STATISTICS Flashcards
(24 cards)
Patients with GBS Patients without GBS
Test Positive 84 2
Test Negative 6 40
A new test for Group B strep has been designed and the results are shown above. Which of the below represents the Negative Likelihood ratio?
0.07
Likelihood Ratios
LR+ = Sensitivity / (1 - Specificity)
LR- = (1- Sensitivity) / Specificity
Sensitivity
Sensitivity is the ability of a test to correctly identify disease.
Sensitivity = True Positives/Number of people with disease
Specificity
Specificity is the ability of a test to correctly exclude a disease
Specificity =True Negatives/Number of people without disease
A study looks at the average systolic blood pressure of patients in antenatal clinic. Out of 200 women the average SBP was 110mmHg and the standard error of the mean was 8mmHg. What is the 95% confidence interval?
95% CI = 94.32 - 125.68
Assuming a normal distribution, we can state that 95% of the sample mean would lie within 1.96 SDs above or below the population mean.
For simplicity or to calculate a quick estimate you can use 2 rather than 1.96
Calculation of 95% CI for mean = (mean - 1.96xSEM) to (mean +1.96xSEM)
Upper CI limit = mean + (1.96 x SEM)
Lower CI limit = mean - (1.96 x SEM)
SEM = 8
SEM x 1.96 = 15.68
95% CI = (110-15.68) to (110+15.68)
95% CI = 94.32 - 125.68
Condition Positive Condition Negative
Test Outcome Positive A B
Test Outcome Negative C D
Negative Predictive Value is
D/C+D
Negative Predictive Value
Negative predictive value = D/C+D or True Negatives/All Negative test results
Study Z gets funding to increase the population size from 1000 to 2000 patients. What effect is this likely to have on the results?
Increasing sample size decreases type 2 errors and increases power of the results.
Sensitivity and specificity of the test should remain constant regardless of sample size.
Types of Error
Type I error
Incorrect rejection of a true null hypothesis
False positive
Detecting an effect that is not present
Type II error
Incorrect acceptance of a null hypothesis
False negative
Failing to detect an effect that is present
Patients with GBS Patients without GBS
Test Positive 84 2
Test Negative 6 40
A new test for Group B strep has been designed and the results are shown above. Which of the below represents the Positive Predictive Value?
PPR = 84/84 +2 = 0.97
Positive Predictive Value
Positive predictive value =True Positives/All positive test results
Patients with GBS Patients without GBS
Test Positive 84 2
Test Negative 6 40
A new test for Group B strep has been designed and the results are shown above. Which of the below represents the Specificity?
Specificity = 40 / 40 +2 = 0.95
Specificity
Specificity is the ability of a test to correctly exclude a disease
Specificity = D / B+D or True Negatives/Number of people without disease
Patients with GBS Patients without GBS
Test Positive 84 2
Test Negative 6 40
A new test for Group B strep has been designed and the results are shown above. Which of the below represents the Positive Likelihood ratio?
LR+ = Sensitivity/ (1-Specificty) = 18.6
Likelihood Ratios
LR+ = Sensitivity / (1 - Specificity)
LR- = (1- Sensitivity) / Specificity
Sensitivity
Sensitivity is the ability of a test to correctly identify disease.
Sensitivity = A / A+C or True Positives/Number of people with disease
Specificity
Specificity is the ability of a test to correctly exclude a disease
Specificity = D / B+D or True Negatives/Number of people without disease
Patients with GBS Patients without GBS
Test Positive 84 2
Test Negative 6 40
A new test for Group B strep has been designed and the results are shown above. Which of the below is the Negative Predictive Value?
NPR = 40/40+6 = 0.87
Negative Predictive Value
Negative predictive value = D/C+D or True Negatives/All Negative test results
Condition Positive Condition Negative
Test Outcome Positive A B
Test Outcome Negative C D
Which of the following represent Type 2 errors?
C
False negatives are type 2 errors
What study is most appropriate to assess the efficacy of a new medication to reduce pain caused by endometriosis?
When assessing treatment or intervention a randomised control trial is the gold standard. Note if assessing prognosis or diagnostic tests Cohort and Cross Sectional Analysis (AKA cross sectional study) respectively are optimal.
Randomised Control Study
Gold standard for clinical trial used to test efficacy of medical intervention
Most appropriate study to assess treatment/intervention
Cross Sectional Study or Analysis
Involves analysis of data for a population at one specific point in time
Descriptive studies
Most appropriate method to assess diagnostic tests
Cohort Study
Form of longitudinal study that compares groups with shared characteristic eg risk factor
It may be impossible or unethical to randomise a group to a prognostic factor
Most appropriate study for assessing prognosis
A small study shows the lifetime incidence of breast cancer in smokers to be 20 in 100. In non smokers the incidence is 10 in 100. What is the absolute risk of breast cancer in smokers?
Don’t get absolute and relative risk confused. Absolute risk is the actual risk. In this case 20 in 100 or 20%.
Relative risk is a ratio of risk comparing risk in the exposed and control groups, in this case relative risk would be 2.
Relative Risk
Relative risk (RR) is the ratio of risk in an exposed group to a non exposed group
RR = Probability of an event when exposed/Probability of event in control group
Group Disease Present Disease Absent
Exposed a b
Control c d
It can be calculated by RR = [a / (a+b)] ÷ [c / (c+d)]
A relative risk of 1 means there is no difference in risk between control and exposed groups.
