Resources - Epidemiology

Epidemiology

 

Epidemiology

From Wikipedia, the free encyclopedia.

 

Epidemiology is the study of the distribution and determinants of health-related states or events in specified populations, and the application of this study to control of health problems (Last 2001).

Epidemiology is the scientific study of factors affecting the health and illness of individuals and populations, and, in this capacity, it serves as the foundation and logic of interventions made in the interest of the public’s health.

The acting epidemiologist works on issues from the practical, such as outbreak investigation, environmental exposure, and health promotion, to the theoretical, including the development of statistical, mathematical, philosophical, and biological theory. To this end, epidemiologists employ a range of study designs from the observational to experimental with the purpose of revealing the unbiased relationships between exposures such as nutrition, HIV, stress, or chemicals to outcomes such as disease, wellness, and health indicators.

 

Epidemiological studies are generally categorized as descriptive, analytic (aiming to examine associations, commonly hypothesized causal relationships), and experimental (a term often equated with clinical or community trials of treatments and other interventions).

 

Epidemiologists work in a variety of settings. Some epidemiologists work "in the field", i.e., in the community, commonly in a public health service, and are at the forefront of investigating and combating disease outbreaks.

 

Epidemiology as causal inference

 

Although epidemiology is sometimes viewed as a collection of statistical tools used to elucidate the associations of exposures to health outcomes, a deeper understanding of this science is that of discovering causal relationships. This conceptualization of epidemiology is difficult to grasp because our internal notions of causation are often poorly developed, frequently being predicated on the notion of a one-to-one relationship. For example, almost everyone would agree that gravity causes a dropped ball to fall towards the ground, but would most agree that drinking one glass of milk a day will cause weight loss? Even very heavy smokers know that their vice causes lung cancer, but only 10% of life-long smokers will get lung cancer. How can this be?

The answer is complex and delves into the philosophical notions of causality, induction, deduction, logic and other dense topics. It is nearly impossible to say with perfect accuracy how even the most simple physical systems will behave, much less the complex field of epidemiology that draws on biology, sociology, mathematics, statistics, anthropology, psychology, and policy. However, for the epidemiologist the key is in the term inference. As epidemiologists, we use gathered data and a broad range of bio-medical and psycho-social theories in an iterative way to generate or expand theory, to test hypotheses, and to make educated, informed assertions about which relationships are causal and exactly how they are causal.

 

Epidemiology as advocacy

 

An alternative aspect of an epidemiologist’s duty is to advocacy for the health of populations, bearing in mind the outpost perspective they have over factors that affect a whole population. In this capacity the epidemiologist is not limited by the strict requirements for scientific accuracy. This of course does not mean that the epidemiologist can advocate for whatever position they please independent of the data, but presentation of the data can take more artistic modes to engender behavior or perspective change. For example, consider these two alternative admonishments against smoking:

 

1. smoking has been consistently linked to health problems such as lung cancer and coronary heart disease in several large prospective studies, this link has been deemed causal by a complex process of induction, consensus, and modeling.

 

2. Smoking will kill you. Although statement one is more accurate, statement two has an air of finality and explicit causation that may help to reduce the rate of smoking, albeit scientific and philosophically questionable.

The best epidemiologist as advocate will consider the broader intellectual landscape beyond the epidemiology and public health literature to render judgment on a course of action for a population, in this manner they are employing a different analytical framework than the Cartesian framework that is more common in scientific epidemiology. However, it is rare for one person to wield the skills and embody the features required to be a leader in both the scientific and advocacy aspects of epidemiology.

 

Measures

  1. Measures of occurrence
    1. Incidence measures
      1. Incidence density (also known as Incidence rate) (Szklo & Nieto, 2000)
      2. Hazard rate
      3. Cumulative incidence
    2. Prevalence measures
      1. Point prevalence
      2. Period prevalence
  2. Measures of association
    1. Relative measures
      1. Risk ratio
      2. Rate ratio
      3. Odds ratio
      4. Hazard ratio
    2. Absolute measures
      1. Risk/rate/incidence difference
      2. Attributable risk
        1. Attributable risk in exposed
        2. Percent attributable risk
        3. Levin’s attributable risk

 

History of epidemiology

 

Dr. John Snow is famous for the suppression of an 1854  outbreak of cholera in London's Soho district. He identified the cause of the outbreak as a public water pump on Broad Street and had the handle removed, thus ending the outbreak. (It has been questioned as to whether the epidemic was already in decline when Snow took action.) This has been perceived as a major event in the history of public health and can be regarded as the founding event of the science of epidemiology.

