Statistics Flashcards
Define specificity
Proportion of patients without the condition who have a negative result
Formula for specificity
TN / TN + FP
Define sensitivity
Proportion of patients with the condition who have a positive test result
Formula for sensitivity
TP / TP + FN
Define positive predictive value
The chance that the patient has the condition if the diagnostic test is positive
Formula for positive predictive value
TP / TP + FP
Define negative predictive value
The chance that the patient does not have the condition if the diagnostic test is negative
Formula for negative predictive value
TN / TN + FN
Define likelihood ratio for a positive test result
How much the odds of the disease increase when a test is positive
Formula for likelihood ratio for a positive test result
sensitivity
/
(1 - specificity)
Define likelihood ratio for a negative test result
How much the odds of the disease decrease when the test is negative
Formula for the likelihood ratio for a negative test result
(1 - sensitivity)
/
specificity
Which significance tests are used on parametric data?
Student’s t- test (paired or unpaired)
Pearson’s product-moment coefficent (for correlation)
Which significance test is used on non-parametric data for comparing ordinal, interval or ratio scales of unpaired data?
Mann-Whitney U test
Which significance test is used on non-parametric data comparing two sets of observations on the same sample (e.g. before and after)
Wilcoxon signed-rank test
Which significance test is used to compare proportions or percentages on non-parametric data?
Chi-squared test
Which significance test is used for measuring correlation in non-parametric data?
Spearman, Kendall rank
Relationship between mean, median and mode in normal distributed data
mean = median = mode
Relationship between mean, median and mode in positively skewed data
mean > median > mode
Relationship between mean, median and mode in negatively skewed data
mean < median < mode
What is the goal of a phase 1 trial?
Determine pharmacokinetics and pharmacodynamics and side effects
Small study on healthy volunteers
What is the goal of a phase 2 trial?
Goals of 2a and 2b trials
How many patients?
Assess dosage and efficacy
2a = dosage 2b = efficacy
Small number of patients affected by disease
What is the goal of phase 3 trial?
Assess effectiveness
Large RCT comparing new treatment to old
What is the goal of a phase 4 trial?
Monitoring long term effectiveness and side effects
Formula for numbers needed to treat
1 / (absolute risk reduction)
How to calculate the absolute risk reduction
Control event rate - experimental event rate
How to calculate the control event rate
number who had outcome with control / total number who had the control
How to calculate the experimental event rate
number who had outcome with the intervention / total number who had the intervention
What is the usual outcome measure for a cohort study?
Relative risk
What is the usual outcome measure for a case-control study?
Odds ratio
What does endemic mean?
Persistent, usual or expected level of disease in a given population
How to calculate the relative risk
EER / CER
How to calculate the relative risk reduction or relative risk increase
EER - CER / CER
What is a type 1 error?
The null hypothesis is rejected when it is true
What is a type 2 error?
The null hypothesis is accepted when it is false
What is the name given to a phenomenon where if a variable is extreme on its first measurement, it will tend to be closer to the average on its second measurement
Regression
What statistical test to use on data with several groups?
ANOVA
What is ANOVA used for?
Comparing the variance of the means
Several groups of data
Normally distributed data
What is a spurious association?
Arisen by chance and not real
What is an indirect association?
Association is due to another factor (a confounding variable)
What is a direct association?
True association not linked by a third variable
What criteria are use to establish causation?
Bradford Hill Causal Criteria
How to calculate standard deviation
√variance
How to calculate standard error of the mean
standard deviation / √n
What is confounding?
A variable that correlates to other variables within a study, leading to spurious results
How can you control confounding at the design stage of an experiment?
Randomisation
How can you control confounding at the analysis stage of an experiment?
Stratification
What is nominal data?
Values put into categories with no order e.g. birth place
What is ordinal data?
Values put into categories which can be ordered e.g. NYHA classification of heart failure
What is discrete data?
Values confined to certain values e.g. number of asthma exacerbations per year
What is continuous data?
Data can take any value within a certain rage e.g. weight
What is binomial data?
Data may take one of two values e.g. gender
What is the delphi method?
Method of asking for then collating expert opinion
Issues with the delphi method
Time consuming
Low response rates
Investigators can mould responses
Expertise of the panelists could be unevenly distributed
What is the hierarchy of medical evidence?
1) Metanalysis of RCT
2) 1x RCT
3) Systematic reviews of cohort studies
4) Cohort studies
5) Systematic reviews of case-control studies
6) Case-control studies
7) Case series
8) Expert opinion
When is hazard ratio used?
