Week 6

Question 1

what is the ranking of the hierarchy of study designs based on their contribution to causal inference from least to most?

Answer 1

case series -> cross-sectional -> case-control -> cohort -> RCT

Question 2

which forms of bias should be considered in a RCT?

Answer 2

selection bias, performance bias, attrition bias, and detection bias

Question 3

how do we control for selection bias?

Answer 3

randomized selection
allocation concealing

Question 4

how do we control for performance bias?

Answer 4

blinding participants

Question 5

how do we control for detection bias?

Answer 5

blinding outcome assessors

Question 6

how do we control for attrition bias?

Answer 6

how was missing data dealt with? hard to control for, but should be considered when interpreting data

Question 7

how do we control for reporting bias?

Answer 7

determine the factors that could result in reporting bias. reflecting on how under or over reporting may occur

Question 8

identify the 2 ways standardized reporting was mandated

Answer 8

The Uniform Requirements for Manuscripts Submitted to Biomedical Journals
Consolidated Standards of Reporting Trials (CONSORT)

Question 9

what are the downsides to simple randomization?

Answer 9

it is less efficient and there’s potential for group imbalance in small trials

Question 10

what factors are considered inappropriate ways to randomize participants?

Answer 10

-test result
-alternating presentation
-case record number
-date of presentation
-age or birthday

Question 11

what is stratified randomization?

Answer 11

study participants are stratified by prognostic factors (gender, age, etc) to reduce group imbalance in smaller trials

Question 12

what is block randomization?

Answer 12

population is split up to account for the effects of a particular variable
ex. group 1 are smokers, they are evenly distributed into treatment group and control group. group 2 are nonsmokers and they are also evenly distributed into treatment group and control group

Question 13

define what minimization means in a study

Answer 13

a complex computer-based method of allocating subjects into balanced groups based on key characteristics

Question 14

T/F allocation concealment is different than blinding

Answer 14

true

Question 15

when does blinding of a study occur? what bias does it prevent?

Answer 15

after subject allocation has been completed and is intended to occur throughout the entire duration of the study to prevent performance and detection bias

Question 16

when does allocation concealment of a study occur? what bias does it prevent?

Answer 16

during allocation to prevent selection bias

Question 17

T/F allocation concealment is always achievable

Answer 17

true

Question 18

T/F blinding is not always achievable

Answer 18

true

Question 19

what form of bias does triple-blinding help prevent?

Answer 19

reporting bias

Question 20

define selection bias
when can it occur during a trial?

Answer 20

Systematic differences between baseline characteristics in the groups being compared
during allocation

Question 21

define performance bias
when can it occur during a trial?

Answer 21

Systematic differences in the care provided to study participants apart from the intervention
during treatment intervention

Question 22

define attrition bias
when can it occur during a trial?
how can it occur?

Answer 22

systematic differences in withdrawals from trial
during followup
drop-out participants or missing data (loss to followup)

Question 23

define detection bias
when can it occur during a trial?
how can it occur?

Answer 23

systematic differences in outcome assessment
during outcome collection (end of trial)
if someone knows or figures out the intervention (treatment or placebo) which can consciously or unconsciously affect the trial outcome

Question 24

an example of attrition bias is incomplete outcome data, what might cause this outcome?

Answer 24

trial stopped early
participants missing outcome assessment appointment
incomplete documentation from participants

Question 25

an example of attrition bias is missing outcome data, what might cause this outcome?

Answer 25

patient data or records lost
patients excluded by investigators
non-adherence to medication regimen

Question 26

how does the intention-to-treat analysis mitigate attrition bias?
what does ITT analysis mean towards the study?

Answer 26

all randomized patients are analyzed according to randomization allocation regardless of protocol deviation or study withdrawal
ITT means the investigators analyze all data without disregarding any data that can skew the outcome results such as lack of medication adherence

Question 27

why is the intention-to-treat analysis good for superiority trials but not for non-inferiority trials?

Answer 27

it may bias non-inferiority trials into showing non-inferiority by making treatment and control arms look similar

Question 28

how is intention-to-treat different from “per protocol”?

Answer 28

if the study only uses data using the “per protocol” method, then they can exclude any significant events that occur prior to the treatment intervention which may affect the analysis of the treatment. using the intention-to-treat method, you include all events during the entirety of the trial regardless of when the treatment was introduced.
For example, if 10 people suffered a stroke before having surgery (the treatment intervention), the per protocol method would not include those 10 people before the surgery and would only analyze participants who have a stroke after receiving surgery. the intention-to-treat method would include both the 10 people who had a stroke before surgery and those after surgery, providing a more accurate representation of if the treatment intervention was significant or not.

Question 29

what percentage of patient loss is considered concerning in regards to determining attrition bias?

Answer 29

20%

Question 30

define what multiplicity is in regards to a study

Answer 30

potential inflation of the probability of making a Type I error due to multiple testing (observing a difference when it does not exist)

Question 31

what is Type 1 error? (a error)

Answer 31

rejecting the null hypothesis when it is true (observing a difference when it does not exist, false positive)

Question 32

in order to adjust for multiplicity, we can use what’s called the Bonferroni method, what is it?

Answer 32

a statistical adjustment method which assumes the null hypothesis is the same for all outcomes as an attempt to correct the inflation of type I error from multiple testing

Question 33

why should calculating the relative risk reduction not be used for primary outcomes?

Answer 33

may exaggerate the importance of the effect and shows a much greater risk difference compared to calculating the absolute risk reduction

Question 34

define what the number needed to treat is
how is it calculated?
what is the ideal NNT?

Answer 34

it is an estimation of how many pts who have received a treatment over a certain amount of time is required before one person would experience a beneficial outcome
1/ARR where ARR is the percent difference between the two study variables (20% risk in placebo - 10% risk in treatment= 10% ARR)
ideal NNT=1 (one person needs to be treated to prevent one bad outcome over a specific period)

Question 35

define what the number needed to harm is

Answer 35

an estimation of the number of patients who would have to receive the treatment over a specific time before one person would experience a negative outcome

Question 36

*only calculate NNT and NNH for statistically significant results