 

Other pioneers include Danish physician P.A. Schleisner, who in 1849 related his work on the prevention of the epidemic of tetanus neonatorum on the Vestmanna Islands in  Iceland. Another important pioneer was Hungarian physician Ignaz Semmelweis, who in 1847 brought down infant mortality at Vienna hospital by instituting a disinfection procedure. His findings were published in 1850, but his work was ill received by this colleagues, who discontinued the procedure. Disinfection did not become widely practiced until British surgeon Joseph Lister "discovered" antiseptics in 1865 in light of the work of Louis Pasteur.

 

In the early 20th century, mathematical methods were introduced into epidemiology by Ronald Ross, Anderson Gray McKendrick and others.  Another breakthrough was the 1956 publication of the results of a British doctor's study, which lent statistical support to the suspicion that tobacco smoking was linked to lung cancer.

 

References

  • Last JM (2001). "A dictionary of epidemiology", 4th edn, Oxford: Oxford University Press.
  • Nutter FW Jr (1999) "Understanding the Interrelationships Between Botanical, Human, and Veterinary Epidemiology: The Ys and Rs of It All. Ecosystem Health 5 (3): 131-140".
  • Szklo MM & Nieto FJ (2002). "Epidemiology: beyond the basics", Aspen Publishers, Inc.

 

 

 

Analyzing Multiple Risk Factors

 

 

The Event

Listed below are some of the initial steps that take place in an outbreak investigation.  This section will provide you with a basis of information to complete this module. 

Background Information


Recently, a national political action committee (PAC) held a fundraising event at a Hotel & Convention Center in Washington, D.C. to select which candidate they would support by providing contributions to the candidate's campaign.

The candidate would be selected by a popular vote that was scheduled to occur at the completion of the annual meeting, which took place during August 28--31.  Upon arrival at the meeting, members of the PAC attended a banquet to recognize the accomplishments of the group during the previous calendar year.

Within 24 hours of the banquet, a number of diners had become ill with diarrhea and at least one of the following symptoms:

-         Headache

-         Fever

-         Abdominal pain

-         Nausea

-         Vomiting

At this point, the local public health department has been notified and is onsite to conduct an investigation.

Identifying an Outbreak


Often, health departments are alerted of a potential outbreak by a health-care provider or the general public.  It is your responsibility to identify whether the event is truly an outbreak or a series of unrelated cases.

Establishing a Case Definition


A case definition is a "set of standard criteria for deciding whether a person has a particular disease or health-related condition, by specifying clinical criteria and limitations on time, place, and person" (1).  In the beginning of an investigation, the case definition can be fairly general, but as information is gathered, the epidemiologist may use a more specific description.

Initial Case Definition


An illness with diarrhea and at least one of the following symptoms: headache, fever, abdominal pain, nausea, and/or vomiting.  Onset of illness or symptoms occurred during August 28--31.

Identification of Cases


Often, the initial cases identified in an investigation represent only a fraction of the affected population.  Therefore, it is important to detect and identify other cases through interviews, questionnaires, surveys of health care facilities, newspaper articles, and other means.

During the next part of the outbreak investigation, the team obtained an occupancy list of all of the guests who were registered at the hotel.  

Using an adapted version of the foodborne outbreak investigation questionnaire developed by CDC, which contained questions pertaining to demographics, clinical information, exposure information, and a 72-hour meal history, the team divided the list and began to collect data in an effort to identify affected cases and systematically organize key information.

The team established that 105 of the 435 hotel guests had become acutely ill and exhibited many of the symptoms listed in the initial case definition.  All of these persons were members of the PAC.

Formulating Hypotheses


The next step in an investigation is to develop an initial hypothesis, based upon the information collected during early interviews and other information  provided by health-care facilities or medical institutions that treated the patients.

