Statistics Flashcards
A new blood test for glioblastoma multiforme
(GBM) has recently been discovered. Which
one of the following best indicates the proportion of patients without GBM in whom the test will be negative?
a. Negative predictive value
b. Positive predictive value
c. Power
d. Sensitivity
e. Specificity
e. Specificity
Sensitivity—proportion of patients with the disease who test positive. High sensitivity¼low false
negative.
Specificity—proportion of patients without
the disease who test negative. High specificity¼low false positive.
Negative predictive value—probability of truly
not having the disease if you are test negative.
Positive predictive value—probability of truly
having the disease if you test positive.
Power—The probability of correctly rejecting
the null hypothesis.
The proportion of patients experiencing
postoperative CSF leak is to be compared
between two differently treated groups.
Which one of the following statistical tests
is most appropriate to evaluate this?
a. Mann-Whitney U test
b. Multiple linear regression
c. Spearman’s correlation coefficient
d. Wilcoxon’s test
e. χ(Chi-squared) test
e. χ(Chi-squared) test
(Chi-squared) test—this is a test of
A new test for cauda equina syndrome (CES)
has 80% sensitivity and 90% specificity. If,
on average 1 in 10 patients with CES symptoms referred to neurosurgery actually has
CES, what are the odds that a patient with
CES symptoms testing positive on the new
test actually has CES?
a. 4 in 10
b. 4 in 15
c. 4 in 20
d. 4 in 25
e. 4 in 5
e. 4 in 5
Odds of CES in patient referred with CES
symptoms and positive test¼pre-test odds in
symptomatic patients x likelihood ratio of
a positive test (sensitivity/(1specificity))¼
1/100.8/(10.9)¼8/10¼4 in 5.
Likelihood ratio for a positive result tell us
how much the odds of the condition increase
when the test result is positive.
Standard deviation of a set of values is best
described as:
a. The measure of spread of the values
around the mean
b. The simplest nonparametric statistical test
c. The square of the variance of the values
d. The square root of the standard error of
the mean
e. The standard error of the mean
a. The measure of spread of the values
around the mean
Standard deviation (SD) of a population is the
square root of the variance. For normally distributed data, 68% of the values are within 1SD of the
mean, 95% within 2SD, and 99% within 3SD.
Standard error of the mean (SEM) is the standard deviation (of the population) divided by the
square root of the number of values in any sample
from that population—it is a measure of how close
the sample mean is likely to be to the population
mean.
A new therapy reduces the rate of infarction from 25% to 5% in hemorrhage patients. What is the number
of patients that need to be treated with the
new therapy to prevent one infarction?
a. 5
b. 10
c. 15
d. 20
e. 25
a. 5
Absolute Risk Reduction (ARR)¼risk in control group—risk
in intervention group¼0.2 (20% reduction in
infarction rate).
Relative risk reduction (RRR)¼(255)/25¼80%.
Relative Risk (RR)¼relative difference in event rate between
intervention and control groups¼(5/100)/(25/100)¼0.2.
Odds Ratio (OR)¼odds of event in intervention group/
odds of event in control group¼(5/95)/(25/75)¼0.157.
Number needed to treat is calculated as
1/ARR¼1/(0.2)¼5.
Number needed to harm (NNH) indicated how many
patients on average need to be exposed to a risk factor
over a specific period to cause harm to one patient who
would not otherwise have been harmed
The power of a study is best described as
which one of the following:
a. The probability of a significant finding
b. The probability of correctly rejecting the
null hypothesis
c. The probability of needing further data at
the end of the study period
d. The probability of obtaining an answer
e. The probably of accepting the null
hypothesis
b. The probability of correctly rejecting the
null hypothesis
The power of a study is the likelihood of it correctly rejecting the null hypothesis when it is false
(or 1 beta), and is normally expressed as a percentage. Power analysis can be used to calculate
the minimum sample size required so that one
can be reasonably likely to detect an effect of a
given size, and vice versa. In most cases adequate
power is accepted as 0.8, with beta 0.2 (Type II
error—false negative rate) and alpha 0.05 (Type
1 error—false positive rate) although these may
vary depending on situation.
