1. What is an exposure? What is a disease? How do we measure them? Epidemiology matters: a new introduction to methodological foundations Chapter 3

2. Seven steps 1.Define the population of interest 2.Conceptualize and create measures of exposures and health indicators 3.Take a sample of the population 4.Estimate measures of association between exposures and health indicators of interest 5.Rigorously evaluate whether the association observed suggests a causal association 6.Assess the evidence for causes working together 7.Assess the extent to which the result matters, is externally valid, to other populations Epidemiology Matters – Chapter 12

3. 1.What is a variable? 2.What are health indicators? 3.What is an exposure? 4.Measuring exposure and disease 5.Summary Epidemiology Matters – Chapter 33

4. 1.What is a variable? 2.What are health indicators? 3.What is an exposure? 4.Measuring exposure and disease 5.Summary Epidemiology Matters – Chapter 34

5. What is a variable? A variable is any measured characteristic of individuals that differs across individuals Epidemiology Matters – Chapter 35

6. Variable examples  Age  Sex  Place of birth  Occupation  Education  Ethnicity  Cigarette smoking  Diet  Alcohol consumption  Blood pressure  Gun ownership  Diabetes  Pancreatic cancer  Depression Epidemiology Matters – Chapter 36

7. 1.What is a variable? 2.What are health indicators? 3.What is an exposure? 4.Measuring exposure and disease 5.Summary Epidemiology Matters – Chapter 37

8. What are health indicators?  Population health is often defined by the absence of the occurrence of disease  Health indicators are typically measures of the occurrence of infections, syndromes, symptoms, and biological or subclinical markers  Health indicators can be measured over the life course and include measures of, for example, disability associated with adverse health states, potential years of life lost due to an illness  Health indicators can also be positive, e.g., well-being Epidemiology Matters – Chapter 38

9. Defining health indicators 1.Binary 2.Ordinal 3.Continuous Epidemiology Matters – Chapter 39

10. Binary health indicators  Variable that takes on two values  Health outcomes: present or absent  Examples  Individual has diabetes  Individual does not have cancer  Individual has Alzheimer’s disease  Individual does not have HIV Epidemiology Matters – Chapter 310

11. Ordinal health indicators  Variable that takes on multiple (>2) graded values  Examples Individual health rating  Question: How would you rate your health?  Response options: Excellent, Good, Fair, or Poor Symptom frequency  Question: How often do you experience night sweats?  Response options: Always, Often, Rarely, or Never Ability to perform health-related activity  Question: How difficult is it to climb a flight of stairs?  Response options: Very difficult, Somewhat difficult, or Not difficult Epidemiology Matters – Chapter 311

12. Continuous health indicators  Variable with continuous response options  Examples  Age  Weeks of pregnancy  Diastolic and systolic blood pressure  Cholesterol level  Viral load  Cancer stage Epidemiology Matters – Chapter 312

13. 1.What is a variable? 2.What are health indicators? 3.What is an exposure? 4.Measuring exposure and disease 5.Summary Epidemiology Matters – Chapter 313

14. Exposure  Any measurable variable that affects or is associated with health  Variable can be from macro social environment to the molecular level  Examples  Policies and laws: areas with higher taxes on alcohol have lower alcohol consumption rates  Biological sex: Men die, on average, younger than women Epidemiology Matters – Chapter 314

15. Types of exposures 1.Acute 2.Chronic or stable 3.Time-varying Epidemiology Matters – Chapter 315

16. Acute exposures  Occur for a relatively short duration  Do not repeat  Examples  Natural disasters  Motor vehicle accident Epidemiology Matters – Chapter 316

17. Chronic exposures  Stable over time  May be present at birth  Examples  Pollution  Poverty  Policies and laws  Biological sex  Race and ethnicity  DNA sequence Epidemiology Matters – Chapter 317

18. Time-varying exposures  Vary across the life course of an individual  Examples  Diet  Exercise  Smoking  Alcohol consumption  Epidemiologists capture variation over time with different measures of exposure Epidemiology Matters – Chapter 318

19. Non-diseased Diseased Epidemiology Matters – Chapter 319

20. Smoking and exercise Epidemiology Matters – Chapter 320

21. Smoking and exercise Epidemiology Matters – Chapter 321

22. Smoking and exercise Epidemiology Matters – Chapter 322

23. Smoking and exercise Epidemiology Matters – Chapter 323

24. Smoking and exercise Epidemiology Matters – Chapter 324

25. Smoking and exercise Epidemiology Matters – Chapter 325

26. Smoking and exercise Epidemiology Matters – Chapter 326

27. Smoking and exercise Epidemiology Matters – Chapter 327

28. Summary: exposure Exposure: wide range of potential variables that individuals are ‘exposed to’  Age  Sex  Education  Water consumption  Individual attendance at lecture today Epidemiology Matters – Chapter 328

29. Summary: exposure Exposure: wide range of potential variables that individuals are ‘exposed to’  Age  Sex  Education  Water consumption  Individual attendance at lecture today Epidemiology Matters – Chapter 329 What type of exposures are these?

