Content uploaded by Eric J. Daza
Author content
All content in this area was uploaded by Eric J. Daza on Jun 01, 2018
Content may be subject to copyright.
Likelihood of Unemployed Smokers vs Nonsmokers Attaining
Reemployment in a One-Year Observational Study
Judith J. Prochaska, PhD, MPH, Anne K. Michalek, BA, Catherine Brown-Johnson, PhD,
Eric J. Daza, DrPH, Michael Baiocchi, PhD, Nicole Anzai, BA, Amy Rogers, OTR/L, Mia
Grigg, MS, MFT, and Amy Chieng, BA
Stanford Prevention Research Center, Department of Medicine, Stanford University, Stanford,
California (Prochaska, Michalek, Brown-Johnson, Daza, Baiocchi, Anzai, Chieng); San Francisco
Department of Veteran Affairs, San Francisco, California (Rogers); Buckelew Programs
Residential Support Services, San Rafael, California (Grigg)
Abstract
Importance—Studies in the United States and Europe have found higher smoking prevalence
among unemployed job seekers relative to employed workers. While consistent, the extant
epidemiologic investigations of smoking and work status have been cross-sectional, leaving it
underdetermined whether tobacco use is a cause or effect of unemployment.
Objective—To examine differences in reemployment by smoking status in a 12-month period.
Design, Setting, and Participants—An observational 2-group study was conducted from
September 10, 2013, to August 15, 2015, in employment service settings in the San Francisco Bay
Area (California). Participants were 131 daily smokers and 120 nonsmokers, all of whom were
Corresponding Author: Judith J. Prochaska, PhD, MPH, Stanford Prevention Research Center, Department of Medicine, Stanford
University, Medical School Office Bldg, Room X316, 1265 Welch Rd, Stanford, CA 94305 (jpro@stanford.edu).
Author Contributions: Drs Prochaska and Daza had full access to all the data in the study and take responsibility for the integrity of
the data and the accuracy of the data analysis.
Study concept and design:
Prochaska, Baiocchi, Rogers.
Acquisition, analysis, or interpretation of data:
Prochaska, Michalek, Brown-Johnson, Daza, Baiocchi, Anzai, Grigg, Chieng.
Drafting of the manuscript:
Prochaska, Baiocchi, Anzai.
Critical revision of the manuscript for important intellectual content:
All authors.
Statistical analysis:
Prochaska, Michalek, Daza, Baiocchi.
Obtained funding:
Prochaska, Rogers.
Administrative, technical, or material support:
Prochaska, Michalek, Brown-Johnson, Anzai, Rogers, Chieng.
Study supervision:
Prochaska, Michalek, Rogers.
Conflict of Interest Disclosures: Dr Prochaska reported providing expert witness testimony in litigation against tobacco companies
and consulting with Pfizer on smoking cessation medications. No other conflicts were reported.
Disclaimer: The article's contents are solely the responsibility of the authors and do not necessarily represent the official views of the
State of California Tobacco Related Disease Research Program; the National Heart, Lung, and Blood Institute; or the Agency for
Healthcare Research and Quality.
Additional Contributions: Richard Johnson, Employment Development Department, Workforce Service Branch for San Francisco
and San Mateo Counties, and Tim McClain, Marin Employment Connection, contributed to the study's Community Advisory Board,
provided input into study measures, facilitated a location for recruitment of job seekers, and participated in interpretation and
dissemination of study findings. They were not compensated for their contributions. Jorge Tapia, Employment Development
Department, Workforce Service Branch for San Francisco and San Mateo Counties, supported the study activities, and Racy Ming,
MA, Marin Employment Connection, supported the grant proposal and provided a location for recruitment of job seekers. They were
not compensated for their contributions. In addition, we appreciate the many managers and staff at both sites for their support with the
study.
HHS Public Access
Author manuscript
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
Published in final edited form as:
JAMA Intern Med
. 2016 May 1; 176(5): 662–670. doi:10.1001/jamainternmed.2016.0772.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
unemployed job seekers. Owing to the study's observational design, a propensity score analysis
was conducted using inverse probability weighting with trimmed observations. Including
covariates of time out of work, age, education, race/ethnicity, and perceived health status as
predictors of smoking status.
Mainoutcomes and measures—Reemployment at 12-month follow-up.
Results—Of the 251 study participants, 165 (65.7) were men, with a mean (SD) age of 48 (11)
years; 96 participants were white (38.2%), 90 were black (35.9%), 24 were Hispanic (9.6%), 18
were Asian (7.2%), and 23 were multiracial or other race (9.2%); 78 had a college degree (31.1%),
99 were unstably housed (39.4%), 70 lacked reliable transportation (27.9%), 52 had a criminal
history (20.7%), and 72 had received prior treatment for alcohol or drug use (28.7%). Smokers
consumed a mean (SD) of 13.5 (8.2) cigarettes per day at baseline. At 12-month follow-up (217
participants retained [86.5%]), 60 of 108 nonsmokers (55.6%) were reemployed compared with 29
of 109 smokers (26.6%) (unadjusted risk difference, 0.29; 95% CI, 0.15-0.42). With 6% of
analysis sample observations trimmed, the estimated risk difference indicated that nonsmokers
were 30% (95% CI, 12%-48%) more likely on average to be reemployed at 1 year relative to
smokers. Results of a sensitivity analysis with additional covariates of sex, stable housing, reliable
transportation, criminal history, and prior treatment for alcohol or drug use (25.3% of observations
trimmed) reduced the difference in employment attributed to smoking status to 24% (95% CI,
7%-39%), which was still a significant difference. Among those reemployed at 1 year, the average
hourly wage for smokers was significantly lower (mean [SD], $15.10 [$4.68]) than for
nonsmokers (mean [SD], $20.27 [$10.54]; F(1,86) = 6.50,
P
= .01).
