Using Nudges to Improve the Delivery of Health Care Mitesh Patel, - - PowerPoint PPT Presentation

▶
using nudges to improve the
SMART_READER_LITE
LIVE PREVIEW

Using Nudges to Improve the Delivery of Health Care Mitesh Patel, - - PowerPoint PPT Presentation

Oct 24, 2022 255 likes •1.53k views

Using Nudges to Improve the Delivery of Health Care Mitesh Patel, MD, MBA, MS February 2020 Director, Penn Medicine Nudge Unit; Ralph Muller Presidential Assistant Professor, Perelman School of Medicine and The Wharton School, University of

slide-1
SLIDE 1

Using Nudges to Improve the Delivery of Health Care

Mitesh Patel, MD, MBA, MS

February 2020

Director, Penn Medicine Nudge Unit; Ralph Muller Presidential Assistant Professor, Perelman School of Medicine and The Wharton School, University of Pennsylvania; Staff Physician, Crescenz Veterans Affairs Medical Center, Philadelphia

Nudgeunit.upenn.edu @PennNudgeUnit

slide-2
SLIDE 2

Disclosures

  • Research funding

– NIH, VA, Donaghue Foundation, Doris Duke Charitable Foundation, Deloitte Consulting

  • Advisory Board Member

– Life.io, Healthmine Services, Holistic Industries

  • Consulting

– Catalyst Health – www.miteshspatel.com

slide-3
SLIDE 3

Human behavior is the final common pathway

Patel, Volpp, Asch. NEJM. 2018

slide-4
SLIDE 4

Human behavior is the final common pathway

Patel, Volpp, Asch. NEJM. 2018

slide-5
SLIDE 5

Human behavior is the final common pathway

Patel, Volpp, Asch. NEJM. 2018

slide-6
SLIDE 6

Human behavior is the final common pathway

Patel, Volpp, Asch. NEJM. 2018

slide-7
SLIDE 7

Human behavior is the final common pathway

Patel, Volpp, Asch. NEJM. 2018

slide-8
SLIDE 8

Medical decision-making is often suboptimal

slide-9
SLIDE 9

Medical decision-making is often suboptimal

  • Sometimes we do too much

– Nearly one-third of health care spending is wasteful and unnecessary

slide-10
SLIDE 10

Medical decision-making is often suboptimal

  • Sometimes we do too much

– Nearly one-third of health care spending is wasteful and unnecessary

  • Sometimes we don’t do enough

– We often fail to practice evidence-based medicine

slide-11
SLIDE 11

Medical decision-making is often suboptimal

  • Sometimes we do too much

– Nearly one-third of health care spending is wasteful and unnecessary

  • Sometimes we don’t do enough

– We often fail to practice evidence-based medicine 90% of clinicians

slide-12
SLIDE 12

Medical decision-making is often suboptimal

  • Sometimes we do too much

– Nearly one-third of health care spending is wasteful and unnecessary

  • Sometimes we don’t do enough

– We often fail to practice evidence-based medicine 90% of clinicians 80% of US adults

slide-13
SLIDE 13

Using nudges to improve our decisions

  • Subtle changes in design that can have an outsized impact on our behavior
  • Remind, guide, or motivate decisions

Patel, Volpp, Asch. NEJM. 2018

slide-14
SLIDE 14

Using nudges to improve our decisions

  • Subtle changes in design that can have an outsized impact on our behavior
  • Remind, guide, or motivate decisions
  • Examples

– Setting the default to the preferred option – Prompting an active choice now rather than delaying the decision – Framing information through increased transparency or social comparisons

Patel, Volpp, Asch. NEJM. 2018

slide-15
SLIDE 15

Nudges are prevalent in other industries

Patel, Volpp, Asch. NEJM. 2018

slide-16
SLIDE 16

Nudges are prevalent in other industries

Patel, Volpp, Asch. NEJM. 2018

slide-17
SLIDE 17

Nudges are prevalent in other industries

Patel, Volpp, Asch. NEJM. 2018

slide-18
SLIDE 18

Nudges are prevalent in other industries

Patel, Volpp, Asch. NEJM. 2018

slide-19
SLIDE 19

Nudges are prevalent in other industries

Patel, Volpp, Asch. NEJM. 2018

slide-20
SLIDE 20

Nudge units are behavioral design teams

  • Systematically test ways to improve decisions and change behavior

Patel, Volpp, Asch. NEJM. 2018

slide-21
SLIDE 21

Nudge units are behavioral design teams

  • Systematically test ways to improve decisions and change behavior
  • UK Behavioral Insights Team

