Work and Mental Illness

Work and Mental Illness: Perspectives from Three National Surveys

Project Investigators: David O'Neill, Ph.D., and David N. Bertollo

PROJECT GOALS
Mental illness reduces the capacity of individuals to work. The size of this reduction and its cost to society have long been of interest. There is also interest in the question of how some individuals with mental illness manage to work in spite of their illness, and whether those with mental illness whose symptoms abate are able to regain their pre-illness levels of work and earning capacity.

The existing literature on the determinants of work among those with mental illness is extensive but tends to focus on follow-up studies of small groups of people with severe mental illness who are either partaking in a cost/benefit analysis of a program to increase their work activity or are part of a long-run longitudinal study of the general "outcomes" of schizophrenia. There are also scattered studies of work loss due to other specific DSM disorders. What is needed is to bring together in one place knowledge about the comparative relationships between mental illness and work loss across the whole spectrum of DSM categories, along with some broad knowledge of what factors determine who works and who does not within a DSM category. Since our knowledge of the aggregate work loss due to mental illness is mainly based on data from a prevalence survey done in 1981-1983, some information based on more recent survey data, even if indirect, would provide some useful comparisons.

This study examines these relationships in three national surveys that take different approaches to the measurement of mental illness. Two of the data sets, the Epidemiological Catchment Area Study (ECA) (1980-1984) and the National Comorbidity Survey (NCS) (1990-1991), employ responses to questions about symptoms to determine current and lifetime prevalence of DSM disorders. In the third data set, the National Health Interview Survey (NHIS), householders were asked directly whether they or anyone in the household had one or more of a short list of specific mental illnesses, which were limited to the more severe DSM disorders. We have identical NHIS supplements for 1989 and 1994.

RESEARCH ACTIVITIES AND RESULTS
Our major findings to date are:

(l) Aggregate comparisons of employment rate differentials between people with and without mental illness among the three surveys suggested a decrease in work loss due to mental illness between 1981 and 1990 and no change between 1989 and 1994. However, the reliability of the evidence for the decrease during 1981-1990 is much less than for the finding of no change between 1989 and 1994.

(2) Our analysis of the change between 1989 and 1994 was based on identical supplementary mental illness questions to the NHIS, and we found an unexpected increase in the prevalence of the population with mental illness sampled by the NHIS. A preliminary analysis suggested that the economic recession of 1990-1992 might have played a causal role.

(3) There is a substantial amount of variation in employment and earnings rates across the DSM categories of disorder. In addition to schizophrenia, which is associated with by far the lowest employment and earnings rates in all surveys, the anxiety disorders (phobia, panic, obsessive-compulsive) and the affective disorder (dysthymia) are associated with relatively low employment and earnings rates. Surprisingly, the DSM categories of major depression and bipolar disorder, usually considered quite severe disturbances, are linked with above average employment rates.

(4) Educational attainment is a major determinant of work among persons with mental illness. It appears to be as strongly correlated with employment rates among persons with mental illness as it is among persons who do not have a mental illness.

(5) ECA data strongly suggest that abatement of symptoms may be associated with a significant recovery of work and earnings capacity by some persons with mental illness.

(6) A logistic multiple regression analysis of the NHIS data suggests that the net effect of educational attainment on employment rates even within this more seriously mentally ill population is large and remains high even after holding constant a number of covariates. The analysis also found that going to an outpatient therapist had a positive net effect on the probability of being employed. However, a variable indicating the use of psychotropic drugs had a negative partial regression coefficient, indicating that taking medication was associated with a lower probability of employment in the NHIS data.

SIGNIFICANCE OF FINDINGS/POLICY IMPLICATIONS
When the prevalence of mental illness in the household sector is measured by asking individuals if they have one or more of the major DSM disorders, it is mainly individuals with serious mental illness who respond. (However, because the household population sampled by the NHIS does not contain community residences and adult homes, the most severely disabled of those with mental illness are not covered by the NHIS.) The resulting data show very low employment and prevalence rates in comparison to the surveys that use extensive interview questions about symptoms to identify those in the sample who have mental illness. The much larger employment and prevalence rates of these surveys may occur because many of the individuals who are classified as having a major DSM disorder may have it in a relatively mild form.

The distinction between the two types of surveys is important, because both are useful for policy makers. For research into the determinants of mental illness and work status among individuals with mental illness, the broad and more inclusive symptom-based survey is much more useful. But for policy makers interested in medium-term changes in prevalence of serious mental illness (the kind usually considered the province of public mental health authorities), the respondent-identification method, which can be conducted rather quickly and cheaply, may yield useful information on trends in prevalence and employment rates for this important subgroup of individuals with mental illness.

The finding of variation in employment rates across DSM categories has implications for the rankings of the categories with regard to seriousness of the illness. Surprising was that for the anxiety disturbances (including obsessive-compulsive disorder) and dysthymia (usually considered the mildest of the affective disorders) employment rates were lower than those of individuals classified as having bipolar disorder and major depression (both repeated episodes and a single episode). In the ECA data, those individuals who had no symptoms for at least a year before the survey had significantly higher labor force performance than those with current symptoms. Moreover, their employment rates and estimated earnings were even higher than those of individuals who never had mental illness. This finding suggests that some individuals with mental illness may recover from their symptoms and return to the labor market at a fairly high level. Recovery may lead to a resurgence in labor market performance, not just a leveling or a minor reversal of a downtrend in performance that began with the onset of a mental illness.

We found that there was an increase in the prevalence of mental illness as measured in the NHIS between 1989 and 1994. Research should be conducted to see if this finding is replicated by other methods. If this is the case, possible causal mechanisms should be investigated. From an economic perspective, we have hypothesized that the "Bush Recession" of 1990-1992 may have had some influence.

Project completed.

Publications and Presentations:

Papers

O'Neill, D. and Bertollo, D. Work and earnings losses due to mental illness: Some broad perspectives from three national surveys. Administration and Policy in Mental Health, 25:3, May 1998.

Proceeding and Conference Presentations

O'Neill, D. and Bertollo, D. "Changes in prevalence and employment status of persons with mental illness: Evidence from the National Health Interview Surveys (NHIS) of 1989 and 1994." Poster presented at the Seventh Annual National Conference of States Mental Health Agency Services Research and Program Evaluation (accepted for publication in their forthcoming proceedings volume).

O'Neill, D. and Bertollo, D. "Work and mental illness: Perspectives from three national surveys," Proceedings: Sixth Annual National Conference on State Mental Health Agency Services Research & Program Evaluation, February 1996, The Evaluation Center at the Human Services Research Institute, 2336 Massachusetts Avenue, Cambridge, MA, 02140.

[Top]