Conclusions and Relevance—To our knowledge, this is the first study to prospectively track
reemployment success by smoking status. Smokers had a lower likelihood of reemployment at 1
year and were paid significantly less than nonsmokers when reemployed. Treatment of tobacco use
in unemployment service settings is worth testing for increasing reemployment success and
financial well-being.
Cross-sectional surveys have shown a consistent association between tobacco smoking and
unemployment. A study of 52 418 construction workers in the 2006-2007 US Current
Population Survey reported a greater likelihood of unemployment among smokers (229
[11.1%]) than nonsmokers (136 [6.4]%).1 In the study's fully adjusted model with sex, age,
education, ethnicity, and household income as covariates, unemployment remained a
significant predictor of current smoking (odds ratio [OR], 1.51; 95% CI, 1.38-1.65). Among
68 501 adults surveyed in the California Health Interview Survey from 2007 to2009,
unemployed job seekers had the highest smoking prevalence (679 [20.9%]) relative to
unemployed individuals who were not seeking a job (2652 [15.9%]) and employed
individuals (7189 [14.8%]); the difference remained significant when adjusting for
demographic factors and other risk behaviors (eg, obesity, binge drinking) (OR, 1.23; 95%
CI, 1.01-1.49).2 Analysis of data from the French National Health Survey from the early
1990s reported a smoking prevalence of 45% among employed menvs 67% for unemployed
men (OR, 2.3; 95% CI, 1.7-3.1).3 Adjusting for demographic and social-psychological
variables, analysis of data from Italy's 2003 Health Determinants Surveillance System with
4002 civilians found that smoking remained a significant correlate of unemployment status
(OR, 2.23; 95% CI, 1.28-3.88).4 Among 7906 jobseekers presenting to employment
Prochaska et al. Page 2
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
agencies in Germany between 2008 and 2009,the smoking prevalence was 57.7% overall (N
= 4328) and exceeded 80% (women, 80.2%; men, 84.8%) for young adults (aged 18-24
years) unemployed more than 24 months.5
While consistent, the extant epidemiologic investigations have been cross-sectional, leaving
it underdetermined as to whether smoking is a cause or effect of unemployment. Tobacco
use among employees is associated with greater health care costs, unproductive time, and
absenteeism.6-8 An employee who smokes costs private employers in the United States an
estimated excess cost (above that for a nonsmoking employee) of $5816 per year.9
Concerned about the health risks and related costs associated with tobacco use, employers
are increasingly taking action to reduce smoking in the workforce.10 Smokers are not a
protected class entitled to special legal protections, based on a 1987 Federal Appeals Court
ruling,11 and hiring policies requiring that employees do not use nicotine are legal in more
than 20 states. Hence, employers can make judgments about tobacco use among prospective
employees. For example, health care and other industries have implemented testing of
applicants' urine for nicotine or cotinine (a nicotine metabolite) as a contingency for
employment. Employers have prohibited tobacco use during working hours, offered
financial incentives for employees to quit smoking, or charged higher medical insurance co-
payments for those who continue to smoke. In many states, employers are able to fire or
discipline employees who smell of tobacco smoke at work.12 Employees who have claimed
nicotine addiction under the Americans with Disabilities Act have not been successful, as
the courts have refused to find that addiction to cigarette smoking is a disability.13
Research has not quantified the economic burden of tobacco use for job seekers. To evaluate
whether tobacco use is indeed a detriment to employability, prospective trials that observe
unemployed smokers and nonsmokers through the job search process are needed. Our study,
using a longitudinal observational design, sought to examine differences in reemployment
success by smoking status during a 12-month period. We hypothesized that nonsmokers
would be more successful than smokers in gaining reemployment. Among those reemployed
at 1 year, we examined hours employed and hourly pay by smoking status. To inform
tobacco treatments for job seekers, we also assessed strategies and motivations for quitting.
Methods
Sample Recruitment
Smokers and nonsmokers were recruited from September 10, 2013,to August 15,2015, from
2 employment development departments in adjacent California counties (1 urban, San
Francisco, and 1 suburban, Marin) serving a combined estimated 7000 clients annually in
the San Francisco Bay Area. Both counties had 100%smoke-free nonhospitality work place
laws that banned smoking of tobacco products in an enclosed space at a place of
employment, with several designated exceptions, and comparable unemployment rates at the
start of recruitment (Marin, 5.0%; San Francisco, 5.2%) and study completion (Marin, 3.5%;
San Francisco, 3.6%). To be eligible, smokers had to report daily smoking with a carbon
monoxide breath sample more than 10 ppm; nonsmokers had to deny tobacco use in the past
year with a carbon monoxide breath sample less than 10 ppm.14 Daily marijuana users were
excluded, as smoked cannabis can elevate carbon monoxide levels. Participants had to be 18
Prochaska et al. Page 3
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
years or older, English literate, unemployed, actively job seeking at the time of study
enrollment, able to provide collateral sources of contact for follow-up, and not actively
planning to relocate out of the area. Recruitment efforts were reactive (via flyers) and
proactive (via onsite outreach).
