Research Study designs

representativeness

how accurately data reflects the occurrence and distribution of a disease in a population

observational studies

observe populations under prevailing conditions

NOT appropriate for assessing causality because they do not involve introducing or modifying a treatment

can compare observations across populations or withing populations over time

no random assignment

representativeness

how accurately data reflects the occurrence and distribution of a disease in a population

observational studies

observe populations under prevailing conditions

NOT appropriate for assessing causality because they do not involve introducing or modifying a treatment

can compare observations across populations or withing populations over time

no random assignment

longitudinal study

observational study over a long period of time; information about subjects is recorded without manipulating the study environment

can last many years

assesses: what is the effect of A on B?

does A cause B?

longtitudinal data

follows large groups of people over a long time; the groups are very similar in many ways, but differ by certain characteristics (ex: female nurses who smoke and those who do not smoke)

a particular outcome (ex: lung cancer) is compared between the groups

same people in each group are assessed multiple times throughout the study

emphasizes correlational research involving looking at the variables over an extended period of time; involves comparison groups (exposed vs non-exposed group)

strengths of longitutidnal

1. can assess many outcomes for one exposure

2. can look at multiple exposures

3. exposure is measured before onset of disease

4. good for measuring rare exposures

5. demonstrates direction of causality

6. can measure incidence and prevalence

weaknesses of longitudinal

1. costly and time consuming

2. prone to bias due to loss to follow-up

3. prone to confounding

4. participants move between one exposure category

5. knowledge of exposure status may bias classification of the outcome

6. being in the study may alter participant’s behavior

7. poor choice for study of rare disease

8. classification of individuals can be affected by changes in diagnostic procedures

cross sectional

observational study at one time point; information about subjects is recorded without manipulating study environment

is A associated with B?

compares different population groups at a single point in time (like a snapshot)

cross sectional data

big gems are indicators

cross sectional strengths

1. quick and easy

2. data on all variables collected simultaneously

3. measures prevalence for all factors under investigation

4. multiple outcomes and exposures can be studied

5. descriptive analyses for generating hypotheses

cross sectional weaknesses

1. difficult to determine if outcome followed exposure in time or exposure resulted from outcome

2. bad for rare diseases or short-lived disease

3. measure prevalence rather than incidence

4. associations can be difficult to interpret

5. susceptible to bias due to low response

case-control study

observational; study population consists of groups who either have (cases) or do not have a particular health problem (control)

investigator looks back in time to measure exposure; exposure is compared amongst cases and controls to determine if exposure causes health condition

is having condition B associated with having been exposed to A?

strengths of case control

1. cost effective

2. retrospective, cases identified in begining, no follow-up

3. effcient for diseases with long latency periods and rare diseases

4. examines multiple exposures

weaknesses of case control

1. prone to bias

2. limited to examining one outcome

3. unable to estimate incidence rate

4. bad for rare exposures

5. time between exposure and disease is difficult to determine

causal designs

assess whether an independent variable causes change in a dependent variable

• nonexperimental

• experimental

• quasi-experimental

nonexperimental

no control group, no random selection, no random assignment, least amount of control

quasi-experimental

often a control group, random selection of participants, random assignments of participants and treatments, decent amount of precision and control

true experimental

control group present, randomly selected participants, random assignment of participants, random distribution of treatments, most control

randomized clinical trial (RCT)

experimental study; participants are allocated randomly to receive one of several clinical interventions

strengths of random clinical trial

1. provides the most substantial evidence of any epidemiology study design

2. best for causality

3. randomizations controls for confounding variables

4. clear temporal sequence

5. strong basis for statistical inference

6. measures incidence and multiple outcomes

weaknesses of RCT

1. ethical constraints

2. expensive and time consuming

3. requires complex design and analysis

4. ineffcient for rare disease

5. subjects may not be representative of entire population