drug lit - bio stats 2 Flashcards
If everything else remains constant, what will happen as the sample size increases?
A
The power of the study increases
B
The study’s alpha increases
C
The difference in the primary endpoint between the two groups increases
D
The standard deviation increases
A
The power of the study increases
the more subjects, the greater the probability that if there is aj significant difference then we will detect it
SD will get smaller because we have more people and so we will be more sure of the results we get
“We calculated a sample size (n = 200) sufficient to detect a 20% difference between the two groups’ cure rates with 80% power and α=0.05.” The study enrolled 189 patients and detected an 18% difference between the two groups’ cure rates. Which one of the following is true?
A
If p < 0.05, the results ARE statistically significant
B
If p < 0.05, the results are NOT statistically significant because the study was underpowered
A
If p < 0.05, the results ARE statistically significant
power relates to type II error, which is a false negative
What does the hazard ratio below tell you?
A
Patients treated with clopidogrel experienced 840% more primary endpoints at 12 months versus ticagrelor
B
Patients treated with ticagrelor experienced 84% of the primary endpoints as patients treated with clopidogrel at 12 months
B
Patients treated with ticagrelor experienced 84% of the primary endpoints as patients treated with clopidogrel at 12 months
HR is similar to RR which is how frequently the dichotomous endpoint
for every 1 person in Clopidogrel, 84 Ticagrelor had it…
What does the odds ratio below tell you?
A
Being prescribed a bisphosphonate DECREASED the odds of developing esophageal cancer
B
Being prescribed a bisphosphonate INCREASED the odds of developing esophageal cancer
B
Being prescribed a bisphosphonate INCREASED the odds of developing esophageal cancer
because the OR is greater than 1
developing esophageal cancer is a dichotomous endpoint
What is the absolute risk reduction?
ARR = % of patients in the control group - % of patients in the experimental group who experienced the outcome
clopidogrel is the control and the ticagrelor is the experiment/intervention so
11.7 - 9.8 = 1.9% ARR
What is the number-needed-to-treat?
NNT = 100/% ARR
ARR = 11.4 - 9.3 = 2.1
100/2.1 = 47.61
but the round up to the nearest whole person so 48
make sure that you read the question and know what is the control and the experiment :)
What is the number-needed-to-harm?
looking at major bleeding
this result is a significant difference due to the p-value of 0.001
NNH: 3.7 - 2.7 = 1
100/1% = 100
What is the relative risk reduction?
A
0.16%
B
0.84%
C
16%
D
84%
RR = % experimental group / % control group
9.8 / 11.7 = 0.83
RRR: the absolute difference between RR and 1, then convert decimal to %
1 - 0.83 = 0.16 = 16%
C
16%
What is the relative risk increase for major bleeding with clopidogrel versus placebo?
A
138%
B
13%
C
38%
D
67%
1.38 - 1 = 0.38 or 38%
RR = 1.38
3.7 - 2.7 =1
Based on the HR and 95% confidence interval shown for the result below, was the difference between the two groups statistically significant?
A
Yes
B
No
A
Yes
because the interval does not cross 1
Patients who are current smokers are significantly more likely to have a myocardial infarction.
A
True
B
False
A
True
because the OR interval does not cross 1
Patients who are former smokers are significantly more likely to have a myocardial infarction.
A
True
B
False
B
False
because the OR crosses 1
Patients taking empagliflozin had significantly better renal function compared to placebo
A
True
B
False
true because it does not cross zero
objective
for calculating and interpreting the absolute risk reduction, absolute risk increase, number, number-needed-to-treat, and number and number-needed-to-harm
absolute risk reduction (ARR)
the proportion of subjects in the control group with an outcome subtracted by the proportion of subjects in the intervention group with an outcome
the % of patients who are spared an outcome as a result of receiving a treatment
only for dichotomous endpoints
ARR = % control group - % experimental group
number need to treat (NNT)
the # of patients who need to receive the intervention instead of the control to avoid 1 additional negative outcome (or achieve 1 additional positive outcome
put into the context of the duration of the study (ex: median duration of treatment)
NNT = 100/% ARR
always round up for NNT
absolute risk increase (ARI)
the % of patients who experience an outcome as a result of reducing the treatment instead of the control
the same as ARR, except ARI is calculated when the intervention increased the probability of an outcome in the treatment vs the control group
usually refers to negative outcomes but can be applied to a positive outcome that intervention increases the probability of, like a cure for an infection
- adverse effects
- negative outcomes
ARI = % intervention group - % control group
number needed to harm (NNH)
also referred as number needed to treat to harm
exactly the same as NNT except calculated when the intervention increased the chance of a harmful outcome in the treatment vs the control group
- adverse effects
- negative outcomes
NNH = 100/ARI
always round down for NNH
Important caveats
only calculate NNT/NNH for dichotomous endpoints with statistically significant results
ARR/ARI will always be a positive number because it refers to the absolute difference between 2 study groups
round UP for NNT
round DOWN for NNH
ARR = % control group / % intervention group
ARI = % intervention group - % control group
NNT = 100/% ARR
NNH = 100/ARI
Relative risk (risk ratio)
