A Sample Glossary

1. Prevention: Primary, secondary, tertiary
According to Gordis, primary prevention consists of a measure taken to prevent disease from occurring, such as immunizing schoolchildren or condom use as protection against an STI. This category of prevention also includes reduced exposure to risk factors in the environment, such as second-hand smoke or asbestos. Many diseases are preventable, but given that many diseases are closely entwined with attitudes and behaviors, change, and thus primary prevention, can prove challenging. An example from my field of interest would be: the prevention of chronic disease through good nutrition and physical activity.
Secondary prevention consists of screening and intervention, as in the case of cancer and hypertension. By identifying a disease early in its course, it may be possible to mitigate its effects. An example from my field of interest would be: screenings to identify clinical indicators, such as high blood pressure, high cholesterol, etc.
Tertiary prevention basically consists of treatment, harm reduction, and disease management to prolong life. An example from my field of interest would be: medication, increased physical activity and change in diet to manage Type II diabetes or cardiovascular disease.

2. Population vs. individual focus
A population focus enables researchers and practitioners to identify trends in disease incidence and distribution. It provides a context against which to compare individual experience of disease. According to Gordis, a population focus enables both diagnosis and prognosis of disease, based on patterns observed over time, in the aggregate. The collection and compilation of population level data enables prevention to be enacted at the primary, secondary and tertiary levels. A focus on individual health enables researchers and practitioners to identify the presence or absence of risk factors. Such an approach also enables one to observe disparities in health status and outcomes within and among groups. An example from my field of interest would be: rates of Type II diabetes over time in the city of Boston versus the progression of diabetes in an individual as indicated by an A1c blood test.

3. Multi-step process in epidemiological reasoning
a. Does association exist between risk factor/social determinant/clinical indicator and disease?
b. If association – causal relationship? Use of descriptive data to observe and determine, combined with establishing a baseline to measure trends over time. Establish a numerator (base population) and count individuals with disease.
c. Conceptualize and enact a prevention action. Do not need full understanding of a causal agent to take preventive steps, e.g. tobacco causes lung and other cancers.
d. Prevention and treatment are not treated as exclusive spheres, but rather are undertaken in concert with one another.
An example from my field of interest would be: The relationship between food and obesity and related clinical indicators and chronic diseases. By having established a baseline, the U.S. has been able to track increases in overweight and obesity over time, as evidenced by the CalorieLab Report on the ‘Fattest States’. And public health professionals do not need a comprehensive understanding of how obesity leads to stroke; in part because able to establish causal relationship between diet and clinical indicators of stroke. Moreover, treatment of diseases can happen simultaneously with primary prevention efforts, screening, and management of clinical indicators – all of which are intended to prevent and manage chronic diseases.

4. Cross-tabulation
According to Gordis, cross-tabulation is a useful method in the case of multiple causal agents – for determining which among them is most likely to be the cause. An example from my field of interest would be: Is gender more salient than age in determining likelihood to smoke cigarettes? Or vice versa.

5. Incidence
The number of new cases in a specific time period. An example from my field of interest would be: the number of Boston residents diagnosed with diabetes for the first time [PT] in 2009.

The number of new cases in the population at a specific time, divided by the population number. An example from my field of interest would be: the number of people with diabetes in March 2009 among various ethnic, education and/or SES groups.

6. Risk
Absolute Risk
The incidence of disease in a population, a certain number. Comparing the risk of one action/exposure against another. Difference in risk is measured by difference in incidence rates. An example from my field of interest would be: incidence of diabetes and cardiovascular disease among smokers.

Relative Risk
The ratio of risk of disease in exposed individuals to the risk in non-exposed individuals. An example from my field of interest would be: looking at the incidence of diabetes and cardiovascular disease among smokers as compared to non-smokers.

7. Target population
Population of interest, an example from my field of interest would be: Latino adolescents in Boston.
Study population
Population of interest defined more narrowly, an example from my field of interest would be: Latino adolescents (ages 15-19) living in public housing in East Boston.
Study sample
Limited number, representative sample of study population. Need sufficient size and responses. Also, subject to bias and error, so need to control for these.

8. Rates
Frequencies of disease (morbidity) and death (mortality). An example from my field of interest would be: morbidity rates of diabetes and associated health complications, as well as mortality rates, i.e. death from diabetes complications.
Measures of morbidity include incidence and prevalence. An example from my field of interest would be: Morbidity Incidence = the overall number of people in Roxbury, MA who developed cardiovascular disease this year. Morbidity prevalence = the number of Latina women in Roxbury who developed CVD this year.
Measures of mortality include case fatality rates, age-adjusted general death rates from all cases, and years of life lost. An example from my field of interest would be: quantifying the impact of CVD related deaths, not just in terms of hard numbers, but measuring those who died prematurely and how pre-maturely the deaths occurred.