The Stanford University Institutional Review Board approved all study procedures,
participants provided written informed consent, and confidentiality was assured. The
computer-delivered surveys, administered at the employment service settings, took 60
minutes at baseline and 30 minutes at follow-up. Participants received up to $100 cash for
their time in the study.
Measures
Participants reported their age, sex, race/ethnicity, educational level, marital status, housing
status, transportation, and height and weight to calculate body mass index (calculated as
weight in kilograms divided by height in meters squared). We assessed criminal history
reportable on a job application; prior treatment for alcohol or drug use, including 12-step
programs; and treatment for psychological or emotional problems. A question about general
health had participants rate their health as fair, poor, good, very good, or excellent. The
Kessler 6 scale assessed psychological distress, scored as low (total score, <5), moderate
(5-12), or high (>12).15
The primary outcome of interest was reemployment at 12-month follow-up We defined
reemployment as current hired work at least 10 hours per week or 40 hours per month.
Among those reemployed, we assessed their hourly wage. At baseline, we assessed the
reason for leaving their last position, duration of unemployment, past year gross income, and
career cluster(s) of interest, categorized per O*Net classifications (part of the American
JobCenter Network [http://www.onetonline.org]) (Table 1).
Measures of tobacco use were usual number of cigarettes per day, the Fagerström Test for
Cigarette Dependence,16 stage of change for quitting smoking,17 past 30-day use of other
tobacco and nicotine products, daily cost of smoking, and preference for menthol tobacco
products. We assessed tobacco-related work experiences (eg, perceived discrimination owing
to tobacco use) with a 4-point scale ranging from strongly disagree to strongly agree, and
work-related expectations as a result of tobacco abstinence (eg, increased productivity) with
a 7-point Likert scale ranging from not likely at all to extremely likely (Table 2).19
We created a measure to assess discretionary spending priorities. Smokers ordered items
based on what they were most likely to purchase, assuming finite resources, using their
discretionary funds, defined as money available after one's bills are paid (Table 3). Possible
rank values ranged from 1 to 13. The online survey system randomly ordered the items for
presentation.
Statistical Analysis
The study was powered to detect an absolute difference in reemployment of 20% between
smokers and nonsmokers at 12-month follow-up (eg, 50% vs 70%). With 2-tailed testing, a
0.05 type I error rate, and 80% retention, a sample size of 120 participants per group
Prochaska et al. Page 4
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
provided greater than 80% power for detecting this group difference. Analysis of variance,
χ2, and gamma tests for ordinal associations tested for group differences in baseline
variables. Among those reemployed at 12 months, we performed analysis of variance tests
for group differences in hours worked and hourly wage by smoking status.
As this was an observational study comparing self-selected groups (smokers vs
nonsmokers), we used a propensity score (PS) design20-25 to mitigate confounding and
investigate the main causal hypothesis of interest. The PS design helped to account for
inherent differences between smokers and nonsmokers that could produce biased estimates
of the effect of smoking on successful reemployment by balancing the distribution of
covariates between smokers and nonsmokers. Under certain assumptions, a PS approach
allows statistical inference to be interpreted as causal inference.22
We conducted an inverse probability weight (IPW) analysis with trimmed observations,
where the weight was the inverse (ie, reciprocal) of the PS; the smoking and nonsmoking
groups were weighted so as to be similar (on average) to each other in baseline
characteristics. Propensity score–based IPWs are used to enhance the internal validity of an
analysis, while survey sampling weights are used to support external validity or
generalizability. Once each observation is weighted by its IPW, the weighted average of the
2 groups are differenced, which estimates the risk difference (RD). In our analysis, the IPW-
adjusted estimand was the causal effect of smoking status on unemployment status.22
Before using IPWs to weight the observations, one must ensure that the 2 comparison
groups do not have members that are completely dissimilar from the other group (ie, the
groups share a common support). Design-based PS analysis starts with careful consideration
of which observational units should be included in the study—using only preexposure
covariates and specifically excluding any information about the outcome information. Thus,
the design-based portion of a PS analysis is distinct from fitting the outcome model, which
is not the immediate goal of the PS approach. One important feature of a PS design is that it
identifies the set of observational units with overlapping PS values; positivity (ie, the
probability of not smoking is strictly between 0 and 1) is a key assumption of the PS
approach. Positivity enforces that (in one particular sense) the exposed group and the
unexposed groups are not distinct in terms of their baseline covariates, thus avoiding
complete confounding by baseline covariates.
To fulfill these criteria, we trimmed observations; that is, we identified and dropped
observations with extreme PS values estimated using variables other than the outcome,
following Crump et al.26 It would not be totally incorrect to compare trimming in
observational studies with the inclusion and exclusion criteria in randomized clinical trials.
One guideline for trimming is that observations with PS values outside of the interval [0.1,
0.9] should be dropped.26 Using both sets of PS values (ie, from nonsmokers and smokers),
our trimming points were instead defined as the highest minimum and lowest maximum PS
values (ie, 0.038 and 0.889 for the 5-covariate model; 0.124 and 0.903 for the 10-covariate
model); hence, overlap was empirically determined using our real data, which closely aligns
with the guideline for trimming. The full trimming-based procedure involves first estimating
the PS model covariates on the full data set, trimming observations, and then re-estimating
Prochaska et al. Page 5
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
these covariates using the trimmed data set; the IPWs for the main analysis are then
constructed using these final PS estimates.
