Active smoking has been recognised as harmful to the smoker for over six decades, since the landmark Doll and Hill publication (Doll 1950), but it was not until 1974 that the medical literature first discussed parental smoking, exposure to environmental tobacco smoke (ETS), and its effect on the child (Harlap 1974). There is now overwhelming evidence that parental smoking is associated with a range of adverse health effects for children (NHMRC 1997). Perhaps its most obvious association is with increased risk, increased severity, and greater likelihood of admission to hospital with lower (Strachan 1997) and upper (Strachan 1998) respiratory tract disease. An increasing body of evidence describes an association between parental smoking and children's increased risk of serious bacterial infections such as meningitis (Iles 2001). In addition, ETS exposure increases health service use and costs (Lam 2001).
Furthermore, parental smoking confers a significantly increased risk for sudden infant death syndrome (SIDS) (Golding 1997). This effect is present regardless of which parent is the smoker (Blair 1999), and is the strongest modifiable risk factor for SIDS. In addition, research across several continents over the last two decades has found children of smokers to have an increased risk of uptake in adolescence, perhaps as a result of role modelling and/or increased access to cigarettes.
Parental smoking is a common but preventable source of infant and childhood morbidity. The World Health Organization (WHO) has identified the need to reduce parental smoking as a key element of action to encourage health and development in early childhood, particularly among those living in difficult social and economic circumstances (WHO 1999). In some countries, strong relationships between socioeconomic status and environmental quality are evident, with strategies to reduce smoking and improve child health outcomes needing to be underpinned by recognition of the limited resources and control some individuals and families have over environmental and social situations.
Infants' and toddlers' exposure to smoking primarily occurs within the home environment, as this is where they spend most of their time. Older children may also be exposed to smoking in a variety of child care and educational settings in which they spend their time. As children increase their time spent in commercial and informal child care settings, the importance of child care workers' behaviours increases. Similarly, the environments in which young children are exposed extend beyond the home to include shopping centres, meeting places, and other social environments.
Tobacco cessation strategies and interventions to reduce environmental tobacco smoke have had mixed success. Systematic reviews have previously demonstrated that individual counselling increases cessation rates (Lancaster 2005) and that simple advice from a physician has a positive effect in triggering quit attempts (Stead 2008). In relation to children's exposure in utero and in the early years, smoking cessation interventions for pregnant women can be effective in terms of reducing smoking (Lumley 2009). Legislation for smoking bans in public places has been introduced in North America, Australia and in some European countries, and has been associated with reduced incidence of acute myocardial infarctions in adults, lower smoking prevalence and high levels of public support (Lemstra 2008), and reduced exposure to ETS in the workplace and in bars and restaurants (CDC 2007; Galan 2008). However, inconsistent effects of such bans on exposure to ETS in the home have been reported, with some detecting minimal change (Galan 2008) and others a significant change in reported exposure (Edwards 2008).
The objectives of this review are:
1. To evaluate the effectiveness of programmes for both the prevention and cessation of smoking by those who interact with children, including parents and other family members, child care workers, and teachers, and the effect on health outcomes in infants, toddlers and young children.
2. To examine and detail the indicators of intervention processes and to identify outcomes of importance to those involved in the care of children and young people.
A priori hypotheses:
1. Smoking cessation and prevention programmes are able to improve carers' knowledge and awareness of the effects of tobacco smoking on the health of children.
2. Smoking cessation and prevention programmes can produce behaviour change in carers, leading to a reduction in children's exposure to environmental tobacco smoke (ETS).
3. Smoking cessation and prevention programmes are able to reduce the short-term illnesses experienced by young children exposed to tobacco smoke (though attribution to a particular type of intervention may be difficult to determine).
4. There is a difference in the effectiveness of interventions aiming simply to change knowledge (and thereby expecting behaviour change to occur) compared with those explicitly aiming to change behaviours by effecting change in attitude or skills.
Criteria for considering studies for this review
Types of studies
Controlled trials with or without random allocation.
In this updated review, we have not evaluated the effects and impacts of recent legislative changes on smoking and ETS exposure, as this has been addressed in a previous review (Callinan 2010). We have therefore decided against including a greater diversity of study designs, such as before and after studies, interrupted time series studies and other methods appropriate to evaluating population level interventions.
Types of participants
People (parents and other family members, child care workers and teachers) involved with care and education of infants and young children (aged 0 to 12 years).
Types of interventions
We included all mechanisms for reduction of children's ETS exposure, and smoking prevention, cessation, and any other tobacco control programmes targeting the participants described above. These included health promotion, social-behavioural therapy, technology, and educational and clinical interventions.
We included studies where the primary aim was to reduce children's exposure to ETS (thereby preventing adverse health outcomes), but where secondary outcomes included reduction or cessation of familial/parental/carer smoking, or changes in infant and child health measures. We also included studies where the primary outcome was reduction or cessation of familial/parental/carer smoking resulting in reduced exposure for children.
We excluded studies of uptake of smoking by minors.
There was no restriction on who delivered the programmes. These could include researchers, general practitioners, midwives, paediatricians, community and hospital nurses, health promotion agencies, tobacco control and anti-cancer organisations and health departments.
Types of outcome measures
The primary outcome measures were children's exposure to tobacco smoke, child illness and health service utilisation, and the smoking behaviours of children's parents and carers. We included studies where the outcome was only parental or carer's smoking status.
We used biological verification of exposure to or absorption of ETS as the 'gold standard', but did not require it as an inclusion criterion. Where biological verification of exposure/absorption conflicted with parental report of exposure, we have taken the biologically verified result as correct.
Outcomes for children
- Exposure to environmental tobacco smoke (ETS): biochemical measures of children's exposure to ETS using air monitoring for levels of nicotine; other measures of ETS (including parent-reported behaviour change, described in next section)
- Absorption of ETS: biochemical measures of children's absorption of ETS through cotinine in urine, blood, saliva, or hair
- Frequency of childhood illness events, respiratory problems (changes in lung function or symptom scores)
- Use of health services: admission to hospital; frequency of use of general practitioners; frequency of medication use
Outcomes for parents and carers
- Behaviour change in relation to children's exposure to ETS: we noted any reported bans or restrictions on smoking at home or in other environments or designated smoking areas outside the home
- Smoking behaviour, including cessation, reduction or uptake. Biochemically validated measures of smoking behaviour (for example thiocyanates, cotinine levels in blood, urine or saliva), or self report.
- Maternal smoking status at postpartum
- Costs and cost-effectiveness associated with interventions and outcomes
We report biochemical confirmation of parental self-reported quit status or changes in behaviour such as moves to smoke outside, but did not exclude studies without this measurement. Biochemical validation was not used in the majority of these studies; however, there is conflicting evidence regarding the validity of self report of smoking status. Some authors suggest it is reasonably accurate in community settings (Dwyer 1986; Velicer 1992; Patrick 1994) whereas others suggest parental self reports of smoke consumption and ETS are frequently under-estimated (Jarvis 1987; Ford 1997; Matthews 1999). For example, in clinical situations where a clinician is the interviewer, social bias may influence the report towards the socially desired response.
Levels of nicotine or its breakdown products, by contrast, are often preferred as a measure of real reductions in smoking or ETS. Smoke exposure can be detected by hair cotinine (Zahlsen 1994; Nafstad 1997; Al-Delaimy 2002a; Al-Delaimy 2002b) and absorption by urinary cotinine (Jarvis 1984; Bakoula 1995). Long-term exposure is best estimated by hair nicotine, whereas urinary cotinine is more informative of short-term exposure. Cotinine is a metabolic breakdown product of nicotine with a half-life of about one day (Haley 1983). The half-life is longer in nonsmokers such as infants and young children (Idle 1990). Cotinine is concentrated in the urine by the kidney and so becomes a sensitive indicator of ETS exposure over the previous few days. Urine creatinine measurements may be used to adjust for urine concentration (Thompson 1990); the urinary cotinine creatinine ratio (CCR) measurement has become a common method for measuring the levels of short-term ETS exposure. Saliva cotinine approximates to blood cotinine concentrations and collection is simple and non-invasive.
