Term
| In relation to the control and experimental groups of an observational study, what contributes to confounding? |
|
Definition
An imbalance of prognostic factors
(different stages of disease, not representative of general population, exposure isn't randomized) |
|
|
Term
| Confounding leads you to under or overestimate the risk or effect of treatment in an observational study? |
|
Definition
Underestimate due to unmeasured factors contributing to the outcome. |
|
|
Term
| What is the core strength of randomized control trials when assessing risk, harm, and prognosis? |
|
Definition
- Because you get to choose the patients in each group (with or without intervention), you can balance prognostic factors in order to develop a cause and effect relationship. |
|
|
Term
Which of the following is NOT a limitation of an RCT for risk, harm and prognosis?
A. Outcomes take a long time
B. Expensive
C. Susceptible to recall bias
D. Requires large sample size
E. Can't randomize patients to harmful agents (unethical) |
|
Definition
C. Susceptible to recall bias
Recall bias is not present in an RTC. It is however present in a retrospective cohort study. |
|
|
Term
| Define referral filter bias: |
|
Definition
Referral filter bias occurs when patients are enrolled in the study because they are already in hospitals and are available. However, they may have other health problems that can creating confounding. |
|
|
Term
| What is the bias called when some of the patients in the study receive a greater amount and diversity of tests to detect a certain outcome? |
|
Definition
Surveillance bias - the patients don't have an equal likelihood of being checked for occurrence. |
|
|
Term
| What is the most common bias associated with prospective cohort studies? |
|
Definition
Recall bias - due to the fact that patients are recruited and monitored over time and their responses can be influence by their memory. |
|
|
Term
| Between Prospective Cohort, Retrospective Cohort, and Case-control Studies, which provides the most control over the data? |
|
Definition
Prospective Cohort studies- because you are acquiring the data rather than compiling it from the past. |
|
|
Term
| What is the analysis called that accounts for baseline differences (confounding) in a prospective cohort study? |
|
Definition
|
|
Term
| How does an adjustment analysis reduce confounding? |
|
Definition
Adjustment analyses compare subgroups in both the exposed and unexposed groups.
For example, if the exposure is smoking, an adjustment analysis would group smokers who exercise at least 4 days a week with nonsmokers who exercise at least 4 days a week and see if it reveals any changes in the data interpretation. |
|
|
Term
| What are the two primary limitations of a prospective cohort study? |
|
Definition
- outcomes take a long time
- patients of the exposed and control groups have unbalanced prognostic factors (confounding) |
|
|
Term
| An adjustment analysis has a correlation coefficient (r) which measures the strength of the relationship between two variables (dependent and independent). What is the range of possible r values and what do each end of the range represent? |
|
Definition
r = -1 to 1.
-1 ~ strongest possible negative relationship between the dependent and independent variable
1 ~ strongest possible positive relatinoship betwen the dependent and independent variable |
|
|
Term
| In relation to the correlation coefficient (r) what value is representative of the null hypothesis? |
|
Definition
|
|
Term
| Regression is used to predict who will get a disease based on the strength of the relationship between what two variables? |
|
Definition
Predictor variable (IE- age, obesity, smoking) and the dependent or target variable (diabetes) |
|
|
Term
| What type of regression is used for dichotomous data? |
|
Definition
|
|
Term
| What type of regression is used for continous data? |
|
Definition
|
|
Term
| Multivariate regression is the most useful in adjusting for what type of confounding? |
|
Definition
differences in baseline characteristics |
|
|
Term
| What is the formula for Relative Risk Ratio (RRR)? |
|
Definition
|
|
Term
| What are the odds for a 4/10 risk ratio? |
|
Definition
|
|
Term
| In a 10 patient study, the treated risk of death is 0.4 and the untreated odds of death is 1.33. What are the Risk and Odds Ratio for Death? |
|
Definition
Risk Ratio for Death
risk treated/risk untreated
0.4/ 0.6 = 0.67
Odds Ratio for Death
0.67/1.33 = 0.5 |
|
|
Term
| An Odds Ratio can be calculated using what formula? |
|
Definition
In a 2 x 2 square...
(a/b)/(c/d) or (a*d)/(c*b) |
|
|
Term
| When an event rate is low, what can we say about odds ratio and relative risk? |
|
Definition
They will approximate each other. |
|
|
Term
| When an odds ratio is around 1, what can we say about the risk ratio and odds ratio? |
|
Definition
They approximate each other. |
|
|
Term
| When an OR is equal to 1, what is the general effect? |
|
Definition
|
|
Term
| Does RR or OR show a greater effect on things like risk reduction? |
|
Definition
Odds Ratios make the effect size appear larger compared to relative risk. |
|
|
Term
| A survival curve with an initial steep descent and a subsequent leveling out would indicate what? |
|
Definition
Poor prognosis early, but a longer survival rate of those who survived the early stages of the disease. |
|
|
Term
| In a prospective cohort study with 100 patients, 6 patients die within the first 5 years. In addition, 11 patients are lost to follow up. What are the crude case fatality rate, worst case scenario, and best case scenario and is the loss to follow up significant? |
|
Definition
Crude Case Fatality Rate
= #died/#follow up = 6/89 = 6.7%
Worst Case Scenario - all lost to follow up died
(6 + 11)/100 = 17%
Best Case Scenario - all lost to follow up lived
6/100 = 6%
Due to the large # of patients lost relative to the number who have died, bias has occurred. |
|
|
Term
| What are the 4 major issues that can effect the validity of a cohort study? |
|
Definition
Confounding due to lack of balance
- fixed with adjustment analysis
Summary statistics
-odds ratios, relative risk, survival curves
Loss to follow up
-worst case/best case scenario
Objective and unbiased outcome measures
- use of blinding |
|
|
Term
Cohort study results can influence what various parts of clinical practice?
|
|
Definition
Therapy decisions
Good prognosis (management of care/lack of therapy)
Very poor prognosis (end of life care) |
|
|
Term
| When are clinical prediction rules the most useful? |
|
Definition
When decision making is complex, clinical stakes are high, or when you can save money without compromising care. |
|
|
Term
| What level of evidence is a clinical prediction rule when it is experiencing narrow validation? |
|
Definition
|
|
Term
| What level of evidence is a clinical prediction rule at when it is undergoing broad validation? |
|
Definition
|
|
Term
| What level of evidence is a clinical prediction rule at when it is being derived? |
|
Definition
|
|
Term
| The highest level of evidence (Level I) is obtained when a clinical prediction rule is undergoing what state of development? |
|
Definition
Impact analysis - rule makes a difference in physician behavior, improves patient outcome, and/or reduces cost. |
|
|
Term
| What is involved with deriving a clinical prediction rule? |
|
Definition
- develop a list of potential predictors
- examine a group of patients for the presence of the predictors and the status of their outcomes
- do a statistical analysis (multivariate) to determine which predictors are good
|
|
|
Term
| What is the primary difference between level III and level II clinical prediction rules? |
|
Definition
- Level II has undergone testing in different setting with patients that belong to different populations resulting in different prevalence of disease. |
|
|
Term
| Why is validation such an important part of developing a clinical prediction rule? |
|
Definition
A rule may be specific to one population group, patient type, or set of unique groups of predictor variables. Validation testing is the key to making succesful clinical prediction rules. |
|
|
Term
| A good clinical prediction rule has what characteristics in regards to setting, physicians, patients, and cost? |
|
Definition
-works in a wide variety of settings
-changes physician behavior
- improves patient outcomes
- reduces cost |
|
|