Analyses were done in R statistical software (R Foundation for Statistical Computing) to
estimate the PS through logistic regression using the glm package, and subsequently the
IPW-adjusted RD for unemployment at 12 months for non-smokers with respect to smokers.
The boot package was used to estimate the SE of the IPW-adjusted RD estimator. The
primary PS was specified using a logit link function and a linear model that included a set of
5 covariates deemed most relevant to reemployment (ie, time out of work, age, education,
race/ethnicity, perceived health status). We also conducted a basic sensitivity analysis
comparing the primary PS results with those of a PS model that doubled the number of
covariates, adding sex, housing stability, reliable transportation, criminal history, and prior
treatment for alcohol or drug use. Including more covariates in a correct PS model with
unknown predictors is generally expected to increase the accuracy of IPW estimators while
decreasing precision. A related tradeoff is that including more covariates in the PS model
tends to lead to less overlap between smokers and nonsmokers; the PS analyst must trim
more, therefore reducing precision. The difference in the proportion of data trimmed
between the primary and sensitivity analyses was 4-fold. Last, we used analysis of variance
to test for group differences in hours worked and hourly wage by smoking status among
those reemployed at 12 months.
Results
Baseline Characteristics
The full sample, 131 smokers and 120 nonsmokers, was 65.7% male (N = 165), with 96
white participants (38.2%), 90 black (35.9%), 24 Hispanic (9.6%), 18 Asian (7.2%), and 23
multiracial or other race (9.2%), with a mean (SD) age of 48 (11) years and mean (SD) body
mass index (calculated as weight in kilograms divided by height in meters squared) of 26.7
(5.9); 129 were never married (51.4%), 99 were unstably housed (39.4%), 72 had received
prior treatment for alcohol or drugs (28.7%) and 106 for psychological or emotional
problems (42.2%), 52 had a criminal history (20.7%), 70 lacked reliable transportation
(27.9%), 92 had a high school degree or less (36.7%), 81 had completed some college
(32.3%), and 78had a college degree (31.1%).A majority (142 [56.6%]) of the participants
were unemployed for more than 6 months; 62.0% (n = 157) reported a gross income in the
past year of less than $20 000. The most common reason for leaving their last employer was
that their contract ended or they were laid off (145 [57.8%]), followed by being fired (38
[15.1%]).
Table 1 compares baseline characteristics of smokers and nonsmokers. Compared with
nonsmokers, smokers were significantly younger and less educated; more likely to be men,
African American or multiracial, never married, unstably housed, and an urban resident; had
a criminal history, unreliable transportation, and prior treatment for alcohol or drug
problems; and reported poorer health. Mental health measures and body mass index did not
differ by smoking status. Smokers were more chronically unemployed and reported lower
past year income than did nonsmokers. Smokers were less likely than nonsmokers to be
seeking employment in business; education and training; health science; marketing, sales,
Prochaska et al. Page 6
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
and service; and science, technology, engineering, or math. The reason for leaving their last
employer did not differ by smoking status.
Smoking Characteristics
At baseline, smokers consumed a mean (SD) of 13.5 (8.2) cigarettes per day for a mean
(SD) of 28 (12) years; 63 (48.1%) smoked only menthol cigarettes, 56 (42.7%) smoked only
nonmenthol cigarettes, and 12 (9.2%) smoked both kinds. A majority (75 [57.3%]) of
smokers reported relighting their extinguished cigarettes, with 33.6% (n = 44) doing so
daily. Fagerström scores were a mean (SD) of 5.6 (1.5), indicating a moderate level of
dependence; 99 of smoking participants (75.6%) smoked within 30 minutes of waking.
Nearly half (62 [47.3%]) of those who smoked cigarettes used additional tobacco products in
the past 30 days, including cigars (30 [22.9%]), e-cigarettes (16 [12.2%]), cigarillos (14
[10.7%]), blunts (10 [7.6%]), pipes (9 [6.9%]), chewing tobacco (5 [3.8%]), hookah (3
[2.3%]), and snus (3 [2.3%]), a form of smokeless tobacco. Few used nicotine replacement
in the past 30 days, including patch (10 [7.6]%), gum (10 [7.6%]), lozenge (2 [1.5%]), and
nasal spray (2 [1.5%]). Participants spent a mean (SD) of $6.49 ($4.35) per day (median,
$5.00) on tobacco. Typical purchasing was by the pack (112 [85.5%]), with 21 [16.0%]
purchasing cigarettes individually (ie, “loosies”), and 8 [6.1%] rolling their own cigarettes.
At baseline, 61 smokers (46.6%) were in the precontemplation stage of change for quitting
smoking, 44 (33.6%) in the contemplation stage, and 26 (19.8%) in the preparation stage.
Table 2 summarizes the prior attempts to quit smoking and abstinence expectancies of the
sample.18 Nearly all (119 [90.8%]) had made a 24-hour attempt to quit smoking in their
lifetime and 85 [64.9%] had done so in the past year. Most (93 [71.0%]) were advised to quit
smoking by a health care professional, yet few used evidence-based cessation approaches.
