Advancing the Science of Behaviour Change
Susan Michie Professor of Health Psychology
University College LondonUK
Connecticut, USA: September 2010
Starting point …
• Interventions to change health-related behaviours– amongst the general population, patients and health
professionals• have had variable success.
• Behaviour change remains a formidable challenge– we lack a clear understanding of
• effective techniques and • mechanisms of change • across behaviours, populations and settings.
To develop more effective interventions …
We need to improve our scientific methods in the following areas:1. Specify intervention content in sufficient detail in trial protocols and
published reports to allow accurate replication and evidence syntheses that can identify “active ingredients” within interventions.
2. Apply theories of behaviour change to intervention development and evaluation, and to evidence syntheses• allows questions to be answered about not just “what worked” in
interventions but “how they worked”.3. Match behaviour change techniques to mechanisms of action to
inform theory-based intervention development and evaluation-based theory development.
4. Define the extent to which interventions, and comparison conditions, use theory and evaluate how this relates to effectiveness.
5. Develop a model of behaviour as a starting point for systematic, theory-based intervention development.
1. Specifying interventions more precisely
• Interventions often “complex” – several, potentially interacting, techniques
• Poorly described– Interventions often described vaguely
e.g. “behavioural counselling” – Where protocols with more detail are
available, terminology is variable
• Impedes replication, implementation, evidence synthesis
Guidelines for specifying interventions
• CONSORT guidelines for reporting RCTs– Evaluators should report “precise details of interventions [as]
actually administered” Moher et al, 2001
• Which precise details?– the content or elements of the intervention– characteristics of those delivering the intervention– characteristics of the recipients,– characteristics of the setting (e.g., worksite)– the mode of delivery (e.g., face-to-face)– the intensity (e.g., contact time)– the duration (e.g., number sessions over a given period)– adherence to delivery protocols Davidson et al, Annals of Beh Med, 2003
Effects of internet behavioral counseling on weight loss in adults at risk for Type 2 diabetes
“feedback on self-monitoring record, reinforcement, recommendations for change, answers to questions, and general
support” (Tate et al. JAMA 2003)
Example of the problem: Descriptions of “behavioural counselling” in two interventions
Title of journal article Description of “behavioural counseling”
The impact of behavioral counseling on stage of change fat intake, physical activity, and cigarette smoking in adults at increased risk of coronary heart disease
“educating patients about the benefits of lifestyle change, encouraging them, and suggesting what changes could be made” (Steptoe et al. AJPH 2001)
Biomedicine vs behavioural science … example of smoking cessation effectiveness
• Intervention content
• Mechanism of action– Activity at a subtype of the
nicotinic receptor where its binding produces agonistic activity, while simultaneously preventing binding to a4b2 receptors
• Intervention content– Review smoking history &
motivation to quit
– Help identify high risk situations
– Generate problem-solving strategies
– Non-specific support & encouragement
• Mechanism of action– None mentioned
Varenicline JAMA, 2006 Behavioural counselling Cochrane,
2005
To improve development and application of interventions to change behaviour ….
• Need a reliable method for describing intervention content in terms of – distinct component techniques– standardised language
• An example developed from 3 systematic reviews of “self-management” interventions to increase physical activity and healthy eating– A 26 item taxonomy
Abraham & Michie (2008). A taxonomy of behavior change techniques used in interventions. Health Psychology.
Behaviour change techniques: reliable taxonomy to change physical activity and healthy eating behaviours
1. General information 2. Information on consequences3. Information about approval4. Prompt intention formation 5. Specific goal setting 6. Graded tasks7. Barrier identification8. Behavioral contract9. Review goals10. Provide instruction11. Model/ demonstrate 12. Prompt practice13. Prompt monitoring 14. Provide feedback
15. General encouragement16. Contingent rewards17. Teach to use cues 18. Follow up prompts19. Social comparison20. Social support/ change21. Role model22. Prompt self talk23. Relapse prevention24. Stress management25. Motivational interviewing26. Time management
The person is asked to keep a record of specified behaviour/s.
This could e.g. take the form of a diary or completing a
questionnaire about their behaviour.
