L02TypesEpidStusy 24-11
L02TypesEpidStusy 24-11
L02TypesEpidStusy 24-11
time
Study begins here
Increasing Knowledge of Descriptive Studies Develop hypothesis
Disease/Exposure
• Advantages
• Useful for hypothesis generation
• Informative for very rare disease with few established risk factors
• Characterizes averages for disorder
• Disadvantages
• Cannot study cause & effect relationships
• Cannot assess disease frequency
Case Report One case of unusual findings
Association - the occurrence of 2 variables more often than would be expected by chance
time
Study only exists at this point in time
Cross-sectional Studies
• Often used to study conditions that are relatively frequent with long duration of
expression (non-fatal, chronic conditions)
• It measures prevalence, not incidence of disease e.g. community surveys
• Not suitable for studying rare or highly fatal diseases or a disease with short
duration of expression
• Disadvantages:
• Weakest observational design, (it measures prevalence, not incidence of disease), prevalent
cases are survivors
• The temporal sequence of exposure & effect may be difficult or impossible to determine
• Usually don’t know when disease occurred
• Rare events or quickly emerging diseases are a problem
Case control study (Retrospective study)
Features:
1. “Observational” design - compare exposures in disease cases vs. healthy controls from same
population
2. Both exposure & outcome have occurred before start of study
2. Study proceed backwards from effects to cause
3. Uses control or comparison group to support or refute hypothesis
4. Involves two populations – case (with disease) & controls (no disease)
5. Unit of study is the individual
6. It is basically a comparison study
7. Cases & controls should be comparable to certain factors – age, sex, social status, occupation
8. Exposure data collected retrospectively
Steps in conducting a Case-Control study
1 Selection of cases • Diagnostic criteria, Eligibility criteria, Source of cases – hospital, population
& selection of • Free from disease under study
controls • As similar to cases as possible except the disease under study
• Group identified before study, who have not been exposed to factors whose influence
is being studied
• Sources – hospitals, relatives, neighborhood, general pop
2 Matching • Ensure comparability
• Process of selection of control with similar to cases – age, sex which are known to
influence outcome
• Confounding factor – which is associated both with exposure & disease and is
distributed unequally in the case & control group.
• Neutralize confounding effect, protect against unexpected strong association between
matching factors & disease
• Pair matching or group matching
time
• ”Case component" = hazard period which is the time period right before the
disease or event onset
• Bias - any systematic error in the determination of association between exposure & disease
• Relative risk estimates may increase or decrease as a result of the bias, reflecting a non-
compatibility between the study and the control group
Types of bias
Confounding bias • Due to unexpected strong association between matching factor & disease
• Can be reduced by matching
Recall bias • Cases may recall better than controls who are healthy
Selection bias • Case and controls may not be representative of the population due to systematic
differences in characteristics between cases & controls
Berkesonian bias • Due to different rates of admission to hospitals for people with disease
Interviewer bias • May happen if interviewer is aware of the hypothesis or knows who the cases are.
May influence the degree of questioning more thorough than controls regarding a
positive history of suspected causal factor
• Can be eliminated by double-blinding
Case control studies
Advantages Disadvantages
Relatively easy to carry out Problem of bias
Suitable to investigate rare disease Cannot measure incidence, only relative risk
Allow study of different etiological factors Unsuitable to evaluate therapy or prophylaxis of a disease
No attrition problem as follow-up not required Representativeness of cases & controls can be a problem
present
future
time
Study begins here
Prospective cohort study
Prospective Study - looks forward, looks to the future, examines future events,
follows a condition, concern or disease into the future
Exposed Outcome
time
Exposed Outcome
Measure exposure
and confounder
variables
Non-exposed Outcome
Baseline
time
Study begins here
Activity:
Can you see the difference between
retrospective case-control & retrospective
cohort study?
