Sherman, Bruce W. MD; Lynch, Wendy D. PhD
* Review previous data on the economic impact of smoking and the current status of smoking cessation benefits.
* Summarize the new findings on the effects of employee smoking on health care and productivity costs, including the estimated additional costs compared to nonsmokers.
* Discuss the implications for employer smoking cessation programs, including the likely effects of new policies under the Affordable Care Act.
Tobacco use continues to be one of the most significant preventable causes of disease, disability, and death in the United States. Smoking is associated with significant adverse health and economic impact at individual and societal levels.1 In the United States, smoking is responsible for an estimated $96 billion per year in health care expenditures, and an additional $98 billion per year in lost productivity due to premature death.2
For employers, smoking adds significantly to organizational health care expenditures.3 Additional costs associated with lost productivity include smoking breaks, illness-associated absenteeism, and presenteeism.3,4 Business costs from increased accidents, workers' compensation (WC) costs, costs associated with early retirement for disability, increased fires and fire insurance costs, as well as facility costs because of smoking all contribute to tobacco-related employer expense.4
Although the percentage of American adults who smoke declined from 20.9% in 2005 to 19.3% in 2010,5 smoking remains a significant employer concern. Employers do recognize smoking cessation as a high-value preventive care service offering.6 Yet, despite these observations, employers have not fully embraced the recommendations for smoking-cessation benefit design established by the Centers for Disease Control and Prevention.2,7 Only 4% of employers provide the full array of recommended program attributes that include counseling, pharmacotherapy, including nicotine replacement, at least two quit attempts per year, and value-based benefit design incentives to lower out-of-pocket costs for program participation.8 To encourage greater adoption, the Patient Protection and Affordable Care Act has mandated first-dollar benefits coverage for employer smoking-cessation programs. Recently proposed regulations will increase the financial incentive employers can incorporate in a smoking-cessation program to 50% of premium costs.9 These directives are likely to increase access to both counseling as well as prescription medications10 and are also likely to increase participation rates in smoking-cessation programs.
This study was performed to quantify and compare more completely the health- and productivity-associated costs for smokers and nonsmokers among the workforce of a single employer. Although other studies have evaluated the impact of smoking on employer health care costs,11,12 productivity,4,11,13 and WC costs,14 to our knowledge, this is the first study evaluating these combined costs in a single population. These findings may help employers more effectively quantify the value proposition for investments in smoking-cessation programs and enhance efforts to engage employees as well as family members in these offerings.
This is a multiyear, cross-sectional analysis using an integrated database created for a single, large US employer. The database linked health care utilization and costs, costs of presenteeism and incidental absenteeism, short-term disability (STD) claims and costs, and WC utilization and costs between 2008 and 2010. This study estimates the cost impact of smoking, using multivariate models that control for demographic variables.
The analysis was limited to a comparison of current smokers and never smokers. The correlation of health care cost reductions with the duration of time after smoking cessation is reasonably well established in the literature.15 Because there was no available measure of the duration since smoking cessation in this population, it was not feasible to treat the former-smoker group as a homogenous population. As a result, former smokers were excluded from analysis.
Subjects in this study were employees of a large multinational company headquartered in the midwestern United States. This employer has created a longitudinal, integrated database through Truven Health Analytics, which includes annual survey information about health and health risks (using the University of Michigan health assessment), as well as medical, absenteeism, and presenteeism data.
Medical, pharmaceutical, WC, STD, and survey data containing both smoking status and presenteeism data for the study population were extracted from the employer's database. The period of time for all data was January 1, 2008, to December 31, 2010. The survey (from University of Michigan16), included in the employer's benefit strategy during 2009 and 2010, had a question about smoking status. The assessment was customized to include additional, specific questions related to employee absenteeism and presenteeism derived from the HPQ-Select questionnaire (Integrated Benefits Institute17). These data were combined into a comprehensive analytic file spanning 3 years (January 1, 2008, to December 31, 2010).