Involves detailed planning of what the person will do including, at least, a very specific definition of the
behaviour e.g., frequency (such as how many times a day/week), intensity (e.g., speed) or duration (e.g., for how long for). In addition, at least one of the following contexts i.e., where, when, how or with whom must be specified. This could include identification of sub-goals or preparatory behaviours and/or specific contexts in
which the behaviour will be performed.
Further work ….
• Extended to 40 techniques reliably identified with improved definitions
Michie et al, British Journal of Health Psychology, in press
• 71 techniques used in smoking cessation interventions
Michie et al, Annals Behavioral Medicine, 2010
• 137 identified by– consulting textbooks for training applied
psychologists in BC techniquesMichie et al, Applied Psychology: An International Review, 2008• 42 in interventions to reduce excessive alcohol use• Three year MRC grant to put taxonomy approach
onto a more scientific basis– Expert coders needed!
Applications of taxonomy approach
1. Identifying active ingredients in interventions– Meta-regression in evidence synthesis
• Physical activity & healthy eating• Audit & Feedback
2. Investigating mechanisms of action (theoretical mediation)– The intervention “ProActive”
3. Designing interventions– A classification based on comprehensive coverage,
coherence and a model of behaviour
Evidence syntheses of BC interventions
• Often find no clear patterns, inconsistent results and small effects
• Synthesis depends on categorising complex, multi-component interventions
• Categories often ad hoc
• Specifying content as BCTs allows theoretically based categorisation & analysis
Identifying active ingredients in interventions
• Usual meta-analysis– overall effect of heterogeneous interventions
• Technique-based meta-regression– similar to traditional regression, except data at study
rather than individual level– classify interventions into component BCTs– meta-regression to investigate effects of
• individual techniques across interventions• theoretically based combination of techniques
What BCTs are effective in interventions to increase physical activity and healthy eating?
• Inclusion criteria– Interventions using behavioural &/or cognitive techniques – in adults– designs experimental or quasi-experimental – outcome measures objective or validated self-report
• 6 electronic databases, 1990-2007• Intervention content analysed using
– a reliable taxonomy of 26 techniques– a theoretically derived combination of techniques
• Random effects meta-analysis and meta-regression– isolates unique contribution of specific techniques to
heterogeneity
Michie, Abraham, et al (2009) Effective techniques in healthy eating and physical activity interventions: A meta-regression. Health Psychology, 28, 690-701
The interventions
• 84 interventions (n=28,838)
• Target behaviour– Physical activity &/or Healthy eating
• Interventions ave. 6 techniques (range 1-14)– Many different combinations
• Effect d=0.37, 95% CI 0.29-0.54
• Very heterogeneous effects (I2=79%)– not explained by 10 moderators examined e.g.
• Setting, population, intervention characteristics, target behaviour
Results
• Only one technique, self-monitoring, had a significant effect for both behaviours across interventions– d=0.57, 14.6% variance
• Next step– Use psychological theory to predict combinations of
techniques that might be more effective– Control Theory suggests how feedback may interact
with other techniques to change behaviour Carver & Scheier, 1982
A Self-regulation (control) Theory: Carver & Scheier, 82
GOALCompare behaviour with standard
Discrepancy noted
Act to reduce discrepancy
Environmental influences
No discrepancy – goal reached
Disengage from goal – give up
SELF-MONITORING/FEEDBACK
GOAL-SETTING
ACTION-PLANNING
Theoretical combination of techniques
• self-monitoring of behaviour• Other core self regulatory processes:
– setting goals– reviewing goals– specifying action plans– feedback on performance
Findings
• Interventions comprising self-monitoring with at least one other “self-regulatory” techniques (n=28)
compared with the other interventions (n=56)• were twice as effective • d=0.60 vs d=0.26
2. & 3. Applying theories of behaviour change and matching BCTs to mechanisms of action
• Re-analysis of a Cochrane review of professional practice (Audit & Feedback)
• ProActive intervention to increase physical activity in those at risk of Type 2 diabetes
Re-analysis of Cochrane review of Audit & Feedback Jamtvedt et al. Audit and feedback: effects on professional practice and health care outcomes. Cochrane Database of Systematic Reviews 2006
• Intervention categorised as “intensive”, “moderate” or “non-intensive”
What is “intensive” Audit & Feedback?
