Parental Risk Perceptions of Child Exposure to Tobacco Smoke

Laura Rosen; Inessa Kostjukovsky


BMC Public Health. 2015;15(90) 

In This Article


Study Design and Recruitment Strategy

The study was a face-to-face survey of parents who visited a large well-baby clinic (Tipat Chalav) in central Israel in order to vaccinate their infants in 2008. All parents of infants up to the age of 14 months were potentially eligible to participate in the survey.

We attempted to recruit an equal number of regular smokers and others, during approximately the same time period. It was important to interview parents with different smoking behaviors concurrently because at the time of the study, a new law for preventing smoking in public places had recently been passed, resulting in broad media coverage and a lively public discourse on the dangers of secondhand smoke exposure.[18] As there were fewer regular smokers than others in the population, simply recruiting on a first-come-first-entered basis could have served to introduce a bias, as those recruited earlier may have had less exposure to the public discussion than those recruited later. We used the following strategy to ensure that regular smokers and others were recruited during the same time period: Whenever a potential participant arrived at the clinic for a regular appointment, the nurse/secretary asked about his/her smoking habits. If the person was a regular smoker, s/he was invited to join the study. Each time a regular smoker was recruited, the next non-regular smoker to arrive at the clinic was recruited for the study. After that non-regular smoker was recruited, a regular smoker was recruited.

Questionnaire Development

We were interested in measuring risk perceptions of parents regarding child exposure to tobacco smoke and knowledge regarding child exposure to tobacco smoke, smoking practices in the home and car, smoking status of respondents, and demographic and socio-economic variables. We built a questionnaire which was influenced by previous work, particularly the work of Brewer,[6] Johansson,[19] Bock,[9] and the Global Youth Tobacco Survey.[20]

Risk Perceptions

We constructed three questions, each corresponding to a dimension of risk perception as defined by Brewer: perceived likelihood, perceived susceptibility, and perceived severity.[6] For perceived likelihood ("the probability that one will be harmed by the hazard") our question was: "In your opinion, is it reasonable that a child exposed to secondhand smoke will get a respiratory tract infection?". For perceived susceptibility ("An individual's constitutional vulnerability to a hazard") the question was: "In your opinion, is a child exposed to SHS more likely to get a respiratory tract infection than other children?" For perceived harm, ("The extent of harm a hazard would cause"), the question asked was: "In your opinion, how much will tobacco smoke in the child's environment affect your child's health?" Severity has also been defined as concerns about clinical or social consequences.[7]

We note that our definition of susceptibility differs from the ones used by Brewer[6] or by Janz.[7] In our framework, susceptibility is by definition due to tobacco smoke exposure, and not due to individual constitutional vulnerability. The difference between the dimensions of likelihood and susceptibility, as defined here, is that likelihood refers to whether someone exposed to tobacco smoke is likely to have an untoward event occur, whereas susceptibility is a comparative question, and asks whether an untoward event is more likely to occur to someone exposed to tobacco smoke than to someone unexposed.

We further note that some previous investigators have chosen to phrase questions in personal terms, while others have chosen to phrase them in general terms. For example, Brewer 2007[6] reported that for assessing severity, Zimmerman et al. used a personal question: "If I had influenza, I would not be able to manage daily activities" while Nichol et al. 1992 used the impersonal question: "Influenza can cause death." In our study, likelihood and susceptibility were asked in general terms, about children exposed to tobacco smoke, while severity was asked in personal terms.

The questions relating to the dimensions of risk perception are presented in Table 1. All questions were asked using a 1–7 Likert scale. Seven point scales have been recommended for risk perception questions.[21] Responses to the three questions were summed to create an overall scale for risk perception.

Knowledge Questions

We also asked about knowledge regarding harm due to TSE. We constructed a scale for knowledge from the five questions about knowledge due to harm to children from tobacco smoke exposure. To create the scale, we reverse-coded Question 23 and then summed the responses.

Smoking Behavior

'Regular smokers' included those who smoked daily or almost daily, while 'others' included occasional smokers (less than almost daily), past smokers, and never smokers.

We chose to classify smoking behavior into these two categories based on results from a pretest of these questions, at which time we saw that the biggest differences in perceptions were between regular smokers and others. Previous researchers have used various categorizations: Drehmer[22] categorized smoking behavior by < =10 vs. >10 daily cigarettes; Chen 2013[10] defined smokers as those who smoked at least 100 cigarettes during their lifetime and at least one in past 30 days; and Wagener 2010[23] defined smokers as those who smoked > =3 cigarettes per day for the past year.

Respondents were asked whether they or family members usually smoked at home (yes/no) or in the car (yes/no).

Additional Variables

Additional potential explanatory variables included: age, gender, years of education, work status (full-time, part-time, or unemployed), country of birth of the parent, family income, marital status (single, married, divorced, widowed), religiosity (Secular, traditional, religious, Ultra-Orthodox), number of children, age of youngest child, and length of lactation (<1 month, 1–3 months, 3–6 months, >6 months).

Validation of Instrument

We ran several pretests of the questionnaire, with participants recruited from a different clinic. The first pretest included 64 participants.[24] Problematic questions were revised. We ran an additional pretest with 10 individuals using a test/retest approach, at an interval of 3–5 weeks. The Pearson correlation coefficients for the test/retest were: Likelihood, r = .55; Susceptibility, r = .53; Severity: r = .75. Full details are reported elsewhere.[24]

Statistical Analyses

We conducted the following analyses:

  1. Smoking status: We compared regular smokers and others on socio-demographic variables. T-tests were used to compare continuous variables and Chi-squared tests were used to compare categorical variables.

  2. Risk perception and knowledge: We calculated Pearson correlation coefficients to study the association between risk perception and knowledge questions and scales, as well as Cronbach's alpha for the risk perceptions and knowledge scales. We compared each of the individual questions and the combined scales by smoking status, using t-tests (as has been done or recommended by other authors when using for a 7-point Likert scale[21,25,26]).

  3. Family smoking in the home and car: We compared smoking behavior in the home and car, and smoking status, using chi-squared statistics.

  4. Statistical models: We examined a. the relationship between PRETS and smoking status, by using multivariate analysis of variance, with PRETS defined as the outcome variable, and smoking status and socio-demographic as explanatory variables; b. the relationship between family smoking in the home and PRETS, using multiple logistic regression, with family smoking in the home defined as the binary outcome variable, and risk perception and parental smoking status as explanatory variables; c. the relationship between family smoking in the home and PRETS, by using multiple logistic regression, with family smoking in the home defined as the binary outcome variable, and risk perception, parental smoking status, and socio-demographic explanatory variables, and d. an exploratory analysis similar to "c" in which we excluded ethnicity from the model.

SAS 9.2 was used for the statistical analysis, and SPSS 21 was used to create the graphs.

Ethical Approval

The study was approved by the Tel Aviv University Ethics Committee, by the Central District Health Bureau, and by the Netanya Regional Health Bureau officials. Written informed consent was fobtained from all participants.