Lectures on Cross Sectional, Case Control, and Cohort Studies

Question 1

What is the Standard Mortality Rate?

Answer 1

The standardized mortality rate (SMR) is the ratio of the number of deaths observed in a population over a given period to the number that would be expected over the same period if the study population had the same age-specific rates as the standard population. If the rate is greater than one, it is interpreted as excess mortality in the study population.

Question 2

What are the 3 important calculations involved in Cross Sectional Studies?

Answer 2

Prevalence Difference
Prevalence Ratio
Prevalence Odds Ratio

Question 3

How do we calculate Prevalence Difference and what does it show?

Answer 3

P1 (a or exposed with outcome / m1 or all exposed) - P2 (all unexposed with outcome / m2 or all unexposed)

Shows prevalence in exposed versus unexposed

Question 4

How do we calculate prevalence ratio and what does it show?

Answer 4

P1 (exposed) / P2 (unexposed)
*Divide prevalence in test group / exposed by prevalence in control group / unexposed

Shows prevalence in exposed versus unexposed

Question 5

How do we calculate prevalence odds ratio and what does it show?

Answer 5

This is the same as a regular odds ratio, ad/bc
(cross product ratio)

Shows prevalence in exposed versus unexposed

Question 6

What are estimators and estimates / point estimates?

Answer 6

An estimator is a function of a sample (a rule that tells you how to calculate an estimate of a parameter from a sample)

An estimate is a value of an estimator calculated from a sample

e.g. Average height of all school children (population is too big to measure accurately). You take a sample of 1000 and use your estimator (sample mean) to calculate the average (56 inches).

Question 7

What is a confidence interval?

Answer 7

A confidence interval defines an upper limit and a lower limit with an associated probability.

The confidence interval based on a single sample indicates that there is a certain probability (percentage) that the derived confidence interval from a future assessment of the data will contain the value of the population parameter.

Question 8

When comparing two confidence intervals, what does overlap mean?

Answer 8

Overlap means no statistical significance

No overlap means statistical significance

Question 9

Who was Janet Lane Claypon (1877-1967)?

Answer 9

was an English physician. She was one of the founders of the science of epidemiology, pioneering the use of cohort studies and case-control studies.[3]

In 1912, Lane-Claypon published a ground-breaking study of two cohorts (groups) of babies, fed cow’s milk and breast milk respectively. Lane-Claypon found that those babies fed breast milk gained more weight, and she used statistical methods to show that the difference was unlikely to occur by fluke alone. She also investigated whether something other than the type of milk could account for the difference, an effect known as confounding.[7]

Lane-Claypon tracked down 500 women with a history of breast cancer – the “cases” – and compared them with 500 women who were free of the disease but otherwise broadly similar, known as “controls”.[8] She showed that breast cancer risk increased for childless women, women who married later than average, and women who did not breast feed. The overall breast cancer risk decreased according to the number of children. For all cases, rapid treatment held the key to survival among women with breast cancer.[9] This study eventually led to the incorporation of risk tables and life expectancy in cancer treatment.[8]

Question 10

What are some traits of effective case control studies?

Answer 10

They have good eligibility criteria so confounding variables are reduced

They use more incident cases than prevalent cases to reduce recall bias and prevent overrepresentation of older disease variations (minimize information bias)

Preferable using only incident cases over a defined period of time

They are the only technique recommended for rare diseases

They use pairs of case matching (e.g. If I have a 26 year old female black case, then I want a 26 year old female black control)

Cases and controls should come from the same population, and controls should be representative of said population

They include more than 1 but less than 4 control groups (to provide power while remaining cost effective)

Question 11

What is a common issue with assessing exposures (exposure estimates) in case control studies?

Answer 11

It has to be assumed that the exposure incurred at the time the disease process began (this may not be valid)

Exposure estimates are subject to recall and interviewer bias (this can be reduced with blind / double blind techniques, careful wording, and reduction of confounding variables)

Question 12

How do we interpret Odds Ratio numbers?

Answer 12

OR>1 means exposure increases disease risk (risk factor)

OR<1 means exposure decreases risk (protective factor)

OR = 1 means the exposure has no association with increasing / decreasing disease risk (not a risk factor or protective factor)

Question 13

What are the general guidelines for interpreting Odds Ratio numbers?

Answer 13

Value Effect of exposure

0.00-0.39 Strong benefit
0.40- 0.59 Moderate benefit
0.60-0.89 Weak benefit
0.90-1.19 No effect
1.20-1.69 Weak hazard
1.70-2.50 Moderate hazard
>2.50 Strong hazard

Question 14

What is recall / response bias?

Answer 14

When participants in a study are systematically more or less likely to recall and relate information on exposure depending on their outcome status, or to recall information regarding their outcome dependent on their exposure (usually more likely to “remember” due to their searching for reason).

Question 15

What is temporal bias?

Answer 15

Temporal bias acts to undermine the validity of predictions by over-emphasizing features close to the outcome of interest.

