Interventions for promoting health can be categorized in various ways. For example, one could focus on developmental stage to contrast childhood interventions with those targeted at adults. Another dimension might be "efficacy," with the aim of sorting interventions according to their likelihood of success. A related dimension would be the expected cost-effectiveness of different types of intervention investments. All of these are important facets, but we have chosen two somewhat different dimensions for emphasis here.
Timing of Prevention Efforts
The first dimension is timing of effort, as adapted from Caplan's classic three-tier definition.( n1) In his model, primary prevention includes interventions that took place before exposure to the risk factor, with the aim of preventing symptoms. Secondary prevention includes interventions that took place soon after the appearance of the earliest symptoms, with the aim of preventing the disorder from becoming severe or chronic. Tertiary prevention, Caplan's last stage, is preventive only in the narrow sense of trying to prevent long-term distress or dysfunction. It consists of rehabilitative efforts that take place after the symptoms have become chronic.
The distinction between primary and secondary prevention is especially pertinent to our analysis. The difference derives from the sequence of events beginning before exposure to an environmental pathogen and proceeding through the process by which individuals respond to these threats, either to become symptomatic or to remain well. The intervention possibilities suggested by this sequence invite a subdivision of Caplan's primary prevention stage.( n2) One subcategory of primary prevention is proactive in that it takes place before pathogens can reach vulnerable individuals. Illustrating this approach, mosquito abatement districts aim to reduce the risk of exposure to mosquito-borne diseases, such as malaria, dengue fever, West Nile fever, or equine encephalitis.
The other primary prevention approach is reactive; it tries to strengthen the individual's resistance to pathogens and hazards. Whether initiated before or after the stressor is experienced, this intervention aims to produce a coping process that will take place before symptoms appear. Illustrating this strategy are the various inoculation programs aimed at preventing diseases from flu to polio and interventions to promote improved prenatal nutrition.
These two approaches differ in their assumptions about the controllability of the pathogen or hazard. Proactive primary prevention presumes that the pathogen can be attenuated or prevented from reaching the individual. Reactive prevention presumes that the pathogen is either unavoidable or not worth avoiding but can be resisted. Subdividing primary prevention in this way yields four prevention stages. These are (a) preventing the experience of the risk factor (proactive primary), (b) preventing the development of symptoms (reactive primary), (c) preventing early symptoms from becoming chronic (secondary), and (d) mitigating the effects of chronic symptoms (tertiary).
Intervention Level
Any attempt to promote group health must consider the social structures and physical environment that surround individuals and influence their health. The fields that focus on group health, (eg, public health, community psychology, social psychiatry, preventive medicine, and health sociology) often argue for organizational interventions that can influence the environment. Just as stressors may be traced to different levels of aggregation from the micro to the macro,( n3) so interventions may be tailored to influence the most relevant level, ranging up to national or international policy. For example, if the incidence of skin cancer is related to rising exposures to ultraviolet radiation passing through holes in the ozone layer, a worldwide effort to protect the ozone layer may well be the most efficient intervention (see the Montreal Protocol on substances that deplete the ozone layer( n4)). Of course, the alternative, not mutually exclusive, is to engage in individual-level behavior modification, whether by one-to-one counseling or by mass media advisories (eg, to get people to use more sunblock or to stay out of the sun).
Intervention levels range along a continuum, from the individual person or family unit through intermediate levels, such as neighborhoods, schools, union locals, and factories, to state, national, and international governments and organizations. For simplicity, we arbitrarily divide this continuum into just two levels--micro and macro. Here, micro refers to the individual and face-to-face groups such as the family; macro refers to higher order organizations ranging up to municipal, state, federal, and world levels. Microinterventions try to change the person, for example, by one-to-one education or therapy. Macrointerventions try to change the larger physical or social environment by changes in infrastructure, social norms, or institutional policies.
