Page of

Outbreak investigations 

Outbreak investigations
Outbreak investigations
Infectious Disease Epidemiology (Oxford Specialist Handbooks)

Clarence Tam

and Walter Haas


Outbreak investigations

The investigation of outbreaks requires specific expertise and is a major task of infectious disease epidemiology. To ensure the best public health outcome during acute outbreak situations, a systematic approach to their timely detection, assessment, investigation, and control is required, together with rapid collation and rigorous interpretation of multiple sources of often imperfect evidence.

This chapter describes the basic principles of outbreak investigations. The epidemiological features of different types of outbreaks are described, together with the use of surveillance for outbreak detection and case ascertainment, the role of the outbreak control team, the major features of epidemiological and environmental investigations, issues in the interpretation of evidence, and considerations in the implementation and evaluation of control measures.

What is an outbreak?

An outbreak is often defined as an increase in incidence of a disease above expected levels in a particular location or population in a given time period. Another common definition is the occurrence of a disease in two or more epidemiologically linked individuals, such as those with a confirmed common source of infection.

Types of outbreaks and epidemic curves

Outbreaks can be common-source, propagated, or both. In common-source outbreaks, a population is exposed to a common source of contamination. Exposure may be restricted to a particular event or point in time (a point-source outbreak), intermittent over an extended period, or continuous. In propagated outbreaks, the infection is transmitted from person to person directly (e.g. influenza) or indirectly (e.g. mosquito-borne transmission of dengue virus (DENV)). Propagated outbreaks can include several waves of transmission resulting from secondary and tertiary spread. A mixed outbreak occurs when a common-source outbreak involves secondary person-to-person spread; this is common for pathogens transmissible through food and the faeco–oral route such as norovirus, hepatitis A virus, and Shigella.

The epidemic curve—a plot of the number of new outbreak cases by the time/day of illness onset—is a crucial feature of any outbreak investigation and can provide valuable information on the type of outbreak, mode of transmission, incubation period, exposure period, and transmission potential of the infection (Figure 3.1). The steps required to investigate an outbreak are summarized in Box 3.1.

Figure 3.1 Types of outbreaks. Point-source outbreaks: if the exposure period is known, the epidemic curve enables the calculation of the modal incubation period and range, which can be useful for narrowing down the list of likely causal pathogens (a); if the incubation period is known, the exposure period can be inferred (b); extended common-source outbreak (c); mixed outbreak with initial point-source and subsequent propagation through person-to-person transmission (d).
Figure 3.1 Types of outbreaks. Point-source outbreaks: if the exposure period is known, the epidemic curve enables the calculation of the modal incubation period and range, which can be useful for narrowing down the list of likely causal pathogens (a); if the incubation period is known, the exposure period can be inferred (b); extended common-source outbreak (c); mixed outbreak with initial point-source and subsequent propagation through person-to-person transmission (d).
Figure 3.1 Types of outbreaks. Point-source outbreaks: if the exposure period is known, the epidemic curve enables the calculation of the modal incubation period and range, which can be useful for narrowing down the list of likely causal pathogens (a); if the incubation period is known, the exposure period can be inferred (b); extended common-source outbreak (c); mixed outbreak with initial point-source and subsequent propagation through person-to-person transmission (d).
Figure 3.1 Types of outbreaks. Point-source outbreaks: if the exposure period is known, the epidemic curve enables the calculation of the modal incubation period and range, which can be useful for narrowing down the list of likely causal pathogens (a); if the incubation period is known, the exposure period can be inferred (b); extended common-source outbreak (c); mixed outbreak with initial point-source and subsequent propagation through person-to-person transmission (d).

Figure 3.1
Types of outbreaks. Point-source outbreaks: if the exposure period is known, the epidemic curve enables the calculation of the modal incubation period and range, which can be useful for narrowing down the list of likely causal pathogens (a); if the incubation period is known, the exposure period can be inferred (b); extended common-source outbreak (c); mixed outbreak with initial point-source and subsequent propagation through person-to-person transmission (d).

Outbreak surveillance and detection

Health authorities may receive reports of suspected outbreaks from numerous sources, including the public, the media, clinicians, and other parts of the health system. Routine surveillance data, including notifications of infectious diseases and laboratory reports of microbiological diagnoses, can aid detection of outbreaks. Surveillance data are generally more reliable than reports from the public and media, as they are collected systematically and have a clinical and/or microbiological diagnosis. Additional information on antimicrobial susceptibility or other phenotypic or genetic characteristics can help to identify an outbreak caused by a specific microbial strain.

