Epi definitions Flashcards
What does the term ‘statistical significance’ mean?
Statistical significance indicates the probability that the data patterns reflect a true relationship within the population, and are not due to random chance
What is a p-value?
The probability of obtaining the observed result if the null hypothesis is true (that there is no real relationship in the population being studied)
What is a confidence interval?
A confidence interval is a range of value calculated from sample data that is likely to contain the true population paramter, with specified level of confidence, expressed as a percentage (such at 95%)
It conveys how precise the measurement is
What is a null hypothesis?
The null hypothesis is a statistical assumption that there is no significant difference between specified populations or variables being compared in a study.
What is a normal distribution?
A symmetical probability distribution in which the frequency of data is characterised by a bell shaped curve, with the majority of observations clustered around the mean, and data points have an equal chance of being above or below the mean
Best suited for continuous data
What is a Poisson distribution?
A statistical distribution that describes the number of events that occur within a fixed interval of time or space, given a known average rate of occcurence, and assuming that the events happen independently.
It applies to when events occur at a constant average rate
Imagine you’re at a bus stop, and you know on average a bus arrives every 15 minutes. The Poisson distribution helps us understand the likelihood of certain events happening within a specific time frame, given that average rate.
For example, it helps answer questions like:
What’s the chance of seeing 2 buses arrive in the next 30 minutes?
How likely is it that no bus arrives in the next 10 minutes?
The Poisson distribution gives us a way to calculate these probabilities based on the average rate of events (in this case, bus arrivals) and the time frame we’re interested in. It’s handy for situations where events happen independently at a constant average rate over time or space, like phone calls to a call center, arrivals at a restaurant, or accidents in a city.
Define sensitivity (of a test)?
the ability of the test to identify people with condition as positive
= true positive/ true positive+false negative (i.e how many true positive tests out of all the real positive cases)
Define specificity (of a test)?
the ability of the test to identify healthy people as negative
= true negatives/true negative + false positive (i.e how many true negative tests out of all the real negative cases)
Define positive predictive value?
the likelihood that the participant has the condition when the test is positive
= true positives/ true positives + false positives (i.e how many true positive tests out of all the positive tests)
Define negative predictive value?
the likelihood that the participant has the condition with the test is negative
= true negative/ true negative + false negative (i.e how many true negatives tests out of all the negative tests)
What does positive predictive value and false positives depend on?
They depend on the prevalence of disease. Low prevalence means false positives are much more likely, which reduces the PPV
What is lead time bias (in screening)?
Disease detected by screening before would have been detected by symptoms – doesn’t actually live longer but looks like it
What is length time bias (in screening)?
Screening more likely to pick up more slowly progressing conditions so it looks like screening is improving survival but just picking up more slow disease, ie outcomes look better compared to those picked up with symptoms.
What is the Absolute Risk Reduction (AAR) or risk difference/attributable risk?
Control event rate - Experimental Event rate
How do you calculate the NNT? How many people need to get intervention to change one outcome?
1/AAR (with AAR being control event rate less experimental event rate)
Convert %s to 0.x