Nearly half (59 [45.0%]) reported that an employer offered them a cigarette or encouraged
them to smoke, 46 (35.1%) were criticized by an employer for smoking, and 11 (8.4%) were
fired owing to tobacco use. Few smokers reported support for quitting smoking from an
employer (9 [6.9%]) or career counselor (10 [7.6%]).
Table 3 summarizes ratings for smokers' discretionary spending priorities. Tobacco was
ranked at the top (lowest score) ahead of basic needs and job-seeking necessities, such as
transportation funds and cellular telephone costs. Nicotine replacement had the lowest mean
rank. Heavier smoking was associated with a higher prioritization (lower score) of tobacco
(Pearson r = –0.19;
P
= .04). Among the top-ranked items, 20 respondents (15.3%) selected
tobacco as their first priority; 19 (14.5%) selected nutritious food, 14 (10.7%) transportation,
and 13 (9.9%) cellular telephone costs.
Reemployment
A total of 217 participants (86.5%) completed 12-month surveys; 89 reported being
reemployed (41.0%). An additional 6 participants (3 nonsmokers, 3 smokers) reported
working less than 10 hours per week. Study retention was comparable for smokers
(109[83.2%]) and nonsmokers(108 [90.0%]) ( , 2.47;
P
= .12). Among those
completing the 12-month survey, 60 nonsmokers (55.6%) and 29 smokers (26.6%) were
reemployed at 1 year. The un adjusted RD in reemployment is 0.29 (95% CI, 0.15-0.42).
Prochaska et al. Page 7
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
In our primary PS analysis, the PS was specified using a logit link function and a linear
model that included the covariates of time out of work, age, education, race/ethnicity, and
perceived health status. Observations with PS less than 0.047 or greater than 0.903 were
trimmed from the original sample of 109 smokers and 108 nonsmokers. The trimmed sample
contained 107 smokers and 102 nonsmokers (3.7% excluded). The RD of reemployment for
nonsmokers vs smokers was 0.30 (SE = 0.09) (95% CI, 0.12-0.48). That is, after controlling
for the 5 covariates of greatest concern and trimming to ensure common support, we
estimate that if all participants in the study changed from being smokers to being
nonsmokers then there would be a 30% increase in reemployment.
In the sensitivity analysis, the model with 10 covariates was fit on a trimmed sample with 82
smokers and 80 nonsmokers (25.3% excluded), yielding an RD estimate of 0.24 (SE = 0.08)
(95% CI, 0.07-0.39). Qualitatively, this sensitivity analysis agreed with our primary analysis,
and the 95% CIs also overlapped, indicating no significant difference in the estimates. A
reduced effect size was found using the 10-covariate model, so additional confounders may
have contributed to the observed difference in reemployment between smokers and
nonsmokers.
Hours and Wages
Participants who were reemployed at 1 year worked a mean (SD) of 32 (22) hours per week,
with no difference by smoking status (F(1,82) = 1.19;
P
= .28). Among those reemployed at 1
year, the hourly wage for smokers was significantly lower (mean [SD], $15.10 [$4.68]) than
for nonsmokers (mean [SD], $20.27 [$10.54]; F(1,86) = 6.50;
P
= .01).
Stabilityof Smoking
Smoking status was generally stable over time: 6 smokers (5.7%) at baseline had quit at 12
months, while 8 baseline non-smokers (7.4%) were smoking (5 of 8 were former smokers at
baseline). Among continued smokers, mean (SD) cigarettes per day declined significantly
over time, from 12.6 (6.3) at baseline to 10.2 (7.4) at 12 months (paired samples tdf = 97 =
3.65;
P
< .001). The reduction in cigarettes per day did not differ by reemployment status
(F1,98 = 0.05;
P
= .83).
Discussion
Although tobacco use has been associated with unemployment in cross-sectional population-
based studies in the United States and Europe, the mechanism and direction of that
association has not been investigated prospectively. Our study examined the association of
smoking with reemployment during a 12-month time frame. In our primary and sensitivity
PS analyses adjusting for covariates of interest, nonsmokers had a significant advantage in
reemployment at 12 months relative to smokers.
Had we randomized participants into groups of smokers or nonsmokers, then we would
conclude that not smoking increased the probability of reemployment at 12 months by 12%
to 48% on average. Given that nicotine is addictive and tobacco use is harmful, it would be
unethical to randomize a participant to smoke vs not smoke. Instead we prospectively
tracked the reemployment success of smokers and nonsmokers in the job-seeking market.
Prochaska et al. Page 8
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
That the groups were preexisting raises concern about residual confounding. We therefore
conclude that there is suggestive evidence that a causal association may exist between
smoking status and reemployment at 12 months.
Among smokers reemployed at 1 year, on average, their hourly income was $5 less relative
to reemployed nonsmokers: $15.10 vs $20.27, a 25.5% difference. Averaging 32 hours per
week, this is a deficit exceeding $8300 annually. Our findings, which were self-reported
among job seekers, are comparable with wage estimates for nonsmokers ($20.71) and
smokers ($17.48) reported by Berman et al9, based on the US Bureau of Labor Statistics,
adjusted to 2010 levels, and discounted at 15.6% for smokers per a report of the Medical
Expenditure Panel Survey.27 Combining our estimated wage gap with the sample's report of
spending about $6.50 per day on tobacco (more than $2300 per year), the findings suggest a
cost, on average, of more than $10 600 annually associated with tobacco use. With nearly 3
decades of smoking and evidence of very low rates of quitting during this 12-month
observational study, the financial losses to smokers are significant.
