Sample
Respondents were from the 7th Wave of Understanding Society, a panel survey based on annual interviews conducted within UK households [27]. In fieldwork spanning 2015–2017, 4534 youth aged 10–15 were eligible for self-completion questionnaires. A total of 3635 youth (80.2% of those eligible) in 2759 households completed questionnaires. Valid survey weights designed to account for household attrition, non-response and over-sampling were available for 3291 youth (90.5% of those responding) and used throughout to render the sample representative of the UK [37]. Multiple imputation of missing values (25 datasets, using an unconstrained model in which all analysis variables predicted all others) enabled inclusion of all observed data from respondents with valid weights [38] (proportions missing for most variables were between 0 and 5.3%, though 18.6% for ethnicity; see Table 1).
Table 1
Sociodemographic patterning of youth vaping and smoking
| Observed N (%) | Missing N (%) | Imputed N (%) | Current Vaping | Ever Smoker | Current Smoker | Smoking Initiation (N = 3075a) |
| Yes % | P-Value | Yes % | P-Value | Yes % | P-Value | Yes % | P-Value |
All | 3291 (100.0) | 0 (0.0) | 3291 (100.0) | 3.4 | - | 7.4 | - | 2.3 | - | 0.9 | - |
| | | | | | | | | | | |
No Vaping | 3069 (96.5) | 112 (3.4) | 3179 (96.6) | | | 5.5 | < 0.001 | 1.2 | < 0.001 | 0.5 | < 0.001 |
Current Vaping | 110 (3.5) | | 112 (3.4) | | | 63.3 | | 31.9 | | 24.1 | |
| | | | | | | | | | | |
Never Smoker | 3022 (92.6) | 27 (0.8) | 3047 (92.6) | 1.4 | < 0.001 | | | | | | |
Ever Smoker | 242 (7.4) | | 244 (7.4) | 29.0 | | | | | | | |
| | | | | | | | | | | |
Non Smoker | 3190 (97.7) | 27 (0.8) | 3216 (97.7) | 2.4 | < 0.001 | | | | | | |
Current Smoker | 74 (2.3) | | 75 (2.3) | 47.5 | | | | | | | |
| | | | | | | | | | | |
Never Smokera | 3022 (99.1) | 27 (0.8) | 3046 (99.1) | 1.4 | < 0.001 | | | | | | |
Initiating Smokera | 28 (0.9) | | 29 (0.9) | 45.8 | | | | | | | |
| | | | | | | | | | | |
Male | 1629 (49.5) | 0 (0.0) | 1629 (49.5) | 4.2 | 0.019 | 7.5 | 0.859 | 2.4 | 0.626 | 1.2 | 0.207 |
Female | 1662 (50.5) | | 1662 (50.5) | 2.7 | | 7.3 | | 2.2 | | 0.7 | |
| | | | | | | | | | | |
Age 10 | 520 (15.8) | 0 (0.0) | 520 (15.8) | 0.8 | < 0.001 | 0.0 | < 0.001 | 0.0 | < 0.001 | 0.0 | < 0.001 |
Age 11 | 596 (18.1) | | 596 (18.1) | 0.0 | | 1.8 | | 0.2 | | 0.2 | |
Age 12 | 561 (17.0) | | 561 (17.0) | 0.7 | | 1.8 | | 0.2 | | 0.2 | |
Age 13 | 493 (15.0) | | 493 (15.0) | 2.7 | | 5.7 | | 0.8 | | 0.7 | |
Age 14 | 588 (17.9) | | 588 (17.9) | 6.9 | | 14.6 | | 4.7 | | 3.1 | |
Age 15 | 533 (16.2) | | 533 (16.2) | 9.2 | | 20.6 | | 7.7 | | 1.7 | |
| | | | | | | | | | | |
England | 2830 (86.0) | 1 (0.0) | 2831 (86.0) | 3.4 | 0.618 | 7.4 | 0.892 | 2.1 | 0.311 | 0.8 | 0.080 |
Wales | 111 (3.4) | | 111 (3.4) | 2.7 | | 6.4 | | 1.8 | | 1.0 | |
Scotland | 263 (8.0) | | 263 (8.0) | 3.0 | | 7.6 | | 3.8 | | 2.4 | |
Northern Ireland | 86 (2.6) | | 86 (2.6) | 5.8 | | 9.3 | | 3.5 | | 0.2 | |
| | | | | | | | | | | |
White UK | 2269 (84.7) | 612 (18.6) | 2764 (84.0) | 3.6 | 0.196 | 7.9 | 0.018 | 2.4 | 0.168 | 1.0 | 0.217 |
Ethnic Minority | 410 (15.3) | | 527 (16.0) | 2.4 | | 4.7 | | 1.4 | | 0.4 | |
