Publication
Cost and benefit analysis of smoking cessation in the workplace
2. Review of the literature
The literature review was undertaken in order to identify the costs of workplace smoking and benefits of introducing a restrictive smoking policy or cessation help. Identification of the key costs and benefits is vital so that they can be measured and then costed, hence the review forms the basis of the estimates presented in Section 4. A summary of a selection of studies investigating different smoking cessation interventions is presented to provide an indication of the effectiveness of different programmes.
2.1 Introduction to the literature
There is a fairly large body of literature exploring the costs of smoking in the workplace. This chapter will review some of the studies which have investigated the costs of smoking to employers, together with a brief review of the effectiveness of different worksite smoking programmes and cessation help interventions. However, the vast majority of the work has been undertaken in the United States and its applicability to the UK is therefore limited. Studies made in a single workplace or geographical area are not generalisable to other populations as the characteristics of subjects will be specific to the study. Problems for interpretation include different rates of smoking prevalence between geographical areas and workforces, different attitudes to smoking, and different characteristics of workers, all of which will obviously affect the cost-effectiveness of any intervention to reduce workplace smoking or to help employees to stop smoking.
Articles included in this review were identified from searches of Medline (1976-1995) and the BIDS Social Science Index databases (1983-1995). The search was based on terms such as “workplace”, “worksite”, “smoking”, “health policy”, “benefits” and “cost benefit analysis”. Copies of the search strategies are presented in Appendix One. In addition, the Applied Social Sciences Index and Abstracts (1988-1995) were searched by hand. This strategy has the advantage of identifying most of the studies published in academic journals but is less helpful in identifying unpublished material. It is recognised that these studies, especially those conducted within workplaces, may be under-represented in this review. However, the resource constraints of the project did not allow a more extensive search strategy to be adopted.
A total of 267 articles were identified from the search strategy (see Appendix). Articles concerned with the clinical aspects of drug testing and articles based on non-specific employee assistance programmes were rejected from the literature identified. The articles selected for this review were those based on the costs and benefits of smoking cessation and the effectiveness of smoking cessation policies. A total of 52 articles were used in the review.
Section One will be structured as follows. In Section 2.2 the major cost categories which result from employee smoking are identified and estimates of the magnitude of the costs identified in the section are presented. In Section 2.3 the estimates of the total cost of employee smoking are outlined, whilst the effectiveness of workplace smoking cessation interventions is presented in Section 2.4.
Cost estimates in U.S. dollars were converted to current UK costs by firstly inflating to 1995 equivalent costs using U.S. inflation rate, and then converting to U.K. pounds by applying the dollar-sterling exchange rate which prevailed at the time of the study.
2.2 The costs of employee smoking
The costs of employee smoking will depend upon the perspective from which the study is undertaken. A societal perspective would include all costs and benefits to society. Such a study would include the value of NHS resources used to treat smoking induced disease and the impacts of passive smoking outside of the work place. This report however focuses on the perspective of the costs and benefits to the employer. The costs of employee smoking are divided into six major categories: absence from work; insurance costs; productivity costs; passive smoking; occupational health; and damage to plant and machinery.
2.2.1 Absence from Work
There is a relatively large body of literature examining the relationship between smoking and workplace absenteeism. However, any cause-effect relationship which may exist between smoking and absence from work is difficult to evaluate because of the numerous potential confounding factors. Determination of a cause-effect relationship would require a random allocation of workers to a smoking (intervention) and non-smoking (control) group. Unfortunately the ‘gold standard’ of a randomised experiment is not ethically feasible, although it is possible to allow for the effects of potential confounders by using statistical techniques such as regression analysis. The studies identified in the literature search have used observational data in order to investigate associations between the levels of smoking and sickness absence. The following section presents a summary of the evidence to date, together with a discussion of their limitations.
Lowe (1963) investigated the link between emergency room visits, absenteeism and smoking amongst 3,341 male workers at the GEC plant in Birmingham, UK between 1956 and 1957. When age standardised, Lowe’s results showed an increase in the number of days absent from work as the number of cigarettes smoked per day increased. This relationship was more pronounced amongst older workers, which is what would be expected given the known relationship between smoking and disease. Frequency of emergency room attendance also increased with number of cigarettes smoked, with a stronger relationship amongst younger workers. Holcomb and Meigs (1972) examined medical absenteeism amongst 226 workers in a US manufacturing plant and also found a positive relationship between the length of time that an individual had smoked and the annual levels of absenteeism. They found more absences with higher daily consumption, with lower absence rates amongst those who had never smoked or had stopped smoking. The relationship between the number of cigarettes smoked and absence rates showed higher rates of absence amongst non-smokers than those who had smoked cigarettes for under 15 years, although the highest rates were amongst smokers who had smoked for over 15 years. The results must however be treated with caution, as they cannot prove a cause effect relationship as confounding factors are not fully explored, and associations were primarily assessed by the use of correlation coefficients. In addition, lower levels of absenteeism were found amongst cigar and pipe smokers.
