Quiz Review Flashcards
What is the MAIN difference between observational and experimental studies?
A. Experimental studies are always prospective while observational studies are always retrospective.
B. In experimental studies, the investigators assign people to recieve an intervention/treatment, while in observational studies the intervention/treatment under study is not assigned by investigators.
C. Experimental studies include a control/comparator group, while obsesrvational studies do not.
D. In experimental studies participants interact with study investigators, but in observational studies study subjects do not have contact with investigators.
B
Which one of the following of the Bradford Hill criteria of causality is a required element to establish causation in a study?
Temporality
Statistical significance of association
Strength of the association
Consistency
Temporality
T/F Causal inference is made by measuring the association between cause and effect in a single study.
False
In clinical research studies, bias refers to…
when an investigators preconceived ideas influences the results reported from a study.
Any type of error that influences study measurements or results.
Any systematic error that influences study measurements or results.
Any chance or random error that influences study measurements or results.
Any systematic error that influences study measurements or results.
Which of the following BEST describes a confounder?
A confounder is another name for the exposure of interest.
A confounder is a third variable that, if not accounted for, can artifically inflate or attenuate the measure of association between an exposure and outcome.
A confounder is any factor that confuses the results of a study.
A confounder is another name for the outcome of interest.
A confounder is a third variable that, if not accounted for, can artifically inflate or attenuate the measure of association between an exposure and outcome.
In randomized controlled trials, the main benefit of randomization is…
to identify study subjects to enroll in a trial
minimize random error by randomly assigning subjects to treatment or control.
maximizing of the statistical power availble to detect a signifcant association.
control for known and unknown confounders by randomly assigning subjects to treatment or control.
control for known and unknown confounders by randomly assigning subjects to treatment or control.
Which of the following is NOT a method used to control for confounding in observational studies?
Multivariable regression
Restriction
Randomization
Matching
Randomization
T/F Case-Control studies can be either prospective or retrospective.
false
T/F In a randomized controlled trial, the benefits of randomization are generally only preserved in an intention to treat analysis.
true
A blinded randomized controlled trial is one in which….
the treatment under study is intended to improve vision of study subjects
study subjects do not know if they are in the experimental treatment or control arm of the trial.
study subjects are not consented until the end of the trial.
study subjects do not know what the researchers are studying
study subjects do not know if they are in the experimental treatment or control arm of the trial.
A study is conducted in the Veterans Affairs Healthcare System to identify the association between medication compliance and hospital admission among heart failure patients. The investigators use the electronic health record data to identify heart failure patients who were hospitalized in 2015, and for each of these patients they identify three heart failure patients who were not hospitalized. The investigators then review pharmacy data in the 90 days prior to identify their fill history for prescriptions as a proxy for medication compliance. This study can best be described as:
Prospective cohort
Case-control
Cross-sectional
Case-Series
Retrospective cohort
Case-control
Which of the following is NOT one of Bradford-Hill’s Criteria for Causality?
Temporality
Consistency
Statistical significance of association
Coherence
Experimental evidence
Statistical significance of association
A distinguishing feature between retrospective cohort studies and case-control studies is that:
For the case-control studies, exposed and unexposed persons are identified before outcome/disease status is identified; whereas for retrospective studies, investigators identify persons with and without the outcome before the exposure status is identified.
Case-control studies are descriptive as they include no comparator group.
For the retrospective cohort studies, exposed and unexposed persons are identified before outcome/disease status is identified; whereas for case-control studies, investigators identify persons with and without the outcome before the exposure status is identified.
One is retrospective while the other is prospective.
For the retrospective cohort studies, exposed and unexposed persons are identified before outcome/disease status is identified; whereas for case-control studies, investigators identify persons with and without the outcome before the exposure status is identified.
T/F A well-designed and conducted case-control study can provide stronger evidence for causal inference than a poorly designed and conducted cohort study.
True
T/F Nested case-control studies provide the same level of evidence to contribute towards causal inference as a traditional case-control study.
False
A case-control study was conducted to identify risk factors for tendon rupture among patients receiving quinolone antibiotics. Among the 50 patients identified with tendon rupture, 20 were identified as concomitantly using corticosteroids. Among the 150 patients with no tendon rupture, 43 patients were concomitantly using corticosteroids. Please select all true responses below:
The relative risk for the association between corticosteroid use and tendon rupture among quinolone users is 1.40.
The odds ratio for the association between corticosteroid use and tendon rupture among quinolone users is 1.66.
The relative risk for the association between corticosteroid use and tendon rupture among quinolone users cannot be calculated in this study.
The odds ratio for the association between corticosteroid use and tendon rupture among quinolone users cannot be calculated in this study.
The odds ratio for the association between corticosteroid use and tendon rupture among quinolone users is 1.66.
The relative risk for the association between corticosteroid use and tendon rupture among quinolone users cannot be calculated in this study.
A cross-sectional study telephone-based survey was conducted among residents of Oregon. Among the 2,000 respondents, 900 were female and 1,100 were male. When asked, 70 females and 101 males reported that they had been diagnosed with diabetes. Select all true statements below.
The relative risk for the association between male sex and diabetes is 1.18.
The overall prevalence of diabetes cannot be calculated in this study.
The overall prevalence of diabetes in this study is 8.6%.
The relative risk for the association between male sex and diabetes cannot be calculated in this study.
The overall prevalence of diabetes in this study is 8.6%.
The relative risk for the association between male sex and diabetes cannot be calculated in this study.