• Intensive A&F– (individual recipients) AND ((verbal format) OR (a
supervisor or senior colleague as the source)) AND (moderate or prolonged feedback)
• Non-intensive– ((group feedback) NOT (from a supervisor or
senior colleague)) OR ((individual feedback) AND (written format) AND (containing information about costs or numbers of tests without personal incentives))
• Moderate– (any other combination of characteristics than
described in Intensive or Non-intensive group)
• No rationale for these categories in the published review or when authors contacted
• Results:– 118 RCTs with objective measures of performance– Variable effects, from 16% decrease to 70% increase– Small effect of “higher intensity” of A & F– No clear trend: Adjusted RR 1.55, 1.11 & 1.45 for high,
moderate and low intensity
• Limited use in informing evidence-based A&F – ‘A&F will continue to be an unreliable approach … until
we learn how and when it works best’ (Foy et al, 2005)
Our re-analysis
• Published reports usually inadequate for meaningful data synthesis
• Therefore, intervention details requested from authors– 59 contacted, 44 responses,16 (27%) sent us additional
information (8 protocols)
• Updated, 12 new RCTs, 69 excluded, 61 reporting 85 comparisons
• Theory and evidence used to identify– Intervention components
• 40 separate components coded, combined into BCTs• Theory-based hypotheses about effectiveness
Gardner, Whittington, McAteer, Eccles, Michie (2010) Using theory to synthesise evidence from behaviour change interventions: The example of audit and feedback. Social Science & Medicine
Coding and analysis
• Interventions coded for:– Goals (yes vs no)
– Action plans (yes vs no)
– Additional intervention BCTs (yes vs no)
– Baseline compliance (continuous)
• Control conditions coded for:– Additional intervention BCTs (yes vs no)
Michie, Prestwich, de Bruin M. (2010) Importance of the Nature of Comparison Conditions for Testing Theory-Based Interventions: Reply to Commentary. Health Psychology.
Self-regulation (control) Theory: Carver & Scheier, 82
GOALCompare behaviour with standard
Discrepancy noted
Act to reduce discrepancy
Environmental influences
No discrepancy – goal reached
Disengage from goal
FEEDBACK
GOAL-SETTING
ACTION-PLANNING
Theory-based hypotheses
• Feedback more effective when goal/target is set• Most effective where goal/target and action plan
1.Feedback only Effective
2.Feedback + goal More effective
3.Feedback + goal + action plan Most effective
Modes of delivery
• Feedback only
• Feedback + goal
• Feedback + goal + action plan
• Additional BCTs– 73 interventions & 34
control arms
Behaviour change techniques
61 comparisons
8 comparisons
3 comparisons
• Non-intensive– ((group feedback) NOT (from a
supervisor or senior colleague)) OR ((individual feedback) AND (written format) AND (containing information about costs or numbers of tests without personal incentives))
• Moderate– (any other combination of
characteristics than described in Intensive or Non-intensive group)
• Intensive A&F– (individual recipients) AND
((verbal format) OR (a supervisor or senior colleague as the source)) AND (moderate or prolonged feedback)
Results
• Meta-analysis:– A&F effective: OR = 1.43 (95% CI = 1.28, 1.61)
– Moderate to strong heterogeneity (I2 = 61%)
• Meta-regression:– Feedback vs Feedback + Goal &/or Action plan
• 61 vs 24 comparisons, adj. p = 1.0• ?power
– Additional BCTs (intervention)• p = .006
The intervention “ProActive” Kinmonth et al, The Lancet, 2008
• Used BCT taxonomy approach to – Assess fidelity of delivery– Evaluate mechanism of action
• By linking intervention content to theory
The intervention and evaluation
• Based on psychological theory and evidence• Specified in detailed protocols/manuals• Delivered by trained professionals in 5 sessions
over 12 months • Results of RCT
– Increased activity by equivalent of 20 minutes per day– No difference between intervention and “control” groups
Implementation “add-on” study
• What worked?• Assess implementation
Hardeman, Michie et al (2008) Fidelity of delivery of a physical activity intervention: Predictors and consequences. Psychology and Health, 23, 11-24.