| | | | | | | | | | | |
Couple Parents | 2437 (75.1) | 45 (1.4) | 2472 (75.1) | 2.6 | < 0.001 | 5.8 | < 0.001 | 1.6 | < 0.001 | 0.7 | 0.011 |
Single Parent | 809 (24.9) | | 819 (24.9) | 5.9 | | 12.2 | | 4.3 | | 1.8 | |
| | | | | | | | | | | |
Parents Never Smokers | 1154 (35.6) | 53 (1.6) | 1174 (35.7) | 3.0 | 0.118 | 5.7 | 0.010 | 1.3 | 0.002 | 0.9 | 0.896 |
Ex-Smoking Parent(s) | 1282 (39.6) | | 1300 (39.5) | 3.0 | | 7.9 | | 2.2 | | 1.0 | |
Current Smoking Parent(s) | 802 (24.8) | | 817 (24.8) | 4.6 | | 9.2 | | 3.8 | | 1.0 | |
| | | | | | | | | | | |
No Parental Vaping | 2857 (88.2) | 53 (1.6) | 2903 (88.2) | 3.1 | 0.014 | 6.8 | < 0.001 | 2.0 | 0.019 | 0.8 | 0.004 |
Parental Vaping | 381 (11.8) | | 388 (11.8) | 5.6 | | 12.3 | | 4.0 | | 2.4 | |
| | | | | | | | | | | |
Parental Education -Degree | 1697 (53.0) | 87 (2.7) | 1741 (52.9) | 3.0 | 0.386 | 6.3 | 0.017 | 1.4 | < 0.001 | 0.6 | 0.137 |
A-Level or equivalent | 658 (20.5) | | 674 (20.5) | 3.5 | | 7.3 | | 2.3 | | 1.3 | |
GSCE or equivalent | 759 (23.7) | | 783 (23.8) | 4.3 | | 9.3 | | 3.9 | | 1.4 | |
No Qualifications | 90 (2.8) | | 93 (2.8) | 3.3 | | 12.7 | | 4.5 | | 0.2 | |
| | | | | | | | | | | |
Managerial/Professional | 1282 (41.1) | 173 (5.3) | 1345 (40.9) | 1.8 | < 0.001 | 6.2 | 0.118 | 1.4 | 0.006 | 0.5 | 0.117 |
Intermediate | 461 (14.8) | | 490 (14.9) | 4.4 | | 7.3 | | 2.5 | | 1.0 | |
Routine | 510 (16.4) | | 536 (16.3) | 5.7 | | 8.2 | | 2.0 | | 1.7 | |
Not employed | 865 (27.7) | | 920 (28.0) | 3.8 | | 8.8 | | 3.7 | | 1.1 | |
| | | | | | | | | | | |
Highest Income Quartile | 463 (14.3) | 48 (1.5) | 472 (14.3) | 1.9 | < 0.001 | 5.9 | < 0.001 | 1.4 | < 0.001 | 0.9 | 0.005 |
2nd Quartile | 836 (25.8) | | 851 (25.9) | 2.1 | | 6.8 | | 2.0 | | 0.3 | |
3rd Quartile | 1079 (33.3) | | 1094 (33.2) | 3.5 | | 6.0 | | 1.4 | | 0.7 | |
Lowest Income Quartile | 865 (26.7) | | 874 (26.6) | 5.5 | | 10.7 | | 4.2 | | 1.9 | |
| | | | | | | | | | | |
aAdolescents who had reported smoking in previous waves of the survey were excluded here (n = 216), so percentages indicate the proportion of those who had never smoked before who initiated smoking in this wave of the survey. |
[Table 1 about here]
Measures
Youth and parents self-reported vaping in response to the question: “Do you ever use electronic cigarettes (e-cigarettes)?” (Yes/No). The present tense “Do you” wording should primarily identify current vaping, though the wording is a little ambiguous and may feasibly have been interpreted by some respondents as “Have you ever used electronic cigarettes?” Our measures of vaping could therefore include both very infrequent and/or ever use in addition to current vaping. Although this was the first time respondents were asked about vaping, smoking was self-reported by youth and parents in this and in earlier waves of the survey. Youth smoking was coded in three binary outcomes: ever, current and initiation (with the latter defined as current smokers who started smoking in the year since the previous survey, i.e. with no indication of smoking from earlier waves). Parental smoking (never, ex, current) and vaping (yes/no) were coded according to the highest level of use from either parent.
