Step 4: Define and Identify Cases

Establish a case definition. Your next task as an investigator is to establish a case definition, or a standard set of criteria for deciding whether, in this investigation, a person should be classified as having the disease or health condition under study. A case definition usually includes four components:

1. clinical information about the disease, 
2. characteristics about the people who are affected, 
3. information about the location or place, and 
4. a specification of time during which the outbreak occurred.

You should base the clinical criteria on simple and objective measures. For example, you might require the presence of an elevated level of antibody to the disease agent, the presence of a fever of at least 101^"F, three or more loose bowel movements per day, or muscle aching severe enough to limit the patient's activities. Regarding the characteristics of people, you might restrict the definition to those who attended a wedding banquet, or ate at a certain restaurant, or swam in the same lake. By time, the criterion might be onset of illness within the past 2 months; by place, it might be living in a nine-county area or working at a particular plant. Whatever your criteria, you must apply them consistently and without bias to all of the people included in the investigation.

Ideally, your case definition should be broad enough to include most, if not all, of the actual cases, without capturing what are called "false-positive" cases (when the case definition is met, but the person actually does not have the disease in question). Recognizing the uncertainty of some diagnoses, investigators often classify cases as "confirmed," " probable," or "possible."

To be classified as confirmed, a case usually must have laboratory verification. A case classified as probable usually has the typical clinical features of the disease without laboratory confirmation. A possible case usually has fewer of the typical clinical features. For example, in an outbreak of bloody diarrhea and severe kidney disease (hemolytic-uremic syndrome) caused by infection with the bacterium E. coli O157:H7, investigators defined cases in the following three classes:

• Confirmed case: E. coli O157:H7 isolated from a stool culture or development of hemolytic-uremic syndrome in a school-aged child resident of the county and who had gastrointestinal symptoms beginning between Nov. 3 and Nov. 8, 1990;
• Probable case: Bloody diarrhea (but no culture), with the same person, place, and time restrictions;
• Possible case: Abdominal cramps and diarrhea (at least three stools in a 24-hour period) in a school-age child resident of the county with onset during the same period (CDC, unpublished data, 1991).

Early in an investigation, a loose case definition that includes confirmed, probable, and even possible cases is often used to allow investigators to capture as many cases as possible. Later on, when hypotheses have come into sharper focus, the investigator may tighten the case definition by dropping the "possible" category. This strategy is particularly useful when you have to travel to different hospitals, homes, or other places to gather information, because it keeps you from having to go back for additional data. This illustrates an important axiom of field epidemiology: "Get it while you can."

Identify and count cases
As noted above, many outbreaks are first recognized and reported by concerned health care providers or citizens. However, the first cases to be recognized usually are only a small proportion of the total number. As a Disease Detective investigating an outbreak, you must therefore "cast the net wide" to determine the true size and geographic extent of the problem.

When identifying cases, you should use as many sources as you can, and you may need to be creative and aggressive in identifying these sources. Initially, you may want to direct your case finding at health care facilities where the diagnosis is likely to be made; these facilities include physicians' offices, clinics, hospitals, and laboratories. You also may decide to send out a letter describing the situation and asking for reports (passive surveillance); or you may decide to telephone or visit the facilities to collect information (active surveillance).

In some outbreaks, public health officials may decide to alert the public directly, usually through the local media. For example, in outbreaks caused by a contaminated food product such as salmonellosis caused by contaminated milk (7) or L-tryptophan-induced EMS (8), announcements in the media have alerted the public to avoid the implicated product and to see a physician if they had symptoms of the disease.

If an outbreak affects a population in a restricted setting, such as a cruise ship, school, or worksite, and if a high proportion of cases are unlikely to be diagnosed (if, for example, many cases are mild or asymptomatic), you may want to conduct a survey of the entire population. In such settings, you could administer a questionnaire to determine the true occurrence of clinical symptoms, or you could collect laboratory specimens to determine the number of asymptomatic cases. Finally, you can ask people who are affected if they know anyone else with the same condition.

Regardless of the particular disease you are investigating, you should collect the following types of information about every person affected:

• Identifying information: This may include name, address, and telephone number and allows you and other investigators to contact patients for additional questions and to notify them of laboratory results and the outcome of the investigation. Addresses also allow you to map the geographic extent of the problem.
• Demographic information: This may include age, sex, race, and occupation and provides the details that you need to characterize the population at risk.
• Clinical information: This information allows you to verify that the case definition has been met. Date of onset allows you to create a graph of the outbreak. Supplementary clinical information may include whether the person was hospitalized or died and will help you describe the spectrum of illness.
• Risk factor information: Information about risk factors will allow you to tailor your investigation to the specific disease in question. For example, in an investigation of hepatitis A, you would look at exposure to food and water sources.

