Empirical Evidence on The Role of Public Warnings on Physicians Drug - - PowerPoint PPT Presentation

▶
empirical evidence on the role of public warnings on
SMART_READER_LITE
LIVE PREVIEW

Empirical Evidence on The Role of Public Warnings on Physicians Drug - - PowerPoint PPT Presentation

Apr 04, 2023 154 likes •568 views

Empirical Evidence on The Role of Public Warnings on Physicians Drug Prescriptions Behavior Pierre Dubois and Tuba Tunel Toulouse School of Economics Harvard Center for Risk Analysis Risk, Perception, and Response Conference March 21, 2014

slide-1
SLIDE 1

Empirical Evidence on The Role of Public Warnings on Physicians Drug Prescriptions Behavior

Pierre Dubois and Tuba Tunçel Toulouse School of Economics

Harvard Center for Risk Analysis Risk, Perception, and Response Conference

March 21, 2014

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 1 / 18

slide-2
SLIDE 2

Introduction

Introduction

Research question: How does physicians’ drug prescription react to scientific information release by public authorities in France?

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 2 / 18

slide-3
SLIDE 3

Introduction

Introduction

Research question: How does physicians’ drug prescription react to scientific information release by public authorities in France? Medical warnings are typically giving news on risk assessments of drugs (side effects, ..) on specific populations and reaction can depend on:

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 2 / 18

slide-4
SLIDE 4

Introduction

Introduction

Research question: How does physicians’ drug prescription react to scientific information release by public authorities in France? Medical warnings are typically giving news on risk assessments of drugs (side effects, ..) on specific populations and reaction can depend on:

prior beliefs that can be heterogenous implying that new information can be positive or negative

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 2 / 18

slide-5
SLIDE 5

Introduction

Introduction

Research question: How does physicians’ drug prescription react to scientific information release by public authorities in France? Medical warnings are typically giving news on risk assessments of drugs (side effects, ..) on specific populations and reaction can depend on:

prior beliefs that can be heterogenous implying that new information can be positive or negative interpretation of the ‘boundaries’ of the ‘specific’ population

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 2 / 18

slide-6
SLIDE 6

Introduction

Introduction

Research question: How does physicians’ drug prescription react to scientific information release by public authorities in France? Medical warnings are typically giving news on risk assessments of drugs (side effects, ..) on specific populations and reaction can depend on:

prior beliefs that can be heterogenous implying that new information can be positive or negative interpretation of the ‘boundaries’ of the ‘specific’ population

News on some drugs can underline ‘lack’ of news on others and generate ‘spillovers’

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 2 / 18

slide-7
SLIDE 7

Introduction

Introduction

Research question: How does physicians’ drug prescription react to scientific information release by public authorities in France? Medical warnings are typically giving news on risk assessments of drugs (side effects, ..) on specific populations and reaction can depend on:

prior beliefs that can be heterogenous implying that new information can be positive or negative interpretation of the ‘boundaries’ of the ‘specific’ population

News on some drugs can underline ‘lack’ of news on others and generate ‘spillovers’ Example: new results on the increase of suicidal thinking in children reported in 2004 for selective serotonin reuptake inhibitors (SSRI)

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 2 / 18

slide-8
SLIDE 8

Introduction

Introduction

Medical information evolves quickly, imposes prescribers to learn quickly about drugs, and is complex to process, sometimes ambiguous

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 3 / 18

slide-9
SLIDE 9

Introduction

Introduction

Medical information evolves quickly, imposes prescribers to learn quickly about drugs, and is complex to process, sometimes ambiguous Antidepressants used in France:

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 3 / 18

slide-10
SLIDE 10

Introduction

Introduction

Medical information evolves quickly, imposes prescribers to learn quickly about drugs, and is complex to process, sometimes ambiguous Antidepressants used in France:

First Generation: Tricyclic Antidepressants (TCAs)

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 3 / 18

slide-11
SLIDE 11

Introduction

Introduction

Medical information evolves quickly, imposes prescribers to learn quickly about drugs, and is complex to process, sometimes ambiguous Antidepressants used in France:

First Generation: Tricyclic Antidepressants (TCAs) Second Generation: Selective Serotonin Reuptake Inhibitors (SSRIs) and Serotonin-Norepinephrine Reuptake Inhibitors (SNRIs), Other (Mianserine, Mirtazapine, Tianeptine)

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 3 / 18

slide-12
SLIDE 12

Introduction

Introduction

Medical information evolves quickly, imposes prescribers to learn quickly about drugs, and is complex to process, sometimes ambiguous Antidepressants used in France:

First Generation: Tricyclic Antidepressants (TCAs) Second Generation: Selective Serotonin Reuptake Inhibitors (SSRIs) and Serotonin-Norepinephrine Reuptake Inhibitors (SNRIs), Other (Mianserine, Mirtazapine, Tianeptine)