When analysing survival over time
What is used to measure general practice workload?
Jarman Underprivileged Area Score
What % of values lie within 1 SD of the mean?
68.3%
What % of values lie within 2 SD of the mean?
95.4%
What % of values lie within 3 SD of the mean?
99.7%
What % of values lie within 1.96 SD of the mean?
95%
What is the power of a study?
Probability of correctly rejecting the null hypothesis when it is false
How is power calculated?
1 - probability of a type 2 error
What is the p value?
the probability of obtaining a result by chance as least as extreme as the one that was observed, assuming the null hypothesis was true
What does reliability assess?
Consistency
What does validity assess?
Accuracy
What error is the p value equal to?
Type 1 error
Equivalence trial
Equivalence margin of -delt to +delta
If CI lies within equivalence margin then drugs have similar effect
What are independent variables?
Something that is purposefully changed by the researcher
What are dependent variables?
The one that is observed and changes in response to the independent variable
What are controlled variables?
All other variables that are not the dependent variable or the independent variable
Do not change throughout the study
What is the population attributable risk?
Reduction in incidence if the population were entirely unexposed
What is the attributable proportion?
Proportion of the disease that would be eliminated if the disease rate were reduced to that of the unexposed group
What is a case control study?
Retrospective
Participants with condition are identified and matched with control
Information then collected on past exposure to a possible causal agent
What is a cohort study?
Prospective
Participants selected on their exposure to a particular agent and followed up to see how many develop disease
Cross-sectional study
‘snap shot’ to find prevalence
What is the standardised mortality rate?
Observed deaths / expected deaths
What does standardised mortality rate of 100 indicate?
Mortality in the group being studied is the same as the standard population
What is the mortality rate?
Number of deaths in a particular population at a particular time
What is the maternal mortality rate?
deaths linked to childbearing per live births per year
What is the infant mortality rate?
Deaths <1 year per live births per year
What is the fetal mortality rate?
Fetal deaths per live births per year
What is external validity?
Degree to which the conclusions in the study would hold for other persons in other places at other times
e.g. ability to generalise
What is internal validity?
How confident we are that the change in the independent variable caused the observed change in the dependent variable
What is face validity?
Does it appear to test what it is meant to?
What is content validity?
Extent to which a test or measure assesses the full content of a subject or area
Methods of assessing validity in qualitative research
Triangulation
Respondant validation
Bracketing
Reflexivity
What is triangulation in qualitative research?
Compares results from 2 or more different methods of data collection, or from 2 or more data sources
Used to assess validity
What is respondant validation in qualitative research?
Investigators account is compared to those of research subjects
Used to assess validity
What is bracketing in qualitative research?
Putting asides ones own beliefs / knowledge
What is reflexivity in qualitative research?
Being sensitive to the ways the researcher and research process have shaped the collected data
What is work up bias?
Clinicians reluctant to order gold standard test if new test is negaive
What is expectation bias?
Observers subconsciously measure/ report data in a way that favours study outcome
What is the Hawthorne effect?
group changes its behaviour because it is being watched
What is procedure bias?
Subjects in different groups receive different treatments
What is anchoring bias?
When the first piece of information influences decisions too much
what is late-look bias?
Gathering information at an inappropriate time
What is lead time bias?
New test diagnoses earlier but there is no effect on the outcome of the disease
What is Berkinson’s/hospital bis?
Occurs in case control studies where hospitalised patients are used as controls
How to calculate risk
number of events / number at risk
how to calculate odds
number of events / number of non-events
How to calculate relative risk (or risk ratio)
Risk in exposed group / risk in unexposed group
How to calculate odds ratio
Odds of having been exposed / Odds in a control group
What does correlation tell us?
How one variable may increase or decrease as another variable changes, but not by how much
What does linear regression tell us?
How much one variable changes when a second variable is changed
What is cost-effectiveness analysis?
Compares interventions by relating costs to a single clinical measure
combines costs and effects into an incremental cost-effectiveness ratio (ICER)
What is cost-benefit analysis?
All costs and benefits are measured in terms of money
What is cost-utility analysis?
Form of cost-effectiveness analysis but allows comparison between different diseases
Uses QALY
What method of analysis uses QALY?
Cost-utility analysis
What does 1 QALY equal?
1 year in perfect health
What does QALY measure?
Gains in life expectancy and health related quality of life
can be negative
What is cost-minimisation analysis?
Aim is to decide the least costly way of achieving the same outcome
