Epidemiology - Quiz 1-2

drraythe's version from 2016-01-29 04:58

Bias, Sampling & Confounding

Question Answer
What is bias?“Any systematic error in an epidemiology study that results in an incorrect estimate of the association btwn exposure dz risk.”
What are the 2 major types of bias?Selection bias & Information bias (can be a combo)
What is selection bias?Distortion in way subjects are selected for study
What is Surveillance bias? Is this selection or information bias?Persons/animals followed more closely by health care providers bc of some exposure are more likely to be diagnosed as a case (this is a selection bias)
What is Non-response bias? Is this selection or information bias?Subject’s [owners] do not participate. In 1 group all individuals respond but only a few in the other (this is a selection bias)
What is an inappropriate comparison group? Is this selection or information bias?Comparison group (controls) does not appropriately represent the population from which cases arose (this is a selection bias)
What are some ways you could prevent selection bias in a study?You could randomly select subjects so that differences btwn groups are minimal, Ensure that response rates are high & Ensure that withdrawal rates are low
What is information bias?Systematic error in the collection of-exposure-or dz data (That results in a mistaken estimate of an exposure’s effect on the risk of dz.)
What is Interviewer bias? Is this selection or information bias?An interviewer interjects his or her bias into interview. Way questions are asked & perceived by respondents affects their response (this is an information bias)
What is Recall bias? Is this selection or information bias?Ability of respondents to accurately remember exposure of pet or use of individuals other than the owner to obtain info on Hx (this is an information bias)
Using data from (inaccurate/incomplete) vet records would be an example of what kinda bias?Information bias
What is Misclassification bias? Is this selection or information bias?When either exposure or dz outcome is misclassified (cases are misclassified as controls or controls as cases) based on inaccurate data collection. (This is an information bias)
In summary, the 3 types of surveillance bias are...Surveillance, Non-response & Inappropriate comparison groups
In summary, the 5 types of information bias are...Interviewer
Inaccurate medical records
Observer bias
Chance vs bias in relation to error?Chance is a RANDOM error, bias is a SYSTEMATIC error & is predictable
What is a sample?Subset of individuals in the defined population.
Why is sample size important?Sample size determination is important in ensuring that our estimate is close to & representative of the true population value
Best method of sampling where each individual in the population has an equal chance of being selected is...Random sampling
What is Simple random sampling?Each subject in a population has an equal chance of being selected. 1 can use a computer to do this.
What is Systematic random sampling?The selection of sampling has a known & equal predefined interval such as select every 4th animal in a herd
What is Stratified random sampling?When the samples are divided into strata & a random sample taken from each strata
What is Cluster sampling?Selection of samples in space & time
What is Convenience sampling?The most accessible samples are collected. (Non-random sampling technique)
What is Purposive sampling?The most desired samples are selected, such as animals known to have the dz of interest (non-random sampling technique)
What is Haphazard sampling?Sampling units are selected w/o any method in mind. (Non-random sampling technique)
What is Confounding?A problem posed in many epidemiological studies is that we observe a true association & are tempted to derive a causal inference when in fact the relationship is not causal... Confounding is the DISTORTION or the MASKING of an association btwn an exposure & a dz bc of a 3rd FACTOR (Potential confounding variables: Age, sex, breed)
What are some things you can do to control for confounding in the DESIGN stage of the study? (3)Randomization (clinical trials only)
What are some things you can do to control for confounding in the ANALYSIS stage of the study? (3)Restriction
Multivariate analysis
Matching as a way to reduce confounding only works when...The confounder is known (match ages if you know age might affect your result)
Restriction as a way to reduce confounding means you will do what? What is the risk?The study to a subgroup that is homogenous for the possible confounder (such as gender). But this has the risk of bias!
What is a temporal relationship?Communicates the simultaneity or ordering in time of events. The evidence provided demonstrates that your cz more than likely preceded the dz of interest
What is a Confounding variable/factor?An independent risk factor for Dz & associated w/ exposure [risk] factor but not a result of it.