Pubic health authority ANSM is in charge of drug authorization and regulation the use of prescription drugs gives usage conditions and recommendations to physicians

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 3 / 18

slide-13
SLIDE 13

Introduction

Information Release and Warnings

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 4 / 18

slide-14
SLIDE 14

Data

Data and Outcome of Interest

Proprietary data of CEGEDIM, a global technology and services company specializing in health care

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 5 / 18

slide-15
SLIDE 15

Data

Data and Outcome of Interest

Proprietary data of CEGEDIM, a global technology and services company specializing in health care Large panel data on exhaustive prescriptions of 386 general practitioners in France between 2000 and 2008

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 5 / 18

slide-16
SLIDE 16

Data

Data and Outcome of Interest

Proprietary data of CEGEDIM, a global technology and services company specializing in health care Large panel data on exhaustive prescriptions of 386 general practitioners in France between 2000 and 2008 Information on physicians, patients and patient visits

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 5 / 18

slide-17
SLIDE 17

Data

Data and Outcome of Interest

Proprietary data of CEGEDIM, a global technology and services company specializing in health care Large panel data on exhaustive prescriptions of 386 general practitioners in France between 2000 and 2008 Information on physicians, patients and patient visits Focus on prescriptions when patient diagnosed by depression for the first time (still no specific information on patient’s responsiveness - learning on experience products)

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 5 / 18

slide-18
SLIDE 18

Data

Data and Outcome of Interest

Proprietary data of CEGEDIM, a global technology and services company specializing in health care Large panel data on exhaustive prescriptions of 386 general practitioners in France between 2000 and 2008 Information on physicians, patients and patient visits Focus on prescriptions when patient diagnosed by depression for the first time (still no specific information on patient’s responsiveness - learning on experience products) Outcome of interest:

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 5 / 18

slide-19
SLIDE 19

Data

Data and Outcome of Interest

Proprietary data of CEGEDIM, a global technology and services company specializing in health care Large panel data on exhaustive prescriptions of 386 general practitioners in France between 2000 and 2008 Information on physicians, patients and patient visits Focus on prescriptions when patient diagnosed by depression for the first time (still no specific information on patient’s responsiveness - learning on experience products) Outcome of interest:

Physician-specific probability to prescribe an antidepressant

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 5 / 18

slide-20
SLIDE 20

Data

Data and Outcome of Interest

Proprietary data of CEGEDIM, a global technology and services company specializing in health care Large panel data on exhaustive prescriptions of 386 general practitioners in France between 2000 and 2008 Information on physicians, patients and patient visits Focus on prescriptions when patient diagnosed by depression for the first time (still no specific information on patient’s responsiveness - learning on experience products) Outcome of interest:

Physician-specific probability to prescribe an antidepressant Choice probability of antidepressants with a Random Coefficient Logit Model (allow heterogeneity of decision makers)

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 5 / 18

slide-21
SLIDE 21

Results

Antidepressant Prescriptions - Means

Table 1: Antidepressant Prescriptions Probabilities at Diagnosis Antidep. All Kids and Young Adults Adults Elderly Group Ages

  • Ado. (2-18)

(19-25) (26-65) (65+) SSRI 29.4% 10.1% 23.8% 31.9% 29.1% SNRI 5.5% 1.0% 4.1% 6.3% 4.6% TCA 2.4% 1.1% 0.9% 2.4% 4.7% Other 5.6% 1.7% 3.3% 5.4% 10.2% None 57.1% 86.1% 67.9% 54% 51.4% Patients 173,207 9,815 19,949 122,178 21,174

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 6 / 18

slide-22
SLIDE 22

Results

Antidepressant Prescriptions Changes

Prescription probability for kids and adolescents decreases after warning for almost all physicians with initial prescription probability above 10 %. Much less changes for other age categories.

. 2 . 4 . 6 . 8 1 P r

  • b

a b i l i t y , A f t e r .2 .4 .6 .8 1 P robability, B efore

Kids and Adolescents (2-18)

. 2 . 4 . 6 . 8 1 P r

  • b

a b i l i t y , A f t e r .2 .4 .6 .8 1 P robability, B efore

Young Adults (19-25)

. 2 . 4 . 6 . 8 1 P r

  • b

a b i l i t y , A f t e r .2 .4 .6 .8 1 P robability, B efore

Adults (26-65)

. 2 . 4 . 6 . 8 1 P r

  • b

a b i l i t y , A f t e r .2 .4 .6 .8 1 P robability, B efore

Elderly (65+)

Probability of Antidep. Prescription

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 7 / 18

slide-23
SLIDE 23

Results

SSRI Prescriptions Changes

Probabilities of SSRI prescriptions are lower for kids and adolescents than for other age groups. It decreases after warning, with heterogeneity.