Search methods for identification of studies
The search was updated by Nia Wyn Roberts, Outreach Librarian, Bodleian Health Care Libraries. We searched the Cochrane Central Register of Controlled Trials [The Cochrane Library, Wiley] (Issue 2011), Medline [OvidSP] (1948 – present), Embase [OvidSP] (1974 – present), CINAHL [EbscoHOST] (1980 – present), PsycINFO [OvidSP] (1967 – present), and ERIC [Proquest] (1966 - present). A search for articles from 2007-2011 was conducted in June 2011. The Trial Search Coordinator searched the CochraneTobacco Addiction Group's Specialised Register. The update searches were conducted in April 2012, and further update searches were conducted in September 2013.
The reports of all references identified as possibly being randomised controlled trials (RCTs) or controlled trials (CTs) were obtained and reviewed. Secondly, reference lists of all identified RCTs or CTs were checked to identify potentially relevant citations. We made enquiries regarding other known published or unpublished studies so that these results could be included in our review.
Data collection and analysis
Two reviewers independently undertook assessment of quality and extraction of study details and results. For this update, RB reviewed all of the studies, MS reviewed three quarters of the studies, and RR, AP and NP each reviewed a proportion of the remaining studies, and compared results. We created a data extraction spreadsheet in Microsoft Excel.
We extracted information on methods, participants, intervention and control conditions, and outcomes. We were particularly interested in aspects of intervention development that may have contributed to a stronger, more appropriate or sustained intervention. We extracted information on the theory underlying the intervention development and content, process indicators and descriptions of community consultation and/or participation in the planning and implementation of the intervention, incentives (if present), and concerns of intervention programmes. We also recorded any information about costs, either in terms of evaluations of cost-effectiveness, or simply where costs were mentioned. Where possible we examined outcomes by gender, age and socio-economic status.
We resolved differences between reviewers' extraction results by discussion or by consultation with a third reviewer. Given the heterogeneity of study design and characteristics, we considered a quantitative estimate of effect to be inappropriate. The synthesis is therefore narrative.
Assessment of risk of bias in included studies
Risk of bias was assessed for all included studies, including those included in previous version of this review, by two authors independently of each other. Risk of bias was categorised as High, Low, or Unclear in accordance with the methods described in the Cochrane Handbook (Higgins 2011) for randomisation, allocation concealment, incomplete data, blinding of outcome assessment, and other bias. For the current update this was undertaken by all authors who undertook data extraction. Risk of bias for previously included studies was assessed by Jamie Hartmann-Boyce from the Cochrane Tobacco Addiction Group. Differences were resolved by discussion.
(1) Sequence generation (checking for possible selection bias)
We have described the methods used to generate the allocation sequence, and have assessed the methods as:
- low risk of bias (any truly random process, e.g. random number table, computer random number generator);
- high risk of bias (any non random process, e.g. odd or even date of birth, hospital or clinic record number); or
(2) Allocation concealment (checking for possible selection bias)
We have described the method used to conceal the allocation sequence in sufficient detail to determine whether intervention allocation could have been foreseen in advance of, or during recruitment, or changed after assignment. We have assessed the methods as:
- low risk of bias (e.g. telephone or central randomisation; consecutively numbered sealed opaque envelopes);
- high risk of bias (open allocation; unsealed or non-opaque envelopes; alternation; date of birth);
(3) Blinding (checking for possible detection bias)
We have described the methods used, if any, to blind study participants and personnel from knowledge of which intervention a participant received. With educational interventions (such as those assessed in this review) it is often not possible to blind participants to group allocation, and hence we did not evaluate blinding based on performance bias but rather based solely on the potential to introduce detection bias. It is possible for outcome assessors to be blind to group allocation and we have noted where there was partial blinding. We have assessed the methods as high risk of bias, low risk of bias, or unclear.
Where findings were objectively measured (biochemical validation, household air nicotine monitors) we assessed blinding as adequate to prevent detection bias.
(4) Incomplete outcome data (checking for possible attrition bias through withdrawals, dropouts, or protocol deviations)
Within each included study, we have described for each outcome or class of outcomes the completeness of data including attrition and exclusions from analysis. We have noted whether attrition and exclusions were reported, the numbers included in the analysis at each stage (compared with the total randomised participants), reasons for attrition or exclusion where reported, and whether missing data were balanced across groups.
(5) Other bias (e.g. selective reporting bias)
We noted any other potential sources of bias that were not related to the above four.
Overall risk of bias
We made explicit judgements about whether studies were at high, moderate, or low risk of bias, according to the criteria given in the Cochrane Handbook (Higgins 2011). With reference to 1 to 5 above, we assessed the likely magnitude and direction of bias and whether we considered it likely to impact on the findings.
Description of studies
We included 57 studies in this review, 21 of which were identified in the most recent update (Van't Hof 2000; Pulley 2002; Culp 2007; Ekerbicer 2007; French 2007; Ralston 2008; Hannover 2009; Hovell 2009; Borrelli 2010; Baheiraei 2011; Butz 2011; Halterman 2011; Herbert 2011; Wilson 2011; Patel 2012; Phillips 2012; Stotts 2012; Chellini 2013; Prokhorov 2013; Ralston 2013; Tyc 2013). The characteristics of the included studies are summarized below. Further detail is available in the Characteristics of included studies table.
A further nine studies were identified for which the outcome data are not yet available, four of which were identified in the previous update (Sockrider 2003; Wilson 2005; Chan 2006b; Johnston 2010; Ortega 2010; Rosen 2011; Chan 2012; Hutchinson 2013; Stotts 2013). Information about these ongoing studies is provided in the Characteristics of ongoing studies table.
Twenty-three studies were excluded from the review (Philips 1990; Meltzer 1993; Murray 1993; Campion 1994; Wilson 1996; Manfredi 1999; Spencer 2000; Cookson 2000; Emmons 2000; Arborelius 2001; Badger 2003; Okah 2003; Morgan 2004; Loke 2005; Turner-Henson 2005; Stepans 2006; Klinnert 2007; Burmaz 2007; Oien 2008; Hovell 2011; Gadomski 2011; Kegler 2012; Winickoff 2013). These were excluded for a variety of reasons, the most common being: study design; participants not meeting inclusion criteria; outcomes not related to environmental tobacco smoke exposure; and no outcome data. Further information is available in the Characteristics of excluded studies table.
Of the 57 studies reported in this review, only seven were targeted at the population or community level. The majority of studies targeted parents within healthcare contexts, with 23 targeting parents in 'well child' settings and 24 reporting interventions in 'ill child' healthcare settings. A further two studies reported on interventions in paediatric clinics but did not designate whether they were in the context of 'well child' or 'ill child' settings, and a further one included both well and ill child visits.
Interventions targeted at population or community settings (for example, communities, schools etc)
This review identified seven eligible studies of interventions targeted at the population or community level. One of them evaluated outcomes for smoking mothers who called a telephone smoking cessation assistance counselling service (Davis 1992). Three studies examined the effectiveness of school-based strategies (Zhang 1993; Elder 1996; Ekerbicer 2007) but used different approaches to limiting children's exposure to ETS. Halterman 2011 also recruited parents of children in school, but specifically targeted asthmatic children ('ill child' setting). A community-based intervention from the USA used trained lay bicultural and bilingual community health advisors to work with Latino families to problem-solve and to develop strategies to lower children's exposure to tobacco smoke in the home (Conway 2004). Herbert 2011 recruited families to participate in the study through five public health nursing offices, eight daycare centres, and kindergartens on Prince Edward Island. Prokhorov 2013 recruited from a cohort of Hispanic Americans, "mano a mano" in Houston, Texas.
Opportunistic interventions targeted at parents of children in the 'well child' healthcare setting
Compared to the relatively few community and population level interventions identified by this review, we found far more studies that evaluated interventions within 'well child' healthcare settings. Twenty-three included studies examined the effect of interventions delivered to parents in this context, and these recruited participants postnatally, at 'well child' health visits or at infant immunisation clinics. Thirteen of these studies were peripartum, recruiting participants via maternity hospitals, from their records, or via midwives and general practitioners (Woodward 1987; Greenberg 1994; Severson 1997; Armstrong 2000; Van't Hof 2000; Emmons 2001; Ratner 2001; Pulley 2002; Schonberger 2005; Wiggins 2005; Culp 2007; French 2007; Hannover 2009). 'Well child' health check visits to a doctor or maternal child health nurse were used by Chilmonczyk 1992; Vineis 1993; Eriksen 1996; Fossum 2004; Zakarian 2004; Abdullah 2005; Kallio 2006; Winickoff 2010; Baheiraei 2011. Chellini 2013 recruited from hospital and public health facility waiting rooms, and also from supermarkets.
