Ineffective, incorrect and unethical: Where next for the TTM and smoking cessation research? Chris Bridle

1
Ineffective, incorrect and unethicalWhere next
for the TTM and smoking cessation
research?Chris Bridle¹Camille Alexis²
Simon Murphy³ ¹Warwick Medical School,
University of Warwick² Department of Primary
Care Population Studies, UCL³ Institute for
Society, Health Ethics, University of Cardiff
Chris Bridle, PhD, CPsychol Institute of
Clinical Education Warwick Medical School
University of Warwick Tel [phone]
Email C.Bridle_at_warwick.ac.uk
www.warwick.ac.uk/go/hpsych
2
Overview

• Background context
• TTM and a framework for intervention
• Available evidence and its limitations
• Solutions for meaningful research
• Our meaningful research
• Interpretation and conclusions

3
Background

• Behaviour influences health
• 25 of UK adults smoke 12 million people
• 120,000 smoking-related deaths in Britain pa
• Health influences policy
• PCTs funded to deliver smoking cessation services
• DoH targets for recruitment and effectiveness
• Policy influences practice
• Stage-based theories dominant in Primary Care
• Transtheoretical Model (TTM)
• Evidence influences ... ?

4
Transtheoretical Model(Prochaska DiClemente,
1983)
Experiential ? Processes of Change ?
Behavioural
Stages of Change
PreCon
Con
Prep
Action
Maint
Situational Self-Efficacy
Self-Efficacy
Decisional Balance
Temptations
Interventions that take account of an
individual's stage of change will be more
effective in promoting behaviour change than 'one
size fits all' interventions
Pros
Cons
5
Available Evidence
3½

• Systematic reviews x 4
• Stage-based interventions (Riemsma et al. 2002)
• TTM interventions for smoking cessation (Spencer
  et al. 2002)
• Stage-based interventions for smoking cessation
  (Riemsma et al. 2003)
• TTM interventions (Bridle et al. 2005)

6
Conclusion

• Little evidence that stage-based interventions
• are more effective than other stage-based
  interventions, non-stage-based interventions,
• or even usual care (Bridle et al. 2005)
• '... but Chris, what does that mean?'
• It means that three decades of research has
  failed to provide an answer to the most important
  question ...
• Does it work?

7
THE Question

• Does the TTM provide a framework for developing
  effective health behaviour change interventions?
• No Wrong, ... lack of evidence of effect is not
  the same as evidence of lack of effect (Doug
  Altman, or someone else who is very clever)
• Yes Wrong, ... lack of good quality evidence
  necessarily prohibits meaningful inferences about
  effectiveness
• Who knows? Correct, ... we dont know because
  there are important gaps and weaknesses in the
  evidence base

?
?
?
8
Why Gaps and Weaknesses?

• Little evidence in an obscure area? 18 RCTs
  published in 2 impact factor journals since Jan
  2006
• Flaws in the design, conduct and analysis of
  trials are repeated consistently incorrect
  research
• Flawed research contributes nothing meaningful to
  the evidence base ineffective research
• Repeated flaws are well recognised and their
  effects are well documented unethical
  research(ers)

Meaningful research fills gaps and strengthens
areas of weakness
9
Inconsistent Evidence

• Limitation / Explanation Solution
• Content theoretically incorrect - Fully tailored
  Vs stage-based
• Delivery conceptually incongruent - Responsive
  and evolving
• Concealed random allocation, etc. - Good
  practice guidelines
• Biased sampling - Demotivated demoralised
  smokers
• Too few participants - a priori sample size
  calculation
• Exposure variable - Computer-generated
  feedback
• Treatment comparability - Standardise
  intervention contact
• Exclusion of Ps from analysis - Include all
  participants ITT
• Use of surrogate endpoints - Behavioural -
  smoking status
• Premature assessment - Long-term effects (6
  months)

10
Present Study
Stage-based V's Fully tailored
Long term effects

• 12-month RCT evaluating the effects of
  differently tailored, computer-generated smoking
  cessation interventions among hard-to-reach
  smokers

Behavioural outcomes
Tailored but standardised intervention content
Demotivated demoralised losers
Essential design features that protect internal
validity are balanced against the demand for a
pragmatic intervention that has external validity
and potential to shape practice patterns within a
large health care system
participants
11
Key (Meaningful?) Questions

• Effectiveness How effective are stage-based and
  fully tailored interventions for promoting
  smoking cessation?
• Theory Is stage progression predicted by
  distinct processes of change and different stages
  of change?
• Practice Can TTM direct resource allocation by
  identifying smokers who are at-risk, unresponsive
  /or hard-to-reach?
• Process Do different tailoring levels have
  different effects at different stages and for
  different people?
• Research Can the study bring meaning to the
  evidence base by identifying sources of effect
  heterogeneity?

