GuiaVETA - Chapter III - Invetigation of outbreaks

Investigation should start immediately after the report since delays could lead to the loss of important data for analysis.

Based on information on the existence of an outbreak and awareness of its spread, initial planning of cooperation among the services involved and the exchange of up-to-date information should take place.

This initial planning should be carried out within a very short space of time (one hour approximately). The following procedure is recommended:

There are ten (10) basic steps that are recommended for investigating a disease outbreak. Although they are listed here in a logical order, you may perform several of these functions simultaneously or in a different order. For example, it is not uncommon to find control and preventive measures instituted soon after the start of an investigation, based only on intuitive reasoning or common sense. What is important is the fact that, in general, the data collected, the analyses done on those data and the control and preventive measures are very similar. Whichever approach is taken, there should be regular meetings and debriefings, and clear assignments of responsibilities.

Table 2 below is provided to assist in giving a clearer picture on disease outbreak investigation along the 10 steps listed, with one or more topics being related to each other.

Once tasks have been assigned, personnel should move as quickly as possible to the places where people have been exposed (whether they fell sick or not) and to the location where the suspected food was prepared and/or consumed. This time factor is crucial to ensuring rapid collection of food, environmental samples and/or clinical specimens from the affected individuals before antibiotics are administered or the food discarded.

An outbreak is verified by comparing the experience with that of previous occurrences of similar cases. Consistently reported data from the routine surveillance system are invaluable in determining that an outbreak is occurring. In the absence of surveillance data, investigators may have to rely on the knowledge of local public health staff.

- As a result of the review of information on cases of FBD amongst patients attending the health services, it reveals an apparent similarity in terms of starting date of the symptoms, number of patients affected, predominant symptoms, suspected food, places where the suspected food was eaten up to 72 hours before the beginning of the symptoms, and any other information of epidemiological interest.

The second step in the investigation is confirmation that there really is an outbreak. This may require a brief review of the clinical findings, or may necessitate laboratory confirmation. Since, however, some laboratory tests may be complex and lengthy, outbreak investigation and some control measures may be undertaken prior to receiving the laboratory results.

An outbreak of FBD may occasionally be mistakenly diagnosed in closed centers, for example in situations such as those caused in Day Care Centers for children and in Nursing Homes by cross-contamination, particularly by highly infectious agents such as Shigella, and the Hepatitis A virus, among others. The opposite may also occur. That is, a possible relationship to water or contaminated food may be concealed. In such cases, only epidemiological research, and the use of the epidemic curve may determine if there is actually an outbreak of FBD.

On the other hand, there may be accusations or reports for which the investigating team may find no corroborating evidence.

When it is clear that there is an outbreak, first find out the number of people affected and then carry out the investigation.

Investigation of cases may also be done if appropriate tabulation methods are available. Where numbers of cases are very high, a sample should be taken.

When an outbreak report has been confirmed, a group of elements such as the following should first of all be communicated to higher levels of the system:

In arriving at a common cause, cases are studied in relation to the time of onset (inclusive of trends and periodic changes), the place of exposure (including travel history), and the characteristics of the individual (age, sex, occupation).

The interview should begin with an explanation to the respondent of the importance of his contribution to the outbreak investigation and the benefits to public health and to the society. After asking about exposure and obtaining the history of the disease, the interviewer should ask more specific questions in order to get particular information to corroborate the responses.

With specific cases of E. coli O157H7, salmonellosis, shigellosis, hepatitis A, ciguatera, fasciolasis or others, a specific file should be prepared, taking only elements of the FBD Surveillance Form 1, to be transferred to a database, for example Epi Info.

Before starting an epidemiological survey, formulate a case definition and state the criteria for inclusion in the list as a suspect or confirmed case. It is essential to decide on the "case" definition, looking at those who would be included in the survey, who would be excluded, and thus who are the people that meet the requirement for investigation. Accordingly, the case definition is often made based on the presenting symptoms. This emphasizes the importance of having a good clinical description in order to develop a proper case definition.

In order to draw up the case definition it is important to take into account Annexes E and I of the present Guide. Examples of proper case definitions are given in Annex K for colitis caused by E. coli O157 and Hemolytic Uremic Syndrome.

An experience of illness amongst two or more persons after ingestion of a common food or drink, and in which epidemiological evidence implicates the food or drink as the source of the illness.

A probable case with laboratory confirmation, with the criteria dependent on the etiologic agent.

In as precise a manner as possible, formulate a hypothesis to act as a guide to the outbreak investigation, and which is to be proven or disproved. The hypothesis should include all clinical laboratory and epidemiological information from the investigation to date, and any known facts about the disease process.

The initial information obtained from case histories and from preliminary inspection of the place where the outbreak occurred may often allow possible description of the event in simple epidemiological terms, and form the basis for establishment of a preliminary hypothesis on the cause of the outbreak, and the degree of risk to the population.

During this stage, it is possible to implement control measures such as withholding of the foods that are involved or suspected, isolation of the food handlers, closure of the establishment, and dissemination of information to the public and to higher levels within the organization. In any case, base conclusions on the pertinent evidence, without relying upon any single distribution or circumstance on its own.