– Launched in 2010 by the United Kingdom’s Government

Patel, Volpp, Asch. NEJM. 2018

slide-22
SLIDE 22

Nudge units are behavioral design teams

  • Systematically test ways to improve decisions and change behavior
  • UK Behavioral Insights Team

– Launched in 2010 by the United Kingdom’s Government – Quickly demonstrated that small changes could lead to significant impact

  • Efficiency: tax payments
  • Health: organ donor consent rates
  • Social good: charitable contributions

Patel, Volpp, Asch. NEJM. 2018

slide-23
SLIDE 23

Nudge units have spread around the world within governments

OECD Research 2018

slide-24
SLIDE 24

Nudge units have spread around the world within governments

OECD Research 2018

slide-25
SLIDE 25

World’s first behavioral design team embedded in a health system

slide-26
SLIDE 26

World’s first behavioral design team embedded in a health system

Mission

To leverage insights from behavioral economics and psychology to design and test approaches to steer medical decision-making toward higher value and improved patient

  • utcomes
slide-27
SLIDE 27

World’s first behavioral design team embedded in a health system

Mission

To leverage insights from behavioral economics and psychology to design and test approaches to steer medical decision-making toward higher value and improved patient

  • utcomes
slide-28
SLIDE 28

Our role

slide-29
SLIDE 29

Impact

50+ projects 12+ specialties 20+ team members

slide-30
SLIDE 30

Impact

50+ projects 12+ specialties

Oncology Cardiology Pediatrics Endocrine Sleep Medicine Internal Medicine Family Medicine Infectious Disease Hospital Medicine Radiation Oncology Emergency Medicine

20+ team members

slide-31
SLIDE 31

Prioritizing opportunities

slide-32
SLIDE 32

Prioritizing opportunities

Right fit for a nudge

slide-33
SLIDE 33

Prioritizing opportunities

Right fit for a nudge Leverages scalable technology

slide-34
SLIDE 34

Prioritizing opportunities

Right fit for a nudge Leverages scalable technology Potential for significant impact

slide-35
SLIDE 35

Prioritizing opportunities

Right fit for a nudge Leverages scalable technology Potential for significant impact

