Parental smoking and child poverty in the UK: an analysis of national survey data

Our analyses combined findings from several national surveys, taking the most recent
available at the time of the study, to estimate the number of children living in relative
poverty by household structure; apply smoking prevalence data to estimate the number
of children living in poor households containing smokers; and then estimate the expenditure
of typical smokers in these households on tobacco. Finally we estimated the numbers
of children drawn into poverty if expenditure on smoking is subtracted from household
income. Where published survey sources did not provide data broken down into the required
level of detail we used conservative assumptions to generate estimates. The study
used publically available data and ethics approval and participant consent were therefore
not required.

Definition of poverty

Poverty was defined as living in a household with an equivalised net household income
before housing costs (BHC) below 60% of the median equivalised net household income
1]. Equivalised income is the sum of income after deductions of income tax, employee
and self-employed national insurance contributions and council tax for all household
members, rescaled to allow for household composition, to reflect the fact that larger
households need more income to maintain the same standard of living. The data were
equivalised using the modified OECD equivalence scale, using an adult couple with
no children as the reference point 17]. In 2011/12 the median equivalised household income per week was £427 BHC. Poverty
was therefore defined as an equivalised income BHC of £256 or less 1].

Numbers of children in poverty

We estimated numbers of children in poverty by household composition using data from
the Department for Work and Pensions’ 2012 Households Below Average Income (HBAI)
report 1]. This draws on data from approximately 20,000 households in the Family Resources
Survey and provides estimates of the number of all children broken down by parental
marital status, and the percentage of those children living in poverty. We combined
these figures with data from the report on the proportion of poor households with
one, two, or three or more children (calculated as 25%, 39% and 36% respectively)
to estimate the number of children living in poverty by parental marital status and
family size. A worked example of these calculations is provided in Additional file
1. In the HBAI report children are defined as those under 16, and those aged 16–19
who are dependent (living with parents and in full time education or in unwaged government
training).

Since the HBAI report does not provide data on the proportions of single parents who
are male and female, we used estimates of these proportions (9% and 91% respectively)
from the Office for National Statistics’ (ONS) 2012 Families and Household survey
to calculate the number of children living in poor households with a single mother
and a single father 18].

Smoking prevalence in poor households

To estimate the proportion of children in poverty with one or more parents who smoke,
we first estimated parental smoking prevalence in these households using data from
the 2012 Opinions and Lifestyle Survey 19]. Since the survey reports do not present smoking prevalence by poverty status, and
no other relevant survey data were available, we made the conservative assumption
that smoking prevalence in households in poverty would be the same as that in households
in the routine and manual occupational socio-economic group. The prevalence of smoking
among men and women in routine and manual occupations in Britain in 2012 was 33% and
32% respectively. In fact these figures are highly likely to underestimate smoking
prevalence among the poor, as among unemployed people the prevalence is substantially
higher (39% in 2012 19]).

Since the 2012 Opinions and Lifestyle survey indicated that smoking rates vary by
marital status as well as socioeconomic group, we weighted the estimates of smoking
prevalence in routine and manual groups by marital status. The survey estimated that
while smoking prevalence in the general adult population was 20%, in adults who were
single, married or cohabiting the rates were 27%, 14% and 33% respectively (these
figures were not available by sex or socio-economic group). We therefore weighted
smoking prevalence in men and women in relation to these figures to estimate smoking
prevalence in low socioeconomic status adults by sex and marital status (see Table 1).

Table 1. Prevalence of smoking by gender and marital status in routine and manual workers (%)

Number of children in poverty by smoking parental marital status and number of children
in household

These weighted smoking rates were then applied to estimate the number of children
in poverty with smoking parents. For single parent households, we simply applied the
smoking rates estimated for single men and women to the number of children in these
households. This gave us an estimate of the number of children in poverty living with
a smoking single mother or father. For two parent households, we needed to estimate
how many contained one smoker and how many contained two. We therefore combined the
prevalence data with estimates from an existing study of smoke-free homes and secondhand
smoke exposure in children in England by Jarvis et al 20]. This study included a nationally representative sample of 13,365 children, including
695 in 2007 on which our estimates were based. While a more recent estimate based
on a larger sample from the whole of the UK would have be preferable, this estimate
was the only suitable one available to us, and is likely to be reasonably representative
of the whole of the UK. From this we calculated that among parents who smoked in two
parent households, 65% were the only smokers, and 35% lived with an adult who also
smoked. A worked example of our calculations of the number of children with smoking
parents is provided in Additional file 1.

The cost of smoking in poor households

We estimated the average weekly cost of smoking to poor households by combining data
on the number of cigarettes smoked per day by routine and manual workers for men and
women with typical costs for manufactured cigarettes and hand rolling tobacco (HRT),
both licit and illicit.

Opinions and Lifestyle Survey data indicate that on average, female and male routine
and manual workers smoke 12 and 13 cigarettes per day respectively 19]. We estimated the number of packets of 20 cigarettes purchased by low-income manufactured
cigarette smokers per week by multiplying the number of cigarettes smoked per day
by seven, and dividing by 20; and the number of packets of HRT purchased by low-income
HRT smokers per week in the same way, but with the assumption that 50 grams of HRT
typically makes approximately 100 cigarettes 21].

To estimate the average weekly spend on manufactured cigarettes and HRT, we combined
our estimated weekly quantities purchased with 2012 Tobacco Manufacturers’ Association
(TMA) data. This indicates that the average cost of a licit packet of 20 cigarettes
was £7.72, and of 50 g HRT £16.11, and that illicit tobacco typically sold for half
the price of licit products 22]-25].

The proportion of type of cigarettes smoked by sex and age was obtained from the OPN
19] (Opinions and Lifestyle Survey). To make calculations more straightforward, smokers
that smoked both packeted cigarettes and HRT were added on to the category they mostly
smoked.

In the UK it is estimated that 73% of female and 59% of male smokers smoke mainly
manufactured cigarettes (66% of women smoke only packeted, and 6% also smoke HRT,
but mainly packeted. 52% of men smoke only packeted, and 7% also smoke HRT but mainly
packeted) 19]. HMRC estimates that 7% of packeted cigarettes smoked are illicit, as well as 35%
of HRT. Based on these figures, we estimated the proportion of smokers purchasing
each type of tobacco (licit packeted, licit HRT, illicit packeted, illicit HRT), and
hence the overall average spend on tobacco products.

It should be noted that our estimate is likely to be an overestimate if cheaper licit
products, illicit and hand-rolled tobacco are disproportionately consumed by those
in poverty.

Effect on poverty rates of subtracting tobacco expenditure from household income

We estimated the number of children effectively drawn into poverty if parental expenditure
on tobacco is subtracted from household income. We calculated how many children are
living in a household where the income is above 60% of the median income, but by less
than the average spend on tobacco.

The HBAI report provides data on households living between 60% and 70% of the median
income; i.e. those living just above the poverty line. We first calculated the number
of children who are living in households between 60% and 70% of the median income.
We then applied the same method used to calculate the number of children in poverty
with smoking parents described above, to estimate the number of children in households
between 60% and 70% of the median income with one or two smoking parents.

We calculated the low income thresholds for these income groups for different household
structures, which showed that the income difference between these income groups was
similar to the average weekly expenditure on tobacco for two smokers calculated in
the previous step. We therefore assumed that all children in two-smoker households
with a household income between 60% and 70% of the median income would be drawn into
effective poverty. Because the spread of the population living between 60% and 70%
of the equivalised median is fairly even 1], we also assumed that half of all children between these thresholds with one smoking
parent in two-parent households, or one smoking parent in a one-parent household,
would be drawn into effective poverty.