Leave a comment

6.4 Principles of outbreak investigation

6.4 Principles of outbreak investigation
Oxford Textbook of Public Health

6.4
Principles of outbreak investigation

Kumnuan Ungchusak

Threat of outbreaks
What is an outbreak?
How can an outbreak be detected?

Health personnel

Laboratory

Official disease notification systems

Newspapers or media

Village health volunteers
Purposes of outbreak investigation

Controlling the current outbreak

Prevention of future outbreaks

Research to provide knowledge of the disease

Evaluation of the effectiveness of prevention programmes

Evaluation of the effectiveness of the existing surveillance system

Training health professionals

Responding to public, political, or legal concern
Components of an investigation team
Prior to the implementation of an investigation

Assessing the existence of the outbreak

Gathering the available basic information

Ensuring that clinical specimens and suspected materials were collected

Obtaining permission and adequate support from the local and national authorities

Field operation plan

Reviewing current knowledge of the outbreak

Steps of outbreak investigation
Co-operation for international outbreak investigation and preparedness
Conclusion
Chapter References

Threat of outbreaks
Knowledge of medicine and diseases has increased enormously during the last few decades. With the advance of knowledge, public health services in many countries can implement effective prevention programmes and are able to protect people from many unnecessary illnesses and death. However, people around the world still suffer and die from various known and unknown outbreaks. Some outbreaks are small and involve only a few persons but some affect more than 10 000 individuals (Fig. 1). During 1997, most countries had at least one infectious disease outbreak and several had as many as eight (WHO 1999). Outbreaks can happen anywhere, from a very remote area where no health facility exists to nosocomial outbreaks in a very sophisticated hospital where hundreds of health personnel are employed. It is a challenge for the government and public health professionals of all countries to detect and control these outbreaks as early as possible. Outbreak investigations also provide the opportunity to discover new aetiological agents, to understand factors that promote the spread of the diseases, and at the same time to identify the weaknesses of existing prevention and health programmes. For these reasons, all public health professionals should have the ability to conduct and support outbreak investigations.

Fig. 1 Large disease outbreaks. (Source: WHO 1999.)

This chapter provides a definition and describes the objectives of outbreak investigation, the methods for planning and conducting an investigation, and what needs to be done after the investigation has been completed. For simplicity, this chapter mainly discusses and presents examples of communicable disease investigations in particular. However, the concepts and principles discussed can be applied to non-communicable diseases as well.
What is an outbreak?
The terms outbreak and epidemic can be used interchangeably. But outbreak is more understandable for most people and conveys a greater sense of urgency. Some epidemiologists prefer to use the term epidemic only in a situation that covers a very wide geographical area and involves large populations. For example, it is possible to use the term ‘outbreak of HIV’ to describe a sharply increasing HIV prevalence rate among commercial sex workers in a city where the normal rate was low in the previous year. But the term ‘HIV epidemic’ can be used when an abnormally high HIV prevalence is found among sex workers in many cities of the country.
In general, the term outbreak is used for a situation when diseases or health events occur at a greater frequency than normally expected in a specified period and place (Fig. 1).
There is often a misunderstanding that only communicable diseases can cause outbreaks. Non-communicable outbreaks such as mass sociogenic illness are sometimes reported as acute outbreaks of unexplained illness, especially in school settings (CDC 1990, 1996).
Because the criteria for judging an outbreak can be very subjective, it is useful to define the term in a more measurable fashion. The criteria for judging that an outbreak has happened can be one of the following.

1.
The occurrence of a greater number of cases or events than normally occur in the same place when compared to the same duration in past years. For example, the epidemic of Kaposi’s sarcoma, a manifestation of AIDS, was confirmed in New York when almost 30 cases were reported in 1981, whereas only two or three cases had been reported in previous years (Biggar et al. 1988).

2.
A cluster of cases of the same disease occurs which can be linked to the same exposure. The term ‘cluster’ is an aggregation of two or more cases which is not necessarily more than expected. For example, three athletes were admitted to hospital with an acute febrile illness and all of them had participated in a triathlon in Springfield, Illinois (CDC 1998a). After receiving this report, the responsible unit started to suspect that an outbreak of febrile illness might be occurring among athletes who participated in the triathlon. The investigation revealed that Leptospira was the cause of the illness.