9. Sensitivity vs. specificity
Sensitivity refers to the ability of a test to correctly identify those who have a disease and not miss them (i.e., provide false negatives) [PT] {those who do not, which helps to reduce the risk of a false positive or negative}. Specificity refers to the proportion of true negatives detected by the test (i.e., thus minimizing false positives) [PT] {, people who don the validity of a test regarding whether a person has a disease or not}. These terms are often applied to screening and diagnostic tests which constitute a form of secondary prevention. An example from my field of interest would be: community screenings for high blood pressure and cholesterol levels. [explain how sensitivity & specificity apply to such screenings, PT]

10. Positive predictive value
This a term also used with regard to screening and diagnostic tests. Seeks to determine what proportion of patients who test positive actually have the disease in question? Can be calculated by dividing the number of true positives by the total number who tested positive (true and false positives). An example from my field of interest would be: the number [proportion, PT] of individuals who test positive for a pre-cancerous condition, who upon further diagnostic work are confirmed to have said condition.

11. Outcome measures
These are designed to evaluate whether a patient/client benefits from the care/intervention received. It’s important not to confuse the outcome measure with the process measure. Criteria to be used in developing outcome measures includes: they should be clearly quantifiable and specific; relatively easy to define and identify, and easily standardized. Seems like stating the obvious to include in the criteria for outcome measures that the population being served must be at risk for the condition for which the intervention is being evaluated. An example from my field of interest would be: a process measure could consist of the number of patients who receive a brochure on the relationship between high blood pressure and diabetes. An outcome measure would be a decrease in the prevalence of diabetes diagnoses among Latina women in a given year.

12. Cohort
A study design in which a selected group of individuals followed over time re: incidence of disease among exposed and non-exposed groups. May be experimental or observed. Employed to guard against potential bias. An example from my field of interest would be: following Boston youth from childhood into adulthood to determine how early nutrition impacts weight, blood pressure, cholesterol and associated disease.

This study design consists of cases, those with the disease, and controls, those without but otherwsie... [PT]. Researchers look at the proportion in each group, the exposed versus the non-exposed, to establish causation retrospectively. An example from my field of interest would be: looking at obese Boston residents in college (in Boston) who grew up in the city neighborhoods to determine variation according to environment and individual behavior.

This type of study design is also known as a prevalence study. It begins with defining a population, getting data on exposure and disease, and then looking at the cases of exposure and disease existing at the time of the study, although timing and duration of the former remain unknown in such studies. An example from my field of interest would be: selecting a group of Asian senior citizens and measuring their blood pressure and cholesterol, as well as having them keep a food diary for August 2009. The basis for comparison would be restricted to this group, for this period of time, without looking at lifetime cumulative effects.

13. Associations:
Consists of looking at outcome measures in the context of the aforementioned study designs.

Odds ratio
A measure of association, which [is related to but different from]{ provides an estimate of} relative risk. [Risk is..., but odds are... , PT] It’s important as a measure of the strength of association, and establishes etiologic relationships. It’s used in place of relative risk in a cohort study (the odds of exposed versus non-exposed groups developing a disease) and case-control studies (the odds that cases were exposed versus the controls were exposed).The probability of an outcome. An example from my field of interest would be: the likelihood that that someone who eats well and engages in physical activity throughout their childhood would develop adult onset diabetes, as compared to another who did not.

Relative risk
This may be used in cohort studies, but cannot be calculated directly in case-control studies, as can only use odds ratio [because...]. Concepts of relative and attributable risk are central to causation and prevention. An example from my field of interest would be: does the relative risk of consuming too much soda impact enough adolescents, in terms of nutrition and health, to merit a tax or ban on sweetened beverages? Also, does the relative risk of CVD among young women merit an education, or screening program?

Attributable risk
Measures difference in risks, the proportion of disease incidence that can be attributed to a specific exposure. A measure of ‘how much of the risk of disease can be prevented if able to eliminate the exposure?’ This can be calculated for target population, i.e. target population, or total population. This has greater application in public health and clinical practice, than does relative risk. An example from my field of interest would be: if junk food is removed from school vending machines and cafeterias how much will student nutrition improve?

14. Bias
A result of error in the design and/or implementation of a study

Approach to organizing data by layers and categories to determine if bias, confounding effect by age, geography, or other characteristics and contexts. An example from my field of interest would be: organizing workers according to job characteristics, in order to determine if there is differentiation in health status and outcomes according to the strata.

An association appears to be causal, but is not. An association is instead correlated, or confounding, typically a third confounding factor that may be a risk factor and associated with the exposure of interest. An example from my field of interest would be: violence levels may be high in neighborhoods where residents do not engage in sufficient levels of physical activity, but violence may not be the cause for sedentary behavior [because there is another variable involved, namely,...].

Involves looking at all strata to see if there is variation in the intensity of an effect [from one stratum to the next]. When strata are unequal, if even one or more strata are amplified (to greater effect), or minimized (to lesser effect), there is an indication of an interaction. An example from my field of interest would be: variation in geographic location of residence may lead to increased blood pressure and/or cholesterol levels. [a different blood pressure to CVD relation]