 |
 |
 |
 |
||
 |
||
| Home | Command Papers | House Papers | Departmental Papers | Search | Site Map | Contact Us | Links | ||
 |
|
|
|
The Health
of Children and Young People
 1 Cigarette smoking
Heather Wardle and Barry Hedges
1.1 Introduction
1.1.1 Background
In 1998, the government produced a White Paper, entitled
Smoking Kills,1 which highlighted trends in smoking behaviour
and established key targets and strategies to reduce both smoking prevalence
and the inhalation of secondhand smoke. The White Paper noted that cigarette
smoking is a major risk factor for health. It is the primary cause of
cancer and heart disease, and contributes to many other fatal conditions
and illnesses. Life expectancy is appreciably reduced for those who
have smoked for most of their lives. The level of smoking in Britain
is therefore a key element in monitoring the health of the nation and
in setting targets for health improvement.
Smoking is strongly associated with the 'gap in healthy
life expectancy between those most in need and those most advantaged'.1
Reducing inequalities in health is a key policy objective, and a reduction
in smoking prevalence among disadvantaged groups would make an important
contribution to this.
The White Paper was particularly concerned with a trend
towards increased cigarette smoking among children and young adults.
Smoking cigarettes in early life may establish a habit that will persist
in later life and thus lead to higher smoking prevalence among adults.
Reducing the level of smoking among children and young adults is therefore
an important health policy aim.
1.1.2 Questions on cigarette smoking
Since its inception in 1991, the Health Survey for
England (HSE) has included questions relating to smoking. This has allowed
a large body of information about trends in cigarette smoking to be
collected. No information about smoking by children, however, was collected
until 1995, when children were first introduced into the survey. In
1997, the survey sample was specially designed to boost the number of
children surveyed. To maximise possibilities for analysis, the 1997
survey results for children (and also for young adults aged 16-24) were
aggregated with those from the two preceding surveys. The report on
the aggregate data from those three surveys is referred to below as
HSE 1997, and comparisons with it are made at appropriate places in
this chapter.
In 2002, the HSE survey sample was again designed to
boost the number of children surveyed, and also the number of young
adults (aged 16-24). The sample has been combined with those of comparable
age from the 2001 survey, and provides the basis for the current report
(HSE 2002).
Information about cigarette smoking is provided by
a number of other surveys, notably the General Household Survey (GHS),
which has plotted adult cigarette smoking prevalence in Great Britain
over several decades.2 Another important source for children
is the survey of Smoking Drinking and Drug Use among Young People in
England (SDD).3 This survey is carried out among secondary
school children aged 11-15 in England. The questions asked in HSE of
those aged 16-24 are the same as in GHS, and questions asked of those
aged 8-15 are similar to those in SDD. For the appropriate age group
(11-15), some comparisons between SDD and HSE will be made in this chapter.
In HSE, information about cigarette smoking for those
aged 8-17 was collected by means of a self-completion questionnaire.
This offers greater privacy for those who did not want to answer questions
in a way that might allow their parents to hear their replies. There
were two versions of the self-completion questionnaire, a brief one
for those aged 8-15 and a longer one for those aged 16 and 17. Children
aged 8-15 were asked if they had ever tried a cigarette, and, if so,
at what age they had first done so. They were also asked how often they
smoked cigarettes (if at all), and the number of cigarettes they smoked
(if any) in the past week. Those aged 16-17 were asked if they had ever
smoked and at what age they first did so. Instead of being asked about
the past week, they were asked general questions about their current
smoking behaviour, including how many cigarettes a day they smoked on
weekdays and at weekends, and the types of cigarette they smoked. If
they were ex-smokers, they were asked whether their smoking had been
regular or occasional. They were also asked if their parents had smoked
regularly when the informants were children.
The information obtained from those aged 18-24 (by
face to face interview) was similar to that obtained from those aged
16-17, with the addition of questions on attempts to give up smoking,
on medical advice received about smoking, and on smoking behaviour in
pregnancy. The questions relating to smoking during pregnancy are not
covered in the present chapter, but are included in the report on Maternal
and Infant Health. (See Section 1.6.2.)
