Background Most smokers initiate before the age of eighteen, and smoking in adolescence is strongly influenced by parental smoking habits. Despite this, few studies have examined factors which may protect teenagers against the risk of having a smoking parent. We investigated whether skills-based components of positive mental health in childhood (Mental Health Competence, MHC) modified the association between parent and teenager smoking, using contemporary data from the UK-representative Millennium Cohort Study (∼18 000 children, born 2000–2002; analytic sample: n=10,133).

Methods Cohort members (CMs) reported whether they had ever smoked cigarettes or e-cigarettes at 14 years(y). A dichotomised variable indicated whether one or both parents reported their own tobacco use when CMs were 11 y. A four-class latent measure of MHC captured learning skills and prosocial behaviours at 11 y: High, High-Moderate, Moderate, Low.

We examined effect modification in two ways. First, we compared risk differences (RD; estimated using binary regression) for CM smoking according to parental smoking, across levels of MHC. Second, we estimated RDs for CM smoking according to combinations of parental smoking and MHC. Confounding by socio-economic and demographic characteristics and parent’s mental health was adjusted for. Survey weights accounted for sample design and attrition; multiple imputation addressed item missingness.

Results Similar proportions of CMs had ever smoked cigarettes (17%) and e-cigarettes (18%), although overlap was moderate (40% who had smoked either had smoked both). CMs were more likely to have ever smoked cigarettes if at least one parent smoked (RD: 16%[13–184]) (baseline: no parents who smoke[11%]). This elevated risk was observed across all levels of MHC, but was greatest for Low MHC (RD: 21%[11–31] (RDs in other MHC groups: 7%–12%). When combining parental smoking and MHC (baseline: no smoking parent, high MHC), those with Low MHC and a smoking parent had a RD of 28%(20–36). This was higher than the sum of RDs in CMs with Low MHC but a non-smoking parent (7%[1–14]) and with High MHC but a smoking parent (11% [7–15]). Thus, Low MHC carried an excess risk. There was little evidence of effect modification by Moderate or High-Moderate MHC. Results were similar for e-cigarettes.

Conclusion The association between parent smoking and teenage smoking initiation was considerably stronger in those with Low MHC in contemporary, UK-representative data. These results require replication in other populations, and at older ages (examining progression to regular smoking), but imply that MHC improvement before transition to secondary school holds potential to buffer against an important smoking risk factor.

The views expressed are those of the authors and not necessarily those of the funders.