Van Tuinen and Land (1986) studied 406 employees from the Missouri Department of Health over a 20 month period. The results showed that the 97 smokers in the Department took an average sick leave of 5.3 hours per month in comparison to the 4.3 hours taken by non-smokers. The result was statistically significant at a 95% confidence level. Using the average departmental salary this was translated to an annual excess cost of $11,931, equivalent to £14,821 at 1995 prices (see Section 2.1 for details of conversion procedure). However, the study failed to identify the amount that each individual smoked or the reasons for taking sick leave. Using the conservative estimate of the total cost to the employer of $601 (£572) per smoker per year (Kristein 1983), the total cost to the Department becomes $73,497 in 1984 dollars (£91,299 at 1995 prices). Kristein’s estimates are based on results of the National Center for Health Statistics’ Health Interview Survey aggregate epidemiological data and population averages, and the estimates for each cost category are presented in the sections below. Van Tuinen and Land also use the estimates of Weis (1981) and conclude that the total cost could be as high as $510,802 (£500,310).
The DuPont study (Bertera, 1991) investigated the impact of six behavioural factors on absenteeism and health care costs amongst 45,976 employees at the DuPont Company in the US. The results showed that smokers had an excess absenteeism of 0.90 days per year (p<0.001) as compared to non-smokers. After controlling for six other risk factors (excess alcohol intake, obesity, elevated cholesterol, high blood pressure, inadequate seatbelt use and lack of exercise) current smokers were found to incur excess illness costs of $960.04 per person (p<0.001) in comparison to non-smokers (£602.27 at UK 1995 prices). The total cost to the company based on the total DuPont US workforce of 96,000 was estimated at $26,174,530 per annum as a result of employee smoking (£16,420,409 at UK 1995 prices).
The Whitehall II Study (North et al., 1993) was based on a population of 6,281 male and 1,434 female non-industrial civil servants in London, with the aim of explaining the differences in rates of sickness absence among workers. The results showed male smokers to have a 46% higher rate of short term absence (periods of seven days or less) and a 81% higher rate of long term absence (periods of greater than seven days). For women the rates were 9% and 37% respectively. However, the lack of baseline data meant the results were difficult to interpret. Wooden and Bush (1995), in a study of the Australian National Health Service, found the probability of absence to be higher amongst ex-smokers and to fall with the number of years since smoking cessation. For example, an individual who ceased smoking 20 years ago would be expected to be 4.5 times less likely to be absent from work than someone who stopped smoking last year. However, a problem with these studies is that the baseline absence rate used in the analysis was not stated. This makes it difficult to apply the findings to other scenarios since the actual time spent absent cannot be determined, as the actual percentage cannot be translated into number of days absent from work.
Jeffery et al. (1993) report on the Healthy Worker Project, a randomized trial in Minnesota, which involved assigning 32 workplaces to either a smoking cessation and obesity reduction programme or a control group. When the results were restricted to the 271 employees who were smokers at the baseline assessment, the reduction in absenteeism was found to be significant. The authors estimated the cost of each sick day saved was $2.60 (£1.71 at 1995 prices) which, if realistic, suggests that work-site smoking cessation programmes offer the possibility of immediate cost savings in terms of absenteeism, besides the other savings such as insurance and long term disability. However, a particular weakness of the study is that absenteeism was based wholly on self report data with no attempt to validate the evidence.
Jackson et al. (1989) attempted to investigate the excess absence between smokers and ex-smokers using a time series control design to investigate patterns of absenteeism. The study population was drawn from the employees of a North Carolina pharmaceutical company. Fifty smokers were matched to the experimental group of 50 ex-smokers, and absences from work for each individual were examined retrospectively. Examination of the data for ex-smokers showed a statistically significant decrease in absences following cessation. A statistically significant difference was also evident between smokers’ and ex-smokers’ absenteeism rates. The study also suggests that absence amongst smokers falls as time since smoking cessation, but fails to detect any relationship between absence and the number of cigarettes smoked. A notable feature of this study is that ex-smokers’ absences tended to decrease before stopping smoking, suggesting that there may be other causes of reduced absenteeism (confounding factors) besides smoking cessation. The issue is complicated by the complex interrelationships between smoking, stressful lifestyles, physical symptoms and upper respiratory tract infections.
An estimate of the impact of absence as a consequence of employee smoking on the U.S. economy as a whole quoted by Serxner et al. (1993) shows the effects to be considerable, with an estimated loss of 81 million working days. These figures are derived from a Surgeon General’s report which stated that smokers tend to be absent from work 33% to 45% more than non-smokers (Royal College of Physicians, 1993).
Studies have provided a range of estimates of absenteeism costs which result from employee smoking. The studies in the review were based on different employment situations, which may explain part of the differences in their findings. It would be expected that costs will vary depending upon employee characteristics, the nature of the workplace, the geographical area and population in question. For example, different levels of smoking prevalence and attitudes to smoking which vary between populations will affect the results of these studies. Each study outlined above will therefore provide estimates specific to the workplace in question, and therefore extreme caution should be exercised when attempting to put such estimates into contexts outside that of the particular study.
2.2.2 The Insurance Costs of Smoking
Many large companies offer health insurance to their employees. If a company pays premiums which are experience rated then its premiums will in part depend on the level of claims in the previous period. With such schemes, therefore, a firm which has made large claims in the past will face high insurance premiums.