Two questions dealt with exposure to other people's smoke. For those aged 0-12, it was established whether they were looked after for at least two hours a week by someone who smoked while looking after them. Those aged 13-24 were asked for how many hours a week they were exposed to other people's smoke. A final question on smoking was asked in the household questionnaire that was completed with a responsible adult in each household in advance of the individual interviews. This question established how many smokers there were in the household, providing another potential measure of exposure to smoke. A saliva sample was taken from informants aged 4 and over, and analysed for cotinine, a metabolite of nicotine which provides an indicator of recent exposure to tobacco smoke (see Section 1.4.1 below).
1.2 Children aged 8-15: self reported
cigarette smoking behaviour
1.2.1 Children's cigarette smoking status At age 8, only a small proportion of children said they had ever smoked (3% of boys, 2% of girls). The proportion increased with age, at first gradually but then more steeply, reaching about a quarter at age 13 and about half at age 15. In the lower half of the age range 8-15 the percentage who had ever smoked tended to be higher among boys than girls, but among those aged 14 and 15 it was higher among girls than boys. This pattern was also seen in Table 6.1 of HSE 1997. Several measures of current smoking activity are available, including cotinine levels (see Section 1.4 below). The two principal self-report measures for children are the proportion reporting that they smoked at least one cigarette a week, and the proportion reporting that they had smoked in the past week. The proportion of children aged 8-15 reporting that they smoked at least once a week was 2% among boys, 3% among girls. It remained small (under 1%) for both boys and girls from age 8 to 12, then increased to 13% for boys and 15% for girls at age 15. The biggest increase occurred between ages 14 and 15. Table 1.1 1.2.2 Children's cigarette smoking
in the past week
The percentage of children aged 8-15 claiming to have
smoked at least one cigarette in the past week was 3% for boys and 4%
for girls, one percentage point higher than claimed to smoke once a
week or more often. The percentage was again low (under 1%) until age
12, climbing by age 15 to 16% for boys and 19% for girls. These results
are similar to those shown in HSE 1997. Table 1.2
1.2.3 HSE, SDD, and self report
There is evidence that self-reports of smoking in HSE
are liable to under-represent true prevalence levels. It is believed
that some children interviewed at home may not wish to risk disclosing
their true level of smoking to their parents, in spite of the steps
HSE takes to protect the confidentiality of their replies. The survey
of Smoking Drinking and Drug Use among Young People in England (SDD)
is, in contrast, carried out by self-completion questionnaire in schools,
which avoids this problem. A comparison of the results for HSE 2002
with those of the 2002 SDD (limited to those aged 11-15, the approximate
age range covered by SDD) shows substantially higher prevalences recorded
by SDD, as shown in Table 1.3. For example, the percentage of boys aged
15 who smoked at least one cigarette a week was 13% in HSE, 20% in SDD,
comparable figures for girls being 15% and 26% respectively. Further
light is thrown on this issue in Section 1.4.2, where self-report is
compared with cotinine levels. Table 1.3
1.3 Self-reported cigarette smoking
among young adults aged 16-24
1.3.1 Young adults' cigarette smoking status 33% of young men and 35% of young women said they were current cigarette smokers. Over half of young adults said they had never smoked cigarettes at all (57% of young men and 52% of young women). 11% of young men and 13% of young women were ex-smokers (having smoked either occasionally or regularly in the past). The current smoking prevalence figures are similar to those shown by the General Household Survey 2001, which showed 33% for each sex.4 HSE 2001 showed 33% for young men and 35% for young women. In the late teens, self-reported cigarette smoking
prevalence continued to increase with age for both young men and young
women, continuing the upward trend seen among children in the early
teens. At age 16, reported cigarette smoking prevalence was 18% among
young men, rising to around 40% from age 22 onwards. Cigarette smoking
prevalence was 26% among young women at age 16, rising to around 40%
from age 20. Table 1.4
1.3.2 Number of cigarettes smoked
by young adults
Current cigarette smokers were classified as light
smokers (smoking under 10 cigarettes per day), moderate smokers (smoking
at least 10 but under 20 cigarettes per day) or heavy smokers (smoking
20 or more cigarettes per day). The proportions of young adults falling
into these light, moderate and heavy categories were respectively 13%,
13% and 5% (young men) and 16%, 14% and 4% (young women). A small proportion
smoked but were unable or unwilling to estimate how many a day they
smoked.