T/F The odds ratio is always an underestimate of the relative risk for a given association.
false
In a retrospective cohort study of newly diagnosed diabetics, the relative risk for diabetes among African Americans compared to non-Latino whites, was 3.0. Based on the data presented here, select all true statements below.
The incidence of diabetes in this study was three times greater among non-Latino whites than among African Americans.
There is no association between race and diabetes risk.
African American race is protective for diabetes compared to non-Latino white race.
African American race is a risk factor for diabetes compared to non-Latino white race.
African American race is a risk factor for diabetes compared to non-Latino white race.
The incidence of pertussis has experienced resurgence in recent years, particularly among infants and young children where the disease can be deadly. In the state of Washington in 2015, the risk difference for incident pertussis among children <1 year of age compared to 1-4 year olds was 0.0007. Based on these data, select all true statements below.
There is no association between age and risk of pertussis.
Age <1 year is protective for pertussis compared to age 1-4 years.
Age <1 year is a risk factor for pertussis compared to age 1-4 years.
The incidence of pertussis among children age <1 year old was 0.0007 times that of children 1-4 years old.
Age <1 year is a risk factor for pertussis compared to age 1-4 years.
Which one of the following is FALSE statement regarding confounders?
A confounder should not be in the causal pathway between exposure and disease.
A confounder is a variable that is causally related to the disease/outcome and also associated with the exposure.
Confounders most commonly result in artificial inflation of the measure of effect of interest.
Confounders can be identified through statistical tests.
The effect of confounders can be controlled for through analytical methods.
Confounders can be identified through statistical tests.
Which one of the following is a FALSE statement regarding effect modification?
Modifiers can be identified through statistical tests.
If the effect of the exposure on the disease/outcome truly varies based on the level or value of some third variable, then that third variable is an effect modifier.
Effect modifiers can be controlled for through analytical methods.
Effect modification cannot be controlled for; rather stratum-specific measures of effect should be reported.
Effect modifiers can be controlled for through analytical methods.
T/F Bias is the result of random error in the measurement of the exposure, outcome, or other variables.
false
T/F We can control for bias in a study by increasing the sample size.
false
A study is conducted of the risk of urinary tract infection (UTI) in pregnant women following birth of the child via cesarean compared to vaginal births. The unadjusted relative risk from the study is 1.3. The investigators stratify the data by whether or not the woman had a urinary catheter placed during hospitalization or not. The relative risk among catheterized patients was 1.2, while the relative risk among non-catheterized patients was 1.1. A statistical test indicated no significant difference between the stratified relative risks. Select all of the following that are true:
The data suggest that placement of a urinary catheter is an effect modifier for the association between cesarean births and UTI.
The statistical test discussed above helps us to test for effect modification.
The statistical test discussed above helps us to test for confounding.
Cesarean birth is associated with an increased risk of UTI after adjusting for catheterization.
The data suggest that placement of a urinary catheter is a confounder of the association between cesarean births and UTI.
The statistical test discussed above helps us to test for effect modification.
Cesarean birth is associated with an increased risk of UTI after adjusting for catheterization.
The data suggest that placement of a urinary catheter is a confounder of the association between cesarean births and UTI.
A case-control study is conducted to identify risk factors for non-fatal car accidents among drivers on rural roads of Indian reservations. Cases, survivors of car crashes on the rural roads of Indian reservations, are each matched to one driver residing in the same vicinity of the crash at that time frame. Cases and controls are surveyed by Indian health services regarding potential risk factors. Cases are asked about the driving conditions at the time of their accident, and controls are asked to recall a driving event within the same time frame and utilize that event to respond to questions. Both cases and controls are asked about weather, condition of their car, sobriety, and any distractions present inside the car or on the road. However, the investigators are concerned that controls will not be able to accurately remember the conditions surrounding their reference driving event. Select all of the following that are true.
If controls do not accurately recall potential risk factors, it could lead to misclassification of these variables.
If controls and cases have differential memory of potential exposures (i.e. risk factors), it could lead to loss to follow-up bias.
If recall bias is present in this study, it is likely to bias associations between the risk factors and non-fatal car accidents away from the null.
If recall bias is present in this study, it is likely to bias associations between the risk factors and non-fatal car accidents towards the null.
If controls and cases have differential memory of potential exposures (i.e. risk factors), it could lead to recall bias.
If controls do not accurately recall potential risk factors, it could lead to misclassification of these variables.
If recall bias is present in this study, it is likely to bias associations between the risk factors and non-fatal car accidents away from the null.
If controls and cases have differential memory of potential exposures (i.e. risk factors), it could lead to recall bias.
T/F If for a given association, the true relative risk is 0.8 and in a study the observed relative risk is 0.2, then this is an example of bias away from the null.
true
T/F If for a given association, the true relative risk is 1.3 and in a study the observed relative risk is 1.1, then this is an example of bias towards the null.
true
A prospective cohort study is conducted to determine if persons taking probiotics are at reduced risk of Clostridium difficile infection following use of antibiotics. Individuals are recruited to participate at a pharmacy; upon dispensing the antibiotic, the pharmacist informs them of the purpose of the study and consents them to participate. Study participants are given a brief baseline survey to assess if they use probiotics, and assess other potential confounders. They are then told that they will be called upon completion of their course of antibiotics and again 7 days later to determine if they experienced a C. difficile infection. Select all of following types of bias that are likely to affect this study?
Lead-time bias
Volunteer bias
Berkson’s bias
Loss to follow-up bias
Recall bias
Volunteer bias
Loss to follow-up bias
T/F Bias can be controlled for using both design and analysis based methods.
false