3.
A single case of disease that has never occurred before or might have a significant implication for public health policy and practice can be judged an outbreak which deserves to be investigated. The first documented case of avian flu (H5N1) in the Hong Kong Special Administration Region in a 3-year-old boy in May 1997 alerted the local authorities and scientists around the world to start a full-scale investigation (Lee et al. 1999).
How can an outbreak be detected?
Public health professionals need to maintain monitoring or surveillance of the disease situation in their local area or country, and also at the international level. It is possible to identify an outbreak by monitoring many sources of information, which will help to detect the abnormal occurrence of disease. Some useful sources are listed below.
Health personnel
Doctors and nurses in a hospital have a good opportunity to observe an abnormal increase in the number of patients with a particular disease or syndrome. An outbreak of suspected mushroom food poisoning in a northern province of Thailand was reported to an epidemiologist during a business telephone conversation with his colleague. The epidemiologist started an investigation and identified the first confirmed outbreak of Clostridium botulism food poisoning associated with home-canned bamboo shoots in the country (CDC 1999a). Without this personal contact, this outbreak would not have been investigated. Thus, public health authorities should maintain a cordial relationship with doctors and hospital staff both in the governmental and private sectors. Conversely, doctors should report all suspected outbreaks to the local public health authorities.
Laboratory
Every laboratory or network can serve as an excellent source of outbreak notification. The avian flu outbreak in the Hong Kong Special Administrative Region was first discovered by the Influenza Surveillance Network, which reported an abnormal influenza, type A (H5N1) (Lee et al. 1999). Without the necessary laboratory capacity, the avian flu might have been overlooked and not triggered a field investigation. A public health professional should communicate regularly with laboratory technicians and vice versa. The laboratory scientists can prevent further spread if they report abnormal findings to the public health authorities regularly and without delay.
Official disease notification systems
Most countries have official systems for notification of cases and deaths from specific diseases from hospitals. The system was designed to detect an outbreak by comparing cases occurring in the current week or month with the average number of cases in the same area during the same period in past years.
For some diseases, like HIV/AIDS, a sentinel surveillance system was designed to monitor and detect abnormal trends in particular sentinel populations and sentinel sites. The first HIV sentinel serosurveillance in Thailand, which started in June 1989, detected that the HIV prevalence among commercial sex workers in a popular northern tourist province was 44 per cent. The finding was very alarming and prompted a field investigation to confirm the high prevalence and to look for risk factors of HIV infection among sex workers (Siraprapasiri et al. 1991). The investigation confirmed the outbreak and revealed the low level of condom use, which led to a recommendation for condom promotion in this high-risk population.
One of the most important functions of epidemiologists and public health professionals is to perform regular analyses of reported disease data. Unfortunately, this task has been neglected and the usefulness of disease reporting systems has been downgraded and often serves only as a vital statistics report. If this neglect of the reporting system can be overcome, the public health system will regain this powerful tool to detect and control outbreaks.
Newspapers or media
In fact, public health professionals learn of outbreaks from the media more often than from the official surveillance system. Newspapers receive outbreak news directly from their journalists or people in the community and are able to report them immediately. The Program for Monitoring Emerging Diseases (ProMED), the prototype for a communications system that monitors emerging infectious diseases globally and an initiative of the Federation of American Scientists and co-sponsored by the World Health Organization (WHO), obtains much of its outbreak news from local or international media. While timeliness is the strength of the media, the validity of the information is often poor and therefore it requires verification.
Village health volunteers
In rural areas where there are no health facilities and communication is limited, village leaders or village health volunteers can often help to recognize an abnormal increase in the numbers of some clinical diseases such as diarrhoea, dysentery, measles, fever, death of unknown aetiology, and so on. For example, the head man in a village of Kachin State in the Union of Myanmar informed the health authorities that seven villagers had died from febrile illness. This information triggered a field investigation, which revealed that malaria was the cause of the outbreak (Dr Myint Win, personal communication, 1999).
Purposes of outbreak investigation
An outbreak investigation can have many purposes as follows.
Controlling the current outbreak
This should be the primary or ultimate goal. If the investigation can start early the findings can guide implementation of appropriate control measures to stop further spread. The avian flu (H5N1) outbreak investigation found a link between infection and illness in chickens and in humans. The same virus was found in both chickens and patients. There were a total of 18 cases and six deaths before the Hong Kong Special Administrative Region decided to kill all 1.5 million chickens in the islands within 3 days to end the outbreak. There have been no cases since (Lee et al. 1999). The key to achieving this goal is to eliminate the delay in detecting the outbreak, to start the investigation as soon as possible, and to implement the appropriate preventive steps indicated by the investigation immediately.
Prevention of future outbreaks
Not all investigations start at the beginning or before the peak of the outbreak. The findings or lessons learned from the investigation may be too late to help fully control the current outbreak but they can still contribute to the prevention of future outbreaks. With good investigation, the weaknesses of the prevention programme can be identified. If recommendations are taken seriously, the chance of recurrence of the same outbreak or other diseases that share common risk factors can be reduced.
Research to provide knowledge of the disease
Information about new diseases and their natural history, clinical spectrum, incubation period, and so on, can often be best learned during an outbreak investigation. The most recent outbreak of encephalitis in Malaysia, which continued until the end of April 1999, and resulted in 257 cases and 100 deaths, prompted an international outbreak investigation which resulted in the discovery of a new nipah virus (CDC 1999b, c). The mode of spread from infected pigs to humans was revealed but there is still much more to be learned.
Evaluation of the effectiveness of prevention programmes
Investigation of an outbreak of disease, which is the target of a public health programme, may reveal weaknesses in that programme. Investigation of an outbreak of vaccine-preventable diseases often identifies populations that have not received the vaccine. For example, the investigation of a measles outbreak that occurred in 1993 in Espindola, a rural community in the Peruvian Andes, revealed that more than a quarter of the 553 residents were affected and that more than 3 per cent of those with measles had died. One year before the outbreak, a national measles campaign targeting children under 15 years of age had been conducted. Although national coverage reported the coverage to be 78 per cent, the investigation found that only 4 per cent of the children in Espindola had actually been vaccinated (Sniadack et al. 1999).
Evaluation of the effectiveness of the existing surveillance system
Some aspects of the surveillance system can be evaluated during an outbreak, such as the timeliness, validity, sensitivity, appropriateness of case definitions, and utilization of the surveillance information.
Training health professionals
The Epidemic Intelligence Service of the United States Centers for Diseases Control and Prevention (CDC) and 20 Field Epidemiology Training Programs around the world use real outbreaks as an opportunity for training as well as to provide service to investigate the causes and determinants of outbreaks.
Responding to public, political, or legal concern
In many situations, the investigation has to be conducted because the media has publicized the complaints of people to politicians, or even rumours. The main objective for this kind of investigation is to verify the outbreak and diagnosis. If it is groundless, the investigator can supply the media with new information that can end the rumours. Conversely, if it is a true outbreak then the investigator needs to decide what steps need to be taken.
In general, for a real outbreak, many objectives can be achieved fully or at least partially. However, the ultimate goal is to control the current outbreak and to prevent future ones. It is unethical for investigators to compromise this ultimate goal with other goals such as training or non-essential research that does not directly contribute to the control activities.
Components of an investigation team
In this chapter, the term ‘investigators’ will represent the people who are directly involved in planning and conducting the outbreak investigation from start to finish. In principle, local health professionals at the district or provincial level should take the role of investigators and start the work as soon as possible. For complicated or difficult field investigations, investigators from additional disciplines or even international experts can provide assistance. It is best to form an investigator team with a single principal investigator in charge of the operation. A good investigative team should include the following people.

1.
A field epidemiologist who is technically competent to conduct field investigations systematically. The field epidemiologist usually serves as one of the primary investigators and should be involved in all the investigative steps.

2.
Disease control people who are experienced in implementing basic disease control measures such as food and environmental sanitation, vector control, vaccination, and so on. If available, an educator who can provide essential knowledge to villagers in clear terms will also be very useful for disease control implementation.

3.
Laboratory technicians who are able to provide basic and advanced laboratory support to the investigative team. Laboratory technicians might not need to travel to the field and collect the specimens by themselves except when a special collection procedure is required.

4.
Specialists in particular areas; for example, a veterinarian would be very helpful for an outbreak investigation of zoonotic diseases. An entomologist is a key member for an outbreak investigation of vector-borne diseases. A social scientist with expertise in qualitative methods will help identify risk behaviours among affected populations and assess the acceptability of the recommended interventions.

5.
Public health administrators, who are good at providing logistic support, mobilizing resources, and providing administrative expertise for the team.

