1 / 29

HaDPop

HaDPop. Measuring Disease and Exposure in Populations (MD) & Introduction to Medical Statistics (MS). Overview. Prevalence (MD) Incidence (MD) Confidence Intervals (MS) Standard Error Error Factor Null Hypothesis (MS) P-values (MS) Ratios (MD) Confounders (MD) SMR (MD). Prevalence.

chase
Download Presentation

HaDPop

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. HaDPop Measuring Disease and Exposure in Populations (MD) & Introduction to Medical Statistics (MS)

  2. Overview • Prevalence (MD) • Incidence (MD) • Confidence Intervals (MS) • Standard Error • Error Factor • Null Hypothesis (MS) • P-values (MS) • Ratios (MD) • Confounders (MD) • SMR (MD)

  3. Prevalence No. of existing cases / No. of persons in the population • A measure of how much of a disease there is (both new and old cases) • Period and point prevalence • It gives a proportion of the population • Useful for studying long term conditions and service provision

  4. Prevalence Example • In a hypothetical office (total 1000 people), 12 were off work with the flu today. • Point Prevalence (today): 12/1000 = 1.2% • Over the past year, 150 took off work due to flu. • Period Prevalence (past year): 150/1000 = 15%

  5. Incidence No. of new cases / In a defined population in a specified time interval (person-years) • A measure of the frequency of new cases (it is a rate) • Useful for tracking infectious diseases and exploring the cause of disease (aetiology) • Person years

  6. Incidence Example • Over the last 5 years, 4000 people have been diagnosed with lung cancer (total population: 200,000) • 4000/(200,000 x 5) = 0.004 or 4 per 1000 per year

  7. Types of Incidence Numerator Disease free - 100 Denominator New Cases = 10 Non Cases = 90 • Incidence Rate – different length of follow up • Cumulative Incidence (or risk) – same length of follow up • Odds of Disease – ratio between having the disease or not Time (t)

  8. Medical Statistics • Statistics are used to estimateinformation about the general population (its not practical to measure everyone!) • This estimate is the known as the observed value and this varies from the true value due to sampling variation • The accuracy of an estimate is calculated using confidence intervals

  9. Confidence Intervals (CI) • The confidence interval is a range of values around the observed value within which the true value lies • The most common range used is the 95%CI, which means 95 times out of 100 the real value will be within that range • The way the are calculated is different depending on the statistics you are using • Proportions (i.e. prevalence) use standard error (SE) • Ratios/rates (i.e. incidence) use error factors (EF)

  10. Standard Error (Prevalence) • Note Accuracy Depends on Sample Size Note 1.96 is a constant used for working out 95% CI’s It changes if you want different CI’s

  11. Standard Error Example • Prevalence of diabetes, sample of 1000 subjects (n = 1000), 243 found to have diabetes (k = 243) • Prevalence = k/n, so = 243/1000 = 24.3% • Standard error (SE) , so = = 0.013

  12. Standard Error Example Cont. • SE = 0.013 • Original prevalence estimate (): 24.3% population had diabetes • = (0.243 – 1.96(0.013), 0.243 + 1.96(0.013) • = (0.218, 0.268) = (21.8%, 26.8%)

  13. Difference between two prevalences

  14. Error Factor (Incidence) • Note Accuracy Depends on No. cases

  15. Error Factor Example • 24 new cases of diabetes per 1000 population per year (i.e. d = 24) • Error factor = exp(1.96 x ) • = exp(1.96 x ) = 1.5 • = (0.0024/1.5, 0.0024 x 1.5) • = (0.0016, 0.0036) • = (16, 36) cases per 1000 p-y

  16. Key Points for CI’s • Proportions (prevalence) • 95%CI = “Estimate ± (‘constant’ x SE)” • Rates/ratios (incidence and SMR’s) • 95%CI= (Estimate/EF , Estimate x EF) • How to calculate the Standard error or Error factor will be given in the exam

  17. Null Hypothesis (H0) • This is used to make a comparison between different groups to see if there is a statistical difference between them • E.g. differences between different drugs • The null hypothesis is when there is no statistical difference between the two groups • Differences – null hypothesis is 0 • Ratios – null hypothesis is 1 • SMR – null hypothesis is 100

  18. Null Hypothesis • If the 95% CI includes the null hypothesis then the data agrees with the null hypothesis can’t be rejected and there is no statistical difference between the two groups • You can never accept the null hypothesis!

  19. P-values • p-values state how likely the results in the study would have occurred by chance if the null hypothesis was true • P-values <0.05 (5%) are good! They mean that the results are statistically significant and that the null hypothesis can be rejected • If the 95%CI overlap with the null hypothesis then p>0.05 and the results are not statistically significant

  20. Relative measures of exposure(Relative risk) see slide 7 • Note – an exposure can be to a treatment, therefore it can be used to find out which treatments are best

  21. Incidence Rate Ratio Example • In one group of 1000 pizza eaters that were followed for 1.5 years (‘exposed’) it was found that 33 new cases (d1) of obesity developed • In another group of 1000 non-pizza eaters that were followed for 2 years (unexposed) it was found that 27 new cases (d2) of obesity developed

  22. Incidence Rate Ratio Example Cont. • In the exposed group: 33/(1000 x 1.5) = 0.022 (or 22 per 1000 pop. per year) • So in the unexposed group: 27/(1000 x 2) = 0.014 (or 14 per 1000 pop. per year) • Error factor = exp(1.96 x ) • = exp(1.96 x ) = 1.66

  23. Incidence Rate Ratio Example Cont. • Estimate (IRR) = 1.57 • EF = 1.66 • 95% CI = (est / EF, est x EF) • = (1.57/1.66, 1.57x1.66) = (0.95, 2.61) • IRR = 1.57 (0.95, 2.61)

  24. Incidence Rate Ratio Example Cont.Interpretation of results • IRR of 1.57 indicates that the observed value indicates a damaging effect of eating pizza on becoming obese on (i.e. >1) • We are 95% confident that the true IRR lies between 0.95 and 2.61. • The 95% confidence interval includes the null hypothesis (IRR=1) and so the result is not statistically significant at the p<0.05 level. • Null hypothesis cannot be rejected. • The results do not indicate an association between eating pizza and obesity.

  25. Absolute Measures of Risk(attributable risk) • Risk difference = risk exposed - risk non-exposed • Attributable Risk = IR exposed - IR non-exposed = events saved per 1,000 • Attributable Risk (%) = Attributable Risk / IR exposed

  26. Examples of attributable risk? • People with lung cancer can be smokers and non smokers • Thus the attributable risk of smoking is the difference between the incidence of smokers and non smokers • Ie the attributable risk is the risk above background risk (the non smokers with lung cancer have suffered from the background risk)

  27. Confounders • A Confounder is a factor that is associated with the exposure under study and independently affects disease risk

  28. Standardised mortality ratio (SMR) • Compares the observed number of deaths in the population under study with the expected number of deaths based on the standard population • It accounts for common confounders such as age and gender • Uses error factors for 95% CI

  29. Thanks for listening • Any questions please Facebook me!!!

More Related