Essential to Understand the Relationship Between Mental Illness and Work
I agree with the exhortation that we need to think of better ways to get at this problem, both as researchers and policy-makers. But this is a focus on the back end - after the period of disability leave. I want to focus here first on key questions that remain about the front end - the onset, or "nature" of mental illness in the workplace. While we have made great advances in understanding the prevalence of mental illness in the workplace during the past two decades, we remain quite limited in our ability to sort out the joint causal relationships between illness and employment. It is clear that one cannot treat mental illness as exogenous in the context of the workplace, and we have made strides in developing better estimates of employment and earnings losses subsequent to episodes of mental illness. But the ability to do this relies on relatively weak information to help us identify causal effects. It is safe to say that our analytical ability to address the complicated causal pathways between mental illness and work substantially exceeds the fertility of current data for yielding new insights into these relationships. This is a view, and a frustration, shared by many researchers in the area (e.g., Wells 2002).
Closing this gap is what I believe should be the first priority of the collective research community. If we hope to understand the complicated relationship between mental illness and work, we simply need more and better observations of people as they live and experience both. Because this relationship is complicated, there really is no substitute for panel data to provide new insight into these relationships. So, I add to the list of priorities, begun by Dewa et al., the collection of a representative panel data set containing both diagnostic information and employment and earnings information. In North America there are very few data sets of this sort. Several longitudinal data sets contain depression scales, but diagnostic information on other psychiatric diseases is rare. The National Comorbidity Survey (NCS) will soon release a follow-up interview of its original cohort. While the NCS contains rich diagnostic information, the "panel" consists of only two interviews, a decade apart. I know of no data set that really is up to the challenge of sorting this out.
Let me turn next to a different issue. As we advance our understanding of the labour market consequences of mental illness, we are implicitly asking the question: What if it were not for mental illness? But it is not obvious exactly what we mean by this question. Should we think about preventing mental illness, so that we might view the productivity and earnings of comparable healthy workers as what we would expect for the mentally ill in the absence of illness? Or should we think about "curing" mental illness, expecting that the mentally ill will regain their pre-morbid levels of productivity and earnings? Alternatively, perhaps we should think about treating the mentally ill - thereby limiting disability and attempting to balance work and illness. In many cases, I think this last question is the most relevant and important to ask. Yet, it is the question we are least equipped to answer. We know precious little about the employment and productivity benefits of treatment. In community-based surveys, little is known about treatment. What we know comes from a few clinical trials of pharmaceuticals and from some important randomized trials of vocational and therapeutic interventions involving groups of severely mentally ill consumers. Thus, we are limited in the types of treatments we can talk about, as well as the population to which we can generalize.
This leads to the second priority I would suggest for advancing our understanding of mental illness in the workplace. It seems to me that many of the ill are willing to talk about treatment, not just symptoms. Can we then construct and include reliable, valid and rich measures of pharmaco-therapeutic and other treatments into panel surveys? If so, the payoffs would be high. First, we would get a better estimate of average benefits of therapy among the ill. Second, we would develop a better estimate of the social value of therapy. Clinical trials tell us about the efficacy of one treatment. But the mentally ill who receive treatment often try many forms of treatment. Community-based samples with valid measures of treatment could provide us with a better sense of the ability of therapy as patients receive it to restore their productivity and employment. Further, if such measures were included in a community-based panel data set, we could learn more about the ways that work can affect therapeutic outcomes.
Next, let me mention a priority relevant to shaping future efforts to collect data, as well as researchers' efforts to analyze available data. It is important to recognize that people lead dynamic lives and that mental illness itself is dynamic. We know that the symptoms and disability associated with various mental illnesses are episodic. Yet, we really know little about the employment and productivity impacts of mental illness over episodes and over time. As the authors describe, we have identified useful relationships between time, disability leave and return to work. But we cannot yet sort out severity versus selection effects.
The dynamism of people's lives creates similar challenges. We know mental illness affects people throughout the life course. Still, we often think of mental illness in the workplace as a static concept - affecting the here and now of workers' lives. But mental illness can affect young people while they prepare for and launch their careers. More needs to be done to identify the effects of mental illness on formal schooling, labour market churning and job search. At the other end of the life course, we do not fully understand the role of mental illness in shaping decisions about withdrawal from the labour market.