Opportunistic interventions targeted at parents of children with health problems
Interventions conducted in the 'ill child' health care setting were reported in 24 studies. Of these, 13 were interventions targeted at the parents of children with respiratory problems (Hughes 1991; McIntosh 1994; Wahlgren 1997; Irvine 1999; Wilson 2001; Hovell 2002; Krieger 2005; Ralston 2008; Borrelli 2010; Butz 2011; Halterman 2011; Wilson 2011; Stotts 2012). Halterman 2011 targeted children in school with asthma, rather than recruiting from a health care setting. Nine studies were conducted in non-respiratory 'ill child' health care settings (Groner 2000; Hovell 2000; Wakefield 2002; Kimata 2004; Chan 2005; Chan 2006a; Hovell 2009; Phillips 2012; Tyc 2013). Patel 2012 and Ralston 2013 targeted children presenting to the emergency department, approximately 40% of whom had a respiratory presenting complaint. Hovell 2000 and Hovell 2009 recruited mothers from a Special Supplemental Nutrition Program for Women, Infants and Children, and looked at the effectiveness of counselling on smoking rates and children's ETS exposure among women of low income, high risk and ethnically diverse backgrounds. A further two studies conducted in paediatric clinics do not make clear whether they are in the context of 'well child' or 'ill child' health visits (Curry 2003; Nuesslein 2006), while Yilmaz 2006 recruited children visiting paediatric clinics for either primary conditions or a 'well child' visit.
Main target of intervention
Reduction of children's ETS exposure can be achieved by encouraging avoidance of children's exposure to cigarettes smoked, such as the child or the smoker moving to a different location, reducing the number of cigarettes smoked by parent or carer, or the smoker ceasing to smoke altogether. The aims of the studies identified by this review were heterogeneous. Only smoking and ETS targets are considered here; other intervention components, such as healthy eating (e.g. Elder 1996), asthma management (e.g. Hughes 1991), or household safety (e.g. Culp 2007) are not described.
Of the 57 included studies, 15 aimed solely for parental or carer smoking cessation or reduction (Vineis 1993; Zhang 1993; Severson 1997; Groner 2000; Emmons 2001; Wakefield 2002; Curry 2003; Kimata 2004; Chan 2005; Wiggins 2005; Kallio 2006; Nuesslein 2006; Ralston 2008; Borrelli 2010; Ralston 2013). Sixteen studies aimed solely for reducing children's exposure to cigarettes smoked (Chilmonczyk 1992; Davis 1992; Elder 1996; Wahlgren 1997; Hovell 2000; Wilson 2001; Pulley 2002; Baheiraei 2011; Butz 2011; Herbert 2011; Wilson 2011; Patel 2012; Stotts 2012; Chellini 2013; Prokhorov 2013; Tyc 2013), while twenty one studies aimed for a combination of parental or carer cessation, reduction or avoidance (Woodward 1987; Hughes 1991; Greenberg 1994; McIntosh 1994; Eriksen 1996; Irvine 1999; Armstrong 2000; Hovell 2000; Conway 2004; Fossum 2004; Zakarian 2004; Abdullah 2005; Krieger 2005; Schonberger 2005; Chan 2006a; Yilmaz 2006; Culp 2007; Ekerbicer 2007; Hovell 2009; Winickoff 2010; Halterman 2011). Five studies aimed to prevent reuptake of smoking postpartum (Van't Hof 2000; Ratner 2001; French 2007; Hannover 2009; Phillips 2012).
All studies aimed to achieve changes in behaviour in some way in order to reduce child ETS exposure. Eight studies did not expressly include an education or knowledge-building component in their interventions, but instead targeted change in attitudes and behaviours (Chilmonczyk 1992; Zhang 1993; Wahlgren 1997; Hovell 2000; Curry 2003; Zakarian 2004; Chan 2005; Nuesslein 2006).
Location of studies
The majority of studies were from high income countries. Thirty-six studies were from North America, with 33 from the USA (Chilmonczyk 1992; Davis 1992; Greenberg 1994; McIntosh 1994; Elder 1996; Severson 1997; Wahlgren 1997; Groner 2000; Hovell 2000; Van't Hof 2000; Emmons 2001; Wilson 2001; Hovell 2002; Pulley 2002; Curry 2003; Conway 2004; Zakarian 2004; Krieger 2005; Culp 2007; French 2007; Ralston 2008; Hovell 2009; Borrelli 2010; Winickoff 2010; Butz 2011; Halterman 2011; Wilson 2011; Patel 2012; Phillips 2012; Stotts 2012; Prokhorov 2013; Ralston 2013; Tyc 2013) and three from Canada (Hughes 1991; Ratner 2001; Herbert 2011). Three studies were from Australia (Woodward 1987; Armstrong 2000; Wakefield 2002), with two from each of the UK (Irvine 1999; Wiggins 2005), Germany (Nuesslein 2006; Hannover 2009) and Italy (Vineis 1993; Chellini 2013). There was one study reported from each of Finland (Kallio 2006), Japan (Kimata 2004), Sweden (Fossum 2004), the Netherlands (Schonberger 2005) and Norway (Eriksen 1996). Ten of the studies in high income countries specifically targeted disadvantaged, low income and/or culturally diverse populations. Seven studies were from low or middle income countries, with four from China (Zhang 1993; Abdullah 2005; Chan 2005; Chan 2006a), two from Turkey (Yilmaz 2006; Ekerbicer 2007) and one from Iran (Baheiraei 2011).
Twenty-one studies targeted mothers only (Chilmonczyk 1992; Davis 1992; Greenberg 1994; Severson 1997; Armstrong 2000; Groner 2000; Hovell 2000; Van't Hof 2000; Ratner 2001; Pulley 2002; Curry 2003; Fossum 2004; Zakarian 2004; Wiggins 2005; Nuesslein 2006; Yilmaz 2006; Culp 2007; French 2007; Hannover 2009; Phillips 2012; Chellini 2013). Hovell 2009 targeted mothers, but invited other family members to participate in counselling. One study (Chan 2006a) targeted fathers through educating their non-smoking wives. Twenty-two studies targeted both parents (Woodward 1987; Hughes 1991; Vineis 1993; McIntosh 1994; Eriksen 1996; Irvine 1999; Emmons 2001; Wilson 2001; Hovell 2002; Wakefield 2002; Conway 2004; Kimata 2004; Abdullah 2005; Chan 2005; Schonberger 2005; Kallio 2006; Ekerbicer 2007; Ralston 2008; Winickoff 2010; Baheiraei 2011; Herbert 2011; Tyc 2013). Zhang 1993 targeted fathers only, Borrelli 2010, Wilson 2011, Patel 2012 and Ralston 2013 targeted a caregiver(s), Elder 1996 targeted teachers only, Wahlgren 1997, Butz 2011 and Stotts 2012 targeted families and Krieger 2005, Halterman 2011 and Prokhorov 2013 targeted households.
We stratified studies according to the age group of the children: infants (less than one year); preschoolers (up to age six); and school age (six to twelve years). Twenty studies examined measures to reduce ETS exclusively for infants (Woodward 1987; Chilmonczyk 1992; Vineis 1993; Greenberg 1994; Severson 1997; Armstrong 2000; Van't Hof 2000; Ratner 2001; Pulley 2002; Fossum 2004; Zakarian 2004; Abdullah 2005; Wiggins 2005; Culp 2007; French 2007; Hannover 2009; Winickoff 2010; Baheiraei 2011; Phillips 2012; Stotts 2012). Measures to reduce ETS for children up to and including preschool age were examined by eight studies (Davis 1992; Eriksen 1996; Hovell 2000; Emmons 2001; Schonberger 2005; Hovell 2009; Herbert 2011; Patel 2012), while measures for children up to and including school age were considered by sixteen studies (Hughes 1991; Zhang 1993; McIntosh 1994; Elder 1996; Irvine 1999; Groner 2000; Wilson 2001; Wakefield 2002; Conway 2004; Kimata 2004; Krieger 2005; Kallio 2006; Ekerbicer 2007; Butz 2011; Halterman 2011; Wilson 2011). Eight studies examined interventions to reduce ETS that included older age groups: Wahlgren 1997 included parents of children aged 6 to 17 years; Hovell 2002 included parents of children aged 3 to 17 years; Chan 2006a included parents of children from birth to 15 years; Yilmaz 2006 included mothers of children under 16 years of age and Borrelli 2010, Chellini 2013, Prokhorov 2013 and Tyc 2013 included children under 18 years of age. Five studies did not provide the ages of the children (Curry 2003; Chan 2005; Nuesslein 2006; Ralston 2008; Ralston 2013).