slide-36
SLIDE 36

Nudges vary in their approach and effectiveness

Patel, Navathe, Liao. JACR. 2019

slide-37
SLIDE 37

Nudges vary in their approach and effectiveness

Patel, Navathe, Liao. JACR. 2019

Effectiveness

slide-38
SLIDE 38

Lessons from the Penn Medicine Nudge Unit

slide-39
SLIDE 39

We are already being nudged whether we know it or not

Patel et al. Ann Intern Med. 2014

slide-40
SLIDE 40

We are already being nudged whether we know it or not

Patel et al. Ann Intern Med. 2014

slide-41
SLIDE 41

We are already being nudged whether we know it or not

Patel et al. Ann Intern Med. 2014

slide-42
SLIDE 42

We are already being nudged whether we know it or not

Brand Generic

Patel et al. Ann Intern Med. 2014

slide-43
SLIDE 43

Generic prescription rates: opt-in versus opt-out

Patel et al. JAMA IM. 2016

75.3%  98.4% P<0.001

slide-44
SLIDE 44

Generic prescription rates: opt-in versus opt-out

Patel et al. JAMA IM. 2016

75.3%  98.4% P<0.001

slide-45
SLIDE 45

Generic prescription rates: opt-in versus opt-out

Patel et al. JAMA IM. 2016

75.3%  98.4% P<0.001

slide-46
SLIDE 46

Generic prescription rates: opt-in versus opt-out

Patel et al. JAMA IM. 2016

75.3%  98.4% P<0.001

slide-47
SLIDE 47

Generic prescription rates: opt-in versus opt-out

Patel et al. JAMA IM. 2016

75.3%  98.4% P<0.001

slide-48
SLIDE 48

Generic prescription rates: opt-in versus opt-out

Patel et al. JAMA IM. 2016

slide-49
SLIDE 49

Generic prescription rates: opt-in versus opt-out

Patel et al. JAMA IM. 2016

75.3%  98.4% P<0.001

slide-50
SLIDE 50

Opioid addiction is linked to the size of the initial prescription

43%

Delgado et al. JGIM. 2018

P<0.001

slide-51
SLIDE 51

Opioid addiction is linked to the size of the initial prescription

43%

Delgado et al. JGIM. 2018

P<0.001

slide-52
SLIDE 52

Opioid addiction is linked to the size of the initial prescription

43%

Delgado et al. JGIM. 2018

P<0.001

slide-53
SLIDE 53

Opioid addiction is linked to the size of the initial prescription

43%

Delgado et al. JGIM. 2018

P<0.001

slide-54
SLIDE 54

Opioid addiction is linked to the size of the initial prescription

43% Implemented Opioid Prescription Defaults in the ED

Delgado et al. JGIM. 2018

P<0.001

slide-55
SLIDE 55

Opioid addiction is linked to the size of the initial prescription

21% 43% Implemented Opioid Prescription Defaults in the ED

Delgado et al. JGIM. 2018

P<0.001

slide-56
SLIDE 56

Opioid addiction is linked to the size of the initial prescription

21% 43% Implemented Opioid Prescription Defaults in the ED

Delgado et al. JGIM. 2018

P<0.001

slide-57
SLIDE 57

Reducing Unnecessary Imaging

slide-58
SLIDE 58

Reducing Unnecessary Imaging

National guidelines recommend that palliative cancer patients do not have imaging to align radiation therapy

slide-59
SLIDE 59

Reducing Unnecessary Imaging

National guidelines recommend that palliative cancer patients do not have imaging to align radiation therapy 80% of these patients at Penn Medicine received daily imaging (e.g. 14 CT scans for a 2-week course of radiation)

slide-60
SLIDE 60

Selecting a nudge intervention

slide-61
SLIDE 61

Selecting a nudge intervention

slide-62
SLIDE 62

Selecting a nudge intervention

slide-63
SLIDE 63

Selecting a nudge intervention

slide-64
SLIDE 64

Selecting a nudge intervention

slide-65
SLIDE 65

Imaging for Palliative Cancer Patients

Sharma et al. JAMA Oncology. 2019

slide-66
SLIDE 66

Imaging for Palliative Cancer Patients

Sharma et al. JAMA Oncology. 2019

slide-67
SLIDE 67

Imaging for Palliative Cancer Patients

Sharma et al. JAMA Oncology. 2019

slide-68
SLIDE 68

Imaging for Palliative Cancer Patients

Sharma et al. JAMA Oncology. 2019

slide-69
SLIDE 69

Imaging for Palliative Cancer Patients

Sharma et al. JAMA Oncology. 2019

68.2%  32.4% P<0.01

slide-70
SLIDE 70

Imaging for Palliative Cancer Patients

Sharma et al. JAMA Oncology. 2019

68.2%  32.4% P<0.01

3000 less imaging tests per year

slide-71
SLIDE 71

Complex Decision Pathways

slide-72
SLIDE 72

Cardiac rehab referral

slide-73
SLIDE 73

Cardiac rehab referral

  • Evidence-based pathway

– Demonstrated to reduce mortality and readmissions by up to 30%

slide-74
SLIDE 74

Cardiac rehab referral

  • Evidence-based pathway

– Demonstrated to reduce mortality and readmissions by up to 30%

  • Referral rates are low

– Only 15% of patients at Penn were referred at the time of hospital discharge – More than 25% of hospitals in the US refer less than 20% of their patients

slide-75
SLIDE 75

Cardiac rehab referral

  • Evidence-based pathway

– Demonstrated to reduce mortality and readmissions by up to 30%

  • Referral rates are low

– Only 15% of patients at Penn were referred at the time of hospital discharge – More than 25% of hospitals in the US refer less than 20% of their patients

  • Manual opt-in process

– Cardiologists had to complete paper form with 12+ fields while on rounds – Patients had to identify a rehab center on their own and check insurance coverage

slide-76
SLIDE 76

Redesigned as an opt-out decision pathway

slide-77
SLIDE 77

Redesigned as an opt-out decision pathway

Automate identification and notification

slide-78
SLIDE 78

Redesigned as an opt-out decision pathway

Automate identification and notification Restructure rounding and discharge process

slide-79
SLIDE 79

Cardiac Rehab Referral Rates Over Time

Adusumalli, Patel et al. Under Review

slide-80
SLIDE 80

Cardiac Rehab Referral Rates Over Time

Adusumalli, Patel et al. Under Review

slide-81
SLIDE 81

Cardiac Rehab Referral Rates Over Time

Adusumalli, Patel et al. Under Review

slide-82
SLIDE 82

Active Choice Framing

slide-83
SLIDE 83

Active choice to prompt decision-making

  • Alternative to relying on the individual to make a

decision

slide-84
SLIDE 84

Active choice to prompt decision-making

  • Alternative to relying on the individual to make a

decision

  • Individual is stopped from proceeding unless they

make a choice between options (Yes or No)

slide-85
SLIDE 85

Active choice to prompt decision-making

  • Alternative to relying on the individual to make a

decision

  • Individual is stopped from proceeding unless they

make a choice between options (Yes or No)

