Intervention Studies - Critical Appraisal Flashcards
What should be considered when appraising the methods of an intervention study?
- Were the groups comparable?
- Was there blinding of patients and assessors?
- Was there complete or near complete follow- up?
- Were the patients selected appropriately?
- Was the intervention applied appropriately?
- Was the comparison intervention appropriate?
- How many correctly guessed which intervention they received?
- Were the outcome measures appropriate?
- Are the participants similar to patients seen in the clinic?
What is required to ensure treated and control groups are comparable?
- Random allocation of subjects to groups
- Randomisation must be concealed from the researcher and the subject
Why is blinding of patients and assessors important?
- Blinding of patients ensures the effect of intervention is not due to placebo effects
- Sham intervention generally used to ensure blinding of patients
- Blinding of assessors ensures no measurement bias - Blinding of providers of care and statisticians will increase the validity of a study further
Why is it important to consider whether there was complete or near-complete follow-up?
- Drop outs in one group may systematically differ in terms of outcomes with dropouts in the other group
- Dropouts of <10% are unlikely to threaten the validity of the study
- Studies that don’t provide data on drop outs should be considered potentially biased
- Analysis by intention to treat should be used to preserve the benefits of randomisation
What should be considered when appraising if patients were selected appropriately?
- Inclusion/exclusion criteria that determine patient characteristics
- But this may make the research inappropriate or may impact on who the results of the research can be generalised to
What should be considered when appraising if the intervention was applied appropriately?
- Content
- Intensity
- Duration
- Frequency
- Time to follow-up
What should be considered when appraising if the outcome measures were appropriate?
- Valid & reliable
- Linked to the intervention
- Clinically meaningful
What should be considered when appraising the results of an intervention study?
- Does the therapy do more good than harm?
- If the intervention provides a clinically worthwhile effect that outweighs the costs/risks of the intervention then it should be considered
What should be considered when appraising the effect size?
- P value
- Continuous outcomes
- Dichotomous outcomes
What is the p value?
- P value tells us if differences between groups are greater than would occur by chance alone
- P value does not tell us if the effects are large enough to be clinically worthwhile
What is a continuous outcome?
- The amount of an outcome is measured
- E.g. Pain scale, gait speed, strength, range of motion
What is a dichotomous outcome?
- Did the event happen? (yes or no)
- E.g. walking independently, nursing home admission, recurrence of pain, death
What is the sampling theory?
- A study statistic, such as the mean, is an estimate of the corresponding population parameter
- It is an estimate because the true value of the population is almost always unknown
What is continuous data?
- The best estimate of the size of the treatment effect is the difference between group means (or medians)
- This could be the difference between means after the treatment period or the difference in mean changes over the treatment period
What is dichotomous data?
- Presented as “Risk” (likelihood)
- Proportion of people in each group who develop the outcome (also called event rate)
