Case-Control Study Design

Selection of cases and controls: Base population Who were the cases? Who were the controls? What was the exposure of interest? What was the exposure ? What was the outcome? What was the data source?

-Children under 17 who died from meningoccal disease -Children under 17 who survived from meningoccal disease -quality of care -Standardized evaluation of emergency medical care. -Death - Hospital and medical records

Strengths of case-control study: 1. Tend to use _______ sample sizes than cohort prospective studies 2. Relative to cohort studies, are they quicker and easier to complete? 3. are they cost effective? 4. Are they useful for studies of rare diseases? 5. Are they useful for diseases with long latency periods (e.g. Cancer)

-smaller -yes -yes -yes, when little is known about the disease -yes

Strengths Case-Control Studies(5)

1.Smaller sample size 2.Good for rare disease/outcome 3.Cheaper and faster to conduct than cohort/RCT 4.Good for disease with long latency (cancer) 5.Can study multiple exposures

TopHat Question. Which of the following best describes a cohort study design? A. Study population who is exposed/not exposed and followed over time for outcome. B. Study population who has outcome/does not have the outcome and followed over time for exposure. C.Study population who has outcome/does not have the outcome and are asked about history of exposure.

A. Study population who is exposed/not exposed and followed over time for outcome.

Berkson's bias can either UNDERESTIMATE or OVERESTIMATE the association between the exposure and outcome.

In this example, Berkson's bias UNDERESTIMATES the odds ratio in this example

we are not assigning treatment/control groups so is what?

Observational

after the population has been chosen and the groups have been made what happens next?

Obtain exposure history

(T/F) Must be selected from the same base or source population as cases

True

Are the two groups, cases and controls, Selected from the same Source population?

Yes

Case-control study is _______ (testing a hypothesis)

analytical

This is a methodological issue in some epi studies - are case-control studies prone to bias due to the design?

are especially prone to bias due to the design

Example of Berkson's bias Study looking at association of smoking and respiratory disease Use hospitalized cases and hospitalized controls Controls are more likely to be smokers than the general population - why?

because they are in the hospital for something that could be related to smoking

what does Case-Control Study design fall on the Causality Hierarchy?

below cohort and above cross-sectional

Two groups are compared, which are they?

cases and controls

A case-control study Identifies association between what?

exposure and outcome by comparing the frequency of exposure in cases vs. controls.

Selection of cases and controls

from total population you take a Base population. then separate that group into two, the case and controls

Do you need a clear definition of a case (define outcome well)?

yes, Many types of heart disease for example - what are you really interested in?

Analytical study Exposure and outcome measured at same time?

NO

Are exposure criteria used to select cases?

NO

what is the distinguishing feature between study designs?

Temporality or Temporal Sequence

(T/F) Investigator decides on number and study base

True

Limitations of case-control study: 1. Temporal sequence between exposure and outcome is ________ 2. Did disease cause exposure? 3. Representativeness of cases and controls often _____ 4. How many outcomes can you study at one time? why? 5. Is this design effective for studying rare exposures - why? 5. Are there biases?

-unclear - unsure therefore not as high causality -unknown -Can only study one outcome at a time because this design STARTS with collecting subjects based on their disease status. - No, Rare exposures - not good, because you may not have enough cases/controls to see even one exposed subject if the exposure is very rare. -yes, Did use of opiates cause pneumonia or did pneumonia cause use increased use of opiates

Limitations Case-Control Studies (5)

1. Temporal sequence unclear 2. Not good for rare exposure 3. Can only study one outcome/disease 4. Susceptible to biases (Recall, Berkson's) 5. Control group selection

How would you design this study? Topic: Birth control use and breast cancer risk Base population source? Case definition? How would you select the controls? Exposure assessment?

Base population source - hospital or several hospitals Case definition - 25-59 years old with first occurrence of primary breast cancer (diagnosed within previous year) and no concurrent cancer or history of cancer) Control definition - same age, no history of cancer admitted to the hospital for nongynecologic nonmalignant conditions unrelated to contraceptive use or reproductive factors. Matched on age and geographic area. So controls could be admitted to hospital for orthopedic disorders, traumatic injury, acute infections, other. Exposure assessment? - want to know when took the birth control, for how long, what brand (for estrogen content) so can get at concentration, duration, time first taken. Can ask self report but perhaps can use pharmacy records? Where to find breast cancer cases? - Cancer registry Case definition - Women with invasive breast cancers diagnosed between 2000 and 2005 Controls - Persons of comparable age and sex from the cancer registry with other types of cancer? Exposure assessment - Self-reported birth control use (questionnaire)

________ - Occurs when using cases and controls from hospitals

Berkson's Bias -controls' exposure history is more like cases than general population

TopHat Question. Which of the following best describes a randomized controlled trial design? A. Study population who is exposed/not exposed and followed over time for outcome. B. Study population who has outcome/does not have the outcome and are asked about history of exposure. C. Study population is randomized to intervention/standard treatment groups and followed over time for the outcome. D. Study population is randomized to have the outcome/to not have the outcome and followed over time for development of exposure.

C. Study population is randomized to intervention/standard treatment groups and followed over time for the outcome.

Why does the case-control design not directly provide incidence data?

Cannot provide incidence data because the people you are selecting already have/don't have the disease. Incidence can only be determined by watching people over time and seeing who does/does not get the disease.

Why does the case-control design not directly provide prevalence data?

Cannot provide prevalence data because we are deliberately selecting people BECAUSE they have the outcome so we are SAMPLING or choosing people solely based on them having the outcome. Prevalence determines how much of a disease is present in a population at a given point in time. So, if we tried to measure prevalence using our data, would we be overestimating or underestimating the prevalence?

Did participants enter the study based on: Outcome

Case-Control

______ - have disease or outcome of interest ______ - comparable to cases but are free of the disease or outcome.

Cases Controls

Identify source or base population for cases

Cases are a random sample of all available cases in source population (or all cases if very rare disease)

Did participants enter the study based on: Exposure

Cohort

Controls DO NOT/DO have the disease/outcome

Do NOT

Overview of a case-control study

Investigator selects one group with disease or outcome (cases) and a comparable group without disease (controls) Outbreak investigations

what does temporality or temporal sequencing mean?

Means the timing of information gathering For the exposure and outcome: -Were they determined at same time? -Was outcome determined first, then exposure? -Or, was exposure determined first, then outcome? (more confident of cause-effect relationship)

Are exposure criteria used to select controls?

NO

Directionality: Measure _______ 1st Measure _______ 2nd

Outcome Exposure

Can you think of other exposure/outcome case-control studies where recall bias might occur?

Smoking and lung cancer -Difficult to remember quantity smoked, when began smoking, etc Maternal alcohol usage and birth defects -Stigma attached to alcohol use Anything during pregnancy and adverse pregnancy outcome - guilt, what did I do to cause this to happen? Did use of opiates cause pneumonia or did pneumonia cause use increased use of opiates

what is bias?

Systematic deviation from the truth

In a Case-Control study, Sampling is based on __________

outcome

this study design is Defined by what?

presence or absence of the outcome

__________ - Inaccurate recall of exposure resulting in exposure measurement error

recall bias -Case may be more likely to recall exposure due to having the disease than the control who does not have the disease

Exposure is determined when?

retrospectively, Sometimes called retrospective study