6.
Public relations person. In certain conditions when the outbreak has caused panic or has gained the intense attention of the public, the investigative team should recruit or appoint a person to be in charge of public relations and press releases. This person should appropriately reassure and not unduly alarm the public.
In practice, all of these team members are not always available at the subdistrict or district level due to limited human resources. Public health professionals and field epidemiologists need to have basic knowledge of all these relevant disciplines and be able to assume tasks if needed.
Owing to the sudden nature of the field investigation, it is better to establish in advance a list of people who will be on call and ready to join the investigative team once an outbreak has occurred.
Prior to the implementation of an investigation
The principal investigator should consider all of the following issues before initiating a field investigation.
Assessing the existence of the outbreak
No matter how the outbreak news was obtained, the investigator should confirm the validity of the information. The best way is to have direct communication with the responsible local health authority or field staff. It is not unusual for the information to be groundless. Sometimes the outbreak did happen but the media incorrectly quoted the name of the place. The investigator should carefully check with all other possible local health authorities in order not to miss the outbreak.
Gathering the available basic information
If the local health authority or field staff confirms the existence of the outbreak, the investigator should ask for additional information related to the situation and control measures being implemented.
Information related to the disease situation

1.
What are the main symptoms and signs of the patients?

2.
By whom and how was the diagnosis made, for example using only clinical or also laboratory evidence?

3.
How many patients were seen and how many have died?

4.
What was the average age of the patients and were there any differences in sex distribution?

5.
Where did the patients come from? Are they clustered or scattered?

6.
When was the increased number first observed and what is the trend at the moment?
Information related to control and response activities

1.
What has already been done in terms of the field investigation and implementation of control activities?

2.
Are there any serious constraints to compromise the field investigation and in implementing control measures?

3.
Who are the key people responsible for the investigation and control activities?
It is not necessary to gather all of this information before leaving for the field but having it will help the investigator to plan an effective investigation.
Ensuring that clinical specimens and suspected materials were collected
It is absolutely vital to contact the doctors who saw the cases and made the diagnosis to obtain relevant clinical specimens such as serum, blood, and so on, for future laboratory tests. The items to which the cases were exposed, such as food and water, should be collected immediately before anything is destroyed unintentionally. The investigator should contact the local and reference laboratories so that the necessary supplies and equipment can be obtained immediately.
Obtaining permission and adequate support from the local and national authorities
The investigative team should ask the permission of the local health authority, and in some situations a national authority. This will create a sense of shared responsibility and partnership. Most of the time, local authorities are pleased to receive assistance. In a few situations, the local authority might be unhappy having outside people for the field investigation because of the sensitive nature of the problem. The investigator needs to convince local authorities that a thorough and good investigation will benefit their organization more than harm it. The investigator should also request field support from the key authorities such as field staff who will facilitate the fieldwork, providing transportation, medication, and so on.
Field operation plan
The investigator needs to have a short meeting among team members to summarize the situation, set up the objectives of the investigation, divide responsibility among team members, and check the readiness of laboratory and logistical support.
It is also important to plan the duration of the field investigation. The investigative team should stay in the field until all investigation processes such as data collection, analysis, interpretation, and the executive summary have been completed. Leaving the field without accomplishing all these objectives will cause delay in the implementation of control measures. Most outbreak investigations should plan to obtain preliminary results and recommendations within 1 or 2 weeks and no later than a month after the investigation begins. This is to make sure that the findings will be in time to assure control of the current outbreak. Additional studies and subsequent investigations can be done later.
This initial plan usually needs to be revised once the team arrives in the field. A normal practice in the field is that the team members should have a meeting at the end of each day to review and plan specific activities for the next day.
Reviewing current knowledge of the outbreak
The investigator or one of the team members should be assigned to review current knowledge of the outbreak. The Control of communicable diseases manual (Chin 2000) is very useful for a quick review of most infectious diseases. Searching the literature from the Internet, using as key words the ‘outbreak and name of the disease’, is useful to learn about previous studies done in different countries and settings.
The investigation team should not spend too much time preparing a perfect plan because of the urgency of the outbreak, but rather should obtain what is most necessary and start the investigation as soon as possible.
Steps of outbreak investigation
An outbreak investigation is an observational study in nature because the events have already happened. Every outbreak investigation needs to start with a good descriptive study followed by analytical studies whenever possible and necessary. Conclusions about the causes, mechanisms, and determinants of the outbreak need to be based on sound epidemiological, clinical, laboratory, and environmental evidence.
A descriptive study can help to identify the risk population and risk area so that immediate interventions can be directed to the most needy people and area. A good descriptive study can also generate hypotheses about how the outbreak has spread and what factors contributed to the abnormal occurrence of the disease. In theory, hypotheses derived from a descriptive study should be confirmed by further analytical study. In reality, this is not always possible because of many constraints.
It is preferable to translate the methodology for outbreak investigation into steps of action. Gregg (1996) has divided the outbreak investigation process into 10 steps. With a slight modification, this chapter will also divide the investigation process into 10 steps (Box 1). Steps 1 to 4 use descriptive epidemiology to generate basic facts and hypotheses, steps 5 to 7 are processes to test hypotheses and make conclusions, and steps 8 to 10 emphasize the important of communication of the results and follow-up of the recommendations.
Box 1 Ten steps to take in an outbreak investigation

1.
Confirm the existence of the outbreak

2.
Verify the diagnosis and determine the aetiology of the disease

3.
Develop a case definition, start case-finding, and collect information on cases

4.
Describe person, place, and time and generate hypotheses

5.
Test hypotheses using an analytic study

6.
Do necessary environmental or other studies to supplement the epidemiological study

7.
Draw conclusions to explain the causes or the determinants of the outbreak based on clinical, laboratory, epidemiological, and environmental evidence

8.
Report and recommend appropriate control measures to concerned authorities at the local, national, and, if appropriate, international levels

9.
Communicate the findings to educate other public health professionals and the general public

10.
Follow-up of the recommendations to assure implementation of control measures

This outline of steps for outbreak investigation does not imply a strict sequence of action. In real outbreak investigation, many steps can happen at the same time depending on the situation.
Step 1: Confirm the existence of an outbreak
The main question is: Is this a true outbreak? Applying the definition of an outbreak outlined above, the investigator should be able to establish or refute the existence of the outbreak. Investigators should review the number of cases with the local health officers or hospital staff and compare it with the number found at the same period recorded in past years.
For example, the outbreak of trichinosis in North Rhine-Westphalia, Germany, was confirmed because there were 52 cases in a 3-month period between November 1998 and January 1999 compared with no more than 10 cases annually during the same time period in the past 10 years (CDC 1999d).
Step 2: Verify the diagnosis and aetiology of the disease
If the number of cases fit the case criteria for the outbreak, the next related questions are:

What is the correct diagnosis and aetiology of the disease?