Surveillance data also enable the comparison of case incidence over a specified time against a baseline, which might indicate an increase in incidence above expected levels. However, the ‘expected’ level of disease may be unclear for rare diseases, while detecting clusters of common diseases over and above the background level may be difficult. In addition, surveillance data sources usually capture only a fraction of all cases in the population, may miss localized outbreaks, and lack the ‘on-the-ground’ context and detail that clinicians treating patients in hospitals or primary care can provide.

Investigating every potential outbreak report is unfeasible, so systematic collation and review of reports are important to prioritize those that require further investigation.

Assessing the public health impact

Assessing the public health threat is crucial to prioritize responses to specific outbreaks and assign adequate urgency and resources. Important considerations include disease severity (e.g. high risk of mortality), setting (e.g. vulnerable hospitalized populations), outbreak size, transmission potential, and feasibility of containment. Annex 2 of the IHR1 provides a decision tree to judge the risk of international spread that would require immediate (<24 hours) communication to the WHO and collaboration with international counterparts for a Public Health Event of International Concern (PHEIC).

Special urgency is warranted if the risk of infection is ongoing and the outbreak source has not been identified or where specific interventions could curb exposure, such as ring vaccination in a measles outbreak or decommissioning and disinfection of water systems to prevent spread of Legionella. Political interest and implications can also play a role.

Prioritizing infectious diseases helps to allocate surveillance resources and support assessment of the potential impact of an outbreak based on known disease properties.2 Emerging pathogens, such as influenza A H7N9 and MERS-CoV, pose a special challenge, because knowledge of their epidemic potential is limited at present, so even a single case of imported disease usually warrants thorough investigation to exclude its involvement in an outbreak3. Outbreak investigations Chapter 6 outlines the investigation of emerging infection outbreaks.

Convening an outbreak control team

Investigation of outbreaks usually requires a multidisciplinary approach, and its coordination is the responsibility of an outbreak control team (OCT). The public health authority in the locality where the outbreak is first detected is generally responsible for evaluating the situation, convening the OCT, and leading the investigation.

The team’s composition should be adapted to the specific outbreak, but it usually comprises a public health specialist or an epidemiologist with field experience, a clinician, a microbiologist, a communications officer, and, where relevant, a representative of the affected institution(s). Additional expertise might be required, e.g. if an environmental source is suspected. Not all of these individuals will be part of the core team, but frequent communication (initially once or twice daily) ensures that no information is lost. Decisions and actions taken by the OCT should be documented to facilitate the exchange of information. The OCT should assess early on whether regional- and national-level epidemiologists should be alerted, e.g. if widespread transmission is likely, and whether to bring in specialist expertise not available locally, such as a reference microbiologist.4 For suspected food-borne outbreaks, the food safety administration should be involved to initiate tracing of food products back through the supply chain.

Mechanisms for active, regular, and transparent communication with relevant authorities, the media, and the public should be established early on. The current status of the outbreak and investigation, affected groups, potential risks, and unknowns should be clearly communicated.

Case definitions

The OCT is responsible for developing clear and appropriate case definitions throughout the investigation. This ensures that cases are systematically identified and reported, and avoids expending resources investigating cases unrelated to the outbreak. Case definitions reflect a balance between sensitivity and specificity. An initial working case definition generally favours sensitivity over specificity, as the public health consequences of missing outbreak-related cases usually outweigh the resource implications of investigating cases not related to the outbreak. As the investigation develops and more information becomes available regarding the clinical and microbiological profile of the disease, the case definition will be tightened, with increasing focus on specificity. The case definition should be easy to interpret and implement, and should include key features of the disease, people affected, and epidemiological circumstances (time period, place, and potential exposure event).

Often information on each individual will be insufficient to classify them definitively as a case; microbiological confirmation may be lacking, and clinical signs and symptoms alone might lack specificity. A hierarchical set of definitions for confirmed, probable, and possible cases can be useful to classify cases based on the strength of evidence (Box 3.2). Where the disease has several clinical manifestations, definitions for a range of outcomes will be required.

Source: data from European Centre for Disease Prevention and control, Influenza case definitions, Copyright © European Centre for Disease Prevention and Control (ECDC) 2005−2015. Available from: Outbreak investigations

Increasingly, microbiological typing and genetic profiling is used to detect unusual strain clusters and improve specificity of case definitions based on genetic relatedness, antimicrobial resistance profiles, and other microbial phenotypes. WGS is a rapidly evolving technique for molecular typing and characterization of outbreak strains in the future. Outbreak investigations Chapters 9 and 10 provide further detail on these techniques.

