The prevalence of smoking
The PrivMort survey asked respondents if they smoked. The available response options were: (1) never smoked, (2) used to smoke but quit, and (3) currently regular smoker. Table 1 shows the prevalence of smoking in Russia and Belarus. More than a half of male respondents in both countries are current smokers. The prevalence of current smoking among women is just under 10% in both countries. The latter is lower rate than those reported in the above-mentioned nationally representative surveys of adult populations (14.4% for Russia and 13.2% for Belarus).
Table 1 Prevalence of smoking among respondents and their parents, percent (PrivMort retrospective cohort study conducted in Russia and Belarus, 2014–2015) Table 1 also shows the prevalence of smoking among the sample of fathers and mothers in Russia and Belarus. Respondents reported their parents’ current smoking behaviour if they were alive at the time of interview, and for deceased parents information on smoking characteristics was collected before their death. Since answer options for parents, as for respondents, include ‘used to smoke but quit’, we are able obtain a good approximation of smoking histories of all parents. By comparing the shares of regular smokers between respondents’ and parental generations, we see that, among men, the prevalence of smoking decreased by about 5 and 2 percentage points, respectively, in Russia and Belarus. Among mothers, the share of those who smoked regularly or smoked and ceased was marginal, at 1.6–2.0 and 1.8–1.3% in considered countries. In both countries, we observe that fathers are less likely to be never smokers than sons, while mothers are more likely to be never smokers than daughters.
Covariates
Table 2 presents the frequencies of covariates in our sample that are used in the multivariate analysis of the intergenerational role of parental influences on smoking. The descriptive statistics suggest that, for most variables, Belarus and Russia are quite similar. We classify parents’ and respondents’ educational attainment in primary, secondary, and tertiary education. Respondents have significantly higher qualifications than their fathers and mothers. In addition, we control for childhood deprivation reported by respondents. The share of individuals stating that they often or constantly went to bed hungry when they were children is about 10% among men and 7–8% among women. We categorise respondents’ age into five groups (42–45, 46–50, 51–55, 56–60, and 61–65) and use this variable to estimate age-adjusted prevalence ratios. Marital status is categorised as single, married, separated/divorced, and widowed. About three-fifths of men are in work, while women are roughly equally distributed between working and not working groups. In both countries, more men than women have experienced long-term unemployment at least once in their lives by looking for work continuously for 6 months. The share of men who have attained supervisory status in their employment is slightly higher than the share of women. Finally, we also account for respondents’ religious denomination.
Table 2 Descriptive statistics of covariates of regular smoking, percent (PrivMort retrospective cohort study conducted in Russia and Belarus, 2014–2015)
Bivariate analysis
In Table 3, we present age-adjusted bivariate prevalence ratios of regular smoking in Russia and Belarus. We observe that father’s regular smoking and smoking cessation are positively associated with their son’s propensity to smoke. Having regularly smoking fathers is also linked to their daughters’ smoking, but the prevalence ratio in Belarus is lower than that observed for sons in this country. On the other hand, mother’s smoking is associated with 2.2 and 2.7 times higher prevalence that their daughters are current smokers in Russia and Belarus. The latter association is not statistically significant among sons. It is also noticeable that the town-level variance in smoking is much higher for women that for men.
Table 3 Age-adjusted bivariate prevalence ratios of regular smoking from multilevel mixed-effect Poisson regressions (PrivMort retrospective cohort study conducted in Russia and Belarus, 2014–2015)
Multivariate analysis
In Table 4, prevalence ratios from multilevel mixed-effect Poisson regressions suggest that smoking patterns of parents are, indeed, associated with individuals’ propensity to smoke after adjustment for all covariates. Significant differences between genders are also apparent. Father’s smoking is a significant predictor of regular smoking among men with prevalence ratios of 1.35 (CI 1.21–1.50) and 1.39 (CI 1.23–1.58), respectively, in Russia and Belarus. The latter associations among daughters are also significant in both countries with prevalence ratios of 1.40 (CI 1.17–1.66) and 1.33 (CI 1.04–1.69). Father’s experience of quitting smoking is positively associated with regular smoking among men but not among women. For the association between mothers and their children, statistically significant prevalence ratios are observed among daughters but not among sons. Having smoking mothers is associated with 1.91 (CI 1.40–2.61) and 2.30 (CI 1.61–3.28) times higher prevalence of being regular smoker among daughters, respectively, in Russia and Belarus. Furthermore, having mothers who quit smoking is associated with 1.83 (CI 1.23–2.72) times higher prevalence of regular smoking in Russia.
Table 4 Age-adjusted multivariate prevalence ratios of regular smoking from multilevel mixed-effect Poisson regressions (PrivMort retrospective cohort study conducted in Russia and Belarus, 2014–2015)
Once parental smoking behaviour is accounted for, parental educational attainment is not significantly linked to offspring’s likelihood of smoking. However, compared to having tertiary educated fathers, women with low educated fathers in Russia are less likely to be regular smokers. Among women in Belarus, we also see that the daughters of low educated mothers in comparison to tertiary educated mothers have 29% lower prevalence of smoking. We do not find that childhood deprivation is systematically and significantly related to smoking. However, Russian men that often went to bed hungry, and Belarusian women who occasionally and constantly went to bed hungry when they were children, are more likely to be regular smokers.
When examined by marital status, single and widowed women and separated/divorced men and women have significantly higher prevalence of smoking than married individuals. Respondents’ own educational attainment is a strong predictor of regular smoking in both countries and across gender. For instance, men with primary education have prevalence ratios of 1.45 (CI 1.27–1.66) and 1.49 (CI 1.36–1.62) of being regular smokers in Russia and Belarus compared with those with tertiary education. Turning to labour market characteristics, being in employment, is associated with lower prevalence of smoking among women in Belarus, while long-term unemployment significantly increases the chances of regular smoking with prevalence ratios of 1.09 (CI 1.00–1.18) and 1.12 (CI 1.01–1.24) for men in Russia and Belarus and prevalence ratios of 1.52 (CI 1.28–1.82) and 1.49 (CI 1.17–1.91) for women in the same countries. Men with supervisory status attainment are also less likely to smoke in Russia. Neither respondent’s religious denomination nor the size of population in towns where the PrivMort retrospective cohort study was conducted is significantly and systematically related to regular smoking.
To show the overall differences in smoking at the population level conditioned by parental smoking patterns, we calculate predicted probabilities. In Fig. 1, we see that accounting for other social background and own socio-demographic and socio-economic variables, as shown in Table 4, men whose father never smoked have a 0.45 (CI 0.41–0.48) probability of being regular smokers in Russia, whereas for men with regularly smoking father, this probability is 14 percentage points higher (0.59 CI 0.56–0.63). This association is even more pronounced in Belarus where having a regularly smoking father is associated with up to 16 percentage points higher probability of being a regular smoker when compared to having a father that never smoked. Figure 2 indicates that women in Russia with never smoking mothers have a probability of 0.08 (CI 0.07–0.10) of being regular smokers against the probability of 0.17 (CI 0.12–0.22) for having regularly smoking mothers. In Belarus, the size of this association is even larger; having smoking mothers is associated with a 12.3 percentage points higher probability of smoking.
Further analysis
In unreported analysis, we have also conducted additional tests. To check if having both parents as smokers carry additional risks for their offspring smoking, we have interacted fathers’ and mothers’ smoking characteristics in multilevel mixed-effects Poisson regressions with the same specifications, as in Table 4. We have not found that there is such an association. Furthermore, through interacting age and parental smoking behaviour, we did not detect any systemic and significant differences in intergenerational transmission of smoking across different age groups in either Russia or in Belarus.