Investigation of Enteric Disease Outbreaks

Most enteric illness is spread by the fecal-oral route, and the history of these diseases
mirrors changes in the social, economic, and physical environments in which we live. By one
estimate, nearly 90% of the global burden of diarrheal disease is attributable to unsafe water
supplies (19). Improved basic sanitation, medical care, and diagnostic capabilities have
helped substantially reduce both the risks for foodborne and waterborne diseases and the
frequency of complications overall. That said, new problems emerge that reflect changes in
both human and microbial populations—not to mention nonhuman reservoir species and the
environment. For example, the population of vulnerable persons is growing as people live
longer or live with immunosuppressive conditions. The rapid international movement of
people and foodstuffs, shifts in food production methods and eating habits, and other factors
all change patterns of food exposure and hence patterns of foodborne disease. Improved
diagnostic tests identify new or newly described pathogens. Many pathogens considered
important sources of morbidity today were unheard of even 40 years ago, including
noroviruses, rotaviruses, Cryptosporidium, Campylobacter, and Escherichia coli O157:H7.
Acute gastroenteritis and other foodborne diseases can be caused by bacteria, viruses,
protozoa, fungi, helminths, prions, and biological or environmental toxins. People are
exposed to these pathogens or toxins in contaminated food or water, by person-to-personcontact,
and by direct contact with animals. Many foodborne and waterborne diseases are
self-limited and characterized by gastrointestinal symptoms such as vomiting and diarrhea.
Others, however, may manifest with or progress to include systemic or neurological disease
that can result in substantial morbidity and mortality, e.g., infection with E. coli O157:H7
(causing hemolytic-uremic syndrome), Helicobacter pylori (linked to gastric
cancers), Listeria (meningitis and miscarriage), Salmonella(increased risk of reactive
arthritis), and Toxoplasma gondii (birth defects).

DISEASE SURVEILLANCE

Enteric disease surveillance is generally the province of public health agencies. Surveillance
includes the collection and analysis of information about disease occurrence and leads to
taking considered action based upon those data. Disease surveillance often is based on
mandatory reporting laws, whereby diagnostic laboratories or clinicians are required to notify
public health agencies about individuals with specified conditions, e.g., salmonellosis or
hepatitis A, as well as unusual clusters of illness.
While conceptually simple, disease surveillance can be a daunting logistical challenge that
places demands on legal, medical, communications, transportation, scientific, and social
infrastructure. Unless otherwise noted, the following discussion pertains to enteric disease
surveillance and outbreak investigation in the United States.
In the United States and many other countries, public health workers collate notifiabledisease
reports in search of broader patterns, often collecting additional information from
patient interviews and other sources. The general public may also contact public health
agencies directly with concerns about individual or apparently clustered illnesses, and these
reports too must be considered for potential significance.
While definitions and usage vary, disease outbreaks in general are comprised of individual
illnesses that can be connected by webs of transmission or by exposures to common sources.
For example, we can refer to an outbreak of norovirus infections at a nursing home, an
outbreak of cholera affecting the population of a region or country, or an outbreak of
salmonellosis resulting from consumption of contaminated peanut butter. In the United
States, over 95% of reported enteric illnesses are not recognized to be part of outbreaks
(12). That said, outbreak-related cases play a disproportionately large role in our
understanding of pathogen transmission and attempts to control it.
Since 1938, the U.S. Public Health Service has collated reports on foodborne and waterborne
disease outbreaks from public health agencies (4). What was once a flow of paper has largely
become an electronic reporting system. From 2004 to 2006, the CDC received~1,200 reports
of foodborne outbreaks involving at least 24,000 people annually
(http://www.cdc.gov/outbreaknet/). Examples of forms, an electronic database used for
foodborne disease outbreaks, and recent data are accessible online
(http://www.cdc.gov/outbreaknet/). Summary information about foodborne and waterborne
outbreak investigations is typically reported to the CDC, which periodically summarizes these
data.
It should be emphasized that the quality of these data is highly variable, which complicates
our ability to summarize them meaningfully. Most outbreaks are probably never recognized
or reported, much less investigated, and those that are reported nationally are likely to be a
biased sample of the whole. For example, large outbreaks or outbreaks involving more
severe illnesses or deaths are much more likely to get a thorough investigation. In the
United States, over half of reported foodborne disease outbreaks are associated with food
eaten outside the home (6). Many investigations are inconclusive, and different people may
summarize similar situations in different ways.
Pathogens most commonly identified in outbreaks reported from 2001 to 2006 are shown
in Fig. 1. It is noteworthy that no etiology was identified for almost one-third of foodborne
outbreaks during that period. Noroviruses were the leading cause of half of those with a
confirmed etiology, and bacteria were responsible for 38%.