The percentage of young men who were heavy smokers
tended to increase with age. It was 2% at age 16, 4% to 6% from 17 to
22, 9% at 23 and 10% at 24. A less marked increase with age was seen
among young women.
Among current smokers, young men smoked an average
of 11.2 cigarettes a day, compared with 10.1 for young women. Table
1.4
1.3.3 Types of cigarette smoked
Young male smokers (18%) were more likely than young
female smokers (12%) to smoke hand rolled cigarettes. A much larger
proportion smoked filter tipped cigarettes (81% of young male smokers
and 86% of young female smokers):
At the lower end of the age range, the proportions
smoking hand rolled cigarettes were below the overall average and the
proportions smoking plain or untipped cigarettes were above average.
The proportion of young male smokers smoking filter tipped cigarettes
did not vary appreciably by age. The proportion of young female smokers
smoking filter tipped cigarettes did vary by age, but not in a consistent
manner. Table 1.5
1.4 Cotinine
1.4.1 Cotinine measurements in the Health Survey
Cotinine is a metabolite of nicotine. It is generally
considered the most useful of various biological markers that are indicators
of smoking. It can be measured in serum or saliva, saliva being used
in this survey. The sample of saliva was taken from those aged 4 and
over during the nurse visit. Cotinine has a half-life in the body of
between 16 and 20 hours, which means that it will detect regular smoking
but may not detect occasional smoking if the last occasion was several
days ago. A level of 15 or more nanograms per millilitre (ng/ml) is
regarded as indicative of smoking, and is unlikely to be due to anything
other than direct inhalation while smoking. Sources of cotinine other
than tobacco can for practical purposes be ignored. While the inhalation
of secondhand smoke can produce measurable levels of cotinine in non-smokers,
it is only rarely likely to produce a reading as high as 15 ng/ml.
The measurement of cotinine in HSE provides an objective check on self-reports of smoking behaviour, which are known not always to be accurate. This is particularly useful in the case of children, who may wish to conceal their behaviour from other household members who may be present at the interview. For this reason, cotinine level has been preferred to self-report in this chapter as the principal variable to be analysed by personal and socio-demographic factors. However, such analyses present a broadly similar picture of between-group variation whether self-report or cotinine levels are used. A valid saliva cotinine measurement is available for approximately 60% of those interviewed in the relevant age group (4-24). Table 1.6 examines the characteristics of these cases, and compares them with the total interviewed sample in respect of age and cigarette smoking status. Among those aged 16-24, the age profile of the cotinine sample was almost an exact match to the age profile of the total sample of young adults. Among those aged 4-15 the cotinine sample was slightly biased towards older children, compared with the total sample in this age group. Table 1.6 1.4.2 Cotinine levels and self
report
Levels of cotinine in saliva are affected by recent
smoking, and not by the individual's overall smoking pattern. Also,
a self-reported claim to be a smoker is not a precise concept, but can
cover a variety of smoking behaviour patterns. The two measures are
thus not measuring the same thing. Nevertheless, they are in fairly
close agreement except in the case of groups of informants for whom
there is reason to believe that they may be tending to under-report
their smoking levels, notably children interviewed in their homes.