Initial Hypothesis


Onset of diarrheal illness of an unknown origin occurred during August 28--31.  The potential exposure to the unknown agent could have occurred at the PAC banquet at Hotel A in Washington, D.C.

Evaluating the Hypothesis


In order to identify the source of the epidemic, investigators must evaluate the relationships between exposures and the disease in question.  The following information will be helpful in evaluating the hypothesis presented on the following pages.

Additional Information


A descriptive analysis of the information gathered from the interviews and questionnaires identified the following characteristics of the distribution of the disease in question.

  • Person - members of the PAC
  • Place -  the PAC banquet on August 28
  • Time - the banquet occurred at 6:00 p.m., and the first onset of illness surfaced at 6:00 a.m. on August 29

This information suggests that attendance at the banquet was the common exposure, and the reported symptoms sound like an outbreak of gastroenteritis, which is commonly caused by food contamination.  

Next, we will discuss methods to organize the data in an effort to begin to make epidemiologic associations.

Potential Exposures

The team was able to determine from the interviews that the potential exposures in this outbreak include all the items served at the banquet:

-         Turkey

-         Gravy

-         Ham

-         Mashed potatoes

-         Green beans

-         Chocolate sorbet

-         Strawberry sorbet

  We need to discover, each of the potential exposures, which is associated with becoming ill.

Risk Factors

All of the potential exposures cited in the scenario so far could be classified as risk factors, which can be defined as "an aspect of personal behavior, lifestyle, or environmental exposure...which, on the basis of epidemiologic evidence, is known to be associated with a health-related condition considered important to prevent".(1)  

Since we are trying to determine what factors are associated with the illness, we must analyze the data in order to measure the frequencies with which a potential exposure can be associated with the illness.  Measures of association may be expressed in several ways.  

Attack Rates

One of the most commonly used measures expresses how many diseases are a consequence of a certain factor.  It can be expressed as an attack rate, which measures the proportion of people in a well-defined population that develop an illness of interest during a limited period of time.

These rates can be calculated by dividing the number of new cases among the population during the limited time period (X) by the population at risk at the beginning of the period (Y).  This number is then multiplied by (100) to obtain a percentage.  The formula can be expressed as:

As you can see, attack rates measure the probability or risk for becoming a case.  So the greater the difference between attack rates for those exposed to a particular risk factor and those not exposed, the higher the probability that the risk factor caused the outbreak.

Let's look at our previous example to calculate the attack rate for the entire population who attended the PAC banquet:

X = the number of ill persons who attended the banquet = 105
Y = the total number of persons who attended the banquet = 235

Risk Ratio

Before we can make any conclusions, we must also calculate the attack rates for the unexposed population (i.e., the population that did not eat the menu item).   In the table below, we have provided you with the attack rate of both the exposed and unexposed population so that we will be able to calculate the risk ratioThe risk ratio compares the amount of disease risk in the exposed population to the amount of risk in the unexposed population.  The formula can be expressed as:

Risk ratios can be evaluated by using the following criteria:

-         RR = 1.0 indicates identical risk in both groups

-         RR > 1.0 indicates an increased risk for the exposed group compared to the unexposed group

-         RR < 1.0 indicates a decreased risk for the exposed group compared to the unexposed group

Selecting a Reference Group

Selection of an appropriate reference or comparison group is critical to determining whether an association exists between the exposure and illness.  Typically, the reference group is the population of individuals that has had no exposure to the risk factor being analyzed.  Ideally, the exposed and unexposed populations should be similar, with the exception of the exposure.  Thus, if no direct association exists between the exposure and the illness, the attack rates will be relatively similar.

Unfortunately, the analysis of food-specific attack rates did not clearly point to any one of the menu items as the source of the outbreak.

However, one of the epidemiologists pointed out that eight persons arrived late to the banquet and were able to eat only dessert.  All of these persons became ill.

Of these eight dessert-only diners:

  • four ate chocolate sorbet
  • four ate strawberry sorbet

In addition, the outbreak investigation team realized through their analysis of the data that ten persons who ate the main course of the meal did not eat dessert.  Of these ten non-dessert diners, only two became ill.