Microinterventions can take place promptly with the simple consent of the individual (eg, to seek medical help to stop smoking) but may, therefore, spread slowly and unevenly. By contrast, macrointerventions require sometimes painstaking cooperation and compromise through political processes (eg, to raise taxes on tobacco products to discourage their use). Once enacted, they may affect whole populations simultaneously. Macrolevel interventions can trigger competition between constituencies with opposing values and are often controversial. For example, the Centers for Disease Control and Prevention has made the epidemic of gun-related injuries and fatalities a public health issue. But this work has engaged it in the heated national debate about gun control and has produced efforts in Congress to cut funding for studies of firearm injuries.( n5) Some in the medical and mental health fields may regard such politically tinged activities as outside of their mandate and beyond their particular expertise.( n6) On the other hand, failing to address community-level stressors means always treating their symptoms instead of their root causes.
Of course, these intervention levels interact with each other. Macrointerventions, even when decided at high levels, ultimately affect the individual. Thus, a state policy that raises tobacco taxes will be felt at the individual level each time a consumer weighs the purchase of a pack of cigarettes. Moreover, the success of individual-level interventions may well depend on environmental conditions influenced by macrointerventions. For example, individual-level advice to intravenous drug users to use clean needles will be effective to the extent that such needles are made available, perhaps as a result of a macrolevel political decision by a city council.
The different ways that mass media can be used to promote health further illustrate this distinction. If the media provide guidance and exhortation to the individual to change (eg, self-help books aimed at helping readers cope behaviorally or cognitively with death or divorce), the intervention operates at microlevel. But if the media are used to change the community (eg, television shows aimed at engendering political support for further regulation of cigarette sales), the intervention operates at the macrolevel.
Matrix
Combining stages of prevention with levels of intervention yields the four rows and two columns shown in Table 1. In this subsection we will describe each cell with illustrations from a variety of perspectives.
Interventions shown in the managing disease row (tertiary prevention) are aimed at helping patients diagnosed with a disorder, such as major depression, AIDS, or coronary heart disease. The micro disease management cell could include all of the traditional medical care and personal self-care interventions intended to help persons with a disease maintain or regain social and economic functioning.
Macrolevel disease management includes those organization-level programs designed to enable or facilitate the provision of individual-level care, including self-help groups or efforts to change group or community attitudes about health behaviors. Such programs could include training offered to employers by prepaid health plans to help managers accommodate employees with AIDS. This training would be in the interest of employers attempting to comply with legislation such as the Americans With Disabilities Act, as well as in the interest of prepaid care providers who want to reduce their care costs.
The secondary prevention category includes efforts to intervene early--after the first appearance of biological signs or symptoms. These programs are particularly important in prepaid health plans because the provider benefits from keeping the patient out of acute care. At the microlevel, this involves early detection (eg, pap smears, x-rays) and early treatment (eg, obesity or cholesterol reduction) to prevent more severe or chronic disease progression. At the macrolevel, this could include educational efforts to inform the community about early signs of disease onset (eg, the first signs of drug abuse in adolescents) to encourage early screening and initiation of treatment.
Reactive primary prevention involves efforts to minimize the risk of developing symptoms of the disease, despite exposure to the risk factor. Such interventions are ecological in the sense that they target the interaction between the environment and individuals by raising the individual's ability to cope with or resist the risk factor. Fore example, the classic individual-level approach is to raise the individual's resistance by inoculation against hepatitis. This intervention is microecological in that it is aimed at individuals or families. Macroecological interventions operate by enabling individuals to cope better with hazards in their environment. Communities can enable and encourage healthy exercise by investing in more playgrounds and recreational programs.
Proactive primary prevention involves efforts to block exposure to the risk factor. Microinterventions in this category are efforts at avoidance. Examples include the use of hypodermic needles designed to protect healthcare workers against exposure to HIV from needle sticks. This intervention is micro because it is designed for and employed by individual members of a relatively small, easily identified group (ie, those who use hypodermic needles to treat persons with blood borne diseases). It is proactive in that it reduces exposure to HIV carried by people with whom the healthcare worker must come into contact.
A macrointervention of the proactive, primary type is illustrated by the screening of the blood supply to prevent HIV or hepatitis from reaching persons in need of transfusions. Although only a small group of people will actually get transfusions, this approach is macro because the small group cannot be identified a priori, and the intervention is conducted at a high organizational level. The intervention, therefore, is designed to protect the entire population by changing the environment. Another example is the use of tuberculosis x-rays to screen food-industry workers. Although this would be a micro, secondary prevention (early detection of disease) from the perspective of the worker, it is a macro, proactive primary prevention because it eliminates a hazard from the environment of the larger population.
