Methods

The annual Health Survey for England (HSE) is designed to provide samples representative of households in England in terms of age, gender, geographic location and sociodemographic circumstances. Each year, a new, random, nationally representative sample is selected using a two-stage stratified process. In the first stage, postcode sectors (the primary sampling unit (PSU)) are selected from the small-user postcode address file (PAF) after stratification by region and proportion of non-manual households. In the second stage, addresses are randomly selected from each of the selected PSUs. The postcode sectors are included in the sampling frame in proportion to the number of addresses within the sector, to ensure each address has the same likelihood of selection in the final sample. This selection procedure is used to obtain as nationally representative a sample as possible that allows efficient data collection. Although clustering increases the measurement error, stratification decreases it. In some years a core sample from the general population is supplemented with a boost sample of respondents from particular population groups. Full details of the HSE methodology are available in previously published reports10 11 12 13 14 15 16 17 18 and online (http://www.ic.nhs.uk/statistics-and-data-collections/health-and-lifestyles-related-surveys/health-survey-for-england). In short, all adults and up to two children in participating households are interviewed in the home, followed by a nurse visit to take biological measures (including saliva samples for cotinine) about 1 week later. Across years, 79% to 66% of eligible households participated in the interview, and in cooperating households 88% to 68% of children were interviewed, had their height and weight measured and saw the nurse. In common with many large surveys, response rates have declined somewhat in recent years. Questions on active smoking are asked every year of all respondents aged 8 years and over. To encourage more accurate self-reports, children aged 8 to 17 years are given self-completion booklets on smoking. Some of those aged 18 to 24 years also self-complete, but other adults are interviewed using a computer-aided schedule. Among children aged 8–15, the question on smoking status asks respondents to select one of the following six categories: “I have never smoked”, “I have only smoked once or twice”, “I used to smoke sometimes but I never smoke a cigarette now”, “I sometimes smoke, but I don’t smoke every week”, “I smoke between one and six cigarettes a week” and “I smoke more than six cigarettes a week”. There was no attempt to assess active smoking in children aged under 8. Adult smokers are defined by a positive response to the screening question “Do you smoke cigarettes at all nowadays?”.

Information about smoking in the home is gathered for the whole household from a single adult respondent at the initial interview. The household reference person or their partner is asked “Does anyone smoke inside this house/flat on most days?” For simplicity, we define as smoke free those homes where the response to this question was “no”, although it is in principle possible that such homes could be mostly rather than completely smoke free. The interviewer was explicitly instructed to count as a “no” instances where smoking by household members was reported, but only outside the home, and to count as “yes” instances where non-household members were reported as smoking in the home.

Cotinine

Cotinine is a sensitive and specific quantitative indicator of the extent of uptake of nicotine over the past few days and is accepted as the best available biomarker of exposure to second-hand smoke.19 Saliva specimens for cotinine were collected from children aged between 4 and 15 from 1996 onwards, and from adults from 1998. The nurse attempted to collect a saliva sample from adults by asking them to keep a dental roll in their mouths until it was saturated and then replace it in the sample tube. Children aged under 8 were given a straw to dribble saliva through into a sample tube.

Cotinine was assayed by a widely applied gas chromatographic method with a detection limit of 0.1 ng/ml.20 A single laboratory carried out all the assays. Regular internal quality controls were run to ensure comparability and reliability of results over time.10 11 12 13 14 15 16 17 18

Sample

For the purpose of this paper we took all available data on non-smoking children aged 4–15 years sampled in the core surveys from 1996 to 2007 inclusive. We excluded data from 1999 and 2004, when only ethnic minority groups had a nurse visit, and from 2000, when cotinine specimens were not collected. Those surveyed in boost samples, for example the boost sample of children aged 2–15 in 2006 had no nurse visit and contained no parental data and ethnic minorities in 2004 were also excluded as they were not representative of the population of England as a whole. We used unweighted data, as weighting for non-response was introduced only in 2003, and published trend tables show that this made minimal difference to observed percentage figures.21 The included data were linked to self-reported smoking behaviour of parental figures living in the same household. We defined non-smoking children as those who reported no current smoking and whose cotinine levels were below a cut-off point of 12 ng/ml for active smoking22 (all aged under 8 were assumed to be non-smokers unless their cotinine levels were over 12 ng/ml).

