US Smoking Trends Among Young Adults with Serious Psychological Distress: 1997–2018

INTRODUCTION

Smoking is responsible for one in three deaths among US adults with serious psychological distress (SPD).¹ Smoking rates in this population remained unchanged over the first decade of the twenty-first century, contrasting with significant reductions in smoking in the general population.^2,3,4 Since 2010, however, efforts to increase access to and receipt of primary and preventative care for individuals with SPD, enhanced tobacco control measures, and new tobacco products may have affected these trends. Because young adult behavior is particularly consequential for long-term trends in smoking-related health outcomes, we focus on 18–35 year-olds and assess whether high smoking rates among those with SPD persisted post-2010.

METHODS

The 1997–2018 National Health Interview Surveys provide annual, nationally representative data on noninstitutionalized US civilian adults. SPD was indicated by a score of ≥13 on the Kessler-6 scale, which screens for serious mental illness in population surveys.

Restricting consideration to young adults (ages 18–35), we considered three binary outcomes: ever-smoker (i.e., smoked ≥100 cigarettes in their life), current smoker (ever-smoker and currently smokes “every day” or “some days”), and former smoker (ever-smoker who is not a current smoker).

First, we estimated current smoking trends as a piecewise function of interview year for individuals with SPD, via sample-weighted non-linear least squares. Results provide point estimates for the year smoking trends changed as well as their slope before versus after that point. Wald tests assessed whether those trends differed.

We then plotted the annual sample-weighted prevalence for each outcome. As drivers of smoking uptake versus cessation may differ, we assessed initiation and cessation’s distinct contributions to smoking trends among young adults with SPD. Specifically, using 2-year intervals for increased precision, we simulated 2017–2018 smoking rates as if only initiation likelihoods changed and, separately, only cessation changed.

RESULTS

Among the 1997–2018 data’s 5,859 young adults with SPD, 5,841 gave their smoking status, 45.3% of whom reported current smoking (95% confidence interval (CI): 43.6, 47.1). Piecewise regressions estimated that annual trends in current smoking for this subgroup changed after 2011 (95%CI: 2009, 2013), dropping by 3.5 (95%CI: −5.1, −1.9) percentage points per year thereafter relative to pre-2011 trends (see Fig. 1 notes).

Figure 1

Young adults’ smoking trends by serious psychological distress (SPD), 1997–2018. Notes: Data on 18–35-year-old respondents to the 1997–2018 National Health Interview Surveys are used to derive sample-weighted annual means and 95% confidence intervals, adjusting for complex survey design. Trends depict the proportion of young adults who are current smokers by year and SPD. SPD is indicated by a K6 score of ≥ 13. Using the same respondent-level data, non-linear least squares models estimate current smoking among young adults with SPD as a piecewise function of year: ${\mathrm{Y}}_{\mathrm{it}}=\left\{\begin{array}{c}{\upbeta}_0+{\upbeta}_1\cdotp \mathrm{year}\kern1em \mathrm{if}\ \mathrm{year}<\mathrm{C}\\ {}{\upbeta}_0+{\upbeta}_1\cdotp \mathrm{C}+{\upbeta}_2\cdotp \left(\mathrm{year}-\mathrm{C}\right)\ \mathrm{if}\ \mathrm{year}\ge \mathrm{C}\end{array}\right.$. This regression finds a trend break ($\hat{\mathrm{C}}$) at 2011.2 (95%CI: 2009.1, 2013.3). Comparing the annual change in smoking before versus after that point (i.e., ${\hat{\upbeta}}_2-{\hat{\upbeta}}_1$), Wald tests find a statistically significant difference between these slopes (−3.5 percentage points per year; 95%CI: −5.1, −1.9).

For young adults with SPD, ever-smoking (initiation) declined steeply post-2011, from 59.3% in 2011 (95%CI: 51.4, 67.2) to 43.1% in 2018 (95%CI: 35.1, 51.2). Smoking cessation doubled in this subgroup, though on a slightly different timeline: rates rose from 15.9% in 2015 (95%CI: 9.3, 22.6) to 33.5% in 2018 (95%CI: 22.5, 44.5).

While 28.7% (95%CI: 23.3, 34.1) of young adults with SPD smoked in 2017–2018, this prevalence would have been 37.4% if only initiation rates had changed post-2010, versus 37.7% if only cessation rates had changed (see Table 1).

Table 1 Young Adult Smoking Trends and Relative Contributions of Initiation vs. Cessation by SPD

DISCUSSION

Though unchanged for over a decade, smoking rates among young adults with SPD fell markedly from 2011 to 2018. Moreover, this trend was driven by both decreases in smoking initiation and increases in cessation. Both changes are encouraging, as estimates suggest that life expectancy for smokers who quit before age 35 is not statistically different from that of never-smokers.⁵

This study is the first to identify changes in smoking trends among young adults with SPD and estimate the relative contributions of initiation versus cessation. Its primary limitation is reliance on self-reported smoking data in a sample that excludes institutionalized populations, where people with SPD are overrepresented. However, consistent under-reporting alone would not explain the observed change in smoking trends: an increase in under-reporting from 2011 to 2018 would be needed to explain our findings. We could not assess concurrent vaping trends, as NHIS did not ask about electronic cigarette use pre-2014.

Potential drivers of these trends range from changes in tobacco control policies—e.g., the 2009 federal cigarette tax increase—to smokers’ substitution towards e-cigarettes,³ to greater emphasis on treating tobacco use in behavioral health care settings, and increased access to such care under the Affordable Care Act. Implications for the burden of tobacco-related disease will depend on the extent to which reduced smoking reflects substitution towards other tobacco products. While current evidence indicates that electronic cigarettes are less harmful than conventional cigarettes, they are not harmless.⁶ Thus, further work to identify the mechanisms behind this shift in smoking trends is needed, to ensure that such reductions continue and assess their health implications.