Case-Control Study
Steps in conducting cohort study
1. Selection of study subjects
General population • When exposure is fairly frequent, If population too large, sample identified
Special select group • Select groups maybe professional eg dentist, lawyer (Doll’s study - done among British doctors
listed in medical register 1951)
• Exposure groups if rare, more economical to select a known cohort who have experienced the
exposure eg radiologist exposed to x-rays
2. Obtaining data regarding exposure
Cohort members • Interviews, mail questionnaires
Review records • Radiation dose, type of surgery, treatment, exposure to risk factor eg. gutka chewing
Medical examination/special tests • Blood test, blood pressure, serum cholesterol, ECG
Environmental surveys • Exposure levels of suspected factor in environment where cohort lived/worked
3. Selection of comparison group
Internal comparison • Single cohort enters study, members may be grouped to compare, compare morbidity/mortality
External • When there is no data on degree of exposure, external control maybe required to evaluate
experience of exposed group eg tobacco chewers & non chewers
• Study & control grp should be similar in demographic & important variables except those under study
Comparison with • If no comparisons are available, then exposed group is compared to general population eg. frequency
general population of cancer among ceramic workers with rate in general population in the same geographic area.
Steps in conducting cohort study (Cont)
4. Follow-up
a. Periodic medical examination of each member Attrition rate (percentage of losses)
- Inevitable due to death, change of residence,
b. Review hospital records
migration, withdrawal
c. Routine surveillance of death records - May cause bias in results
d. Mailed questionnaire, phone calls, periodic home visits etc - For good results, at least 95% follow up is necessary
5. Analysis
Incidence • Directly determined in those exposed & non exposed
rates
Estimation Relative risk (RR) Attributable risk Population-attributable risk
of risks - Ratio of incidence of disease among exposed & non- - Difference in incidence - Incidence of disease in
exposed rate of disease total population minus
- Direct measure of strength of association between between exposed & incidence of disease
suspected cause & effect non exposed among non-exposed to
- RR=1 suggest no association, RR>1 is positive - Indicate the extent of suspected causal factor
association, RR<1 is negative association disease that can be
- RR=2 indicate incidence rate 2X higher among exposed attributed to exposure
- The large the RR, the greater strength of association
between suspected factor & the disease
Cohort studies
Advantages Disadvantages
1. Incidence can be estimated 1. Need large number of people (unsuitable for investigation of uncommon
disease)
2. Possible outcomes related to 2. Take longer time – patients/investigator may die, difficulty to keep large
exposure can be studied number under medical surveillance
3. Provide direct estimate of relative 3. Administrative problems – loss of expensive staff, funding, extensive
risk record maintenace
4. Dose-response ratio can be 4. Loss of original cohort – migration, patients may loose interest in
calculated study/refuse to give information
5. Certain biases can be minimized 5. Comparison grp may not be representative of population
6. There may be changes in diagnostic criteria over time
7. Expensive
8. Study itself may change people’s behavior
9. Ethical issues
10. Practicality – limited number of factors related to disease outcome
Compare & contrast
Case control study Cohort study
1. Proceeds from effect to cause 1. Proceeds from cause to effect
2. Start with the disease 2. Start with people exposed to RF/suspected cause
3. Tests if suspected cause occurs more 3. Tests whether disease occurs more frequently
frequently in those with disease than those exposed than in those not exposed
those without disease
4. First approach to testing hypothesis 4. Testing of precisely formulated hypothesis
5. Involve fewer subjects 5. Involve large number of subjects
6. Relatively quick results 6. Long follow-up, delayed results
7. Suitable for study of rare diseases 7. Inappropriate when disease/exposure under
investigation is rare
8. Yields only RR 8. Yields incidence rate, RR & attributable risk
9. Does not yield info about disease other 9. Can give info about more than one disease
than the selected study outcome
10. Relatively inexpensive 10. Expensive
Experimental epidemiology
Aims:
1. To provide scientific proof of etiological or risk factors which may permit the
modification or control of disease
• Subjects in the study who receive no treatment or a different treatment - comparison group.
