Download Crunch-Time Manual: EPAID 7120 Epidemiologic Methods 2 Final Exam Prep. A Comprehensive and more Exams Public Health in PDF only on Docsity!
Epidemiologic Methods 2 Final
Exam Prep.
A Comprehensive Exam Study Guide
for a Guaranteed Top Score with
Grade A+
What are two common models of causation? - ANSmodel of sufficient and component causes and the counterfactual DAGs - ANSvisual presentation of causal assumptions Bradford Hill Criteria - ANSstrength, consistency, temporality, biological gradient, coherence, experiment, analogy, specificity, plausibility; not a checklist strength - ANSA small association does not mean that there is not a causal effect, though the larger the association, the more likely that it is causal. Consistency - ANSConsistent findings observed by different people in different places with different samples strengthens the likelihood of an effect. Temporality - ANSThe effect has to occur after the cause Biological gradient - ANSGreater exposure should generally lead to greater incidence of the effect. In other cases, greater exposure leads to lower incidence. Coherence - ANSCoherence between epidemiological and laboratory findings increases the likelihood of an effect but "... lack of such [laboratory] evidence cannot nullify the epidemiological effect on associations". Experiment - ANS"Occasionally it is possible to appeal to experimental evidence". Analogy - ANSThe effect of similar factors may be considered. Specificity - ANSCausation is likely if there is a very specific population, at a specific site, with a specific disease with no other likely explanation. Plausibility - ANSA plausible mechanism between cause and effect is helpful types of causal relationships - ANSdirect & indirect cause - ANSAn event, condition, or characteristic that preceded the outcome, and had the event, condition, or characteristic been different in some way, the outcome would not have occurred at all or not for some time later. sufficient cause - ANSA minimal set of conditions and events that are sufficient for the outcome to occur. (e.g. exposure by itself leads to an outcome or set of exposures together leads to the outcome) necessary cause - ANSis a particular type of component cause that is required for the outcome to occur. (remove exposure and do not get the outcome) counterfactual - ANSwhat you do not actually observe (e.g. Does Lipitor (cholesterol- reducing medicine) change the risk of heart disease? Iheart disease that would have been observed if these persons had not taken Lipitor = counterfactual (not actually observed). Directed Acyclic Graphs (DAGs) - ANSLike conceptual frameworks, DAGs are visual representations of "the world" These visual representations use a common set of "rules" to facilitate communication and collaboration among investigators One approach to translating causal pathways into diagram, that be can be a useful tool for epidemiologic studies of causal associations
Epidemiologic Methods 2 Final
Exam Prep.
A Comprehensive Exam Study Guide
for a Guaranteed Top Score with
Grade A+
are a set of arrows drawn along a timeline, characterizing [theorized] causal and temporal relationships between variables. There can never be a cycle (hence acyclic) because we can never go back in time." What are the rules of causal diagrams? - ANS1) no cycles
- nodes (measured and unmeasured variable)
- path (association)
- directed path (theorized causation) For diagram to be "causal" (i.e. allow determination of bias) must include ALL common causes of every variable on the diagram open pathway - ANSA - > B - > C (obesity to diabetes to cardiovascular disease) blocked pathway - ANSA->B<-C (obesity - >diabetes<-congenital hyperinsulinism) collider - ANSa node with two arrow heads pointing at it decendant - ANSE (diabetes) is a descendant of B (glycated hemoglobin A1c) mediators - ANSaffected by exposure AND on causal pathway between exposure & disease AND cause of disease; translate part of the effect of exposure on disease; we do not want to adjust for mediators confounders - ANSassociated with the exposure, AND associated with disease, AND not on the causal pathway; often want to adjust for confounders direct causal pathway - ANSe.g. obesity causes MI indirect causal paths - ANSconnect exposure and outcome of interest through other variable(s) (e.g. obesity causes MI through High BP) confounding - ANSdistortion of the estimated effect of an exposure on an outcome due to the presence of of a common cause of the exposure and the outcome (e.g. taking aspirin and risk of a heart attack (more elderly people take aspirin and more elderly people have a heart attack). What are the major problems for causal inference? - ANS1. Were there confounders?
