EPID 1 - Measuring Disease Occurrence, Cross-SEctional, and Ecologic Studies

Card Set Information

EPID 1 - Measuring Disease Occurrence, Cross-SEctional, and Ecologic Studies
2013-10-16 09:02:52

Show Answers:

  1. 1. What do you need to make sure an epidemic is actually an epidemic? (2)

    2. To measure disease occurrence, what must you take into account? Define

    3. What must you consider in epidemiological studies (in measuring disease occurrence)?

    4. What are the 3 main types of data sources? When can you use each in the natural history of disease?

    5. What influences the rate of disease we measure?
    1. SURVEILLANCE DATA - need endemic data to determine if there's an epidemic & potential source

    2. Natural history of disease; development & course of disease in an individual over a period of time

    3. Sources of data we have and how to capture people at each level of wellness and sickness

    4. Interviews, physician records, hospital records

    5. Sources of data.
  2. 1. Rates vs. proportions
    2. What 3 elements does a rate contain?
    3. Why is it important to capture rates?
    1. Rates tell us how fast a disease is occurring in a population; proportions tell us what fraction of population is affected

    2. (1) Disease frequency (2) Unit size of population and (3) time period during which an event occurs.

    3. We don't want to only capture mortality - we want to catch people earlier in the disease process!
  3. 1. Define incidence rate

    2. What is it a measure of? For what? Over what?

    3. Why is it a measure of the above?

    4. What is it typically multiplied by?

    5. Why does each individual's person-years vary?

    6. What is the total # of person-years here?
    1. Def - number of new cases that occur during a specified time period in a population at risk at that time.

    2. It's a measure of risk for a defined population over the time that the population was followed.

    3. Because it is a measure of events (transition from nondiseased to diseased state)

    4. Per 1,000 (but can be 10,000, etc)

    5. Because some individuals may drop out

    6. 25
  4. 1. Define prevalence

    2. Is duration of disease taken into account?

    3. Is prevalence a measure of risk?

    4. What are the two types of prevalence? Define each and give equations
    1. Prevalence - the proportion of affected individuals present in a population at a specific time.

    2. No

    3. No


    Point prevalence - prevalence of disease at a specific point in time (# of persons ill/population at-risk)

    Period prevalence - defines how many people have had the disease at ANY time during a certain period (# of persons ill/average population during a time period!!!)
  5. Below are a list of questions. What are they asking about?

    1. Do you currently have asthma?

    2. Have you ever had asthma?

    3. Have you had asthma during the last 3 years?

    - Point prevalence in May?
    - Incident cases in 2008?
    - Point prevalence in Sept?
    - Period prevalence for 2008?
    1. Point prevalence

    2. Cumulative/lifetime

    3. Period prevalence

    4. 5; 2; 3; 5
  6. 1. How does incidence relate to prevalence?

    2. What sorts of new measures will lead to increase in prevalence? Example?
    1. Prevalence = Incidence x Duration of Disease

    2. If a new measure/treatment treats disease that previously killed patients, but doesn't cure it (ARTs and HIV; Diabetes + Insulin)
  7. (1) How is prevalence used in public health? (3)

    (2) What do we need to determine if a risk factor leads to a disease?
    (1) 1. Measures burden of disease in a community

    2. Helps determine resources needed

    3. Helps in planning of health services

    (2) Incidence rates!!!
  8. What are problems with incidence/prevalence measures? 2;3

    • (1) how to define who has the disease -
    • different criteria can lead to different numbers of cases (i.e.,
    • rheumatoid arthritis or dementia).

    • (2) Source of data (if someone has
    • disease but no symptoms, will not count as case!; hospital records may
    • not be complete; not all cases get admitted to hospital)

    • (1) Undercounting of population
    • (2) Misclassification of ethnic groups
    • (3) Need to ensure that denominator is actually at risk for the disease (i.e., uterine cancer - only females in denominator, but what about women with hysterectomy?)
  9. When looking at graphs of disease occurrence, what should you pay attention to?
    Whether graph is showing proportion or INCIDENCE (i.e., rate). It can be misleading - sometimes, proportion may increase but incidence stays the same or vice versa.
  10. 1. What are the 4 ways to measure death?

    2. What are problems with mortality data? (1)

    3. What are limitations of mortality data? (3)
    1. Mortality rate (total deaths/mid-year population)X100,000

    2. Infant mortality rate (total deaths/total liveborns that year)

    3. Case fatality rate (actually a proportion) = # of deaths over specific time after disease onset/diagnosis/# of individuals with disease.