There are many reasons to investigate outbreaks. The most obvious is to arrest an ongoing
problem. Recalling a contaminated product from the marketplace or excluding a typhoid
carrier from work as a food handler can prevent illness and even save lives. Given the
inherent delays in disease reporting, many outbreaks are over before public health agencies
learn about them; the benefits of investigating these clusters are more indirect.
Investigations often reveal systemic problems that could cause illness in the future;
correction of these problems reduces those risks. Examples might include substandard
operation of a drinking-water treatment facility or inadequate temperature control in a
peanut roasting operation. Outbreaks also provide an opportunity for public health
education—education that may be targeted to the individual who manages a restaurant or to
the broader community through media contacts. The identification or better characterization
of the root causes of outbreaks, often assessed as aggregate data from many investigations,
also helps drive the agenda of regulatory agencies, businesses, and others trying to develop
more effective practices and regulations.
Public health agencies in most countries have a legal authority and responsibility to
investigate certain kinds of diseases and most clusters of illness (i.e., outbreaks) (Table 1).
In the United States, that legal authority rests primarily at the local and state levels, and the
vast majority of outbreaks are investigated by agencies at those levels. Suspected disease
outbreaks must be reported to the public health department in most states, though this
requirement is often unappreciated or ignored by clinicians and others.
a This table provides a general outline of typical responsibilities for different agencies
involved in investigating food- and waterborne outbreaks. All states have unique food and
water safety laws, policies, and organizational structures that will affect investigations, and
many other agencies and organiza-tions may play important roles in certain situations.
Additional information is available at the “Gateway to Government Food Safety Information”
(http://www.foodsafety.gov/) and the Environmental Protection Agency website
(http://www.epa.gov/ebtpages/water.html).
To thoroughly investigate an outbreak, epidemiologists need to collect a diverse range of
information. Public health authorities have substantial legal authority to gather such data
that usually goes hand in hand with state laws protecting the confidentiality of information
collected in the context of an investigation. In practice, however, sweeping legal authorities
are rarely invoked; epidemiological investigations typically rely on the voluntary cooperation
of all parties. Regulatory agencies such as the FDA and the USDA have clearer police powers
that can be utilized when circumstances demand them, but these powers typically apply only
to commercial entities.
The capacity to conduct public health investigations varies with the resources, experience,
and interest of the agencies involved and in practice is highly variable. Moreover, some
states are hobbled by a cumbersome legal structure that makes it difficult to move quickly in
a coordinated fashion. The success of many multistate investigations in the United States
depends to a surprising degree on the involvement of one or more of a small number of
states with expertise in outbreak investigation. Realistically, not all clusters merit a thorough
investigation; resources are always finite. One must consider the severity of the disease, the
community affected, and the likelihood that useful data can be obtained from an
investigation. Basic laboratory testing and interviews of a limited number of ill persons may
provide sufficient information to identify the source of the outbreak and take appropriate
control measures. In many cases, however, more extensive epidemiological investigation is
indicated.

Outbreak Detection

A foodborne outbreak can be defined as “the occurrence of two or more cases of a similar
illness resulting from ingestion of a common food” (4). There are analogous surveillance
definitions for outbreaks due to drinking or recreational water (23, 24). In practice, most
intrahousehold clusters are not pursued unless the disease is severe. To recognize unusual
clusters of illness, public health officials must have knowledge of what “normal” or baseline
rates of a disease are in the affected community. These may be available from historical
surveillance data.
In general, potential outbreaks come to the attention of health authorities through one of
three routes. First, people often contact public health agencies directly to report illness
clusters, e.g., after a common restaurant meal or a wedding, or among persons at a nursing
home, school, or prison. These reports come from physicians, infection control practitioners,
institutional staff, and often from private citizens. Second, public health agencies identify
clusters through review and follow-up of routinely collected surveillance information,
including reports of legally notifiable diseases and subtyping of specimens submitted to
public health laboratories. Third, recognition may come in response to queries from other
public health agencies doing “case finding” as part of investigations in their jurisdictions.
One of the first responses to the recognition of a potential outbreak is assessing whether
something “real” has actually occurred. Surveillance reports are often incomplete,
misleading, or erroneous, and many “clusters” dissolve under scrutiny. Outbreak clusters are
often obvious, but sometimes it is difficult to separate minor fluctuations in rates (i.e.,
“noise”) from true increases due to a common-source outbreak (“signal”).