• How did it work?• Link component techniques to theory
Michie, Hardeman et al (2008) Investigating Theoretical Explanations for Behaviour Change: The Case Study of ProActive. Psychology and Health, 23, 25-39.
Intervention techniques
1. Give information2. Elicit questions3. Summarise message4. Set goals5. Self-monitor 6. Build motivation 7. Action plans
8. Use prompts 9. Use rewards 10. Build support 11. Review goals 12. Build habits13. Relapse prevention14. Generalise skills
Theories1. Theory of Planned Behaviour2. Relapse Prevention Theory3. Self-regulation Theory4. Operant Learning Theory
The implementation process
Theories of behaviour change
Techniques in manual
Delivery of techniques by professional
Participant response to intervention
Physical activity
Method
• 27 participants selected to study in depth • Intervention manual specified 208 behaviours
for delivering the 14 techniques• Intervention sessions tape recorded &
transcribed• All discussion in sessions relevant to
behaviour change was reliably coded into techniques and theories– Both of professionals and of participants
Question: How did the intervention work?
• 27 participants selected to study in depth
• Tape recorded and transcribed sessions
• All discussion in sessions relevant to behaviour change was reliably coded into techniques and theories– Both of professionals and of participants
Percentage of techniques delivered by professionals 45%
0
10
20
30
40
50
60
70%
Variation in implementation
0
10
20
30
40
50
60
70
session1
session2
session3
session4
fac 1 fac 2 fac 3 fac 4
Sessions: p<0.001 (Page test)Facilitators: p<0.001 (Kruskal-Wallis test)
Consensus of at least 3 of 4 raters for:
Techniques Theory
Build motivation
Give information Theory of Planned Behaviour
Set goals
Develop action plans
Self-monitoring
Review goals
Self-regulation Theory
Use rewards
Use prompts
Build support
Generalise skills
Build habits
Operant Learning Theory
Prepare for setbacks Relapse Prevention Theory
How were techniques distributed over the theories? (a) in protocol (b) delivered
0
5
10
15
20
25
30
35
% protocol
delivered
Theory of P
lanned
Behaviour
Self-re
gulation
Theory
Operant L
earning
Theory Relapse
Preventio
n Theory
Process linking theory and behaviour change
Theories of behaviour change
Techniques in manual
Delivery of techniques by professional
Participant response to intervention
Physical activity
How was the intervention received by participants?
Participants talk about behaviour change or maintenance was reliably coded into 17 components of four theories e.g.
Example from transcript
Theoretical component
Theory
Thinking about benefits of activity e.g. losing weight
Attitude Theory of Planned Behaviour
Parking car further away so has to walk further
Action plan Self-regulation Theory
Asking partner to remind him
Cue to action Operant Learning Theory
How was (a) professional (b) participant talk about behaviour distributed over the theories?
05
101520253035404550
% professionals
participants
Theory of P
lanned
Behaviour
Self-re
gulation
Theory
Operant L
earning
Theory Relapse
Preventio
n Theory
Which theories best accounted for change?
Although Self-regulation theory is the basis of the most commonly delivered
intervention techniques,Operant learning theory may be a better explanation for behaviour
change among participants
To develop more effective interventions …
We need to improve our scientific methods in the following areas:1. Specify intervention content in sufficient detail …2. Apply theories of behaviour change to intervention
development and evaluation, and to evidence syntheses
3. Match behaviour change techniques to mechanisms of action …
4. Define the extent to which interventions, and comparison conditions, use theory and evaluate how this relates to effectiveness.
5. Develop a model of behaviour as a starting point for systematic, theory-based intervention development.
4. Use of theory
• Many literature reviews conclude that behavioural interventions based on theory are more effective than those that are not
– E.g. Albarracin et al (2005), Downing et al (2006), Fisher & Fisher (2000), Gehrman & Hovell (2003), Glanz & Bishop (2010), Jemmott & Jemmott (2000), Kim et al (1997), Noar (2008), Wingood & DiClemente (1996)
• HOWEVER– Researchers of reviewed studies generally state
rather than demonstrate a theoretical base for the intervention
Need a method ….
• of assessing the extent to which interventions are based on theory
• Theory Coding Scheme: Checklist of 19 items in following categories:– is a theory mentioned – are relevant theoretical constructs targeted? – is theory used to select recipients or tailor
interventions?– are relevant theoretical constructs measured?– is theory tested?– is theory refined?