. 2 . 4 . 6 . 8 1 P r

  • b

a b i l i t y , A f t e r .2 .4 .6 .8 1 P robability, B efore

Kids and Adolescents (2-18)

. 2 . 4 . 6 . 8 1 P r

  • b

a b i l i t y , A f t e r .2 .4 .6 .8 P robability, B efore

Young Adults (19-25)

. 2 . 4 . 6 . 8 P r

  • b

a b i l i t y , A f t e r .2 .4 .6 .8 P robability, B efore

Adults (26-65)

. 2 . 4 . 6 . 8 1 P r

  • b

a b i l i t y , A f t e r .2 .4 .6 .8 1 P robability, B efore

Elderly (65+)

Probability of SSRI Prescription

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 8 / 18

slide-24
SLIDE 24

Mixed Logit Results

Mixed Logit Results - Means of Marginal Effects

Depression Mild Severe

  • Prob. (%)

Change

  • Prob. (%)

Change Prescription Age group Before After Before After No Antidepressant Kids (2-18) 70.6 72.8 +2.2% 14.3 15.9 +1.6% Young Adults 17.0 14.6

  • 2.3%

1.9 1.7

  • 0.2%

Adults 17.9 15.0

  • 3.0%

2.3 1.9

  • 0.3%

Elderly 15.9 11.1

  • 4.8%

1.2 0.8

  • 0.4%

SSRI Kids (2-18) 16.7 14.2

  • 2.5%

47.5 43.2

  • 4.3%

Young Adults 18.4 19.7 +1.3% 26.9 26.9

  • 0.1%

Adults 20.1 18.4

  • 1.7%

26.9 23.6

  • 3.3%

Elderly 20.4 20.9 +0.5% 24.2 23.8

  • 0.5%

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 9 / 18

slide-25
SLIDE 25

Mixed Logit Results

Mixed Logit Results - Means of Marginal Effects

Depression Mild Severe

  • Prob. (%)

Change

  • Prob. (%)

Change Prescription Age group Before After Before After SNRI Kids (2-18) 3.3 3.7 +0.4% 10.6 12.4 +1.7% Young Adults 20.8 22.6 +1.8% 21.2 23.1 +1.9% Adults 20.4 22.9 +2.5% 22.3 24.6 +2.3% Elderly 21.9 22.5 +0.6% 27.0 26.2

  • 0.9%

TCA Kids (2-18) 2.6 2.7 +0.1% 7.2 7.9 +0.7% Young Adults 21.1 21.7 +0.6% 21.5 22.3 +0.8% Adults 22.8 22.0

  • 0.8%

25.8 24.4

  • 1.4%

Elderly 19.2 22.0 +2.8% 20.0 22.3 +2.3% Other Kids (2-18) 6.8 6.5

  • 0.2%

20.3 20.6 +0.2% Young Adults 22.7 21.3

  • 1.3%

28.5 26.1

  • 2.4%

Adults 18.8 21.7 +2.9% 22.7 25.4 +2.8% Elderly 22.6 23.6 +0.9% 27.5 26.9

  • 0.6%

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 10 / 18

slide-26
SLIDE 26

Mixed Logit Results

Change in c.d.f. of Predicted Probabilities

No Drug

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Kids

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Young Adults

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Adults

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Elderly

No Antidepressant - Mild Depression

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 11 / 18

slide-27
SLIDE 27

Mixed Logit Results

Change in c.d.f. of Predicted Probabilities

No Drug

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Kids

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Young Adults

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Adults

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Elderly

No Antidepressant - Severe Depression

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 12 / 18

slide-28
SLIDE 28

Mixed Logit Results

Change in c.d.f. of Predicted Probabilities

SSRI

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Kids

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Young Adults

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Adults

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Elderly

SSRI - Mild Depression

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 13 / 18

slide-29
SLIDE 29

Mixed Logit Results

Change in c.d.f. of Predicted Probabilities

SSRI

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Kids

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Young Adults

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Adults

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Elderly

SSRI - Severe Depression

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 14 / 18

slide-30
SLIDE 30

Mixed Logit Results

Change in c.d.f. of Predicted Probabilities

SNRI

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Kids

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Young Adults

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Adults

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Elderly

SNRI - Mild Depression

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 15 / 18

slide-31
SLIDE 31

Mixed Logit Results

Change in c.d.f. of Predicted Probabilities

SNRI

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Kids

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Young Adults

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Adults

. 2 . 4 . 6 . 8 1 .2 .4 .6 .8 1 Choice Probability c.d.f. without news c.d.f. with news

