Presentation on theme: "Hadpop Calculations. Odds ratio What study applicable? Q. It is suggested that obesity increases the chances on an individual becoming infected with erysipelas."— Presentation transcript:
Odds ratio What study applicable? Q. It is suggested that obesity increases the chances on an individual becoming infected with erysipelas. A sample of 165 obese people was taken with 68 having suffered previously with erysipelas. This compares with a sample including 257 normal people with 61 having been infected. Construct a table and work out the odds ration of developing erysipelas as a consequence of obesity. Interpret your result.
Odds ratio Odds = A x D / B x C =68x197 / 97x 61 =2.3 This is greater than 1, and we can therefore conclude that being exposed to this infection increases the odds of developing obesity. Case control study – identify outcome, people with infection, compare exposure obesity. E.g (HAVE AIDS, DO ANAL?) erysipelasNo erysipelas obese68 (A)97 (B) normal61 (C)197 (D)
IRR What study? What purpose? Over a 3 year period 1500 former employee’s of ‘asbestos uk’ were followed up. They were separated into 2 cohorts, 1000 who were exposed to aspestos, 500 who weren’t. In the exposed cohort 600 developed mesothelioma, in the unexposed 3 developed mesothelioma. This compares to the national average of 120 people per year per 1000. Q. Calculate the IRR and interpret.
IRR Cohort Identify exposure – compare outcome, e.g (ANAL? – AIDS) Exposed = 3 x 1000 = 3000 people years Not = 3 x 500 = 1500 py IRR exp, 600/3000 = 200 per 1000 py IRR not exp, 3 / 1500 = 2 per 1000 py, 200/2 Answer = 100 Greater than 1, is a much higher incidence
SMR What purpose? Q.120 people die in Scotland from cardiac related problems. This compares with an average of 33. Calculate the SMR and interpret.
SMR Ans = 120 x 100 / 33 =363.6 Greater than 100, so we can conclude that living in Scotland's increases your risk of cardiac related mortality.
Causality Identifies the chance that an exposure caused a disease. Using the IRR from the previous question calculate the confidence intervals for this study. Interpret Error factor = 3, (a small study has a big error factor) Hint Ub, Lb,
CI UB = er x ratio, =3 x 100, 300 LB = ratio / er, = 100/3, 33
Generic things to say about C95% I can say with 95% certainty that the true underlying value lies between x – y, the range makes me think... The value for the null hypothesis is 1, this is included so is a possible value, p>0.05 and i fail to reject the null hypothesis. The result is not statistically significant 1 is just included in the range, and so there is a tendency to significance 1 is not in the range, i can reject the null hypothesis, p<0.05 the results are significant
This example Range is 300-33. Large range, say with 95% certainty true value lies in this range. Does not include 1, reject the null, p<0.05 the results are significant