This eight-cell typology could be applied, with varying degrees of fit, to a wide variety of factors that influence health. In the following section, we apply it to one such factor--economic stress. Specifically, we focus on that part of social ecology known as the "economy" and the risk factors stemming from adverse employment change such as job loss. The economy is an appropriate focus for applying our framework. There are precedents for studying its effect on health and ample evidence for its impact.( n7, n8) It is especially relevant to primary prevention because it is, to varying degrees, both controllable and predictable.

Background
Adverse employment change refers both to job loss and to becoming less adequately employed (eg, poverty wages or involuntary part-time work). Together, unemployment and inadequate employment can be termed underemployment.( n9) The risk of underemployment varies with both personal (eg, ability, training) and ecological characteristics (eg, the local economic climate). The latter is popularly indexed by the unemployment rate, which is the ratio of those not working but wanting and actively seeking work, to the labor force (sum of employed and unemployed).
Note that employment status differs conceptually from two other health-related economic stress constructs--class and poverty. Social class refers to one's relative socioeconomic position and is usually indexed by such stable characteristics as years of education or usual occupational status. These variables may be considered causally before employment change, in the sense that they influence, but do not completely govern, the risk of underemployment.
On the other hand, poverty refers to a hardship level of income, which can be regarded as an outcome of both enduring class variables and momentary employment status changes, among other factors. Both low income and income inequality within a community have been linked to poor health.( n10, n11) Employment status change is viewed here as a more proximal precipitating factor superimposed on more distal predisposing factors, such as individual class and economic climate. Job loss or downgrading of employment offers a better focus than the others for intervention. Employment status is more malleable than class status, which is usually fixed by early adulthood as a function of parental class and educational attainment. It also offers a more politically feasible lever for influencing income than do options that rely on direct income redistribution.
Studies of adverse employment change and health have focused primarily on unemployment, operationalized as either personal job loss in individual-level studies or variations in the unemployment rate in aggregate-level studies. The literature on the health effects of unemployment has been reviewed many times in the past 2 decades.( n7, n8, n12-n15) The findings reported in this literature have converged on a number of conclusions. First, although health can influence employment status (ie, reverse causation), studies using the best longitudinal designs confirm that job loss has adverse effects on health, controlling for preexisting health status. Unfortunately, because of the variety of health effects and the complex causal processes that produce them, there is no consensus on a single estimate of the amount of harm that unemployment can cause. Second, unemployment is associated with not just one health outcome but, rather, with many kinds of health outcomes. Affective symptoms, such as depression, are commonly reported after job loss.( n16, n17) These symptoms often rise above garden-variety distress to reach diagnostic case levels. One study of unemployed men who were normal at the time of job loss reported that 14% developed clinical psychiatric disorders; a further 17% reached borderline levels over 6 months of unemployment.( n18) Another commonly reported effect of unemployment is alcohol misuse.( n19) In one study, the researchers found that becoming unemployed was associated with a sixfold increase in the risk of clinical alcohol abuse.( n20) Unemployment has also been linked to violence to others( n21-n23) and to oneself in the form of suicide.( n24)
Unemployment as an acute stressor has more often been linked with such maladaptive behaviors or psychological outcomes as those just identified than with more organic or slow-developing physical health outcomes. Some researchers have claimed links to specific health outcomes, such as heart disease.( n25-n28) More often, unemployment is linked to nonspecific physical illness or mortality.( n29-n34)
Although the literature on becoming inadequately employed is more recent and sparse, some evidence exists for harmful health effects paralleling those of job loss.( n35) Thus, an intervention that reduces exposure to adverse employment change would be likely to lower the risk of several different health problems simultaneously.
Third, the relationship between adverse employment change and health is complex and not simply a matter of a direct effect of job loss on the unemployed worker's health. From a population perspective, an economic recession not only raises the risk of job loss, but may also degrade the quality of work for those who retain their jobs. With fewer job options and more competition for scarce jobs, workers may feel obliged to stay in stressful or economically inadequate jobs. Thus, the health costs of a recession will be paid partly, perhaps even largely, by those still working( n36) or not in the labor force.( n37) Moreover, there may also be interaction effects in which higher unemployment rates in the community exacerbate the harmful effects of personal unemployment and reduce the chances of timely reemployment.