What can be done immediately to prevent new cases from occurring?
Knowing the exact diagnosis and aetiology of the disease will help to establish appropriate preventive measures immediately. This will protect susceptible people and allow the team to start education of villagers to avoid the risk factors. For example, many adults in a remote village were sick with fever, muscle and joint pain, rash over the body, and so on. If the diagnosis and aetiology of the disease is unknown, the local public health officials will find it very difficult to educate the public or implement effective preventive programmes. Until the serology of some patients showed sharply rising immunoglobulin M antibodies to dengue virus, control measures to destroy the larvae of the Aedes mosquito, the vector of dengue, which breeds in water containers, could not be started.
In many situations, an unknown or unclear diagnosis can cause panic due to rumour. This is demonstrated by an outbreak of pneumonic plague in Surat, India, in 1994 and of encephalitis in Malaysia in 1999, which resulted in a severe panic among local people and foreign tourists. Thus it is very important to obtain the exact diagnosis as rapidly as possible.
Investigators should have basic knowledge of the clinical diagnosis and how to confirm the aetiology of suspected diseases by well-established laboratory techniques. It is recommended that investigators should visit and talk with some patients, review and visualize the signs and symptoms, and hold discussions with the attending doctors. The information from this step will help to develop a case definition to facilitate active case-finding. The information on the aetiology will also help to interpret the findings from the later descriptive study and establish a causal association.
The investigators should also visit the laboratory facilities and ask for either positive or negative results of the testing of specimens. It is not necessary that cases have to be laboratory confirmed, but at least some of the apparent clinical cases or deaths should be confirmed. Once there are some laboratory-confirmed cases, the investigator will find it more reasonable to assume that other cases with the same clinical manifestation in the same period and location are the same disease.
It is unfortunate that specimens from patients are often thrown away when the primary results (either positive or negative) are obtained. The investigators should plan with the doctors and laboratory technicians to do further laboratory investigation on the specific strain, and to establish drug sensitivities, genetic markers, and so on. Many new laboratory technologies, such as serological testing, culture and isolation, and molecular techniques, are very powerful for diagnosis and tracking the connection between cases.
Step 3: Develop an appropriate case definition, start case-finding, and collect information on cases
At this stage, the investigator needs to answer at least three questions.

Who should be counted as cases?

Are there more undetected cases in the hospitals and in the community?

What are the characteristics of cases?
To answer these three questions the investigators must follow these three small steps.
Developing an appropriate case definition
It is important that the investigator should develop a case definition, which will be applied consistently during the investigation. The definition should be sensitive or adequate at the beginning in order to capture all actual cases. A good case definition for investigative purpose should be time and place specific. The case definition should not include any specific suspected exposure that the investigator plans to verify. This would create selection bias when the investigator wants to generate the hypothesis or test the hypothesis in the following steps. Using the information from the previous steps, the investigator can divide the case definition into different levels. For example, in an investigation of leptospirosis among athletes participating in the triathlon in Illinois and Wisconsin in 1998, the following definitions were used (CDC 1998b).

1.
A suspected case is a triathlon participant who manifests at least two of the following symptoms or signs: chills, headache, myalgia, diarrhoea, eye pain, or red eyes during the period of 21 June to13 August 1999. These criteria were based solely on a set of clinical signs and symptoms and were rather loose.

2.
A confirmed case is a suspected case who has laboratory confirmation by serology (an immunoglobulin M ELISA and microagglutination test titre greater than or equal to 400), or a positive tissue immunohistochemical test, or a positive culture of Leptospira.
In some instances, the investigator might need to apply a definition of ‘probable case’. A probable case is not fully confirmed by specific laboratory testing but has unique clinical or preliminary laboratory test results. The definition of the same disease in different investigations can be slightly different based on the availability of the laboratory support.
Active case-finding
In places where there is a good surveillance system, cases from all hospitals will be reported to the epidemiology unit at the district or provincial level. Investigators can apply the case definition and count the number of cases and review data that are collected on the reporting form. If there are enough cases and basic information on cases has already been gathered then investigators can start to do descriptive epidemiology.
Conversely, if only a few cases have been seen at the health facility, the investigators should plan to search for cases in the community. The investigator has to start a process called active case-finding. The objective of active case-finding is to have enough cases to analyse. At the same time, this case-finding will give a better picture of the magnitude of the outbreak. For example, in the outbreak of leptospirosis among triathlon athletes in Wisconsin and Illinois in 1998, only three cases were reported from the hospital at the beginning. The investigators co-operated with many state health departments by making a telephone call survey and in the end they succeeded in interviewing 1194 athletes. Among them, 110 or 9 per cent had an illness that met the suspected case definition of leptospirosis. The investigators also obtained acute serum from 70 cases, 24 of whom were confirmed as having leptospirosis by both immunoglobulin M and microagglutination tests (WHO 1978).
In the above example, it sounds easy to do active case-finding by telephone interview. However, investigators need to try their best to do active case-finding whenever it is necessary no matter how difficult it is. Very often, the investigators have to visit many hospitals in the outbreak area and review medical records by themselves. Sometimes, they have to search the cases in each village by doing interviews from house to house. This is the real nature of outbreak field investigations and the way people learn epidemiology. This active case-finding in the community will also provide another two benefits as follows.

1.
Control measures can be implemented if the aetiology of the disease is known and treatable. During an investigation using active case searching in a village that reported seven deaths from malaria illness in Kachin State in the Union of Myanmar, the investigator found 94 probable cases and 53 microscopically confirmed malaria infections. All of the probable and confirmed cases found by this active process were treated. Without this measures there might be more deaths later on (Dr Myint Win, personal communication, 1999).

2.
Rapid environmental assessment can be started during the visit to the affected families or villages. From this direct interview with the cases, the investigators can develop some hypotheses and implement some necessary interventions immediately such as sanitation improvement in food handlers and treatment of water to prevent water-borne outbreaks.
In situations in which the outbreak is not localized but widespread, the investigator might need to use the media to alert the public about the outbreak. People can then avoid suspected exposures and see a health-care provider if they have developed symptoms compatible with the case definition.
Collecting information on cases
For each case, the investigator should collect at least four types of information.