Laboratory Investigation

The response to preventive measures and the treatment of clinical infections varies markedly
depending on the etiologic agent involved. It is therefore concerning that the etiologic agents
were not identified for half of foodborne disease outbreaks reported to the CDC from 2004 to
2006. While many factors can impede an investigation, laboratory testing is usually
necessary to confirm an etiology. Stool specimens were not collected for laboratory testing in
two-thirds of foodborne disease outbreaks of unknown etiology occurring at seven sites in
1998 and 1999 (12).
To meet most definitions of a confirmed outbreak, the etiologic agent should be isolated from
the stool of two or more ill persons or from the epidemiologically implicated food (Table 3).
In a few situations, such as mushroom poisoning, ciguatera fish poisoning, or other chemical
intoxications, it is sufficient to document the clinical syndrome among affected
persons. Staphylococcus can also be problematic because the organism may not be viable in
stool or food samples, and most laboratories cannot test for enterotoxin. Thus, investigators
must collect sufficient numbers of specimens (potentially including stool and other clinical
specimens and also food, water, or environmental specimens) and handle them appropriately
to ensure that laboratory testing identifies the etiologic agent. Investigators should consult
early with their public health laboratory regarding appropriate collection and testing of
samples.

Most private clinical laboratories currently cannot test specimens for norovirus, which is the
most common cause of foodborne disease in the United States. Most state health department
laboratories offer PCR (reverse transcriptase PCR [RT-PCR]) testing for norovirus and can
help coordinate appropriate testing of specimens if this agent is suspected as the cause of an
outbreak. Likewise, few private laboratories can serotype Salmonella isolates or definitively
identify enterotoxigenic E. coli, non-O157:H7 Shiga-toxigenic E. colistrains, or staphylococcal
enterotoxin.
Guidelines for collecting appropriate specimens during an outbreak investigation are listed
in Table 4. It can be difficult to collect adequate specimens for laboratory testing. In general,
the concentration of etiologic agents decreases with time after onset, putting a premium on
prompt specimen collection, but many pathogens are sometimes detectable days or even
weeks after symptoms resolve. Investigators may need to convince reluctant persons of the
social benefits of providing stool specimens and will need to arrange the necessary logistics,
including distribution of collection materials, aliquoting, and transportation. Investigators
should promptly contact private laboratories that may have received specimens from
outbreak-associated patients to ask that the original material be held for possible additional
testing.
b Wrap the packaged samples in sealed, waterproof containers (i.e., plastic bags). Label each
specimen container in a waterproof manner. Batch the collection and send in overnight mail
to arrive at the testing laboratory on a weekday during business hours unless other
arrangements have been made in advance with the testing laboratory. Contact the testing
laboratory before shipping, and give the testing laboratory as much advance notice as
possible so that testing can begin as soon as samples arrive. When etiology is unclear and
syndrome is nonspecific, consider collecting all four types of specimens.
c For more detailed instructions on how to collect specimens for specific parasites, please go
tohttp://www.dpd.cdc.gov/dpdx/.
d For more detailed instructions on how to collect specimens for viral testing, please go
tohttp://www.cdc.gov/mmwr/PDF/RR/RR5009.pdf.
e The containers have been tested for the presence of the chemical of interest before use.
f Unused specimen collection containers that have been brought in to the field and subjected
to the same field conditions as the used containers. These containers are then tested for
trace amounts of the chemical of interest.
Where applicable, samples of suspected vehicles of infection (e.g., food or water) should be
collected as soon as possible after an outbreak is recognized. Actual testing is not always
practical, but getting specimens preserves the option of later testing. Ideally, investigators
will collect specimens from the batches or lots of food or water that patients actually ate or
drank before becoming ill. If this is not possible (as is often the case), products as similar as
possible are the next best thing. If investigators suspect that the source of an outbreak is
contaminated packaged food, they should collect unopened packages from the implicated lot.
Positive (and sometimes negative) specimens are often critical evidence in outbreaks caused
by commercial products.
Laboratory subtyping is critical to many outbreak investigations, particularly those with
bacterial etiologies. “Subtyping” is a generic term referring to any method that improves the
specificity of the description of an etiologic agent. For example, a Salmonella isolate could be
characterized by serotype (e.g., Enteritidis or Heidelberg), pulsed-field gel electrophoresis
(PFGE) or other restriction fragment length polymorphism patterns, multilocus variable
number tandem repeat analysis (MLVA) pattern, DNA sequence, or other characteristics.
When methodologies are standardized, subtyping data can be shared to identify potential
matches between isolates at different laboratories. PulseNet
(http://pulsenetinternational.org) is an example of an international network that effectively
shares such information. Specific subtyping can be invaluable: two salmonellosis reports in
the same week may be normal, but two Salmonella enterica serovar Hvittingfoss cases in the
same week in most jurisdictions would be an unlikely coincidence. On closer inspection,
those matching isolates might turn out to be from the same household (or even from the
same individual), but such matches are often the first indication of an outbreak. Subtype
matches not only help link scattered cases but also can provide a way to exclude similar but
perhaps unrelated cases (Fig. 4). Refining case definitions in this manner can greatly
increase the statistical power of an analytic epidemiological study.
Some outbreaks involve multiple variants of a pathogen. Investigations may start because of
an increase in one PFGE pattern, but multiple patterns may eventually be recognized to be
part of the same cluster, reflecting the genetic diversity or microevolution of the etiologic
agent in vivo. Outbreaks that involve gross environmental contamination may even include
multiple species (13, 18).