Unemployed workers with less education and fewer job prospects may be particularly distressed in economic downturns, when fewer openings are available.( n38) Of course, interactions may also entail moderating processes. Some researchers have reported little or no adverse effect from unemployment, attributing these negative findings to such buffering factors as the worker's capacity for proactive coping( n39) and the support of the surrounding environment.( n40) The duration of unemployment, like any other stressor, might also effect many aspects of the individual's life.
Research findings suggest the existence of indirect spillover effects from workers to spouses( n41, n42) and children.( n43, n44) This complex interplay between environmental factors and individual experiences involving direct, interactive, and indirect causal pathways makes it difficult (even impossible, given the present state of knowledge) to account fully for the health costs of adverse employment change.
Applying the Typology to Economic Stressors
Although the epidemiologic literature connecting adverse employment change to health is substantial, the reports of carefully evaluated interventions in this field are relatively few. The research is too sparse to anoint a single best practice with anything like a cost-benefit efficiency rating. Instead, it may be helpful to catalog the various types of interventions that have been attempted or suggested and warrant further consideration. We illustrate the intervention typology with examples from the employment stress field. The matrix resulting from combining four stages of prevention with two levels of analysis is summarized in Table 2.
The lines forming Table 2 may imply clearer distinctions between categories than actually exist. The blurring between macro- and microlevels can occur at each of the prevention stages and is related to the fact that the same macrolevel interventions can operate in two different ways. First, macrointerventions can operate by reinforcing individual-level interventions. For example, advice to individuals to control their driving speed can be reinforced by the threat of punishment for violating state speeding laws. The resulting change in the individual's driving style helps reduce the risk of accident through the driver's loss of control. Second, macrointerventions can operate to modify the environment to promote health without any response on the part of individuals. For example, the same highway speed limit can reduce the average speed of other drivers; which, in turn, should reduce the risk of chain-reaction collisions to any individual driver, regardless of his or her speed.
Microlevel Interventions
Medical and mental healthcare providers are accustomed to working with individuals alone or in small family or therapy groups. Thus, interventions in the micro column may seem most familiar. Of these, the most traditional appear in the medical care cell of the disease management row, where helpers wait for individuals to experience not just the stressor (eg, job loss) but the actual symptoms of disorder before providing assistance. Each disorder (eg, alcohol abuse, depression) that can arise from employment stress can also be caused by other hazards, and each can be handled by an appropriate treatment regime independent of the initial cause. It is always more effective to know and respond to underlying causes, and economic considerations may be especially salient when employment is an important causal or moderating factor.
An example of a microlevel intervention intended to manage illness that could otherwise be made worse by economic insecurity would be programs designed to help persons with chronic physical or behavioral problems remain employed. It has been reported that persons with such illnesses are more likely to leave the labor force and apply for disability programs when the labor market contracts than when it is stable or expansive. Employers are believed to view persons with chronic illness to be relatively inefficient workers. These workers may take comparatively more sick days and use more health services than their colleagues. They are, however, hired and kept as long as more productive workers are not available or are available only for higher wages. When the demand for labor slackens, employers are less likely to hire chronically ill persons and more likely to lay them off. It appears that some of these persons then use their illness to justify claims for disability income.( n45) A public or private insurer must pay for medical care if the chronically disabled person successfully obtains workers' compensation. These costs could be assigned to the employer and the prepaid healthcare provider if the worker is laid off. This assignment of costs might motivate employers and providers to offer programs that help chronically ill persons manage their disease and stay employed.( n46)
The micro, early intervention cell involves efforts to provide short-term counseling soon after the first appearance of symptoms. The standard approach consists of crisis intervention designed to identify the hazard that precipitated the crisis and to avoid a maladaptive resolution of the problem.( n1) This widely used counseling approach has been recommended for dealing with adverse reactions to job loss,( n47) but it is equally suitable for dealing with other types of losses.