In HSE 1997 (Chapter 6, Section 6.5), a detailed assessment was made of the relationship between cotinine levels and self-reported smoking behaviour among children and young people.5 It was found that a very high proportion of those claiming to be smokers had saliva cotinine levels of 15 ng/ml or more, as would be expected, while only a very small proportion of those who claimed never to have smoked reached a level of 15 ng/ml or more. Appreciable proportions of those who claimed to be ex-smokers, however, did have cotinine levels of 15 ng/ml or more. The data from HSE 2002 presents a broadly similar picture, confirming the general correspondence of cotinine levels with self report but pointing to a minority who smoke but do not admit to doing so. Among boys aged 8-15, the proportions with a cotinine level of 15 ng/ml or more were 1% of those claiming never to have smoked, 9% of those who said they had smoked only once or twice, and 20% of those who said they used to smoke but not now, comparable figures for girls being 1%, 6% and 20%. Among young men aged 16-24, the percentage was 8% among those who claimed never to have smoked or smoked only occasionally, and 37% among those who said they used to smoke regularly. Comparable figures for young women were 8% and 18%. Tables 1.7, 1.8
Children's cotinine levels and self reported cigarette smoking
The proportion of children aged 4-15 who had a cotinine level of 15 ng/ml or more was 4% for both boys and girls. Table 1.9
Only those aged 8-15 were asked to report their smoking,
so to compare self-report with cotinine, it is necessary to isolate
this group by excluding those aged 4 to 7. Of both boys and girls aged
8-15, 5% had a cotinine level of 15 ng/ml or more. This may be compared
with the 2% of boys and 3% of girls who said they smoked every week
(or 3% and 4% respectively if those who said they smoked occasionally
are included). The difference is consistent with other evidence that
children in HSE under-report their smoking. (Table not shown.)
Cotinine levels and self-reported cigarette smoking among young adults
aged 16-24
As reported above, 33% of young men and 35% of young
women aged between 16 and 24 were said they were current cigarette smokers
(Table 1.4). 35% of both young men and young women had a cotinine level
of 15 ng/ml or more. Compared with children, there is thus relatively
little tendency among young adults to under-report smoking levels. Insofar
as it happens at all, it tends to be among men, and among the younger
members of this age group. (Details not shown.) Table 1.10
1.4.3 Cotinine level, by age
Self-report, discussed above (Section 1.2.1), indicated
very little active smoking among children aged 8-12, self-reported cigarette
smoking prevalence being less than 1%, and not starting to increase
significantly until age 13. The proportion with a cotinine level of
15 ng/ml or more followed a broadly similar pattern, staying at or below
1% from age 6 to age 11 (slightly higher figures were found at ages
4 and 5: see Section 1.5.1 below). After age 11, the proportion with
a cotinine level of 15 ng/ml or more was higher than the self-reported
prevalence. For example, whereas self-report showed a cigarette smoking
prevalence of 13% for boys at age 15, 20% of boys aged 15 had a cotinine
level of 15 ng/ml or more. Comparable figures for girls at age 15 were
15% self-report, 19% a cotinine level of 15 ng/ml or more. Table
1.9
Among young men, the proportion with a cotinine level
of 15 ng/ml or more increased from 22% at age 16 to 37% at age 18, fluctuating
thereafter and reaching just over 40% at age 22-24.
Among young women, the proportion with a cotinine level
of 15 ng/ml of more increased from 28% at age 16 to 36% at age 18, fluctuating
thereafter and reaching 39% at age 22-24. The patterns for both sexes
are thus similar. Table 1.10
Figure 1A shows the way both these measures increase
with age for both males, and females. In order to smooth out random
variation between the relatively small samples in each age year, the
data are presented in the form of moving averages of three age years.
It will be seen that the difference between self-report and cotinine
measurement is broadly similar for both sexes, but generally slightly
greater, from the mid teens, among males than females. The two measures
begin to diverge as soon as smoking prevalence starts to rise, but eventually
converge (convergence being more complete, and occurring earlier, among
females than males, as was also seen in HSE 1997). Tables 1.9,
1.10, Figure 1A
Figure 1B uses the cotinine data to compare the age
patterns for each sex. It will be seen that this indicator suggests
a close similarity between males and females in the way their smoking
increases with age, though with some tendency for prevalence to be higher
among females than males in the mid teens, as also shown in the self-report
data. Figure 1B
1.4.4 Cotinine levels in socio-economic
sub-groups
Cotinine level by
equivalised household income quintile
Cotinine levels were estimated for those aged 4 or
more. The analysis in this section thus includes a number of children
at pre-smoking ages. The logistic regression model discussed in Section
1.7 below, however, includes only those aged 13 or more.