Classical outbreaks occurring in closed communities may be easier to investigate, since the food consumed and the various risk factors tend to be the same and the people affected may share similar experiences. When however, there are isolated cases or the disease has an extended incubation period, the epidemiological study is more difficult.

During the interview process, responses should not be suggested to the respondent. Instead, questions should be clearly asked so that the respondents can describe the disease and symptoms in their own words. The interviewer must always be aware that errors in response, whether voluntary or involuntary, due to misunderstanding, misinformation, misrepresentation, and faulty memory, can and will occur.]

Large numbers of both cases and healthy respondents, used as controls, could make it impossible to work on letter-sized sheets. By using standard investigation forms from the initial collection of information, separate out the variables that are important possible etiological factors. Use may be made of existing guidelines and of special forms that may be devised. It is therefore recommended to prepare a single registration sheet on larger sized paper, following the guidelines of the FBD Surveillance Form 2.

Some responses will have to be obtained through deduction since some respondents may be sensitive to certain questions. It is therefore also recommended that the survey be conducted in private. Where doubts arise, indirect questions can be asked that will corroborate the responses that are needed, such as visits to a specific place, recent meetings, same type of food ingested, etc.

During the description of diseases by respondents, the interviewer should take into account Annex H, dealing with signs and symptoms. This should never be used to enquire about the full range of symptoms, but to verify those symptoms marked by an asterisk in relation to the specific disease being investigated.

The signs and symptoms that appear in the first two columns refer to chemical substances and to poisonings. Those in the third, fourth and fifth columns are related to generalized and localized enteric diseases respectively. Those in the sixth column refer to infections of the central nervous system.

Sick people will report only a limited number of signs and symptoms, but if a disease seems to fall within one of these categories, the other symptoms should be mentioned and the patient’s responses noted.

To further facilitate the survey, if nausea and vomiting are the predominant symptoms observed, food consumed within the six hours prior to the appearance of the first symptoms should be investigated and agents such as Staphylococcus, Bacillus cereus, agents associated with vomiting and/or poisoning by chemical substances could be considered. With chemical substances, be aware of acidic foods packaged in metallic containers that, through a process of `leaching`, may yield ions to the food, or the accidental or incidental addition of chemical substances such as nitrites, pesticides, etc.

When diarrheal diseases and abdominal pains predominate in the absence of fever, foods consumed between 6 and 20 hours before the disease should be investigated. The agents could be Clostridium perfringens or Bacillus cereus (diarrheal type).

When symptoms such as diarrhea, chills and fever are present, then foods consumed between 12 and 72 hours previously should be investigated and the agents could be E. Coli, Salmonella, or Norwalk Virus type.

With incubation periods in excess of one week the most likely agents would be: Salmonella typhi, Fasciola hepática, Cryptosporidia sp or Giardia lamblia, among others. In such cases it is unnecessary to survey food consumed within the last 72 hours. Instead, in accordance with calculations obtained through the epidemic curve and taking into account the possible incubation period of the disease, investigation would focus on:

The maximum number of people should be interviewed during the investigation. However, when numbers are very high and there are insufficient resources, a sample for interviewing may be selected as follows:

During the interviewing of the sick, attempts should be made to find other individuals who may have had some connection, whether in time, place or person, in order to increase the number of respondents. In particular, a check should be made of recent complaints or reports from doctors, etc. that could be related.

Every epidemiological study should have a control group, since otherwise there can be no statistical analysis. Ideally an equal number of persons should be surveyed who had not fallen sick but who were exposed to the same conditions as those who fell sick.

The signs and predominant symptoms contribute to determining if the causative agent of the outbreak is the result of a toxin, an enteric infection, a generalized infection, a localized infection, or a neurological disease. Its utilization has also been previously mentioned in the request for reviews, and for this reason, in addition to its usefulness in determining the reviews, this information should be sent to the laboratory.

The percentage analysis of the symptoms and signs determines the greater frequency, and is used to define the case of FBD in the outbreak.

The incubation period is the time that elapses from the ingestion of the contaminated food up to the presentation of the first signs and symptoms of the disease. It is determined from knowledge of the exposure time and through calculation of the incubation period of each case from the survey. The incubation period may vary and the extent depends on individual susceptibility, the agent, the quantity of the food consumed, and the size of the innoculum in the food, among other causes. The calculation of the median of the incubation period helps to decide if the disease under investigation is a food poisoning or a microbiologically induced infection, and helps to prepare a hypothesis on the causative agent and thus suggest the most appropriate laboratory tests (Annex D).

An epidemic curve is a graph that presents the distribution of cases over time, from the date of the first symptoms and considering all those affected in the disease outbreak. Use of a bar chart is recommended, where each case is represented by a small square. The unit of time employed in the design of the graph depends on the period covered by the outbreak, which will vary according to the disease in question. For example, a scale in days or weeks would be used for hepatitis A and a scale in hours for staphylococcal food poisoning.

The epidemic curve helps determine if the outbreak originated from a common source, such as food or water, (Figures 3, 4 and 5), or was spread from person to person, figure 6:

In the following examples, the figures on the vertical axis represent the number of cases; those on the horizontal line indicate the days of the month or the hours of the day.

