week 4- observational study designs

analytic studies

designed to examine etiology and causal associations → observational studies

observational studies types

cohort, case-control, cross-sectional, ecological

If observations are made looking forward they are called ____

prospective observational

If observations are made looking back from existing data

retrospective

Cross-sectional:

made at one point in time

Two types of analytic studies

experimental studies and observational studies

experimental studies

the intervention is under the control of the researcher

observational studies

the researcher simply observes the participants at one point in time (cross-sectional) or over time (longitudinal)

case control observational study

case-subjects are selected based on a defined outcome, and a control group of subjects is selected separately to serve as a baseline with which the case group is compared

cross-sectional studies

observational study designed to collect data on an outcome and intervention variables of interest at one point in time

ecological studies

large groups are compared without having information on the individuals within the group (using population data)

historical control group

clients in the previous 6 months

concurrent control groups

clients in another department

internal validity

the extent to which the results reflect the truth about what happened within just this study

internal validity evaluated by

sample identification ad selection process. exposures. results are not due to confounding variables

external validity

the extent to which the results reflect the truth about what may happen outside this study, but in similar situations

external validity evaluated by

sample similarity to target population, setting, protocol of study

confounding variable

factor that influences both the cause and potential effect

For a variable to be "true" confounder it has to meet the following conditions

relationship with exposure and outcome. uneven distribution in comparison groups (non-random designs)

common confounders

Co-intervention: participants receive other interventions at the same time as the intervention of interest

bias

result of a systematic error in a study design, can not be controlled for statistically

sampling/selection bias

reference population have the same probability of inclusion in the sample, bias is possible

maturation bias

effect might be due to changes that have occurred naturally over time

measurement bias

systematic error arising from inaccurate measurement of subjects

response bias

differences between individuals who choose to respond or not

convenience bias

sample taken from individuals that are conveniently available

bias vs. confounders

confounders can be statistically controlled for. once bias is present, no way to adjust for it in results