Statistical analysis

Since the distribution of cotinine concentrations in non-smokers is positively skewed, we subjected the data to logarithmic transformation, first assigning a value of 0.05, half the limit of detection, to undetectable concentrations. We report geometric mean concentrations and their 95% confidence intervals. Having established that there was a linear trend between year and log cotinine, we fitted linear trend terms across years to examine changes in log cotinine over time. All analyses were performed using the statistical software package SPSS V.14 (SPSS, Chicago, Illinois, USA).

Results

There were a total of 13 365 confirmed non-smoking children with measured saliva cotinine across all years. Table 1 shows the number in each year and reported parental smoking. There was a trend with time towards lower prevalence of parental smoking, with the proportion of children with two parents smoking declining from 11% in 1996 to 5% in 2007. There was a corresponding increase in the proportion of children with no smoking parent(s), from 60% in 1996 to 69% in 2007.

Table 2 shows the percentage of children living in a smoke-free home by year of survey and whether or not parents were self-reported current smokers. Where neither parent smoked, or there was a lone non-smoking parent, the overwhelming majority of homes were smoke free, ranging from 95% in 1996 to 99% in 2007. The proportion of homes that were smoke free when one parent was a smoker almost doubled across years, rising from 21% in 1996 to 37% in 2007. There was an even more marked increase where both parents were smokers, from 6% of homes being smoke free in 1996 to 21% in 2007.

The impact of parental smoking and smoke-free homes on children’s measured cotinine concentrations is shown in table 3 and illustrated in fig 1. As in previous studies, geometric mean cotinine values were strongly related to parental smoking habits (across all years combined, geometric mean cotinine levels were 0.23 ng/ml (95% CI 0.23 to 0.24) in children with non-smoking parent(s), 1.15 ng/ml (95% CI 1.10 to 1.20) when one parent smoked and 2.12 ng/ml (95% CI 2.00 to 2.20) when both parents were smokers. However, the observed values were greatly affected by whether or not the child lived in a home that was reported to be smoke free (table 3 and fig 1). For children with non-smoking parent(s) and living in a smoke-free home the mean across all years was 0.22 ng/ml, but 0.80 ng/ml when there was smoking in the home. The corresponding values for children with one smoking parent were 0.37 ng/ml and 1.67 ng/ml, and for those with two smoking parents 0.70 ng/ml and 2.46 ng/ml. Thus living in a smoke-free home offered children a considerable, but not complete, degree of protection against exposure to parental smoking.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Geometric mean saliva cotinine by parental smoking and whether or not home is smoke free; non-smoking children aged 4–15 surveyed as part of the Health Survey for England, 1996–2007.

There were strong trends across years for declines in geometric mean cotinine levels in children in smoke-free homes for the children of smokers and of non-smokers. In children with non-smoking parents, the largest group of children, the observed mean declined from 0.29 ng/ml in 1996 to 0.10 ng/ml in 2007, an absolute decline of 0.19 ng/ml and a relative decline of 65%, (linear trend across years F = 430.9 df 1,8027, p<0.001). In 2007 54% of these children had undetectable cotinine, compared with 16% in 1996. There were greater declines in cotinine in children living in smoke-free homes with two smoking parents, in whom an absolute fall of 0.87 ng/ml and relative decline of 77% was recorded (F = 16.7 df 1,141, p = 0.001), but no significant decline in children with one smoking parent (F = 3.39 df 1,890, p = 0.07). By 2007, just under 20% of children with smoking parents but living in smoke-free homes had undetectable cotinine, up from 11% in 1996.