Equivalised household income is a measure of household income that takes account of the number of persons in the household. The percentage with a cotinine level of 15 ng/ml or more increased as equivalised household income decreased, among both children and young adults. For the highest and lowest income groups respectively, the proportions were: boys 1% and 6%, girls 2% and 7%, young men 30% and 45%, young women 27% and 49%. Table 1.11, Figures 1C, 1D
Cotinine level by NS-SEC of household reference person
NS-SEC is a classification of social position that
was introduced in the 2001 census. It has similarities to Registrar
General's Social Class. Informants are assigned to an NS-SEC category
based on the current or former occupation of the household reference
person. The five categories presented here are
managerial and professional, intermediate (e.g. clerical, administrative,
sales), small employers and own account workers, lower supervisory and
technical, and semi-routine and routine occupations.
For children aged 4-15, the proportion with a cotinine
level of 15 ng/ml or more was highest in lower supervisory/technical
and routine or semi-routine households, 5% and 6% respectively for boys,
4% and 7% for girls. It was lowest for those in managerial/professional
households, 2% for boys and 3% for girls.
1.6 Cigarette smoking and alcohol
consumption by young adults
Table 1.19 shows that there is an association, in young
adults, between smoking and alcohol consumption. The table divides the
sample into five categories according to their reported average frequency
of drinking over the past year. These five categories are a condensed
version of the eight categories in Table 2.8 in Chapter 2. The proportion
with a cotinine level of 15 ng/ml or more was 16% among young men who
said they never drank alcohol, rising across the drinking frequency
categories to 66% among those who said they drank almost every day.
The two comparable figures for young women were 19% and 51%. In part,
this relationship reflects the fact that both smoking and drinking increased
with age in the late teens, but as will be seen in the following section,
the relationship persisted, and was very marked, after controlling for
age. Table 1.19
1.7 Factors associated with smoking:
logistic regression
Multivariate logistic regression has been used to examine
the independent factors that are associated with having a cotinine level
of 15 ng/ml or more. Unlike simple standardisation, which only adjusts
for one variable such as age, this regression technique adjusts for
several explanatory variables simultaneously. Forward stepwise regression
models were used to identify variables that were significantly related
to the outcome. At a second stage these significant variables were entered
in the logistic regression. Separate models were run for men and women
aged 13 to 24 years, the lower age limit being chosen so as to concentrate
the analysis on the years in which there is a significant amount of
smoking.
Table 1.20 shows the adjusted odds ratios for factors that are associated with having cotinine levels of 15 ng/ml or more. All odds ratios are compared with a reference category (OR=1). They show how many times greater (if more than 1.0) or less (if less than 1.0) the odds are than for the reference category. 95% confidence intervals are shown for each odds ratio. If the interval does not include one, there is a significant difference between that odds ratio and that for the reference category. The p value shown indicates whether, taken as a whole, the variable is a significant predictor of the dependent variable. The variables included in the model were age (in three bands), reported average frequency of alcohol consumption, equivalised household income quintile, NS-SEC of household reference person, and the area-based Index of Multiple Deprivation (IMD). After controlling for the other variables in the model, age, alcohol consumption frequency and equivalised household income were significantly associated in both sexes with a cotinine level of 15 ng/ml or more. In addition, IMD was significant among young men (but not young women), and NS-SEC of the household reference person was significant only among women. The odds of having a cotinine level of 15 ng/ml or more were almost twice as high among males aged 16 or over, and at least twice as high among females aged 16 and over, as among those under 16. Odds ratios increased with increased drinking frequency. Those drinking at least once a week (but less than every day) had odds around four times as great as for those never drinking alcohol, and for those drinking (almost) every day the odds ratio was even higher. Compared with those in the bottom quintile (the lowest income category), men and women in all other income categories were less likely to have a cotinine level of 15 ng/ml or more. The differences between these four quintiles were less than their collective difference from the bottom quintile, especially in the case of women where the highest three income quintiles had approximately the same odds ratios. Although the overall result for NS-SEC was not significant in the case of men, the odds ratio for men in semi-routine and routine households was higher than for those in managerial or professional households (the reference category). In the case of women, those in lower supervisory and technical households, as well as those in semi-routine and routine households, had higher odds than those in managerial and professional households. Among men, area deprivation (IMD) was significantly associated with having a cotinine level of 15 ng/ml or more. Relative to the least deprived quintile, the odds were significantly higher in the two most deprived quintiles. For women, IMD was not overall a significant predictor, but the odds ratio between the two most deprived and the least deprived quintile was borderline significant. Table 1.20
1.8 Trends in smoking prevalence
over time
Trends in children's
smoking
In considering the trend data reported below, it should
be remembered that HSE 1997 refers to aggregate results for 1995-1997
(1996-1997 in the case of cotinine data), while HSE 2002 refers to aggregate
results for 2001-2002.