Thirty of the 57 studies expressly employed a theoretical framework in the design and/or development of the intervention. Eleven studies used motivational interviewing (Emmons 2001; Curry 2003; Chan 2005; French 2007; Hannover 2009; Borrelli 2010; Baheiraei 2011; Halterman 2011; Phillips 2012; Stotts 2012; Ralston 2013). Four used a social learning model (Greenberg 1994; Elder 1996; Conway 2004; Fossum 2004) and six used the stages of change component of Prochaska's transtheoretical model (Abdullah 2005; Krieger 2005; Ralston 2008; Winickoff 2010; Patel 2012; Ralston 2013). Ralston 2013 used stage of change and motivational interviewing. Social cognitive theory was used by two studies (Krieger 2005; Borrelli 2010). McIntosh 1994 developed the activities for the parent manual based on behaviour modification theory. Wahlgren 1997 tailored the programme to individual families, and incorporated a number of behavioural modification techniques, including stimulus control, shaping, personal feedback, and contingency contracting. Groner 2000 employed the health belief model, and Wakefield 2002 used a harm minimisation approach, based on previous research indicating that restrictions produced significantly lower urinary cotinine levels. Ratner 2001 utilised Marlatt's relapse model. Chan 2006a used Fishbein's theory of reasoned action and Ajzen's theory of planned behaviour in the development of their educational intervention. Hovell 2009 used the behavioural ecological model for development of the counselling intervention. Herbert 2011 used a family-centred assessment and intervention model to empower families to reduce cigarettes smoked in the home. Winickoff 2010 referred to a number of theories as informing the development of their intervention: the transtheoretical stages of change model, together with social learning theory; health beliefs model; cognitive behavioural theory; Wagner's chronic care model and behavioural and systems theory. Tyc 2013 used behavioural contracting, problem solving and social reinforcement.
Acceptability of intervention to participants
Four studies appeared to have involved consultation with potential participants as part of the development of the intervention (Hughes 1991; Davis 1992; Hovell 2000; Borrelli 2010). Davis 1992 employed focus groups with smokers and nonsmokers to understand their beliefs and attitudes towards smoking and cessation in order to develop improved self-help materials. Borrelli 2010 conducted focus groups to better understand Latino culture and modify the motivational interviewing technique accordingly.
Process indicators provide important information regarding the integrity of the way in which interventions were implemented. However, they were well described in only 22 of the 57 studies (Chilmonczyk 1992; Davis 1992; Greenberg 1994; McIntosh 1994; Eriksen 1996; Severson 1997; Hughes 1991; Hovell 2000; Emmons 2001; Hovell 2002; Wakefield 2002; Fossum 2004; Zakarian 2004; Abdullah 2005; Wiggins 2005; Culp 2007; Hannover 2009; Hovell 2009; Borrelli 2010; Winickoff 2010; Stotts 2012; Tyc 2013). More specifically, seven studies reported that they maintained regular monitoring and support with those responsible for providing the intervention (Hughes 1991; Greenberg 1994; Emmons 2001; Culp 2007; Hannover 2009; Hovell 2009; Borrelli 2010), and twelve reported that they evaluated the extent to which participants received, read, undertook or adhered to the intervention as intended (Davis 1992; McIntosh 1994; Severson 1997; Hovell 2002; Wakefield 2002; Zakarian 2004; Abdullah 2005; Wiggins 2005; Culp 2007; Hovell 2009; Winickoff 2010; Stotts 2012). Among those that commented on the monitoring of study implementation, one study (Severson 1997) recommended the need to prompt the providers over the course of the study to ensure appropriate implementation. One study (Fossum 2004) reported the collection of qualitative data on the opinions of the nurses delivering the intervention.
Biological verification of children's exposure
Twenty studies used biological evidence of children's ETS absorption, measuring cotinine in urine or saliva (Woodward 1987; Chilmonczyk 1992; Greenberg 1994; McIntosh 1994; Irvine 1999; Hovell 2000; Wilson 2001; Hovell 2002; Wakefield 2002; Conway 2004; Kimata 2004; Zakarian 2004; Kallio 2006; Ekerbicer 2007; Hovell 2009; Baheiraei 2011; Butz 2011; Halterman 2011; Wilson 2011; Tyc 2013), and ten studies used environmental monitors of children's exposure to ETS (Wahlgren 1997; Hovell 2000; Emmons 2001; Hovell 2002; Zakarian 2004; Hovell 2009; Borrelli 2010; Butz 2011; Stotts 2012; Prokhorov 2013). Five of the ten used passive sampling nicotine monitors as a primary study outcome (Emmons 2001; Borrelli 2010; Butz 2011; Stotts 2012; Prokhorov 2013). Butz 2011 also measured particulate matter in the child's bedroom and living room. The remaining five used air nicotine monitors to either promote or verify the accuracy of parent report of smoking behaviours. Wahlgren 1997 reported using air nicotine monitors in a room where greatest exposure to ETS was reported for two weeks prior to clinic visits to verify parent report of cigarette consumption. Hovell 2000, Hovell 2002, Zakarian 2004 and Hovell 2009 all used inactive air nicotine monitors placed in three rooms where children’s greatest ETS exposure was reported, to promote accurate self report of smoking behaviours by mothers. These studies also placed active air monitors in a selected proportion of the total sample: Hovell 2000 in a randomly selected half of the sample; both Hovell 2002 and Zakarian 2004 in 20% of the sample; and Hovell 2009 in a randomly selected 24% of the sample at six months. Zakarian 2004 reported randomly selecting these homes and placing the monitors in the homes one week before data collection, while Hovell 2002 did not report how the 20% of homes were selected but reported that they were used only for baseline and post-test measures. Cost was given as a reason for not using active air nicotine monitors across the whole sample.
Six interventions used feedback to parents of biological evidence of children's ETS absorption as a stimulus for parental behaviour change (Chilmonczyk 1992; McIntosh 1994; Wilson 2001; Wakefield 2002; Ekerbicer 2007; Wilson 2011). Twenty-one studies used biological validation of parental smoking cessation, measuring cotinine in urine, saliva or serum (Woodward 1987; Irvine 1999; Hovell 2000; Hovell 2002; Fossum 2004; Zakarian 2004; Abdullah 2005; Nuesslein 2006; Kallio 2006; French 2007; Hovell 2009; Winickoff 2010; Phillips 2012; Tyc 2013) and/or expired carbon monoxide (Emmons 2001; Ratner 2001; Curry 2003; Abdullah 2005; Schonberger 2005; Borrelli 2010; Stotts 2012).
Length of follow-up
In this review we determined length of follow-up as being from completion of intervention to time of data collection. Length of follow-up is important to determine, as it affects the extent to which sustainability and long-term outcomes can be assessed. While short-term reductions in children's ETS exposure have some benefit to children's health outcomes, the ultimate goal is for long-term and sustained change in order to maximise the positive impact on children's health and well-being as they grow and develop.
Twelve months or more
Seventeen studies included in this review reported a follow-up of at least 12 months from the end of the intervention (Hughes 1991; Vineis 1993; Elder 1996; Severson 1997; Irvine 1999; Ratner 2001; Hovell 2002; Curry 2003; Conway 2004; Krieger 2005; Schonberger 2005; Wiggins 2005; Chan 2006a; Kallio 2006; Hannover 2009; Hovell 2009; Prokhorov 2013).
Six to twelve months
Shorter follow-up periods of between six and twelve months were reported by a further 18 studies (Davis 1992; Zhang 1993; Greenberg 1994; Wahlgren 1997; Groner 2000; Hovell 2000; Emmons 2001; Wilson 2001; Wakefield 2002; Zakarian 2004; Abdullah 2005; Yilmaz 2006; Culp 2007; Ekerbicer 2007; Ralston 2008; Wilson 2011; Patel 2012; Tyc 2013). Wahlgren 1997 debriefed participants at the six-month follow up, and reported ongoing follow up 8 and 18 months after that.