Traditionally, we collect the information described above on a standard case report form, questionnaire, or data abstraction form. We then abstract selected critical items in a table called a "line listing." In a line listing, each column represents an important variable, such as name or identification number, age, sex, and case classification, while each row represents a different case, by number. New cases are added to a line listing as they are identified. This simple format allows the investigator to scan key information on every case and update it easily. Even in the era of microcomputers, many epidemiologists still maintain a hand-written line listing of key data items and turn to their computers for more complex manipulations of data. Here is a portion of a line listing that might have been created for an outbreak of hepatitis A.

S*=Sclera;, N=Nausea; V=Vomiting; A=Anorexia; F=Fever; DU=Dark urine; J=Jaundice; HAIgm=Hepatitis AIgM antibody test

Back to Top

 
Step 5: Describe and Orient the Data in Terms of Time, Place, and Person

Once you have collected some data, you can begin to characterize an outbreak by time, place, and person. In fact, you may perform this step several times during the course of an outbreak. Characterizing an outbreak by these variables is called descriptive epidemiology, because you describe what has occurred in the population under study. This step is critical for several reasons. First, by becoming familiar with the data, you can learn what information is reliable and informative (e.g., the same unusual exposure reported by many of the people affected) and what may not be as reliable (e.g., many missing or "don't know" responses to a particular question). Second, you provide a comprehensive description of an outbreak by showing its trend over time, its geographic extent (place), and the populations (people) affected by the disease. This description lets you begin to assess the outbreak in light of what is known about the disease (e.g., the usual source, mode of transmission, risk factors, and populations affected) and to develop causal hypotheses. You can, in turn, test these hypotheses using the techniques of analytic epidemiology described later in Step 7: Evaluate Hypotheses.

Note that you should begin descriptive epidemiology early and should update it as you collect additional data. To keep an investigation moving quickly and in the right direction, you must discover both errors and clues in the data as early as possible.

Characterizing by time
Traditionally, we show the time course of an epidemic by drawing a graph of the number of cases by their date of onset. This graph, called an epidemic curve, or "epi curve" for short, gives a simple visual display of the outbreak's magnitude and time trend. The following example depicts the first outbreak of Legionnaires disease, in Philadelphia, Pennsylvania, in 1976.

Insert EPI Curve
An epidemic curve provides a great deal of information. First, you will usually be able to tell where you are in the course of the epidemic, and possibly to project its future course. Second, if you have identified the disease and know its usual incubation period, you may be able to estimate a probable time period of exposure and can then develop a questionnaire focusing on that time period. Finally, you may be able to draw inferences about the epidemic patternfor example, whether it is an outbreak resulting from a common source exposure, from person-to-person spread, or both.

How to draw an epidemic curve
To draw an epidemic curve, you first must know the time of onset of illness for each person. For most diseases, date of onset is sufficient; however, for a disease with a very short incubation period, hours of onset may be more suitable. The number of cases is plotted on the y-axis of an epi curve; the unit of time, on the x-axis. We usually base the units of time on the incubation period of the disease (if known) and the length of time over which cases are distributed. As a rule of thumb, select a unit that is one-fourth to one-third as long as the incubation period. Thus, for an outbreak of Clostridium perfringens food poisoning (usual incubation period 10-12 hours), with cases during a period of only a few days, you could use an x-axis unit of 2 or 3 hours. Unfortunately, there will be times when you do not know the specific disease and/or its incubation period. In that circumstance, it is useful to draw several epidemic curves, using different units on the x-axes, to find one that seems to show the data best. Finally, show the pre- and post-epidemic period on your graph to illustrate the activity of the disease during those periods.

Interpreting an epidemic curve
The first step in interpreting an epidemic curve is to consider its overall shape, which will be determined by the pattern of the epidemic (e.g., whether it has a common source or person-to-person transmission), the period of time over which susceptible people are exposed, and the minimum, average, and maximum incubation periods for the disease.

An epidemic curve with a steep up slope and a gradual down slope, such as the illustration above on the first outbreak of Legionnairesdisease, indicates a single source (or "point source") epidemic in which people are exposed to the same source over a relatively brief period. In fact, any sudden rise in the number of cases suggests sudden exposure to a common source. In a point source epidemic, all the cases occur within one incubation period. If the duration of exposure is prolonged, the epidemic is called a "continuous common source epidemic," and the epidemic curve will have a plateau instead of a peak. Person-to-person spread (a "propagated" epidemic) should have a series of progressively taller peaks one incubation period apart.