- Were these confounders measured accurately?
- Was the analysis done appropriately?
- sampling and measurement processes adequate residual confounding - ANSconfounding that remains after inadequate control, e.g. because of measurement error in measuring the confounder, improper categorization or analysis Definitions for confounding - ANS1. "mixing of effects" leading to a change in estimate
- "classical" definition
- "counterfactual" model
- "collapsibility" definition mixing of effects - ANSthe effect of the exposure of interest is mixed together with the effect of another variable, leading to an incorrect effect estimate ("change in estimate") anticonservative confounding - ANSoverestimation of effect (away from the null effect) conservative confounding - ANSunderestimation of effect (towards null effect)
Epidemiologic Methods 2 Final
Exam Prep.
A Comprehensive Exam Study Guide
for a Guaranteed Top Score with
Grade A+
Very useful for study design Gave rise to powerful statistical methods (g-estimation of structural models, IPW of marginal structural models) Analysis of time-dependent confounding in longitudinal studies Pros and cons of Collapsibility definition - ANSGives us ways to statistically test Does not require prior knowledge Limited value for identifying confounders Helpful to examine how previously identified confounders "behave" in your data unmeasured confounding - ANSif a confounder has not been measured we can't easily adjust for it How are confounders identified? - ANSThe identification of confounders requires expert or substantive knowledge about the causal network of which exposure and outcome are part. Attempts to select confounders solely based on observed statistical associations may lead to bias. knowledge-based approaches to identify potential confounders - ANSpublic health or clinical subject knowledge; literature; conceptualization (e.g. DAGs) data-driven approaches - ANSassociations within own study and adjusted versus crude estimates (collapsibility) conditioning - ANSadjusting for a variable; In an open pathway, if you condition on a non- collider, you block the path. In a closed pathway, if you condition on a collider, you open the path. In a closed pathway, if you condition on the descendant of a collider, you open the path. data driven approaches within own study - ANScomparison at baseline (poor guide alone to decide exchangeability) and association with outcome What are the limitations of significance testing comparisons at baseline? - ANSA mix of effect and sample size Does not "prove" comparabilityCharacteristics may not cause the outcomeBivariate viewNot very informative with respect to outcome expectations of the groups What are the limitations of significance testing association of potential confounders with outcome? - ANSDoes not prove causality Does not show that a factor is a confounderDoes not consider the definition of confounding irrespective of being univariate or multivariateA mix of effect and sample size What are the pros and cons of knowledge-based approaches? - ANSConsider multi- dimensional nature of confoundingDecision separate from own dataBased on body of evidenceMinimizes risk of over- and under-adjustmentRequire judgment (subjective)Not easy to report/replicate What are the pros and cons of data-driven approaches? - ANSMay confirm associationsDo not require judgment if conditions pre-definedEasy to report/replicateMay not capture the definition for confoundingMay be based on a single studyStatistical significance testing alone is a poor guide to identify confoundingRisk of manipulation of analysis How do you control for confounding at the design stage? - ANSrestriction, matching, and randomization
Epidemiologic Methods 2 Final
Exam Prep.