30. Summary: exposure Exposure: wide range of potential variables that individuals ‘exposed to’  Age continuous chronic  Sex binary chronic  Education ordinal chronic  Water consumption binary time-varying  Individual attendance at lecture today binary acute Epidemiology Matters – Chapter 330

31. 1. Duration of exposure 2. Latency and critical windows Epidemiology Matters – Chapter 331 Characterizing exposures

32. Exposure duration  Duration that individual is exposed matters for production of adverse health for certain exposures Epidemiology Matters – Chapter 332

33. Exposure duration, examples Smoking  Smoking a cigarette is unlikely to have long-term health consequences  Smoking > a pack of cigarettes per day for 40 years is likely to have long-term health consequences Trans fat  One trans fat and calorie laden meal is unlikely to affect health  Years of unhealthy eating is likely to accumulate to adversely impact health Epidemiology Matters – Chapter 333

34. Exposure timing  Timing of the exposure across the life course may also be important for the production of health  Core concepts: Latency and critical windows Epidemiology Matters – Chapter 334

35. Exposure timing, examples  Latency Low birth weight associated with the development of chronic diseases in adulthood  Critical window Extreme caloric restriction during first trimester of fetal development associated with schizophrenia development in adulthood Epidemiology Matters – Chapter 335

36. Examples, exposure timing Epidemiology Matters – Chapter 336

37. Epidemiology Matters – Chapter 337 Examples, exposure timing

38. Epidemiology Matters – Chapter 338 Examples, exposure timing

39. Epidemiology Matters – Chapter 339 Examples, exposure timing

40. Epidemiology Matters – Chapter 340 Examples, exposure timing

41. 1.What is a variable? 2.What are health indicators? 3.What is an exposure? 4.Measuring exposure and disease 5.Summary Epidemiology Matters – Chapter 341

42. Measuring exposure and disease  In previous sections we have conceptualized the exposures and health indicators of interest  Now we are interested in measuring these factors  Good measurement of variables is critical for epidemiologists Epidemiology Matters – Chapter 342

43. Measurement example  Research question  Are individuals who have depression more likely to be overweight than individuals without depression?  Measuring depression  Constellation of symptoms  Condition characterized by disabling feelings of hopelessness, sadness, and loss of interest in activities  Measuring overweight  Obesity = Body Mass Index (BMI) ≥ 30 Epidemiology Matters – Chapter 343

44. Measurement 1.Be clear about the construct being measured 2.Assess the reliability of the measures 3.Assess the validity of the measures Epidemiology Matters – Chapter 344

45. Measurement example, clarity 1.Be clear about the construct being measured  Depression: validated scale  Obesity: BMI ≥ 30 2.If measurements include respondent answered questions, make sure questions are easily interpretable, short, clear, and precise.  Instead of “Are you depressed?”  Try “In the past week have you felt happy most of the time?” Epidemiology Matters – Chapter 345

46. Reliability and validity of measures Epidemiology Matters – Chapter 346

47. Reliability and validity of measures Epidemiology Matters – Chapter 347 Not valid or reliable Valid and reliable Reliable not valid

48. Reliability and validity of measures Epidemiology Matters – Chapter 348 Not valid or reliable Scale does not work Valid and reliable Scale works perfectly Reliable not valid Scale consistently weighs people 5 pounds more than they weight

49. Dimensions of reliability  Test-retest reliability: Would the respondent answer the question similarly if asked at ≥ 2 time points?  Internal consistency: Are all the items used to assess the construct indicative of that construct?  Inter-rater reliability: Would ≥ 2 independent raters all rate the response the same? Epidemiology Matters – Chapter 349

50. Measurement validity Questions to consider when assessing validity  What is the gold standard?  What are the sensitivity and specificity of our measure as compared to the gold standard? Epidemiology Matters – Chapter 350

51. 51 Sensitivity, key question: Among those who have blood cotinine ≥300 ng/mL, what proportion report that they smoke ≥20 cigarettes per day?  ≥20 cigarettes per day self-report smokers with ≥300 ng/mL cotinine / all smokers with ≥300 ng/mL cotinine  20/(20+10)=0.67 or 67% Interpretation: 67% of people who actually smoked a pack of cigarettes in the past 24 hours reported that they smoked a pack of cigarettes in the past 24 hours Measurement, sensitivity

52. 52 Specificity, key question: Among those who have blood cotinine <300 ng/mL, what proportion report that they smoke < 20 cigarettes per day?  <20 cigarettes per day self-report smokers and <300 ng/mL cotinine / all with <300 ng/mL cotinine  168/(2+168)=0.99 or 99%. Interpretation: 99% of people who actually did not smoke a pack of cigarettes in the past 24 hours reported that they did not smoke a pack of cigarettes in the past 24 hours Measurement, specificity

53. Summary: sensitivity and specificity  Provides an assessment of the validity of our measures  Sensitivity: proportion who are accurately identified as positive on the measure  Specificity: proportion who are accurately identified as negative on the measure  Requires a gold standard Epidemiology Matters – Chapter 353

54. Measurement, validity Questions to consider when assessing validity  What is the gold standard?  What are the sensitivity and specificity of our measure as compared to the gold standard? Epidemiology Matters – Chapter 354 What if there is no gold standard?

55. 1.What is a variable? 2.What are health indicators? 3.What is an exposure? 4.Measuring exposure and disease 5.Summary Epidemiology Matters – Chapter 355

56. In summary  Conceptualization and measurement of health in populations is critical to improving population health  Health indicators are presence of disease, symptoms, syndromes, disability, wellness, quality of life, and other health-related states  Exposures are potential influences on these health- related exposures and can be acute or chronic, long or short in duration, have impact only at a critical point in human development or accumulate 56Epidemiology Matters – Chapter 3

57. Seven steps 1.Define the population of interest 2.Conceptualize and create measures of exposures and health indicators 3.Take a sample of the population 4.Estimate measures of association between exposures and health indicators of interest 5.Rigorously evaluate whether the association observed suggests a causal association 6.Assess the evidence for causes working together 7.Assess the extent to which the result matters, is externally valid, to other populations Epidemiology Matters – Chapter 157

58. epidemiologymatters.org 58Epidemiology Matters – Chapter 1