Microinterventions do not need to wait for symptoms after such stressors as job loss, but they could prevent illness by making the ecology of workers more salutary through readying people to respond effectively to their environmental challenges, for example. Ecological coping interventions are reactive primary preventive efforts mounted immediately after or even in anticipation of the stressor. These include individual or small group programs that try to prepare people to cope with the stress of such events as job dislocation. One approach for improving the general resilience of workers to all kinds of stress is the hardiness-training program.( n48) This technique involves 10 weekly 1 1/2-hour sessions and can be applied before the stressor is experienced.
In one experiment, utility company managers were randomly assigned to hardiness training, relaxation/meditation, or a passive listening control group. The hardiness trainees reported significantly more hardiness and job satisfaction and significantly less strain and self-reported illness.( n48) Whether such training could inoculate individuals against the stressful effects of adverse job change remains to be evaluated.
More narrowly targeted at the stress induced by unemployment are programs designed to help laid-off workers. Such out-placement programs typically enhance job-seeking skills and offer counseling to reduce the likelihood of adverse psychological or physiological outcomes. One group approach based on cognitive behavioral therapy principles and designed for long-term unemployed (more than 1 year) is delivered in 7 weekly 3-hour sessions.( n49) Compared with a social-support control group, this intervention not only significantly reduced psychiatric symptoms, but also increased the likelihood of finding full-time jobs at 4-month follow-up (34% versus 13%). Programs of this type are not common, but they have demonstrated their efficacy in both the near and longer term. For example, the JOBS program, a group program delivered in 5 sessions of 4 hours each, shows that although targeted at recently unemployed workers (less than 13 weeks), it has a lasting inoculation effect for those who find reemployment, then face the setback of losing work again.( n50) Compared with their counterparts in the control group, JOBS respondents who temporarily found and then lost a job showed significantly lower depressive symptoms.
Such programs may be offered to laid-off workers as a benefit negotiated by unions or because an employer feels obligated to do so. They may also be financed by philanthropic organizations. They would probably be more common if laid-off persons kept their employer-provided health insurance or if the care was prepaid. This circumstance could provide an economic incentive for care providers to offer such packages to laid-off persons because the risk of acute illness resulting from job loss would be reduced.
However, such programs face problems other than funding. First, efforts to offer counseling to laid-off workers are often met with resistance, perhaps because of the double stigma of being unemployed and being unable to cope with it. In one program for workers at a closing auto plant, only 1% sought help for personal and emotional stress and just 5% for alcohol and substance abuse.( n51) In an evaluation of the JOBS program, just 54% of those screened and invited for participation took advantage of the intervention.( n50) On the other hand, evidence has also been advanced that those most in need of such interventions self-select to take part( n52) and that the intervention may work best for those at highest risk.( n53)
A second kind of challenge comes from the societal perspective. If the local job market has too few openings for the current pool of unemployed workers, such psychotherapeutic efforts may only determine which of the unemployed get the few available jobs without substantially changing the total number who remain without work.( n54) Some evidence suggests that such programs' success will vary to the extent that the surrounding labor market provides job opportunities.( n55)
Finally, microlevel interventions in the proactive primary prevention stage involve efforts to avoid adverse employment change rather than to cope with it. Such efforts can include family decisions about which schools children will attend and what courses of study they will pursue as they relate to the future likelihood of employment and economic security. These activities are so distal from the expertise and daily responsibilities of healthcare providers as to seem beyond the purview of the medical field. Nevertheless, it has been argued that the best way for individuals (microlevel) to avoid stressful employment crises (proactive primary prevention) is to prepare people while they are in school for the jobs of tomorrow.( n56) Proposals for and evaluations of educational reforms constitute macrolevel interventions that could complement and enable such individual-level efforts. However, the literature from this domain is both too large and too tangential to consider in this article.
Early failure in the job market (eg, unemployment at age 21) has been traced in long-term panel studies to a variety of childhood indicators of personal and social capital. At ages 3 to 5 years, these include low family occupational status, low intelligence, and a difficult temperament. At ages 7 to 9 years, these include living in a single-parent family and family conflict. By the age of 15 years, contributing factors include the lack of a school certificate, low reading achievement, and low school involvement.( n57)
From a societal perspective, individual-level efforts to use family and educational resources to help one's children avoid underemployment may prove uneven and help sustain social inequities. Parents who can invest more money and make better informed decisions about their child's training can help pass along their higher occupational status and steadier employment. Parents with few such resources may instead experience the intergenerational transfer of underemployment.