Surveillance during an outbreak

Surveillance may need to be intensified to ascertain outbreak-related cases, particularly if the outbreak is widely disseminated, illness is severe, or the risk of onward transmission is high (see Outbreak investigations Chapter 2). The OCT should alert relevant public health authorities and clinicians and communicate case definitions and reporting procedures. Active surveillance may be required when timely and comprehensive case ascertainment is necessary (e.g. to initiate prompt treatment, containment, or contact tracing). The OCT will contact key personnel at predefined intervals to enquire about new cases of disease. Active surveillance is more feasible when cases are ascertained at a limited number of sites such as tertiary care centres or clinics within a defined location. A minimum set of information should be collected on each case using standardized forms, including contact details, age, sex, residential location, occupation, date of illness onset, laboratory confirmation, hospitalization, outcome, and common risk factors for the suspected pathogen.

Descriptive epidemiology

The epidemic curve should be updated daily to understand the course of the epidemic and to determine its magnitude, whether transmission is increasing or decreasing, and whether secondary or continuous transmission is occurring. The key clinical and epidemiological features of cases should be described, using the principles of person, place and time. The timing, location, and population affected can yield valuable clues regarding outbreak aetiology and control implications. Clustering of cases in time might point to a specific event at which transmission was initiated, such as a function or public gathering. Spatial clustering might indicate a common environmental exposure, while common characteristics of cases can indicate shared risk factors such as consuming food from the same venue, attending the same school, or belonging to a high-risk group.

Preliminary investigations

In-depth interviews with initial cases (or proxy respondents) can provide timely information to develop hypotheses regarding outbreak aetiology. Trawling questionnaires are often used to collect information on a comprehensive list of possible exposures before the onset of illness. Shared behaviours or exposures among initial respondents can lead to specific hypotheses and inform the development of standardized questionnaires for further investigations using analytical study designs. Careful consideration of the intended analysis is needed when designing the questionnaire to ensure that it captures sufficient breadth and detail on exposures, exposure information is specific to avoid misclassification biases, and adequate information is collected on likely confounders and other relevant features such as dose response, vaccination status, or use of personal protective equipment.

Contact tracing

Contact tracing serves to follow up exposed and potentially infected contacts to prevent further spread and to identify the source of infection. Exposure definitions should be standardized, and different risk categories are often assigned, based on relationships to the index case (e.g. household vs community contact) and the intensity and duration of contact. For TB, recommendations are fairly standardized internationally, and contact tracing is prioritized using a combination of diagnostic information (smear or culture positivity) and cumulative exposure time. The risk of contacts progressing to active disease is also considered.5

Environmental investigations

Environmental investigations are especially important for pathogens with non-human reservoirs of infection. For example, contamination of cooling systems with Legionella and aerosolized spread has caused large outbreaks, sometimes infecting people living several kilometres away from the source.6,7 Environmental investigations are also important for investigating nosocomial spread via fomites or contaminated medical devices—as in outbreaks among patients in neonatal intensive care. For food-borne outbreaks, inspection of food premises, supply chain integrity, and compliance with established food hygiene standards are integral to the investigation. Detailed knowledge of infection vehicles, pathogen habitats, and transmission routes is necessary to direct environmental investigations. In certain contexts, other information, such as meteorological data on temperature, humidity, and wind direction and velocity, can be useful. Consideration should also be given to the frequency, timing, and location of environmental sampling and the sensitivity of pathogen detection from these samples.

Analytical studies in outbreak investigations

In most instances, descriptive analysis of the data will suffice to inform outbreak control strategies and protect groups at highest risk. Where evidence from initial epidemiological and environmental investigations is insufficient to implement adequate control measures, the OCT may decide to conduct an analytical study. Such studies require considerable resources, so the purpose and aims of the study should be clearly defined and agreed by members of the OCT and other partners involved in the investigation. Analytical studies might be particularly justified when:

  • there is a public health imperative to identify the source of infection and prevent further cases

  • the investigation could yield novel information regarding the epidemiology or natural history of the disease such as a previously unrecognized risk factor, the effectiveness of an intervention, or the infection serial interval.

Analytical studies involve the use of a comparison group to establish associations between suspected exposures and illness. Cohort studies and case control studies are most commonly used. The choice of study design depends on the specific circumstances. Outbreak investigations Chapter 4 provides further details on these study designs, including their limitations.