You wish to assess rare adverse effects
thought to occur in less than 0.1% of patients
taking a new antiepileptic. What type of
study is most appropriate for assessing these
effects?
a. Phase I
b. Phase II
c. Phase III
d. Phase IV
e. Phase V
d. Phase IV
Which one of the following is a parametric
statistical test?
a. Multiple linear regression
b. Spearman’s correlation coefficient
c. Student’s t-test
d. Wilcoxon’s test
e. χ2 (Chi-squared) test
d. Wilcoxon’s test
In a clinical trial of a new treatment for pituitary adenoma which one of the following are
most likely to result in a type II error?
a. False hypothesis
b. Large effect size
c. Normally distributed data
d. Small sample size
e. Use of multiple statistical tests
d. Small sample size
A type II error is the mistaken acceptance of the
null hypothesis when it is incorrect (false negative). This is most likely to occur when the power
of a study to detect differences is small, e.g. small
sample size, small effect size. A type I error (false
positive) is if multiple statistical tests are used
without any correction for multiple comparisons.
Recall bias is most likely to be an issue in
which one of the following?
a. Case-control studies
b. Cross-over trials
c. Meta-analysis
d. Prospective cohort studies
e. Randomized controlled trial
a. Case-control studies
Recall bias is most likely to occur when participants are asked about their experience, risk factors or behavior after they have been diagnosed
with the disease under study
In a given population, which one of the following best reflects a change in the balance
of etiological factors of a particular disease?
a. Five year mortality rate
b. Incidence
c. Period prevalence
d. Point prevalence
e. Standardized mortality ratio
b. Incidence
A new therapy for cerebral venous sinus thrombosis finds that it reduces the mortality rate from 4.4% to 3%. Which one of the following best describe the absolute risk reduction?
a. 0.0014%
b. 0.014%
c. 0.14%
d. 1.4%
e. 14%
d. 1.4%
Which one of the following best explains
higher death rates in head injury patients
treated with thiopentone infusion?
a. Attrition bias
b. Confounding bias
c. Detection bias
d. Publication (reporting) bias
e. Researcher bias
f. Selection bias
f. Selection bias
A trial of a new antibiotic for meningitis finds
that it reduces the mortality rate from 35% to
20%. What is the relative risk?
a. 0.0015
b. 0.015
c. 0.15
d. 1.5
e. 15
c. 0.15
Which one of the following terms best
describes data collected about systolic blood
pressure of patients presenting with incidental intracranial aneurysms
a. Bayesian variable
b. Continuous interval scale variable
c. Continuous ratio scale variable
d. Nominal variable
e. Ordinal variable
c. Continuous ratio scale variable
Which one of the following types of test is
best used to estimate the effect and statistical significance of gender on survival in a cohort
of patients with subarachnoid hemorrhage?
a. Cox regression model
b. Kaplan-Meier survival curve
c. Linear regression
d. Logistic regression
e. Spearman’s rank correlation
a. Cox regression model
Regression analysis is a technique for finding the
relationship between variables, one of which is
dependent on the other.
A new brain scan to diagnose radiation
necrosis (RN) is performed on 200 patients
with suspected RN. Of 20 people eventually
diagnosed with RN, 5 tested positive on the
new scan. In addition, 45 people without
RN tested positive with the brain scan. What
is the sensitivity and specificity of the test?
a. Sensitivity 25, specificity 75
b. Sensitivity 25, specificity 90
c. Sensitivity 75, specificity 25
d. Sensitivity 90, specificity 25
e. Sensitivity 90, specificity 75
a. Sensitivity 25, specificity 75
- Test tocompare survivalbetween twounpaired
groups using Kaplan-Meier estimates.
Statistical tests:
a. Analysis of variance (ANOVA)
b. Bonferroni correction
c. Cox proportional hazards regression
model
d. Fisher’s exact test
e. Friedman’s test
f. Kruskal-Wallis test
g. Linear regression
h. Log-rank test
i. Logistic regression
j. Mann-Whitney U test
k. Multiple linear regression
l. Pearson correlation coefficient
m. Spearman’s rank correlation coefficient
n. Student’s t-test
o. Wilcoxon’s signed-rank test
p. χ2 (Chi-squared) test
h. Log-rank test
The selection of the most appropriate test generally depends on hypothesis, type of data (continuous or categorical variables), number of groups
(two vs. more than two), whether data is normally
distributed and whether data is independent
(unpaired) or dependent (paired; same individual
or possibly matched individuals). The tables
below should be used as guides only.
- Test to compare paired blood pressure
measurements before and after starting
nimodipine.
