making measurements meaningful
play

Making Measurements Meaningful Brian J. Zikmund-Fisher, PhD - PowerPoint PPT Presentation

Making Measurements Meaningful Brian J. Zikmund-Fisher, PhD University of Michigan Department of Health Behavior & Health Education Department of Internal Medicine Center for Bioethics & Social Sciences in Medicine Health Informatics


  1. Making Measurements Meaningful Brian J. Zikmund-Fisher, PhD University of Michigan Department of Health Behavior & Health Education Department of Internal Medicine Center for Bioethics & Social Sciences in Medicine Health Informatics Program @bzikmundfisher

  2. Assumptions In Science Communication Assumption: People Lack Information (The Deficit Model) Assumed Solution: Give them more!

  3. Language Numeracy Literacy Barriers Emotionality of Volume of Health Relevant Situations Information

  4. More ≠ Better!

  5. Evaluability

  6. Good or Bad? Amend B. Welcome to Jasorassic Park, 1998, p.36.

  7. Good or Bad? Amend B. Welcome to Jasorassic Park, 1998, p.36.

  8. Evaluability of Risk Information

  9. Imagine Robert

  10. Imagine Robert Your 10-year risk of cardiovascular disease is: 11.22%

  11. “Am I at high risk, or not?”

  12. Evaluability of Laboratory Test Results

  13. Can Patients Use This?

  14. What Is Out of Range?

  15. What Is Out of Range?

  16. “Am I at high risk, or not?”

  17. Evaluability of Exposure Information

  18. Buying a Home… Radon: 6 pCi/L

  19. Drinking Water… Lead: 7 ppb

  20. “Am I at high risk, or not?”

  21. Problem #1: Numbers

  22. Problem #2: Lack of Meaning

  23. So now what? What can we do to help?

  24. Step 1: Visual information

  25. Robert’s Risk Created at iconarray.com

  26. Tables Zikmund-Fisher BJ, et al. Graphics help patients distinguish between urgent and non-urgent deviations in laboratory test results. Journal of the American Medical Informatics Association 2017;24(3):520-528.

  27. Table vs. Number Line Zikmund-Fisher BJ, et al. Graphics help patients distinguish between urgent and non-urgent deviations in laboratory test results. Journal of the American Medical Informatics Association 2017;24(3):520-528.

  28. Lines with More Meaning Zikmund-Fisher BJ, et al. Graphics help patients distinguish between urgent and non-urgent deviations in laboratory test results. Journal of the American Medical Informatics Association 2017;24(3):520-528.

  29. Near-Normal Results vs. Extreme Results Near-Normal Extreme Zikmund-Fisher BJ, et al. Graphics help patients distinguish between urgent and non-urgent deviations in laboratory test results. Journal of the American Medical Informatics Association 2017;24(3):520-528.

  30. % with No Difference in Perceived Urgency Platelets ALT Creatinine (135 vs 25 x10 9 /L) (80 vs 360 U/L) (2.2 vs 3.4 mg/dl) Table 26.5 56.3 43.7 Simple Line 17.5 21.3 27.7 Block Line 19.0 20.2 28.7 Gradient 15.8 14.8 24.0 Line Zikmund-Fisher BJ, et al. Graphics help patients distinguish between urgent and non-urgent deviations in laboratory test results. Journal of the American Medical Informatics Association 2017;24(3):520-528.

  31. Step 2: Gist-full information

  32. Fuzzy Trace Theory (Brainerd and Reyna, 1995) Verbatim Gist vs memory

  33. Fuzzy Trace Theory (Brainerd and Reyna, 1995) Verbatim Gist vs memory What the heck is “gist”?

  34. Know Your “Commander’s Intent” Heath & Heath, Made to Stick , 2007

  35. Cancer Screening Test Decisions

  36. Colorectal Cancer Screening

  37. Colorectal Cancer Screening … at age 50

  38. Colorectal Cancer Screening … at age 50 … vs. at age 75 (with multiple comorbidities)

  39. Benefits vs. Harms Image from study materials for “Promoting Veteran-Centered Colorectal Cancer Screening” (I01 HX001278-01); SD Saini, PI.

  40. Gist Processing Image from study materials for “Promoting Veteran-Centered Colorectal Cancer Screening” (I01 HX001278-01); SD Saini, PI.

  41. Setting the Context in Visual Displays

  42. Hemoglobin A1c Unpublished graphic from 1 R01 HS021681, BJ Zikmund-Fisher, Principal Investigator.

  43. Same (?!?) Result Unpublished graphic from 1 R01 HS021681, BJ Zikmund-Fisher, Principal Investigator.

  44. Scale Matters Unpublished graphics from 1 R01 HS021681, BJ Zikmund-Fisher, Principal Investigator.

  45. Test Results for Diagnosed Patients

  46. Goals for Test Results Scherer AM, Witteman HO, Solomon J, Fagerlin A, Exe NL, Zikmund-Fisher BJ. Improving understanding of test results by substituting (not adding) goal ranges. Poster presentation to the Society for Medical Decision Making, Vancouver, BC, Canada, October 23, 2016.

  47. Goals for Test Results Scherer AM, Witteman HO, Solomon J, Fagerlin A, Exe NL, Zikmund-Fisher BJ. Improving understanding of test results by substituting (not adding) goal ranges. Poster presentation to the Society for Medical Decision Making, Vancouver, BC, Canada, October 23, 2016.