Children's self reported smoking behaviour did not differ greatly between HSE 1997 and HSE 2002. In HSE 1997, 3% of both boys and girls aged 8-15 reported smoking one or more cigarettes a week, compared with 2% for boys and 3% for girls in 2001-2002. The proportion claiming to have smoked cigarettes in the past week remained unchanged at 3% for boys, 4% for girls (aged 8-15). An analysis of children's cotinine levels (age 4-15) also failed to show any appreciable trend. Overall, 4% of boys and 5% of girls in 1996-1997 had a cotinine level of 15 ng/ml or more. In 2001-2002 the proportions were 4% for both boys and girls. Mean cotinine levels decreased, but not significantly. There was a decrease of three percentage points among boys at age 14 and 15, and of five percentage points among girls at age 14 and 15, but given the sample sizes in these age years, these decreases are not statistically significant. Tables 1.21, 1.23 The Surveys of Smoking, Drinking and Drug Use among Young People6 show a relatively low level of smoking in 1988, followed by a moderate increase up until about 1996, a subsequent decrease, and then fluctuations without clear trend.
Trends in young adults'
smoking
There was a significant decrease in cigarette smoking
by young men aged 16-24 between 1997 and 2002. Among young men, self-reported
smoking prevalence decreased from 37% to 33%, accompanied by a decrease
in the proportion with a cotinine level of 15 ng/ml or more, from 40%
to 35%. The proportion reporting heavy smoking (20 or more cigarettes
a day) fell from 8% to 5%.
Among young women changes were in the same direction, but less marked and not statistically significant: self-reported prevalence 36% down to 35%, percentage with a cotinine level of 15 ng/ml or more 37% down to 35%, proportion reporting heavy smoking 6% down to 4%. Tables 1.22, 1.24 These results can be put into context by looking at other data. A long-term decline in smoking halted in the late 1980s, as shown in Figure 1J, which
is based on General Household Survey (GHS) data for those aged 16-24.
It should be noted that since 1998 GHS has introduced a weighting system,
so that data relating to years up to 1997 are not strictly comparable
with those from 1998. However, within this narrow age group the effect
of the weighting is likely to be too small to affect the trend pattern
shown. Figure 1J
The GHS data in Figure 1J suggest the possibility that after the decline halted, there was a subsequent, but temporary, increase in young adults' smoking prevalence, peaking, around 1996 to 1998, although the fluctuations due to the relatively small sample sizes aged 16-24 make this interpretation uncertain. It is, however, supported by Figure 1K, which shows self-reported smoking prevalence figures from HSE from 1993 to 2001, for those aged 16-24. Prevalence increased up to about 1998, and then decreased. The fall in young adults' smoking prevalence between HSE 1997 and HSE 2002 could thus represent a falling-back after a temporary increase. It is worth noting that there has been some change in the age distribution of children between 1997 and 2002, with a relative increase in the number of older children and a corresponding relative decrease in the number of younger children. Since older children smoke more, such a change would have the effect of increasing the prevalence of smoking among children as a whole even when there is no actual increase for each age year separately. Adjusting for this does not materially affect the picture.   |
|||||||||||||||||||||||||||||||||||||||||||||||||||