Evidence indicates that smokers do tend to use more medical services than non-smokers. Fielding (1986) estimated the average smoker’s utilisation to be up to 50% higher than a non-smoker. Therefore, if insurance premiums are experience rated then premiums will be higher for businesses with a large proportion of smokers. This is supported by Dawley et al. (1991) who argue that much of the incentive for businesses to reduce employee smoking in the United States has come from massive increases in health care costs.
Kristein (1983) reports health care insurance excess costs of between $105 and $204 per annum (£100-£194 at 1995 prices) for the average adult smoker. The estimate is based on the annual excess-illness cost (direct excess medical-care spending due to smoking) divided by the number of estimated number of smokers in the United States.
Although a significant cost category in American research, excess health insurance costs are only relevant to companies which provide health insurance to employees. This is widespread in the USA, but tends to be limited to only the largest employers in the UK. However, any savings in health care resources in the UK as a result of employer smoking policies would enter the cost benefit equation as a benefit of society as a whole.
Historically, a large number of industrial fires have resulted from employee smoking, which increases the risk of accidents due to fire. Luce and Schweitzer (1978) estimated $5 (£7.72 at 1995 prices) per smoking employee per year for non residential fire losses. Smokers have also been shown to have twice the accident rate of non-smokers (Yusta and De Guevara, 1973 (in Kristein, 1983) and this may be due in part to loss of attention, smoking hand being occupied, eye irritation and cough.
Smokers can raise life assurance costs due to the early disability and early retirement costs imposed on businesses. For example, earlier retirements and disability claims will increase the premiums which must be paid. Therefore, large firms either self insuring or with experience rated employee insurance premiums have a potential to offset part of the cost of their smoking cessation programme through reduced premiums.
2.2.3 Productivity
Employee smoking results in direct productivity costs. Such costs include time taken by ‘smoking rituals’, inefficiency and errors due to excess carbon monoxide levels in employees and eye irritation (Kristein, 1983). Workers dexterity may also be affected if their smoking hand is occupied. Kristein’s 1983 estimate was a productivity loss of $80 per year per smoker (£92 at 1995 prices). However, Walsh and McDougall (1988) argue that the productivity argument may be incorrect if cigarette smoking helps addicted workers to concentrate, although there is no direct evidence to support this view. Hocking et al. (1991) also reported that productivity losses could be incurred due to time losses as smokers must leave work to go outside to smoke. Under these assumptions, studies attempting to evaluate productivity gains may be questionable if these negative effects are excluded. Smoking bans may also reduce the morale of smokers.
A further cost in this category is productivity losses due to smoking related deaths and retirements. High unemployment may mean that the employer can replace workers but training costs will be incurred (Nelson, 1986). The actual training cost will depend upon the type of labour used.
2.2.4 Involuntary Smoking
The dangers of ‘passive smoking’ have long been known. Non-smoking employees suffer both physical and psychological damage from the tobacco smoke produced by their smoking colleagues. The psychological effect can be to reduce work effort. In 1988 some 28.5 million non-smoking workers in the U.S. were employed in workplaces with few or no smoking restrictions (Repace et al., 1993). Studies have shown (White and Froeb, 1980) that passive smoking can have the same impact upon a non-smoker’s health as light smoking (1-10 cigarettes per day). By applying the risk of cancer for light smokers to non-smokers Kristein estimated that passive smoking cost the average business between $53 and $105 (1980 dollars) in terms of insurance (£50-£100 at 1995 prices). In terms of absenteeism, accidents and productivity costs this was $27 to $56 (£26-£53 at 1995 prices).
Improved ventilation has been one means used to remove tobacco smoke from some workplaces. However, such ventilation systems are extremely expensive to install. In addition, the fresh air which replaces the smoky air must be heated which adds to the firm’s heating bill (Nelson, 1986).
2.2.5 Occupational Health
In certain occupations tobacco smoke can have the effect of making workplace agents more harmful to employee health (Coultas et al., 1992; Strasser, 1991). Such an interaction can occur between tobacco smoke and substances such as cyanide, acetone and lead (Kristein, 1983), nickel (Magnus et al., 1982) and chloromethyl ethers (Weiss et al., 1979). Although the evidence is limited, a synergism has been shown to exist between cigarette smoke and asbestos (Saracci, 1987). This factor is made more significant by the fact that the workers most likely to smoke are also those with the highest risk of exposure to such harmful industrial agents (Walsh et al., 1988). Exposure to the synergistic effects of multiple agents can cause employee disability and consequently claims for large amounts of compensation, which can be averted by preventing workplace smoking. Industries in which smoking has been shown to raise occupational health risks are uranium mining, gold mining, asbestos and rubber production.
2.2.6 Damage to Plant and Machinery
Smokers impose clean up costs upon employers, such as clearing away ash, cigarette butts and matches. Damage to equipment, for example cigarette burns and falling ash, can also impose significant costs on employers. Tobacco smoke can cause particular damage to sensitive machinery, including computers, telephone switchboard equipment and fine measuring equipment (Nelson, 1986). Redecoration costs will also be higher in workplaces permitting smoking, as environmental tobacco smoke damages surfaces and wall-coverings.