Socioeconomic position (SEP) was measured at the household level (taking the more advantaged responses from couple parents) based on highest educational level (degree or higher, A-Level or equivalent, GCSE or equivalent, or no qualifications), occupational status using NS-SEC codes (managerial or professional, intermediate, routine, or not employed), and household income, equivalised for household composition and split into quartiles. For ease of presentation, SEP measures were treated as ordinal when assessing confounder balance. Indicators of gender (male vs female), ethnicity (White UK vs ethnic minority), family structure (couple vs single parents), UK country (England, Wales, Scotland, and Northern Ireland) and interview date to account for temporal trends in smoking and vaping during fieldwork were also included, with youth age (in years) as a continuous variable.
Statistical analyses
We estimated the ATEs and ATTs using a propensity weighting procedure, which is designed to balance measured confounders across the main exposure groups, i.e. youth who did and did not vape, and youth with parents who did and did not vape [36, 39]. This involves first running logistic regression models to predict each exposure, based on measured confounders (identified a priori). Gender, age, ethnicity, family structure, household SEP, UK country and interview date were treated as potential confounders throughout, as was parental smoking (except when stratifying on this variable). For estimating effects of youth vaping, parental vaping was included as an additional confounder.
The predicted probability of each individual’s observed exposure status can then be used to calculate weights designed to help estimate ATEs and ATTs. Table 2 details these calculations. ATE weights re-weight both exposed and unexposed respondents to resemble the total sample (with regards to measured confounders), while ATT weights re-weight the unexposed respondents to resemble the exposed group. Prior to using these weights to estimate effects, validity of the weights was assessed by examining mean differences in confounders associated with the relevant exposure [39]. Weights were deemed valid if confounder differences, expressed in standard deviation units, were reduced close to zero (with differences < 0.2 considered close to 0). Models predicting exposure probability initially used main effects of confounders only, but where imbalance remained the model was revised by adding interactions terms and then re-assessed. Improvements in confounder balance from model revisions were balanced against sufficient overlap of propensity distributions between exposed and unexposed groups by confirming that mean ATE weights were close to 1 [39, 40] (the same is not expected of mean ATT weights). Deviations from this would suggest that some individuals were being assigned extreme weights, indicating risk of making inferences not strongly supported by the available data.
Table 2
Calculation and interpretation of propensity weights
Estimand | How predicted probabilities of exposure are used to calculate weightsa: | Re-weighting of confounding characteristics in exposure groups: | Implied Contrast for Youth Vaping | Implied Contrast for Parental Vaping |
Numerator | Denominator | Unexposed | Exposed | | |
Average Treatment Effect (ATE) | P | P^ | Resemble sample characteristics | Resemble sample characteristics | No youth vape vs. All youth vape | No parents vape vs. All parents vape |
Average Treatment Effect among the Treated (ATT) | 1 if exposed, 1-P^ if unexposed | 1 if exposed, P^ if unexposed | Resemble Exposed group | Unchanged | No youth vape vs. Vaping by youth who did so | No parents vape vs. Vaping by parents who did so |
P = Overall, unadjusted probability of individual’s observed exposure level. |
P^=Predicted probability of individual’s observed exposure level conditional on confounders. |
aWhen investigating effects of parental vaping within strata of parental smoking, both P and P^ were additionally conditional on parental smoking. |
[Table 2 about here]
Exposure effects were estimated in ATE- and ATT-weighted logistic regressions of each outcome on the exposure of interest. For comparison, we also present associations weighted for sample selection only (labelled “sample weighted associations”). Standard errors were adjusted for clustering of youth within households. Z-tests were used to compare differences in effect estimates between strata of parental smoking and between ATEs and ATTs [41].
Analyses of smoking initiation excluded 216 youth who had reported ever smoking in previous survey waves. These prior smokers were older, more likely to be vaping and to have single parents. Since this could introduce selection bias, these differences were reduced by additional weighting back to the total sample for analyses of initiation.
It is important to emphasise that the resultant effect estimates may not necessarily reflect the true effects of interest. For example, while our method aims to balance measured confounders between exposure groups, our effect estimates may still be biased by unmeasured confounders. For this reason, we calculate e-values for each point estimate and for the lower limit of the confidence interval [42]. E-values represent the minimum strength of association (OR in our analysis) that a set of unmeasured confounders would need to have with both the outcome and exposure of interest (independent of measured confounders), in order to explain away the association, or cause its lower confidence interval to include the null (if it already includes the null the e-value for the lower limit will be 1). We also include e-values for the sample-weighted associations, to indicate how much these were reduced by the adjustments made for measured confounders.