National and state efforts to equalize educational opportunities are well justified on equity grounds and would be likely to have the indirect effect of promoting health through the prevention of unemployment. But the long-term health benefits of such educational interventions, although difficult to trace, appear in many measured health differentials by middle age.( n58) For obvious political reasons, the task of such educational reform, whatever its health benefits, is unlikely to be assigned to the already complicated health sector.
Macrolevel Interventions
Hopes for sparing people the stress of job loss and underemployment typically turn to the management of the larger economy. Such macrolevel, proactive primary prevention would have the great virtue of being environmental, that is, applying to all without requiring individual efforts that are likely to be inadequate for many. The fiscal and monetary actions necessary to carry out such an ambitious goal would have to take place at the national level.
Guaranteed employment was written into the constitution of at least one nation:
Article 4: Citizens of the USSR have the right to work (that is, to guaranteed employment and pay in accordance with the quantity and quality of their work, and not below state-established minimum), including the right to choose their trade or profession, type of job, and work in accordance with their inclinations, abilities, training and education, with due account of the needs of society.

That the USSR no longer exists may be partially attributed by some to the economic policies that pursued this utopian goal. The United States' counterpart is found in the Full Employment and Balanced Growth Act of 1978, which declared as national goals the promotion of full employment and the right to the opportunity for paid employment. However, it stopped short of making the federal government the employer of last resort.
By default, efforts to reduce unemployment or inadequate employment are incremental or local. One incremental approach to lower unemployment is work sharing or having the employed work fewer hours so that more people can have some work. This was one of the arguments for 1998 French legislation reducing the statutory work week from 39 to 35 hours. Whether this will succeed in decreasing unemployment remains to be seen, but even if it does, it may have the unintended effect of increasing underemployment in the form of involuntary part-time work. Incremental approaches to decrease underemployment are possible through minimum-wage legislation. Currently the US minimum wage is $5.15 per hour, but a number of states and cities have raised their local minimum wage in an effort to assure workers of a living wage. However, critics of this approach argue that raising the minimum wage (in an effort to decrease inadequate employment) will raise unemployment as some employers are forced to lay off workers whose marginal rate of return falls below the new minimum. In short, economic policies to assure full and adequate employment are not clear-cut.
Outside of the political process, workers have found other ways to protect themselves from job loss. These include contracts negotiated between unions and employers that preclude layoff except in extreme economic circumstances and allow it only for workers who lack seniority. These contracts, which have been negotiated for coal miners and some auto workers, are "environmental," as opposed to ecological, because they are not designed to change the behavior of those for whom they were designed. On the contrary, they are intended to make it possible for the beneficiary to maintain as much of his or her regular behavior as possible. Such macroenvironmental interventions for reducing adverse job change derive from economic rather than public health policy, and the techniques for managing the economy clearly fall outside the area of behavioral medicine.
Macrolevel interventions to make workers hardier in the face of insecure or lost employment are less obvious than those at the microlevel. One macrointervention could involve mailings to workers in contracting industries alerting them to the risks inherent in layoff and to services that are available. It has been reported, for example, that the stress of being unemployed distracts individuals from their regular health maintenance behaviors.( n60)
Telephone calls to the families of the unemployed or to those employed in contracting industries reminding them of mammogram, prenatal care, and other preventive behaviors could counteract, in part, the distraction of coping with unemployment and its sequelae.( n61) Another type of macroecological intervention could be to use union, church, or neighborhood groups to organize social support networks for those who are laid off. Such interventions have been reported to help pregnant women, for example, cope with stress and reduce the possibility of low birthweight babies.( n62)
A whole class of government programs to help the unemployed cope with job loss also falls into this category of macrolevel, reactive primary prevention. Together, these massive programs constitute the social safety net and fall in the political more than in the health-services realm. One of the most familiar is unemployment insurance, which was enacted in 1935 to provide partial compensation to workers who lose their jobs. In recent decades, however, the proportion of American unemployed workers who receive such compensation has fallen sharply because of factors such as demographic shifts and the decline in union membership.( n63)
A second strand in the safety net is federal financing of training and job-search services for the unemployed mandated by the Job Training Partnership Act (JTPA) passed in 1982.( n64) Although it does not provide health or mental health counseling for the unemployed, the JTPA does provide hundreds of millions of dollars per year in programs to help workers get reemployed.