A Comprehensive Exam Study Guide
for a Guaranteed Top Score with
Grade A+
How do you control for confounding at the analysis stage? - ANSstandardization, stratified analysis, regression and propensity score methods restriction - ANSSelection of subjects that have the same value (or nearly same value) of a variable that might be a confounder. Pros and cons of restriction - ANSpros - very powerful tool to control for confounding & built into the design of study cons - slows down recruitment, limits assessment of heterogeneity of effects across subgroups, may limit applicability of results beyond study, concerns about "representativeness" of study population for other populations matching - ANSSelection of controls so that they have the same value of a variable that is a potential confounder as the case does(i.e. "match" characteristics of a case, like age, to its control(s)) Randomization - ANSIs an explicit method that uses the play of chance to assign participants to comparison groups in a trial Exposure decision is removed from participant and health care provider Groups should end up similar on known and unknown prognostic factors at baseline avoids baseline confounding, but there is no guarantee exchangeability in terms of outcome expectations direct standardization - ANSSum of stratum-specific rates is weighted by the person-time distribution of a standard population; External age distribution (from the standard population) is applied to stratum-specific incidence rates in both populations. The resulting rates have the age-distribution of the standard population. Therefore, the magnitude/value of the age-adjusted rate has little "true" meaning (since it is based on a standard population that presumably isn't of direct interest). Instead, rates are mainly good for comparison. When would you want to show unadjusted values? - ANSnarrow description of group (e.g. narrow age group); need to understand what you want to achieve with data; if you are interested in age as a confounder look at age specific rates. If you want to look at trends with age look at the crude data. When do you use internal vs. external direct standardization? - ANScan use internal standardization if you are only interested in the trend, but generally use the external standardization. How can age standardized rates be interpreted? - ANSAge-standardized rates can be interpreted as the hypothetical rate of the disease that would have happened if the observed age-specific rates would occur in a population with the same age-distribution as the standard population What does the magnitude of the age-adjusted rate depend on? - ANSabsolute magnitude of the age-adjusted rate depends on the choice of the standard population
Epidemiologic Methods 2 Final
Exam Prep.
A Comprehensive Exam Study Guide
for a Guaranteed Top Score with
Grade A+
Standardized" (and other kinds of adjusted) measures are weighted averages, with weights chosen to improve comparability - ANStrue The observed ("crude") rate is also a weighted average of subgroup-specific rates, weighted by the size of the subgroups - ANSTrue - crude data are weighted by the person-time distribution Comparability of weighted averages depends on similarity of weights - ANSTrue Crude rates are "real", standardized rates are hypothetical - ANSTrue When is direct standardization commonly used? - ANSDirect standardization is commonly used to adjust for confounding when comparing vital statistics data across populations When is indirect standardization commonly used? - ANSare commonly used to adjust for confounding in occupational cohorts that compare the mortality of workers to that of the general population What is a limitation of indirect standardization? - ANSA limitation of indirect standardization is that standardized event rates cannot be easily compared across populations Stratified analysis - ANSEven if studies ultimately require more complicated analyses, stratified analysis is an important interim tool. It familiarizes the investigator with the distribution of key variables and patterns in the data in a way that other approaches cannot." stratum - ANSPartition of the sample in a way that is-Exhaustive-Mutually exclusive nEvery participant belongs to one stratum onlynSome examples for strata:-Age group-Family income group-Age group and family income group Steps in stratified analysis - ANSStart with crudetable and crudeeffect estimate Identify a confounding variable or an effect measure modifier Partition the sample according to this variable Calculate an effect estimate separately for each stratum Examine the crudeand the stratum-specificestimates If stratum-specific estimates are different: - ANSEMM-present stratum-specific estimates crude estimate differs from the stratum-specific estimates & stratum-specific estimates are similar: - ANSconfounding calculate a combined effect estimate (weighted average with 95% CIs and P value) Mantel-Haenszel method - ANSweighted average of stratum-specific odds ratios statistically adjusted effect estimate (adjusted for k confounders); gets rid of the problem of having to multiply or divide by a zero and now you just have to add properties of the Matel-Haenszel method - ANSBetter statistical properties than some other methods that apply weights (e.g. generalized inverse variance method) Works well with finely stratified data Works if there are 0s in the stratum specific tables Assumes that the OR is constant across strata (need to check for heterogeneity of ORs) May be applied to other measures of association (not only to the OR, but also IRR etc.) pros and cons of stratified analysis - ANSGood way to get to know the data Adjusts for confounding and may be used for the initial evaluation of effect measure modification (EMM)
Epidemiologic Methods 2 Final
Exam Prep.