When the exposure time and the time when the disease appeared are known, the period of individual incubation can be calculated directly and then used to calculate the average.

If only the time when the disease first appeared is known and the epidemic curve suggests a point of origin of the outbreak, an inference can be drawn concerning the average incubation period, and thus the probable time of exposure can be calculated through the epidemic curve:

Figure 7: Determination of the average of the incubation period and probable time of exposure in the point of origin of the outbreak

If the microorganism and the moment of appearance of the disease are known and the epidemic curve suggests a point of appearance of the outbreak, then the probable exposure time can be determined by the epidemic curve in the following example:

Figure 8: Determination of the probable exposure period in the point of origin of the outbreak with a known pathogen

Determining the suspected food through calculation of the specific attack rate. Retrospective cohort analysis.

When one specific food in a meal has been determined as the cause of an outbreak or when an event is suspected, a table to determine the attack rate of each specific food is prepared (Table 4). Retrospective cohort analysis is used when the group of people who attended the event or meal is known and can be interrogated about the disease and the exposure.

The attack rate table for a specific food compares the attack rate among the patients who ingested the specific food in an event or meal with the attack rate of patients who were at the event or meal but did not eat the food in question.

For each food, the total number of cases that fell ill should be noted, including those who consumed the food and those who did not. Continuing with the same food, those who did not consume it are also noted, and within that group the numbers who fell ill and the numbers who did not.

The attack rate is calculated firstly from amongst the number of those who ate the food. It involves counting all those who ate the food and fell ill, divided by the total number exposed to eating the food (i.e. those who fell ill and those who did not), multiplied by 100 { e.g. - E/(E+NE)x100 }. Next, the attack rate is calculated from amongst those that did not eat the food. The number is taken of those who did not consume the food but became ill and this is divided by the total (i.e. those who became ill and those who did not) of all those who did not eat multiplied by 100 { e.g. - E/(E+NE)x100 }. The difference is found between those who consumed and those who did not consume the food. The food that has the highest percentage of who ate and became ill coupled with the least percentage who did not consume the food but became ill can be inferred to be the suspected food.

Arrangement of the data is important for all forms. It is recommended that a summary table be prepared that shows, for each food, how many sick persons ate and did not eat it, and similarly how many healthy people ate and did not eat it. See the example below:

The total number of persons should agree with the total number of persons in the table that was prepared in obtaining the attack rate.

The difference between the attack rates is (+ 81) which, being the highest number, indicates that pork is the suspect food.

This calculation is carried out for each food or beverage, and the differences in the rates for each food and the results are entered in the corresponding column.

The food that has the highest attack rate for people who ingested and the lowest attack rate for people, who did not, i.e. the food with greatest difference between the two rates, becomes the suspected food. For example, in the previous table the attack rate for people who ingested pork was 81% and the rate for those who did not ingest pork was 0%.

The difference between these two rates (percentage difference) was (81), greater than the difference for any of the other foods.

A much more exact way of identifying the responsible food is through the calculation of the relative risk (RR) which provides a better guide to the identification of the vehicle than the percentage of difference. The relative risk (RR) is sought for all foods with a high percentage of difference.

The relative risk demonstrates that the attack rate for those who consumed pork was (81) times higher (81/0) than for those who did not.

Some people who did not ingest the food or beverage but nonetheless became ill are sometimes tabulated as patients. Possible explanations are that some of them may have forgotten which foods they ate, some may have become ill from other causes and some may display psychosomatic symptoms rather than symptoms induced by an etiologic agent. On the other hand, agents that are infectious in low doses may have been contaminated through intermixing. It is also not unusual to include in the table people who ingested contaminated food but did not become ill. A reasonable explanation of this would be:

The case-control study is employed when all those who were exposed to the risk cannot be identified or when only a proportion of sick persons (cases) and healthy people (controls) can be interviewed about their exposure.

For both cases and controls, the percentages of those who consumed a specific food and those who did not ingest the food are calculated. The two percentages are compared, and the attributable risk is found as verification.

Usually only a proportion of the sick or those subject to risk can be selected for comparison because not all cases and controls can be identified or used as respondents. The odds ratio inspires greater confidence and provides a better guide in the identification of the correct vector than can be obtained from the difference of percentages.

When two foods have a similar attack rate a stratified analysis comparing the attack rate for each food is recommended.

As shown by the following table, the stratified analysis comparing the attack rate for specific analysis (cross table), the attack rate for those who ate and those did not (e.g. turkey) is compared with the attack rate for those who did not eat the other food, perhaps the salad. The total values in the table fit values in the table of the attack rates of specific foods, but the entries in the table of stratified analysis must be obtained from data in FBD Surveillance Form 2 or from individual case histories, FBD Surveillance Form 1.

The highest attack rate was found among those eating turkey (73% and 75%) and the lowest rates among those not eating turkey, (0% and 8%). This comparison provides additional evidence that the turkey was the vehicle of the outbreak.