  48. Goals for Test Results Scherer AM, Witteman HO, Solomon J, Fagerlin A, Exe NL, Zikmund-Fisher BJ. Improving understanding of test results by substituting (not adding) goal ranges. Poster presentation to the Society for Medical Decision Making, Vancouver, BC, Canada, October 23, 2016.

  49. Goals for Test Results Scherer AM, Witteman HO, Solomon J, Fagerlin A, Exe NL, Zikmund-Fisher BJ. Improving understanding of test results by substituting (not adding) goal ranges. Poster presentation to the Society for Medical Decision Making, Vancouver, BC, Canada, October 23, 2016.

  50. Test Results for Monitoring

  51. Harms Alanine Aminotransferase (ALT): 80 IU/L Standard Range: 10-40

  52. Showing the Possible Range Zikmund-Fisher BJ, Scherer AM, Witteman HO, Solomon J, Exe NL, Fagerlin A. Providing harm anchors in visual displays of test results can mitigate patient perceptions of urgency about near-normal values. Journal of Medical Internet Research, 2018.

  53. Harm Anchors Zikmund-Fisher BJ, Scherer AM, Witteman HO, Solomon J, Exe NL, Fagerlin A. Providing harm anchors in visual displays of test results can mitigate patient perceptions of urgency about near-normal values. Journal of Medical Internet Research, 2018.

  54. Harm Anchors “Many doctors are not concerned until here” Zikmund-Fisher BJ, Scherer AM, Witteman HO, Solomon J, Exe NL, Fagerlin A. Providing harm anchors in visual displays of test results can mitigate patient perceptions of urgency about near-normal values. Journal of Medical Internet Research, 2018.

  55. Increased Sensitivity with Harm Anchors 6 5 Perceived Alarm 4 3 2 1 Near Value Extreme Value Simple Harm Anchor Zikmund-Fisher BJ, Scherer AM, Witteman HO, Solomon J, Exe NL, Fagerlin A. Providing harm anchors in visual displays of test results can mitigate patient perceptions of urgency about near-normal values. Journal of Medical Internet Research, 2018.

  56. One Last Example

  57. Hemoglobin A1c 7.7 7.6 7.5 7.4 Percent 7.3 7.2 7.1 7 0 5 10 15 20 Months

  58. Hemoglobin A1c 7.7 7.6 Central 7.5 Message: 7.4 Percent Varies 7.3 7.2 7.1 7 0 5 10 15 20 Months

  59. 7.7 7.6 7.5 Percent 7.4 7.3 7.2 7.1 7 0 10 20 Months

  60. 12 10 Percent 8 6 7.7 7.6 = ? 4 7.5 0 10 20 Percent 7.4 Months 7.3 7.2 7.1 7 0 10 20 Months

  61. 12 10 Percent 8 6 7.7 7.6 = ? 4 7.5 0 10 20 Percent 7.4 Months 7.3 7.2 7.1 8 7 0 10 20 6 Percent Months 4 2 0 0 10 20 Months

  62. 12 10 Percent 8 Varies! 6 7.7 7.6 = ? 4 7.5 0 10 20 Percent 7.4 Months 7.3 7.2 7.1 8 7 0 10 20 6 Percent Months 4 2 0 0 10 20 Months

  63. 12 Stable! 10 Percent 8 Varies! 6 7.7 7.6 = ? 4 7.5 0 10 20 Percent 7.4 Months 7.3 7.2 7.1 8 7 0 10 20 6 Percent Months 4 2 0 0 10 20 Months

  64. 12 Stable! 10 Percent 8 Varies! 6 7.7 7.6 = ? 4 7.5 0 10 20 Percent 7.4 Months 7.3 7.2 7.1 8 7 0 10 20 6 High! Percent Months 4 2 0 0 10 20 Months

  65. 12 Stable! 10 Percent 8 Varies! 6 7.7 7.6 = ? 4 7.5 0 10 20 Percent 7.4 Months 7.3 7.2 7.1 8 7 0 10 20 6 High! Percent Months 4 Scaled to data variations 2 0 0 10 20 Months

  66. Scaled to 12 population Stable! variations 10 Percent 8 Varies! 6 7.7 7.6 = ? 4 7.5 0 10 20 Percent 7.4 Months 7.3 7.2 7.1 8 7 0 10 20 6 High! Percent Months 4 Scaled to data variations 2 0 0 10 20 Months

  67. Scaled to 12 population Stable! variations 10 Percent 8 Varies! 6 7.7 7.6 = ? 4 7.5 0 10 20 Percent 7.4 Months 7.3 7.2 7.1 8 7 0 10 20 6 High! Percent Months 4 Scaled to data variations 2 Scaled to zero 0 0 10 20 Months

  68. 12 11 10 Good 9 or 8 Bad? 7 6 5 4 0 5 10 15 20

  69. 12 11 10 Good 9 or 8 Bad? 7 6 5 Standard Range 4 0 5 10 15 20

  70. 12 11 10 Good 9 or 8 Bad? 7 Target Range for Patients with Type 2 Diabetes 6 5 4 0 5 10 15 20

  71. My Commander’s Intent

  72. “We need to design for the way people ARE, not the way we wish they were” - Holly O. Witteman

  73. People only process or remember one thing

  74. You can’t change this fact!

  75. BUT…

  76. BUT… You get to choose!

  77. Use context to create ONE message based on THEIR needs

  78. Thank You! bzikmund@umich.edu @bzikmundfisher

Recommend


More recommend