1 / 91

Causation?

Causation?. Tim Wiemken, PhD MPH CIC Assistant Professor Division of Infectious Diseases University of Louisville, Kentucky. Overview. 1. Testing for an Association. 2. Other Measures of Association. 3. Confidence Intervals. Overview. 1. Testing for an Association.

monet
Télécharger la présentation

Causation?

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. Causation? Tim Wiemken, PhD MPH CIC Assistant ProfessorDivision of Infectious Diseases University of Louisville, Kentucky

  2. Overview 1. Testing for an Association 2. Other Measures of Association 3. Confidence Intervals

  3. Overview 1. Testing for an Association 2. Other Measures of Association 3. Confidence Intervals

  4. Testing for Association 1. Develop hypothesis Null hypothesis: There is no association Alternative hypothesis: There is an association

  5. Testing for Association 1. Develop hypothesis

  6. Testing for Association 2. Choose your level of significance α value What P-value will you consider statistically significant? Usually 0.05 - arguments for bigger/smaller

  7. Testing for Association 3. Choose Your Test Call your statistician. • A bad test gives bad results. • A good test may give bad results (bad data?). • A good statistician may tell you if the results are bad, but cannot always tell you if the data were bad.

  8. Testing for Association Chi-squared Test Will tell you if there is an association between two variables

  9. Testing for Association Chi-squared Test Will tell you if there is an association between two variables Measures observed versus expected counts in study groups

  10. Testing for Association Chi-squared Test Will tell you if there is an association between two variables Measures observed versus expected counts in study groups Must have adequate sample size

  11. Testing for Association Chi-squared Test 2x2 table – categorical data

  12. Example

  13. Example Hospitalized CAP Patients HIV+ HIV- Dead Alive Dead Alive Does HIV Have an Effect on Patient In-Hospital Mortality?

  14. Example Does HIV Have an Effect on Patient In-Hospital Mortality? Null Hypothesis Significance Level What Test?

  15. Example Does HIV Have an Effect on CAP Patient In-Hospital Mortality? Assuming a cohort study… + HIV: 118 - HIV: 2790 + HIV + died : 12 - HIV + died : 257

  16. Example Does HIV Have an Effect on Patient In-Hospital Mortality? Assuming a cohort study…

  17. Example Does HIV Have an Effect on Patient In-Hospital Mortality? Assuming a cohort study… + HIV: 118 - HIV: 2790 + HIV + died : 12 - HIV + died : 257

  18. Example Does HIV Have an Effect on Patient In-Hospital Mortality? Assuming a cohort study…

  19. Example Do they? No! P>0.05

  20. Example Where to publish?

  21. Example

  22. Example CAP Patients with H1N1 Obese (BMI ≥30) Lean (BMI <30) Dead Alive Dead Alive Does Obesity Have an Effect on Patient Mortality?

  23. Example Is obesity related to mortality in CAP patients? Null Hypothesis Significance Level What Test?

  24. Example Do obese patients with CAP due to H1N1 die more than lean patients? Assuming a cohort study… + Obese: 1004 + Lean: 2530 + Obese + died : 317 + Lean + died : 509

  25. Example Do obese patients with CAP due to H1N1 die more than lean patients? Assuming a cohort study…

  26. Example Do obese patients with CAP due to H1N1 die more than lean patients? Assuming a cohort study… + Obese: 1004 + Lean: 2530 + Obese + died : 317 + Lean + died : 509

  27. Example Do obese patients with CAP due to H1N1 die more than lean patients? Assuming a cohort study…

  28. Example Do they? Yes! P≤0.05

  29. Overview 1. Testing for an Association 2. Other Measures of Association 3. Confidence Intervals

  30. Measures of Association 1. Risk Ratio Used for cohort studies or clinical trials Gold standard measure for observational studies Answers: How much more (less) likely is this group to get an outcome versus this other group?

  31. Example How much more likely is an obese person with CAP due to 2009 H1N1 to die than a lean person?.

  32. Example How much more likely is an obese person with CAP due to 2009 H1N1 to die than a lean person?. Risk of death in obese group: 317 / 317+687= 0.316

  33. Example How much more likely is an obese person with CAP due to 2009 H1N1 to die than a lean person?. Risk of death in obese group: 317 / 317+687= 0.316 Risk of death in lean group: 509 / 509+2021= 0.201

  34. Example How much more likely is an obese person with CAP due to 2009 H1N1 to die than a lean person?. Risk of death in obese group: 317 / 317+687= 0.316 Risk of death in lean group: 509 / 509+2021= 0.201 Risk Ratio: 0.316 / 0.201 = 1.57

  35. Example Interpret the Risk Ratio Who wants to interpret a risk ratio of 1.57?

  36. Example Obese patents with CAP due to 2009 H1N1 are 57% (1.57 times) more likely to die than lean patients. Interpret the Risk Ratio

  37. Example If the mortality rate for lean patients is 8%, the mortality rate for obese patients is 57% higher than this: 8%*1.57 = 12.6% Interpret the Risk Ratio

  38. Example

  39. Example CAP Patients Empiric Atypical Pathogen Coverage No Empiric Atypical Pathogen Coverage Dead Alive Dead Alive Does Empiric Atypical Pathogen Coverage Have an Effect on Patient Mortality?

  40. Example Do those patients who have empiric atypical pathogen coverage die less often than those without atypical coverage? Assuming a cohort study… + Atypical : 2220 - Atypical : 658 + Atypical + died : 217 - Atypical + died : 110

  41. Example Do those patients who have atypical pathogen coverage die more often than those without atypical coverage? Assuming a cohort study…

  42. Example Do those patients who have empiric atypical pathogen coverage die less often than those without atypical coverage? Assuming a cohort study… + Atypical : 2220 - Atypical : 658 + Atypical + died : 217 - Atypical + died : 110

  43. Example Do those patients who have atypical pathogen coverage die more often than those without atypical coverage? Assuming a cohort study…

  44. Example Interpret the Risk Ratio Anyone??

  45. Example Interpret the Risk Ratio Those with atypical coverage are 42% less likely to die as compared to those without atypical coverage

  46. Example What does that mean? 8% x 0.58 = 4.64 Just multiply original risk by the risk ratio!

  47. Example Even Better: Number Needed to Treat 1/Absolute Risk Reduction (ARR) ARR = Unexposed Risk – Exposed Risk

  48. Example Even Better: Number Needed to Treat ARR = Unexposed Risk – Exposed Risk ARR = Risk w/out atypical coverage – Risk w/atypical coverage

  49. Example Even Better: Number Needed to Treat

  50. Example Even Better: Number Needed to Treat 16.7 = unexposed risk

More Related