Cohort studies

Cohort studies are particularly useful for investigating point-source outbreaks in well-defined, relatively small populations such as a food poisoning outbreak among guests at a catered event. Investigators would interview all attendees to collect information on foods consumed at the event and subsequent illness. Associations between exposures and illness are analysed using exposure-specific attack rates and ratios (Table 3.1).

Table 3.1 Food-specific attack rates (ARs) and attack rate ratios (ARRs) in an outbreak of Clostridium perfringens food poisoning among guests at a hotel dinner party

Food item

Ate food item

Did not eat food item




AR (%)



AR (%)

Beef stew
















Green salad
















Cured meat sausage








The food-specific attack rate (AR) is the risk of disease among individuals eating a particular food item (the percentage of individuals eating a particular food item who subsequently fell ill). This is compared with the risk of disease among those who did not eat that particular food by means of a food-specific attack rate ratio (ARR; actually a ratio of risks). In the example, in this table, the AR is 42/47 = 89% among those eating beef stew and 1/14 = 7% among those not eating beef stew. Comparing the two gives an ARR of 89/7 = 12.51. People eating the beef stew were therefore 12.51 times more likely to become ill than those who did not eat the stew. A food item with a high food-specific AR and large ARR is a likely candidate for a contaminated food vehicle, as this implies that the food is strongly associated with illness and that it was consumed by a large proportion of cases.

Adapted with permission from Wahl E, Rømma S, Granum PE. A Clostridium perfringens outbreak traced to temperature-abused beef stew, Norway, 2012. Euro Surveill. 2013;18(9):pii=20408. Available online: Outbreak investigations

Case control studies

Case control studies are more suitable when cases are disseminated in space and/or time and the population at risk is difficult to define. In other situations, the population at risk may be well defined but too large to investigate using a cohort approach, so a case control study may be conducted by recruiting all, or a random subset of, cases and an adequate control group. Risk factors are identified by assessing exposure-specific odds ratios (ORs).

Appropriate choice of controls is important to minimize bias and depends on the specific outbreak. Because outbreaks often affect specific population subgroups, restriction and matching are commonly used for selection of controls to ensure that they reflect the population from which cases arise. Examples include outbreaks of varicella-zoster infection among patients with cancer, TB in schools, and syphilis detected at a clinic for STIs. For community-wide outbreaks, potential sources of controls include individuals registered at the same medical clinics as cases but who do not have the disease or civil registers such as the electoral roll or population registries. Random-digit dialling is a popular method for selecting controls when a sampling frame is unavailable but is increasingly challenging (and potentially biased), given recent increases in the use of mobile phones. For reasons of expediency, some investigators ask cases to nominate peers as controls. Case-nominated controls are easier to identify, and acquaintance with the case can be an incentive to participate. However, acquaintances are likely to be similar to cases in terms of area of residence and certain behaviours that might be related to infection risk, so overmatching is a potential problem and could limit the ability to detect the exposure of interest.

Inference in analytical studies

Statistical evidence of the strength of association between exposures and illness can be assessed using p values and 95% confidence intervals for effect measures (ARRs or ORs), which are available in most statistical software (see Outbreak investigations Chapter 13). Mantel–Haenszel stratified analysis or multivariable regression models can be used to adjust for confounding factors. Matched study designs require specific analysis methods such as McNemar’s test for matched pair designs or conditional logistic regression.

Outbreak size is usually a limiting factor in the analysis. Study power in an outbreak investigation is limited by the number of cases affected and the resources available for recruitment of controls. With modest case numbers, statistical support may be weak, yielding borderline p values and wide confidence intervals. However, analytical studies of outbreaks aim to provide evidence for timely public health action, not estimation of parameters. Statistical evidence and the strength of the association should be interpreted alongside other evidence, including results of microbiological and environmental investigations and the known biology of the suspected pathogen. For outbreaks of food-borne diseases, evidence from an analytical study implicating a specific food vehicle, identifying the same pathogenic strain in cases and implicated foods, and corroborating evidence of contamination from environmental investigations are considered the gold standard for establishing causation. Potential sources of bias, summarized in Table 3.2, should be considered when interpreting evidence from analytical studies.

Table 3.2 Common sources of bias in analyses of outbreaks

Type of bias


Poor recall

Respondents’ recall of exposures may be inaccurate, particularly for organisms with a long incubation period, such as Giardia or hepatitis A virus, or if there is a delay between onset of illness and the interview.