Odds Ratio
Odds ratio (OR) = [a÷b] ÷ [c÷d]
Study Y involves a trial of 72 patients. It finds improved outcomes in the treatment group (being treated with Drug B) than those taking placebo. The null hypothesis (that Drug B makes no difference to outcomes) is rejected.
Several larger studies involving several thousand patients clearly demonstrate Drug B makes no statically significant difference to outcomes. What type of error has occurred in Study Y?
This is an example of a Type 1 error.
Type I error
Incorrect rejection of a true null hypothesis
False positive
Detecting an effect that is not present
Patients with GBS Patients without GBS
Test Positive 84 2
Test Negative 6 40
A new test for Group B strep has been designed and the results are shown above. Which of the below is the Sensitivity?
Sensitivity = 84 / 84 +6 = 0.93
Sensitivity
Sensitivity is the ability of a test to correctly identify disease.
Sensitivity = A / A+C or True Positives/Number of people with disease
Evidence from a panel of experts is what level of evidence
Level IV = Evidence from a panel of experts
This is the weakest form of evidence
Levels of evidence range from I to IV as shown below
Ia - Evidence from Meta-analysis of Randomized Controlled Trials
Ib - Evidence from at least one Randomized Controlled Trial
IIa - Evidence from at least one well designed controlled trial which is not randomized
IIb - Evidence from at least one well designed experimental trial
III - Evidence from case, correlation, and comparative studies.
IV - Evidence from a panel of experts
Evidence from a single randomised control trial is what level of evidence
Ib
Levels of evidence range from I to IV as shown below
Ia - Evidence from Meta-analysis of Randomized Controlled Trials
Ib - Evidence from at least one Randomized Controlled Trial
IIa - Evidence from at least one well designed controlled trial which is not randomized
IIb - Evidence from at least one well designed experimental trial
III - Evidence from case, correlation, and comparative studies.
IV - Evidence from a panel of experts
A study looked at 40,000 women at a maternity unit in an area with a population of 520,000. The number of identified VTE episodes was 60 with a standard deviation of 20. Using the figures what is the standard error of the mean (SEM)?
0.1
Standard Error of the Mean = Standard Deviation / square root of sample size
SEM = SD/ square root of n
SEM = 20/200 (200 is the square root of 40,000)
SEM = 0.1
To calculate SEM therefore the only figures you need are the sample size and the standard deviation.
The Standard Error (SE) measures the amount of variability in the sample mean OR another way to put it is the SE is the standard deviation of the sampling distribution of a statistic. It indicates how closely the population mean is likely to be estimated by the sample mean.
Which of the following studies is most appropriate to assess a prognosis of a disease?
Cohort study
Condition Positive Condition Negative
Test Outcome Positive A B
Test Outcome Negative C D
Which of the following represent Type 1 errors:
False positives are type 1 errors
B
Study X involves a trial of 40 patients. It finds no difference between the control (treated with placebo) group and treatment groups (being treated with Drug A). The null hypothesis ( that Drug A makes no difference to outcomes) is accepted.
Several larger studies involving several thousand patients clearly demonstrate Drug A makes a statically significant difference to outcomes. What type of error has occurred in Study X?
This is an example of a Type 2 error.
Type II error
Incorrect acceptance of a null hypothesis
False negative
Failing to detect an effect that is present
Which of test require the population to be of normal distribution?
Tests may be parametric or non-parametric
Parametric tests assume a normal distribution of population data for the variable being tested and are used for testing variables within a population that are interval or ratio e.g. height, temperature or age.
Non-Parametric tests can be used for data that is not normally distributed within a population or is of nominal or ordinal value e.g. eye colour or marital status
Parametric Test Non-Parametric Test
Pearson* Spearman*
t-test Mann Whitney
Analysis of Variance (ANOVA) Chi-squared
f-test Wilcoxon Signed Rank
z-test Fisher Exact Probability
Kruskal Wallis
Friedman
Evidence from meta-analysis of RCTs is what level of evidence?
Ia
You are carrying out analysis of a new test for DVTs and decide to plot a ROC curve. What are the axes of the curve?
ROC Curves plot Sensitivity against 1-Specificty. They are a useful measure of how good a test is at distinguishing those with and without disease. The greater the area under the curve the better the test.
A small study looks at medical records of deceased women to see how many smoked and how many were diagnosed with breast cancer. The results are below:
Group Breast cancer Disease Free
Smokers 15 65
Non-smokers 11 77
Using these figures what is the relative risk of breast cancer in smokers?
RR = 1.5
First calculate the prevalence in the control and exposed groups
In smokers: 15/ (15+65) = 0.1875
In Non-smokers 11/ (77+11) = 0.125
Then divide the prevalence in the exposed by the control group
0.1875/0.125 = 1.5
Relative Risk
Relative risk (RR) is the ratio of risk in an exposed group to a non exposed group
RR = Probability of an event when exposed/Probability of event in control group
Which study is most appropriate to assess a diagnostic test?
Cross Sectional Study or Analysis
Involves analysis of data for a population at one specific point in time
Descriptive studies
Most appropriate method to assess diagnostic tests
