Epidemiology- study for quiz 3 – Flashcards

Observational or Experimental Observational studies Involves simply observing what occurs in nature No external manipulation of the study Experimental design allows use of control techniques: Randomization Masking/blinding to reduce bias Observational studies considered to be "natural" experiments Experimental studies considered to be "true" experiments Observational: Analytic - hypothesis testing, descriptive - only describes Experimental studies - analytic, always hypothesis testing

Observational vs. Experimental Designs Observational Investigator does not manipulate exposure status of exposure study participants. Are "observing" what occurs naturally. Most epidemiologic studies are observational. Experimental Investigator manipulates the exposure. Equivalent to a laboratory experiment. Considered to be the "gold standard" for establishing causality.

Observational Observational Ecologic studies Cross sectional Case-control Nested case-control Cohort Retrospective cohort Prospective cohort Experimental Randomized trials Clinical Community Nonrandomized (quasi-experimental)

Descriptive Generate hypotheses (are not hypothesis testing) Usually have no comparison group Focus on patterns of disease Analytic Test a hypothesis Usually include a control or comparison group Use a more complex design by person, place, and time

"Factors That Influence the Choice of Study Design Nature of the research question How much is known about the topic Nature of the disease (induction period) Frequency of disease or exposure Resources available (cost, personnel, time) Available data Ethical issues" TRUE. A cohort study would be the best design choice when little is known about a rare exposure. about a rare exposure.

FALSE. A Case-Control study would be the best design choice when little is known about a rare disease.

FALSE.

TRUE. A cohort study would be the best design choice when you want to learn about the multiple effects of an exposure.

TRUE. A retrospective cohort study is more efficient than a prospective cohort study for studying diseases with a long latent and induct

FALSE. Cohort studies are the most sensible design for examining many exposures in relation to a single disease.

FALSE. The ideal comparison group for a cohort study would consist of exactly the same individuals in the exposed group had they not been exposed.

Ecologic Studies: Design Based on information at population level Association between exposure and disease is based upon average exposure levels and average disease rates. Studies of group characteristics No individual data Ecologic Studies: Advantages Availability of data Can be done quickly, with limited resources Analysis is fairly simple May facilitate a range of exposure levels Exploratory, hypothesis generating Can answer population-level questions Ecologic Studies: Limitations More subject to confounding than individual risk studies Confounding is the mixing of effects between the exposure of interest and other risk factors. Difficult to assess temporality Did exposure precede the outcome? Subject to the ecological fallacy Ecological Fallacy: Improper interpretation of results Individual associations attributed to group-level data

Cross-sectional study design Begin with a defined population Gather data on Exposure and disease Four groups are possible: 1. Exposed- have disease 2. Exposed- do not have disease 3. Not exposed- have diease 4. Not exposed- do not have disease Cross-Sectional Studies: Design "Snapshot": Looks at prevalence of disease and/or exposure at one time point Begins with a population base and collects individual-level data Does not choose participants based on disease or exposure status Does not follow individuals over time Persons are classified as: Diseased or nondiseased, and/or Exposed or unexposed Cross-Sectional Studies: Descriptive or Analytical May be either descriptive or analytical: Descriptive: to describe the health status of the population Analytical: to relate exposure to disease 1. Identify Target Population Usually you don't know exposure and disease status of individuals in population until you collect data. 2. 2. Data Collection Governmental surveys (e.g., NSFG, BRFSS) Other nationwide surveys (e.g., GSS) Other telephone surveys Random digit dialing (RDD) surveys Occupational settings Cross-Sectional Studies: Analysis Conceptualized as: Odds of exposure among diseased vs. nondiseased (exposure-odds ratio); or Odds of disease among exposed vs. unexposed (disease-odds ratio) Either yields the same OR, which is a prevalence odds ratio Cross-Sectional Studies: Advantages Relatively quick and cheap Generalizability (depending on data!) Do not have to know disease/exposure status ahead of time Can collect detailed data exposures, outcomes, and confounders Cross-Sectional Studies: Limitations Time sequence (temporality) between exposure and disease can be difficult to infer Potential for confounding Selection bias "Survival" bias Representativeness of volunteers Often rely on self-reported information