    4. Proportionate mortality: # of deaths from disease over specific time/total deaths over specific time

    2. Generally based on certificate of death - subject to coding bias that varies by year, by country, etc

    3. (1) Unclear cause of death in multiple chronic illnesses, (2) lack of standardization with diagnostic criteria, and (3) stigma associated with certain diseases may lead to inaccurate reporting.
  11. What are possible explanations of trends in mortality that are artifactual? (Produced by Humans) (Num - 4, Denom - 3)
    • Numerator:
    • (1) Change in diagnosis
    • (2) Errors in age
    • (3) Changes in coding rules
    • (4) Changes in classification


    • (1) Errors in counting populations
    • (2) Errors in demographic classifications
    • (3) Differences in the population at-risk
  12. What are possible explanations of trends in mortality that are real? (4)
    (1) Changes in survivorship without changes in incidence

    (2) Change in incidence

    (3) Change in age composition of the populations

    (4) Combination of the above factors
  13. 1. What should you use if a disease impacts someone but they're not dying from it? What is the #1 cause of this?

    2. Define infant mortality rate

    1. YPLL (Years of Potential Life Lost); Unintentional injury

    2. # of deaths <1/# of live born children during the year

    1. Mortality rate = deaths/mid-year population = 269/(32500/2)

    2. 269/32800!!!! Keep in mind that MORTALITY RATE uses mid-year population, while INFANT MORTALITY RATE uses all liveborn children!
  14. 1. What is a frequent use of epid?

    2. What is a frequent sequence of studies in human populations? (5)

    3. How are epid studies categorized? (2)
    1. To determine if a factor leads to increase/decrease of disease

    2. Clinical observations --> available data --> case-control studies --> cohort studies --> RCTs

    3. Based on design (i.e., cohort/longitudinal/prospective vs. case control vs. cross-sectional) or unit of analysis (individual vs. ecologic)
  15. 1. What is the unit of analysis/observation for ecologic studies? Examples? (3)

    1.5. What is measured for each group? (2) Then what is obtained?

    2. What are the 2 types of ecologic studies?

    3. What does statistical analysis of ecologic studies entail? What's an important concept? What's considered "good" of this concept?

    4. Limitations of ecological studies? (5)

    5. Define the last one. What are the implications of this?

    6. Strengths/advantages of ecologic studies? (4)
    1. Groups (countries, hospitals, states, neighborhoods, regions etc)

    1.5. Exposure & outcome for each group. THEN correlations are obtained bt exposure rates and disease rates among different groups or populations.

    2. Ecologic comparison study - examines exposure rates and disease rates among different groups over same time period

    Ecologic trend study - examines changes in exposure/changes in disease within same group

    3. Obtaining a correlation describing linear relationship bt 2 variables - extent of relationship is quantified by correlation coefficient (R value). Can be between -1 and +1 but a good correlation is between 0.2-0.3 in studies.

    4. Limitations:

    (1) No assessment of disease risk (2) no time order (3) no individual risks (4) quality control of data (5) ecologic fallacy

    5. Ecologic fallacy - making observations at a group level may not represent exposure-disease relationship at the individual level

    Conclusions obtained from ecologic study may be the reverse of those from a study collecting data on individual subjects.

    6. (1) Low cost/convenient (2) can use previously collected data (3) can generate hypotheses for analytical epid studies/get individual data later on (4) Some exposures may vary very little or not at all within a given group
  16. 1. What type of study is a cross-sectional study?
    2. Individual or ecological level?
    3. How many periods of observation?
    4. When are exposure & disease histories collected?
    5. What type of sampling is used ? (2)
    1. Prevalence

    2. Individual

    3. 1

    4. Simultaneously

    5. Both probability and non-probability sampling used.
  17. 1. What are CS studies used for? (3)

    2. Draw what it would look like.

    3. Draw how analysis would go down (2)

    • 1. Hypothesis generation for future studies
    • 2. Good for future planning (snapshot of issues plaguing community)
    • 3. Estimation of magnitude/distribution of health problem

  18. CS STUDY

    1. How to pick study population? (3)
    2. How to measure exposure/disease? (4)
    3. When is CS study best?
    4. What must be controlled for? When? How? What must have been done?
    1. (1) sample of people/households in defined geographical area (2) may be defined by study question as in occupational/regional exposure or (3) random sampling of an area

    2. Questionnaires, records, lab tests, physical measurements

    3. Best suited for factors that do not change as a result of a disease (i.e., blood group)

    4. Confounders during statistical analysis by statistical analysis. Must have collected data on potential confounders.
  19. CS Study:

    1. What are limitations? (5)

    2. What are advantages? (5)

    • 1. No incidence data/disease risk
    • 2. Cannot establish temporality/causality bt exposure and disease
    • 3. Can only review prevalent cases (not ones that are quickly cured or those who quickly died)
    • 4. Cannot study low prevalence diseases
    • 5. Exposure status at the time a person is included in study may not reflect exposure status at beginning of disease process


    • 1. Quick/inexpensive
    • 2. Hypothesis generating
    • 3. Based on general population, not just those seeking medical care
    • 4. Good for planning
    • 5. Good for multiple diseases