2.3 Total cost of employee smoking estimates
The total cost of employee smoking estimates are derived by combining the above categories to produce a single figure. If both costs and benefits are included in the analysis this figure will show whether the effect is an overall net cost or net benefit. A number of studies have attempted to provide a total cost figure.
Weis (1981) attempted to calculate the total cost of smoking in the workplace. The estimated costs were $220 (£215 at 1995 prices) as a result of excess absenteeism, $230 (£225) medical care costs, $765 (£749) due to lost earnings, $90 (£88) for inflated insurance premiums, $1,820 (£1,782) due to time lost on the job, $500 (£490) for property damage and depreciation, $500 (£490) for maintenance and $486 (£476) for the effects of involuntary smoking. This gave a total cost of $4,611 (£4,515) per smoking employee per month.
A more conservative estimate of between $336 and $601 at 1980 prices (£320-£572 at 1995 prices) per smoking employee per month was made by Kristein (1983). The estimate was low because the costs of property maintenance and depreciation were excluded and used lower estimates for the costs of involuntary smoking and time lost on the job.
Kristein’s estimates were inflated to 1986 prices and used as the cost per employee in the computer simulation set up by Swank et al. (1988). The model was based on introducing a smoking cessation programme in a population of hospital nurses, an occupational group which tends to have a high prevalence of smoking. Six scenarios are presented with different programme success rates and proportions of incoming smokers. The net economic impact of the programme ranges from -$284,000 to $684,000 (-£293,099 to +£705,914 at 1995 prices) with the largest economic gain in the scenario with the lowest incoming staff smoking rate. The paper demonstrates how the proportion of incoming staff who are smokers and the turnover rate are important in the determination of the economic impact of a smoking cessation programme, and therefore a programme which yields economic benefits in one workplace need not have the same effect in another workplace with different characteristics. Swank et al. also note that whilst some economic benefits can be gained from simply banning smoking in the workplace, which is the least expensive policy, this should only be part of the employer’s approach. In order to reduce the costs of ill health and associated absenteeism, it is necessary to eliminate smoking both on and off the job.
Nelson (1986) attempted to calculate the economic costs of smoking to employers in Northern Ireland. Absenteeism as a result of smoking related illness is estimated to cost employers an annual loss of almost £40 million at 1984 prices (£66.5 million at 1995 prices). This is supplemented by a loss of £10 million (£16.6 million) due to smoking related deaths and retirements. In organisations where smokers can smoke in designated areas, the ‘smoking ritual’ is estimated to take 30 minutes per day, which would cost employers over £84 million (£139.6 million). An estimate of five minutes daily is estimated for workplaces where employees can smoke ‘on the job’. The cost estimate here is approximately £13.5 million (£22.4 million). Damage to plant and machinery must also be added to these costs, including fire damage attributable to smoking of almost £2 million. The overall cost of employee smoking to employees in Northern Ireland was estimated to be £100,704,000 (£167.4 million).
A pilot study of a smoking cessation programme at the Hawaii Telephone Company in 1990 (Serxner et al., 1993) estimated that the intervention paid for itself in one year and saved an additional $350 (£230 at 1995 prices) per year for each individual remaining smoke free after the study period. This figure excludes productivity and morale benefits. However, the sample size of 12 is extremely small and these are likely to be the more motivated workforce members due to their self selection into the study.
2.4 Effectiveness of smoking cessation programmes in the workplace
The following provides a review of studies which are concerned with the ‘effectiveness’ of different worksite smoking policies. The review includes only specific studies of smoking cessation interventions in the workplace. Attempts were made to identify studies which included evidence of the cost effectiveness of the intervention.
2.4.1 Cessation Help in the Workplace
Smoking cessation programmes have incorporated several different strategies. These have included self-help manuals, support groups, media coverage and incentives such as lotteries or weekly bonuses.
A cost-effectiveness analysis of the smoking cessation programme at the DuPont Company, Wilmington, Delaware is reported by Bertera et al. (1990). Approximately 1,400 employees in a large office complex were given the opportunity to attend stop smoking group meetings. Of the 110 employees which attended, 90 employees filled out cards showing interest in one of the two cessation methods. Twenty-seven employees received either a self-help kit containing a manual containing a quit manual, a maintenance manual and an audio cassette. The remaining 43 employees signed up for the clinic based approach to quitting. The 70 employees provided baseline and 18 month follow up data by self report. After 18 months, 3 of the 27 (11%) participants in the self help group were abstinent compared to 9 of the 43 (21%) in the clinic based group. However, the results were not statistically significant. A cost analysis was attempted by the authors. The total cost of the self help programme was estimated at $448.50 (£294.66), whilst the clinic intervention was costed at $1,385 (£909.94). The cost per successful quit in the self help group was therefore $149.50 (£98.22) in comparison to the $153.89 (£101.10) in the clinic group. The authors also attempted to attach costs to the number of employees who managed to reduce their cigarette consumption by at least 33%. Cost per person was estimated at $44.85 (£29.47) in the self help group and $197.86 (£129.99) at the clinic.