Case-Control Studies: Design Subjects are ascertained based on their disease status. Cases = Diseased Controls = Nondiseased Compare exposure history between cases and controls. When Is It Desirable to Conduct a Case-Control Study? When the disease is rare Example: Studying risk factors for brain cancer When exposure data are expensive or difficult to obtain When disease has long induction and latent period Example: Cancer, cardiovascular disease When little is known about the disease Example: Early studies of AIDS Case-Control Studies: Types Community based Incidence case-control study Prevalence case-control study Cohort based (nested case control) How do you chose cases of disease in a case-control study? 1. Choose Cases of Disease Identify newly diagnosed (incident) cases of disease Importance of case-definition Or, choose prevalent cases Sources: Death certificates, disease registries, hospital discharge records, insurance records, medical lab records, records of physicians/other medical providers, community recruitment 2. Selection of Controls Definition: A sample of the source population that gave rise to the cases Purpose: To estimate the exposure distribution in the source population that produced the cases and controls "Would" criterion: A member of the control group who gets the disease being studied "would" end up as a case in the study. Exposed and unexposed controls should have the same probability of selection. Sources of Controls: General Population Used mainly when cases are selected from a defined geographic population Sources: random digit dialing, residence lists, drivers' license records Advantages and Limitations of General Population Controls Advantages: Investigator is usually assured that they come from the same base population as the cases. Limitations: Hard to contact and get cooperation May remember exposures differently than cases (recall bias) Time consuming Expensive Sources of Controls: Hospital Used most often when cases are selected from a hospital population What are the advantages and limitations of using Hospital controls Advantages Same selection factors that led cases to hospital led controls to hospital Easily identifiable and accessible (less $ than pop-based controls) Accuracy of exposure recall comparable to that of cases since controls are also sick More willing to participate than pop-based controls Limitations Hospital catchment areas may differ for different diseases. Since hospital based controls are ill, they may not accurately represent the exposure history in the population that produced the cases. Hospital Controls: What Illnesses? What illnesses make good hospital controls? Illnesses that have no relation to the risk factor(s) under study. Selection of Controls: Other Issues May want to use more than one control group to avoid biases Desirable to control confounding factors Address confounding during design (matching) or during analysis (stratification, regression) Obtain Data on Exposure(s) and Potential Confounders Sources: interviews, mailed surveys, medical, employment, or other records Use comparable methods to obtain data from cases and controls

Nested Case-Control Study Design Performed after cohort study Choose all cases of disease among cohort Identify sample of controls among cohort without disease Analyze previously collected data on exposures and potential confounders Compare prevalence of exposure between cases and controls Nested Case Controls: Advantages Efficient Don't have to calculate exposures on entire cohort, just on cases and selected controls Controls come from same population as cases Minimizes control selection biases Example: Pesticides and Breast Cancer Hypothetical cohort study: Prior exposure to pesticides through self-reporting 89,949 women Longitudinal, 8-year study period 1,439 women develop breast cancer Pesticide/DDE levels analyzed for cases and sampled controls Characterized as "high exposure" or "low exposure"

Cross-sectional study design Begin with a defined population Gather data on Exposure and disease Four groups are possible: 1. Exposed- have disease 2. Exposed- do not have disease 3. Not exposed- have diease 4. Not exposed- do not have disease Cross-Sectional Studies: Design "Snapshot": Looks at prevalence of disease and/or exposure at one time point Begins with a population base and collects individual-level data Does not choose participants based on disease or exposure status Does not follow individuals over time Persons are classified as: Diseased or nondiseased, and/or Exposed or unexposed Cross-Sectional Studies: Descriptive or Analytical May be either descriptive or analytical: Descriptive: to describe the health status of the population Analytical: to relate exposure to disease 1. Identify Target Population Usually you don't know exposure and disease status of individuals in population until you collect data. 2. 2. Data Collection Governmental surveys (e.g., NSFG, BRFSS) Other nationwide surveys (e.g., GSS) Other telephone surveys Random digit dialing (RDD) surveys Occupational settings Cross-Sectional Studies: Analysis Conceptualized as: Odds of exposure among diseased vs. nondiseased (exposure-odds ratio); or Odds of disease among exposed vs. unexposed (disease-odds ratio) Either yields the same OR, which is a prevalence odds ratio Cross-Sectional Studies: Advantages Relatively quick and cheap Generalizability (depending on data!) Do not have to know disease/exposure status ahead of time Can collect detailed data exposures, outcomes, and confounders Cross-Sectional Studies: Limitations Time sequence (temporality) between exposure and disease can be difficult to infer Potential for confounding Selection bias "Survival" bias Representativeness of volunteers Often rely on self-reported information Case: were exposed, were not exposed - Have the disease Controls: Were exposed, were not exposed - Do not hae the disease Case-Control Studies: Design Subjects are ascertained based on their disease status. Cases = Diseased Controls = Nondiseased Compare exposure history between cases and controls. Case-Control Studies: Types Community based Incidence case-control study Prevalence case-control study Cohort based (nested case control) How do you chose cases of disease in a case-control study? 1. Choose Cases of Disease Identify newly diagnosed (incident) cases of disease Importance of case-definition Or, choose prevalent cases Sources: Death certificates, disease registries, hospital discharge records, insurance records, medical lab records, records of physicians/other medical providers, community recruitment 2. Selection of Controls Definition: A sample of the source population that gave rise to the cases Purpose: To estimate the exposure distribution in the source population that produced the cases and controls "Would" criterion: A member of the control group who gets the disease being studied "would" end up as a case in the study. Exposed and unexposed controls should have the same probability of selection. Sources of Controls: General Population Used mainly when cases are selected from a defined geographic population Sources: random digit dialing, residence lists, drivers' license records Advantages and Limitations of General Population Controls Advantages: Investigator is usually assured that they come from the same base population as the cases. Limitations: Hard to contact and get cooperation May remember exposures differently than cases (recall bias) Time consuming Expensive Sources of Controls: Hospital Used most often when cases are selected from a hospital population What are the advantages and limitations of using Hospital controls Advantages Same selection factors that led cases to hospital led controls to hospital Easily identifiable and accessible (less $ than pop-based controls) Accuracy of exposure recall comparable to that of cases since controls are also sick More willing to participate than pop-based controls Limitations Hospital catchment areas may differ for different diseases. Since hospital based controls are ill, they may not accurately represent the exposure history in the population that produced the cases. Hospital Controls: What Illnesses? What illnesses make good hospital controls? Illnesses that have no relation to the risk factor(s) under study. Selection of Controls: Other Issues May want to use more than one control group to avoid biases Desirable to control confounding factors Address confounding during design (matching) or during analysis (stratification, regression) Obtain Data on Exposure(s) and Potential Confounders Sources: interviews, mailed surveys, medical, employment, or other records Use comparable methods to obtain data from cases and controls