  • Design

– Prompt decision when you have the physician’s attention – Increase saliency of advantages and disadvantages for each option – Make it easy to say yes

slide-86
SLIDE 86

Active choice to prompt decision-making

  • Alternative to relying on the individual to make a

decision

  • Individual is stopped from proceeding unless they

make a choice between options (Yes or No)

  • Design

– Prompt decision when you have the physician’s attention – Increase saliency of advantages and disadvantages for each option – Make it easy to say yes

slide-87
SLIDE 87

Influenza vaccination rates

Patel et al. JGIM. 2017

slide-88
SLIDE 88

Influenza vaccination rates

Patel et al. JGIM. 2017

slide-89
SLIDE 89

Influenza vaccination rates

Patel et al. JGIM. 2017

slide-90
SLIDE 90

Influenza vaccination rates

  • Active choice prompt targeted to

physicians and medical assistants

Patel et al. JGIM. 2017

slide-91
SLIDE 91

Influenza vaccination rates

  • Active choice prompt targeted to

physicians and medical assistants

  • Difference-in-difference analysis

– 6.6 percentage point increase (P<.001) – 37.3% relative increase in vaccination rates

Patel et al. JGIM. 2017

slide-92
SLIDE 92

Influenza vaccination rates

  • Active choice prompt targeted to

physicians and medical assistants

  • Difference-in-difference analysis

– 6.6 percentage point increase (P<.001) – 37.3% relative increase in vaccination rates

  • Expanded to other primary care

practices at Penn Medicine

Patel et al. JGIM. 2017

slide-93
SLIDE 93

Influenza vaccination rates

  • Active choice prompt targeted to

physicians and medical assistants

  • Difference-in-difference analysis

– 6.6 percentage point increase (P<.001) – 37.3% relative increase in vaccination rates

  • Expanded to other primary care

practices at Penn Medicine

– Number of notifications reduced

Patel et al. JGIM. 2017

slide-94
SLIDE 94

Influenza vaccination rates

  • Active choice prompt targeted to

physicians and medical assistants

  • Difference-in-difference analysis

– 6.6 percentage point increase (P<.001) – 37.3% relative increase in vaccination rates

  • Expanded to other primary care

practices at Penn Medicine

– Number of notifications reduced – Redirected to medical assistants to template

  • rders for physicians

Patel et al. JGIM. 2017

slide-95
SLIDE 95

Influenza vaccination rates after an active choice intervention

Kim, Patel et al. JAMA Network Open. 2018

slide-96
SLIDE 96

Influenza vaccination rates after an active choice intervention

Kim, Patel et al. JAMA Network Open. 2018

slide-97
SLIDE 97

Influenza vaccination rates after an active choice intervention

Compared to Control Adjusted difference

9.5 percentage points 95% CI: 4.1 – 14.3 P<0.001

Kim, Patel et al. JAMA Network Open. 2018

slide-98
SLIDE 98

Nudges for Population Health

slide-99
SLIDE 99

Nudges for Population Health

40,000 patients at Penn Medicine meet national guidelines for statin therapy

slide-100
SLIDE 100

Nudges for Population Health

40,000 patients at Penn Medicine meet national guidelines for statin therapy Only about 50% have ever been prescribed a statin

slide-101
SLIDE 101

Study design

  • Sample

– 96 PCPs from 32 practice sites comprising 4774 patients eligible but never prescribed a statin

  • Randomized, controlled trial

– Usual care – Active choice dashboard – Active choice dashboard + peer comparison feedback delivered once by email

Patel et al. JAMA Network Open. 2018

slide-102
SLIDE 102

Study design

  • Sample

– 96 PCPs from 32 practice sites comprising 4774 patients eligible but never prescribed a statin

  • Randomized, controlled trial

– Usual care – Active choice dashboard – Active choice dashboard + peer comparison feedback delivered once by email

– Below median: compared to the median – Above median: compared to 90% percentile – >90th percentile: recognized as top performer

Patel et al. JAMA Network Open. 2018

slide-103
SLIDE 103

Study design

  • Sample

– 96 PCPs from 32 practice sites comprising 4774 patients eligible but never prescribed a statin

  • Randomized, controlled trial

– Usual care – Active choice dashboard – Active choice dashboard + peer comparison feedback delivered once by email

– Below median: compared to the median – Above median: compared to 90% percentile – >90th percentile: recognized as top performer