Epidemiology investigates the associations that may exist between the state of health or disease of a population and the factors related to these states. By association is understood the relation that may exist between two or more factors, events, or circumstances in the production of a given phenomenon. There are `causal` associations and `non-causal` associations.

Causal association: exists between two categories of events when, on altering the frequency or the quality of one, a change is produced in the frequency or the quality of the other. If increasing the first factor leads to an increase in the other one, then there is a ‘positive causal association’; while if increasing the first factor produces a reduction in the other, then there is ‘negative causal association’.

Non-causal association: exists when association between the two events depends on a third factor.

Primary cause and secondary cause: A primary cause is one that produces a direct effect, while a secondary cause produces its effect through intermediate stages.

Sufficient cause and necessary cause: It is important to differentiate between a necessary cause (presence of danger, such as the infectious agent) and sufficient cause, which are those elements such as the nutrients, pH, moisture, temperature, time, etc. that permit the multiplication of the germ and the production of toxins.

Association in time: exists, for example, when cases of a disease, with similar characteristics, emerge within a short space of time.

Place association: exists when people obtained food from a single place, ate in the same establishment, attended a single event, live in a single place, etc.

Association of people: suggests shared personal characteristics such as age group, sex, ethnic group, occupation, social group, or religion.

When any of these associations appear obvious, other people who could have been at risk through association of time, place, or person with those who became sick should be interviewed.

At this stage of the study, it is reasonable to carry out a preliminary evaluation of the collected data and lay out a hypothesis of `causal` factors to determine if the preliminary hypothesis is maintained or a new one put forward. A brief informal meeting with the members of the team at the outbreak location can be used to organize all the data collected to date for subsequent analysis.

the vehicle involved;
the probable time of exposure to the contaminated food;
the mode of transmission of the causative agent and the source whether single or multiple.

On the basis of the data analysis it is possible to determine the seriousness of the disease and its prognosis, the number of exposed people (who shared the meal), the number who became ill, the suspect food, the contributing factors and other factors.

If during research it is considered that either the magnitude of the outbreak, or the aspects to be investigated, exceed the capacity of the equipment, then the involvement of other levels of the organization or of outside experts should be requested.

During the outbreak investigation, there should be a continuous search for and survey of all sick and healthy persons associated in time, place or person. Medical reports and complaints from people and other sources should be reviewed in order to detect new cases.

The research procedure may vary depending on available human resources, and the sequence of actions might vary depending on current needs. Although additional procedures may be required, the principles and techniques described will be sufficient for most investigations.

In deciding if the observed association constitutes a causal relationship between the exposure and the disease, the following questions should be answered:

There are predisposing factors inherent in the collection of epidemiological data and the association between disease and exposure. These include selection, information, and confusion influence. An example of erroneous selection is when the procedure used to select the controls differs from that used to select the cases. The identified cases are sometimes those which have required medical care, while the less severe cases of illness have not come to the attention of the physician. In addition, the laboratory-confirmed cases represent only a part of the total number of people who became ill during the outbreak, consequently, the less severe cases are never recognized, which results in an underestimation of the number of sick persons.

The information can be influenced during interviews based on incomplete classification of the exposure of cases and controls resulting from an inefficient search for persons who were afflicted.

In addition, the cases and controls may describe their experiences differently: For example, persons who know or suspect that their illness is food-borne in origin may claim that they ate a food that they really did not, or that they consumed a greater quantity than they actually did, while the controls may not remember.

Confusion can be caused by a second food or activity that is related to the disease and to the vehicle in question, but that is not in fact the cause. Error from such sources can sometimes be removed by calculating the specific rates. The biases may consist of a mixture of major and minor factors and, collectively, can change the course of the outcome.

There are two commonly used measures of disease association - relative risk and attributable risk. The choice depends on the manner in which the data are analyzed. The relative risk (RR) is calculated in cohort studies, while the attributable risk (AR) is calculated in case-control studies. Both calculations begin with a 2 X 2 contingency table that compares groups of patients with exposed and unexposed persons.

An example is presented in the following table: Both can be interpreted as consumers who have "X" times more risk of becoming ill than non-consumers. Very small samples may result in inaccurate measurement of disease and exposure.

RR=1: There is no difference in becoming ill between the exposed and the unexposed.

RR<1: The exposed group is at lower risk of becoming ill than the unexposed group.

RR>1: The exposed group is at greater risk of becoming ill than the unexposed group.

In this comparison the people who ate pork and became ill were at greater risk (approximately 80 times) than those who did not eat pork.

This shows that there is an association between those exposed and those who fell ill, but it is not, however, proof of causality. This calculation assumes that other risk factors for those who ate pork (exposed) and those who did not eat (unexposed) are approximately equal.

The odds ratio in calculating risk is applicable in situations where it is not possible to obtain data from all those who were exposed to a potential danger. In such studies, the histories of those presenting with the disease (cases) are compared with the histories of exposure of a similar population (in terms, for example, of age, living in the same neighborhood, attending the same event or other shared conditions) who did not become ill. It is impossible to calculate the real risk of a case-control study, but the attributable risk is used as an estimate of the risk.