1.
Identifying information: name, hospital number, contact person, and address of contact. This additional information will help to avoid duplication of enumerated cases. The investigator can also maintain communication with these cases when more information is needed.

2.
Demographic information: age, sex, occupation, religion, ethnicity, area of residence, place of work, and so on. This is important information which will help to determine the characteristics and distribution of cases.

3.
Clinical information: the symptoms and signs, the date of onset, the duration of illness, and results of diagnostic procedures. These data will help to confirm that this is a true case, provide the pattern of clinical manifestations, and also the distribution of cases by time.

4.
Suspected risk factors: the investigators can ask for a history of exposure to some factors before disease developed. The timing of interest usually is one incubation period if the aetiology is known or suspected. Questions about contacts with other patients who have similar clinical symptoms are also very helpful. If the diagnosis is not known, the investigator might collect this group of information in a qualitative manner.
The investigator should develop a questionnaire as a tool to collect the relevant information from the hospital or during active case searching in the community. In practice, some of the information will not be available or not of good quality. The investigators should indicate all data that are in doubt.
Step 4: Describe the outbreak in person, place, and time, and hypotheses formation
In this step the investigators need to answer the following questions.

What are the main clinical features?

Who is the population at risk?

What are the risk factors?

What are the most likely explanations of how the outbreak began?
The simple approach is to analyse clinical information from each case and see the distribution of factors in terms of person, place, and time. The analysis should be done using rates rather than absolute numbers. The investigator needs to obtain the denominators from an available source or to estimate them. Using rates, the investigator can compare and determine the populations and areas of highest risk. With the advent of computers, many software programs are available to analyse this data. A popular one is Epi Info which is a public domain software from the American CDC which was designed specifically for field investigations. In the absence of a computer the investigators can still analyse the data manually. Individual questionnaires can be compiled into a line listing, which includes important variables of all cases. With this line listing simple counts can be made. In this way, the investigator will gain knowledge about populations and areas of risk. Resources and control measures can then be directed to the risk populations and risk areas. The enumeration will also produce information for hypothesis formation to explain how and why the outbreak happened.
Clinical manifestation of cases
Signs and symptoms of cases can be analysed in percentages and shown in a summary table. In an outbreak of unknown aetiology, the clinical information will help to differentiate the diagnosis. For an outbreak in which the aetiology is already known, the investigators still need to compare the clinical information found in the investigation with previous knowledge. Any discrepancy found, such as the attack rate, mortality rate, severity, and so on, should be carefully examined because this might indicate that a new strain or different specific host response is occurring. The high mortality rate of approximately 40 per cent in the nipah encephalitis outbreak in Malaysia 1999 (CDC 1999c) indicated that this outbreak might not be due to the usual endemic encephalitis.
Index case
The investigators should analyse the characteristics of cases by sex, age, occupation, ethnicity, and so on. At first this can be done by examining the proportion of all cases, but the specific attack rate by age and sex will be more useful for comparisons and hypotheses formation. The rates will provide useful indicators of the possible aetiology of the outbreak. In the Ebola haemorrhagic fever outbreak in Zaire in 1976, all ages and both sexes were affected but females slightly predominated. Age- and sex-specific attack rates indicated that adult females had the highest attack rate. This finding suggested that parenteral injection with unsterilized syringes and needles given during antenatal care in the local hospital was the means of transmission (WHO 1978).
The outbreak of nipah encephalitis in Malaysia in early 1999 was initially thought to be due to Japanese encephalitis alone. However, after careful analysis of the descriptive information, it was clear that the cases were mainly male, adult, working in pig farming, and of Chinese ethnicity. This descriptive information did not fit the pattern for Japanese encephalitis, which affects mainly children of both sexes with no preference for a particular ethnicity. The investigator then began to suspect other organisms and to hypothesize that working in pig farming increased the risk of becoming ill.
Index and outlier case In infectious diseases, the first case on the epidemic curve or the index case is important because of the possibility that he or she brought the disease into the community. Cases that appear at the very beginning or at the end of the epidemic should also be given careful attention. These cases are called ‘outlier cases’ and can provide important information about the source and the way in which the disease is spread. Fruit juice from a vendor in a school was suspected as the source of a food poisoning outbreak involving Methomyl, a carbamate insecticide, in a secondary school in Bangkok, Thailand. This hint came from the first case, which was a worker sending ice to the school vendor. He was kindly given a cup of juice by the vendor and became ill without taking any other food items from the school. Without this information, it would have been quite difficult for the investigators to suspect any particular food or drink from the list of eight food items sold that morning. Thus the very first and very last cases in the epidemic curve should be critically appraised. The first case might not be the true index case because of misdiagnosis, an unrelated case to the epidemic, and so on. The late cases may be due to misdiagnosis, being an unrelated case to the epidemic, or a secondary case that had a different exposure than the majority of the cases in the epidemic.
Place
The investigators can calculate the attack rate of cases by different places. These can be place of residence, place of work or place of exposure, and so on. Places with a high attack rate often indicate the source of infection or contamination. A spot map showing the location of cases can give a very good idea of the source as had been demonstrated by Snow in his classical investigation of a cholera outbreak in the Golden Square area of London between August and September 1854 (Snow 1936). In that investigation Snow found most cases clustered around the Broad Street pump. From this information, Snow deduced that the Broad Street pump was probably the primary source of the outbreak.
If cases are scattered in many places, the investigators should explore the secular pattern of the case over time. This will indicate whether the outbreak started in one area and spread to other areas or whether people living in different places had a common exposure.
Time
The aim is to show the occurrence of cases over time and look for a pattern of occurrence. In general, there are two major types of outbreaks, a common source and a propagated source. The way to differentiate these two patterns of outbreak is to draw an epidemic curve which shows the number of cases (on the y axis) over time of onset (on the x axis) (Fig. 2). The epidemic curve of each outbreak will suggest whether the mode of spread is by common source or person to person.

Fig. 2 Epidemic curve of food poisoning in secondary school, Bangkok, Thailand, 1981 (n = 101). (Courtesy of Dr Somsak Wattanasri, Division of Epidemiology, Ministry of Public Health, Thailand.)

A common source outbreak This kind of outbreak happens when people get the infection by exposure to the same source of infection. For example, a group of people contract hepatitis A infection because they eat the same contaminated food served during a wedding party. A common source outbreak can be divided into a point common source and a continuous common source.

