Module 5 Flashcards
What are some health-related purposes of observational research?
- Answer questions of test hypotheses about a single variable rather than a statistical relationship between 2 variables
- Answer research question about a non-causal statistical relationship between variables
- If independent variable cannot be manipulated or participants cannot be randomized - Causal relationships can ONLY be examined using observational designs
- Answer broad and exploratory research questions
What is the goal of observational research?
To identify areas for further study of to draw inferences about relationships by collecting information about exposures (Independent variables) and outcomes (dependent variables)
What are some “flaws” of observational research?
no randomization and no control over the environment
Why must alternate explanations be considered for observational quantitative studies?
No randomization
No control over the environment
Do observational quantitative studies always test a hypothesis?
no, not if it is descriptive
What is epidemiology?
Study of how and why different patterns of health and disease occur among various subgroups in a population
What are the 3 Ds in the 3D definition of epidemiology?
Distribution
Determinants
Deterrents
What is distribution?
Frequency and patterns of morbidity and mortality in a given population or community
What are determinants?
Factors that contribute to morbidity and mortality in human populations
What are deterrents?
Factors that prevent of reduce morbidity and premature mortality in human populations
What are the 2 subdivision of observational epidemiology?
Analytic
Descriptive
What is descriptive epidemiology?
Describe the frequency and distribution of determinants and deterrents of morbidity and mortality in a given population or community by person, place, and/or time
- Used to help inform health programs and services, identify health-related issues and trends, and suggests hypotheses for further study
- Who What When Where -
What is analytic epidemiology?
Observational methods are used to uncover the relationship between suspected determinants or deterrents or morbidity and mortality (exposures) in a given population of community
Used to identify risk factors for disease and explain disease patterns in a population
- How Why-
Why are observational studies conducted?
- Collect descriptive information (how many people have a disease)
- Study the potential effects of exposures on disease outcomes
- Investigate a problem without interfering with the natural setting
- Learn about health risk or benefits in humans when a experiments study is not feasible or unethical
What is clinical equipoise in experimental study and why is it important?
There can be neither enough enough for or against a treatment to preclude its use as an intervention in research. If this isn’t met only observational study can examine the relationship between exposures and outcomes.
eg: you can’t make people smoke to study effects of cigarettes OR you cant withhold insulin from Type 1 diabetic to see what happens.
What is the main assumption about disease distribution?
They do not distribute randomly in populations, but distribute in relation to their determinants; therefore, determinants of diseases can be identified by studying their distributions
Describe sampling in Observational studies
They take samples from a study population. These subsets are “randomly selected” (each member of the population has a known probability of getting selected)
In epidemiology, what are exposures?
Independent variables
Usually risk factors, but are understood as ANY factor that might influence health positively or negatively
In epidemiology, what are outcomes?
Dependent variables
Usually diseases, but are understood as ANY type of health state (pregnancy, health behaviours, perception of well-being)
What is an “Association” in epidemiology?
Mathematical relationship representing the amount of dependence between 2 variables (exposure and outcomes)
What are the 3 types of associations in epidemiology?
Spurious
Non-causal
Causal
What is a spurious association?
The association is simply not true. Can be caused by random error or by bias
What are non-causal associations?
Real associations statistically, however it is due to other factors eg One did not cause the other
What are causal relationships?
When changes occur in the occurrence of the outcome can be shown to be the direct result of an exposure
According to Hill’s postulates, causality is most likely when there is (8) ?
- Strength of association
- Consistency
- Specificity
- Temporality
- Biological gradient
- Biological Plausibility
- Coherence
- Experimental Evidence
- Analogy
What are the 2 types of common measure in observational epidemiology6
Measure of Occurrence
Measure of Association
What are the 2 measure of occurrence?
Prevalence
Occurrence
What is prevalence?
Number of events/disease cases in a population. Usually reported as a proportion but can also be reported as odd
What are the 2 common types of prevalence?
Point prevalence (at a specific point in time) Period prevalence (during a period of time)
What is incidence?
Number of NEW or INCIDENT cases of the specified outcome in a population during a specified period of time
What are the 2 common types of incidence?
- Incidence proportion
Proportion of an initially outcome-free population that develops the specified outcome during a speficied period of time - Incidence rate
Speed at which the outcome is observed; it is a measure of incidence that directly incorporates time. Divide incidents by person-time
What is the use of prevalence?