Odds of exposure among the cases a/c ad
OR= ------------------------------------------ = -------- = -----
Odds of exposure among the controls b/d bc

Using a 2 X 2 table of contingency to calculate the risk will show whether the people who became ill had a higher probability of having eaten the food in question than those who did not become ill.

OR = 1: There is no difference in exposure between cases and controls, the presentation examined was not in any way associated with the disease.

OR < 1: Cases had less probability of exposure to the suspicious agent than the controls.

In the last example, the risk of exposure was greater for the cases than for the controls. As a result, the odds of exposure (consumption of pork) were greater for the group that became ill, and pork was probably the vehicle for the etiologic agent.

In order to identify the possible elements in a disease outbreak, it will be necessary to develop a most highly technical inspection team, employing an epidemiological approach and the principles of the HACCP System. The places where the suspicious food was produced, processed, packed, prepared, transported, stored, and served should be investigated, this being an important element in a review of the food and its preparation.

Investigation of the source of contamination, and the factors of contamination, survival, and multiplication are carried out from the endpoint, that is, from where the outbreak occurred and hence it is necessarily a very exhaustive retrospective study.

The investigating personnel should, first of all, introduce themselves to the manager or administrator of the center, informing them of the objective of the visit and inspecting all the locations where the suspected food was produced, processed, prepared, or served, to determine the events or activities that contributed to the FBD outbreak, as well as to take the preventive measures necessary for controlling the outbreak and avoiding repetition.

Activities and controls on critical operations should be verified before they are modified, and food samples obtained before they can be discarded.

Routine inspections frequently do not detect the processing factors that are critical to production of a Food Borne Disease. In addition, inspections may have been carried out months before at a time when operations, the food processed, and the personnel were different. They may have been done also at a time when the production levels were lower, and no rules were being violated, as they would be during times of increased pressure of work.

Inspection of an establishment where an outbreak has occurred should be carried out with extreme rigor and given sufficient time to evaluate all possible processes, starting from the beginning of the process and culminating with cleaning and disinfecting.

During the review of food processing, from receipt to serving, workers at each station should be interviewed and any observations should be compared with the procedures and versions outlined in meeting with management. If the two are incompatible, the one in agreement with epidemiological judgement is more likely to be what actually occurred.

The inspection should obtain maximum information on the management of the operations and the handling of the food implicated. For example:

If the investigation is related to products associated with slaughtered animals, the cleaning and washing of the animals, and methods of skin removal, deboning, cutting, and freezing of the carcasses in the abattoir’s environment should be monitored. It is important to evaluate the available quantity and quality of water since this has often been a source of contamination.

In a processing plant the recording and evaluation of critical limits in the processes used for heating, cooling, freezing, drying, fermentation, acidification, smoking, packing, storage, etc. should be taken into account.

Information should be collected on the ingredients or components added to food that has undergone heat treatment.

The physical state of transportation equipment should be evaluated, as well as cleaning and temperature and the possible transportation of contaminated products.

Animal carcasses can be contaminated during slaughter and processing by agents such as Salmonella, Campylobacter jejuni, Yersinia enterocolitica, Clostridum perfringens, Staphylococcus aureus, or other pathogens as a result of being themselves colonized or because they are contaminated during some of the processes.

If any of these agents is suspected of producing an outbreak, samples should be taken of the meat or segments, equipment with which they have been cut, and waste or dripping from the drains, all of which will help in identifying the source of contamination.

Swabs from working surfaces (chopping tables, blenders, other utensils) that come in contact with food can also help in establishing a relationship in transmission of contaminants.

The origin of foods in establishments that serve or sell food, in markets and residences must be investigated. In order to achieve this, the following should be checked - reception, contracts, invoices, health certificates, inspection and temperature records, food preparation, cooking, handling after cooking, storage of hot foods, cooling, warming and serving. In all cases, temperature records should be checked and the data obtained used to prepare a flow chart of every product investigated, taking into account when the food was prepared, the ingredients used and the source or origin of any significant ingredient.

The workers involved in preparation of the food under investigation and the process employed should be identified.

Initial food sampling, should be non-specific in order to avoid loss of hazardous foods. However, those that appear to have the highest attack rates in the survey of specific foods should be given priority. Sampling of raw materials and products being processed is also important. Rooms or warehouses where products similar to those responsible for the outbreak may have been stored should be checked.

The results obtained should be interpreted with care since, depending on the place where the contamination occurred, growth can occur according to the steps in the process, the type of food, the atmospheric temperature and the length of time the food was in storage.

If no samples or remnants of the foods that have been clearly implicated are available then samples of the foods prepared subsequently under the same conditions should be taken.

According to the situation, it might be necessary to sample both ready-to-eat foods and products in preparation.

In the entire sampling process, only clean and disinfected knives, spatulas, and other utensils should be used, or erroneous results might be obtained.

Samples thus collected should be placed in glass containers with covers or in sealed plastic bags, and kept refrigerated until analyzed. If there are no refrigeration facilities, then the samples should be stored on ice. The samples should never be frozen because certain bacteria (such as the Gram-negatives and vegetative forms of Clostridium perfringens) die rapidly during the storage in freezing conditions. When filling in the shipping form, note the temperature of the sample, the time elapsed since preparation of the food, the identification code given and the sequence number given by the sample unit. The shipping department should save the codes, dates and time of the sampling, the type of sample, required proofs, etc. For more details, see Annex C.