A Comprehensive Exam Study Guide
for a Guaranteed Top Score with
Grade A+
Your adjusted estimate will lie between your stratum-specific estimates Difficult to accommodate many confounders or more than one effect measure modifier simultaneously Can not accommodate continuous variables (need to group Anticonservative confounding - ANSOverestimation of effect (away from null effect); stronger association than they would be if we accounted for confounding Conservative confounding - ANSUnderestimation of effect (towards null effect); weaker than they would be if accounted for confounding Qualitative confounding - ANSInference changes qualitatively(crosses null effect or reaches the null) Positive confounding - ANSfor ratios >1: overestimate for ratios <1: underestimate negative confounding - ANSfor ratios >1: underestimate for ratios <1: overestimate What is Effect Measure Modification? - ANSWhere ≥2 risk factors modify the effect of each other with regard to the occurrence or level of outcome Heterogeneity of effects - does the effect of the exposure on the outcome differ across strata?oCausal interaction - does being exposed to two factors increase the risk of outcome more than expected if both factors were acting individually? How to evaluate EMM - ANSThe effect of an exposure differs by categories of another "modifying" variableTo see if there is EMM, evaluate if there is homogeneity or heterogeneity of effects 1.Identify exposure, outcome, and potential EMM 2.Calculate the effect (OR, RR, PR etc.) in each strata of potential EMM oHomogeneous or Heterogeneous3.Formally test 4.Decide how to present and interpret results Immigrative selection bias by diagnostic bias - ANSPhysicians aware of possible associations between exposure and disease may follow exposed persons more closely and detect cases earlier Can evaluate EMM - ANSoIn anystudy designoUsing any risk measure Can account for EMM in - ANSStudy design: plan for stratified enrollment & analysisoAnalysis: usually explore EMM hereoCritical to check even if don't expect Quantitative EMM - ANSEach of the stratified effect measures suggest increased risk (or both suggest decreased risk) but of different magnitudes Qualitative EMM - ANSOne suggests increased risk the other suggests decreased risk Additive EMM - ANSThe joint causal effect of two variables is different from what would be expected by "adding" the independent effects of each exposure on the chosen measurement scale Scale for EMM - ANSScale for EMM Scale used depends on effect metric oRatio (multiplicative measure) multiplicative scale oDifference (additive measure) additive scale
Epidemiologic Methods 2 Final
Exam Prep.
A Comprehensive Exam Study Guide
for a Guaranteed Top Score with
Grade A+
Bias in epidemiology - ANSSystematic (as opposed to random) deviation of results or inferences from the truth Selection bias - ANSBiased estimate of an exposure-outcome association resulting from selection of study participants as an effect of the exposure and the outcome (or causes of the exposure and the outcome).exposure-outcome association is conditioned on study participation (depends on how participants are selected intoor stayin a study)often results in non-comparability of populations Selection bias - result - ANSSelection bias - result Exposure-outcome association is different for those who participate in the study and for those who are theoretically eligible for the study Selection bias in case-control studies - ANSOccurs when study participants are selected based on exposure status case-control studies are particularly prone to selection bias because the outcome is already known so it can be easy to select by exposure status, you need to select participants independent from the outcome!! Emigrative selection bias - ANSSelection out of the study population; Difficult to control for in the analysis Immigrative selection bias - ANSSelection into the study population; Very difficult to control for in the analysis because you need to know something about your source population and how it is different from study population, but we don't study the source population so it is very difficult to correct this bias at the analytic stage. Various mechanisms can lead to immigrative selection bias - ANSAscertainment biasParticipation bias- Diagnostic bias- Berkson's bias- Survival bias- Self-selection- Non- response- Healthy worker effect Various mechanisms can lead to emigrative selection bias - ANSDifferential losses to follow- up- "Depletion of susceptibles"- Competing risks- Participant not trackable- Withdrawal (e.g. adverse effects)- Protocol deviation (e.g. per protocol analysis in RCTs) Immigrative selection bias by diagnostic bias - ANSPhysicians aware of possible associations between exposureand disease may follow exposed persons more closely and detectcases earlier Berkson's bias - ANSselection bias that may occur in hospital-based case-control studies when the combination of exposure and disease increases the risk of admission to the hospital; Result: Systematically higher exposure odds among hospital cases than among hospital controls Immigrative selection bias by survival bias - ANShappens we we are studying very aggressive diseases; people we sample with those diseases are more likely to be slow progressors; e.g. for chronic disease we are more likely to sample people who have a milder disease.