The study population consisted of active employees and spouses/partners, aged 18 to 64 years, and with at least one full year of medical and/or disability coverage who had completed the survey, providing a valid indication of smoking status. Medical care utilization and cost analyses included dependents. Work-related outcomes (absenteeism, disability, and presenteeism) were limited to employees. For medical care utilization analyses, women with a medical claim for pregnancy were excluded.
Demographic descriptors were available from enrollment files, employment files, and survey responses. These included age and sex, ethnicity (classified as white or other), income, and educational achievement (college degree or not). Subsequent references to demographic variables include all the aforementioned, including income and education.
Definition of Smoking Status
Smoking status was determined on the basis of self-reported smoking status from two sources: benefits enrollment information and the annual survey. As part of the benefits enrollment process, individuals were required to declare current smoking status to be eligible for a nonsmoking premium discount. The health assessments were utilized to identify prior smokers who responded affirmatively to the question: “How would you describe your cigarette smoking habits?” The response options were still smoke, used to smoke, never smoked. The smoking status for each respondent was based on his or her response to the question. The number of cigarettes smoked on a daily basis was not included as a variable in this analysis. Respondents who considered themselves prior smokers were not required to meet a defined minimum time since smoking cessation. Concurrence of the self-reported benefits enrollment smoking status during each of the 3 eligibility years of “Yes” or “No” and 2 years of survey responses indicating smoking status as “Still smoke” or “Never smoked,” respectively, was required for categorization of individual smoking status.
Annual health care costs were calculated using a 3-year average total, adjusted for inflation by using the 2010 medical consumer price index. Total allowed costs included employee copayments and deductibles, as well as payments made by the employer. To remove the effect of outliers, individuals having an average health care cost more than $100,000 were capped at that level. Other health care utilization outcomes included rates of the use of emergency department (per 1000 covered lives), inpatient admissions (per 1000 covered lives), office visits, and the number of times the person filled a medication prescription. Outpatient drug costs were adjusted using the medical commodities consumer price index.
Presenteeism was measured using relevant questions from the HPQ-Select, which were included as part of the employer health assessment. Two-year HPQ-Select results (2009 and 2010) were annualized and averaged. Nonproductive time was converted to missed workdays by using the HPQ-Select absolute productivity question regarding usual performance in the past month. Responses were converted to a percentage, scaled, and then averaged over 2 years. Missed workdays were derived by multiplying the percentage amount by the typical number of paid workdays each year (240). Missed workdays were then monetized, multiplying hourly compensation figures from the Bureau of Labor Statistics ($27.75) by 8 to calculate the daily wage, which was then used to quantify the annual presenteeism cost for each employee.
Data on absenteeisms were collected using relevant questions from the HPQ-Select, also included in the employer health assessment during 2009 and 2010. Responses were annualized and averaged. Absent days were then monetized by multiplying the imputed number of absent days by the average daily compensation figures from the Bureau of Labor Statistics ($27.75 times 8). This figure was multiplied by 12 to calculate the annual absenteeism cost for each employee. Rates of STD and WC claims were averaged across 3 years, 2008 to 2010.
Lost-time costs for STD and WC claims included health care costs and wage replacement payments. These latter payments were attributed to the calendar year the claim began, regardless of the disability duration. These results were not adjusted for covariates in the analysis, as described hereafter.
Three general outcomes were investigated: the effect of smoking status on (1) health care costs, (2) specific types of health care utilization, and (3) other worker outcomes, such as absenteeism, disability, injury, and presenteeism. Also, health care expenses and other worker outcomes were combined to create a more comprehensive measurement of the employer cost of smoking. To adjust for population differences between smokers and nonsmokers, several covariates were included in the analysis. Demographic variables were included as covariates, including age, sex, education, income, and ethnicity.
All statistical analyses were performed using STATA (Stata Corp, College Station, TX). Basic differences between groups were compared by using t test (means) and chi-square test (percentages). Multivariate regression models were performed to estimate the effects of smoking on cost and utilization outcomes. Because most study subjects had nonzero health care costs across the 3-year period, a single model was used to analyze health care costs. Nevertheless, because of the number of zero values in other outcomes, two-part regression models were used for absenteeism, presenteeism, STD, and WC. The first part of the two-part regression model estimated the difference in the rate of (having any) events, using a logit model. The second part of the regression model estimated the difference in cost for those having a claim, using a General Linear Model with gamma distribution.18 Final cost differences were estimated using bootstrapping.19–21 A P < 0.05 was considered significant.