The choice of tests on the samples collected depends on the information obtained from the epidemiological survey, in particular predominant symptoms, incubation period and the food that presents the greatest difference in the attack rate. Other epidemiological factors should be taken into account considering the presence of chemical or biological agents in that environment.

Although the laboratory staff should be included in the investigating team, if for some reason they were not, they should be notified in order to coordinate sample shipments according to criteria agreed on in the study. Whenever possible, samples should be transported by means that will prevent deterioration in the agents.

When Staphylococcal poisoning is suspected, swabs should be taken of the nostrils, open wounds, etc. of all who handled the suspect food. Swabs should also be taken of any superficial injuries after disinfecting the surface or, in the case of an abscess, through aspiration with a syringe.

Each specimen should be placed in a separate test tube containing a sterile preservative solution or transport medium for dispatch to the laboratory.

When there are indications that the outbreak was caused by Salmonella, Shigella or other organism that causes enteric infections, rectal swabs should be collected from people who handled the suspect food.

Another method that is not highly recommended because of the possibility of errors, is to give to each person who handled the suspected food a container for a sample of the `specimen`. In any case the person should be duly oriented.

Anyone who thinks that he/she may be responsible for an outbreak could deliver a `specimen` from another person, thus falsifying the test.

Isolation of a pathogenic microorganism from a handler’s fecal `specimen` and from the suspected food does not support the immediate conclusion that the worker was the source since the worker could have consumed the same food and therefore be a victim rather than a carrier. An epidemiological history of the handler showing evidence of a skin infection or of gastrointestinal or respiratory problems before or during the preparation of the suspected food could be more incriminating.

The temperature of food should be taken at the start of processing or warming, specifying the maximum temperature reached and the time elapsed until the temperature falls below 55ºC (131ºF).

The temperature is taken during processing and storage, recording the sequence of operations. If it is felt that the environmental temperature could have an effect on the product, then this too should be measured and recorded.

The temperatures and length of storage of foods should stay within a range that inhibits rapid multiplying of bacteria. 4ºC (39,2ºF)

The average period needed by food to reach a safe cooling temperature, in industrial, commercial and domestic circumstances, should be evaluated.

The temperature of foods under investigation should be observed and measured since they may sometimes be stored near a heat source and may then reach an ideal temperature for incubation. Foods could possibly remain at those temperatures for an extended period inducing an alarming increase in bacterial agents.

One of the most important factors in the prevention of the FBD is that the food that is to be cooled or preserved reaches an appropriate temperature 4ºC (39,2ºF) in the shortest possible time. To this end the dimensions of the container used to keep it cool and the volume of the food should be measured. Very wide or very deep containers should be avoided since these may hinder rapid cooling, thus promoting rapid microbial growth. The average cooling time and the potential for bacterial growth should be calculated.

Take into account the use of protective devices that prevent contamination and production of disagreeable odors but which also impede rapid cooling.

It is also necessary to check the placement of containers in the refrigerator (which can influence cooling or cross-contamination) and if the establishment uses forced ventilation or any other type of rapid cooling, for example, frozen water.

Preparation of the thermal curve is achieved by fixing the measures of time and temperature in a Cartesian system where the vertical coordinate "Y" denotes temperature and the horizontal axis "X", the time. The suggested temperature guidelines are:

A temperature of 46ºC (114,8ºF) is ideal for the growth of the Clostridium perfringens, and 30^ºC (86ºF) is ideal for Bacillus cereus and other pathogens.

Data should be interpreted on the basis of the optimum growth temperature for the microorganisms of that class and temperature at which they can multiply. Also, based on the highest temperature reached and the time/temperature exposure, the heating and cooling curve can be interpreted to determine if the pathogens in question could survive the heating process, if there was insufficient cooking, or if contamination occurred subsequently during storage or cooling.

For preparation of the "flow chart" there should, above all else, be a form filled in during the previous steps and which lists the source of foods and ingredients, the people who participated in the preparation, the procedures employed, the thermal processes specifying temperatures and their duration, records, the potential sources of contamination during preparation, and the conditions of time and temperature to which the foods were exposed from their preparation until they were served.

The composition of foods, the ingredients of which indicate possible contamination by the probable agent should be checked for potential factors of contamination, survival and proliferation of the agents.

When the outbreak occurs in service centers the suspected food should be investigated with regard to whether the food was prepared hours or days before being served and if a method different from that of other days was followed in the processing, acquisition of raw material, preparation, production or preservation.

For each food involved and according to the information obtained and evaluated, a flow chart should be prepared which will in many cases, allow for the pinpointing of errors made during the process. Take into account not what is prescribed but the actual practices of workers who often make modifications that are not recommended.

