Epidemiological Study Designs: A Comprehensive Overview, Exams of Epidemiology

Download Epidemiological Study Designs: A Comprehensive Overview and more Exams Epidemiology in PDF only on Docsity!

Epidemiological Study Designs

Course: Infectious Disease Epidemiology

(EPI 522 )

Harvard University

Lecture Objectives  Distinguish strengths and limitations of cross- sectional, ecological, case control and cohort methods of observational studies  Understand the differences between study designs and the use of appropriate study designs to answer specific research question.

Cross sectional Study  Can be descriptive (disease distribution) or comparative/analytical (disease determinants)  Exposure and disease measured at the same time.  Often sample of the general population collected.  Prevalence of outcome amongst exposed and unexposed computed.  Prevalence odds ratio calculated.

Total Population (T) E+D+ E-D- E+D- E-D+ Sample (t) E+=Exposed, E-=Non-exposed, D+=diseased, D-=Non-diseased

Computations in Cross-sectional studies Diseased Non-diseased Total Exposed (^) a b a+b Non-exposed (^) c d c+d Total (^) a+c b+d a+c+b+d Prevalence of disease in exposed= a/a+b (Pe) Prevalence of disease in non-expose c/c+d (Po) Prevalence Ratio=Pe/Po

Cross-sectional Study– Example  Research Question: Does the level of physical activity affect the risk of Coronary Heart Disease (CHD)?  Compared with individuals with high physical activity, how high is the risk of CHD among individuals with low physical activity? CHD No CHD Total Low Physical Activity 56 389 High Physical Activity 12 388 Total

Advantages  Relatively inexpensive and can be conducted over a short time.  Results can be easily generalized (if sample is randomly selected)  Good for diseases with long duration and low Case Fatality Rate.

Disadvantages  not efficient for rare diseases- large sample required.  not good for diseases of short duration or periods of exacerbation and remissions.  difficult to establish temporal sequence – (can not separate cause and effect)  excludes those who die or gets cured quickly.  can not measure disease incidence

Ecological/Correlational Studies

 Aggregate (group) and NOT individual

data on exposure and outcome are

collected

 Prevalence of an exposure is correlated

with prevalence of an outcome.

 A correlation coefficient (value (-1 to +1)

is computed

Ecological Studies—Examples

 Per capita daily fat consumption &

incidence of colon cancer

◦ What is the source of data on fat consumption and colon cancer?

 Mammography rates & incidence of

breast cancer

 TB case-finding rate & adult

mortality

Ecological Studies—Main Limitation  Can not establish a causal link between exposure and outcome ◦ Ecological fallacy– making inferences about individual phenomenon, process or outcome based on group data (Selvin, 1958)

Ecological Studies—Strengths and uses

 Fast, may be able to take advantage

of previously collected data

 Can be used to generate hypotheses

or test the plausibility of a new

hypothesis

 Can establish the direction and

strength of relationship between

exposure and outcome

Total Population (T) E+D- E-D- E+D+ E-D+ Sample (t) Follow-up E+=Exposed, E-=Non-exposed, D+=diseased, D-=Non-diseased Follow-up 2005 2005 2015 2015

Design Considerations  Must exclude prevalent cases and those not at risk of a specific outcome  Must decide criteria for detecting outcome and how frequently exposed and unexposed people should be followed  Must decide length of follow up to allow the outcome presentation.