Animal studies
• Induce epidemics in animal under lab conditions
• Experimentally reproduce human disease in animals to confirm etiological hypothesis
• Test efficacy of preventive & therapeutic measures of vaccine/drugs & completing natural
history of disease
Advantages Disadvantages
1. Experimental animals can be bred in 1. Not all human disease can be
labs & manipulated according to needs reproduced in animals
of investigator
2. Experimental animals multiply rapidly 2. All conclusions derived from animal
experiments are not applicable to
humans
3. Ethical clearance available in many 3. Animal rights issues have been raised
countries
Human experiments
• One best method to test preventive & therapeutic measures
• James Lind (1747) – citrus fruits in scurvy
• Edward Jenner (1796) – cowpox
• Finlay & Reeds (1881 – 1900) – mosquito home nature of yellow fever
• Issues are ethical & logistic considerations
2 types:
a. Randomized Control Trial (RCT)
b. Non-randomized/non-experimental trials
Randomized Controlled Trials
• A design with subjects randomly assigned to “treatment” and “comparison” groups
• Provides most convincing evidence of relationship between exposure and effect
• Not possible to use RCTs to test effects of exposures that are expected to be harmful, for
ethical reasons
• “gold standard” of research designs
• Provides most convincing evidence of relationship between exposure and effect
e.g. Trials of hormone replacement therapy (HRT) in menopausal women found no
protection for heart disease, contradicting findings of prior observational studies
• Disadvantages
– Very expensive
– Not appropriate to answer certain types of questions
• it may be unethical e.g. to assign persons to certain treatment or comparison groups
RANDOMIZATION
outcome
Experimental Design Intervention
no outcome
Study
population
outcome
Control
no outcome
baseline
future
time
Study begins here (baseline point)
What is randomization?
i. extensively used in human clinical trials & other biological experiments.
ii. prevents selection bias ie each patient has equal chance of receiving any of the treatments under study
iii. insures against the accidental bias
iv. produces comparable groups - ie. alike in all important aspects except for the intervention each group receives
v. eliminates bias in treatment assignments
vi. it permits use of probability theory to express likelihood of chance as a source for the difference of end outcome
Block randomization
• Designed to randomize subjects into groups that result in equal sample sizes.
• Ensure a balance in sample size across groups over time.
• Blocks are small & balanced with predetermined group assignments, which keeps numbers of subjects in each group similar at all times.
• The block size is determined by the researcher & should be a multiple of the number of groups (with two treatment groups, block size of either
4, 6, or 8).
• Blocks are best used in smaller increments as researchers can more easily control balance.
• After block size has been determined, all possible balanced combinations of assignment within the block (equal number for all groups within
the block) must be calculated.
• Blocks are then randomly chosen to determine the patients’ assignment into the groups.
• Although balance in sample size may be achieved with this method, groups may be generated that are rarely comparable in terms of certain
covariates. For example, one group may have more participants with secondary diseases (e.g., diabetes, multiple sclerosis, cancer,
hypertension, etc.) that could confound the data & may negatively influence the results of the clinical trial. Therefore important to control for
these covariates because of serious consequences to interpretation of results. Such imbalance could introduce bias in statistical analysis &
reduce power of study. Hence, sample size & covariates must be balanced in clinical research.
Stratified randomization
ii. Double blind – neither doctor nor participant aware of group allocation & treatment
received
iii. Triple blind – doctor, participant & statistician not aware of groups or treatment allocation
*This method is unsuitable if drug of interest cures the disease, drug is effective only during
certain stage of disease or disease changes radically during period of study
Non-randomized control trials
RCT not always possible due to:
• Ethical/administrative reasons
• Some preventive measures can be applied to groups/community
• Require long time
• Expensive because disease frequency low or natural history is long
Types of non-randomized trials
1. Uncontrolled trials – useful in initial evaluation whether a specific therapy has value in a particular disease or
to investigate adverse reaction
2. Natural experiments – since experiments not possible in human, epidemiologist seeks to identify natural
circumstances eg smokers, non-smokers, migrants, religious or social groups, families, earthquakes