Less than six months
Long-term effectiveness was particularly difficult to assess in studies with follow-up periods of six months or less. McIntosh 1994 reported follow-up periods that varied between four and six months. Stotts 2012 reported a follow-up period of six months from baseline, but it was unclear what the follow-up was post-intervention. Twenty studies used a follow-up time of less than six months (Woodward 1987; Chilmonczyk 1992; Eriksen 1996; Armstrong 2000; Van't Hof 2000; Pulley 2002; Fossum 2004; Kimata 2004; Chan 2005; Nuesslein 2006; French 2007; Winickoff 2010; Borrelli 2010; Baheiraei 2011; Butz 2011; Halterman 2011; Herbert 2011; Phillips 2012; Chellini 2013; Ralston 2013).
Twenty-eight of the 57 studies mention conducting a power calculation in the design of their studies (Woodward 1987; Greenberg 1994; McIntosh 1994; Severson 1997; Wahlgren 1997; Irvine 1999; Armstrong 2000; Groner 2000; Hovell 2000; Emmons 2001; Wakefield 2002; Conway 2004; Krieger 2005; Schonberger 2005; Wiggins 2005; French 2007; Ralston 2008; Hannover 2009; Hovell 2009; Borrelli 2010; Baheiraei 2011; Butz 2011; Halterman 2011; Wilson 2011; Phillips 2012; Chellini 2013; Prokhorov 2013; Ralston 2013). Of these McIntosh 1994, Wahlgren 1997 and Borrelli 2010 explicitly mention that the statistical power of their study was limited by the small sample size.
Risk of bias in included studies
To meet inclusion criteria for this review, studies had to be controlled trials. Risk of bias assessment has been completed in this update for all of the included studies. This assessment is summarised in Figure 1 and Figure 2. In this review we considered bias arising from detection bias from misclassification by self report (assessed as part of blinding) particularly important, as well as bias arising from inadequate randomisation, although the review considers controlled studies without randomisation.
|Figure 1. Risk of bias summary: review authors' judgements about each risk of bias item for each included study.|
|Figure 2. Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.|
The method of randomisation was rarely described in sufficient detail to permit assessment of whether the allocation was concealed at the time of trial entry. For example, it was common for studies to merely state that participants were randomised. Quasi-randomisation was not uncommon even in large trials. Overall, 34 studies had high or unclear risk of bias from poor randomisation or lack of randomisation. Forty-one studies had a high or unclear risk of bias from allocation concealment with allocation concealment not described in many of the studies.
Blinding (detection bias)
Very few trials had any blinding of participants or providers, largely due to pragmatic issues associated with administering an educational intervention. We have noted in the Characteristics of included studies tables where there was blinding of outcome assessors. Those trials without adequate blinding of outcome assessors or that used a subjective measure of outcome assessment have been classified as “high risk of bias” in this review. Overall, 17 studies had high or unclear risk of bias from blinding of outcome assessment.
Incomplete outcome data
Attrition from withdrawals and exclusions from trials were common and the reasons for these were often not clearly specified. Attrition is potentially a serious risk of bias in these studies. Levels of attrition in each study, and information about any intention to treat analysis have been provided in the Characteristics of included studies. Overall 18 studies had high or unclear risk of bias due to incomplete outcome data.
Other potential sources of bias
Twelve studies were thought to be at high risk of an "other potential source of bias". In six of these studies this related to a difference in the baseline characteristics of groups: two were related to the possibility of contamination between groups; two were related to a lack of intention-to-treat analysis and another two were related to selective reporting.
Effects of interventions
Results are reported by outcome and by setting and child age below. Specific intervention types are discussed within individual outcomes, and are discussed more generally in the Discussion.
Tobacco smoke exposure outcomes
Of the 57 studies, 14 reported success in achieving reduced children's ETS exposure between intervention and control groups, six with biochemical or environmental measures of children's ETS exposure (biological verification of cotinine in urine or saliva of child, or environmental monitors) (Wahlgren 1997; Emmons 2001; Kimata 2004; Borrelli 2010; Baheiraei 2011; Prokhorov 2013) and eight without such measures (Zhang 1993; Armstrong 2000; Curry 2003; Abdullah 2005; Schonberger 2005; Yilmaz 2006; French 2007; Phillips 2012). Of these, eight were judged to be at high risk of bias, two at low risk of bias, and four at unclear risk of bias. A brief summary of outcomes can be found below, with further details of outcome measures in the section Analysis 1.1.
Of the six studies with biochemical or environmental measures of children's ETS exposure, two reported urinary cotinine measures in children, and four recorded household air nicotine with monitors. Three of these six studies used motivational interviewing techniques, one used intensive counselling, one used "fotonovelas" and a comic book, and in one the intervention was described only as parents agreeing to stop smoking. Following an intensive counselling intervention, Wahlgren 1997 reported parental reduction of 1.1 cigarettes per day smoked in the presence of the children for the control group, and 2.2 cigarettes per day for the intervention group; a greater reduction had occurred prior to the intervention. There was no validation by measurement of children’s exposure or absorption via cotinine, or validation of the parental reports, and the clinical significance of such a fall is unclear. The environmental monitors were placed in one room with the "heaviest exposure" and did not find a significant difference between groups. Kimata 2004 achieved a reduction in urinary cotinine levels in children in the cessation group compared to the controls at one month of 285±43 ngmL
Eight studies reported success based on parents' report of smoking cessation, with or without salivary cotinine verification, or reduction in smoking in the presence of children but without verification of children's ETS exposure. These studies employed a range of interventions ranging from school based interventions (children writing letters to their fathers urging them to quit), intensive counselling, a home visiting programme, education and advice, and an intervention based on the Behavioural Action Model (BAM). Zhang 1993 used a school-based intervention and reported proportion of fathers who quit smoking for at least 180 days as 800/9953 (11.7%) for the intervention group and 14/6274 (0.2%) for the control group. At follow-up, Armstrong 2000 reported smoking in house around infant (maternal self report) for the intervention group as 8.6% and control group as 23.8%, where the intervention group received a home visiting programme. Curry 2003 reported smoking abstinence at 12 months as 13.5% in the intervention group, following a brief motivational message and telephone counselling, and 6.9% in the control group. Abdullah 2005 used telephone counselling and reported a biochemically validated quit rate of 47/444 (10.6%) for the intervention group and 21/459 (4.5%) for the control group at six months. Schonberger 2005 reported 52% (14/27) of postnatal mothers quit smoking in the intervention group, compared to 28% (8/30) in the control group at six months follow-up, where the intervention group received home visits. Yilmaz 2006 had two intervention groups which had discussions about the effect of smoking on child or maternal health. Quit rates at follow-up were: child intervention group 24.3%; mother intervention group 13%; control 0.8%. French 2007 used motivational interviewing and at six month follow-up 26 (22%) of the intervention group and 9 (10%) of the control group were saliva cotinine verified non-smokers. Phillips 2012 used motivational interviewing for both groups, and the intervention group also received information about infant bonding. The study reported that at eight weeks post-partum, there were significantly more smoke-free mothers post-partum in the intervention (81%) compared with the control (46%) group.
Forty-two studies failed to detect an intervention effect on ETS outcomes (Woodward 1987; Hughes 1991; Chilmonczyk 1992; Davis 1992; Vineis 1993; Greenberg 1994; McIntosh 1994; Elder 1996; Eriksen 1996; Severson 1997; Irvine 1999; Groner 2000; Hovell 2000; Van't Hof 2000; Ratner 2001; Wilson 2001; Hovell 2002; Pulley 2002; Wakefield 2002; Conway 2004; Fossum 2004; Zakarian 2004; Chan 2005; Krieger 2005; Wiggins 2005; Chan 2006a; Kallio 2006; Nuesslein 2006; Culp 2007; Ekerbicer 2007; Ralston 2008; Hannover 2009; Hovell 2009; Winickoff 2010; Butz 2011; Herbert 2011; Wilson 2011; Stotts 2012; Chellini 2013; Patel 2012; Ralston 2013; Tyc 2013). In two of these studies based on intensive counselling there was significant reduction in self reported parental smoking without a corresponding reduction in children’s urinary cotinine measurements (Hovell 2000; Hovell 2009). Halterman 2011 only measured child health outcomes. In Culp 2007 the intervention group received home visits, and whilst there was no significant reduction in smoking, the other outcome of relevance to our review was mothers' knowledge of the effects of smoking on child development. At 12 months, 2 out of 6 questions were answered better by the intervention group.