Michie S, Prestwich A. (2010) Are interventions theory-based? Development of a Theory Coding Scheme. Health Psychology, 29,1-8.
Application of the framework: to address …
• To what extent are interventions said to be theory based, actually theory based?
• Are theory-based interventions more effective than those not explicitly based on theory?
• Is intervention effect associated with– particular theories– particular use of theories
• e.g. to select techniques, target participants
• What is the association between theoretical base and intervention content?
Test of Theory I:meta-regression of smoking cessation counselling
• 3 Cochrane reviews– Individual, group, telephone behavioral support
• Inadequate intervention descriptions– Authors contacted
• 70 item taxonomy of BCTs for smoking cessation interventions
• Theory Coding Scheme
Stavri, Beard, Whittington, Michie & West, 2010
Results: published intervention descriptions
• For 14 studies where protocols available, more BCTs reported in the protocols (t(14)=3.63, p<0.01)
5.9
2.3
0
1
2
3
4
5
6
7
Number of BCTs
Protocol
Publisheddescription
Results: application of theory (n=92)
37
13
15
12
6
0
0 10 20 30 40
Is theory mentioned?
Are relevant theoretical constructstargeted?
Is theory used to select recipients ortailor interventions?
Are relevant theoretical constructsmeasured?
Is theory tested?
Is theory refined?
Number ofstudies
Theory use and effectiveness
• No association between theory use and intervention effectiveness
• Types of theory – 27 Transtheoretical Model – 8 Relapse Prevention Model – 1 Lazarus & Folkman’s model of stress– 1 compared Relapse Prevention Model with Health Belief
Model
Results: BCTs associated with intervention effectiveness
• Two BCTs were associated with intervention effectiveness in psychological interventions– Facilitate restructuring of social life (β=.56, R²=13.18%,
p=.02)– Advise on/facilitate use of social support (β=.37,
R²=3.45% p=.02)
Test of Theory II: meta-analysis of internet based interventions
Webb, Joseph, Yardley, & Michie (2010). Using the Internet to promote health behaviour change: A systematic review and meta-analysis of the impact of theoretical basis, use of behavior change techniques, and mode of delivery on efficacy. Journal of Medical Internet Research, 12, e4.
Use of Theory
k Q 95% CI d+
Theory used to select recipients 3 2.84 0.15 to 0.52 0.33Group of intervention techniques linked to group of constructs
6 9.85 0.03 to 0.43 0.23
Theory used to select intervention techniques 37 191.40
0.13 to 0.29 0.21
Targeted construct mentioned as predictor of behaviour
18 60.07 0.11 to 0.31 0.21
Theory used to tailor intervention techniques to recipients
11 67.75 0.07 to 0.34 0.21
Theory or model of behaviour mentioned 30 161.33
0.11 to 0.28 0.19
At least one intervention technique is linked to theory
19 93.65 0.09 to 0.29 0.19
Intervention based on single theory 12 57.13 0.05 to 0.32 0.18All constructs are linked to intervention techniques 10 47.70 -0.02 to 0.37 0.18
At least one construct is linked to an intervention technique
18 70.63 0.07 to 0.27 0.17
All intervention techniques are linked to theory 2
Use of theory associated with effect size
-0.20
-0.10
0.00
0.10
0.20
0.30
0 1 2 3 4 5 6 7
Extent of use of theory: number of items
Effec
t siz
e (d
)
Effect size = -.07 + .03
Currently applying Theory Coding Scheme to ….
• 146 papers reporting interventions to increase physical activity and healthy eating– 3 systematic reviews UCL, Leeds, Aberdeen
• Total sample size– intervention groups 12,282– control groups 12,248
• Watch this space …..
Prestwich, Whittington, Sniehotta & Michie, in preparation
To develop more effective interventions …
We need to improve our scientific methods in the following areas:1. Specify intervention content in sufficient detail in trial protocols and
published reports to allow accurate replication and evidence syntheses that can identify “active ingredients” within interventions.