EPIDEMIOLOGICAL APPROACHES TO OUTBREAK
INVESTIGATIONS

While there is much variation in practice, there is a general logic that underlies most
successful outbreak investigations. These steps are outlined in Table 5. Depending on
circumstances, some steps may be more implicit than explicit, and they do not necessarily
occur in a neat sequence.

In any investigation, the most immediate priorities are to implement appropriate measures
to control the outbreak, mitigate associated morbidity, and prevent recurrences. When
outbreaks appear to be ongoing, public health officials may institute substantial and
occasionally controversial control measures before the investigation is complete and before
all desired data are available. Such measures may include ordering (or recommending)
product recalls, confiscating products, excluding food handlers or ill persons from work,
closing retail food establishments or implicated venues, and publicly notifying persons who
may have been exposed. Because such interventions can have important medical, emotional,
and economic implications, the larger investigative team must assimilate available data
rapidly and communicate effectively with each other and the public. Such collaboration is
imperative to ensure that the public health benefits are maximized while the collateral
damage is minimized.
Good communication among epidemiologists, laboratorians, environmentalists, and other
partners is essential to successful investigations. Most health department jurisdictions have
environmental health specialists (also known as sanitarians or environmentalists) or
regulatory staff with expertise in the technical aspects of food handling, inspection of food
establishments, tracing of food distribution, environmental specimen collection, water safety,
and other issues that are often critical to successful investigations.
Epidemiologists use case definitions to consistently include or exclude people from a cluster
or study group. Case definitions typically include criteria for symptoms, time of onset, and
the time and place of potential exposure, e.g., “any person reporting vomiting or diarrhea
within 5 days of consuming food from restaurant A from 2 to 5 March” or “any U.S. resident
with an isolate of Salmonella enterica serovar Rissen and PFGE pattern XYZ1234 reported in
2010.” Case definitions are investigative constructs, not biological verities, and may be
modified during the course of an investigation as the need arises; there can even be multiple
case definitions serving different purposes.
Epidemiologists collect clinical information about potential cases, including signs and
symptoms, onset time, and indices of severity such as duration of illness, hospitalization, and
fatalities. Individual case data are often displayed in a spreadsheet or database layout as a
“line list,” and epidemiologists pore over these in search of patterns. When outbreaks
manifest by matching laboratory isolates (e.g., E. coli O157:H7 cases with the same PFGE
pattern), the pathogen is known at the outset, and there may be less interest in some
clinical details. Investigations that stem from citizen reports, on the other hand, often rely on
thorough symptom profiles to classify potential cases.
After confirming the existence of an outbreak, investigators interview affected persons to
identify demographic characteristics of cases, the nature and timing of symptoms, and
potential exposures of interest. Systematically collected information is much more useful
than desultory anecdotes. Well-designed forms and questionnaires—and skilled
interviewers—provide structure to that information and are essential to transforming raw
reports into analyzable data. Good questionnaire design requires training, insight, and
experience and can be surprisingly time-consuming. Outbreak investigations often move
quickly, and template questionnaires developed in advance that can be quickly modified to fit
a given situation can be very useful. The best templates allow for quick and flexible
modification, deployment in multiple modes (e.g., paper and electronic) for a variety of
audiences, rapid data entry, and easy abstraction of data for tabulation and statistical
analyses.
A careful consideration of the cases’ demographics (age, sex, race, and ethnicity) can narrow
the range of plausible vehicles. For example, salmonellosis outbreaks caused by fresh
produce (e.g., lettuce, spinach, and sprouts) typically manifest with disproportionate
numbers of female patients in the age group from 20 to 50. Outbreaks focused among young