The third strand in the federal safety net is the Worker Adjustment and Retraining Notification Act (WARN) that requires prenotification of workers before their plants shut down. By giving workers more time to prepare for and cope with their job loss, such advance notice appears to decrease the duration of the unemployment episode.( n63)
Taken together, do such programs reduce the health costs of unemployment? Indirect positive evidence comes from research showing that societies with more generous safety nets and cultural attitudes experience less adverse reactions to job loss. In one study, for example, researchers reported that college-educated Dutch youth showed no effect of employment status on mental health symptoms.( n40) This was attributed not only to the generous unemployment benefits they received but also to two cultural factors. These are the institutionalization of unpaid work (blurring the status distinction between them) and the "normalization" or public acceptance of unemployment in the Netherlands.( n40) One implication of this analysis is that macrointerventions of the reactive, primary prevention type might consist of both greater financial investments in the social safety net and changes in community attitudes that make the unemployed feel less unworthy.( n65)
Macro secondary prevention stage interventions dealing with economic stress are scarce in practice, but they can be sketched in principle. They might facilitate, at the community level, the early identification and treatment of people reacting symptomatically to adverse employment change. For example, in towns experiencing sharp rises in unemployment (eg, as a result of the closing of a major plant), health-service providers could warn relevant institutions to be on the alert for stress reactions. Paternal job and income loss appear to have adverse spillover effects on children,( n66) and these effects might well be detected early by alert teachers and school counselors. Thus, advisories could be sent to the local school system along with offers of outreach teams of consultants from the behavioral medicine sector. Although different in detail, such community-level economic crisis interventions might parallel the kinds of emergency responses mobilized immediately after disasters such as tornadoes or school shootings.
Macro tertiary prevention programs aim at facilitating the management of symptoms associated with unemployment. One example is federal legislation assuring the portability of health insurance. The Consolidated Omnibus Budget Reconciliation Act of 1985 (COBRA) mandates that employers who provide group insurance must offer the option of continued coverage to their employees who lose or leave their jobs. However, because the former employee must pay a larger premium for this continued coverage without the employer's subsidy, many unemployed workers and their families go without health insurance. This only compounds the problem because many employed people work for employers who do not provide group insurance. A better accounting of the costs incurred by the public resulting from the health effects of layoff might make it economically rational to prevent illness by subsidizing the employees' purchase of health insurance.

Professionals with a reasonable claim to reducing illness in populations will naturally want to be part of the expanding effort to reduce illness among groups covered under prepaid, group health plans. The motivation is partly economic. It is clear in the case of behavioral medicine, for example, that the opportunities to practice in traditional fee-for-service settings are not growing as rapidly as those in managed care.
We believe that the knowledge base of behavioral medicine is at the intersection of psychology and risk-factor epidemiology. Mastering this intersection well enough to help individuals manage the behaviors that put them at risk of contracting new or aggravated illness is a considerable challenge. We are skeptical about whether the field can be further expanded to include interventions that manage the environmental pathogens or risk factors to which various populations are exposed. We suggest, therefore, that behavioral medicine should focus its efforts on the four cells in the micro column and the lower two cells of macro column of Table 1.
The temptation for "mission creep" into the environmental manipulation and ecological enabling cells should be resisted. Those cells should be left to experts in public health, environmental design, health economics, and organizational behavior. Although gifted individuals may be able to function effectively across all of these fields, we doubt that professionals can be routinely trained to do so.
The assumption that professionals cannot be routinely trained to function effectively across the cells in Table 1 implies that the admonition against mission creep is symmetric. Professionals trained in public health, environmental design, health economics, and organizational behavior should not be designing programs that assume an understanding of psychology, behavior change, or clinical intervention.
Although mission creep is dangerous, it may be less threatening to the health of the population than "mission shrink." This problem, which also arises from economic self-interest, is one in which professionals ignore the contribution of other groups to risk management and practice as if their skills are all that are needed to further population health. Mission shrink can lead to interventions with unintended consequences that could have been averted if other experts were consulted.