2. Apply theories of behaviour change to intervention development and evaluation, and to evidence syntheses• allows questions to be answered about not just “what worked” in
interventions but “how they worked”.3. Match behaviour change techniques to mechanisms of action to
inform theory-based intervention development and evaluation-based theory development.
4. Define the extent to which interventions, and comparison conditions, use theory and evaluate how this relates to effectiveness.
5. Develop a model of behaviour as a starting point for systematic, theory-based intervention development.
Designing interventions
• Start from an analysis of the nature of the behaviour to be changed
• unless we understand the nature of the behaviours that need to change, unlikely our interventions will be maximally effective
Designing interventions ctd …
• Use a systematic approach to selecting from the range of interventions and policies available– Need a framework that meets criteria of
• comprehensive coverage,• coherence
– categories mutually exclusive and same level of specificity • linked to a model of behaviour
– Systematic review identified 18 existing frameworks, none met all these criteria
Michie, van Straalen & West 2010
The Behaviour System: Behaviour emerges from interactions between ….
Opportunity
Capability
Motivation Behaviour
Psychological or physical ability to enact the behaviour
Reflective and non-reflective mechanisms that activate or inhibit
behaviour
Physical and social environment that enables the behaviour
The Behaviour System: CMOB
Opportunity
Capability
Motivation Behaviour
Capability, Motivation and Opportunity must be present for a Behaviour to occur
The system is in dynamic equilibrium and a change in behaviour may require a sustained change in one or more of the other elements
The Behaviour Change Wheel: A system for choosing interventions and policies
Behaviour source
Interventions: activities designed to change behaviours
Policies: decisions made by authorities concerning interventions
Capability
Opportunity
Mo
tiva
tio
n
Physical
Psychol-ogical
Reflec-tive
Non reflect-ive
Physical
Social
Motivation
Beliefs about what is good and bad, conscious intentions and decisions as per e.g. Theory of Planned Behaviour
Emotional responses, desires and habits resulting from associative learning and physiological states
Non-reflective
Reflective
Reflective-Impulsive Model, Strack & Deutsch, 2004
PRIME Theory of Motivation, West, 2006
The Behaviour Change Wheel
Behaviour source
Intervention categories
Policy categories
Per
suas
ion
Education
Coercion
Environmental
restructuringInce
ntiv
isat
ion
Enablement/ resources
Capability
Opportunity
Mo
tiva
tio
n
TrainingR
estriction
Modelling
Physical
Psychol-ogical
Reflec-tive
Non reflect-ive
Physical
Social
Behaviour system influenced by an
intervention system which is enabled
and/or supported by a policy system
Intervention categoriesEducation Imparting knowledge e.g. on health
risks
Persuasion Using communication to induce belief or knowledge
Incentivisation Creating expectation of reward
Coercion Creating expectation of punishment or cost
Training Imparting skills
Restriction Reducing availability
Environmental restructuring
Changing the physical context
Modelling Providing an example for people to aspire to
Enablement/ resources
Increasing means/reducing barriers
The Behaviour Change Wheel
Behaviour source
Intervention categories
Policy categories
Regulation
Env
ironm
enta
l/so
cial
pla
nnin
g
Communication/
marketing
LegislationFiscal
Per
suas
ion
Education
Coercion
Environmental
restructuringInce
ntiv
isat
ion
Enablement/ resources
Capability
Opportunity
Mo
tiva
tio
n
TrainingR
estrictionService provision
Guidelines
Modelling
Physical
Psychol-ogical
Reflec-tive
Non reflect-ive
Physical
Social
Policy typesCommunication/ marketing Using print, electronic,
telephonic or broadcast media
Guidelines Creating documents that recommend or mandate practice
Fiscal Using the tax system
Regulation Establishing rules or principles of behaviour or practice
Legislation Making or changing laws
Environmental/ social planning
Designing and/or controlling the physical or social environment
Service provision Delivering a service
Conclusions: If we are to ….
• most efficiently build evidence about BC interventions– need to specify intervention content and link to theory
• design more effective interventions to change behaviour need to increase understanding of – nature of behaviour – range and nature of interventions/policies– mechanisms of action
• and links between these
More information from [email protected]
“He who loves practice without theory is like the sailor who boards ship without a rudder and compass and never knows where he may cast”
Leonardo Da Vinci, 1452-1519