Dr Jo Holliday

1
Dr Jo Holliday Graham Moore Professor Laurence
Moore
2
Study Aim

• CHETS Wales was commissioned by the Welsh
  Assembly Government to examine the impact of the
  Welsh smokefree legislation on SHA exposure
  amongst children. It aimed to asses
• Perceived exposure to secondhand smoke
• Displacement of parental smoking from public
  places into the home
• Perceived social norms relating to smoking

3
Study design

• Repeated cross-sectional study of nationally
  representative year 6 students in state
  maintained primary schools across Wales
• Baseline data collection carried out between
  February and March 2007, before the ban was
  introduced.
• Follow-up data collection carried out with a
  different group of students between February and
  April 2008.

4
School sample

• 118 schools were approached between December 2006
  and March 2007
• Schools unable to participate were replaced by
  another from within the same strata
• 76 schools agreed to take part in the study
• 1 school dropped out prior to collection of
  pre-legislation data

5
School characteristics

• The proportion of students entitled to free
  school meals ranged from 1.0 to 70.8 (mean
  entitlement 18.9).
• The number of year 6 students ranged from 6 to
  104 (mean 37) pre-ban and from 7 to 119 (mean 37)
  post-ban.
• 17 schools were denominational
• 22 were either Welsh medium or bilingual

6
Student sample

• Within each school, one year 6 (final year of
  primary school age 10-11) class was randomly
  selected to participate in the study, with all
  students in the class being involved
• Approximately 1,750 students included in the
  study at each time period (pre- and post-ban).
  The mean number of students who took part in in
  each school was 25 pre-ban (range 5-51) and 26
  post-ban (range 7-77).

7
The survey

• Parental approval to participate in the study for
  all students obtained through letters sent to the
  parents or carers of potential participants
• Students asked to provide
• questionnaire data (on own smoking behaviour and
  exposure to secondhand smoke in public places)
• saliva samples (for cotinine analysis to
  biochemically measure exposure to secondhand
  smoke).

8
Student response
9
Description of sample
10
Results

• Section 1
• Population change in SHS exposure
• Section 2
• Displacement of parental smoking into the home
• Section 3
• Perceptions of smoking as a normative behaviour

11
Population change in SHS exposure - statistical
analyses

• Distributions of pre and post legislation samples
• Modelled estimates of geometric mean cotinine
  concentrations from linear regression pre and
  post legislation
• Multinomial logistic regression models examining
  change in the distribution of cotinine from low,
  medium and high tertiles.
• Multinomial logistic regression models examining
  change in self-reported frequency of SHS exposure
  (i.e. about every day, sometimes, never, I dont
  know)
• Design adjusted chi-squared analyses for
  perceived exposure the previous day in various
  private and public locations

12
Cotinine distributions in pre and post
legislation samples

• Almost half of samples contained cotinine
  concentrations below limit of detection (0.1
  ng/ml)
• An approximately 5 increase in the percentage of
  children below this threshold was observed after
  introduction of legislation.
• Less clear change is apparent at higher ends of
  the distribution

13
Cotinine distributions in pre and post
legislation samples and change in mean cotinine
concentrations
14
Summary and limitations of linear analysis

• Whilst declines between survey years are
  observed, particularly at the bottom of the
  distribution, change in geometric means between
  years was not significant.
• BUT, as cotinine concentrations were low, with
  47 below the limit of detectability, linear
  analysis relied heavily upon imputation.
• This degree of reliance on imputation may
  compromise the validity of linear analyses.
  Hence, these were supplemented with categorical
  multinomial logistic regression analyses

15
Categorical analyses of population change in
cotinine concentrations
Frequencies and percentages of children providing
saliva samples with concentrations allocated to
each of three approximately equal tertiles pre
(2007) and post (2008) introduction of smoke-free
legislation in Wales.
Relative risk ratios from design adjusted
multinomial logistic regression models, for the
relative likelihood of children being categorised
to 1st or 3rd tertiles of salivary cotinine
distribution, (both unadjusted and adjusted for
age and family affluence), after the introduction
of smoke-free legislation.
16
Change in perceived SHS exposure in private
locations
17
Change in perceived SHS exposure in public
locations
18
Change in perceived frequency of SHS exposure
Perceived frequency of exposure to SHS before and
after smoke-free legislation in Wales.
Relative risk ratios from design adjusted
multinomial logistic regression models for the
likelihood of a child reporting perceived SHS
exposure about every day or never (both
unadjusted and adjusted for age and family
affluence), after smoke-free legislation in
Wales.
19
Summary of key findings

• Whilst no significant change in overall mean
  cotinine concentration was observed, there was
  significant movement from the middle of the
  distribution to the lower end.
• Perceived exposure in public though not private
  locations fell significantly.
• Perceived frequency of exposure also declined
  significantly

20
Displacement of parental smoking into the home -
statistical analyses

• Change in percentage of children reporting
  parent figures smoking in the home assessed using
  design adjusted chi-squared analyses.
• Linear and categorical analyses of cotinine
  concentration and perceived exposure repeated for
  each of the following subgroups based on
  childrens reports of parental smoking behaviour.
  