An economically disadvantaged group, with most earning less than $20 000 gross in the past
year (and residing in the San Francisco Bay Area), participants reported relighting
extinguished cigarettes; a preference for menthol tobacco, which is often priced more
cheaply; purchasing single cigarettes, which are illegal; and smoking cigars and cigarillos,
which are taxed at a lower rate and are available for individual sale. Regulatory efforts to
ban menthol, increase taxes, and enforce bans on individual sales of all forms of tobacco
may help promote cessation among job-seeking smokers. Notably, smokers in our sample
prioritized tobacco as more important than items relevant to job seeking, such as
transportation costs, cellular telephone service, and grooming needs.
Study limitations included the exclusion criteria and sample size, which, while powered for
the main outcome, did not allow for tests of association within career clusters. Participants
were English-literate, not intending to relocate in the next 12 months, and residing in the San
Francisco Bay Area, a geographical location with a very low smoking prevalence and
probably unusually high stigma about smoking. As such, study findings may not be
generalizeable to all job-seeking smokers in all regions. Although limited by its
observational design, our study yielded novel findings.
Conclusions
Employment development departments are well placed for reaching tobacco users and
addressing tobacco-related health and economic disparities. Our research team is now testing
the effect of a tobacco cessation intervention on time to reemployment in a randomized
controlled trial with job-seeking smokers. As a “one-stopshop” for employment resources,
employment service agencies could raise awareness of tobacco-related costs, wage losses,
health harms, and associations with lower reemployment success and serve as a connector to
low-cost cessation services such as state quit-lines.
Prochaska et al. Page 9
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Acknowledgments
Funding/Support: This study was supported by two grants from the State of California Tobacco-Related Disease
Research Program: a Pilot Community-Academic Research Award 21BT-0018 and a Research Award 24RT-0035.
Postdoctoral training grant T32 HL007034 from the National Heart, Lung, and Blood Institute supported Drs
Brown-Johnson and Daza. Grant KHS022192A from the Agency for Healthcare Research and Quality supported Dr
Baiocchi.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection,
management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and
decision to submit the manuscript for publication.
References
1. Okechukwu C, Bacic J, Cheng KW, Catalano R. Smoking among construction workers: the
nonlinear influence of the economy, cigarette prices, and antismoking sentiment. Soc Sci Med.
2012; 75(8):1379–1386. [PubMed: 22795358]
2. Prochaska JJ, Shi Y, Rogers A. Tobacco use among the job-seeking unemployed in California. Prev
Med. 2013; 56(5):329–332. [PubMed: 23415765]
3. Khlat M, Sermet C, Le Pape A. Increased prevalence of depression, smoking, heavy drinking and
use of psycho-active drugs among unemployed men in France. Eur J Epidemiol. 2004; 19(5):445–
451. [PubMed: 15233317]
4. De Vogli R, Santinello M. Unemployment and smoking: does psychosocial stress matter? Tob
Control. 2005; 14(6):389–395. [PubMed: 16319362]
5. Freyer-Adam J, Gaertner B, Tobschall S, John U. Health risk factors and self-rated health among
job-seekers. BMC Public Health. 2011; 11:659. [PubMed: 21854611]
6. Max W. The financial impact of smoking on health-related costs: a review of the literature. Am J
Health Promot. 2001; 15(5):321–331. [PubMed: 11502013]
7. Bunn WB III, Stave GM, Downs KE, Alvir JM, Dirani R. Effect of smoking status on productivity
loss. J Occup Environ Med. 2006; 48(10):1099–1108. [PubMed: 17033509]
8. Goetzel RZ, Carls GS, Wang S, et al. The relationship between modifiable health risk factors and
medical expenditures, absenteeism, short-term disability, and presenteeism among employees at
Novartis. J Occup Environ Med. 2009; 51(4):487–499. [PubMed: 19337132]
9. Berman M, Crane R, Seiber E, Munur M. Estimating the cost of a smoking employee. Tob Control.
2014; 23(5):428–433. [PubMed: 23733918]
10. Asch DA, Muller RW, Volpp KG. Conflicts and compromises in not hiring smokers. N Engl J Med.
2013; 368(15):1371–1373. [PubMed: 23534547]
11. Palmer v Liggett & Meyers Co, 825 F2d 620 (1st Cir 1987).
12. Jones JW, Novick WM, Sade RM. Should a medical center deny employment to a physician
because he smokes tobacco products? Ann Thorac Surg. 2014; 98(3):799–805. [PubMed:
25193184]
13. Brashear v Simms, ea 138 F Supp 2d 693 (SDCD Md 2001).
14. Brose LS, Tombor I, Shahab L, West R. The effect of reducing the threshold for carbon monoxide
validation of smoking abstinence— evidence from the English Stop Smoking Services. Addict
Behav. 2013; 38(10):2529–2531. [PubMed: 23773961]
15. Prochaska JJ, Sung HY, Max W, Shi Y, Ong M. Validity study of the K6 scale as a measure of
moderate mental distress based on mental health treatment need and utilization. Int J Methods
Psychiatr Res. 2012; 21(2):88–97. [PubMed: 22351472]
16. Fagerström K. Determinants of tobacco use and renaming the FTND to the Fagerström Test for
Cigarette Dependence. Nicotine Tob Res. 2012; 14(1):75–78. [PubMed: 22025545]
17. Prochaska JO, DiClemente CC, Norcross JC. In search of how people change: applications to
addictive behaviors. Am Psychol. 1992; 47(9):1102–1114. [PubMed: 1329589]
18. Fiore, MC.; Jaén, CR.; Baker, TB., et al. Treating Tobacco Use and Dependence: 2008 Update.
Rockville, MD: U.S. Department of Health and Human Services. Public Health Service; 2008.