In the flow chart, each operation is represented by a rectangle, inside of which is the name of the operation and other information pertinent to the operation. The arrow shows the direction of flow and within each rectangle a symbol representing the best estimate of the probable type of contamination, probability of survival or destruction during treatment or other processing designed to render pathogens or toxic substances harmless; or the probability of multiplication. The temperature and duration of each process must be measured, specifying the measurements of the container and the thickness of the food stored. Flow Chart FBD Surveillance Form 8 is recommended for this purpose.

Some foods are quite likely to be contaminated from source with Campylobacter jejuni, C. perfringens, Salmonella, E.coli, Staphylococcus aureus, Yersinia enterocolitica or other pathogens.

Contamination can occur subsequently through handlers or from equipment where the food is prepared, such as tables, meat grinders and other kitchen utensils.

Cross-contamination might have occurred through the washing of contaminated table, meat grinder, knife, or kitchen towels, as well as from the hands of workers who first handled raw, and then cooked, food.

The exposure time may vary, since some spores can survive cooking, particularly when there has been incomplete thawing. Bacterial growth might have occurred in very deep and large containers during cooling. Vegetative cells might have survived re-heating and subsequent storage on the hot/holding table.

All information should be transferred to a graph, to be confirmed from conversations with the people involved, through observation of subsequent processing (if this takes place), through food analysis, and according to information on critical control points.

Review the forms for date, time, registered temperature, and people who carry out the monitoring, annotations on the deviations from the critical limits, annotations of the corrective measures taken where deviations occurred, and evidence that the annotations could have been falsified. Falsification may be characterized by frequent recording of an exact critical limit, highly similar daily recordings, or records that are uniform or illogical as suggested by experience of typical records.

Everyone involved in the acquisition, storage, handling, and processing of foods should be interviewed.

Workers who think that they might be criticized or experience punitive action for their possible role in the outbreak will not always give a true account of how the food preparation processes work. If the information obtained does not meet the objectives of the investigating team, the investigation should continue among other people aware of the phases through which the food underwent during preparation or serving. Various versions will be obtained until the most logical means of contamination, and of survival and multiplication emerge, given the epidemiological characteristics of the outbreak.

Research should always be done so that it reflects the flow of the food from reception until being served. Each worker should describe the operations that were carried out and from these descriptions it should become clear whether the workers understand the elements of `food safety` in their work.

While questioning workers, it should be borne in mind that they could be the source of pathogens such as Staphylococcus aureus, which can be found in the nostrils, skin and human feces; or Shigella, Salmonella sp, Salmonella typhi, virus Hepatitis A, Clostridium perfringens, found in feces. Note that Clostridium perfringens can be found in the feces of healthy persons.

It is very important to ask about recent illnesses among handlers or members of their family, especially those with an obvious link related to the incubation period. Absentism among workers should be reviewed for the likelihood that they may have been caused by diarrheal or related diseases. This may establish a connection of the diseases of the workers or their family members with the "case" definition created for the outbreak study.

Where workers are able to report a disease with an incubation period that coincides with the outbreak, a form should be filled out for signs, symptoms, and foods consumed up to three days before the disease.

During the clinical review, the presence of pimples, skin eruptions, boils, earache, infected wounds, or other disorders not visible to the naked eye should be investigated.

The cleaning of equipment and utensils is fundamental to the prevention of cross-contamination between raw foods, which come in contact with the equipment, and foods that were treated and therefore up until then free from microorganisms. The cleaning of the equipment and utensils should be observed and methods and facilities established for cleaning and disinfecting. As an element of control, samples can be collected through swabbing equipment, utensils or surfaces that are, or have been, in contact with the food, and storing them under refrigeration until analyzed.

The pH of foods is a very important factor in bacterial proliferation and thus its control is important in many cases, as with mayonnaise and other marinated foods. Improper control of this parameter has been a cause of outbreaks.

The various factors that contributed to the contamination of the food, to the survival of the agent through improper treatment, as well as the factors that permitted the proliferation will gradually be revealed during research. The principal factors in contamination, survival, and proliferation that should be considered in the corresponding reports are annexed in order to guarantee homogeneity of information. (Annex G).

In every outbreak there are household members who did not eat the suspect food but became ill, and others that ate it and did not become ill. This may occur for the following reasons:

All results from sampling foods, `specimens` and the environment should be compared with the epidemiological information obtained.

Results of the sanitary inspection of the place where the outbreak occurred should be available and the research data used to test the hypothesis formulated during the investigation. Each of the following factors should be considered with regard to the suspected agent.

Test samples never take the place of direct observations made by a keen observer, however if they are taken in the correct place and time and are analyzed by experienced personnel, they mayl provide invaluable information. A sample that is poorly taken, transported, or analyzed sometimes gives a negative result, which does not mean that the food is free from the microorganism. Detection is always more likely with higher levels of contamination and where there are no competitive flora. Thus, when the level of contamination is thought to be low, the number of samples should be increased. Bear in mind also that the infective dose of some agents is extremely low.

Contamination of a food is rarely homogeneous and solid foods should therefore be thoroughly ground or mixed. Occasionally in preparing a food, only one part of it comes in contact with the contaminated part of the container, or only one part is contaminated by contact with the dirty hand of a handler.