Statistical tests:
a. Analysis of variance (ANOVA)
b. Bonferroni correction
c. Cox proportional hazards regression
model
d. Fisher’s exact test
e. Friedman’s test
f. Kruskal-Wallis test
g. Linear regression
h. Log-rank test
i. Logistic regression
j. Mann-Whitney U test
k. Multiple linear regression
l. Pearson correlation coefficient
m. Spearman’s rank correlation coefficient
n. Student’s t-test
o. Wilcoxon’s signed-rank test
p. χ2 (Chi-squared) test
n. Student’s t-test
The selection of the most appropriate test generally depends on hypothesis, type of data (continuous or categorical variables), number of groups
(two vs. more than two), whether data is normally
distributed and whether data is independent
(unpaired) or dependent (paired; same individual
or possibly matched individuals). The tables
below should be used as guides only.
- Test to compare occurrence of postoperative
CSF leak after foramen magnum decompression when two different methods for closure were used in 10 patients.
Statistical tests:
a. Analysis of variance (ANOVA)
b. Bonferroni correction
c. Cox proportional hazards regression
model
d. Fisher’s exact test
e. Friedman’s test
f. Kruskal-Wallis test
g. Linear regression
h. Log-rank test
i. Logistic regression
j. Mann-Whitney U test
k. Multiple linear regression
l. Pearson correlation coefficient
m. Spearman’s rank correlation coefficient
n. Student’s t-test
o. Wilcoxon’s signed-rank test
p. χ2 (Chi-squared) test
d. Fisher’s exact test
The selection of the most appropriate test generally depends on hypothesis, type of data (continuous or categorical variables), number of groups
(two vs. more than two), whether data is normally
distributed and whether data is independent
(unpaired) or dependent (paired; same individual
or possibly matched individuals). The tables
below should be used as guides only.
- The proportion by which an intervention
reduces the risk of an event
Statistical terms:
a. Alpha
b. Beta
c. Confidence interval
d. Degrees of freedom
e. Hazard ratio
f. Intention to treat
g. Meta-analysis
h. NNT
i. Nonparametric
j. Odds ratio
k. Parametric
l. Power
m. Relative risk reduction
n. Two tailed test
o. Type I error
p. Type II error
m. Relative risk reduction
- Enables the null hypothesis to be rejected
whether the new treatment is better or worse
the current treatment (i.e. in either direction)
Statistical terms:
a. Alpha
b. Beta
c. Confidence interval
d. Degrees of freedom
e. Hazard ratio
f. Intention to treat
g. Meta-analysis
h. NNT
i. Nonparametric
j. Odds ratio
k. Parametric
l. Power
m. Relative risk reduction
n. Two tailed test
o. Type I error
p. Type II error
n. Two tailed test
- Where patients are included in the analysis
according to the group into which they were
randomized, even if they did not actually
receive the treatment.
Statistical terms:
a. Alpha
b. Beta
c. Confidence interval
d. Degrees of freedom
e. Hazard ratio
f. Intention to treat
g. Meta-analysis
h. NNT
i. Nonparametric
j. Odds ratio
k. Parametric
l. Power
m. Relative risk reduction
n. Two tailed test
o. Type I error
p. Type II error
f. Intention to treat
- Thoroughly examines a number of valid studies on a topic and mathematically combine the results using accepted statistical methodology to report the results as if it were one large study.
Study Design:
a. Case series
b. Case-control study
c. Cohort study
d. Cross-sectional design
e. Cross-over design
f. Experimental study
g. Meta-analysis
h. Observational study
i. Prospective study
j. Randomized controlled trial
k. Retrospective study
l. Systematic review
g. Meta-analysis
- Focus on a clinical topic and answer a specific question. An extensive literature search
is conducted to identify studies with sound
methodology. The studies are reviewed,
assessed for quality, and the results summarized according to the predetermined
criteria of the review question.
Study Design:
a. Case series
b. Case-control study
c. Cohort study
d. Cross-sectional design
e. Cross-over design
f. Experimental study
g. Meta-analysis
h. Observational study
i. Prospective study
j. Randomized controlled trial
k. Retrospective study
l. Systematic review
l. Systematic review
- Describes the relationship between diseases
and other factors at one point in time in a defined population.
Study Design:
a. Case series
b. Case-control study
c. Cohort study
d. Cross-sectional design
e. Cross-over design
f. Experimental study
g. Meta-analysis
h. Observational study
i. Prospective study
j. Randomized controlled trial
k. Retrospective study
l. Systematic review
d. Cross-sectional design
- Individual randomized controlled trial
Levels of evidence:
a. 1a
b. 1b
c. 1c
d. 2a
e. 2b
f. 2c
g. 3a
h. 3b
i. 4
j. 5
b. 1b
- Systematic review of case-control studies
Levels of evidence:
a. 1a
b. 1b
c. 1c
d. 2a
e. 2b
f. 2c
g. 3a
h. 3b
i. 4
j. 5
g. 3a
