Term
|
Definition
| A grouping of different data into similar categories; it usually involves randomization at the level of groups and not individuals. |
|
|
Term
|
Definition
| The time from an exposure to disease outcome. This is not a type of bias, but instead shows how long you must have an exposure or intervention before one appreciates an outcome. The latent period is especially important in chronic disease epidemiology. |
|
|
Term
| When designing a study, there are 3 methods of controlling confounders: |
|
Definition
| matching, restriction, and randomization |
|
|
Term
| When analyzing a set of data, there are 2 methods for controlling confounders: |
|
Definition
| stratification and modeling. |
|
|
Term
| The correlation coefficient: |
|
Definition
| A number from -1 to 1 that assesses the linear relationship between two variables. 0 is the null value, and means that there is no association. A positive value indicates that there is a positive correlation b/w two variables (ie, as one increases so does the other), while a negative value indicates the opposite. The closer the value is to its margins, the stronger the association. |
|
|
Term
|
Definition
| when the effect of a main exposure on an outcome is modified by another variable. This is not a bias. |
|
|
Term
|
Definition
| This is a cousin to the odds-ratio, however, while the odds-ratio looks at the odds of getting a disease if you have an exposure versus the odds of getting a disease if you don't have an exposure, the hazard ratio looks at the odds of getting an adverse outcome if you have an intervention for your disease versus the odds of an outcome if you have no intervention for your disease. |
|
|
Term
|
Definition
| A subgroup of selection bias that occurs when the treatment regimen selected for a patient depends on the severity of the patient's condition. This bias negates the benefit of randomization. |
|
|
Term
| How is randomization achieved and why do we do it? |
|
Definition
| Randomization is to eliminate bias from treatment assignments. We do this by striving for: equal patient group sizes, low selection bias, and a low probability of confounding variables. Confounding variables can be explored though baseline characteristics of patients. If patients have very similar baseline characteristics in similar randomization groups, it is hard to rule out confounding variables. |
|
|
Term
|
Definition
| similar results on repeat measurements. |
|
|
Term
|
Definition
| 2 or more experimental interventions, each with 2 or more variables that are studied independently. |
|
|
Term
| Observer's and ascertainment bias |
|
Definition
| bias due to misclassification of the outcome due to flaws in the study design |
|
|
Term
| Attributable risk percent: |
|
Definition
| The excess risk in a population that can be explained by exposure to a particular risk factor. |
|
|
Term
|
Definition
| Most easily derrived from the relative risk: ARP = (RR-1)/RR |
|
|
Term
|
Definition
| a method of selecting controls so that youevenly distribute confounding variables in a study population. E.g., picking controls from the same neighborhood as your subjects. |
|
|
Term
| Generalizability (or external validity): |
|
Definition
| The applicability of the obtained results beyond the cohort that was studied. It answers the question, "how generalizable are the resuls of a study to another population?" For example, if you included only middle-aged women in your study, you would not be able to apply the results to middle-aged men. |
|
|
Term
|
Definition
| This test is used to compare the proportions of a categorized outcome. I.e., if we had 102 that were 'low' and 45 that were 'high,' chi squared will compare those proportions. A 2x2 table may be used to compare the observed value with the expected value. |
|
|
Term
|
Definition
| compares 3 or more variables |
|
|
Term
| In a bell curve, what percentage of the results are within 1 SD of the mean? |
|
Definition
|
|
Term
|
Definition
|
|
Term
|
Definition
|
|
Term
|
Definition
|
|
Term
|
Definition
|
|
Term
|
Definition
When there is an incorrect assumption or conclusion of prolonged apparent survival and better prognosis due to a screening test.
Whenever you see a screening test and improved outcome, beware of this bias. Early detection may seem like it improves outcome, when instead it diagnoses diseas earlier. It does not change the prognosis of a disease at any given stage. |
|
|
Term
|
Definition
| A study where investigators have a RISK FACTOR and are looking for downstream effects from that risk factor. (Analyzed with relative risk.) This study may be prospective or retrospective. |
|
|
Term
|
Definition
| When investigators have a DISEASE, and they are looking for exposures in diseased versus undiseased people that happened in the past. (Odds ratio.) |
|
|
Term
|
Definition
| This test compares the means of 2 variables, but populations are studied instead of sample variance. (Limited applicability.) |
|
|
Term
|
Definition
| a measure of the incidence of a disease. Risk shows the probability of getting a disease over a certain time period. R = diseased subjects/subjects at risk in a population. (This will used one side of your 2x2 table.) |
|
|
Term
|
Definition
the risk of developing a disease in an exposed group divided by the risk of developing a disease in an unexposed group.
A RR greater than 1 indicates a positive association between the studied risk factor and the studied outcome. A RR less than one shows a negative association. |
|
|
Term
| When can the odds ratio approximate relative risk? |
|
Definition
| When a diseas is not very prevalent in the population. |
|
|
Term
| Criteria for confounding variables: |
|
Definition
(a) the confounding variable is associated with the independent variable
(b) the confounding variable is a risk factor for the outcome of interest. |
|
|