Epidemiologic Methods 2 Final
Exam Prep.
A Comprehensive Exam Study Guide
for a Guaranteed Top Score with
Grade A+
healthy worker effect - ANSStudy of association of occupational exposures by comparing the outcomes of workers and the general population People who are employed are, on average, healthier than those unemployed Hazard by occupational exposures may be masked by the generally better health status in the working as compared with the general population competing risks - ANSA competing risk is an event whose occurrence either precludes the occurrence of another event under examination or fundamentally alters the probability of occurrence of that event; When the frequency of previously common causes of death is reduced, competing risks must become more prominent other diseases that had been rare became more prevalent because people had previously died from respiratory infections. drop-out - ANSParticipant not trackable (e.g. moved)- Participant did not adhere to treatment protocol (in trials)- Participant did not complete follow-up assessments (loss to follow-up)- Participant withdrew informed consent options for limiting missing data in trial design - ANSelect a target population that is not adequately served by current treatments - increases adherence Run-in period (active treatment) for all study participants - only those who tolerated treatment and adhered undergo randomization Shorten the follow-up period for the primary outcome For long-term efficacy trials: Consider a randomized withdrawal design (only patients on active medication who did not drop out are randomized to receive further active treatment or placebo) Options for limiting missing data: Trial conduct - ANSSet acceptable target rates for missing data and monitor the trial for those targets closely Provide incentives for complete data for investigators and participants as long as they meet ethical standards Provide continuous access to effective treatments after trial completion and before treatment approval Collect information from participants regarding the likelihood of dropping out and use this information to reduce drop-outs Minimize immigrative selection bias - ANSIs predominantly a problem of observational studies, in particular case-control studies RCTs aim to control for selection bias by concealment of the random allocation list see lectures 16 & 17 (clinical trials) Select cases and controls independent of exposure status!Cases/controls in the study population should have same odds of exposure as cases/controls in source/target population see lectures 18 & 19 (case-control studies) Hard to correct if immigrative selection bias occurs
Epidemiologic Methods 2 Final
Exam Prep.
A Comprehensive Exam Study Guide
for a Guaranteed Top Score with
Grade A+
bias in studies of congenital malformations(for example: cases are more likely to remember the intake of harmful substances than controls) Example: differential measurement error (cohort study) - ANSResearch question: Incidence of emphysema among smokers and non-smokers?→ If smokers seek more medical attention than non-smokers (worried, bronchitis, ...), emphysema may be diagnosed more frequently in smokers as compared with non-smokers→ Effect of smoking on emphysema would be overestimatedResult: Different accuracy of outcome measurement in exposed and unexposedHow to improve measurement: comparable follow-up Independent error - ANSMeasurement error in the variable in question (e.g. the outcome) is not associated with the errors in measuringother variables(e.g. the exposure) Dependent error - ANSMeasurement error in the variable in question (e.g. the outcome) isassociated with the errors in measuringother variables(e.g. both exposure and outcome derived from same questionnaire) Quantification of measurement error Dichotomous variables - ANSSensitivity, specificity Kappa statistic Quantification of measurement error Categorical variables - ANSSpearman correlation coefficient Kappa statistic Quantification of measurement error Continuous variables - ANSCoefficient of variation Intraclass correlation coefficient (reliability coefficient) Non-differential, independent measurement error of exposure - ANSEffect estimate is biased towards the null ("attenuated") if the test is informative What factors affect the extent of bias? - ANSThe measure of association usedThe prevalence of the exposureSensitivity and specificity (and the specific pair)The magnitude of the true effectThe risk of the disease Non-differential, independent measurement error of disease - ANSAttenuates estimate of association if sensitivity is <100% Attenuates estimate of association if specificity is <100% usually biases the association towards the null Differential, independent measurement error of disease - ANSBias can go in any direction The magnitude can be substantial, even with relatively good sensitivities and/or specificitiesIn cohort studies, an important concern is differential classification of disease as a function of exposure an lead to an over- or underestimation of an association regression fallacy - ANSI suspect that the regression fallacy is the mostcommon fallacy in the statistical analysis of economic data. regression towards the mean - individual level - ANSis the phenomenon that if a (continuous) variable is extreme on its first measurement, it will tend to be closer to the average on a second (or subsequent) measurement. Regression towards the mean causes - ANSIntra-individual variabilityRandom measurement error
Epidemiologic Methods 2 Final
Exam Prep.