Prochaska et al. Page 10
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
19. Hendricks PS, Wood SB, Baker MR, Delucchi KL, Hall SM. The Smoking Abstinence
Questionnaire: measurement of smokers' abstinence-related expectancies. Addiction. 2011; 106(4):
716–728. [PubMed: 21205053]
20. D'Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to
a non-randomized control group. Stat Med. 1998; 17(19):2265–2281. [PubMed: 9802183]
21. Morgan, SL.; Winship, C. Counterfactuals and Causal Inference: Methods and Principles for Social
Research. 2nd. New York, NY: Cambridge University Press; 2015.
22. Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for
causal effects. Biometrika. 1983; 70(1):41–55. DOI: 10.1093/biomet/70.1.41
23. Rosenbaum, PR. Observational Studies. 2nd. New York, NY: Springer; 2002.
24. Rosenbaum, PR. Design of Observational Studies: Springer Series in Statistics. New York, NY:
Springer; 2010.
25. Rubin DB. Causal inference using potential outcomes: design, modeling, decisions. J Am Stat
Assoc. 2005; 100(469):322–331. DOI: 10.1198/016214504000001880
26. Crump RK, Hotz VJ, Imbens GW, Mitnik OA. Dealing with limited overlap in estimation of
average treatment effects. Biometrika. 2009; 96(1):187–199. DOI: 10.1093/biomet/asn055
27. Cowan B, Schwab B. The incidence of the healthcare costs of smoking. J Health Econ. 2011;
30(5):1094–1102. [PubMed: 21820747]
Prochaska et al. Page 11
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Key Points
Question
Does reemployment success differ by smoking status?
Findings
In this 2-group, 12-month prospective study with 251 unemployed job seekers,
nonsmokers were 30% more likely on average to be reemployed at 1 year relative to
smokers. Among those reemployed at 1year, the average hourly wage was $5 higher for
nonsmokers than smokers.
Meaning
Given the disparities in reemployment by smoking status, treatment of tobacco use in
unemployment service settings is worth testing for increasing reemployment success and
financial well-being.
Prochaska et al. Page 12
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Prochaska et al. Page 13
Table 1
Characteristics of Unemployed Job Seekers at Baseline
Characteristic
Valuea
P Value
Nonsmoker (n = 120) Current Smoker (n = 131)
Age, mean (SD), y 49.3 (11.9) 46.2 (10.8) .03
Male sex 63 (52.5) 102 (77.9) <.001
Ethnicity
Non-Hispanic white 56 (46.7) 40 (30.6)
<.001
African American 23 (19.2) 67 (51.1)
Hispanic 17 (14.2) 7 (5.3)
Asian or Pacific Islander 16 (13.3) 2 (1.5)
Multiracial or other 8 (6.7) 15 (11.5)
County
Suburban 62 (51.7) 32 (24.4) <.001
Urban 58 (48.3) 99 (75.6)
Marital status
Never married or single 56 (46.7) 73 (55.8)
.01 Married or cohabitating 25 (20.8) 10 (7.6)
Divorced, separated, or widowed 39 (32.5) 48 (36.6)
Education, mean (SD), y 14.6 (2.7) 12.9 (2.6) <.001
High school degree or less 27 (22.5) 65 (49.6)
<.001 Some college 36 (30.0) 45 (34.4)
Completed college degree 57 (47.5) 21 (16.0)
Housing
Own, rent, or live with family 93 (77.5) 59 (45.0) <.001
Transitional or unhoused
b
27 (22.5) 72 (55.0)
Lack of reliable transportation 22 (18.3) 48 (36.6) .001
Criminal history 17 (14.2) 35 (26.7) .01
Prior treatment for drug or alcohol use 18 (15.0) 54 (43.5) .001
Prior mental health treatment 47 (39.2) 59 (46.1) .27
Kessler 6 scale
None or mild psychological distress 52 (43.3) 50 (38.2)
.28 Moderate psychological distress 56 (46.7) 59 (45.0)
Severe psychological distress 12 (10.0) 22 (16.8)
Perceived health status
Poor or fair 15 (12.5) 37 (28.2)
<.001 Good 38 (31.7) 54 (41.2)
Very good or excellent 67 (55.8) 40 (30.5)
BMI, mean (SD) 27.3 (6.7) 26.1 (5.0) .11
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Prochaska et al. Page 14
Characteristic
Valuea
P Value
Nonsmoker (n = 120) Current Smoker (n = 131)
Chronicity of unemployment, mo
0-3 51 (42.5) 21 (16.0)
<.001
>3-6 14 (11.7) 23 (17.6)
>6-12 24 (20.0) 39 (29.8)
>12 31 (25.8) 48 (36.6)
Past year gross income, $
<10 000 47 (39.2) 65 (49.6)
.03
10 000-20 000 20 (16.7) 25 (19.1)
21 000-40 000 22 (18.3) 26 (19.8)
>41 000 31 (25.8) 15 (11.5)
Career clusters
c
Agriculture, food, and natural resources 8 (6.7) 16 (12.2) .14
Architecture and construction 16 (13.3) 17 (13.0) .93
Arts, audio and video technology, and communications 17 (14.2) 10 (7.6) .10
Business, management, and administration 31 (25.8) 14 (10.7) .002