1.
A point common source outbreak occurs when there is one single source that exists for a very short time and all cases have common exposure to it in that same particular period.

2.
A continuous common source outbreak occurs when there is only one source, which provides continuous or intermittent exposure over a longer period.
Epidemic curve of a point common source outbreak The epidemic curve shows a sharp increase of many cases suddenly followed by a rapid decline, though not as rapid as at the beginning of the epidemic. Another criteria to judge a point common source is that the first case and the last case usually happen within one incubation period.
If the aetiology and the knowledge of the incubation period of the disease are known, the investigators can then roughly estimate the probable time of the initial exposure. This can be done by identifying the peak of the epidemic from the curve and counting back on the x axis the equivalent of the incubation period. The investigator can also use the first case and count back on the x axis a duration of the minimum incubation period, which will also give a rough estimate of the time of exposure.
If the aetiology of the disease is not known, but the epidemic curve fits well with a point common source outbreak, the investigator can estimate the average incubation period if he or she knows the time of common exposure. In the outbreak of Methomyl food poisoning in a school, 101 students became ill after eating food in the school around 7.30 a.m.; the average incubation period was estimated to be 60 min using the duration between the time of known exposure to the peak of the epidemic curve as shown in Fig. 2.
Epidemic curve for a continuous common source outbreak The epidemic curve shows an abrupt increase in the number of cases but, instead of having a peak and a decline within one incubation period, new cases persist for a longer time with a plateau shape instead of a peak before decreasing. However, if there are many peaks or irregular jagged curves this suggests an intermittent common source.
A propagated outbreak
This kind of outbreak is caused by a transmission from one person to another person which requires direct contact such as touching, biting, kissing, or sexual activities.
Epidemic curve of propagated source outbreak The epidemic curve shows a slow increase in the number of cases with progressive peaks approximately one incubation period apart. The investigators might observe an abrupt decrease in new cases because everyone has already been infected. The span from the first to the last cases will also last longer than several incubation periods.
The outbreak of measles in 1993 in Espindola, a rural area in Peru, clearly showed a propagated pattern (Fig. 3). The outbreak began following the arrival of a family from Ecuador in July. Two children of this family developed measles during the journey. A welcome party for the family was held, after which 10 people who attended the party developed measles 1 to 2 weeks later. Several of these new measles cases also attended either a baptism or funeral at the local church. During the subsequent week another 11 cases who attended the baptism or funeral developed similar symptoms (Sniadack et al. 1999).

Fig. 3 Measles by date of symptom onset, Espindola, Peru, 1993. (Source: Sniadack et al. 1999.)

Step 5: Testing the hypotheses by analyses
In an outbreak of infectious disease, the investigator needs to answer the following questions.

What is the aetiology of the disease?

What is the source of infection?

What is the pattern of spread?

What are the risk factors for an individual to get the disease?

What are the determinants of the outbreak or the factors which when combined together result in the outbreak?
The aetiology of the disease should be derived from the laboratory study. The pattern of spread can be identified by a careful descriptive study as described above. Sometimes, the descriptive study is not as simple as expected and does not give enough clues. The investigator must patiently re-examine the descriptive information, carry out more active case-finding, and reanalyse the data. The investigator also needs to observe directly the place, lifestyle, or behaviour of the case. These additional investigations usually help the investigator to generate some reasonable hypotheses. For example, a cholera outbreak in a home for mentally disabled children persisted over a 1-month period and the epidemic curve showed a pattern of person-to-person spread. After looking into the attack rate in different buildings in the home the investigator found that some buildings had a very high attack rate. When the investigator visited the home he discovered that the building with the high attack rate was very crowded. Two or three children had to share one single bed. The children also bathed in a small room where they were exposed to faeces from other children. With this direct observation, the hypothesis was formulated that crowding promoted the spread of the disease. In this example, the hypothesis was very reasonable and might not need to be tested. Improving the crowded condition was recommended and action was taken immediately.
In another example, the hypothesis needed to be tested by an analytical study design. The most common one is a case–control study. The investigator needed to define the cases and measure the rate of finding the suspected factors found among cases. He or she then compares this with the rate found among appropriately selected controls. The case definition for the analytical study may be more specific than used for descriptive study in order to reduce misclassifying non-cases as cases. Conversely, the controls may also need to be tested to avoid classifying non-apparent cases as controls.
In the outbreak of Methomyl food poisoning among secondary school students described above, the investigator had already identified 101 cases. Because a high percentage of cases had drunk fruit juice, the investigator hypothesized that drinking fruit juice from one of the food vendors in the school was the risk factor. The investigator decided to conduct a case–control study to prove this hypothesis (Table 1). He selected another 107 students without any diarrhoea, nausea, or vomiting from the same class but with the same gender as the controls. The investigator then asked which of the eight food and drink items the cases and controls took that morning. The odds ratio of drinking any of the three fruit juices calculated from this analytical study was three to seven times higher and was statistically significant. This finding established a significant statistical association between drinking fruit juices and becoming ill.

Table 1 Odds ratios and 95% confidence intervals for specified food items found in a case–control study: Methomyl food poisoning investigation in School A, Bangkok, 7 July 1981

In this same investigation, the investigator also conducted an historical cohort study, which is another type of analytical study. Out of the total 1544 students, he first identified 416 who ate the breakfast in the school. These 416 students were defined as the cohort population and were interviewed for their consumed food and the occurrence of the illness after eating. The attack rate, or incidence, of becoming ill among those who ate each food item was compared with that in those who did not eat. The relative risk could then be calculated as shown in Table 2.