A Comprehensive Exam Study Guide
for a Guaranteed Top Score with
Grade A+
Prevention and correction of regression to the mean (study design) - ANSDo not rely on a single measurement at the extremes of distributions Repeated measurementsRandomized trial Accounts for regression to the mean (similar extent across groups) Prevention and correction of regression to the mean (analysis) - ANSCompare change in patients at tail of baseline distributions (e.g. lowest 20%) to those not at tailAnalysis of covariance 753 Approaches to minimize information bias (study planning) - ANSExplicit definition of what to measure-Data needed for measurement-Different data sources-Synthesis of data for the measurement-Selection of questionnaires-Pilot testing-Certification of personnel Approaches to minimize information bias (study conduct) - ANSData monitoring and audits- Re-Certification of personnel Approaches to minimize information bias (data analysis) - ANSCorrection of measurement error challenges with multiple data sources - ANSHow to synthesize information?• How to handle conflicting information obtained from different sources?→ pre-specify how exactly you are addressing these issues in your protocol→ adjudication local adjudication - ANSReview of documents by (local) health care professional(s) central adjudication - ANSReview of documents by adjudication committee; Considered a "gold standard"Often seen in large trialsCan (should?) also be done in observational studiesExperts decide independently (in view of the complete documentation) whether an event happened or notDifferences are discussed in meetings (mechanism of consensus must be pre-specified) pros and cons of local adjudication - ANSFeasibilitySufficient if little conflicting data, few data sources, and clear guidance on how to adjudicate Not done identically for all patients Systematic differences across study sites possible pros and cons of central adjudication - ANSUniform classificationAdequate if events require judgment (e.g. cause-specific mortality, COPD exacerbations)May reduce sample size requirements Expensive Critical questions: selection of a questionnaire - ANSWhat is the study population? What should be measured? What's the purpose of the instrument? Observe change in persons over time? - ANSEvaluative questionnaire Observe differences across participants/ populations? - ANSDiscriminative questionnaire Evaluative instruments - ANSSpecific to the participants' problems (be cautious with 'generic' questionnaires)Address problems than can change over timeOffer at least 5 answer options (e.g. extremely poor to excellent)Have high reproducibility (ICC close to 1)
Epidemiologic Methods 2 Final
Exam Prep.