Measure of disease burden at a point in time or over a period of time
What is the use of incidence proportion?
Identify risk factors by identifying groups in which new cases are more likely to occur
What is the use of incidence rate?
Identify risk factors by identifying groups in which new cases occur more quickly
What are the 2 measures of association?
Relative Risk
Odd ratio
What is Relative Risk?
of times that the exposed group is more likely to develop the outcome compared to the unexposed group
What does it mean when the relative risk = 1?
Risk of the outcome occurring is equal in each group (exposed and not exposed)
What does it mean when the relative risk < 1?
Risk of the outcome occurring is LOWER in the EXPOSED group. Exposed group is 1/X times less likely to develop the outcome than not exposed group
What does it mean when the relative risk > 1?
Risk of the outcome occurring is HIGHER in the EXPOSED group. Exposed group is X times more likely to develop the outcome than not exposed group
What do we mean by “measure of association”?
Ways of quantifying the relationship or association between an exposure and an outcome
What is the equation for relative risk?
Incidence in exposed group/incidence in group not exposed
a/(a+b))/(c/(c+d)
What is the Odd Ratio?
Summarizes the relationship between an exposure and an outcome - Measures relative odds: odds that an outcome will occur given a particular exposure compared to the odds of the outcome occurring in the absence of that exposure
What is the equation of Odds Ratio?
odds outcome will occur if exposed/odd outcome will occur if not exposed
(a/c)/(b/d)
What does it mean if OR=1?
Indicative of NO association between the exposure and the outcome
What does it mean if OR>1?
Indicative of a positive association between exposure and outcome - odds of the exposure among those with the outcome is greater than the odds of exposure among those without the outcome
What does it mean if OR<1
Indicative of a negative association between exposure and outcome - Odds of exposure among those with outcome is less than the odds of exposure among those without the outcome
What is a “rare” outcome?
Prevalence proportion is less than 5% of the study population
What does the interpretation of Odds ratio depend on?
How rare the outcome of interest it and where prevalent or incident (new) events/outcomes are studied
How do we interpret the odds ratio if the disease is rare and NEW cases/outcomes are studied?
Odds ratio will be reasonable approx of the relative risk and can be interpreted as reflecting excess “risk”
Reasonable to say a significant finding of an OR>1 indicates those with the exposure factor are at greater risk of the disease
How do we interpret the odds ratio if the disease is NOT rare and/or outcomes are not limited to new events?
Odds of the exposure among those with the outcome is greater than the odds of exposure among those without the outcome
What are the relative strengths brackets for inverse and positive association?
Weak Association
Inverse: 0.71-0.99
Positive: 1.01-1.50
Moderate Association
Inverse: 0.41-0.70
Positive: 1.51-3.00
Strong Association
Inverse: 0.00-0.40
Positive: >3.01What is the difference between descriptive and analytic observational studies?
Descriptive epidemiology seeks to describe the occurrence of exposure and outcomes (Who, what, when, Where)
Analytics epidemiology seeks to analyze the association between exposures and outcomes (how and why)
Describe descriptive observational studies
Describe the occurrence of and exposure and/or outcome in a population. Who What When Where. May suggest future direction to explore, but limited in the ability to answer the “why” due to the lack of an explicit comparison groupq
Describe analytic observational studies
Explain the how and the why
Test a hypothesis
Has a formal comparison group
What is a cohort study?
Basic premise is that the researcher selects a group of people with different levels of exposure and follows them through time to see what happens to them - All free of the outcome at the time exposure status is determined
Can be prospective of retrospective
What is the difference between a prospective or retrospective study?
Prospective: the present (where the investigator is) corresponds to the time of the exposure and the outcome is in the future
Retrospective: exposure has happened in the past, outcome may already be known. Investigator select participants based on past exposure
What are the strengths of a cohort study?(4)
Establish temporality
Determine incidence
Study multiple effects of a single exposure
Suitable for studying rare exposures
What are the weaknesses of cohort studies? (5)
- Time consuming and expensive
- Inappropriate for studying rare outcomes
- Potential for exposure misclassification (change in status during follow-up or inadequate data)
- Potential for outcome misclassification due to advances in disease detection during follow up
- Potential for diagnostic suspicion bias
What is a prospective cohort study?