The authors draw two major conclusions. Firstly, in the light of the similar costs of the programmes, employees should be offered the choice of the two programmes in order that the largest number of potential participants is reached. Secondly, basing the cost of employee smoking on the estimates of Kristein (1983), if the programme costs per quitter are $150 as the study suggests, a smoking cessation programme offers a return of $4 for every $1 invested. Therefore smoking cessation policies appear to offer a cost-beneficial investment for employers.
It should, however, be noted that this study has serious limitations. Firstly, employees are not assigned to the groups randomly, raising the question of biases in the estimates of effectiveness. Secondly, the method of attributing costs to the programmes is not presented. Thirdly, data are based on self report with no attempt to validate the results, which as the authors concede may lead to employees giving what they believe to be socially desirable responses. Fourthly, the cost-benefit ratio of 4:1 is based on the assumption of constant returns to scale, in that there is the implicit assumption that it costs the same amount to help every smoker stop smoking. In reality it is likely that such a programme would exhibit diminishing returns so that the 4:1 return is not attainable as the policy is expanded. This would be the case where it is easier to help the first few smokers to stop smoking, whereas a considerably higher amount of resources would be needed to help the more dependent smoker. Finally, the differences in quit rates between the groups were not statistically significant, which should be one of the first requirements of such a study. Hence it must be noted that this study should be interpreted with caution.
A study of 400 smokers at the Dow Chemical Company was undertaken by Danaher (1980), where ex-smokers received a $1 (£0.95 at 1995 prices) bonus for abstinence and chances to win a quarterly $50 (£47.55) bonus and a boat. The results reported showed 76% of the study population remaining abstinent, although no validation of the self report data was provided. Klesges et al. (1986) found evidence of higher cessation rates in cessation programmes with competition for monetary prizes. The study included one US loan institution and four banks. The banks were awarded monetary prizes for the highest take up of cessation help and the highest level of cessation. No competition was included for the loan institution. Results showed higher levels of abstinence in the competitive environment (16% abstinence at six month follow up) in comparison to the non-competitive scenario (7% abstinence). Klesges, Cigrang and Glasgow (1987) review seven incentive/competition programmes and find the average post-test cessation rate to be 44.8% with 26% at follow up (in Jason et al., 1989). The authors suggest that programmes including incentive components may motivate smokers and achieve higher outcome rates.
Worksite support groups provide a useful addition to workplace smoking cessation programmes. Jason et al. (1987) analysed a programme using self help manuals in combination with television broadcasts and smoker support groups. The comparison group received just the manuals and television schedules. At the end of the programme 41% of smokers receiving the full package were abstinent, compared to 21% without the worksite support group. Unfortunately the abstinence rates in the two groups were not statistically significant at 12 month follow up. The effectiveness of group meetings and incentives were also investigated by Jason et al. (1989). The twice weekly 1-hour group meetings and incentive package of a $50 (£39 at 1995 prices) lottery at each group meeting were provided as a supplement to a core programme of self-help manuals, television and support groups. Thirty-eight corporations were randomly assigned to one group or the other. The results showed that 42% of smokers provided with incentives and group meetings were abstinent post-treatment in comparison to just 15% in the core group. The figures were 26% and 16% respectively at 12 month follow up. However, the authors do state that the results should be treated with caution until a longer follow up evaluation can be made.
Li (1984) randomly assigned 576 employees at a naval shipyard to groups with either a simple warning not to smoke or brief behavioural counselling. Results showed the counselling to produce a statistically significant reduction in smoking, with 8% abstinent in the counselling group after 11 months compared to 4% in the group that had received a warning only. Self report data was validated by measuring expired air carbon monoxide.
A videotape was employed as the treatment intervention by Sutton et al. (1984) as 61 employees were randomly assigned to either a treatment group receiving a videotape on smoking or a control group receiving a tape on seatbelts. The results were 14% abstinence in the smoking group and 0% abstinence in the seatbelt group after a 3 month follow up. Sutton (1988) extended the 1984 study by offering nicotine gum to a randomly chosen 50% of the group who had seen the smoking videotape but were still smoking at the 3 month follow up. At 12 month follow up, 22% of the 32 individuals using the gum were abstinent compared to 2% abstinence in the control group who were not offered gum.
Hymowitz et al. (1991) investigated the effect of offering an ‘enriched milieu’ of health education and promotion activities as well as worksite smoking policies would increase the long term effectiveness of a smoking cessation programme. The comparison group was offered only a group quit-smoking programme. The sustained abstinence for the full programme was 50% as compared to 44% for the group only programme. However, the results failed to achieve statistical significance and hence although in the expected direction, could not support the study hypothesis.
Windsor et al. (1988) used a 2x2 factorial design to evaluate the effectiveness of self-help smoking cessation methods. All 378 individuals in the study received either a self-help manual or cessation skills training/social support. Within each group subjects were then randomly assigned to receive a monetary incentive of $25 ($19 at 1995 prices) after 6 weeks abstinence with a further $25 after 6 months or no monetary incentive, hence creating a factorial study design. The manual only was therefore the basic intervention, to be compared to the manual plus monetary incentive, social support and social support plus monetary incentive. The results showed that the skills training/social support was significantly more successful than the manual only programme regardless of the monetary incentive suggesting that the monetary incentives were not effective. At the one year follow-up none of the participants cited the monetary incentive as the main reason for quitting. However, this may be due to the arbitrary nature of the amount as different amounts of recompense were not evaluated. It should also be noted that monetary incentives are likely to have different impacts depending upon whether they are offered as a recruitment aid or as a reward for abstinence. This is because a recruitment incentive is likely to attract smokers most committed to abstinence and hence would appear to exaggerate the effect of treatment.
Three different smoking cessation interventions were compared by Omenn et al. (1988) at the U.S. Department of Energy in Hansford, Washington. The Multiple Component Program combined behavioural skills training and stress management in an intensive three week intervention. Relapse Prevention was an eight week interactive programme focusing on the relapse period. The third intervention was a seven day self-help Minimal Treatment Program. Employees were randomised to the three programs in either group or self help format depending upon their preferred format. Results from the study showed quit rates at the end of the program for the MCP (60.8%) to exceed the RPP (36.8%), which in turn was greater than the MTP (11.8%). However, after a twelve month follow up there were no significant differences between programs. Quit rates showed the same rankings in the self help format, although the rates were consistently lower than the group format. Self report results were corroborated by cotinine saliva samples, but this was of limited value due to missing samples.
It should be noted that although smoking cessation policies may be successful in reducing the number of smokers at a worksite, the effects are not necessarily translated immediately into cost savings for the employer. In a recently published study, Wooden and Bush (1995) investigated the relationship between absence and the time since smoking cessation in a sample of 4,821 ex-smokers as part of the 1989-90 Australian Health survey. The results showed that ex-smokers were more likely to be absent from work in the first year following cessation and that the probability of absence declined with the amount of time since stopping smoking. A positive relationship between smoking cessation and work attendance was discovered, but the effects are only fully realised after a significant time period. For example, an individual who ceased smoking 20 years ago would be expected to be 4.5 times less likely to be absent from work than someone who stopped last year.
The effects of a smoking ban and employee smoking cessation programmes at the John Hopkins Hospital were investigated by Waranch et al. (1993). The hospital had 8,742 full-time employees, of which an estimated 2,000 were smokers. The four cessation programmes offered were a multi-component group, a one hour clinic, brief individual counselling and an American Lung Association manual. From a total of 262 employees participating in one of the programmes it was possible to contact 186 for the one-year follow-up (71%), of which 26 reported that they had not smoked for a year. The results were verified by CO measures for 22 of these, indicating a one year cessation rate of 8.4%. At the follow-up there was a significant decrease in the number of cigarettes smoked. The one hour clinic and the multi component groups had the highest number of quitters (21.7% and 12.5% of participants respectively). However, it is very difficult to draw firm conclusions from the results as the programmes attracted employees with significantly different characteristics (mean number of cigarettes smoked, mean number of quit attempts, mean Fagerström score and mean CO level (ppm)). Since these characteristics are likely to affect the effectiveness of the programme, this would introduce considerable biases into the results. Furthermore, the interventions are non-comparable due to the different strategies used to recruit participants, although these differences are not explained. The authors did note that the three self/minimal-help programmes required the least staff-time input but did not present a cost analysis of the interventions.
2.4.2 Other Worksite Smoking Policies
The relationship between workplace smoking policies and cigarette consumption was investigated by the California Tobacco Survey (Woodruff et al., 1990). The results found the prevalence of regular smokers to be lower in workplaces with restrictive smoking policies. Regular smokers in smoke free workplaces consumed less cigarettes than those workers in workplaces with no restrictions. This was supported by Wakefield et al. (1992) in a comparison of cigarette consumption on work and leisure days for employees at workplaces with and without smoking policies. Findings showed that employees subject to workplace smoking restrictions smoked fewer cigarettes on workdays compared to leisure days. The comparison group of employees without workplace smoking restrictions was found to have no difference between work and leisure day consumption.
Sorensen et al. (1991) investigated the effect of a non-smoking policy at the New England Telephone Company in 1986. The results showed that 21% of the 375 respondents who were smokers had stopped smoking after a period of 20 months. Eighteen per cent said they had not smoked for at least three months. Of the 375 smokers, 32 cited the policy as the reason for abstinence. The study does show that a worksite no-smoking policy can be effective in helping smokers to stop smoking, but is limited due to the lack of a comparison group.
The Group Health Cooperative, Puget Sound, was the focus of a study by Rosenstock et al. (1986). In 1986 the GHC introduced a three stage policy to prohibit smoking from its 35 facilities. The programme impact was assessed by a mail survey of a representative sample of 687 company employees. Amongst the 447 respondents, three reported having stopped smoking during or soon after the programme and 67 smokers indicated a desire to stop smoking, 61% of these reporting smoking fewer cigarettes since implementation of the smoke-free policy. Twenty-nine per cent of current smokers reported they were now smoking less and attributed the reduction to the policy. The average reduction in cigarettes smoked was two per day (p<.003). The results support the reductions in the consumption of cigarettes identified in other studies following a restrictive worksite smoking policy, but must be treated with caution for three major reasons. Firstly the study is based upon self report data which is uncorroborated. Secondly, the survey response rate of 65% may introduce some bias to the results, and thirdly there is no control group against which to compare the results. However, the authors provide some useful recommendations for the introduction of restrictive smoking policies. For a policy to be successful they suggest that:-
the new policies are introduced gradually.
provisions are made to offer opportunities for dissidents to express their feelings.
the limitations of employee influence are clearly communicated.
