding at which gastrointestinal illness occurred, the questionnaire would include questions about sex, age, vomiting and diarrhea, every food item served or available (identified from menu's and the party coordinator), drinks, ice, and edible party favors. If the biological agent norovirus was suspected, exposure to public vomiting, other events such as the rehearsal dinner, and/or individuals known to be ill might also be asked. (An example of a form from a cohort study of a business conference is included in Appendix D.)

To conduct a case-control study, an investigator develops a questionnaire that covers all of the suspected sources and routes of transmission. Epidemiologists know from experience and knowledge of epidemiological patterns when to include items (e.g., intravenous drug abuse, food and water

consumption, places visited, sexual practices, exposure to sick or dead animals or people, and travel history). Case-control studies invariably include age and sex, markers for socioeconomic status, race/ethnicity, occupation, disease history, and prior immunizations, in addition to questions on the exposures of interest in a specific investigation. The investigator then assembles a set of individuals with disease (cases) and a set without disease (controls). In the design of a case-control study, attention is given to matching controls to cases on known confounding variables such as socioeconomic status, age, and sex to remove these influences from the analysis.

The investigator then administers the questionnaire to each of the cases and controls. Required data may be collected or verified from medical records. The odds ratio (OR) for each factor is calculated, which is the ratio of the incidence rate in exposed individuals relative to that among unexposed individuals (Rothman and Greenland, 1998). If the OR is equal to one, it suggests that the factor is not causing the illness.

The investigation of the hepatitis A outbreak described in Chapter 2 involved a case-control study of food items served in the restaurant. Investigators interviewed individuals with hepatitis A and controls without hepatitis A who either had dined with case patients at Restaurant R or were identified through credit card receipts as having dined at Restaurant R during October 3 through 6. They found an OR of 24.2 for consumption of mild salsa with green onions, and an OR of 5.2 for consumption of chili con queso with green onions (CDC, 2003a). An OR of 24.2 indicated that people who dined at restaurant R and subsequently developed hepatitis A were 24.2 times more likely to have eaten mild salsa with green onions than were people who dined at the same restaurant but did not develop the disease.

Case-control studies depend on the ability of people to remember key historical details accurately such as what they ate. For the outbreak of hepatitis A, the investigators obtained a food history for a period of two to six weeks before onset of symptoms because the incubation period of hepatitis A is long. For this reason, investigators need to move quickly to develop and administer outbreak questionnaires. Investigators follow standard methods of interviewing to minimize bias in how they ask questions and to minimize recall and prevarication bias on the part of the interviewee (Kalter, 1992).

4.3. Outbreak Characteristics

This section discusses how investigators elucidate the following outbreak characteristics: biological agent; source; route of transmission; size; and, when the disease is new or unusual, the disease process itself.

4.3.1. Biological Agent

The causative biological agent is perhaps the single most important characteristic of an outbreak. It has immediate implications for treating the sick and focuses the search for the source and route of transmission as each biological agent has propensities and limitations in the environments in which it can reside and the mechanisms by which it can be transmitted. The investigators of Legionnaire's disease, AIDS, mad cow disease, Lyme disease, and Nipah virus did not know the causative biological agent and had great difficulty finding the sources and routes of transmission.

Although the biological agent is often known at the time that an outbreak is detected, for diseases that have recently crossed species or for rare diseases that clinicians do not routinely test for, it may not be known. Importantly, the recent trend toward monitoring surveillance data of lower diagnostic precision (e.g., sales of diarrhea remedies or numbers of individuals with flulike symptoms) has increased the number of situations in which the biological agent is not known when an outbreak is detected. In these situations, the differential diagnosis may be large (Table 3.2). When the biological agent is not known, investigators use the clinical symptoms of affected individuals to select laboratory tests to narrow down and ultimately identify the biological agent.

A significant amount of laboratory work may be required to identify the biological agent. As in the case of Legionnaire's disease, Lyme disease, and Nipah virus in which the organism was previously unknown, it may take considerable time to isolate the organism. Identification of a difficult-to-identify organism is largely a process of elimination. Laboratories use cultures, serological tests, immunohistochemistry, and nucleic acid probes to search for known organisms that are most

TABLE 3.2 Biological Agents and Toxins of Concern for a Large-Scale Aerosol Release

Biological Agent


Early Clinical Presentation


Swine Influenza

Swine Influenza

SWINE INFLUENZA frightening you? CONCERNED about the health implications? Coughs and Sneezes Spread Diseases! Stop The Swine Flu from Spreading. Follow the advice to keep your family and friends safe from this virus and not become another victim. These simple cost free guidelines will help you to protect yourself from the swine flu.

Get My Free Ebook

Post a comment