12
Expert system / computer programme

• Tailored, but standardised content
• Message generation for TTM variables and
  validated by key stakeholders
• Algorithms to combine messages based on TTM
  scores and tailored to demographic factors
• Pilot evaluation of feedback messages (i.e.
  intervention content) using current smokers
• Readability high, non-significant (ns)
  difference
• Personal relevance moderate, ns difference
• 7-day recall significantly higher for fully
  tailored

13
Conditions

• Stage-based (I1) Partially tailored feedback
  based only on the participants stage of change
• Fully tailored (I2) Feedback tailored to TTM
  variables and relevant participant demographic
  factors
• Usual care (C) NHS leaflet 'Giving Up for Life,
  which contains some stage-based information and
  advice

LREC insisted we provide our no intervention
control group with erm an intervention
Tossers
The study benefited greatly from the insightful
guidance provided by the LREC
14
Feedback Example Contemplation
Temporal Relevance

• Stage-based Stopping smoking at any age
  increases life expectancy. The benefits of
  stopping begin immediately - blood circulation
  improves (20 minutes) blood oxygen levels
  increase to normal (8 hours) the body is
  nicotine-free (48 hours) breathing is easier and
  energy levels increase (72 hours).
• Later, circulation improves and exercise becomes
  easier (2-10 weeks) lung efficiency increases by
  5-10 (3-6 months) risk of having a heart attack
  is halved (5 years) risk of lung cancer is
  halved and risk of heart attack is the same as
  someone who has never smoked (10 years).'

Fully tailored Stopping smoking before you are
30 years old will increase your life expectancy
to the same as someone who has never smoked. Once
you stop smoking your chances of having a heart
attack begin to fall after only 8 hours, and your
risk is halved after 5 years and, after 10 years,
your risk is the same as someone who has never
smoked. Stopping smoking halves the risk that
your children will be smokers, and having a
non-smoking spouse doubles your chances of
quitting permanently compared to smokers whose
spouse is also a smoker.
Personalised Self-relevance
Fear Arousal
Response Efficacy
Normative beliefs
Social Support
15
Outcomes

• Effectiveness
• Primary outcome Self-reported abstinence defined
  as having smoked no more than 5 cigarettes since
  previous follow-up (3 months)
• Secondary outcomes 7-day point-prevalence
  abstinence (no tobacco use), no. of 24-hour quit
  attempts, and stage movement (progression or
  regression)
• Effect mediators
• Treatment factors no. of feedback reports
  provided, and uptake and use of co-interventions,
  e.g. return to NHS SSS, NRT, non-nicotine
  pharmacotherapy
• Psychological factors change from baseline in
  situational self-efficacy for smoking abstinence,
  and pros and cons of behaviour change, i.e.
  decisional balance

16
Sample Sample Size

• Adult smokers who failed to quit after starting
  the NHS Stop Smoking Service's (SSS) cessation
  programme
• Database of failed quitters in 5 PCTs random
  selection of 500 from each database limited to
  previous 12 months
• a priori calculation based on predicted
  difference in proportions () of self-reported
  abstinence
• 3, 10 and 20 for C, I1 and I2 respectively
• 80 power to detect 7 difference at .05 sig.
  level (n220) with an anticipated attrition rate
  of 25 (n55)
• Sample size required n275 per group, i.e. N825

17
Design Overview

• Randomised controlled trial with concealed
  allocation stratified by stage of change
• Data collected via mailed questionnaire at
  baseline and 3, 6, 9 and 12 months
• Feedback mailed within 2 weeks of receiving
  questionnaire data
• Non-response protocol 2 tel. attempts 1 mailed
  reminder
• ITT assessment for treatment effect using
  logistic regression controlling for PCT and SoC
• Pre-specified subgroup analysis for potential
  moderators and mediates