• Neither parent figure smokes in the home,
• mother figure only smokes in the home,
• father figure only smokes in the home,
• both parents smoke in the home.

21
Change in percentage of parent figures smoking in
the home
Numbers and percentages of children assigned to
each category relating to parental smoking
behaviour within the home (Chi-square3.15,
P0.03)
22
Change in mean cotinine concentrations by
parental smoking in the home
Geometric mean cotinine concentrations and 95
confidence intervals by number of parent figures
who smoke, adjusted for age and family affluence,
pre (2007) and post (2008) introduction of
smoke-free legislation in Wales
23
Percentages of children in each tertile of the
distribution before and after legislation by
parental smoking in the home
Frequencies and percentages of children allocated
to each tertile of the cotinine distribution by
i) number of parent figures who smoke, and ii)
number of parent figures who smoke within the
home.
24
Significance of change in cotinine distributions
after legislation by parental smoking in the home
Relative risk ratios from design adjusted
multinomial logistic regression models, for the
relative likelihood of children being categorised
to 1st or 3rd tertiles of the cotinine
distribution, adjusted for age and family
affluence, after the introduction of smoke-free
legislation.
25
Self reported frequency of SHS exposure by
parental smoking in the home
Frequencies and percentages of children reporting
each of four levels of frequency of SHS exposure,
pre (2007) and post (2008) introduction of
smoke-free legislation by i) number of parent
figures who smoke, and ii) number of parent
figures who smoke within the home.
26
Significance of change in perceived frequency of
exposure by parental smoking in the home
Relative risk ratios from design adjusted
multinomial logistic regression models for the
likelihood of a child reporting perceived SHS
exposure about every day or never, by number of
parent figures that smoke, adjusted for age and
family affluence, after smoke-free legislation in
Wales.
27
Summary of key findings

• There was a significant decline in the proportion
  of children reporting that parents smoke in the
  home.
• Percentages of children in the lowest third of
  the distribution increased for children with no
  parent figures who smoked in the home.
• Perceived frequency of exposure fell for those
  with no parent figures who smoked in the home.
• There was no evidence of displacement of parental
  smoking into the home.

28
Change in the perception of smoking as a
normative behaviour statistical analyses

• Change assessed using design adjusted chi-squared
  analyses
• Childrens estimates of prevalence of smoking
  amongst adults
• Childrens estimates of prevalence of smoking
  amongst children
• Childrens reports of the visibility of smoking
  outside public places

29
Change in estimates of smoking prevalence
Perceived social norms for child and adult
smoking pre (2007) and post (2008) legislation
(frequencies, percentages and p-values from
design adjusted chi-squared statistics)
30
Change in perceived visibility of smoking
Frequency and percentage responses to the item
How often do you see people smoking in the
street outside buildings (for example, outside
pubs, restaurants or offices)?
31
Summary

• After the introduction of smoke free legislation,
  SHS exposure fell significantly, only amongst
  those with low exposure at baseline, and amongst
  those whose parents did not smoke in the home.
• Though the home remained the main source of
  exposure, no evidence of displacement of smoking
  into the home was found.
• Though perceived exposure in the home did not
  change, there was a significant decline in the
  percentage of children reporting that both
  parents smoked in the home.
• Prevalence of SHS exposure in public locations
  fell, though this affected small numbers of
  children at baseline.

32
Study strengths and limitations

• Strengths
• Large, representative sample.
• Validated measures of SHS exposure
• Data collected by trained research staff
• Limitations
• Large proportions of children provided samples
  below the limit of detection, and hence linear
  analyses relied on imputation.
• No comparison group observed changes might have
  occurred without the legislation.
• Focuses only on one year group strength for
  internal validity, but impacts may differ for
  other age groups.

33
Future directions

• Forthcoming secondary analyses of CHETS data
• Impact of smoke free legislation on health
  inequalities SES subgroup analyses.
• Impacts of legislation on the occurrence of
  asthma and symptoms.
• Important areas for primary research
• Monitoring of longer term impacts of legislation
• Development of interventions to protect children
  from home-based SHS exposure (e.g. through
  assisting parents smoking cessation and
  encouraging home smoking bans).

34
The CHETS team

• Principal Investigator - Laurence Moore
• Project Manager - Jo Holliday
• Clerical Assistants - Anna Hamilton, Hayley
  Collicott
• Research Assistants - Heather Rothwell, Claire
  Pimm, Margaret Humphreys, Keith Humphreys, Rosie
  Salazar, Nancy West, Ellie Byrne, Rachel Clark
• Data analyst Graham Moore
• For further information contact Dr Jo Holliday
  hollidayj1_at_cardiff.ac.uk 029 2087 9053