Chapters 7, 8, and 9 (Experimental, Cohort, and Case Control Studies)

Question 1

What are the different types of cohort study populations

Answer 1

1. Open or dynamic, defined by a changeable characteristic. Members may come and go. Measured by incidence.
2. Fixed: caused by an irrevocable event. Does not gain members but may lose them. Measured by incidence.
3. Closed: caused by an irrevocable event. Does not gain or lose members. Measured by cumulative incidence.

Question 2

Where did the term “cohort” come from?

Answer 2

The latin word “Cohors,” meaning a group of soldiers

Question 3

What are the three types of cohort study temporality

Answer 3

Prospective
Retrospective
Ambi-directional

Question 4

What is the process of using “placebos” in a cohort study?

Answer 4

the use of special matching criteria when selecting unexposed subjects
(e.g. using individual matching, for each 50 year old black female case, they choose a 50 year old black female control)

Question 5

What are confounding variables?

Answer 5

Often present in cohort studies

A confounding variable, also called a confounder or confounding factor, is a third variable in a study examining a potential cause-and-effect relationship. A confounding variable is related to both the supposed cause and the supposed effect of the study (e.g. cigarette smoking often accompanied by alcohol use - which is the cause of the outcome?)

Question 6

How are sample sizes for cohort studies chosen?

Answer 6

To maximize the power of a study, and NOT to mimic the frequency in a general population (e.g. a rare disease might occur in 5% of a population but is 50% of a cohort study)

Question 7

What are the two main types of cohorts?

Answer 7

Special cohorts: assembled to study the health effects of rare exposures such as uncommon occupational chemicals or rare diets (e.g. 7th Day Adventists do not eat pork and most follow a lacto-ovo diet)

General cohorts: assembled for common exposures such as use of oral contraceptives, dietary factors, or risky behaviors such as alcohol use

Question 8

What is the counterfactual ideal?

Answer 8

The ideal comparison group in a cohort study would be a group that was exactly the same as the exposed group, except that they would be unexposed. This is referred to as the “counterfactual ideal,” because it is impossible for the same person to be both exposed and unexposed at the same time.

Question 9

What are the 3 types of comparison groups in cohort studies?

Answer 9

Internal - most effective, individual matching, but may be hard to find
General Population - easy to find, but may have lack of comparability
Comparison Cohort - fairly comparable, but often have other exposures that may confound results - some compare cohorts to past studies instead of another current sample

Question 10

What is the “healthy worker effect”?

Answer 10

a special type of selection bias, typically seen in observational studies of occupational exposures with improper choice of comparison group (usually general population)
a relatively healthy working population is being compared with the general public, including ill and healthy people

Question 11

What sources of information are used for different types of cohort studies?

Answer 11

Study of medical exposures / outcomes: health records
Study of occupational exposures: employment records
Study of lifestyle: interviews and questionnaires

Existing records are limited, so most only use data collected specifically for study: questionnaires, physical exams, lab tests, biological specimens, environmental monitoring, etc.

Question 12

What are examples of biological specimen frequently studied for exposures?

Answer 12

blood, urine, bone, and toenail samples

(e.g. levels of lead in blood, selenium / mercury levels found in toenail clippings)

Question 13

What are useful tools for the collection of outcome information in Cohort studies?

Answer 13

state and national disease registries
department of vital records
& the same records used for exposures (medical & lab, study specific records)

Question 14

How does a biological marker relate to an outcome / exposure?

Answer 14

It is considered an intermediate outcome because it is on the pathway from exposure to the appearance of clinical disease.

(e.g. CD4 lymphocytes decline can indicate HIV)

Question 15

How do cohort studies maintain follow up?

Answer 15

For baseline studies, usually collecting detailed baseline information including full name, address, phone number, SSN, date and place of birth, and contact info for family, friends, and physicians

Mail, Additional Mail, Address Correction Mail, and National Change of Address System

Question 16

What is “person time” in a cohort study?

Answer 16

Follow up time (ends when individual dies or study ends - or often when diagnosis occurs)

NOT years of exposure or induction or latency (empirical latent period)

Question 17

What does cohort study analysis typically measure?

Answer 17

The length of time between the causal action of an exposure and the eventual diagnosis of disease

Question 18

What are two special types of cohort studies?

Answer 18

Proportional Mortality Ratio (PMR)

Standardized Mortality Ratio (SMR)

These are used to compare the mortality experience of the exposed group with that of the general population

Question 19

When should cohort studies be used?

Answer 19

when researchers want to learn about multiple effects of an exposure or when an exposure is rare

Retrospective cohort should be used when the outcome of interest has a long induction and latent period

Question 20

What are weaknesses of cohort studies?

Answer 20

They are inefficient for studying rare outcomes
Prospective cohort studies are inefficient for studying diseases with long induction and latent periods
Retrospective cohort studies are vulnerable to bias

Question 21

What are different types of experimental studies?