A Comprehensive Exam Study Guide
for a Guaranteed Top Score with
Grade A+
- knowledge of subject matter (application of correct dose of drugs in a clinical trial, choice of window of exposure in a longitudinal study, planning of intervals of measurement, planning of procedures for outcome ascertainment) What are typical barriers to internal validity? - ANS-Vague definition of the target population
- Inability to define the source population
- Problems with the process for obtaining the study population What does external validity depend on? - ANS-knowledge of methods (composition of study population-inclusion and exclusion criteria)
- knowledge of subject matter (Are circumstances of exposure (e.g., duration, route, level, etc.) similar?-Are susceptibility factors similar?-Are other related factors similar?-Would the same results be expected: for children? across disease stages? for other countries with different health systems? Is there a statistical test for external validity? - ANSThere is no statistical test or automated method to ensure that inferences are applicable to other populations-Ultimately a matter of informed judgment For an association to be generalizable to a different target population - ANSthe exposure- outcome association in the study population should be the same as in this different target population. Just comparing population characteristics - ANSis not enough to judge external validity (generalizability). If there is reason to believe that an association is different in a different population - ANSwe need to study that different population before generalizing. What is expected of source and study population exposure-disease associations? - ANSReminder: Sourceand study populations are those expected to have the same exposure- disease association (or outcome expectation) as the target population What does not have to be true for the source population? - ANSHowever, the source population does not have to be representativeof the target population When is representativeness necessary? - ANS-Descriptive (survey) research
- Describe the occurrence of factors or disease in the population in a given place at a defined point or period in time•Burden A study that is free from bias has - ANSinternal validity Internal validity means - ANShat inferences/associations are the same across the study, source, and target population External validity - ANSis often difficult to determine. It describes how inferences can be applied to other populations Descriptive epidemiology - ANSis always observational, never experimental; observational studies may be descriptive or analytic.
- Important for monitoring disease and risk factor burden.
- Help direct funding for research and public health intervention.
- Inform how well public health/clinical strategies are working.
- Identify new health problems as they emerge.
Epidemiologic Methods 2 Final
Exam Prep.
A Comprehensive Exam Study Guide
for a Guaranteed Top Score with
Grade A+
Analytic Study - ANS-A study conceived to examine hypothesized causal relationships and to make causal inferences.
- An analytic study is usually concerned with identifying or measuring the effects of risk factors or with the health effects of specific exposures or interventions.
- Contrast with descriptive study, which does not test causal hypotheses. What are observational studies useful for? - ANSidentifying risk factors for health outcomes.
- Etiology studies
- Translational studies When conducting research on risk factors with the goal of translating to a future intervention, it is useful to - ANSframe the exposure and outcome in a way that can be translated to an RCT design. when designing an RCT based on observational findings, it is important to - ANSlook closely at what comparisons the observational study actually made. Cohort Study: - ANSAnalytic epidemiological study in which subsets of a defined population can be identified who are, have been, or in the future may be exposed or not exposed - or exposed in different degrees - to a factor or factors hypothesized to influence the occurrence of a given outcome. A common feature of a cohort study - ANSis comparison of incidences in groups that differ in exposure levels. The denominators used for analysis may be persons or person-time. Cohort Study: Populations - ANSCohort is a group of people who have a common experience or grouping (e.g. age cohort (birth cohort), risk cohort (cohort of smokers)), general population, clinic based open cohort - ANSAll members are not followed for the entire time. closed cohort - ANSoften theoretical; Everyone enters at the same time. dynamic cohort - ANSNew members are joining over follow-up. Prospective cohort - ANSthe protocol and cohort are established before people are followed - i.e., all events occur after the cohort is established retrospective cohort - ANSthe protocol and cohort are assembled after events have already occurred. Follow-up of cohort starts in the past and events are determined from past information fixed cohort - ANScannot gain members after a defined time (or event) Is a cohort the right design? - ANSoEvaluate an exposure with many outcomes oExposure is rare oDoes not work well for rare outcomes Is establishing a cohort feasible? - ANSoExpensive oCan take a long time to conduct Establishing a Cohort : Study Population - ANSWho will be in your cohort, what is the target population, define eligibility criteria, determine the number of individuals needed to answer the study question (power) Convenience sample - ANSEnroll all individuals who meet eligibility criteria; Non probabilistic; potentially may not reflect your target population
Epidemiologic Methods 2 Final
Exam Prep.