In all, 14 of these 42 studies used biochemical measures of children's ETS exposure (child urinary or salivary cotinine levels) while the rest used self reports of smoking behaviour, with or without salivary cotinine verification. Interventions used in these studies were varied, 14 used more intensive counselling approaches, including four that used motivational interviewing. Other interventions included brief advice or counselling (nine studies), feedback of a biological measure of children's ETS exposure (six studies), feedback of maternal cotinine (one study), telephone smoking cessation advice or support (two studies), educational home visits (eight studies), group sessions (one study), an information kit and letter (one study), a booklet and no smoking sign (one study), and school based policy and health promotion (one study). Some studies employed more than one intervention.
Household air quality
Five studies reported household air nicotine as a primary outcome measure (Emmons 2001; Borrelli 2010; Butz 2011; Stotts 2012; Prokhorov 2013). Emmons 2001 used motivational interviewing and telephone counselling, and reported reduced household air nicotine measurements over time in the intervention groups. As there was no change in the number of cigarettes per day smoked, nor in the cessation rate, the implication of the difference was that parents and carers had changed smoking location and had moved outside to smoke. Borrelli 2010 reported a significant decrease in nicotine concentrations as measured by home monitors in the BAM (intervention to increase self-efficacy) but not PAM (motivational interviewing) group at 3 month follow-up. Butz 2011 had three groups which all received asthma education: one group received air cleaners, another group received air cleaners and a health coach, and a control group. They combined results of both intervention groups, who received air cleaners, and compared air quality with the control group. The results suggested that the intervention groups had significantly lower mean particulate matter concentrations compared to the control group. There were no significant differences in air nicotine levels. Stotts 2012 used motivational interviewing, and found no significant differences in environmental nicotine monitors between groups. Prokhorov 2013 reported a significant decrease in nicotine concentrations for the intervention group,which received a comic book and "fotonovelas" for the "high exposure" room but not the "low exposure" room, whilst the decrease in the control group was not significant.
Child health outcomes
Fourteen studies explicitly aimed to improve child health outcomes (Hughes 1991; Greenberg 1994; Armstrong 2000; Wilson 2001; Pulley 2002; Kimata 2004; Krieger 2005; Schonberger 2005; Wiggins 2005; Culp 2007; Borrelli 2010; Butz 2011; Halterman 2011; Wilson 2011) and a fifteenth (Wahlgren 1997) measured child health outcomes although they were not a primary outcome variable (see Analysis 1.1). Of these, in the majority (nine studies), the child health outcome of interest was asthma related (symptom scores, quality of life, functional morbidity, symptom free days, and asthma related health services utilisation). In two studies, the health outcome of interest was respiratory illnesses and another two reported health service utilisation alone, community services in one and hospital admissions and emergency visits in another. One study measured a change in neurotrophin levels though which neurotrophins were measured is not specified.
Nine studies found improvement in child health outcomes. Hughes 1991 embedded an intervention to reduce children’s ETS exposure in a study of a comprehensive asthma education intervention. The outcome was improved asthma control but no change in exposure to ETS. Greenberg 1994 targeted ETS exposure in infants less than six months of age, and aimed to reduce the incidence of lower respiratory tract illness and the prevalence of respiratory symptoms. For infants of smoking mothers it demonstrated a lower prevalence of persistent symptoms in the intervention group (17.8%) compared with control group (30.9%; risk difference 13.1%; 95% CI: 1.0 to 27.0%). There was no difference in the incidence of illness. Wilson 2001 examined the effect of an intervention targeting smoking behaviour change and asthma education on health care utilisation and asthma hospitalisations, and explored other measures of asthma control. It demonstrated a reduction in the prevalence of children making more than one acute care asthma visit in the year following the intervention. Given that there was no apparent benefit of the smoking-related counselling on smoking related outcomes, it is likely that it was the asthma education that achieved the improvement in asthma morbidity, rather than the smoking behaviour programme. Kimata 2004 found that cessation of smoking had no effect on the skin wheal responses or plasma neurotrophins in normal children, but achieved a significant reduction in skin wheal response, responses to house dust mite, cat dander and lower neutrophil levels for those with atopic eczema/dermatitis syndrome. Neurotrophins are a subset of growth factors with a range of functions throughout the body and include nerve growth factor and brain-derived neurotrophic factor (Lackie 1999). This was the only study identified by this review to consider neurotophin levels, and it does not specify which particular neurotrophins were measured. Krieger 2005 delivered a community home intervention to address conditions affecting childhood asthma, and reported that the high-intensity intervention group had a clinically significant improvement in paediatric caregiver asthma quality-of-life score and a decline in urgent health service utilisation, but no significant difference in symptom-free days, compared to the low-intensity intervention group. However, they did not achieve a statistically significant intervention effect for carer report of smoking in the home or report of no smoking allowed in the home, so child health intervention effect is probably due to other aspects of the intervention. Culp 2007 conducted home-visits with the goal of promoting the health and development of first-time mothers and infants, and found that there were no significant differences between groups on number of hospital admissions or emergency room visits. At 12 months, intervention mothers were more likely to make use of health department clinics for well child care as compared to control group (p = 0.04). Borrelli 2010 reported that the child’s level of functional morbidity due to asthma decreased significantly (p<.001) in both the BAM (intervention to increase self-efficacy) and PAM (motivational interviewing) groups over time. Butz 2011 reported that after combining the two group that used air cleaners, children assigned to those groups had a significant increase in symptom-free days during the past 2 weeks; 1.36 compared with 0.24 symptoms-free days for control group children from baseline to follow-up. Halterman 2011 used motivational interviewing to counsel the primary caregiver and an additional smoker who spends the most time with the child, with observed inhaler administration at school by nurse. The study only measured child health outcomes and found a significant improvement in many asthma-related outcome measures in the intervention compared to the control group. Further details can be found in the Analysis 1.1 table.
Five studies did not detect a significant intervention effect on child health outcomes (Wahlgren 1997; Armstrong 2000; Pulley 2002; Wiggins 2005; Wilson 2011). See Analysis 1.1 for more details. Armstrong 2000 used a broader intervention which included education about smoking near infants as a Sudden Infant Death Syndrome (SIDS) prevention strategy in a post-natal nurse home visiting programme aimed to improve the quality of maternal-child attachment, maternal health and child health parameters. At 12 months there was no statistically significant difference between the groups for immunization status or for rates of utilisation of community services. Of the other four studies, two used home visits and two used more intensive counselling methods (one of which included cotinine feedback).
Schonberger 2005 reported associations of exposure to passive smoking with parentally reported asthma symptoms without group allocation. It is therefore not possible to determine an intervention effect on child health outcomes.
Results according to child age
A similar proportion of studies in each age bracket detected intervention effects. Four of the twenty studies which examined measures to reduce ETS exclusively for infants detected an intervention effect (Abdullah 2005; Baheiraei 2011; French 2007; Phillips 2012). Two of the eight studies examining measures to reduce ETS for children up to and including preschool age demonstrated an intervention effect (Emmons 2001; Schonberger 2005). Nine of the 24 studies examining measures to reduce ETS for children up to and including school age and older demonstrated an intervention effect (Zhang 1993; Greenberg 1994; Wahlgren 1997; Kimata 2004; Krieger 2005; Yilmaz 2006; Borrelli 2010; Halterman 2011; Prokhorov 2013).
Results according to setting
In the 'ill child' respiratory setting, four of 13 studies demonstrated an intervention effect (Wahlgren 1997; Krieger 2005; Borrelli 2010; Halterman 2011). Krieger 2005 and Halterman 2011 showed a significant effect on child health outcomes but not on tobacco smoke exposure outcomes. Three of these four studies used intensive counselling or motivational interviewing, whilst one used a community home intervention with elements of education and behaviour change. Of the nine studies that did not demonstrate an intervention effect, three used intensive counselling, one used motivational interviewing, one used a motivational health coach in addition to air cleaners, two used brief counselling methods and two used home visits.