Answer 21

Individual
Community
Preventative (prophylactic agent is given to healthy or at risk individuals to prevent outcome) - primary prevention

Therapeutic (given to diseased individuals) - secondary prevention

Parallel (each group given one treatment concurrently - Group A gets Treatment B while Group B gets Treatment A; then they switch after washout period )

Crossover (each group gets treatment one after the other)
Simple (one treatment per group)
Factorial (2+ treatments per group)

Question 22

What is the typical protocol for new drug trials?

Answer 22

Phase 1 drug trial on small number of healthy individuals

Phase 2 drug trial on larger number of diseased individuals

Phase 3 drug trial (usually 3 years) showing advantage of experimental therapy over standard one (e.g. improved survival), usage, dosing, side effects (most drugs get approved here)

Phase 4 drug trial to examine rare, slowly developing adverse events

Main exception: drugs being developed for serious diseases with few treatment options

Question 23

What is statistical power?

Answer 23

the ability of a study to demonstrate an association if one exists

Question 24

What is blocking in experimental design?

Answer 24

In the statistical theory of the design of experiments, blocking is the arranging of experimental units that are similar to one another in groups (blocks). Blocking can be used to tackle the problem of pseudoreplication.

Question 25

What is stratification in experimental design?

Answer 25

Stratification means that the person or computer randomizing patients to each group tries to assign roughly equal numbers of patients with similar health or tumor characteristics -confounding factors- to each type of treatment. The factors to be stratified for are identified from the outset, as part of the study rules.

Question 26

What is another term for “blind” or “double blind?”

Answer 26

Masked or double masked

Question 27

What are single, double, and triple masking techniques?

Answer 27

Single: participants are masked

Double: participants and investigators giving treatment are masked

Triple: participants, treatment givers, and investigators monitoring are masked

Question 28

How do we measure outcomes in experimental trials?

Answer 28

Primary outcome (the primary one being studied)
Secondary outcomes (other end point factors)

Surrogate outcome measure (measuring a different factor to determine the primary)

Question 29

What population is described in applications of “generalizability”?

Answer 29

The reference population

Question 30

What are situations in which a case control study is desirable?

Answer 30

1. exposure data are difficult or expensive to obtain
2. the disease is rare
3. the disease has a long induction and latent period
4. little is known about the disease
5. the underlying population is dynamic

Question 31

Is a case control the opposite of a cohort study?

Answer 31

No, though they measure the same concept in the end (comparison between exposed and unexposed groups)

Question 32

Why do epidemiologists prefer incidence over prevalence?

Answer 32

Usually more interested in factors that lead to the development of the disease than the factors that affect the duration of the disease

Question 33

When do epidemiologists have no choice but to use prevalence over incidence?

Answer 33

Example: when studying birth defects since malformed fetuses are very difficult to study

Temporality is confusing - are issues caused by the defect, surviving in utero, or both?

Question 34

What are important factors of case definition and selection?

Answer 34

1. case definition should clearly stratify diseased and nondiseased
2. Efficient and accurate sources should be used to find cases
3. Incidence is preferable to prevalence
4. Partial case ascertainment (using some and not all disease cases in a population) is fine as long as the source population can be defined

Question 35

What are important factors for selecting controls?

Answer 35

1. representative of the population that gave rise to cases
2. provide info on exposure distribution in source population
3. Use the question “If a member of the control group actually had this disease, would he end up as a study case?”
4. Use many sources to identify controls, including: general population, hospital and clinic rosters, death certificates (dead controls, generally not recommended), and family / friends
5. Remember that each source type has advantages / disadvantages

Question 36

What ratios are considered effective for case : control?

Answer 36

Up to 4:1 (beyond that it is not cost effective)

Question 37

What are the 3 main methods for sampling controls in a Case Control study?

Answer 37

1. Survivor sampling (using survivors of cohort study, most common)
2. Case Base or Cohort sampling (sampling population at risk)
3. Risk set sampling (longitudinally sampling controls during 10 year follow up - e.g. a skin cancer case is diagnosed, an at risk control from the cohort is added in)

Question 38

What are common sources of exposure information?

Answer 38

1. Questionnaires (must be carefully designed)
2. Preexisting Records (inexpensive but may lack detail or uniformity)
3. Biomarkers (not common due to lack of precision and validity, expensive)

Question 39

What is a case crossover study?

Answer 39

new variant of case control study used when brief exposure causes a transient change in risk of a rare acute onset disease

The period after exposure is called the Hazard Period, exposure frequency at this time is compared to control

Cases are their own controls

Question 40

What are effective applications of case control study design?

Answer 40

Evaluation of Vaccine Effectiveness

Evaluation of a Prevention Program

Investigation of an Outbreak of Disease

Question 41

What are strengths and weaknesses of case control studies?

Answer 41

Strengths:
good for rare diseases or diseases with long induction and/or latent periods
can evaluate multiple exposures for one disease (good for diseases that are not well known)

Weaknesses:
inefficient for rare exposures
vulnerable to bias
may have poor info on exposures
difficult to infer temporality between exposure and disease