After inclusion and exclusion criteria are applied, 14,387 employees and spouses were eligible for the demographic and health care analyses. Of them, 2448 were smokers (17% of employees) and 11,939 were nonsmokers. Approximately 18% of the smokers were spouses.
Population demographics of the two groups, smokers and nonsmokers, are shown in Table 1. Although similar in sex, smokers differed significantly from nonsmokers in other attributes. A higher percentage of smokers than nonsmokers were white (92.6% vs 86.5%) and a lower percentage of smokers had college degrees (39.4% vs 68.4%) or earned more than $50,000 per year (41.4% vs 61.3%).
Smokers used some types of health care services at a higher rate than nonsmokers over the 3-year period. Table 2 shows estimated utilization rates for both unadjusted and adjusted regression models. Adjusted estimated rates of emergency department visits were higher for smokers than nonsmokers (393 per 1000 vs 249 per 1000; P < 0.0001), as were inpatient admissions (50.8 per 1000 vs 38.7 per 1000; P < 0.05). The adjusted number of medication refills per year was significantly higher for smokers (9.9) than for nonsmokers (8.8) (P < 0.0001); however, the adjusted number of office visits did not differ between groups (P = 0.11).
Total health care costs were higher for smokers than nonsmokers in each year from 2008 to 2010 (see Fig. 1). The 3-year, inflation-adjusted difference was $659 (P = 0.0097). After controlling for all demographics, income, and education, estimated costs for smokers was $4704 compared to nonsmokers' $4333 (P = 0.06).
A higher number of smokers than nonsmokers reported being absent from work (unadjusted: 17.4% vs 15.8%). In addition, smokers reported more days absent (8 days) than nonsmokers (6 days). After adjusting for demographics, differences in both prevalence (P = 0.03) and duration (P < 0.01) of absenteeisms remained significant. Total absenteeism costs were estimated to be $674 higher for smokers than nonsmokers ($2697 vs $2023) (P < 0.01), as shown in Table 3.
As shown in Table 3, results from two-part regression models for presenteeism were mixed. Fewer smokers reported having any decrement in work performance (presenteeism) compared with nonsmokers (75.5% vs 81.2%). Nevertheless, of those reporting any presenteeism, smokers reported a significantly higher level of performance impairment. After adjusting for demographic variables, then combining these effects, smokers averaged significantly higher presenteeism cost than nonsmokers ($2167 vs $1941; P < 0.05).
Neither of the two-part models (rate of claims and size of claims) estimating differences in WC claims was significant. As shown in Table 3, the combined model estimated a difference of $101, which was also not significant (P = 0.13).
Results from the two-part models estimating differences in STD costs between smokers and nonsmokers showed mixed results. Smokers had a significantly higher adjusted, estimated rate of claims. Nevertheless, the adjusted cost per claim was not significantly different. Overall, combining these two findings, smokers and nonsmokers, the difference of $11 was not significant (P = 0.66).
Total expenditures for smokers and nonsmokers, controlled for demographics, income, and education, are listed in Table 3. Combining medical, absenteeism, presenteeism, STD, and WC, smokers had an estimated annual cost of $9969 compared with $8586 for nonsmokers. The total difference was $1383. Restricting to cost components where statistical significance was P < 0.10, the difference was $1271. Restricting to a significance level of P < 0.05 lowers the overall difference to $900.
This study provides a broader view of the financial burden of employee smoking for a single large employer. To our knowledge, this is the first time that employer-related health care and productivity costs have been characterized within a single population. By linking specific program cost components with individual smoking status and health care utilization and costs, a more comprehensive perspective of employer health care and productivity costs was able to be generated. Specific program areas in this analysis included medical and pharmacy costs, self-reported lost productivity, STD, and WC.