Recall bias

Exposure recall may be more accurate among cases than healthy respondents, because their illness has prompted them to think more carefully about possible exposures.

Selection bias

In case control studies, selection bias can occur when the control group does not adequately reflect the population from which cases arise. An example would be an investigation in which cases are recruited from a genitourinary medicine clinic, but controls are selected from the community, as it is not possible to know whether community controls would have attended the clinic had they developed the disease in question.

Participation bias

The incentive to participate in an outbreak investigation may be less for those who have not been ill, so respondents could be a selected, non-representative sample of healthy individuals.

Social desirability bias

Respondents may not be willing to provide information on risk factors that may be sensitive or stigmatizing, such as sexual activity, drug use, or potentially illicit behaviours, or they may provide responses that they feel are more socially acceptable.

Implementation and evaluation of control strategies

The OCT should decide and coordinate the implementation and evaluation of adequate control strategies. For any given strategy, consideration should be given to existing evidence for its effectiveness, resources and infrastructure available for implementation, and methods used for its evaluation. For example, a community-wide outbreak of TB would require consideration of resources and priorities for contact tracing, region-specific evidence for adherence to and effectiveness of chemotherapy in different risk groups and mechanisms for its delivery, the use of laboratory resources for molecular typing to identify chains of transmission, and mechanisms for prompt case finding and detection.

The OCT is responsible for declaring the end of an outbreak. The team should consider whether control strategies implemented are adequate to prevent new cases or whether there is a continuing threat to the population. The OCT should also discuss implications for:

  • changes to existing control policy

  • changes to guidelines for private enterprises, public bodies, and the general public

  • procedural issues to facilitate future investigations

  • legal issues arising from the outbreak

  • future mitigation of risk

  • further research into specific areas.

The OCT should arrange to complete a written report of the outbreak investigation, including implications and recommendations.


Outbreak investigations require a balance between epidemiological rigour and pragmatism, as circumstances often dictate the quality of available information and the scope and sophistication of the investigation. Geoffrey Rose’s view that epidemiologists work with dirty hands, but a clean mind, and make concessions to rigour, always aware of the implications of each concession,8 is no truer than when applied to outbreak situations. The acute nature of outbreaks, their capacity to cause great social disruption, and the need to identify the causes and implement adequate control measures rapidly means that investigators must often work with, and make decisions based on, descriptive analysis of imperfect data. Development of clear questions and systematic assessment of all the strands of evidence, with its associated limitations, should be guiding principles for making sound inferences. Fortunately, circumstances often work in our favour; for many outbreaks, the exposure effect is strong and can be detected in analytical studies with a modest sample size, and results should be robust against various forms of bias, particularly when considered in the context of other evidence. However, because outbreaks often affect specific population subgroups and result from unusual circumstances, results are not easily generalizable to other settings.


1. World Health Assembly (2005). Revision of the International Health Regulations, WHA58.3. 2005. Available at: Outbreak investigations (accessed 19 January 2015).

2. Balabanova Y, Gilsdorf A, Buda S (2011). Communicable diseases prioritized for surveillance and epidemiological research: results of a standardized prioritization procedure in Germany, 2011. PLoS One, 6, e25691.Find this resource:

3. Reuss A, Litterst A, Drosten C (2014). Contact investigation for imported case of Middle East respiratory syndrome, Germany. Emerg Infect Dis, 20, 620–5.Find this resource:

4. Gregg M (2008). Field epidemiology, 3rd edn. Oxford University Press, New York.Find this resource:

5. European Centre for Disease Prevention and Control (ECDC). Risk assessment guidelines for infectious diseases transmitted on aircraft (RAGIDA)—tuberculosis. ECDC, Stockholm.Find this resource:

6. Nguyen TMN, Ilef D, Jarraud S (2006). A community-wide outbreak of Legionnaires’ disease linked to industrial cooling towers—how far can contaminated aerosols spread? J Infect Dis, 193, 102–11.Find this resource:

7. White PS, Graham FF, Harte DJG, et al. (2013). Epidemiological investigation of a Legionnaires’ disease outbreak in Christchurch, New Zealand: the value of spatial methods for practical public health. Epidemiol Infect, 141, 789–99.Find this resource:

8. Coggon D, Rose G, Barker D (2003). Measurement error and bias. In: Coggon D, Rose G, Barker D (eds). Epidemiology for the uninitiated. BMJ Publishing Group, Chennai, pp. 21–8.Find this resource:

Copyright © 2022. All rights reserved.