Cases that stand apart (called "outliers") may be just as informative as the overall pattern. An early case may represent a background (unrelated) case, a source of the epidemic, or a person who was exposed earlier than most of the people affected (e.g., the cook who tasted her dish hours before bringing it to the big picnic). Similarly, late cases may be unrelated to the outbreak, may have especially long incubation periods, may indicate exposure later than most of the people affected, or may be secondary cases (that is, the person may have become ill after being exposed to someone who was part of the initial outbreak). All outliers are worth examining carefully because if they are part of the outbreak, their unusual exposures may point directly to the source. For a disease with a human host such as hepatitis A, for instance, one of the early cases may be in a food handler who is the source of the epidemic.

In a point-source epidemic of a known disease with a known incubation period, you can use the epidemic curve to identify a likely period of exposure. This is critical to asking the right questions to identify the source of the epidemic.

Characterizing by place
Assessment of an outbreak by place provides information on the geographic extent of a problem and may also show clusters or patterns that provide clues to the identity and origins of the problem. A simple and useful technique for looking at geographic patterns is to plot, on a "spot map" of the area, where the affected people live, work, or may have been exposed.

A spot map of cases in a community may show clusters or patterns that reflect water supplies, wind currents, or proximity to a restaurant or grocery store. On a spot map of a hospital, nursing home, or other such facility, clustering usually indicates either a focal source or person-to-person spread, while the scattering of cases throughout a facility is more consistent with a common source such as a dining hall. In studying an outbreak of surgical wound infections in a hospital, we might plot cases by operating room, recovery room, and ward room to look for clustering.

If the size of the overall population varies between the areas you are comparing, a spot map, because it shows numbers of cases, can be misleading. This is a weakness of spot maps. In such instances, you should show the proportion of people affected in each area (which would also represent the rate of disease or, in the setting of an outbreak, the "attack rate").

Characterizing by person
You determine what populations are at risk for the disease by characterizing an outbreak by person. We usually define such populations by personal characteristics (e.g., age, race, sex, or medical status) or by exposures (e.g., occupation, leisure activities, use of medications, tobacco, drugs). These factors are important because they may be related to susceptibility to the disease and to opportunities for exposure.

Age and sex are usually assessed first, because they are often the characteristics most strongly related to exposure and to the risk of disease. Other characteristics will be more specific to the disease under investigation and the setting of the outbreak. For example, if you were investigating an outbreak of hepatitis B, you should consider the usual high-risk exposures for that infection, such as intravenous drug use, sexual contacts, and health care employment.

Summarizing by time, place, and person
After characterizing an outbreak by time, place, and person, you need to summarize what you know to see whether your initial hypotheses are on track. You may find that you need to develop new hypotheses to explain the outbreak.

Back to Top

Step 6: Develop Hypotheses

In real life, we usually begin to generate hypotheses to explain why and how the outbreak occurred when we first learn about the problem. But at this point in an investigation, after you have interviewed some affected people, spoken with other health officials in the community, and characterized the outbreak by time, place, and person, your hypotheses will be sharpened and more accurately focused. The hypotheses should address the source of the agent, the mode (vehicle or vector) of transmission, and the exposures that caused the disease. Also, the hypotheses should be proposed in a way that can be tested.

You can develop hypotheses in a variety of ways. First, consider what you know about the disease itself: What is the agent's usual reservoir? How is it usually transmitted? What vehicles are commonly implicated? What are the known risk factors? In other words, simply by becoming familiar with the disease, you can, at the very least, "round up the usual suspects."

Another useful way to generate hypotheses is to talk to a few of the people who are ill, as discussed under Step 3: Verifying the Diagnosis. Your conversations about possible exposures should be open-ended and wide-ranging and not confined to the known sources and vehicles. Sometimes investigators meet with a group of the affected people as a way to search for common exposures. Investigators have even found it useful to visit the homes of people who became ill and look through their refrigerators and shelves for clues.

Descriptive epidemiology often provides some hypotheses. If the epidemic curve points to a narrow period of exposure, ask what events occurred around that time. If people living in a particular area have the highest attack rates, or if some groups with particular age, sex, or other personal characteristics are at greatest risk, ask why. Such questions about the data should lead to hypotheses that can be tested.

Back to Top | Step 7: Evaluate Hypotheses