When Is It Desirable to Conduct a Case-Control Study? When the disease is rare Example: Studying risk factors for brain cancer When exposure data are expensive or difficult to obtain When disease has long induction and latent period Example: Cancer, cardiovascular disease When little is known about the disease Example: Early studies of AIDS

Nested Case-Control Study Design Performed after cohort study Choose all cases of disease among cohort Identify sample of controls among cohort without disease Analyze previously collected data on exposures and potential confounders Compare prevalence of exposure between cases and controls Nested Case Controls: Advantages Efficient Don't have to calculate exposures on entire cohort, just on cases and selected controls Controls come from same population as cases Minimizes control selection biases Example: Pesticides and Breast Cancer Hypothetical cohort study: Prior exposure to pesticides through self-reporting 89,949 women Longitudinal, 8-year study period 1,439 women develop breast cancer Pesticide/DDE levels analyzed for cases and sampled controls Characterized as "high exposure" or "low exposure"

COME BACK TO THIS

Blinding or Masking in Study Design Blinding is a technique used to minimize information/ascertainment bias. Single-blind study: Participants don't know whether they're in placebo or intervention group, but researchers do. Double-blind study: Neither the researcher nor subject knows the intervention assignment. Unmasked on completion of trial or if study is halted early. Placebos in Study Design A form of masking in treatment. A sham or fake treatment. Psychological impact is important for boosting perception that participant is receiving the intervention. Placebo control trials are widely used. May be unethical at times. May not be possible to mask the group with a placebo because of the type of intervention. Possible side effects or obvious differences between groups Compliance Compliance is adherence to intervention (e.g., taking prescribed pills). A large amount of noncompliance in the intervention group will make it similar to the comparison group. Noncompliance diminishes true differences in the study population. Eliminating Bias in Study Design and Conduct Goals Adequate sample size Masking/blinding High follow-up rate High compliance rate

Retrospective vs. Prospective Study Designs Retrospective studies Rely heavily on records Disadvantageous: past records must be complete and thorough Prospective studies Collect data in contemporaneous manner Reliance on past records unnecessary Follow-up methods are the same for both. Retrospective studies are challenging due to lapses of time since exposure. Timing of Cohort Studies: Retrospective Design - Starts at current date, looks backwards in time -Exposure and outcome have already occurred - Effective for diseases with long induction/latency period - Attempt to accurately measure exposure is very difficult Timing of Cohort Studies: Prospective Design More time-consuming and expensive Starts at current date, moves forward in time May have to follow up with population for years or decades Not efficient for diseases with long latency periods Better confounder data and estimate of exposure Less susceptible to selection bias An Ambidirectional Cohort Study A cohort study may also be ambidirectional , meaning that there are both retrospective and prospective phases of the study. Ambidirectional studies are much less common than purely prospective or retrospective studies, but they are conceptually consistent with and share elements of the advantages and disadvantages of both types of studies.