Bias's Flashcards
Selection bias
systematic differences between baseline characteristics of the groups (minimised by randomisation)
Performance bias
Systematic differences between groups in care provided other than the intervention of interest (minimised by blinding and having set protocol)
Detection bias
Systematic differences between groups in how outcomes are determined (minimised by blinding and having objective outcomes and standardising assessment)
Recall bias
A systematic error caused by differences in the accuracy of recollections retrieved by study participants
Publication bias
Only studies with positive results are published, not the neutral or negative studies (means can overestimate the effect of the treatment or intervention)
Language bias
Studies with positive findings are more likely to be published in an English-language journal and are also more quickly than those with inconclusive or negative findings
Power
Probability of picking up a significant difference where one exists ( the number needed to avoid a type II error)
Adequate power is generally 80% (0.8)
Factors associated:
- Sample size
- Alpha level (alpha level is the probability of rejecting the null hypothesis when the null hypothesis is true)
- Variability of outcome measure e.g. SD (lower variation = higher power)
- Minimum clinically significant difference
- Estimated attrition rate
Type 1 error
Wrongful rejection of the null hypothesis (false positive)
Causes may include: sampling error, data dredging & confounding
Type 2 error
Wrongful acceptance of the null hypothesis (false negative)
Can be caused by an underpowered study
Confounder
A confounder has a triangular relationship between the exposure & outcome, but is it not along the causal pathway
Reduced through matching and randomisation (accounts for unknown confounders) and restriction (inclusion/exclusion which removes effect from study but reduces generalisability)
Can also be reduced through standardisation (e.g. age-standardised risk) and subgroup analysis (but losses power) or multivariate analysis (allows you to study multiple confounders)
Interval validity
How well a study is designed to answer their question and hence the extent to which we can trust the reported outcome
A measure of how methodologically robust a study is and how well systematic bias is eliminated/accounted for
External validity
How generalisable the outcomes of the study are to the target population (compared to the study population)
What is within participant comparison
Participants are assessed before and after an intervention
Analysis is of the same participant
What is a cross-over trial
Participants receive both the intervention and control in a random order
Often is separated by a washout period
What is an N-of-1 trial
A single subject trial where an individual is the sole observation
Provides optimal intervention for an individual (e.g. optimal dose)
What is a factorial design
Study that investigates multiple independent variables on an outcome measure (both separately and combined)
What is a surrogate endpoint
When a biomarker (often easy to measure) is used to predict the likelihood of a clinical outcome
Pros = reduces required follow-up period and sample size and allows measurement when ideal outcome measure is excessively invasive or unethical
Cons = assumes a direct and guaranteed correlation between the biomarker and clinical outcome
What is a composite endpoint
Where multiple clinical events/outcomes are combined to form the primary outcome e.g. composite cardiovascular endpoints (unstable angina, stroke and MI)
Pros = allows for higher event rate so need shorter trial and fewer participants suitable when individual events occur infrequently
Cons = distribution of events may be unclear, main driver may be less severe/clinically relevant
P value
The calculated probability that a particular outcome has occurred due to chance
The calculated probability that the null hypothesis is true
Confidence interval
A range of values between which the true population value lies 95% of the time
Intention to treat
Statistical approach where all subjects are included in the analyses as members of the groups to which they are allocated, regardless of whether they completed the study or not
Pros = More closely reflects real life, Accounts for effect of adverse events & reduces attrition bias, Can help baseline characteristics of both groups to remain similar
Cons = Imputed value may be inaccurate (e.g. last observation carried forward)
Per protocol analysis
Where the final analysis only includes the patients that completed the treatment they were originally allocated to
Pros = accurately reflects the effects of treatment, useful in non-inferiority trials
Cons = subject to attrition bias - control and intervention groups may no longer have similar characteristics
Number needed to treat
Number of participants required to take a medication/have an intervention (compared with the control) to see one positive event
Is 1/ARR
Precision
How much agreement there is between repeat measurements
Accuracy
How close measurements are to the true value
Absolute risk
Incidence rate of the outcome
Absolute risk reduction
Incidence rate in the control group – incidence rate in the intervention/exposure group
CER – EER
Relative risk
The risk in the experimental group relative to that in the control group
EER/CER
Relative risk reduction
The risk reduction in the experimental group, relative to that in the control group
(CER-EER)/CER
Odds ratio
Likelihood that an outcome will occur given a particular exposure, compared to the likelihood of the outcome occurring without the exposure
Hazard ratio
is the relatively likelihood of the event occurring in the treatment vs control subjects at any given time point
Sensitivity
How well the test is able to detect those with the disease
True Positive (correctly detected with disease) /True Positive +False Negative (total with disease)
Specificity
How well the test is able to rule out those without the disease
True Negative (correctly detected without disease) /True Negative + False Positive (total without disease)
Positive predictive value
The percentage of people that test positive, that truly have the disease
True Positive (correctly detected with disease / True Positive +False Positive (total that tested positive)
Negative predictive value
The percentage of people that test negative, that truly do NOT have the disease
True Negative (correctly detected without disease)/ True Negative + False Negative (total that tested negative)
Systematic bias
Anything that impacts the study result in a non-random way and leads to an incorrect estimate of effect or association
Attrition bias
Systematic differences in withdrawals from the trial
Reduced by intention-to-treat analysis
Confounding bias
Refers to the presence of a factor that has a triangular relationship with the exposure and outcome, but is not along the causal pathway
Berkson bias
Form of selection bias, in which the sample is taken from a subpopulation (not the general population)
For example, participants being recruited from the hospital, rather than also community settings
Advantages and disadvantages of an RCT
Advantages:
- Gold standard for identification of treatment effects
- Prospective design that can infer causality
- Compares otherwise identical groups
- Allows for meta analyses later on
- Unbiased distribution of confounders
- Blinding more likely
Disadvantages:
- May be underpowered
- Expensive: time and money
- Volunteer bias
- Ethically problematic at times
- Surrogate outcomes may not reflect outcomes that are important to patients
Advantages and disadvantages of cohort studies
Advantages:
- Identifies causal and temporal relationships
- Gives information on prognosis
- Good for rare exposure
- Can answer questions about aetiology
- ethically safe;
- Allows determination of relative risk
Disadvantages:
- Expensive, time consuming
- Not suitable for rare disease
- Not suitable for disease with long latency periods
- exposure may be linked to a hidden confounder;
- blinding is difficult
- randomisation not present
- Attrition bias if subjects drop out (large cohort required)
Advantages and disadvantages of case-control study
Advantages:
- Identifies multiple risk factors for rare diseases
- Suitable when long time between exposure and outcomes
- Cheap and fast
- Used when unethical to conduct an RCT
Disadvantages:
- Relies on quality of records
- Selection bias (outcomes not established a priori)
- Recall bias
- Does not identify temporal association
Advantages and disadvantages of cross-sectional study
A study that examines the relationship between diseases (or other health-related characteristics) and other variables of interest as they exist in a defined population at one particular time (ie exposure and outcomes are both measured at the same time). Best for quantifying the prevalence of a disease or risk factor, and for quantifying the accuracy of a diagnostic test - is descriptive and may be a survey
Advantages:
- Identifies trends
- Cheap and fast
Disadvantages:
- Static
- Does not deduce causality on association at best