18
Participant Characteristics
Characteristic () Conditions Conditions Conditions
Characteristic () Control (n280) I1 Stage (n290) I2 Full (n280)
Mean age, y 49 47 48
Female 62 61 65
British, White 87 84 83
Employed (FT) 75 77 72
Single, live alone 64 63 67
Education (school) 56 58 57
SES (low 4-5) 63 65 62
19
Baseline TTM Variables X SoC
Variable (mean SD) PreCont. n122, 14 Cont. n481, 57 Prep. n247, 29
Dependency ? 5.5 2.1 5.3 2.3 4.6 2.2
Self-Efficacy ? 15.6 4.6 19.3 6.2 19.8 6.7
Decisional Balance ? -2.3 3.2 0.6 3.3 2.1 3.2
PoC Behavioural ? 18.3 5.4 22.7 6.1 25.6 7.3
PoC Experiential ? 26.7 7.5 31.2 7.1 33.3 7.5
20
Effectiveness Outcomes Primary and secondary
outcomes as proportions ()
Control Control Control Control Stage-Based Stage-Based Stage-Based Stage-Based Fully Tailored Fully Tailored Fully Tailored Fully Tailored
Outcome 3 6 9 12 3 6 9 12 3 6 9 12
P-Abs¹ 9 7 5 4 16 10 8 5 6 11 14 12
7-Abs 11 7 6 7 26 10 5 5 12 15 17 12
24-Quit 18 11 8 5 29 22 11 7 16 9 10 8
Stage 18 10 5 2 17 11 6 4 33 27 23 14
Stage - 16 12 8 5 17 17 10 16 10 6 4 4
¹Accululative data all other data represent new cases ¹Accululative data all other data represent new cases ¹Accululative data all other data represent new cases ¹Accululative data all other data represent new cases ¹Accululative data all other data represent new cases ¹Accululative data all other data represent new cases ¹Accululative data all other data represent new cases ¹Accululative data all other data represent new cases ¹Accululative data all other data represent new cases ¹Accululative data all other data represent new cases ¹Accululative data all other data represent new cases ¹Accululative data all other data represent new cases ¹Accululative data all other data represent new cases
21
Effect Estimates
Odds of being a non-smoker at 3, 6, 9 and 12 months Odds of being a non-smoker at 3, 6, 9 and 12 months Odds of being a non-smoker at 3, 6, 9 and 12 months Odds of being a non-smoker at 3, 6, 9 and 12 months
Contrast OR (95 CI) Contrast OR (95 CI) Contrast OR (95 CI)
Follow-up I1 (C) I2 (C) I2 (I1)
3 months 1.79 (1.06, 3.03) 1.51 (0.79, 2.88) 0.36 (0.21, 0.65)¹
6 months 1.18 (0.67, 2.09) 1.41 (0.79, 2.51) 1.08 (0.63, 1.86)
9 months 1.56 (0.81, 2.98) 3.08 (1.69, 3.59) 1.98 (1.17, 3.33)
12 months 1.04 (0.48, 2.26) 3.89 (1.81, 8.31) 2.54 (1.33, 4.88)
¹I1(I2) OR 2.77 (1.54, 4.98) ¹I1(I2) OR 2.77 (1.54, 4.98) ¹I1(I2) OR 2.77 (1.54, 4.98) ¹I1(I2) OR 2.77 (1.54, 4.98)
22
Conclusions

• Stage-based intervention is significantly more
  effective than both usual care and fully tailored
  intervention in short-term - 3 months
• In the longterm stage-based intervention is no
  more effective than Usual Care and significantly
  less effective than fully tailored intervention
• Short-term evaluation of stage-based intervention
  may overestimate positive effects and
  underestimate, or miss, longer term adverse
  consequences
• Fully tailored intervention significantly more
  effective in promoting long-term behaviour change
  and stage progression, i.e. 6 months
• Beneficial effects of fully tailored intervention
  evident only in long-term assessment

23
Where, or What Next?

• Research moratorium Periodic cessation of
  research and evidence assessment to guide more
  meaningful research
• Methods reporting / quality Supplement CONSORT
  with guidelines for reporting intervention
  content
• Mechanisms of action (MoA) Identify active
  ingredients to maximise clinical and
  cost-effectiveness
• Effect assumption Habitual behaviours are not
  driven by attitudes and beliefs, but by
  contextual cues to action
• Intervention is effective for 50 of people 100
  of the time focus on antecedent attitudes and
  belief ?
• Intervention is effective for 100 of people 50
  of the time focus on concurrent context of
  change ?

24

• Thank you.
• Any questions now or later?
• Chris Bridle, PhD, CPsychol
• Institute of Clinical Education
• Warwick Medical School
• University of Warwick
• Tel [phone]
• Email C.Bridle_at_warwick.ac.uk
• Web www.warwick.ac.uk/go/chrisbridle