Incidence, Prevalence and 95% Confidence Intervals Tom Walker and Matt Newman
Prevelance Prevalence – proportion/percentage of current sufferers. (Point) prevalence - in winter, 4.4% of population has common cold at any given time. Period prevalence – 20% of UK adults suffer from the flu each year (12-monthly prevalence). Lifetime prevalence – Over 33% of people born today will be diagnosed with cancer during their lifetime.
Incidence Incidence is rate at which new cases appear. Formally: number of new cases per person per year. Incidence = E.g. Coventry has population people develop osteoporosis over 2 years. Incidence = (7.8 cases per 1000 people per year). People observed x years observed New cases observed
Incidence Calculation Warwick university has a population of 20,000 students. Every 4 years they produce 600 of the Uks finest doctors. What is the incidence of doctors coming out of Warwick University.
Incidence Incidence = x 4 years Incidence = 7.5 per 1000 people
95% Confidence Interval Due to sampling variation observed and true values are going to be different. We use a calculation to show how mathematically confident we are, that the true value lies within the an upper and lower interval. If we keep sampling the same population, 95% of the confident intervals we create will contain the true value.
Building 95% Confidence Intervals II For estimated proportion p the 95% CI is: [p x SE, p x SE] ^ ^ ^ LEARN!
Example: we test 100 people for blood type O. 40 people are found to have it.
Building 95% Confidence Intervals II For estimated proportion p the 95% CI is: [p x SE, p x SE] So in blood type example, 95% CI is [0.4 – 1.96 x 0.049, x 0.049] = [0.304, 0.496] ^ ^ ^
Wtf is with the standard error? You don’t have to learn the standard error calculation! It’s a mathamatical calculation to discern how spread out the sample is from the mean. The larger the standard error, the further from the true value an observed value can be. Therefore a bigger standard error means wider interval.
Null Hypothesis The null hypothesis is the hypothesis of no effect. I.e, our results show no relationship between cause and effect. If the observed value falls outside the confidence interval, we can reject the null hypothesis And therefore the result is statistically significant. If the observed value falls inside the confidence interval, we can’t reject the null hypothesis, and the result is statistically insignificant.