Education and training 17 (14.2) 7 (5.3) .02
Finance 11 (9.2) 5 (3.8) .08
Government and public administration 15 (12.5) 9 (6.9) .13
Health science 19 (15.8) 9 (6.9) .02
Hospitality and tourism 15 (12.5) 28 (19.1) .16
Human services 20 (16.7) 20 (15.3) .76
Information technology 10 (8.3) 9 (6.9) .66
Law, public safety, corrections, security 12 (10.0) 13 (9.9) .98
Manufacturing 7 (5.8) 6 (4.6) .65
Marketing, sales, and service 33 (27.5) 15 (11.5) .001
Science, technology, engineering, math 9 (7.5) 1 (0.8) .006
Transportation, distribution and logistics 16 (13.3) 13 (9.9) .40
Other (eg, “would take any job”) 4 (3.3) 6 (4.6) .61
Reason last employment ended
Laid off or contract work ended 72 (60.0) 73 (55.7)
.36
Fired 14 (11.7) 24 (18.3)
Quit 7 (5.8) 5 (3.8)
Relocated 10 (8.3) 6 (4.6)
Other (eg, medical, pregnant, legal) 17 (14.2) 23 (17.6)
Abbreviation: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared).
a
Data are presented as number (percentage) of participants unless otherwise indicated.
b
Includes homeless, single residency occupancy, halfway house, or therapeutic community.
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Prochaska et al. Page 15
c
Career clusters based on O *Net classifications, part of the American JobCenter Network.
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Prochaska et al. Page 16
Table 2
Past Attempts to Quit, Encouragement to Quit, and Abstinence Expectancies of Current
Smokers at Baseline
Characteristic Valuea
24-h Quit attempt
Lifetime 119 (90.8)
Past year 85 (64.9)
Lifetime 24-h quit attempts, median (IQR), No. 4 (2-7)
Past year advice to quit
Any health care professional 93 (71.0)
Physician 80 (61.1)
Co-worker 20 (15.3)
Social worker 18 (13.7)
Nurse 37 (28.2)
Mental health professional 17 (13.0)
Other medical professional 17 (13.0)
Friends 66 (50.4)
Family members 61 (46.6)
Significant others 19 (14.5)
Career counselor or caseworker 10 (7.6)
Employer 9 (6.9)
Work experiences with smoking
b
Discriminated against as a smoker 60 (45.8)
Harder to get a job because a smoker 38 (29.0)
Hide smoking
At work 53 (40.5)
At home 28 (21.4)
Quit strategies
Cold turkey 91 (69.5)
Gradual reduction 56 (42.7)
Nicotine replacement
c
36 (27.5)
Quit smoking class or program
c
18 (13.8)
E-cigarettes 14 (10.7)
Acupuncture 11 (8.4)
Hypnosis 6 (4.6)
Tobacco quitline
c
5 (3.8)
Health professional counseling
c
4 (3.1)
Bupropion
c
4 (3.1)
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Prochaska et al. Page 17
Characteristic Valuea
Varenicline
c
2 (1.5)
Abstinence expectancies
d
Feel a sense of accomplishment 79 (60.4)
Would be more productive 54 (41.2)
Would be sick less often 53 (39.4)
Have more control over their life 54 (41.2)
Have less trouble finding work 35 (26.7)
Abbreviation: IQR, interquartile range.
a
Data are presented as number (percentage) of participants unless otherwise indicated.
b
Percentage who answered somewhat to extremely likely.
c
Evidence-based approach recommended by US Tobacco Treatment Clinical Practice Guidelines.18
d
Percentage who answered agree or strongly agree.
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Prochaska et al. Page 18
Table 3
Discretionary Spending Priorities Among Job-Seeking Smokers
Item Rank, Mean (SD)aMedian
Tobacco
b
5.02 (3.32) 4
Nutritious food 5.24 (3.56) 4
Transportation funds (eg, gasoline, bus fare)
c
5.65 (3.71) 5
Cellular telephone
c
5.70 (3.57) 5
Grooming care (eg, shave, haircut)
c
6.48 (3.48) 6
New clothing
c
6.73 (3.43) 7
Entertainment (eg, movies, magazines) 7.22 (3.47) 7
Prescription medications 7.47 (3.92) 8
Dental appointments 7.66 (3.26) 8
Nonemergency medical appointments 8.01 (3.61) 9
Gifts for others 8.27 (3.33) 9
Alcohol or nonprescribed drugs 8.67 (3.74) 10
Nicotine replacement therapy
b
8.87 (3.54) 9
a
Possible rank values ranged from 1 (greatest priority) to 13 (lowest priority). Participants who were current smokers at baseline were asked to
order items based on what they were most likely to purchase, assuming finite resources, using their discretionary funds, defined as money available
after one's bills are paid. The items were presented in random order.
b
Items directly related to smoking.
c
Items directly related to job seeking.
JAMA Intern Med
. Author manuscript; available in PMC 2016 November 16.
