Critical numbers Flashcards
What is the difference between an experimental and observational study?
Experimental: researchers have intervened in some way
Observational: researchers haven’t intervened, just observed
What are 3 types of observational study?
Retrospective
Cross-sectional
Prospective
How do randomised controlled trials work?
Randomly allocate ptps to different interventions and follow up
Experimental
Prospective
What are some variations of RCT?
Cluster: ptps randomised in groups, not individually (e.g. a whole gp centre)
Crossover: ptps receive both interventions in a random order
Multi-arm: 2+ interventions evaluated in single study
Adaptive design trials: accruing info is used to inform planned design adaptations
Pros of RCTs
Gold standard
Reduced confounding potential
Reduce bias
Determine causal effects
Cons of RCT
Randomisations can be unfeasible or unethical
Require expert management and oversight
Expensive
How do cohort studies work?
Non-randomised
Observational
Typically prospective
Sample split into exposed/unexposed
Pros of cohort studies
Useful when random allocations not possible
Can work for rare exposures (select ptps exposed)
Can examine multiple outcomes
Cons of cohort studies
May require long follow up
Can be expensive
Not ideal for rare outcomes
How do case-control studies work?
Non-randomised
Observational
Retrospective
One group have condition already, other group don’t
Pros of case-control
Faster as use past data
Useful for rare outcomes- select ptps on basis of outcome
Cheaper
Cons of case-control
More prone to bias or poor quality data
Harder to show causal relationship
Not ideal for rare exposures
Features of cross-sectional studies?
Non-randomised
Observational
Single time point
Pros of cross-sectional studies
Relatively quick
Cheap
Can assess multiple exposures/outcomes
Cons of cross-sectional studies
Susceptible to bias
Cannot prove causality
Not ideal for rare exposures/outcomes
What are ecological studies, +pros and cons
Observational study of a group
Pros: large scale, can quantify geographical or temporal trends
Cons: ecological fallacy, can’t make inference at individual level
Types of categorical variables
Binary- 2 categories e.g. + or -
Ordinal- categories with natural order
Nominal- categories without order, e.g. blood group
Types of numeric variables
Discrete- only certain numerical values,e.g. number of children
Continuous- any value within a range, e.g. height
Define proportion
Number with a characteristic or outcome divided by total number
scale 0-1
Define odds
Number with an exposure or outcome divided by number without
Ratio of probability of event occurring vs not
probability divided by 1-probability
What is a rate?
Frequency per another unit of measurement, e.g. events per 100 people per year
Example of risk difference and risk ratio
risk of hypertension in aspirin group = 0.12
risk of hypertension in placebo group = 0.35
risk difference = 0.12 - 0.35 = 0.23, risk reduction 23%
risk ratio = 0.12/0.35 = 0.34, lower risk as less than 1
Example of odds ratio
odds of hypertension on aspirin = 0.13
odds of hypertension in placebo = 0.55
odds ratio = 0.13/0.55= 0.24
How to work out mean, median and mode
mean - sum the values divide by count, typically reported with st dev
median- order values then take midpoint, typically reported with range
mode- most common value, not used very often practically
if normal distribution, mean and median will be same
What is standard deviation?
Describes dispersion of values around the mean
x̄ = mean and sd = s when describing a sample
μ = mean and sd = σ when describing a population
What are range and interquartile range?
Range = lowest value and highest value
IQR = 25th-75th centile, central 50% range
What is right skewed distribution?
Positive
Median is less than mean
Tail on right hand side
What is left skewed distribution?
Negative
Mean less than median
Tail on left side
What % lie with 1sd, 2sd and 3sd of the mean (when symmetric)?
1sd = 68% of values
2sd = 95% of values
3sd= 99.7% of values
What is correlation?
A measure of linear relationships between variables
Quantified by correlation coefficient r
r between -1 and 1
What is statistical inference?
Use information from a sample to make inference about a population
Need to account for uncertainty
Derive estimates of population parameters and test hypotheses
What is central limit theorem?
the distribution of sample means approximates a normal distribution as the sample size gets larger, regardless of the population’s distribution
What is standard error?
Indicates how different a sample mean is likely to be from the population mean
Used for estimating
Tells us precision of estimation
Smaller standard error = more precise
Equation for standard error of the mean
SD divided by √n
bigger sd, bigger standard error
bigger the sample size, smaller standard error
What is a confidence interval?
a range around a measurement that conveys how precise the measurement is
gives plausible range
What factors affect confidence interval width?
Variability in the sample (SD)
Sample size (n)
The desired level of confidence
- typically we use 95% but it could be 90%, 99%, etc
What does a probability of 1 mean?
Event is certain to happen
What is the null hypothesis?
assume this hypothesis is true until we see sufficient evidence to the contrary
typically the theory we want to disprove
What is the p value?
the probability of seeing an effect of the observed magnitude or greater if the null hypothesis were true
What happens if p value is smaller than significance level?
Reject null hypothesis
Features of regression
Allows one variable to be predicted from the other
Order matters (predict y from x)
Can handle multiple predictors
Variables don’t have to be numeric
Features of correlations
Quantifies the linear association between two numeric variables
Variable order doesn’t matter (correlate x~y or y~x)
What is regression?
a statistical technique that relates a dependent variable to one or more independent (explanatory) variables
Advantages of multiple regression
- Can include both numeric and categorical predictors
- Explore relative strength of prediction
- can adjust or control for confounding
- reduce bias from chance covariate imbalance
- explore interactions between predictors
- make predictions based on combinations of risk factors
What is multiple regression?
