Using a computer software program, the epidemiologists have analyzed the food history data from the questionnaires and have constructed the following attack rate table.

FOOD ITEMS

95% Confidence Interval

Ill

Attack Rate

30

Analytic epidemiology is defined as the study of the determinants of disease or reasons for relatively high or low frequency in specific groups.  Analytic epidemiology answers questions regarding why the rate is high or low in a particular group.  Observations of differences lead to formation of hypotheses.

Analytic Studies

There are basically two types of studies: experimental and observational.  In an experimental study, the exposure has not occurred yet.  The investigator controls the exposure in the study groups and studies the impact.  For example, he may immunize one group with an experimental vaccine that has been developed for a disease and compare the frequency with which the disease develops to the control group (which had no modification).  In an observational study, the exposure has already occurred.  The exposures and outcomes are observed and analyzed, not created experimentally.  Observational studies are often more practical and continue to provide the major contribution to our understanding of diseases.  There are two main types of observational studies: cohort (prospective) and case-control (retrospective) studies.

In a cohort study, a group of people who share a common experience within a defined time period (cohort) are categorized based upon their exposure status.  For example, individuals at a work place where an asbestos exposure occurred would be considered a cohort.  Another example would be individuals attending a wedding where a foodborne illness occurred.  Cohort studies have well-defined populations.   Often, cohort studies involve following a cohort over time in order to determine the rate at which a disease develops in relation to the exposure.  

In a cohort study, relative risk is used to determine whether an association exists between an exposure and a disease.  Relative risk is defined as ratio of the incidence rate among exposed individuals to the incidence rate among unexposed individuals. 

To calculate the relative risk, you would use the following formula: (a/a+b) / (c/c+d) where:

a = the number of individuals with a disease who were exposed. 

b = the number of individuals without a disease who were exposed.

c = the number of individuals with a disease who were NOT exposed. 

d = the number of individuals without a disease who were NOT exposed. 

In a case-control study, the sample is based upon illness status, rather than exposure status.  The researcher identifies a group of people who meet the case definition and a group of people who do not have the illness (controls).  The objective is to determine if the two groups differ in the rate of exposure to a specific factor or factors.   

In contrast to a cohort study, the total number of people exposed in a case-control study is unknown.  Therefore, relative risk cannot be used.  Instead, an odds ratio or risk ratio is used.  An odds ratio measures the odds that an exposed individual will develop a disease in comparison to an unexposed individual.  Please click the button below to learn how to calculate an odds ratio.

To calculate an odds ratio, you would use the following formula: ad/bc

where:

a = the number of individuals with a disease who were exposed. 

b = the number of individuals without a disease who were exposed.

c = the number of individuals with a disease who were NOT exposed. 

d = the number of individuals without a disease who were NOT exposed. 

Below is an example…

If a researcher selects 50 Lyme disease cases and 100 controls for a case-control study, and the results indicated that 45 cases and 10 controls recently hiked in a national forest, the odds ratio would be inserted into the 2×2 table below:  

The odds ratio would be calculated as follows:

Odds ratio = (45 x 90) / (10 x 5) = 81

Interpretation of Odds Ratios and Relative Risk

A relative risk or odds ratio that is approximately equal to 1.0 indicates that there is no association between the exposure and the outcome.  If the relative risk or odds ratio is significantly greater than 1.0, then the outcome and exposure are positively associated.  If the relative risk or odds ratio is significantly less than 1.0, then the outcome and exposure are negatively associated and the exposure is referred to as being protective.  For example, exercise may be negatively associated with lung cancer because individuals who smoke are less likely to exercise.

To determine if a value is statistically significant, confidence intervals are often calculated using computer software programs.  A 95% confidence interval is defined as a range of values that has a 95% probability of containing the value being estimated (e.g. odds ratio or relative risk).  For example, if the 95% confidence interval for the odds ratio of 81 in the above example is 23.5-302.9, then it tells you that there is a 95% probability that the odds ratio will be between 23.5 and 302.9. 

Confidence intervals that are above 1.0 and DO NOT include 1.0 are statistically significant and may indicate that a food item is contaminated.  For example, a confidence interval of 1.1 – 7.9 is significant because 1.1 (the left number of the confidence interval) is above 1.0.  The number 1.0 is NOT between 1.1 and 7.9.

Confidence intervals that include 1.0 are NOT significant and indicate that the food item is probably NOT contaminated.  Using pepsi as an example, the attack rate table indicates that the 95% confidence interval equals: .8 – 10.5.  Because 1.0 is between .8 and 10.5, it includes one and therefore is probably NOT contaminated.

To identify the contaminated food item you need to identify the food items that have significant confidence intervals and pick the food with the highest relative risk

To view an example of how to calculate a relative risk, click here

Often these values are put into the following 2×2 table:

The attack rate is a form of incidence in which the numerator is the number of new cases of a health problem during an outbreak, and the denominator is the population at the beginning of the period.  Food-specific attack rates are frequently used in foodborne outbreak investigations to compare those who ate a specific food with those who did not eat the food.  A high attack rate among persons who ate a specified food suggests that a food is associated with the illness.  A low attack rate among persons who ate the food suggests that the food is not associated with the illness.  The risk difference is the difference in attack rates (i.e. the percent ill among those who ate a specified food minus the percent ill among those who did not eat the food).  Usually, the risk difference is large for the contaminated food and small for other foods.  For example, the risk difference for cheese in the table below is 74 – 56 = 18.

A statistically significant association between an exposure and a disease does not necessarily mean that there is a cause-effect relationship between the exposure and illness.  The association could reflect biases in the design, conduct, or analysis of the study.  The association may also occur because the exposure and the disease are related to some common underlying condition.  Please click here to view the criteria that are widely used to evaluate whether an association is causal.

 http://www.epidemiolog.net/evolving/AnalyticStudyDesigns.pdf