Table 2 Relative risk (RR) and 95% confidence intervals for specified food items found in retrospective cohort: Methomyl food poisoning investigation in School A, Bangkok, 7 July 1981

The results of a case–control study and the historical cohort study gave the same conclusion that the source of the outbreak was fruit juice A, B, and C. The historical cohort study took more effort but gave information on the incidence and the relative risk, which better showed the association of exposure and illness.
Step 6: Environmental or other studies to supplement the epidemiological findings
Although an analytical study might be able to confirm the hypothesis, the investigator still needs to find environmental or other evidence to support and explain the epidemiological evidence.
In an outbreak of unknown illness in a rural village of Egypt in which the cases developed severe abdominal pain, persistent vomiting, and generalized weakness, the investigator was able to detect abnormally high blood lead levels among the cases. The analytical study revealed an association between high blood level and eating flour from one mill factory. The mill implicated in the outbreak was visited. Upon arrival at the mill, the investigators noted a lead smelting pot in the corner of the mill. Lead was used by the miller to attach the crosspiece to the grinding stone. Occasionally, the lead would break off and contaminate the flour. The miller reportedly used about 2 kg of lead per year. Analysis of grain from the mill showed no lead; however, lead was found in flour from the surface of the mill stone and in samples of flour after grinding was complete (Abdel-Nasser et al. 1996).
Step 7: Establishing the causes of the outbreak
Once the hypothesis has been tested and other necessary studies have been done, the investigators can make their conclusions about the causes of the outbreak. This conclusion is very important because actions should follow. Many restaurants or factories have had to be closed because they were implicated as the source of the outbreak. Outbreak investigations are naturally faced with constraints of time and other uncontrolled conditions, which do not favour a perfect design and methodology. The findings potentially include both random and systemic errors. Before making any conclusion, it is worthwhile for the investigator to examine carefully the weaknesses or limitations of the investigation. In principle, the investigator must identify the cause of the outbreak based on the agreement of the following four pieces of evidence.
Laboratory evidence
The aetiology of the disease has been identified in the patients and in the suspected source of infection. If the investigator cannot identify the aetiology from the suspected source, it is still possible to use some marker to support their conclusions. For example, although a Vibrio cholerae type 01 could not be cultured from the drinking water supply of the affected community, the observation of human coliform bacteria in the water helped to indicate that human faeces had really contaminated the water.
Clinical evidence
It is necessary to verify that the clinical manifestations and the incubation period are compatible with the aetiology reported from the laboratory.
Environmental evidence
The examination of the environment should reveal the possibility that the causative factor can pass from the source to the cases. A cooking environment which is dirty or located close to the toilet is convincing evidence for food being contaminated by faeces. Crowding in the bedroom is very convincing evidence for a respiratory tract disease outbreak.
Epidemiological evidence
This epidemiological evidence found in the descriptive and analytical studies should clearly explain the following aspects:

pattern of spread as described by epidemic curve

statistical strength of association between exposure and developing the disease

dose–response relationship, which demonstrates a higher strength of association when the exposure is increasing

exposure should precede illness.
In the example of the Methomyl food poisoning discussed above, the investigator identified the three fruit juices as the implicated food. In the case–control study (Table 1) cases were many times more likely to have drunk the fruit juice than healthy students. In the retrospective cohort study (Table 2), students who drank either of the three fruit juices were almost twice as likely to become sick as those who did not drink them. Methomyl was identified as the aetiology of this outbreak. It was found in sugar that was used to prepare the implicated fruit juices. Methomyl could produce the clinical signs and symptoms found in the cases and also matched the incubation time of approximately 60 min found in the investigation. Because the food vendor also used Methomyl in her household, it was possible that she accidentally contaminated the sugar with Methomyl since the chemical is colourless.
It is not uncommon to see some disagreement in the evidence from different sources. In this event, the investigator needs to verify the validity of each piece of evidence and discuss the data with people who have expertise in that particular area in order to obtain more information.
Step 8: On-site reporting and recommendations for concerned authorities
The most important step which should lead to a timely response is to report the findings to the responsible individuals both at the local and national levels so that they can take the appropriate action. Keeping this in mind, the investigators need to complete two tasks before leaving the outbreak area.

1.
Complete the analysis and data interpretation. Leaving the field without completing these tasks will reduce the sense of urgency necessary to finish the work. Data are easier to retrieve when in the field rather than instructing the field people to send them to the investigative team later.