Fake Backend

  • Data from clinical warehouse
  • Study team sent email to PCP
  • PCP uses secure website to submit orders
  • Study team templates orders in EHR
  • PCP logs into EHR and signs orders
  • Prescription sent to pharmacy

electronically

  • Study team sends letter to patient

Patel et al. JAMA Network Open. 2018

slide-104
SLIDE 104

Description of health system initiative

Patel et al. JAMA Network Open. 2018

slide-105
SLIDE 105

Description of health system initiative

Patel et al. JAMA Network Open. 2018

slide-106
SLIDE 106

Description of health system initiative

Patel et al. JAMA Network Open. 2018

slide-107
SLIDE 107

Automated patient dashboard

Patel et al. JAMA Network Open. 2018

slide-108
SLIDE 108

Automated patient dashboard

Patel et al. JAMA Network Open. 2018

slide-109
SLIDE 109

Automated patient dashboard

Patel et al. JAMA Network Open. 2018

slide-110
SLIDE 110

Automated patient dashboard

Patel et al. JAMA Network Open. 2018

slide-111
SLIDE 111

Automated patient dashboard

Patel et al. JAMA Network Open. 2018

slide-112
SLIDE 112

Automated patient dashboard

Patel et al. JAMA Network Open. 2018

slide-113
SLIDE 113

Dashboard Utilization

Patel et al. JAMA Network Open. 2018

slide-114
SLIDE 114

Dashboard Utilization

Active Choice (N=32) Active Choice w/Peer Comparisons (N=32) Accessed Dashboard 16/32 (50.0%) 12/32 (37.5%) Submitted Decisions 4/32 (12.5%) 8/32 (25.0%) Signed Prescription Orders 2/32 (6.3%) 8/32 (25.0%)

Patel et al. JAMA Network Open. 2018

slide-115
SLIDE 115

Dashboard Utilization

Active Choice (N=32) Active Choice w/Peer Comparisons (N=32) Accessed Dashboard 16/32 (50.0%) 12/32 (37.5%) Submitted Decisions 4/32 (12.5%) 8/32 (25.0%) Signed Prescription Orders 2/32 (6.3%) 8/32 (25.0%)

Patel et al. JAMA Network Open. 2018

slide-116
SLIDE 116

Dashboard Utilization

Active Choice (N=32) Active Choice w/Peer Comparisons (N=32) Accessed Dashboard 16/32 (50.0%) 12/32 (37.5%) Submitted Decisions 4/32 (12.5%) 8/32 (25.0%) Signed Prescription Orders 2/32 (6.3%) 8/32 (25.0%)

Patel et al. JAMA Network Open. 2018

slide-117
SLIDE 117

Active Choice Dashboards to Increase Statin Prescribing

Patel et al. JAMA Network Open. 2018

slide-118
SLIDE 118

Active Choice Dashboards to Increase Statin Prescribing

Patel et al. JAMA Network Open. 2018

slide-119
SLIDE 119

Active Choice Dashboards to Increase Statin Prescribing

Patel et al. JAMA Network Open. 2018

slide-120
SLIDE 120

Active Choice Dashboards to Increase Statin Prescribing

Active choice and peer comparisons 5.8 percentage points 95% CI: 0.9 to 13.0 P<0.01

Patel et al. JAMA Network Open. 2018

slide-121
SLIDE 121

Growing interest in implementing nudge units in health care

Changolkar, Patel et al. NEJM Catalyst. 2019

slide-122
SLIDE 122

Summary: Nudging Medical Decision-Making

slide-123
SLIDE 123

Summary: Nudging Medical Decision-Making

  • Medical decision-making is often suboptimal
slide-124
SLIDE 124

Summary: Nudging Medical Decision-Making

  • Medical decision-making is often suboptimal
  • Design of choice environments influences our behavior

– We are already being nudged but are often unaware of it – Strategic attention to align design with our goals

slide-125
SLIDE 125

Summary: Nudging Medical Decision-Making

  • Medical decision-making is often suboptimal
  • Design of choice environments influences our behavior

– We are already being nudged but are often unaware of it – Strategic attention to align design with our goals

  • Nudge units can improve the delivery of health care

– Systematic approach to design, implement, and test interventions – Steer decisions towards higher value and better patient outcomes

slide-126
SLIDE 126

Using Nudges to Improve the Delivery of Health Care

Nudgeunit.upenn.edu @PennNudgeUnit

Mitesh Patel, MD, MBA

Email: mpatel@pennmedicine.upenn.edu Twitter: @miteshspatel Website: www.miteshspatel.com