Background

Employer interest in worksite smoking cessation reflects two reasonably independent considerations: the direct adverse health and productivity effects of cigarette smoking on smokers, and the negative consequences of environmental tobacco smoke (ETS) for nonsmokers. While the latter were first suggested two decades ago,9 the evidence indicting ETS as a serious cause of disease is exclusively a phenomenon of the past decade.10-12 Major new contributions to this body of evidence have emerged only in the past year.13-14 The health consequences of ETS have motivated much of the public concern about smoking in public places and have prompted increasing numbers of employers to restrict or ban smoking in the workplace.1 Employers have acted both from altruistic concern for employees' welfare and from a heightened sensitivity to legal considerations (avoidance of lawsuits and,in several jurisdictions, compliance with new clean indoor air laws).

Worksite smoking cessation programs are a component of many of the new employment-based smoking restriction policies. However, the origin of most existing worksite smoking cessation programs, and quite possibly the driving force in contemporary consideration of their adoption, is the belief that smoking employees impose substantial costs on their employers due to the direct effects of smoking on health and productivity. Fostered initially by two analyses that developed this conclusion almost a decade ago,15,16 the belief derives from a well-documented body of evidence demonstrating that, at each age, smokers use more medical care and experience more sick days than do nonsmokers.1,17 The logical conclusion, encouraged by the authors of the aforementioned studies on the employment costs of smoking, is that successful smoking cessation programs will return an economic dividend to employers, while improving the health of their workers. Assisting employees in quitting smoking appears, therefore, to present employers with an opportunity to attain the corporate ideal: doing well by doing good.

Contemporaneous with the growth in smoking cessation programs has been an emerging "worksite wellness" movement, with behavior-change programming covering a myriad of health practices. A recent national survey found that two-thirds of worksites with 50 or more employees offered one or more health promotion activities, with the proportion approaching 90 percentof worksites with more than 750 employees. Smoking cessation is an integral component of many corporate health promotion programs and was the leading individual activity provided by the surveyed worksites.18

Corporate interest in health promotion reflects a more general society-wide appreciation of the role of individual behavior in health and, increasingly, in health care costs, an interest in the U.S. typically dated from the 1979 publication of the Surgeon General's Report on Health Promotion and Disease Prevention .19 A growing body of evidence indicts a variety of unhealthy behaviors, ranging from substance abuse to lack of exercise, as a cause of higher health care costs and greater job absenteeism.20,21 In a business environment plagued by high and rapidly escalating health care costs, and an increasingly competitive world marketplace, reports of the costs of unhealthy behaviors, and the prospects of taming them, naturally have attracted considerable attention. Thus, while surveys find a variety of factors motivating adoption of worksite health promotion programs, the expectation of a positive economic yieldinevitably ranks high on the list.5,6 Furthermore, many observers believe that financial return is a particularly important consideration to firms that have not as yet adopted health promotion programs, as compared with those having programs in operation; survey evidence is consistent with this view.6

The logic underlying worksite health promotion has considerable intuitive appeal: employees engaging in unhealthy behaviors have documented excess health care and other costs; health promotion interventions are typically inexpensive per participant and can produce behavior change; consequently, individually-low-cost interventions can change behavior and thereby yield economic dividends in the form of reduced health care costs, absenteeism, etc.4 Following this logic, the most ardent supporters tout worksite health promotion (WHP) as a (if not "the") major solution to the rapid and persistent growth in corporate health care costs, and as an investment yielding broader returns as well, including reductions in absenteeism and increases in productivity.22,23

As appealing as this argument sounds, it is not free of logical flaws, and to date it has not been supported by adequate empirical evidence. Logically, the argument fails to weigh the amount of behavior change resulting from interventions against the total cost of the interventions. Furthermore, it does not address the fact that changed behavior patterns may not produce health (and other) outcomes identical to those of people who have never engaged in the unhealthy behaviors. The argument ignoresthe fact (and role) of employee turnover: unless all benefits obtain immediately (which no one suggests), some proportion of later benefits will be "exported" outside of the firm when program participants leave the firm. Finally, as commonly presented, the argument ignores the potential economic "downside" to the firm of WHP success: if health promotion prevents premature death, it may increase the population of retirees to whom pensions must be paid and supplemental health insurance provided. (Of course, as with benefits, some of this later cost will be "exported" due to employees leaving the firm.)

To date, research on the health and economic implications of WHP programs is limited in quantity and generally poor in quality.8 Most of the published commentary is found in the trade literature, where evaluation of the subject reflects "hearsay" and "wish bias", rather than sound analysis. In the scholarly literature, reports on quasi-experiments in large corporations yield useful information,24-26 but such studies are limited in number, subject to serious research design problems,8,27 and inherently unsuited to evaluate complicated questions pertaining to health and economic implications, particularly those produced several years following introduction of a WHP program.4 Limitations of these quasi-experiments include their inability to control for all of the "noise" in the system (a problem that is exacerbated over time), difficulties in identifying truly representative control groups, and the expense and time required to undertake the studies and derive and interpret findings. In particular, issues related to the longer term effects of WHP programs cannot be addressed by such analysis absent the passage of the "term" in question. If this is a matter of years, the answer to the question (were it possible to derive it from such a study, which is most unlikely) would come too late to be useful.