Analyses of Multiple Risk Factors

"It is important to note that when several risk factors are being considered simultaneously, the non-exposed group should be defined as those with none of the risk factors under evaluation".(2)  In this case, the reference group is the population of individuals that were not exposed to (did not eat) the risk factors being analyzed.

We have established the group that ate no dessert as the "risk-free" reference group, as they have many of the same characteristics of the exposed population but did not eat any of the dessert items. 

Previously, when the risk ratios were calculated for the two dessert items, they were 1.32 for chocolate sorbet and 1.07 for strawberry sorbet.  There is a difference between these figures and those calculated above (2.25 for chocolate sorbet and 2.00 for strawberry sorbet).  There are several reasons for this discrepancy listed below.

-         Two high-risk populations were compared because the comparison (unexposed) group included all those who ate the other implicated food item (minus those who ate both items).

-         The resulting attack rate for the comparison (unexposed) group was relatively similar to the attack rate for the exposed population.

-         Therefore, the risk ratios were relatively low and did not appear to be statistically significant.

Bottom of Form

Knowing when to refocus and reanalyze like this is not easy; it is something that investigators generally learn from judgment and experience.   It is also essential to stay open and alert to clues.  A key clue was that among a small number of people who ate nothing but dessert, all became ill, and among a small number who ate no dessert, few became ill.

Further Investigation

After a thorough analysis of the data, the investigation team hypothesized that the primary contributing factors to the illness were associated with eating chocolate sorbet or strawberry sorbet.  

Early in the investigation, stool specimens had been taken from several persons.  In addition, the team also took stool samples from all of the kitchen staff on duty during the preparation of the food for the banquet.  All of the samples were sent to the local health department lab to be cultured for enteric pathogens.

Revising the Hypothesis
Once you have characterized the outbreak by time, place, and person and analyzed the data to determine risk factors associated with the disease, it is helpful to revisit the initial hypothesis and to determine if any changes need to be made on the basis of the information gathered so far.  Much of this information can be derived from descriptive epidemiology and the analyses of the individual risk factors.

For instance, we are able to deduce that the vehicles for transmission of the illness were chocolate sorbet and strawberry sorbet.  Unfortunately, at this point the causative agent (such as a microorganism, chemical substance, or form of radiation, whose presence, excessive presence, or relative absence is essential for the occurrence of a disease) is still unknown, and additional data will therefore need to be gathered after the revised hypothesis is developed.

Our revised hypothesis could read "The illness was caused by consumption of either chocolate sorbet and/or strawberry sorbet consumed at the Political Action Committee Banquet at 6 p.m. on August 28.

 

When the results were returned from the lab, the results indicated that:

-         No kitchen staff tested positive for any enteric pathogens,

-         Stool cultures from the early onset cases tested positive for Salmonella Typhimurium,

-         Cultures from the leftover chocolate sorbet and strawberry sorbet yielded the identical strain of Salmonella to the persons tested at the banquet.

Based upon this new information, the investigation team revised their case definition to read:

-         An illness with diarrhea (multiple loose bowel movements within a 24-hour period) and at least one of the following symptoms: headache, fever, abdominal pain, nausea, and/or vomiting 

OR

-         A positive stool culture for Salmonella Typhimurium

AND

-         Onset of illness or symptoms occurred during August 28--31 and the patient attended the PAC banquet.

In addition, local sanitarians also conducted an environmental analysis of the kitchen facility and found no indication of Salmonella Typhimurium on any of the equipment or utensils that were used in the storage, preparation, and service of the meal.  In addition, they also found that the kitchen staff practiced good hygiene and safe food-handling practices.

Understanding the Outbreak

Many outbreak investigations are not straightforward, and this one is no exception.  Another aspect of evaluating a hypothesis is to review reference material on the disease or illness to determine whether the situation is similar to other reported incidents.  Often, the Emerging Infectious Diseases (EID) journal is a valuable resource for finding incidents, trends, and specific disease information. (EID is available at: //www.cdc.gov/ncidod/EID/.