In the 'ill child' non-respiratory setting, two of eleven studies showed an intervention effect (Kimata 2004; Phillips 2012). For Kimata 2004 the intervention was not described, whilst Phillips 2012 used motivational interviewing for both groups, and the intervention group also received information about infant bonding. Of the nine studies that did not demonstrate an intervention effect, three used brief counselling methods, four used more intensive counselling, including one study that used motivational interviewing, one used a booklet and one used cotinine feedback.
In the clinical setting (not designated 'well child' or 'ill child'), one study out of two demonstrated an intervention effect (Curry 2003). This study used a brief motivational message and motivational interview, with follow-up telephone counselling. Nuesslein 2006 did not find an intervention effect, and used parental cotinine feedback.
In the clinical setting (both 'well child' and 'ill child') Yilmaz 2006 demonstrated an intervention effect, with smoking cessation interventions aimed at the child or mother's health. There were no other studies in this group.
In the 'well child' clinical setting, six of the twenty-three studies demonstrated an intervention effect (Armstrong 2000; Emmons 2001; Abdullah 2005; Schonberger 2005; French 2007; Baheiraei 2011). Three of these six studies used motivational interviewing, two used home visiting interventions and one used telephone smoking cessation counselling. Of the 17 studies that did not demonstrate an intervention effect, five used brief counselling methods, five used intensive counselling methods, including one that used motivational interviewing, four used home visits, one used cotinine feedback, one used telephone counselling and one used an information kit and letter.
In the community setting, two of the seven studies showed an intervention effect (Zhang 1993; Prokhorov 2013). Zhang 1993 was one of four studies in a school setting. Prokhorov 2013 used fotonovelas and a comic book for their intervention group. Of the five studies that did not demonstrate an intervention effect, two used telephone counselling, two were school-based, and one used group sessions.
Benefit among participants in comparison groups: A possible 'study effect'
In 32 of the 57 studies, there was reduced children's ETS exposure for study participants regardless of assignment to intervention or control groups (Woodward 1987; Hughes 1991; Davis 1992; Vineis 1993; Elder 1996; Eriksen 1996; Severson 1997; Wahlgren 1997; Irvine 1999; Groner 2000; Ratner 2001; Wilson 2001; Hovell 2002; Wakefield 2002; Curry 2003; Fossum 2004; Abdullah 2005; Chan 2005; Krieger 2005; Chan 2006a; Kallio 2006; Nuesslein 2006; Ekerbicer 2007; Hovell 2009; Winickoff 2010; Halterman 2011; Herbert 2011; Wilson 2011; Chellini 2013; Prokhorov 2013; Ralston 2013; Tyc 2013).
Biological validation of parents' self report
Of the 20 studies with biological evidence of child ETS absorption, 12 allowed an assessment of validation of parent-reported change in exposure versus child ETS absorption (Greenberg 1994; McIntosh 1994; Hovell 2000; Wilson 2001; Hovell 2002; Wakefield 2002; Kimata 2004; Zakarian 2004; Kallio 2006; Hovell 2009; Baheiraei 2011; Tyc 2013). Of these studies, four did not show a discrepancy between reported exposure and an objective measure of absorption (Wilson 2001; Wakefield 2002; Kimata 2004; Kallio 2006). Kallio 2006 reported that parent serum cotinine values showed that parents reported smoking habits accurately but did not provide data. Of the studies using environmental monitors of child exposure to ETS, Wahlgren 1997 and Hovell 2009 allowed an assessment of validation of parent-reported change in exposure versus objective measure. Wahlgren 1997 did not demonstrate a correlation between parental report and environmental monitoring, whilst Hovell 2009 reported a significant moderate correlation. For Hovell 2009 however, the results showed a significant reduction in child secondhand smoke exposure associated with the intervention according to reports, but not according to child urinary cotinine. Tyc 2013 also noted a significant decrease in reported child secondhand smoke exposure but not in child urinary cotinine in the intervention group. Borrelli 2010 noted that, according to monitors in the home, but not on the child, there was a significantly greater reduction in exposure to children in the BAM (intervention to increase self-efficacy) group, although there were a higher quit rates in the PAM (motivational interviewing) group. This was thought to have occurred due to a greater change in the number of cigarettes smoked in front of the child in the BAM group, rather than considering the monitors as a validation measure.
Cost data and cost effectiveness
Twelve of the studies made some reference to costs. However this was generally limited to some statement of implementation costs; McIntosh 1994 mentioned the cost of the manual and Severson 1997 mentioned the staff and intervention cost per person of the intervention. Conway 2004 and Wiggins 2005 also mentioned the costs of implementing the intervention but indicated that further analysis of cost effectiveness was not conducted due the lack of intervention effect. Krieger 2005 reported reduced urgent healthcare costs during the two months before the exit interview for those receiving the intervention relative to the comparison group, but did not provide an extensive cost benefit analysis.
The evidence for success in reducing children's exposure to tobacco smoke is drawn from 14 studies. Seven of these were conducted in or from a clinical setting and employed an intensive counselling-based approach or motivational interviewing (Wahlgren 1997; Emmons 2001; Curry 2003; French 2007; Borrelli 2010; Baheiraei 2011; Phillips 2012). Phillips 2012, however, used motivational interviewing for both groups with additional infant bonding information for the intervention group. While individual studies reported evidence of success for the following types of interventions, further research is needed to confirm their findings: a school-based curriculum-based approach (Zhang 1993); intensive home visiting programme for at-risk mothers that included education about preventive child health (Armstrong 2000); smoking cessation telephone counselling to mothers recruited through 'well child' clinics (Abdullah 2005); the provision of brief educational information to parents of sick children in a clinical setting (Schonberger 2005); education provided by nurses to mothers attending 'well child' visits about the impact of smoking on either their own or their child's health (Yilmaz 2006); and culturally sensitive "fotonovelas" and a comic book (Prokhorov 2013). A further successful study only reported that parents agreed to stop smoking, and does not describe further detail (Kimata 2004). The remaining studies which demonstrated no evidence of intervention effect on reducing children's exposure to tobacco smoke were also conducted in clinical and community settings. Halterman 2011 found a significant improvement in asthma related outcomes in the intervention group, using observed medication administration and motivational interviewing regarding environmental tobacco smoke.
Of the 42 studies that did not find an intervention effect to reduce children's ETS exposure: 14 used more intensive counselling approaches, including 4 that used motivational interviewing. Other interventions included brief advice or counselling (9 studies), feedback of a biological measure of children's ETS exposure (6 studies), feedback of maternal cotinine (1 study), telephone smoking cessation advice or support (2 studies), educational home visits (8 studies), group sessions (1 study), an information kit and letter (1 study), a booklet and no smoking sign (1 study), and school based policy and health promotion (1 study). Some studies employed more than one intervention.
There is no clear evidence of success in reducing children's exposure to tobacco within various clinical settings: respiratory settings (four of the thirteen were successful); non-respiratory 'ill child' (two of eleven); non-peripartum 'well child' (two of ten); and peripartum 'well child' (four of thirteen) settings. In the community setting, two of the seven studies were successful (one of which was a school-based interventions). Three of the eight studies which focused primarily on change in participants' attitudes and behaviours rather than knowledge were among the more successful interventions.
Strategies which are effective in the adult healthcare setting may not be generalisable to the paediatric setting. Brief advice for adult smokers when they attend clinical services for their health has a positive effect in triggering quit attempts (Stead 2008). This effect was not detected in the trials of interventions for parents attending clinical paediatric or child health services. However, this finding might also suggest that either a different sort of brief intervention should be employed or that this context should not be used for brief advice. It is also possible that the studies were underpowered to detect a small effect. Examination of the dynamics of the doctor-child-parent relationship may assist the development of brief strategies with a greater likelihood of success in this clinical setting. Given that there are unknowns about the doctor-child-parent interaction there is potential for interventions in this setting to cause harm. One study found a trend for mothers in the intervention group to smoke more than controls after the intervention (Irvine 1999). Several studies used only one-tailed t-tests to look for statistical significance. Where there is potential to cause harm, even if the hypothesis is unidirectional, two-tailed tests of significance should always be employed. Hovell 2009 undertook a regression analysis to examine the factors associated with the longest participant smoking quit attempts following counselling. The odds for the longest quit attempt were significantly increased when the participants had made a 24 hour quit attempt in the year prior to baseline, had tried a greater number of methods to quit in the past, and had reduced permissiveness of home smoking. Significant associations were not found between longer quit attempts and level of education, heaviness of smoking or the smoking status of the partner.
