Following review of existing measures and discussion with experts, caregivers, and patients, content was generated for four domains: 1) living situation; 2) IADLs; 3) productive activities; and 4) social functioning. The final SOFI was constructed with items informing domain scores, and an interviewer-completed global rating for each domain.
Psychometric characteristics of the SOFI were evaluated in a sample of 104 community residing patients with schizophrenia and their informants. Test–retest reliability was evaluated in a sub-sample of patient–informant dyads using ICC; all values were > 0.70 for both patient-interviews (SOFI-P) and informant-interviews (SOFI-I). Inter-rater reliability ICCs ranged from 0.50 to 0.79 on a different sub-sample. The SOFI demonstrated adequate construct validity based on correlations with the PSP (range 0.58 to 0.76; p < 0.0001) and the QLS (p < 0.001). Some correlations between SOFI and PETiT scores were low to moderate (p < 0.05). Discriminant validity was supported based on SOFI score comparisons for patient groups based on PANSS and BACS scores (p < 0.05); SOFI scores differed between borderline and moderately ill patients as measured by the CGI-S (p < 0.05).
The SOFI expands on existing measures and more comprehensively captures functioning of patients in the real world than other performance-based (proxy) measures. The SOFI has good evidence supporting reliability and construct validity, and may be a useful measure of functional outcomes in schizophrenia.