Some of the areas that will help with the investigation are:

-         Typical signs and symptoms

-         Modes of transmission

-         Exposures

In the example of the PAC Banquet, the majority of ill persons interviewed met the established symptoms of Salmonella Typhimurium, such as:

-         Diarrhea, fever, and abdominal cramps

-         Incubation period of 6 -- 72 hours after infection

-         Illness usually lasts 4 -- 7 days

-         Transmitted through contaminated food, water, or contact with infected animals

However, the fact that Salmonella Typhimurium is not typically found in the ingredients used in chocolate sorbet and strawberry sorbet led the investigation team to consider the possibility of deliberate tampering.  

When the investigators first suspected intentional food tampering, they contacted local law enforcement officials, who in turn notified the Federal Bureau of Investigation (FBI).  

Information obtained in the ensuing investigation revealed that a supporter of one of the candidates owned the company that supplied desserts to many of the food service facilities in the area.  This person later admitted to trying to sabotage the election by inoculating Salmonella Typhimurium into the desserts in an attempt to alter the results of the popular vote for funding.  The saboteur and colleagues skipped the banquet and had planned to show up for the vote while others were ill. 

If you recall, there were two persons reported that they did not eat the dessert.  As it turns out, when asked again, they indicated that they forgot to mention taking a "few bites" from their spouses' desserts.

Also, four hotel employees who were not employed in the kitchen, were identified as having Salmonella Typhimurium.  These employees became ill after they ate desserts that had been left over from the banquet.

In the end, all of the cases were accounted for and associated with either chocolate sorbet or strawberry sorbet.

Summary

Many outbreak investigations are not straightforward, and therefore it is important to look for clues that will help you identify what factors are associated with the illness in question.  To accomplish this, you should select an appropriate reference group and analyze the data for measures of association.  The following is a brief description of the steps involved in this process.

Selecting a Reference Group
Selection of an appropriate reference or comparison group is critical to determining whether an association exists between the exposure and illness.  Typically, the reference group is the population of individuals that has had no exposure to the risk factor being analyzed.  Ideally, the exposed and unexposed populations should be similar, with the exception of the exposure.  Thus, if no direct association exists between the exposure and the illness, the attack rates will be relatively similar.

Calculating Rates and Ratios
To determine what factors are associated with the illness, we must analyze the data to measure the frequencies with which a potential exposure can be associated with the illness.  There are several ways to express measures of association in cohort studies.

-         Attack rates are calculated by dividing the number of new cases among the population during the limited time period (X) by the population at risk at the beginning of the period (Y).  This number is then multiplied by (100) to express in a percentage.  The formula can be expressed as:

-         Risk ratios compare the amount of risk associated with an event such as a disease or death in the exposed population to the amount of risk in the unexposed population.  As a result, they are calculated by dividing the attack rate in the exposed population by the attack rate in the unexposed population.  The formula can be expressed as:

Analyzing Multiple Risk Factors
When several risk factors are being considered simultaneously, it is often necessary to establish an experimental control, where there is an absence of exposure to all of the risk factors being analyzed.  By establishing an experimental control, a uniform denominator is created and the resulting risk ratios can be compared.  The following can assist with comparison of risk ratios.

-         RR = 1.0 indicates identical risk in both groups

-         RR > 1.0 indicates an increased risk for the exposed group compared to the unexposed group

-         RR < 1.0 indicates a decreased risk for the exposed group compared to the unexposed group

Identifying the correct association will the illness does not always occur on the first attempt and knowing when to refocus and reanalyze often requires experience and judgment and might require advanced analysis.  Often, investigations are an iterative process in which a hypothesis is established, tested, and revised several times before association(s) can be established.  Overall, it is also essential to stay open and be alert to clues

Certisafety Section Home Page

Copyright ©2000-2019 Geigle Safety Group, Inc. All rights reserved. Federal copyright prohibits unauthorized reproduction by any means without permission. Disclaimer: This material is for training purposes only to inform the reader of occupational safety and health best practices and general compliance requirement and is not a substitute for provisions of the OSH Act of 1970 or any governmental regulatory agency. CertiSafety is a division of Geigle Safety Group, Inc., and is not connected or affiliated with the U.S. Department of Labor (DOL), or the Occupational Safety and Health Administration (OSHA).