a comparison of the probability of an event/outcome happening in the treatment group vs the control group
RR = 1 means the probability of the event happening in the treatment group is the same as the probability of the event happening in the control group
RR < 1 means the probability of the event happening in the treatment group is less than the probability of the event happening in the control group
RR > 1 means the probability of the event happening in the treatment group is more than the probability of the event happening in the control group
RR = % treatment group / % control group
OR = (A/B) / (C/D)
RR = (A/A + B) / (C/C + D)
relative risk calculation and interpretation
RR = % intervention group / % control group
odds ratio
a measure of association between exposure and an outcome
- the odds that an outcome will occur in the presence of exposure, compared to the odds that the outcome will occur
OR is similar to RR byt RR is calculated using probability and OR is calculated using the ratio of odds
OR = means the exposure did not affect the odds of the outcome e
OR < 1 means the exposure is associated with lower odds of the outcome
OR > 1 means the exposure is associated with a higher odds of the outcome
calculating the odds ratio
disease (case). no disease (control)
exposed A B
unexposed C D
OR = odds that a case was exposed/odds that a control was exposed
OR = (A/D) / (B/C)
hazard ratios
a comparison of the probability of events in a treatment group vs the probability of events in a control group
used for survival analysis - measures if patients in the treatment group are progressing faster or slower than patients in the control group an outcome
- death/other adverse outcome (ex: stroke, heart attack)
- positive outcome like cancer-free survival or clinical cure
HR = 1 means no difference between the treatment vs control group
HR < 1 means the treatment group was less likely to have the outcome vs the control group
HR > 1 means the treatment group was more likely to have the outcome vs the control group
relatiev risk reduction (RRR)
the % of risk that is removed as a result of the treatment compared to the control
calculated by subtracting the RR from 1 and converting the decimal to %
- when RR is greater than 1 (ex: the intervention increased the risk of an outcome versus the control the RRR is the absolute difference between the RR and 1 (and it is reported as the relative risk increase)
the disadvantage of RRR
it doesn’t tell the magnitude of absolute risk
the solution
absolute risk reduction
- more clinically relevant than relative risk reduction
- allows you to calculate the number needed to treat and the number needed to harm
both relative risk reduction and absolute risk reduction are important components of reporting clinical study results
a reminder about P-values
will tell you “Yes or no, are results statistically significant”
need to take into account reasons for the increased risk of type I and type II errors when interpreting
multiple ways to calculate
95% confidence interval
imagine you want to find out if the true chance of getting heads on any single coin toss = 0.5 (50%)
toss the coin 2 times
- get 1 heads, 1 tails
toss a coin 100 times
- get 50 heads, 50 tails
confidence intervals
given a study result, the confidence interval gives an estimate of the range of values that the true difference between 2 study groups lies between
tells you “Yes or no, are the results statistically significant”
- also tells you “How sure of this study result am I”
the smaller the study sample, the wider the confidence interval
the larger the study sample, the more narrow the confidence interval (more likely that the observed difference is close to the true difference)
95% confidence intervals
given a study result, the 95% confidence interval shows the range of plausible values that the true difference lies between
- 95% of the time, the true results lie in the range
- if the study were repeated 100 times, the result would be within that range 95 times
- based on 1 study sample, provides an estimate of the true result in the larger population
using confidence intervals to determine statistical significance
dichotomous (yes/no) endpoints
- if the 95% CI crosses 1, it is NOT statistically significant
- if the 95% CI does NOT cross 1, it is significant significant
using confidence intervals to determine statistical significance
calculate the relative risk reduction/relative risk increase for a 95% confidence interval and decide for yourself if the relative risks are still clinically significant
for statistically significant results, are the results still clinically significant at the lowest end of the 95% confidence interval
- if no, then the results might not be clinically meaningful
for nonsignificant results, are the results clinically significant at the highest end of the 95% confidence interval
- if yes, maybe a new study with better power should be conducted
formulas to know
range
power
relative risk
hazard ratio
odds ratio (OR) and the 2X2 table
Relative risk reduction
absolute risk reduction
absolute risk increase
NNT
NNH
relative risk = % intervention / % control same as hazard ratio (?)
OR = AD/BC - use the 2X2 table :)
RRR = 1 - RR
RRI = 1 - RR
ARR = % of patients in the control group - % of patients in the experimental group who experienced the outcome
so ARR = % control - % intervention
ARI (absolute risk increase) = % of patients in the control group - % of patients in the experimental group who experienced the outcome, except when the intervention increased the probability of an outcome
so ARI = % control - % intervention
NNT = 100/ARR - round up
NNH = 100/ARI - roun down