There is insufficient evidence of the effects on child health indicators of efforts to change child exposure to ETS. Where studies showed a beneficial effect on child health outcomes this could not always be related to an intervention aiming to reduce children’s exposure to ETS (e.g. Culp 2007, Halterman 2011) as there might have been a range of interventions in addition to an intervention to reduce children’s ETS exposure or there may have been a change in measured health outcomes without a corresponding change in ETS exposure outcomes.
There are major differences between those studies which aim to reduce children's exposure to ETS while potentially leaving parental smoking levels unchanged, and those studies which aim to encourage parents to stop or to reduce smoking. A third category may be studies which aim to encourage parents to stop or to reduce smoking, but qualify this with a compromise position of reducing children's exposure to ETS if parents do not cut down or quit. Any interventions which reduce children's exposure are beneficial for the child, although they still expose children to the harm of an increased risk of smoking themselves in adolescence. They also do nothing to improve health outcomes for parents.
There are relatively high rates of smoking cessation in pregnancy, both spontaneously and with clinical interventions (Lumley 2009). With high relapse rates postnatally among women who have quit in pregnancy (Lelong 2001), prevention of relapse for this group is an obvious means of preventing ETS exposure for their children. Of five studies examining an intervention to prevent smoking relapse postpartum, two (French 2007; Phillips 2012) showed a significant beneficial effect for reducing relapse. Ratner 2001 and Van't Hof 2000 identified risk factors for relapse. Risk factors identified by Ratner 2001 were having a partner who smoked and a higher number of sticks smoked per day prior to quitting, whilst prolonging breast feeding and a higher score on a scale measuring "mental health" were protective. Van't Hof 2000 found that a lower level of confidence to maintain cessation, a lower level of family and friends' encouragement to maintain cessation and a higher number of family and friends who smoked were all associated with significantly higher odds of relapse postpartum. Further work in this area will make an important contribution.
Overall, 32 of the 57 studies demonstrated reduced child exposure to ETS for participants, regardless of assignment to intervention or control groups, which suggests that the studies may be describing the natural history of smoking among parents. Parents may reduce their own smoking or their children's exposure over time, possibly as a result of social pressures. Indeed the prevalent social trend in many developed countries over the last decade has been of increasing community concern about exposing nonsmokers to ETS (although arguably more so among nonsmokers than among active smokers).This is especially true for adults in the workplace and public spaces such as bars and restaurants, particularly in North America, Australia and some countries within the EU, where total smoking bans for these settings are increasingly being legislated. Campaigns and community concern about children's exposure to ETS at home and in cars has also increased. It is possible that these studies have recorded parents responding to this social trend by limiting their children's exposure in the home. This being the case, studies need to aim not just for a reduction in children's ETS exposure, but for a greater than background reduction in ETS exposure. In order for a study to produce a significant effect, the impact of interventions must be greater than the comparison groups' rate of decline. It may also be that as most studies used comparison groups rather than control groups (i.e. no cessation or avoidance advice and no information), the comparison interventions may have been more effective than anticipated. As the studies have generally involved comparison groups receiving a limited intervention rather than being strict control groups, this is certainly possible. Moreover, measurement of tobacco smoke exposure outcomes alone may produce an intervention effect and thus be an important component of any intervention.
Limitations of methods employed
Parent reports and reliability
Of the 20 studies which used objective measures of children's ETS exposure or absorption, four showed no discrepancy between parental report of children's exposure and the biological measure. Achieving parental or carer smoking cessation would result in reductions in ETS exposure for the child, in addition to obvious benefits for the ex-smoker. The child harm minimisation approach in this context aims to change adult smoking location or amount, but does not aim for cessation. There is insufficient evidence to comment on whether the parental or carer cessation approach, or the child harm minimisation, is the strategy most likely to lead to reduction of children's ETS exposure. If they were equally effective, adult cessation would be the preferable strategy, because of the benefits to the adult, as well as elimination of the negative role modelling associated with smoking, and would therefore be the preferable strategy.
Small sample sizes
Many of the included studies had small sample sizes, and only half of studies (n = 28) studies reported a power calculation. This results in difficulty establishing whether the intervention did not appear to reduce children's ETS exposure as the sample size was too small. The heterogeneity of study designs and characteristics has rendered quantitative analysis inappropriate in this review.
Length of follow-up
The included studies had varying length of follow-up and we used the longest reported follow-up for the results. Some studies did, however, have short lengths of follow-up with 20 studies reporting a follow-up of less than six months. It is difficult to determine the sustainability and long-term effectiveness of interventions where the study follow-up is short. Indeed, of the studies with longer follow-up, some did show an initial difference between intervention and control group that was not sustained at the final follow-up period.
Implications for practice
Implications for research
Jamie Hartmann-Boyce from the Cochrane Tobacco Addiction Group for assessing risk of bias for previously included studies. Study investigators Susan Blake, Sophia Chan, Ayman El-Mohandes, Michele Kiely, Thurman Allen Merritt, Anne Turner-Henson, and Jonathan Winickoff for providing information about their studies to the review team. Ruchi Baxi and Mohit Sharma wrote the review as part of their role as NHS Specialty Registrars in Public Health.
Funding support for the original review (Roseby 2002) from the Australian National Health & Medical Research Council (Trainee Research Scholarship [RR]), Murdoch Children's Research Institute, VicHealth (Public Health Research Fellowship [EW]) and for the previous update (Priest 2008) from the Cochrane Public Health Group and McCaughey Centre is gratefully acknowledged.
Thank you to authors of the previous versions of the review. Rona Campbell was involved in the development of the original review and the previous update, and extracted data from papers for the previous versions and edited both the original review and the previous update. Grace Ferguson-Thorne extracted data and assisted with editing for the previous review update.
Data and analyses
- Top of page
- Authors' conclusions
- Data and analyses
- What's new
- Contributions of authors
- Declarations of interest
- Sources of support
- Differences between protocol and review
- Index terms
Appendix 1. MEDLINE (Ovid SP) search strategy
Appendix 2. EMBASE (Ovid SP) search strategy
Appendix 3. CINAHL (EbscoHOST) search strategy
Appendix 4. PsycINFO search strategy
Appendix 5. ERIC (Proquest) search strategy
Appendix 6. Cochrane Library (Wiley) search strategy
Last assessed as up-to-date: 20 February 2014.
Protocol first published: Issue 3, 1999
Review first published: Issue 3, 2003
Contributions of authors
RB was involved in coordinating the current review update, extracted data and co-wrote and edited the current review update.
MS was involved in coordinating the current review update, extracted data and co-wrote and edited the current review update.
PW was involved in coordinating the current review update, developing the original review and previous update, extracted data for the original and previous update, and edited the original review and the updates.
RR was involved in coordinating the original review, wrote the original review, and extracted data for the original review and the updates, and edited the updates.
AP was involved in the development, data extraction and editing of the original review and updates.
EW was involved in coordinating the original review and the updates, extracted data for the original review, and edited the original review and the updates.
NP was involved in coordinating the previous update, wrote the previous update, and extracted data for the previous and current updates.
NS was involved in the development, and data extraction for the original review and previous update, and involved in editing of the original review and updates.
Declarations of interest
No conflict of interest known.
Sources of support
- The McCaughey Centre, Melbourne School of Population Health, University of Melbourne, Australia.
- National Health & Medical Research Council, Australia.
- Murdoch Children's Research Institute, Australia.
- VicHealth (Victorian Health Promotion Foundation), Australia.
Differences between protocol and review
Some secondary outcomes removed from methods section in most recent version, as not addressed in recent versions or in current version. These are:
- Knowledge, attitudes and beliefs of carers about the effects of passive smoking or ETS for self or children
- Participants' views of the intervention
- Measures of anxiety, depression, guilt, stress/locus of control, health, and well-being/health-related quality of life
- Measures of family functioning
Medical Subject Headings (MeSH)
*Caregivers; *Family; Age Factors; Controlled Clinical Trials as Topic; Environmental Exposure [prevention & control]; Smoking [*prevention & control]; Smoking Cessation; Tobacco Smoke Pollution [*prevention & control]
MeSH check words
Child; Child, Preschool; Humans; Infant; Infant, Newborn
* Indicates the major publication for the study