Descriptive Studies

Question Answer
Internal validity could be summed up as 1 question, which is? Also, what are the factors to consider when thinking about internal validity?"Were things done properly?" think about bias, confounding, chance variation, Have the causal criteria been established (8) ← that list of Evans criteria of causation
External validity could be summed up as a question, like...Can the results be applied to populations other than that which was studied? (Are the results consistent w/ other evidence?)
Case reports & Case series...descriptive or analytical study?Descriptive
Cross-sectional study...descriptive or analytical study?Descriptive
Correlational (ecologic) study....descriptive or analytical study?Descriptive
Cohort (prospective or retrospective) study...descriptive or analytical study?Analytical
Case control study......descriptive or analytical study?Analytical
Randomized trials study....descriptive or analytical study?Analytical
Descriptive studies describe patterns of Dz according to... (3)Where, Who & When
What are some advantages to descriptive studies?Cheap, fast, easy to collect info....this study is often used when little is known about the dz
What is the goal of Descriptive studies?The goal of these studies are to describe the occurrence of dz in populations & generate hypotheses
What is the goal of analytical studies?Seek to id & explain the czs of dz & usually assign a numerical value to (quantifying) the effect of a potential risk (exposure) factor
What is a case report?Describes a single case
What is a case series?Report which describes a group of cases
What are the strengths/pros of a case report/series?Quick, cheap, may be 1st indication of an epidemic or newly recognized dz
What are the weaknesses/cons of a case report/series?There is no comparison to compare the results to, so you cant generalize the findings & is limited by small size.
What are Cross-sectional studies? Aka?Aka "prevalence" studies. Snapshot of health at any defined point in time or very short period of time... Description of attributes of a population based on a sampling/ SURVEY from the population.
A 2x2 table in epi measures the association btwn what 2 things?An exposure factor (+ or-) & the occurrence or absence of dz (dz status vs exposure status)
What are independent variables?The characteristic being observed or measured that is hypothesized to effect the occurrence of dz...Aka proposed causal factor
What are dependent variables?The manifestation of dz whose variation we seek to explain or account for by the influence of independent variables.... aka dz
LOOK AT 2x2 TABLE ON SLIDE 24!Understand it
What are the strengths & weaknesses of a Cross-sectional study?STRENGTHS: Quick, easy, cheap surveys (snap-shot from a sample) & you can calculate prevalence
WEAKNESSES: Can’t assess cz & effect, as looking at 1 point in time. May miss “exposures” that vary over time.
Ecologic studies are aka & aka? What does an ecologic study look at?Aka Correlational, aka Aggregate-level studies. Evaluate exposures & don a group rather than an individual level & a common approach is to make geographical correlations (such as the prevalence of Tuberculosis (TB) in UK herds).... Uses demographic data & reportable dz information on herds that are collected by State & federal agencies
Explain how an ecological fallacy is createdAn ECOLOGICAL FALLACY is committed if an assumption is made that the association found at the herd/group level is also true on the individual level (we want to avoid this)
What are the strengths & weaknesses of ecologic studies?STRENGTHS: Allows you to look at patterns & form hypotheses & is quick & cheap often using already available information. WEAKNESSES: Unable to link exposure to dz, nor control for confounding factors, nor test hypotheses. Ability to correlate does not imply a valid statistical association

Analytical Studies

Question Answer
What do analytical studies look at? What does it allow you to determine?Starts w/ data collection > leads to analysis & interpretation. Determine the strength, importance & statistical significance of epidemiological associations
Cohort study-observational or non-observational?Observational
Case-control study-observational or non-observational?Observational
Clinical trials-observational or non-observational?Non-observational (interventional studies-therapeutic/prevention)
What is a cohort?A cohort is a group of animals or persons who share a common experience w/in a defined time period. (Ex: birth cohort born in same year)
The best method in comparing Dz *incidence* btwn 2 cohort’s means you would...START w/ identifying “exposed” cohort & “non-exposed” cohort & then measure “dz” occurrence in each &compare (so follow up is from present to future)
What is the RR? When is it used?Relative Risk, or Risk Ratio. it is used when the following question is asked: How many more times more (or less) likely are exposed individuals to get the dz relative to un-exposed individuals?.... The dz is RR times more likely to occur among those exposed to the suspected risk factor than among those w/ no exposure.
Which value of RR is associated w/ exposure probably not being associated w/ the risk of Dz? What if exposure is-likely-to be associated w/ the risk of Dz?If exposure & risk of dz are not associated, the RR will be very close to 1. If exposure IS linked to risk of dz, the number is greater than or smaller than 1....and the farther away from one, the stronger the association
If RR is 1, less than 1, or greater than 1, what is the implication?1= no association btwn exposure & dz. If greater than one, it is a positive association, which means it is possibly causal. If the RR is less than one, it is a negative association & it is possibly a protective factor against the dz
What is the shortcoming of a RR (risk ratio) & how do we compensate for this?CANT KNOW IF RR's association is statistically significant. Use Confidence intervals (CIs) to determine that
If you have a 95% CI (confidence interval) for your RR (risk ratio), what does this mean?A 95% CI will incld the true RR for the population, 95% of the time
If the 95% CI inclds the null value of 1.0, what would you think about the RR?1= no difference, aka no association btwn exposure & dz. If this happens, then the RR could be due to chance & we say that the RR is NOT statistically significant at the 95% CI
When is a RR statistically sig? Insig?Insignificant: if the CI (confidence interval) contains the 1 value of the RR (1=no association btwn dz & exposure) then it is statistically insig. If the 95% CI falls on 1 or the other side of the 1 RR value, then it IS statistically sig.
If RR = 0.87 & 95% CI is 0.54 – the RR sig or not?There is a negative association (maybe protective) btwn the exposure & the dz that is not statistically significant at the 95% level
What are the strengths of a prospective cohort study?You can measure incidence, Can describe temporal relationships & if the cohort is a prospective study specifically, can collect data on possible confounders
What are the strengths of the retrospective cohort studies?Less time-consuming, easier & less costly that prospective as data already present. Advantage can be taken of computerized medical databases & kennel or breed club registries
What is the major strength that both pro & retrospective cohort studies have?Useful when the exposure (proposed risk factor) is RARE bc w/ a cohort, you already have a group which has experienced the exposure (ex: Chernobyl radiation). Also, Can examine the multiple likely outcomes/ dz of a single exposure
In general, what is the weakness of cohort studies?Inefficient for rare dz. would require follow up of cohorts for a long time
Weakness of a prospective cohort study?
THIS SET IS NOT FINISHED, SORRYThe quiz isn’t that bad. Know the basics for this section