2.
Present the main findings with recommendations. Some findings may be very sensitive because they reflect the weaknesses or mistakes of the health or other authorities. The investigator needs to select the appropriate approach, either formal or informal, with the responsible people. Leaving the field without giving the information that the team obtains from the investigation will do harm to some other people who could be prevented from becoming ill if the information were known. In many urgent situations, the findings and recommendations might be given to the responsible people repeatedly during the outbreak investigation process instead of waiting until the end. The investigators also need to present or communicate the findings and recommendations to the national authorities as soon as possible. The recommendations for action should be based on the findings from the investigation. These might include the following aspects.
What can be done to control this outbreak?
With timely investigation, some interventions can be implemented to stop further spread. In the outbreak of measles in a rural community of Peru, measles vaccine was given to children who were not measles cases aged between 6 months and 15 years regardless of their previous immunization status. Using knowledge about the complication of measles in previous studies, the investigators estimated that the action prevented 87 cases of diarrhoea and 46 cases of pneumonia, and averted five deaths (Sniadeck et al. 1999).
How to improve the interventions
Commonly, interventions have to be implemented while the investigation is on going. The investigator should review the measures and give unbiased opinions regarding the appropriateness, the target group, the timing, and the area of implementation. In a cholera outbreak without adequate investigation, some health authorities tried to use mass treatment with tetracycline or doxycycline for all members in the affected community. A careful investigation can help indicate better, more targeted interventions and discourage this kind of mass treatment.
How to prevent future outbreaks
In many instances, the interventions cannot be implemented for the current outbreak but the findings can be used to set up new practices or policies. These recommendations can help prevent future outbreaks. In Thailand the recommendation was made to change Methomyl from a colourless powder to a blue powder following the investigation of food poisoning described above. The Egyptian government, after reviewing the results of the investigation of lead contamination in flour mills, agreed to ban the use of lead in privately owned flour mills.
How to improve the investigation
The investigator should review the performance of the investigation and identify the weaknesses of the methodology or the field operation so that improvements can be made. These might include a more appropriate case definition, a better design, improved laboratory support, a different team composition, less cost, a shorter time course, and so on.
How to improve surveillance
The best time to evaluate the surveillance system is during an outbreak investigation. In most outbreaks, the investigator has to review data from existing surveillance systems. With this direct involvement, the investigator will be able to evaluate the timeliness, completeness, validity of diagnosis, sensitivity of the system, and utilization of the surveillance information.
Step 9: Dissemination of the information
In addition to on-site reporting, the investigator should disseminate the information to educate the public health community and the general public. There may be many other communities that are also prone to a similar outbreak. The information will raise the awareness of health and government authorities to assess their own situation and implement some measures to prevent possible outbreaks. The dissemination of information should be done in a timely manner through weekly or monthly reports. Release of important findings through the mass media is also very useful to educate the public. Before releasing the investigation results through the media, the investigator must make sure that all the facts will be delivered in a constructive manner and will not result in blame to any organization. The investigator should also report the investigation results in an international journal or bulletin such as the Weekly Epidemiological Record of the WHO or the Morbidity and Mortality Weekly Report of the CDC. This kind of practice is necessary to alert health professionals in other countries and keep them informed of the problem.
Step 10: Follow-up to ensure implementation of control measures
Finally, the investigator should follow-up the investigation by maintaining close communication with the local health authorities. An absence of new cases for at least two incubation periods of the infectious disease under investigation could suggest that the outbreak is subsiding. A good investigator should follow up on the recommendations. An outbreak investigation is a waste of time if good recommendations have not been implemented. The investigator should learn the reasons why the recommendations were or were not implemented. If the recommendations were implemented, the investigator can also learn the impact by observing the trend of the diseases.
Co-operation for international outbreak investigation and preparedness
The world today is especially prone to outbreaks because of frequent cross-border movement, civil war and migration, rapid transportation, international trading, tourism, and so on. An outbreak in one country, therefore, can spread to other countries very easily. An outbreak anywhere in the world must now be treated as a threat to all countries. It is important for each country to build up its capacity for surveillance, outbreak preparedness, and investigation. Besides the CDC Epidemic Intelligence Services, many countries have started training programmes and established a medical detective unit which will be ready to investigate all kind of outbreaks. These training programmes have different names, such as the Field Epidemiology Training Program in about 20 countries, and the Epidemiology European Training Program in the European Community. The WHO has set up an outbreak verification network, a worldwide network of laboratories and reporting sites which collect information on reported and rumoured outbreaks nationally and worldwide. It is a wise investment, and commitment is beginning to be made by international communities to collaborate in information exchange and to help investigate outbreaks together.
Conclusion
Outbreak investigation is an essential function of public health professionals who care for the well being of the community. It is an opportunity to gain new knowledge of diseases and to discover the weaknesses of current public health practices and systems. A good public health professional must always be alert to the possibility of outbreaks. Normal surveillance systems or unofficial sources such as the mass media can be a good source for detection of an outbreak. Before starting the field investigation, the investigator should organize the team, review previous knowledge, prepare the technical and management aspects, and start the investigation as soon as possible. The investigation can be conducted by following the 10 steps outlined above. The investigation usually starts by confirming the existence of the outbreak, verifying the diagnosis, gathering case information, doing descriptive epidemiology, formulating and testing the hypothesis when necessary, doing environmental surveys to supplement epidemiological evidence, and providing timely on-site reporting of the findings with practical recommendations to local and national responsible authorities. Competent outbreak investigation combines sound scientific knowledge and good management skills to resolve the crisis or unexpected situations. It is not enough to understand or memorize these steps. Direct participation in conducting the investigation is needed to gain the necessary skill. A good investigator should be a field-oriented person with good levels of perseverance, scepticism, and common sense. The investigator should not end his or her work with the report but should follow up on the recommendations and continue surveillance of the problem vigorously. In the future more and more joint international investigations will be needed. The rich countries must help the poor build up their epidemiology capacities in order to detect and stop outbreaks before they get out of control. With this co-operation the world will be a safer place amidst the threat of disease outbreaks.
Chapter References
Abdel-Nasser, M.A. et al. (1996). Outbreak investigation of an unknown illness in a rural village, Egypt, 1996. Field Epidemiology Training Programmes, Cairo.
Biggar, R.J. et al. (1988). AIDS-related Kaposi’s sarcoma in New York City in 1977.New England Journal of Medicine, 318, 252.
CDC (Centers for Disease Control and Prevention) (1990). Mass sociogenic illness in a day-care center, Florida.Morbidity and Mortality Weekly Report, 39, 301–4.
CDC (Centers for Disease Control and Prevention) (1996). Outbreak of unexplained illness in a middle school—Washington, April 1994. Morbidity and Mortality Weekly Report, 45, 6–9.
CDC (Centers for Disease Control and Prevention) (1998a). Outbreak of acute febrile illness among athletes participating in triathlons—Wisconsin and Illinois 1998. Morbidity and Mortality Weekly Report, 47, 585–8.
CDC (Centers for Disease Control and Prevention) (1998b). Update: leptospirosis and unexplained acute febrile illness among athletes participating in triathlons—Illinois and Wisconsin, 1998. Morbidity and Mortality Weekly Report, 47, 673–6.
CDC (Centers for Disease Control and Prevention) (1999a). Foodborne botulism associated with home-canned bamboo shoots—Thailand, 1998. Morbidity and Mortality Weekly Report, 48, 437–9.
CDC (Centers for Disease Control and Prevention) (1999b). Outbreak of Hendra-like virus—Malaysia and Singapore, 1998–1999. Morbidity and Mortality Weekly Report, 48, 265–9.
CDC (Centers for Disease Control and Prevention) (1999c). Outbreak of nipah Virus—Malaysia and Singapore, 1999. Morbidity and Mortality Weekly Report, 48, 335–7.
CDC (Centers for Disease Control and Prevention) (1999d). Trichinellosis outbreaks—North Rhine-Westfalia, Germany, 1998–1999. Morbidity and Mortality Weekly Report, 48, 488–92.
Chin, J. (ed.) (2000). Control of communicable diseases (17th edn). American Public Health Association, Washington, DC.
Gregg, M.B. (1996). Conducting a field investigation . In Field epidemiology (ed. M.B. Gregg), pp. 44–59. Oxford University Press.
Lee, S.Y., Mak, K.H., and Saw, T.A. (1999). The avian flu (H5N1): one year on. Public Health and Epidemiology Bulletin, 8, 1–8.
Siraprapasiri, T., Thanprasertsuk, S., Rodklay, A., et al. (1991). Risk factor for HIV among prostitutes in Chiangmai, Thailand. AIDS, 5, 579–82.
Sniadack, D.H., Moscoso, B., Aguilar, R., et al. (1999). Measles epidemiology and outbreak response immunization in a rural community in Peru. Bulletin of the World Health Organization, 77, 545–52.
Snow, J. (1936). Snow on cholera. Oxford University Press, London.
WHO (World Health Organization) (1999). Report on infectious diseases. WHO/CDS/99.1, p. 39. WHO, Geneva.
WHO (World Health Organization) (1978). Report of an international commission. Ebola haemorrhagic fever in Zaire, 1976. Bulletin of the World Health Organization, 56, 271–90.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: