The study collected data on patient employment and behaviors from the following sources: personal interviews, vocational activities log and employment documentation. Employment measures drew on and integrated information from these sources. The major employment outcome was the attainment of a paid job.
To provide a single overall index of vocational outcomes, they were combined into a four-level, ordinal measure of highest level of vocational attainment:
1. Attained a competitive job (highest).
2. Attained an informal job.
3. Performed constructive activity other than paid employment.
4. None of the above (lowest).
CES counselors left during the study period. Patients who received no counseling (N=24) were excluded from this sample. The interview follow up rate for follow up at 12 months was 91%. The main analytical technique was multiple logistic regression with statistical significance set at p < .05 (2-tailed). Vocational intervention condition was introduced into the analysis as an indicator variable (CES=1, standard=0). The statistical control for period of intervention, as described above, is also an indicator variable (earlier cases = 1, later cases = 0). All analyses of employment outcomes included the baseline measure of the outcome variable as a covariate. Since the study had no dollar measure of earnings prior to baseline, work in the prior 6-month period was used as the covariate in the analysis of earnings as an outcome. Patient baseline variables were examined to determine whether any were significantly associated with both an employment outcome and vocational intervention assignment (CES vs. standard counseling); none were. Thus, none of these variables was a potential confounder in the regression analysis. Work in the pre-study period was included as a covariate in the multivariate analysis, however, to enable the potential independent effect of the CES intervention on employment outcomes to be determined.