Parental Risk Perceptions of Child Exposure to Tobacco Smoke

Laura Rosen; Inessa Kostjukovsky


BMC Public Health. 2015;15(90) 

In This Article



We approached 135 potential participants, of whom 132 (97.8%) agreed to participate.

Most participants (80.3%) were mothers, with the mean age of around 30. Sixty (45.5%) were daily smokers, 6 were almost-daily smokers (4.5%), 11 (8.3%) were occasional smokers, 12 (9.1%) were past smokers, 10 (7.6%) were nonsmokers who had at some point experimented with smoking, and 33 (25%) were never smokers. Comparisons between regular smokers and others on socio-demographic and other variables are shown in Table 2. The following significant differences were found: Fathers, single or divorced parents, and secular or traditional parents were more likely to be regular smokers. Regular smokers had fewer years of education than did others.

Risk Perceptions and Knowledge

Correlations between the three dimensions of risk perception were statistically significant (p < .0001). The correlations were .64 for likelihood and susceptibility, .60 for likelihood and severity, and .75 between susceptibility and severity. Cronbach's alpha for the three components – likelihood, susceptibility, and severity -- was .85.

Table 3 presents descriptive statistics and comparisons for the three components of risk perception and the combined score. Regular smokers were less aware of risk than were others (Combined score: Regular Smokers: Mean:14.03, Standard Deviation (STD):4.67, Others: Mean: 16.65, STD: 3.39. Differences between regular smokers and others were significantly different for all three dimensions of risk (Likelihood p-value:.0440 Susceptibility p-value:.0002 Severity p-value:.0001), and the combined score (p = .0003). Among the three dimensions, the biggest absolute difference, and the most extreme p-value, occurred for the severity dimension.

Descriptive statistics regarding knowledge are also presented in Table 3. The combined knowledge construct had a Cronbach's alpha of .75. Parents who smoked regularly scored lower on knowledge for each individual question (including q23 once it was transposed) and for the combined measure. The differences between regular smokers and others were statistically significant for the questions on child disease and development, but not for the question about child mortality.

The associations between the combined scores for risk perceptions and knowledge were statistically significant for both regular smokers (r = .538, p < .0001) and others (r = .465, p < .0001).

Smoking Behavior in Home and Car

Most (78.8%) regular smokers, and a minority (16.7%) of others, reported that they or family members usually smoked in their homes. Half (50.0%) of regular smokers, and a minority of others (15.2%) reported that they or their family members usually smoked in cars. Most families of regular smokers (57/66 = 86.4%) smoked in the home, car, or both. A minority of families of others (14/66 = 21.2%) smoked in the home, car, or both.

Statistical Models

Full results from statistical models 1, 3, and 4 are presented in Table 4. Model 1 shows the results of the statistical analysis of the effects of smoking status and other variables with risk perceptions: The R-squared for the multiple regression, with the combined risk perception scale as the response variable, was .24. Regular smokers evaluated risk perceptions as lower than did others (LSMeans, Regular smokers: 14.01, Others: 16.04, p = .0158). Ethnicity was statistically significant, with parents of Russian origin showing lower risk perceptions than others (LSMeans Russians: 13.26, Others: 15.54, Israeli: 16.29, p = .0019). None of the other variables reached statistical significance.

Models 2, 3, and 4 concern the relationship between risk perceptions and family smoking in home, while controlling for different sets of variables. In Model 2, just two covariates were included: combined risk perception, and parental smoking status. Parents who smoked regularly, and respondents with lower levels of risk perceptions, were more likely to live in families where smoking took place in the home (Regular smokers vs. Others: OR = 15.63, CI: [6.40,38.15], p < .0001), Risk perceptions: OR = 0.89,CI: [0.79,0.99], p = .0350). In Model 3, regular smokers (OR(regular smoker vs. others) = 45.00, CI: [9.98,202.94], p < .0001), and those who didn't work (OR(No vs. full-time) = 4.45, CI: [1.16,17.02]), were more likely to live in families where smoking occurred in the home than did others. Ethnicity was also associated with smoking in the home (OR (Russia vs. Israel) = 4.03, CI: [0.93,17.37], OR(Other vs. Israel) = 0.41, CI [0.09,1.96], p = .0433). Risk perceptions were not significantly associated with family smoking practices in the home (p = .3224).

Model 4 is similar to Model 3, but does not include the variable for ethnicity. Most of the effects are of similar size and significance levels as in Model 3, with the exception of risk perception. The effect size of risk perception is similar in Models 2, 3, and 4 (Model 2: OR = .89, CI:[0.79,0.99], p = .0350; Model 3: OR = 0.93, CI: [0.81,1.07], p = .3224; Model 4: OR = .88, CI: [0.78,1.003], p = .0560), however, the p-value in Model 2 is statistically significant, in Model 4 is borderline significant, but in Model 3 is not statistically significant.