Depending on the level (α) chosen for type I error, the difference in costs between smokers and nonsmokers was $900 to $1383 per year, which equates to a 10% to 16% higher total cost. Most of the added cost for smokers is attributable to lower productivity and lost time.
Health Care Costs
Three-year, inflation-adjusted health care costs were consistently higher for smokers than for nonsmokers. Controlling for age and sex, the difference was significant, at 9.2% (P = 0.031). Nevertheless, after controlling for race, income, and education in addition to demographics, the magnitude of the incremental difference between smokers and nonsmokers was smaller, at 8.2% (P = 0.06). The observed difference between these results and previous findings may result from differences in the respective study populations. Higher smoking rates and lower health status are prevalent in lower socioeconomic groups.22 In our study, a higher socioeconomic status and active employment status of the population may help explain why the smoking-associated increase in health care costs was smaller than noted elsewhere.
These results are generally consistent with prior publications that have demonstrated an incremental increase in health care costs associated with smoking.12,23 These have presumably been due to the direct effect of smoking-related health conditions, with increases in hospitalization rates12,24,25 and outpatient visits.12,24 In this study, a similar pattern was observed, with a significant increase in hospitalization rates, emergency department use, and prescription utilization noted among smokers after adjustment for demographics, income, and education. Outpatient visit utilization rates were not statistically different between the two groups.
Although an explanation for the observed health care utilization patterns is beyond the scope of this study, evidence indicates that smokers are less likely to have a personal health care provider26 and utilize immunizations27 or other routine preventive care services.28 Taken together, these results suggest that smokers may have a more reactive approach to utilization of health care services, which may contribute to increased inpatient and emergency department use.
Several studies have documented productivity loss among smokers in comparison with nonsmokers, which has been attributed to both illness-related absenteeism and presenteeism. Regarding absenteeism, Halpern et al13 demonstrated that among a population of airline reservation workers, smokers missed 6.16 days of work per year because of sickness (including smoking-related acute and chronic conditions), compared with nonsmokers, who missed 3.86 days of work per year. Similarly, among a diverse group of employers, illness-related absenteeism accounted for 6.7 workdays per year among smokers, in comparison with 4.4 workdays per year for nonsmokers.4
Smoking-related presenteeism seems to be a more significant contributor to lost productivity relative to absenteeism. Using a self-reported presenteeism measurement instrument, annualized losses of 9.6 workdays per year among smokers, in contrast to 5.4 workdays per year among nonsmokers, have been reported.4 Presenteeism-related lost productivity from smoking also includes time lost because of smoking breaks. With an average of 6 minutes for a smoking break, employees smoking five cigarettes a day during work hours may work 3 weeks less per year than nonsmokers.29
When absenteeism and presenteeism components are combined, the difference between smokers and nonsmokers becomes even more significant. The American Productivity Audit of the US workforce found that tobacco use was one of the greatest causes of lost worker production time, which was directly related to the amount smoked. Lost productivity estimates for workers who reported smoking one pack of cigarettes per day or more were 85% higher than that observed for nonsmoking employees or employees who had previously quit smoking.30
When these combined lost productivity sources were quantified on the basis of the equivalent value of hourly wages and benefits, the total annual employer cost was estimated at $1897 (2001 dollars) ($2824 in 2011 dollars [adjusted for medical inflation]),2 and $1807 (2005 dollars) ($2262 in 2011 dollars [adjusted for medical inflation]).4 In the present study, lost productivity was the greatest contributor to smoking-associated employer costs, with significance persisting after controlling for demographics, income, and education. These values are consistent with prior reports, with presenteeism representing the most substantial component of productivity loss.