children are less likely to be caused by vegetables. The spatial and temporal distribution of
cases also is instructive. Illnesses scattered across the country are unlikely to be caused by
products with local or regional distribution (e.g., milk). Products with a short shelf life (e.g.,
fresh produce) tend to cause shorter-lived outbreaks than processed foods with a long shelf
life (e.g., peanut butter).
Different kinds of outbreaks call for different types of questionnaires. Even in a single
investigation, more than one questionnaire may prove useful as things progress. The desire
to be comprehensive must be balanced by practical considerations, including ease of use and
acceptability to interviewees and interviewers. Complete answers to reasonably limited
questions are generally preferred over incomplete answers to an unreasonable number.
Questionnaire designers should be mindful of the population for which it is intended, the
mode of administration, the sophistication of those collecting the data, and —often
overlooked—how data will be entered and analyzed. Questionnaires can be deployed via
telephone, in a face-to-face interview, or as self-administered paper or electronic forms;
each choice has advantages and disadvantages. Interviewers must know how to introduce
the study to participants and answer questions about it. Data collectors must also be given
guidelines for obtaining data and be familiar with standard definitions so that information is
elicited consistently and completely, thereby minimizing errors and potential bias.
Simple event-centric outbreaks (e.g., church suppers, wedding receptions, and many
restaurant clusters) lend themselves to short questionnaires. Menus are obtained, other
exposures (e.g., water) are assessed, and with a good template, questionnaire design is
often straightforward. When cases are more scattered in place and time, suspicions may
arise about commercially distributed food products. These can be complex, multijurisdictional
investigations with large public health and economic consequences. Such investigations often
evolve over weeks or even months, with different questionnaires used at different stages of
the process. Broad hypothesis-generating questionnaires may be used initially to identify
food items or other exposures that deserve additional scrutiny. There are several models for
approaching these kinds of outbreaks that have proven effective in practice (8, 21).
Interviews with persons who were not sick (controls) are often necessary to provide a
comparison group. If the population affected by the outbreak is well defined, such as
attendees of a church picnic or patrons of a single restaurant within a 3-day period, a
retrospective cohort study can be performed. In this situation, all members or a
representative sample of the affected group (the cohort) are interviewed to assess whether
they were ill and what items they consumed. Rates of illness among persons with and
without certain exposures are compared with appropriate statistical methods to identify
exposures associated with illness (3,10). If the affected population is not well defined or a
cohort study is not practical, a case-control study may be performed. In such outbreaks,
persons with the illness of interest (cases) are compared to those without it (controls). In the
absence of formal control data, other information may be better than nothing in assessing
case exposure data: restaurant service records (e.g., how many people got salads on
Monday night), brand name market share data, or even available survey data (<3% of
people drink unpasteurized milk in any given week).
To keep the investigation focused, data about cases should be regularly organized,
summarized, and shared with colleagues. The occurrence of new cases over time (usually by
day or week of symptom onset) can be represented graphically as an “epidemic curve” (Fig.
5 and 6), and the geographic distribution of cases can be plotted on a map. The line list, the
epidemic curve, and the map can give the investigator clues about the source of an
outbreak. Localized cases may reflect an exposure at a single time and place (a “point
source” outbreak), e.g., patrons of a restaurant on a single day; such outbreaks are often
associated with mishandling, undercooking, cross-contamination, or poor hygiene locally. In
contrast, cases widely scattered geographically suggest the distribution of a commercial
product (or a convergence at a national meeting, event, or vacation venue).

No comments:

Post a Comment