Borland et al. (1990) found evidence of a reduction in cigarette consumption following a worksite no-smoking policy in the Australian Public Service. The study used the 391 respondents reporting being smokers both before the policy was introduced and 5 to 6 months after its inception. Average cigarette consumption fell by 7.0 cigarettes per day, which was only partly compensated by a 0.7 increase outside working hours. Furthermore, no evidence of higher compensation amongst heavier smokers was found. The authors conclude that the heaviest smokers are those who are most likely to benefit from a worksite smoking ban as they are the ones which find cessation most difficult, and a worksite smoking ban provides a means by which cigarette consumption can be reduced. The results are also supported by Brenner and Fleischle (1994) who observed a lower average number of cigarettes smoked per day amongst employees at worksites with smoking restrictions in a cross sectional study of 931 employees in Southern Germany.
A study of the impact of a ban on smoking at the John Hopkins Children’s Center in Baltimore in 1987 was undertaken by Becker et al. (1989). The Center was a 200 bed acute care hospital with 951 full-and part-time staff. Surveys of self reported smoking behaviour, numbers of staff smoking in the visitors lounge, the numbers of cigarette butts in ashtrays inside the Center and environmental nicotine vapour concentrations were made before and after the ban. Six months after the ban was imposed the number of smokers smoking at work had decreased from 82% to 43%. The number of visitors and staff smoking in public areas which was observed to be 53% one month before the ban had decreased to 0% six months after the ban. Counts of cigarette butts dropped from an average 940 per day pre-ban to 19 per day six months post-ban. Environmental nicotine vapour also declined from 13 µg/m3 one month pre-ban to 0.51 µg/m3 six months post ban in nine of the lobby lounges, although no statistically significant change was found in the restrooms and the outpatient clinics. There was no statistically significant decline in the prevalence of smoking or the average number of cigarettes smoked per day.
Waranch at al (1993) in their study of the smoking ban and cessation help at the John Hopkins Hospital (see above) undertook a before and after survey of the smoking behaviour of the whole workforce. A questionnaire was sent to all 8,742 full time workers of which 2,877 returned a usable questionnaire both at the six month pre-ban and six month post-ban mailings. The mean number of cigarettes smoked per day decreased from 16.4 to 13.1 (p<0.0001) and the mean number smoked at work fell from 7.8 to 3.1 (p<0.0001). The results should be treated with caution as data are based on self-report with no validation. Furthermore, no attempt was made to collect information on those not responding to the questionnaire, which reduces the generalisability of the findings.
It should be noted that whilst restrictive workplace smoking policies may be successful in reducing environmental tobacco smoke at the workplace, such policies may not succeed in improving employee health as smoking outside the workplace may be unaffected or even increase (Gottlieb, 1990).
| Reference |
Cessation Policy |
Results |
| Sutton et al (1984) |
Random assignment to: (1) Videotape onseatbelts (control) (n=33) and (2) Videotape on smoking (n=28). |
Group (1), abstinence 14% and Group (2)abstinence=0%. |
| Li (1984) |
US Naval shipyard: Group 1, warning not tosmoke. Group (2), warning and counselling |
Group 1, abstinent 4% at 11 months. Group(2), abstinent 8% at 11 months. Validated by exhaled CO. |
| Klesges et al (1986) |
5 US worksites; One loan company and fourbanks. Loan Co. received basic programme. The four banks entered intocompetition, with monetary prizes for (1) Greatest uptake and (2) Greatestabstinence rate. |
Six month follow up showed 16% had stoppedsmoking in the competitive worksites, comapred to 7% in the non-competitivesite. |
| Jason et al. (1987) |
43 corporations assigned to either (1) TV andself help or (2) TV, self help and support group. |
After programme 41% abstinent in (2) and 21%in (1): After 18 months, abstinence |
| Windsor (1988) |
University of Alabama, Birmingham, U.S. 2x2factorial design: Self help or cessation training, then randomly assigned toreceive monetary incentive or not. |
At 1 year follow up, monetery reward cited asthe main reason for quitting. However, incentive may also affect recruitmentand bias the study. |
| Sutton (1988) |
Extended 1984 study by offering nicotine gumto randomly chosen 50% of the group who had seen the smoking videotape (n=79).No offer of gum to the control group (n=82). |
Those using gum achieved 22% abstinence at 12month follow up, compared to 2% abstinence in the control group. |
| Omenn et al (1988) |
US Department of Energy, Washington 402 smokers allocated to either: (1) Multiple component programme, (2)Relapse prevention or (3) Minimal Treatment Programme. Programmes available ineither group or self help format |
Quit rates after the programme were Group(1), 60.8%; Group (2), 36.8%; Group (3), 11.8%. At 12 month follow up thedifferences between programmes failed to achieve levels of statisticalsignificance. |
| Jason et al. (1989) |
38 workplaces assigned to either (1) Selfhelp, TV, support or (2) Self help, TV, support and incentives. |