There was also a tendency for cotinine levels to decline in children living with smoking parents in a home where smoking was allowed. By 2007, the observed mean in those with one smoking parent had declined by 0.46 ng/ml to 1.35 ng/ml from 1.81 in 1996 (linear trend F = 14.0, df 1,2776, p<0.001), and in those with two smoking parents by 0.67 ng/ml to 2.18 from 2.85 (linear trend F = 22.9, df 1,1120, p<0.001).

Smoking parents living in smoke-free homes were considerably lighter smokers than parents who smoked in the home. Across all years combined, their usual daily cigarette consumption averaged 8.2 cigarettes per day, compared with 16.2, and their average cotinine was 188.8 ng/ml compared with 318.9 ng/ml.

Discussion

Our findings show that a rapidly increasing proportion of parents in England who smoke have elected not to smoke in the home in recent years and this has resulted in marked reductions in their children’s exposure to second-hand smoke, as indexed by saliva cotinine concentrations. Where one parent smokes the proportion of smoke-free homes has approximately doubled from (going from 21% in 1996 to 37% in 2007), and where both parents smoke has tripled (from 6% to 21%). These are robust findings based on large and representative samples of the general population, and the availability of measured cotinine values serves to validate the home’s reported smoke-free status and to quantify the impact on children’s exposure. While some US studies have provided estimates of the prevalence of smoke-free homes where parents are smokers,23 24 25 26 our study appears to be the first in the UK and the first to quantify the impact on children’s measured exposure.

The degree of protection conferred on children when smoking parents refrain from smoking in the home is not complete, as their cotinine concentrations remain substantially elevated above those seen in children in smoke-free homes with non-smoking parents. Nevertheless, the mean cotinine in children with smoking parents but smoke-free homes in 2007 was similar to that observed in 1996 in children with non-smoking parents in smoke-free homes. Importantly, children whose parents smoked but did not allow smoking in the home on average had similar or lower cotinine levels than those whose parents did not smoke but allowed smoking.

Parents who smoke, but not in the home, are considerably lighter smokers than those who smoke inside the home; their cigarette consumption and their nicotine intake from smoking are lower by more than 50%. It is plausible that lower nicotine dependence may have been one of the factors enabling them to adopt a smoke-free home, but it is also possible that eliminating smoking in the home facilitated a reduction in their cigarette consumption and smoke intake.

Our definition of a smoke-free home had some limitations, based as it was simply on the response to the question “Does anyone smoke inside this house/flat on most days?” From this we could not determine whether there was a formal household policy, or whether the de facto smoke-free status was the result of a series of informal decisions on the part of the adults residing there. Previous studies in the USA4 5 and Australia7 have asked more explicitly about household rules and policy. We could also not distinguish between homes that were completely or mostly smoke free. However, our operational definition of smoke free was given good support by measured cotinine concentrations.

We observed a considerable reduction in children’s exposure to second-hand smoke between 1996 and 2007 even when one or both parents were smokers and there was regular smoking in the home. Some of this decline was no doubt attributable to reductions in exposure outside the home due to general societal changes. However, the magnitude of the reductions was too great to be wholly explained by this. It is tempting to speculate that smoking parents may have been influenced by the wider move towards smoke free, so that, even though they continued to smoke in the home, they made increasing efforts to limit smoking in the presence of their children.

Our results show that children’s exposure to second-hand smoke has continued its marked secular decline in recent years. Several factors have probably contributed to this. Parental smoking prevalence has declined, and there have been reductions in smoking in society more generally that have led to lower exposure for children from non-smoking homes and where parents are smokers. In addition, smoke-free homes have become much more common in households containing smokers, perhaps indicating that smokers reason that if smoking is harmful in public places, then it is harmful in the home too, and that restrictions on smoking in the home are called for just as much as restrictions on smoking in public. It is striking that the years immediately preceding the introduction of the legislative ban on smoking in public places were characterised by a surge in the percentage of smoking parents going smoke free in the home. It is too early to say with confidence what the impact of the 2007 legislation will be, as our data include only the first few months of the ban’s implementation. However, given the strong existing trend, it seems entirely plausible that the move towards smoke-free homes will gain added momentum, rather than experiencing a setback.