Simulation: An Alternative Approach to the Problem

Given the limitations of the dominant research approach to assessing the effects of WHP programs, we have initiated development of a computer model intended to simulate the workings of WHP programs in the corporate setting over a period ranging from one to 50 years. The attraction of a simulation model is that it permits rapid, inexpensive investigation of a wide variety of questions pertaining to the effects of an intervention like a worksite smoking cessation program. Given a properly constructed model and valid parameter values, investigators can examine any programmed question in a matter of minutes. The simulation lacks all of the sources of "noise" that contaminate quasi experiments, and it permits virtually instant appraisal of intervention effects occurring "years" in the future. Furthermore, it permits investigators to vary parameter valueswith a push of a button; an experiment, in contrast, is locked into its parameters for its duration. As such, investigators can perform sensitivity analyses, whereas analysts of quasi-experiments are not generally afforded that opportunity.

To illustrate, a simulation model could determine with ease whether, for example, a 20 percent smoking cessation rate significantly reduces the net economic benefit of a worksite program compared with a 25 percent rate. An experiment, in contrast, is "stuck" with the rate actually experienced, and must await passage of a year or two to determine it. The simulation model can examine the impact of the 20 percent rate (or both rates, for that matter) at the end of one year or five years or 25 years, again within minutes. The experiment can appraise these impacts, if at all, only after the real-time passage of the number of years of interest.

While the strengths of computer simulation are speed, low cost, and flexibility, the potential weakness is the failure of the model to capture sufficient elements of the "real-world" phenomenon of interest to produce valid and reliable results. Thus the critical words in the first paragraph in this sectionare, "Given a properly constructed model and valid parameter values." Smoking cessation represents an unusually desirable case for simulation modeling precisely because the epidemiologic, cessation program, and economic data are so well-developed.8 Excellent data are available on the impacts of smoking, both health and economic, the behavioral effectiveness of cessation programs, and the health consequences of cessation.1,29 For this reason and others (notably, the importance of the problem and the expertise of the principal investigator), we have selected smoking cessation as the first health promotion activity to investigate in this simulation model framework.

Given the potential virtues of the simulation approach to examining worksite health promotion, it is somewhat surprising to find a dearth of previous simulation-based research. In part, this likely reflects the relative novelty of the WHP issue; in part, it represents the lack of substantial academic interest in the subject: the vast majority of concern with this issue, and formal analytical work, have come from the business community, where observational studies and quasi experimentation are familiar and comfortable, while computer simulation may not be.

Several health promotion "entrepreneurs" have developed simulation models, but the models familiar to the principal investigator are exceedingly simple and fail reliability and validity tests. Few if any of their results are published. To our knowledge, only one well-developed simulation has been reported in the published literature, namely the work of Leviton.30 This model presents a useful look at the short-term health and economic implications of several WHP interventions. The model is structured to examine effects up to five years only, and thus cannot consider the long-term consequences of interventions (e.g., for future pension liabilities and retiree health benefits); nor does it deal with employee turnover as we propose to model it. Leviton's model is also restricted to examination of the most basic impacts of WHP on the firm (as is our prototype model; inclusion of additional impacts is a significant objective of the research, as described in the next section of this proposal). Nevertheless, the Leviton simulation represents the published state of the art, and has served as a source of considerable insight for us as we have conceptualized the work that we wish to undertake.

Alternative Interests

As noted above, the objective of this research is to develop useful new knowledge of the health, economic, and workforce demographic implications of WHP programs, from both the corporate and social perspectives. As issues, the health and economic implications to the firm are familiar to all parties interested in worksite health promotion (although the long-term economic implications generally are not4). In contrast, to our knowledge the issue of demographic implications has never been previously raised in discussions of WHP, in part because the orientation of all previous analyses has been exclusively the interests of the firm.

By demographic implications, we refer to the effects of health promotion programs on the age structure of the workforce, and the size and composition of the population of employees. Given increasing concern about the costs of retirees, andprojections of the changing composition of the entering workforce in the next century, the demographic implications of worksite programs ultimately may prove to be as interesting and important as the short-term estimates of health and economic consequences. Without contemplating the effects of alterations in health behaviors, one demographer has projected that, by the year 2030 (when fully a quarter of the U.S. population will be over 60 years old), there will be two million fewer workers age 16 to 24 than there are today, and 25 million more age 35 to 54.33 Furthermore, demographic characteristics of new entrants to the labor force, other than age, will be changing radically. Nonwhite workers are expected to represent twice as many new entrants as their current share of the workforce and immigrants treble their current share. White women are projected in one study to represent a substantially larger share of new entrants than are white males.34

The exclusive focus of previous WHP research on the immediate interests of the firm has also precluded attention to the broader societal health and economic implications of worksite programs, i.e., the consequences that transcend the interests of the firm. Because, as noted above, some of the benefits of WHP programs will be "exported" outside of the firm as workers retire or migrate to other jobs, we intend to investigate the social health and economic consequences of WHP programs, in addition to their direct implications for the firm.

Consider, for example, that a worksite smoking cessationprogram may benefit the firm by reducing smoking among employees and its sequelae; but it will also benefit the community as a whole in that some employees who are helped to quit smoking will leave the firm but remain members of the community. Indeed, very preliminary calculations suggest that over a 20-year period, perhaps as many as half of a firm's employees who successfully quit smoking may have left the firm, taking their health and associated benefits with them. Through use of our model, we will be able to compare and contrast firm-specific and community-wide consequences of smoking cessation. As discussed in the concluding section of this Project Description, this comparison may have direct utility to public policy makers interested in facilitating private sector initiatives in health promotion.