More than one predictor
e.g. a+b1height+ b2age +
What is prognosis?
the forecast of future outcomes
What is critical appraisal?
the process of assessing and interpreting evidence by systematically considering its validity, results and relevance to your own context
Ways to reduce bias in RCTs?
- Randomly allocating participants to receive an intervention or not
- Blinding participants, investigators and assessors to the different treatments (where possible)
- Including a control group to compare the intervention against
What is a systematic review?
A research article in which existing evidence on a topic is systematically identified, appraised and summarised according to predetermined criteria
What is a meta-analysis?
Statistical method for combining evidence from different sources
What is heterogeniety?
a measure of variation between different studies
What is sensitivity analysis?
analysis to test the robustness of the findings of primary analysis – looks at the effect of assumptions or variations in approach.
What is sample bias?
sample does not represent population of interest
What is recall bias?
inaccurate recall of past events/exposures/behaviours
What is information bias?
incorrect measurement e.g. miscalibrated machine
What is the Hawthorne effect?
participants change their behaviour when they know they are being observed
What is attrition bias?
differential dropout from studies
e.g. sicker participants drop out so our outcome is only measured on healthier participants
What does a confounding variable do?
obscure the real effect of an exposure on an outcome
What does risk difference mean?
difference in proportions between groups
0 = no difference
What is the risk ratio (aka relative risk)?
risk in one group divided by risk in the other, greater than 1= higher risk in groups of interest, less than 1= lower risk in group of interest
What is the odds ratio?
odds in one group divided by odds in the other
When will median and mean be the same?
When distribution is normal
What does parametric mean?
make distributional assumptions
Stages of hypothesis testing using stats
Define null and H1
Choose significance level
Perform appropriate stats test
Find p value
What is regression?
a statistical technique that relates a dependent variable to one or more independent variable
Regression general eqtn
y=a+bx
y variable being predicted
x variable predicting (independent)
a constant/intercept
b gradient/coefficient
What is univariable regression?
One predictor
What is used to establish causality in non randomised trials?
Strength of association
Consistency
Specificity
Temporality
Biological gradient
Plausibility
Coherence
Experiment
Analogy
What is positivism based on?
objectivity
scientific method
Empiricism
What does interpretivism believe?
social reality can only be understood through social constructions such as language, consciousness and shared meanings and understandings
What is quantitative methodology?
Emphasises quantification in collection and analysis of data
Deductive approach – theory testing
Based on positivism
Views social reality as external and objective
What is qualitative methodology?
Emphasising words, rather than numbers
Inductive approach – generating theories (does not claim ‘truth’ status)
Based on interpretivism – understanding the ways in which individuals and groups interpret their world
Features of quantitative research
Deductive (theory testing)
Large, random samples
Generalisability, representativeness
‘Objective’ instruments (attitude/ outcome scales)
Results as number and statistics
Infer to population
‘Distance’ between researcher and subjects
Emphasis on following original research plan
Features of qualitative research
Inductive (develop theory)
Small, purposive samples
May, or may not, be representative
Less structured instruments (interviews)
Results as words and concepts
Do not infer to population
Reflexivity and attention to individual participants
Flexibility of approach
Examples of qualitative data
Interview transcripts
Focus groups transcripts
Field notes
Documents (reports, minutes, plans, blogs, social media posts)
Film, photographs, photo-elicitation.
Diary entries
What is critical appraisal?
the process of assessing and interpreting evidence by systematically considering its validity, results and relevance to your own context
What questions could be asked in critical appraisal?
Is the research question clear and focussed?
Is the study design appropriate?
What are the strengths/weaknesses?
Is there potential for bias? Has it been addressed?
Are the analysis methods appropriate?
Have the results been interpreted correctly?
Are the findings relevant to your practice?
What does PICO stand for?
Population
Intervention
Comparator
Outcomes
What does internal validity mean?
Accurately measuring those within the trial
What does external validity mean?
Generalisable outside trial
What is the CASP checklist?
Is the basic study design valid for a randomised controlled trial?
Was the study methodologically sound?
What are the results?
Will the results help locally?
How do RCTs attempt to limit bias?
Randomly allocating participants to receive an intervention or not
Blinding participants, investigators and assessors to the different treatments (where possible)
Including a control group to compare the intervention against
Features of a systematic review
Transparent, systematic methods make the process replicable
Bias is addressed by assessing each individual study for bias
What are the steps of a systematic review?
Specify research question
Develop search strategy and inclusion/exclusion criteria
Identify relevant studies
Assess quality and risk of bias
Extract data from each study
Pool and interpret results
Answer research question
What is a meta-analysis?
Statistical method for combining evidence from different sources
What is a sensitivity analysis?
analysis to test the robustness of the findings of primary analysis – looks at the effect of assumptions or variations in approach
What is heterogeniety?
a measure of variation between different studies
What are the PRISMA guidelines for?
guidelines aimed to improve the reporting of systematic reviews
