Evidence-Based Practices: Shaping Mental Health Services Toward Recovery

Supported Employment

Simple Employment Outcome Measures

The following gives an outline for measuring employment outcomes at a program level. Staff and/or administrators in a supported employment program or in an agency seeking to monitor its own progress over time could be assigned to collect the information. The methods are also suitable for a state agency seeking to monitor a group of programs across the state.

Monitoring outcomes is important for any evidence-based practice. For supported employment, the main outcome is competitive employment. Although there are many aspects of competitive employment that would be desirable to know, the primary outcome of interest is whether or not a consumer is working or not in competitive employment.

The definition of competitive employment includes the following:

pays at least minimum wage
the employment setting includes co-workers who are not disabled
the position can be held by anyone, that is, one does not need to be a member of a population with a disability to hold the job

Some employment programs may choose to measure other types of employment in addition to competitive employment. The system below can be adapted to do so (e.g., use different codes for agency-run business), but every addition to a reporting database compounds the complexity of one’s method.

The employment reporting database

We strongly recommend the prospective collection of data. Although in theory one can retrospectively collect program activity over a prior period of time, our experience is that retrospective data collection, especially when the time period is long and the number of consumers to track is large, is susceptible to clerical and data entry errors. We recommend that the reporting grid be updated at a regularly scheduled meeting, ideally at least weekly. To minimize data entry errors one individual should be assigned the responsibility to update the database. This person obtains the information directly from the employment specialists.
The basic format for recording employment data is an EXCEL spreadsheet, (see Employment Reporting Grid), with the rows consisting of consumer names (or ID #s, depending on issues of confidentiality) and the columns consisting of weeks. The second column gives the admission date into the program and the third column gives the termination date if the consumer is closed.

A start date must be chosen for data collection. The names consist of all active consumers in the program as of the start date. In each cell is recorded a “W” for working or a “N” for not working in a particular week. Working means that the person actually worked in that week. New names are entered at the bottom of the list as they are added to the roster. An end date for data reporting—e.g., one year after the start date—must also be chosen and then comparisons can be made.

A prototype for the database is attached.

Employment outcomes

The Employment Reporting Grid permits the calculation of the following:

Percentage of consumers who were employed at any time during follow-up.

The numerator consists of the number of people who worked at least one week during follow-up. Denominator consists of the number of people who were active at least one week during the follow-up.

Percentage of consumers employed at follow-up.

Numerator = all employed at follow-up. Denominator = all active at follow-up.

Average weeks worked among clients enrolled in program.

Numerator = total number of weeks of employment across all consumers. Denominator = total number of consumers enrolled at any time.

Refinements to the database

As described above, the database captures very basic information about employment outcomes. For agencies seeking more detailed information, the methodology can be modified to record number of hours worked each week as the cell entry, rather than simply Work/No Work. If this information is recorded, then average hours employed during follow-up can be calculated. Another refinement would be to record types of jobs held. For many purposes, a running list of each new job obtained would be adequate.

Further points to consider

The calculations of rates are sensitive to admissions and dropouts. For example, if there are many admissions just prior to the end date for the evaluation, and most are not employed, then the employment rate may be artificially low. There could be statistical corrections that could be made (that is, adjust for total time available to work), but the calculations can be complicated and defeat the goal of “simple” employment measures.

In some cases programs or agencies may choose to assess employment rates beyond the supported employment program and may choose to select consumers based upon such variables as program participation, diagnosis, functional assessment. For example, a center may choose to study the total population of consumers with severe mental illness attending the community support program. If this is the case, the Employment Reporting Database should be adjusted accordingly, and the method of data collection may need to be adapted.

Graphing employment outcomes

We recommend that programs implementing an EBP graph their employment outcomes over time using the data from the grid.

Employment Reporting Grid

Name/ ID #	Admission Date	Termination Date	Week Starting
			1/1/03	1/8/03	1/15/03	1/22/03	1/29/03	etc.

Back to Supported Employment