Cohort is created based on current or past exposure and followed into the future to determine if the outcome develops.
Considered to provide the strongest observational evidence
What are some characteristics of a prospective cohort study? (4)
- Study INDIVIDUAL characteristics
- All participants at the onset of the study are free of the outcome
- Measure association of exposure and outcome based on risk
- Can be used to determine the etiology and/or prognosis of a disease
What are the steps to prospective cohort study design? (4)
1- Subset of the population is identified that is free of the study outcome
2- Participants are classified into exposed and not exposed
3- Follow up period needs to be for long enough that sufficient amount of the study outcomes have occurred
4- Occurrence of the outcome (risk or the rate of the outcome) in those exposed is compared to the occurrence of the outcome in those not exposed
What is a retrospective cohort study?
Historical cohort constructed using past records. Exposure status is determined based on information recorded in the past. Cohort is followed into the present or more recent past to determine outcome status
What are the strengths of retrospective cohort studies?
- Considered to be practical and economical when good exposure record exists and outcome status cant be stacked
- Avoids long periods of follow-up
What are the weaknesses of retrospective cohort studies?
- Limited to studying mortality of cancer outcomes given the limited access to universal records
What is a case-control study?
Target population is chosen and two groups of participants are selected based on the outcome being present of absent. Cases (with outcome) are compared to a group of controls (without outcome). Past exposure is collected from both groups
What are some characteristics of case-control study? (5)
- Study INDIVIDUAL characteristics
- Selection of participants is based on outcome independent of exposure status
- Cases: group of participants with the outcome
- Controls: group of participants without the outcome
- Efficient to study the relationship between exposures and rare outcomes
What are the steps of a case-control study?
1- Everything has already happened and the investigators select participants based on their outcome status
2- Select a group of cases and a group of controls. Precise source and the mechanism for selecting cases and controls may be different but hey should come from the same population
3- Past information on exposure is collected for both groups and then compareg. Association between exposure and outcome is based on odds. The odds of exposure in cases is compared to the odds of exposure in controls
What are the strengths of a case-control study? (5)
- Quick and inexpensive
- Appropriate for studying rare outcomes
- Can be conducted with moderate numbers of subjects
- Allow multiple exposure to be examined in the same study
- Suitable for studying chronic disease with long induction or latency periods
What are the weakness of a case-control study? (5)
- Incidence and prevalence cannot be determined
- Not appropriate for studying rare exposures
- Information on prior exposure or potential confounders may not be readily available, accurate, or of the same quality between cases and controls
- May be difficult to identify comparable case and control groups causing a greater potential for bias
- May be difficult to establish temporality
What is the main threat to internal validity in observational studies?
systematic errors = due to inaccuracies in the design or conduct of the study
What are the 2 types of systematic errors?
- Bias
- Extraneous factors
What are the 2 types of bias that can be encountered in observational studies?
- Selection Bias
- Information Bias (aka measurement bias)
What is information bias?
Aka measurement bias, is an error in measurement whereby the exposure and outcome are classified differently in the groups being compared.
If error is the same across both groups = non-differential
If error is different across both groups = differentia
What can cause information bias? (3)
- Recall error (one group is more likely to recall an exposure than another)
- Detection error
- Lack of accuracy of tools
What is selection bias?
Systematic errors that results from procedures used to select participants and from factors that influence participation in the study
What can cause selection bias? (3)
- Healthy Worker Effect
- Volunteer Bias
- Loss to follow-up bias
What is the healthy worker effect?
Potential bias caused by the phenomenon that sicker individuals may choose work environments in which exposures are low - usually seen in prospective studies
What is the volunteer bias?
When those who choose to participate are different from the source population with regard to their probability of exposure (case-control) or outcome (cohort)
What is loss to follow-up bias?
When participants can no longer be located of when they choose to no longer participate in the study - will introduce bias if those who are lost are systematically different from those who remain
What are component causes?
An exposure that causes a disease through its participation in casual mechanisms that typically include a combination of causes
How do you determine external validity in observational studies?
1- Population sampling: Ensure population sampling is representative of the population of interest
2- Prevalence of component causes: Careful understanding of the distribution of component causes across populations
What are the 3 activities of reproducibility?
1- Reproducing the process of investigations
2- reproducing the results of investigations
3- reproducing the interpretation of the results