Contamination in a given food can increase or decrease in accordance with factors intrinsic and extrinsic to the food and the place in question. Bacterial multiplication is unlikely in solid food but is far more likely in liquid or semisolid foods.

Among the extrinsic factors, is storage temperature that can vary even in a single room or refrigerator. In the same way, improper cooling may occasionally permit multiplication of agents within a food because heat transfer, resulting either from insufficient cooling, the size of the container or other factors, makes for very variable bacterial multiplication.

It should be borne in mind that multiplication of saprophytic bacteria might hinder the isolation of pathogens.

The correlation established from the results of food sampling should be interpreted carefully since the time that elapsed between the ingestion of the food and the taking of the sample could alter the bacterial counts, depending on how the food was preserved.

In the majority of outbreaks the agent is never identified because of the non- collection of clinical `specimens` at the right time, because specimens were incorrectly stored or transported, insufficient quantities of sample were taken, or the search for the responsible agent in the outbreak was not carried out.

When high counts of mesophilic `aerobic` microorganisms are present two kinds of situations may have occurred. The first is that the raw food or ingredient contained high microorganism concentrations and the ingredient or product in question was untreated, or was insufficiently treated to reduce the microbial concentration. The second is that the food was stored at temperatures at which enough bacterial growth could occur and spores that survived could germinate and the resulting cells multiply. Pathogens, if they are present, may or may not multiply since the normal flora could inhibit multiplication of pathogens in foods that grow in the soil or are exposed to it during harvesting. In these foods one can expect the presence of bacteria from the soil. Marine products are more likely than other foods to exhibit the presence of marine microorganisms.

Bacteria characteristic of fecal contamination such as: coliforms, heat tolerant or fecal coliforms and germs of the family Enterobacteriaceae usually originate in raw animal foods and their presence in thermally treated food suggests retro-contamination. High concentrations suggest that multiplication occurred after treatment.

In general, the total number of coliforms, fecal coliforms, E. coli, and organisms of the family Enterobacteriaceae serve as indicators of post-treatment contamination. Salmonella has been used as an indication of surviving the thermal process such as in sausages, or pasteurized egg.

The presence of Salmonella and E. Coli is often due to the cross-contamination of meats from contaminated surfaces or equipment.

Staphylococcus aureus can be used as an indicator of improper handling of treated food.

Counts of these microorganisms and colonies of mesophilic aerobes can also indicate abuse in the time-temperature indicator.

Other pathogens such as B. cereus are examined qualitatively or quantitatively in rice and other cereals, grains and milk; Vibrio parahemoliticus in fish and shellfish; C. perfringens in meats, cooked poultry, and grains.

Epidemiological information can suggest the need to examine certain foods for specific pathogens or indicator microorganisms.

Epidemiologically, the presence of some pathogens (Salmonella, Shigella, E. Coli) in the implicated food is sufficient for confirmation; however, for other pathogens such as Staphylococcus aureus and Clostridium perfringens values higher than 100,000/g or ml must be found for confirmation (see Annex F)

The information obtained from the laboratories should be discarded when it indicates that causative agents of the outbreak are microorganisms that are indicators rather than pathogens, as happens when the following are found: a count of `aerobic` mesophiles, coliforms, fecal or heat-tolerant coliforms, Enterobacteriaceae, etc. The presence of these in a food, even in large quantities, does not indicate that the food was the cause of the outbreak.

Definitive typing is fundamental when we wish to identify or "draft" the origin of contamination. To confirm the involvement of a suspected food, the same organisms, toxins, or chemical markers should be found in the implicated food as in the `specimens` from patients. The organism can be identified by serotype, phagetyping, analysis of drugs antimicrobial resistance and more.

When clinical `specimens` are not available, a vehicle can be identified, at least circumstantially, through the detection of toxic substances (such as Zinc or botulinum toxin), through isolation of a significant number of specific pathogens (such as 100,000 ufc/g or more of Staphylococcus aureus or Clostridiun perfringens,) from the food or by the recovery of enteric pathogens, such as Salmonella, E.coli, Shigella, from a food through enrichment techniques.

The food from which these isolations are made will be `epidemiologically` suspect as a result of analysis of the attack rate or because of the case-control survey, and the symptoms reported by the person should be consistent with the isolations of agents found in the implicated food.

`Typing` and other epidemiological tracers are vital to achieving confirmation of the agent, as can be seen in Annex 6.

Although in practice it is difficult to achieve a complete investigation of all the factors present in the outbreak, the research team should make the maximum effort to collect the most complete information for its value in the prevention of future outbreaks.

Before making final recommendations the hypothesis formulated should be confirmed. Annex 6 presents the criteria for confirming an outbreak of FBD associated with bacteria, virus, or parasites when the cases have clinical profiles characteristic of the disease. Otherwise, there is the risk of taking ineffective control measures.

Based on the conclusions, a report should be prepared that may be distributed immediately, and can then be published. It should be sent to the next level of the health administration, to the laboratory, and to the Food Safety service. If the outbreak was produced by a food that was widely distributed in several establishments and as a result the people who ate it were at risk, the distribution should be ascertained and all levels of the health infrastructure (even internationally) informed so appropriate control measures may be taken as rapidly as possible.