Finally, let me turn to Dewa et al.'s very helpful discussion of the burden of mental illness in the workplace. Mental illness can limit the ability to find a job, show up at a job or be productive if present. As the authors make clear, the cost implications of this depend on whether you consider lost earnings or the cost of finding a replacement worker. Though they say little about this, a third perspective of importance is the larger social costs - which raise a host of complicated issues. One of these is that mental illness involves social expenditures in the form of disability and health benefits, in addition to the loss of the productivity of the ill. But this raises a further issue that the authors do not address. One way to assess the burden that mental illness places on the workplace is to assess the cost of finding a replacement for the ill worker, either temporarily or permanently. This friction-cost approach to assessing costs of illness is common in Europe but is used little in North America. In previous work, I have suggested that this approach is more sensible in European countries with high unemployment rates, where one might think of illness as resulting in a re-distribution of employment among workers (Marcotte and Wilcox-Gök 2001). I remain convinced that a friction-cost approach is less compelling in the relatively tight North American labour markets - and in any case, from the perspective of the ill, the costs of illness are huge. But it is fair to say that it is not as clear as Dewa et al. claim that costs to employers, or society, of mental illness are larger that the sum of losses to the ill themselves. On this note, Dr. Goering has published two of the few papers to apply a friction-cost approach to estimate the costs of mental illness in Canada. She and her colleagues find that friction-cost estimates of the cost of schizophrenia are much smaller than the sum of the losses to the ill themselves (Goeree et al. 1999a; Goeree et al. 1999b).
This last point can be integrated with the priorities suggested earlier to round out an agenda for the community of researchers and policy analysts that is both ambitious and humbling. Our first task must be to work to acquire better information if we hope to understand the nature of mental illness in the workplace as well as the employment, productivity and earnings implications of mitigating mental illness. Even if that can be done, we need to develop further our analytical understanding of the full economic costs of mental illness in the workplace, as well as the distribution of those costs.
Clearly there is a lot on the collective agenda - so much that one could be overwhelmed by how much we do not know, despite decades of good work by smart people. But I take heart in the knowledge that what we do know about mental illness in the workplace today is leaps and bounds beyond what we knew just one or two decades ago. The amount and quality of data available to us now would have seemed like a fantasy even very recently, and our ability to analyze it has progressed rapidly. I am optimistic that in the coming decades we can continue to advance our understanding of the complicated and important relationship between mental illness and employment.
About the Author(s)
Dave E. Marcotte, PhD
Associate Professor of Public Policy,
University of Maryland, Baltimore County
Dewa, C.S., A. Lesage, P. Goering and M. Caveen. 2004. "Nature and Prevalence of Mental Illness in the Workplace." Paper prepared for the Canadian Institutes of Health Research Invited Workshop on Mental Health and the Workplace. April 28-29, 2004. Toronto, ON.
Goeree, R., B.J. O'Brien, G. Blackhouse, K. Agro and P. Goering. 1999a. "The Valuation of Productivity Costs Due to Premature Mortality: A Comparison of the Human-Capital and Friction-Cost Methods for Schizophrenia." Canadian Journal of Psychiatry 44: 455-63.
Goeree, R., B.J. O'Brien, P. Goering, G. Blackhouse, K. Agro, A. Rhodes and J. Watson. 1999b. "The Economic Burden of Schizophrenia in Canada." Canadian Journal of Psychiatry 44: 464-72.
Marcotte, D. and V. Wilcox-Gök. 2001. "Estimating the Employment and Earnings Costs of Mental Illness: Recent Developments in the United States." Social Science and Medicine 53: 21-27.
Murray, C.J. and A.D. Lopez. 1997. "Alternative Projections of Mortality and Disability by Cause, 1990-2020: Global Burden of Disease Study." Lancet 349: 1498-504.
Wells, Kenneth. 2002. "Science Discovery in Clinician-Economist Collaboration: Legacy and Future Challenges." Seventh Carl Taube Lecture, National Institute of Mental Health, 11th Biennial Research Conference on the Economics of Mental Health. Bethesda, MD.
Be the first to comment on this!
Personal Subscriber? Sign In
Note: Please enter a display name. Your email address will not be publically displayed