A Comprehensive Exam Study Guide
for a Guaranteed Top Score with
Grade A+
Wasted exposure - ANS"Wasted" exposure is exposure time that does not biologically contribute to the outcome. Including wasted exposure in exposure metrics dilutes the exposure effects lagged exposure - ANSDon't want to include the exposure that occurred with the disease dx. I.e. don't care as much about smoking right before dx. outcomes in cohort studies - ANScan be discrete (non-repeatable (birth) or repeatable(MI)) or biomarker level or change passive follow-up - ANSlinkage to external systems (requires unique identification information: full name, birth date, SSN)
- Registry
- Governmental records (birth certificate, National Death Index, Social Security)
- Medical or occupational records active follow-up - ANS-interaction with participants/ proxies
- Interviews
- Examinations
- Diaries
- Monitoring device
- Biomarker measurements Risk set - ANSAt each time of outcome of interest, we can identify other cohort members who are still at risk for the outcome. Immortal Person Time Bias in a Cohort Study - ANS-Typically arises when cohort entry criteria is dependent on survival.
- A form of bias in which persons are credited for time at risk during which the effect cannot possibly occur; therefore these participants have an artificially lower event rate.
- This can occur across an entire cohort or differentially by exposure status (e.g. if inlcusion criteria is that a person must have worked somewhere for 5 years, you don't include those 5 years because they are part of the inclusion criteria) Ghost Time in a Cohort Stud - ANSWhen there is imperfect mortality ascertainment in aging cohorts, participants continue to contribute person-time to the cohort dataset after their death. This becomes an issue for aging cohorts, when mortality is particularly high. Limitations of a cohort - ANS-Can be expensive
- Especially if large size or long follow-up
- Often less expensive than RCT, initially
- Not efficient when there is
- Long disease process/time to outcome
- Rare outcome
- For rare exposure - measuring exposure on everyone in cohort may not be warranted
Epidemiologic Methods 2 Final
Exam Prep.
A Comprehensive Exam Study Guide
for a Guaranteed Top Score with
Grade A+
Strengths of a cohort study - ANS-No temporal ambiguity
- Direct measure of risk due to an exposure
- Standardized collection of exposure & outcome
- Serial exposure measurements
- time-varying effects
- Examine multiple outcomes
- Can describe: natural history, pathogenesis, latency or incubation period, survival clinical trial - ANSplanned experiment designed to assess the efficacy of a treatment in man by comparing the outcomes in a group of patients treated with the test treatment with those observed in a comparable group of patients receiving a control treatment, where participants in both groups are enrolled, treated and followed over the same time period." Clinical equipoise - ANSpre-requisite for a clinical trial; can only randomize if we are uncertain which treatment is better. exists when there is genuine uncertainty within the professional community as to which of the two treatment arms is superior." randomization - ANS-Is an explicit method that uses the play of chance to assign participants to comparison groups in a trial.
- Is a process where treatment assignment follows no describable deterministic pattern, but follows a probabilistic distribution. What is the effect of randomization? - ANS1) Exposure decision is removed from participant and health care provider(Avoids confounding by indication)
- Groups should end up similar with respect to known and unknown prognostic factors(should avoid baseline confounding, but there is no "guarantee")
- Exchangeability of response distributions under identical exposures
- Randomization comes closest to simulating the counterfactual Does randomization guarantee groups will be comparable? - ANSnot always the case, it is extremely likely that, especially small groups, may not be comparable. If groups differ by baseline characteristics, does that mean randomization did not work? - ANSif we do randomization right it always works but they can still differ at baseline (its random, but that doesn't mean that we will like the outcome). The options to control for confounding in a trial - ANS1. restriction
- adaptive randomizatin
- adjustment for baseline variables (at analysis stage) Concealment of random allocation - ANSProcess for preventing disclosure of (random)treatment assignments to study participants and to those who request or administer the assignment.
- prevents selection bias
- Keeps the randomization sequence separate from trial enrollment and unknown to the enrollers
