In the history of CICM exams, this has only come up once: in Question 23 from the second paper of 2008, where we were called upon to define the types of error.
- Results from a lack of precision in study conduct
- Reduced by meticulous technique and large sample size
Type 1 error
- This is a "false positive".
- The null hypothesis is incorrectly rejected (i.e. there really is no treatment effect, but the study finds one)
- The alpha value determines the risk of this happening. An alpha value of 0.05 - same as the p-value - so there is a 5% chance of making a Type 1 error.
Type 2 error
- This is a "false negative"
- The null hypothesis is incorrectly accepted (i.e. there really is a treatment effect, but you fail to find it)
- The power (1-beta) determines the risk of this happening. Thus at a beta of 0.2, there is a 20% chance of making a Type 2 error.
Minimisation of error
- Construct a well-designed RCT, which is the gold standard of research
- Choose an appropriate sample size to power the study
- Choose an appropriate effect size (which is an arbitrary decision, but one made rationally)
- Minimise bias by blinding and allocation concealment
- Conduct a sequential trial with a built-in interim analysis at a predetermined point in the study
- Use correct tests to analyse the data
- Report mean with standard deviation
- Dont assume statistical significance equates to clinical significance
- Give explicit details of the study design and statistical analysis
- Publish even the negative results