Dr. Salkever is professor of public policy at the University of Maryland-Baltimore County (UMBC), Public Policy Building, Room 418, 1000 Hilltop Circle, Baltimore, MD 21250 (e-mail: salkever@umbc.edu).

An evidence base is accumulating for supported employment for persons with severe and persistent mental disorders, particularly for supported employment based on the individual placement and support (IPS) model (1; see page 1391). This evidence base is the platform for codifying IPS with fidelity measures and for efforts to more widely disseminate supported employment based on IPS principles.

In this group of articles on employment, several remind us that even when an intervention deserves the appellation "evidence-based," further questions about its comparative and overall efficacy and effectiveness are still worth asking.

The multisite Employment Intervention Demonstration Project (EIDP) generated what is perhaps the most widely read evidence of supported employment success (2). However, studies in this issue by Macias and colleagues (page 1406) and Schonebaum and coauthors (page 1416) analyzing the same EIDP data from the site in Worcester, Massachusetts, contradict earlier findings and report that employment services provided to members of a clubhouse achieved outcomes similar to those of IPS supported employment programs. This finding is even more surprising because the clubhouse was compared with a supported employment program that was fully integrated into an assertive community treatment (ACT) program—and full integration with mental health treatment is believed to be most effective for supported employment. (Bond [1] has conjectured that the ACT model may take the idea of integration too far.) At a minimum, reconciling these findings requires recognition that "control" conditions in the EIDP varied widely by site. The same may be true for the experimental interventions; most sites were not described as IPS programs (3). Defining "treatment as usual" in our evidence base can be tricky, and generalizing from results from multiple sites with varying interventions may be hazardous.

Other interesting issues are raised by these papers. One is the importance of job duration as an outcome measure (page 1416). The IPS job duration results have been weak (1). A second is the ability of different program models to move clients into employment who are not strongly work oriented at intake, which Macias and colleagues (page 1406) argue may be greater for clubhouse programs. We do not have evidence on the likely take-up rate of supported employment programs if they were more widely available. Similarly, we do not know whether the rate for clubhouse programs would in fact exceed rates for supported employment. If that turns out to be so, having both types of services available may be the best option. A third issue, raised by Casper and Carloni (page 1430), is the extent to which take-up rates would increase if providers of psychiatric services were more aware of consumers' desires to work.

The clear but qualified success of IPS-based supported employment invites further efforts to improve it and assess it against the most promising alternatives, which may include a clubhouse model. At least two directions for improvement are now under active investigation. One, highlighted by McGurk and Mueser (page 1406), is combining IPS services with cognitive remediation strategies. A second is to offset problems in access to and provision of high-quality psychiatric care, including medication management, which Cook (page 1391) notes may constitute barriers to employment. The Social Security Administration has recently funded a major trial, the Mental Health Treatment Study (www.ssa.gov/disabilityresearch/mentalhealth.htm), that will randomly assign SSDI recipients with mental disorders to IPS services combined with medication monitoring by using the protocols of the Texas Medication Algorithm Project. This study will also substantially expand financial coverage of behavioral health services for the experimental group. The results will be valuable in assessing the degree of synergy between improvement in access to and quality of care and IPS services.

Among the many directions that might be taken in transforming the mental health system to remove barriers to employment, expanding access to and improving the effectiveness of supported employment and other promising models of employment services should surely rank very high on our policy priority list.