Elderly

SNRI - Severe Depression

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 16 / 18

slide-32
SLIDE 32

Conclusion

Main Results

Targeted population by the warning (kids and adolescents):

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 17 / 18

slide-33
SLIDE 33

Conclusion

Main Results

Targeted population by the warning (kids and adolescents):

Reduction of SSRI prescription (-2.5% to -4;3%) without drug substitution for mild depression but partial switch to SNRI, TCA for severe depression

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 17 / 18

slide-34
SLIDE 34

Conclusion

Main Results

Targeted population by the warning (kids and adolescents):

Reduction of SSRI prescription (-2.5% to -4;3%) without drug substitution for mild depression but partial switch to SNRI, TCA for severe depression

For older age categories: heterogenous responses with some increase and some decrease, spreading prescription probabilities.

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 17 / 18

slide-35
SLIDE 35

Conclusion

Main Results

Targeted population by the warning (kids and adolescents):

Reduction of SSRI prescription (-2.5% to -4;3%) without drug substitution for mild depression but partial switch to SNRI, TCA for severe depression

For older age categories: heterogenous responses with some increase and some decrease, spreading prescription probabilities. Potential explanations: some physicians interpret warning on SSRI for kids and adolescents (which is clearly a ‘bad’ news about SSRIs for kids and adolescents) as ‘bad news’ for other age groups too and

  • thers as ‘good news’ for SSRIs in other age groups (they may think

if there was bad news for other age groups they would be informed).

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 17 / 18

slide-36
SLIDE 36

Conclusion

Main Results

Targeted population by the warning (kids and adolescents):

Reduction of SSRI prescription (-2.5% to -4;3%) without drug substitution for mild depression but partial switch to SNRI, TCA for severe depression

For older age categories: heterogenous responses with some increase and some decrease, spreading prescription probabilities. Potential explanations: some physicians interpret warning on SSRI for kids and adolescents (which is clearly a ‘bad’ news about SSRIs for kids and adolescents) as ‘bad news’ for other age groups too and

  • thers as ‘good news’ for SSRIs in other age groups (they may think

if there was bad news for other age groups they would be informed). As warning does not recommend substitutes: different reactions substituting to other drugs or not.

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 17 / 18

slide-37
SLIDE 37

Conclusion

Policy Implications

Even if prescription by professional, warnings are not fully respected and reaction is heterogenous (studies like Witteman et al. on vaccination decisions may be useful even for physicians)

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 18 / 18

slide-38
SLIDE 38

Conclusion

Policy Implications

Even if prescription by professional, warnings are not fully respected and reaction is heterogenous (studies like Witteman et al. on vaccination decisions may be useful even for physicians) Warning affecting drug substitutes, authority should be careful in communication and it should depend on market structure

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 18 / 18

slide-39
SLIDE 39

Conclusion

Policy Implications

Even if prescription by professional, warnings are not fully respected and reaction is heterogenous (studies like Witteman et al. on vaccination decisions may be useful even for physicians) Warning affecting drug substitutes, authority should be careful in communication and it should depend on market structure Warning affecting non targeted populations means that risk communication should clearly define the population and communicate

  • n other population because relative ‘lack of news’ can also be

misinterpreted

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 18 / 18

slide-40
SLIDE 40

Conclusion

Policy Implications

Even if prescription by professional, warnings are not fully respected and reaction is heterogenous (studies like Witteman et al. on vaccination decisions may be useful even for physicians) Warning affecting drug substitutes, authority should be careful in communication and it should depend on market structure Warning affecting non targeted populations means that risk communication should clearly define the population and communicate

  • n other population because relative ‘lack of news’ can also be

misinterpreted Effect may depend on physician ‘prior’ (as for subjective risk perception (Flores and Loomes), physicians are also influenced by

  • wn experience) and on ‘ambiguity’ of warning (see Bansback et al.
  • n imprecision communication)

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 18 / 18

slide-41
SLIDE 41

Conclusion

Policy Implications

Even if prescription by professional, warnings are not fully respected and reaction is heterogenous (studies like Witteman et al. on vaccination decisions may be useful even for physicians) Warning affecting drug substitutes, authority should be careful in communication and it should depend on market structure Warning affecting non targeted populations means that risk communication should clearly define the population and communicate

  • n other population because relative ‘lack of news’ can also be

misinterpreted Effect may depend on physician ‘prior’ (as for subjective risk perception (Flores and Loomes), physicians are also influenced by

  • wn experience) and on ‘ambiguity’ of warning (see Bansback et al.
  • n imprecision communication)

Effect may depend on whether physicians are Bayesian updating (like in Rheinberger and Hammitt)

Dubois - Tunçel (TSE) Warnings and Prescription March 21, 2014 18 / 18