Sometimes researchers don’t know if a factor caused a disease or a disease caused a factor

Question 16

What are advantages / disadvantages of case control studies?

Answer 16

Advantages of case control studies

1. Excellent way to study rare disease and diseases with long latency
2. Relatively quick
3. Relatively inexpensive
4. Requires relatively few subjects
5. Can often use existing records
   Can study many possible causes of a disease

Disadvantages of case control studies

1. Relies on recall or existing records about past exposures
2. Difficult or impossible to validate data
3. Control of extraneous factors incomplete
4. Difficult to select suitable comparison group
5. Cannot calculate incidence (and consequently risk, which is preferred to odds)
6. Cannot study mechanism of disease

Question 17

What is the difference between a case report / series and a study design?

Answer 17

Case report / series have no sample or control group and no measure of association

All study designs have measures of association

Question 18

What are some other names for Cohort Studies?

Answer 18

Incidence study
Longitudinal study
Forward Looking Study
Follow Up Study
Concurrent Study
Prospective Study

Question 19

Describe the temporality of the 3 main study types

Answer 19

Question 20

Describe confounding variables versus mediators

Answer 20

Question 21

Describe basic analysis in a cohort study

Answer 21

Question 22

What are the 3 main types of outcomes in a Cohort study

Answer 22

Death
Disease (diagnosis)
Recovery
(and sometimes loss of follow up)

Question 23

What are disease markers?

Answer 23

Diagnostic markers are biological parameters that aid the diagnosis of diseases. Examples include the presence of particular genetic variants and the levels of particular proteins in body fluids.

These are NOT the same as causative agents

Question 24

What is the general set up of a cohort study?

Answer 24

Start with an at-risk population
Determine eligibility and exposure criteria (presence of absence of risk)
Follow up and track incidence events

No true randomization (those at risk are chosen by nature)

Question 25

What is a historical cohort study?

Answer 25

In a retrospective (historical) cohort, investigators use preexisting data to identify exposed and unexposed individuals in the past, without regard to outcome status, and trace these individuals forward, up to and possibly including the present, to determine incident outcomes

(usually occupational studies about previous exposures)

Question 26

Describe the temporal schematic of historical, concurrent, and ambi-directional cohort studies:

Answer 26

Question 27

What is the formula for Relative Risk?

Answer 27

Incidence rate in exposed / incidence rate of non-exposed

RR = a / ni or a / (a + b)
c / no c / (c + d)

Question 28

What is the measure of association used for Cohort Studies?

Answer 28

Relative Risk

Question 29

How is Attributable Risk Calculated?

Answer 29

Attributable risk (AR) is the risk difference between exposed and unexposed groups. In equation form (Table 1) it is Risk(exposed)−Risk(unexposed)=[A/(A+B)]−[C/(C+D)].

Question 30

What is the difference between Relative Risk and Attributable Risk?

Answer 30

Question 31

What is selection bias?

Answer 31

Selection bias is a distortion in a measure of association (such as a risk ratio) due to a sample selection that does not accurately reflect the target population.

Question 32

What is a prevalent form of selection bias in Cohort Studies?

Answer 32

Loss of follow up

Question 33

What is confounding bias?

Answer 33

A systematic distortion in the measure of association between exposure and the health outcome caused by mixing the effect of the exposure of primary interest with extraneous risk factors.

Question 34

What are some advantages of cohort studies?

Answer 34

Can measure disease incidence
Can study the natural history
Provides strong evidence of casual association between E and D (time order is known)
Provides information on time lag between E and D
Multiple diseases can be examined
Good choice if exposure is rare (assemble special exposure cohort)
Generally less susceptible to bias vs. CCS

Question 35

What are some disadvantages of cohort studies?

Answer 35

Large samples needed
It can be difficult to find appropriate controls, or sometimes controls at all
Drop out due to long follow up is a threat to validity
Changes over time in exposure / diagnostic criteria confounds data
Expensive and time consuming (prospective)
Requires substantial data (retrospective)
Inefficient for rare diseases

Question 36

What are examples of well known cohort studies?

Answer 36

Framingham Study
Nurses’ Health Study
Womens’ Health Initiative
Physicians’ Health Study

Question 37

What is cause specific mortality rate versus proportionate mortality versus case fatality rate

Answer 37

CAUSE-SPECIFIC DEATH RATE is the number of deaths from a specified cause per 100,000 person-years at risk.

Proportionate mortality (PM) was computed as follows: PM = (number of deaths due to a specific cause/total number of deaths) × 100;

Case Fatality Rate: is the proportion of people diagnosed with a certain disease, who end up dying of it. Unlike a disease’s mortality rate, the CFR does not take into account the time period between disease onset and death.

Question 38

Describe the 4 major outbreak curves

Answer 38

Question 39

Describe the features of a propagated outbreak curve

Answer 39

Question 40

Describe the natural history of disease timeline

Answer 40

Question 41

Z scores for common confidence intervals

Answer 41