Smoking adversely impacts wound healing and, as such, can be reasonably anticipated to increase WC claims costs for occupational injuries. This study lends weak support to this hypothesis, with a not statistically significant trend toward higher WC case prevalence and costs among smokers. To our knowledge, this study is the second to report on the impact of smoking status on WC costs. A 2001 study of long-term employees reported an average payment of $2189 in WC costs for smokers, compared with $176 for nonsmokers.14 Our findings revealed comparable costs for nonsmokers, but WC costs for smokers were considerably lower. Several factors may account for this difference. First, as the authors of the earlier study noted, there was a significant skewing of WC costs, such that 10% of the participants were responsible for 54.5% of the observed medical and lost-time expenditures. Given the small population of high-risk employees (including smokers) sustaining a work-related injury (n = 14), it is possible that one or more high-cost cases were included in previous studies, thereby significantly inflating the results. In addition, the small sample size in that study precluded any further results adjustment for comorbid conditions or other factors that might impact injury severity.
Total Health Care and Productivity Costs
Combining the incremental additional health care and productivity costs attributed to smokers, the total annual cost of employee smoking is $1383 per individual, as measured in this study. In comparison, the Centers for Disease Control and Prevention estimated this cost to be $3383 (1999 dollars) ($5426 in 2011 dollars [adjusted for medical inflation]) in combined medical costs and lost productivity.31 Although costs associated with lost productivity are comparable, the difference in health care costs is notable. This finding may result from differences in the population demographics between the two analyses or may reflect a different illness-related burden between the study group and the Centers for Disease Control and Prevention estimate, which represents data from a broader population.
There are a number of limitations to this study. First, the results reflect the results of a single employer and may not be generalizable to other employee populations, particularly those in other types of businesses. Importantly, this study includes only individuals (employees and their associated spouses) who were actively employed during the period of data collection and excluded individuals older than 65 years. This bias may underestimate the extent of smoking-related illness, because individuals with significant illness may no longer be active members of the workforce.
Also, the productivity data included in the study originate from self-reported data collected in the context of health assessment completion. As such, although most of the workforce completed the health assessment, this population may not necessarily represent the findings from the broader employee population, thereby introducing self-selection bias into the study. It is also possible that those individuals completing the health assessment were more engaged in their work and were less likely to experience absenteeism or presenteeism, thereby underreporting the experience of those not completing the health assessment. In addition, this study design does not account for elements of benefit design that may have impacted health care utilization. Nevertheless, these were applied consistently across the entire workforce and would not have been expected to selectively adversely impact either the smoker or nonsmoker populations.
Support for smoking cessation has been recognized as a high-value health benefit offering,6 and has been included among first-dollar covered preventive care benefits in the Patient Protection and Affordable Care Act.32 For employers to be willing to proactively endorse use of this benefit, a clear demonstration of the business value of smoking cessation is essential. This study provides an additional support for the business value of investments in smoking cessation.
Notably, although surveys suggest that many individuals are interested in smoking-cessation programs,33 participation rates remain low. These data from a single employer provide support for enhanced communication and the development of additional strategies to increase individual engagement in smoking-cessation programs. If enacted, the proposed regulations to increase financial incentives for smoking cessation to 50% of health care costs34 will likely increase individual participation in smoking-cessation activities.
Analysis of health- and productivity-related costs associated with smoking among health beneficiaries of a single large employer revealed significant differences in health care utilization patterns, as well as absenteeism and presenteeism, but not for health care costs. Overall, employees who smoke were estimated to cost employers $900 to $1383 more than their nonsmoking counterparts. Employers seeking to improve workforce performance may want to expand their approach to smoking-cessation efforts as part of their communications and benefit design strategy.
The authors thank Truven Health, Inc, for its role in providing the statistical analysis.
1. Leistikow B. The human and financial costs of smoking. Clin Chest Med. 2000;21:189–197.
2. Centers for Disease Control and Prevention. Smoking-attributable mortality, years of potential life lost, and productivity losses—United States, 2000–2004. MMWR Morb Mortal Wkly Rep. 2008;57:1226–1228.
4. Bunn WB III, Stave GM, Downs KE, Alvir JM, Dirani R. Effect of smoking status on productivity loss. J Occup Environ Med. 2006;48:1099–1108.
5. Centers for Disease Control and Prevention. Current cigarette smoking prevalence among working adults—United States, 2004–2010. MMWR Morb Mortal Wkly Rep. 2011;60:1305–1309.