An example of mission shrink might arise in the case of behavioral medicine and unemployment. The scenario begins when a local economy is slowed by the loss of a basic industry and more than the usual number of persons are laid off. Some laid-off persons keep their prepaid health plans by paying the premiums themselves, whereas others become eligible for public or private (eg, union-sponsored) programs typically provided through prepaid plans. The managers of organizations that provide healthcare are aware that families in which a job has been lost are at elevated risks for illness and using health services. These managers decide to introduce programs to reduce illness in the affected population in the hope that such interventions would be cheaper than providing medical care after illness occurs or is made worse by the stress of unemployment. Persons trained in behavioral medicine are hired to design the programs. These individuals plan programs intended to inoculate individuals against the adversity of job loss and give them skills that improve their job-finding skills.( n49)
These seemingly salutary circumstances could have adverse effects unanticipated by behavioral medicine programmers. One arises when the local job market has significantly fewer openings than needed to absorb job losers. Behavioral medicine programs under such circumstances may influence who among the unemployed get the available jobs, but the total number who remain jobless in the labor market stays the same.( n54) The prevalence of illness induced by unemployment in the population, therefore, is unaffected by the behavioral medicine intervention. The illness is simply redistributed.
A related effect is that behavioral medicine approaches may reduce support for public policies that inhibit layoffs (eg, progressive unemployment compensation taxes on employers such that those with higher levels of layoff pay higher rates). If the association between illness and layoff is viewed as arising primarily from the worker's failure to cope without symptoms, which is the likely view in behavioral medicine, the implication is that the stressor is unavoidable. Unemployment may be tolerated as a way to avoid presumably worse circumstances (eg, runaway inflation), but it is rarely unavoidable. Indeed, those who provide healthcare to the unemployed may prevent more pain and suffering by joining efforts to manage the local economy than by counseling job losers.( n67)
A more ominous aspect of mission shrink arises from a convergence of interests among those in behavioral medicine, the institutions that heighten risk, and the many who are at risk. A simple example of this circumstance might be mission shrink in the reduction of smoking. Managed care may appropriately motivate some in behavioral medicine to seek employment as experts at reducing the cost of smoking-related illness in populations. It could be argued, however, that intervening directly with individuals to influence their reactions to cigarettes is no more effective at reducing smoking in the population than raising taxes on tobacco or constraining the opportunities to smoke in public places.
The tobacco industry clearly prefers behavioral medicine to tax and regulatory approaches. Smokers as a population may similarly prefer the behavioral-medicine approach. Those who want to quit would want help from behavioral medicine and want their insurance to pay for it. Those who want to smoke would not want the price of tobacco to increase or the opportunity to use it to be constrained. They, therefore, would also prefer that their insurers adopt behavioral-medicine approaches rather than advocate more macroapproaches. The interests of behavioral medicine, if shaped by mission shrink, the tobacco industry, and many smokers, therefore converge to support individual-level programs in managed care. The interests of the remainder of the population, however, might be better served if prepaid health plans had their lobbyists, health economists, and attorneys advocate for salutary taxes and regulatory policies.
The example of job loss implies a similar outcome from mission shrink. The firms that lay workers off prefer behavioral medicine approaches because macroapproaches imply greater government or union involvement in personnel decisions. Workers who have been laid off would prefer programs that offer them help rather than approaches that focus on protecting the job security of the employed. Although programs that do both are possible, they are particularly objectionable to industry. The path of least political resistance, therefore, is to support the behavioral-medicine approach. How do we avoid the perils of mission creep and shrink? One way is for the consumers of risk-reduction expertise to be aware of both phenomena. These consumers are increasingly likely to be the managers of prepaid group health plans. It is in their economic interest to understand the limits of each profession's abilities, and these interests would be served by understanding our typologies.
Another way to avoid mission creep and shrink is to organize, rather than exploit, the intersection of skills among the professions that claim to reduce the risk of illness. This organizing might require cooperation that is not in the short-term interest of some groups. Researchers supported by universities, foundations, and the government may therefore have to begin the task of exploring the intersections and offering models of cooperation and training that avoid both mission creep and shrink. We hope that the typology we have offered in this article lends impetus to the effort.