Post treatment 42% abstinent in (2) and 15%abstinent in (1). At 12 month follow up, 26% in (1) and 16% in (2). |
| Bertera et al. (1990) |
DuPont, Delaware: 27 in self help, 43 in clinic based approach |
After 18 months 3/27 had stopped smoking inself help, 9/43 had stopped in clinic group. |
| Hymowitz et al. (1991) |
Six US white collar work sites randomlyassigned to either: (1) Full program of cessation help (n=131) or (2) Grouphelp only (n=121) |
Group (1), 50% abstinence, Group (2) 44%abstinence. Results failed to reach statistical significance. |
| Waranch et al (1993) |
Johns Hopkins Hospital, Baltimore. 262 smokers participated in either (1) Multicomponent programme (n=88); (2)American Lung Association manual (n=59) or Lifesign (n=59); (3) One hourclinics (n=23); (4) Brief individual counselling (n=33). |
186 were contacted at one year follow up.Highest quit rates were in Group (3), abstinence=21.7% and Group (1),abstinence=12.5%. Problem with the study is that the groups werenon-comparable, hence confounding factors may influence the results. |
2.5 Implications for Scotland
The categories of costs incurred by employers as a result of employee smoking may depend upon the characteristics of the workplace and workforce in question and the type of smoking policy to be introduced. The magnitude of the benefits yielded by a workplace smoking policy will also depend upon the prevalence of smoking in the particular workplace. For example, a policy is likely to produce larger benefits in a workplace with a higher smoking prevalence as opposed to one with low prevalence where the potential for benefits is limited. The prevalence of cigarette smoking is high in Scotland in comparison to the rest of Great Britain. The 1992 General Household Survey (HMSO, 1994) showed 34% of Scottish men and women to be smokers, this being the highest rate for the regions of Great Britain (see Table 1.1). The smoking prevalence in Scotland is considerably higher than the 26.5%(1) in the U.S., where most of the evaluation studies have been conducted. This suggests that there is are greater potential benefits to be gained from smoking policies in Scotland.
The major costs of employee smoking are summarised in Table 1.2, along with an assessment of the relevance of different items to Scottish workplaces. The impact of employee smoking upon absenteeism, the health of non-smoking colleagues and occupational health issues will be common to both U.S. and Scottish workplace smoking policies. The health insurance category is less applicable to Scotland as it will only be relevant to employers which offer their workforce health insurance packages. Higher insurance premiums for fire are more likely to be applicable if the premium is related to whether the employer has a restrictive smoking policy. The productivity category is ambiguous, as in the U.S. scenario. If workers have to stop work to smoke elsewhere then productivity may be adversely affected. Conversely, if smoking helps worker concentration then productivity may be aided, although against this must be offset possible negative effects on non-smoking workers.
Table 1.1: Prevalence of smoking by sex and standard region
| Smoking prevalence (% smoking cigarettes) |
| Region |
All persons |
Men |
Women |
| England Wales Scotland Great Britain |
28 32 34 28 |
29 32 34 29 |
27 33 34 28 |
| Source: General Household Survey 1992. HMSO 1994. |
Population base: persons aged 16 and over |
Table 1.2: Costs of employee smoking in the workplace
| Cost category |
Relevance to Scotland |
Comments |
| Absenteeism Insurance costs: Health Insurance costs: Fire Productivity Passive Smoking Occupational health Damage to plant and machinery |
Yes Yes Yes Yes Yes Yes |
If employer pays experience rated health insurance premiums If employer pays fire damage insurance premium related to workplace smoking policy. Effect may be positive or negative |
(1) US Department of Health and Human Services (1990) Healthy People 2000. The 1987 smoking prevalence has been adjusted to 1992 by the average decrease in smoking prevalence of 0.5% per annum.
2.6 Conclusions
Research shows that smokers impose significant costs on employers through higher rates of absenteeism, inflated insurance premiums, reduced productivity and higher maintenance costs. Costs are also imposed on non-smoking colleagues through the inhalation of environmental tobacco smoke.
The effectiveness of workplace smoking cessation policies has been estimated at 20-30% (Katz et al., 1990), although this will depend upon many variables such as the population being targeted, the smoking prevalence amongst this population and the particular policy in question. Studies also tend to show that workplace smoking restrictions reduce the consumption of cigarettes and policies involving competition and monetary rewards also appear to perform well.
The vast majority of the literature is from the United States, and hence the applicability to the United Kingdom is unknown. Whilst the major categories of economic costs of employee smoking are likely to be similar, their magnitude will certainly differ. The different specifications of the smoking restrictions and smoking cessation policies which have been evaluated also make it very difficult to provide meaningful comparisons of results. Caution must therefore be taken when interpreting these studies and especially when attempting to apply results to countries outside the United States.