8. National Business Group on Health. Tobacco use treatment. Moving Science into Coverage: An Employer's Guide to Preventive Services. Available at: www.businessgrouphealth.org/preventive
. Published September 23, 2011. Accessed September 10, 2012.
10. Melek S. The Synergistic Effect of PPACA and MHPAEA: Implications for Smoking Cessation Therapies. Denver, CO: Milliman Inc; 2011.
11. Kowlessar NM, Goetzel RZ, Carls GS, Tabrizi MJ, Guindon A. The relationship between 11 health risks and medical and productivity costs for a large employer. J Occup Environ Med. 2011;53:468–477.
12. Javitz H, Zbikowski S, Swan G, Jack L. Financial burden of tobacco use: an employer's perspective. Clin Occup Environ Med. 2006;5:9–29.
13. Halpern M, Shikiar R, Rentz A, Khan Z. Impact of smoking status on workplace absenteeism and productivity. Tob Control. 2001;10:233–238.
14. Musich S, Napier D, Edington D. The association of health risks with workers' compensation costs. J Occup Environ Med. 2001;43:534–541.
15. Musich S, Faruzzi SD, Lu C, McDonald T, Hirschland D, Edington DW. Pattern of medical charges after quitting smoking among those with and without arthritis, allergies, or back pain. Am J Health Promot. 2003;18:133–142.
18. Manning WG, Morris CN, Newhouse JP, et al. A two-part model of the demand for medical care: preliminary results from the health insurance study. In: van der Gaag J, Perlman M, eds. Health, Economics, and Health Economics. Amsterdam, the Netherlands: North-Holland; 1981.
19. Manning W, Basu A, Mullahy J. Generalized modeling approaches to risk adjustment of skewed outcomes data. J Health Econ. 2005;24:465–488.
20. Barber J, Thompson S. Analysis of cost data in randomized trials: an application of the non-parametric bootstrap. Stat Med. 2000;19:3219–3236.
21. Johnston J, DiNardo J. Econometric Methods. 4th ed. New York, NY: The McGraw-Hill Companies Inc; 1997.
22. Adler N, Newman K. Socioeconomic disparities in health: pathways and policies. Health Aff. 2002;21.
23. Penner M, Penner S. Excess insured health care costs from tobacco-using employees in a large group plan. J Occup Med. 1990;32:521–523.
24. Kahende J, Adhikari B, Maurice E, Rock V, Malarcher A. Disparities in health care utilization by smoking status—NHANES 1999–2004. Int J Environ Res Public Health. 2009;6:1095–1106.
25. Eisner M, Iribarren C. The influence of cigarette smoking on adult asthma outcomes. Nicotine Tob Res. 2007;9:53–56.
26. Hughes M, Hannon P, Harris J, Patrick D. Health behaviors of employed and insured adults in the United States, 2004–2005. Am J Health Promot. 2010;24:315–323.
27. Pearson W, Dube S, Ford E, Mokdad A. Influenza and pneumococcal vaccination rates among smokers: data from the 2006 Behavioral Risk Factor Surveillance System. Prevent Med. 2009;48:180–183.
28. Vander Weg M, Howren M, Cai X. Use of routine clinical preventive services among daily smokers, non-daily smokers, former smokers, and never-smokers. Nicotine Tob Res. 2012;14:123–130.
29. Tsai S, Wen C, Hu S, Cheng T, Huang S. Workplace smoking related absenteeism and productivity costs in Taiwan. Tob Control. 2005;14(suppl I):i33–i37.
30. Stewart W, Ricci J, Chee E, Morganstein D. Lost productivity work time costs from health conditions in the United States: results from the American productivity audit. J Occup Environ Med. 2003;45:1234–1246.
31. Centers for Disease Control and Prevention. Annual smoking-attributable mortality, years of potential life lost, and economic costs—United States, 1995–1999. Morbid Mort Wkly Report. 2002;51:300–303.
33. Centers for Disease Control and Prevention. Quitting smoking among adults—United States, 2001–2010. Morbid Mortal Wkly Rep. 2011;60:1513–1519.
Copyright © 2013 by the American College of Occupational and Environmental Medicine