Types of study design Flashcards
Cohort studies
select disease-free individuals
classify the individuals by their exposure
follow-up and measure number of people with disease and extent of exposure.
calculate incidence rates for each exposure group
Tells you about disease aetiology and disese prognosis
Types of comparison in cohort studies
Internal: Individuals split into sub-cohorts based on exposure and compared using Incidence Rate Ratio (IRR). Requires large cohort study, prone to greater random variation
External: comparison of exposed cohort with an external reference population normally using Standardised Mortality Ratio (SMR). Prone to inadvertent selection bias and unmeasured confounders
Case control studies
Aim to identify whether an exposure is associated with a disease
Cases are identified as a group of people who have the disease
Controls are identified who do not have the disease
Cases and controls interviewed to compare exposure frequency in two groups. Enquires about exposures that happened before the event.
Analysed using Odds ratio
Advantages of case control studies
Suitable for rare diseases (start with cases, no need for follow up)
Can look at multiple exposures at once.
Quick and cheap
Disadvantages of case control studies
Not suitable for rare exposures
Only possible to estimate relative risk, not incidence rates.
Does not tell you about the prevalence or incidence of disease
Prone to selection bias
Prone to recall bias
Confounding
Outline the steps involved in an RCT
Identify, recruit, consent and maintain two comparable groups of participants
Randomise: one group to be given a new treatment and the other the standard treatment
Follow-up: two groups followed up equally an aim to maintain all participants
Assess: using same criteria of outcome in both groups
Compare: outcomes in the two groups to see how big the difference is and whether this difference is attributable to the intervention
Explanatory trial
Uses As-treated analysis
Used to establish the physiological potency of the drug in comparison with standard drug.
Non-compliers are excluded and assessed with the control group
Problem: two groups no longer randomised. Non-compliers are likely to be different to compliers which would be related to their response. Leaves study open to confounding
Pragmatic trial
uses intention-to-treat analysis
Gives a more realistic indication of the likely impact of routine clincal use of the drug. Doctors cannot control if a patient will take the drug or not
Allows randomisation to be preserved, problems due to confounding are avoided.
Clinical equipoise
For a treatment to be considered for clinical trial, there must be reasonable uncertainty or genuine ignorance about whether it or the comparator is better.
To do otherwise would be unethical
Blind trials
Attempts to remove placebo effect
Single blind trials: either the patient, clinician or assessor does not know the treatment allocation
Double blind: two or more of the patient, clinician or assessor does not know the treatment allocation
Explain the disadvantages of non-randomised control trials and the use of historical controls
Historical controls: all patients in trial receive new treatment and compared with patients who previously received the standard treatment. Historical group may be less well defined in selection, differ in prognosis or standard treatment and have poor information on confounders. New treatment group may be subject to selection bias and reporting bias
Non-randomised control trials are where patients are selected to receive the new or standard treatment. Subject to selection bias, reporting bias and confounding.
What is a prevalence survey
Survey where people in a population are examined for the presence of a condition of interest. The proportion of the population that has the condition tells you the prevalence of the disease Also known as cross-sectional studies
Things to consider when designing a prevalence survey
Defining a case
Defining the whole population
Defining your sample
Choice of examination
Response rates
Give 4 reasons why differences may exist between different prevalence surveys
(e.g. HSE, QoF)
Difference in case definition
Time-point of survey
Proportion of the population sampled (sampling frame)
Difference in sampling technique
Differences in the sample populations (age, ethnicity)
Difference in response rate
Analysis of case control studies
Odds Ratio
Odds of exposure in cases/odds of exposure in controls
axd/bxc
Nested Case Control Studies
Combines both cohort studies and case control studies
Cases are selected from participants in a cohort that have experienced a certain event and the controls are selected from participants in the cohort who have not experienced that event.
Advantages of nested case control studies
Can calculate incidence rate
Population that the controls should be selected from is already defined
Eliminates recall bias associated with conventional case control studies.
More detailed information can be collected on the smaller group of participants compared to in a cohort study.
Cost-effective alternative to a full cohort study
Definition of a clinical trial
Planned experiment involving patients which id designed to determine the most appropriate method of treatment of future patients with a given medical condition.
Allows new treatments to be evaluated scientifically and safely
Difference between observational and intervention studies
Observational studies:
Patients receive normal medical care and then patient outcomes are observed and relationships to risk factors are assessed
Intervention studies:
Patients receive treatment in a designed experiment and the effect of the clinical intervention is assessed.
What are the advantages of random allocation in clinical trials?
Gives patients equal chance of receiving each treatment
leads to groups that are likely to have similar characteristics by chance (minimises known and unknown confounders)
reduces selection bias if patients enter a trial before randomisation
non-randomisation may result in selection bias, reporting bias and confounding between treatment groups.
What is the placebo effect?
The psychological benefit that arises from being looked after or cared for more carefully or differently, or from receiving a ‘new’ treatment’. This effect can result in an improved clinical condition and a better outcome even if the new treatment is ineffective.
How does blinding eliminate the placebo effect?
Blinding in a study means that a group of people involved in the study are prevented from knowing information which could introduce bias
Clinical trials can be single or double blinded.
Reduces placebo effect and assessor/interviewer bias
What are placebos and why are they used?
A placebo is an inert treatment which looks identical to the drug under study but has no biological effect.
Used to remove the ‘placebo effect’ i.e. the psychological benefit from being looked after better or being on a ‘new’ treatment. Also help maintain blinding and acts as a blank control tto compare an active drug vs. no drug.
Systematic Review
Assembles and critically appraises all the available data on a particular subject.
Studies are included systematically using predefined methods (inclusion and exclusion criteria) which appraise the quality of that data and help minimise bias.
Resuts from similar studies (normally RCTs) are pooled, which incrases the sample size and power of the study. Statistical analysis is then carried out on the data. (Meta analysis)
Benefits of using systematic reviews
Data from RCTs cant always be used in clinical practice because they are very restricted
Systematic reviews are rigorous, objective and reproducible
Pooling data increases the statistical power
Enables more actionable conclusions on the benefits of an intervention
Allows clinicians to make evidence-based judicious descisions
How is data synthesised in a meta analysis?
Continuous data - weighted mean difference
Continuous data from standardised scales - standardised mean difference (gives you an effect size)
Dichotomous data: relative risk
How are effect sizes interpreted?
- 2 - small effect size
- 5 - medium
- 8 - large effect size
Limitations of systematic reviews
Synthesis of the data may oversimplify important distinctions between primary studies regarding inclusion and exclusion criteria or the nature of an intervention
Reviews of similar topics may appear to reach different conclusions depending on the precise form of the review question
Systematic reviews dependent on the data currently available
Findings from systematic reviews are not always consistent with the finding of large scale RCTs
What are funnel plots?
Way of assessing whether the results of a review have been influenced by publication bias.
Trial specific effect (OR, RR) is plotted against a measure of prescision (standard error).
If the plot is symmetric, this means that there is probable no publication bias, whereas asymmetric plots suggest there is.
What is a forest plot?
A plot of the quantitative results of each study with the results of the meta analysis shown. Results of the trial are given with 95% CI
What is a regression analysis?
AIms to isolate the effect of a single determinant or causal factor on a single disease from a number of factors associated with a disease.
Eliminates the effect of cnfounding factors
Assumed that the factor most strongly related statistically is the key factor.
Not suitable for estimating effect in complex causal pathways, multifactorial diseases and assesting holistic effects.
Define number needed to treat
The average number of patients who need to be treated to prevent one additional bad outcome
The ideal NNT is 1, where everyone improves with treatment and no one improves with control. The higher the NNT, the less effective is the treatment
