INTRODUCTION: The Social Functioning Scale (SFS) was designed to evaluate social functioning essential to schizophrenic patients. Its length may be difficult to use in clinical practice. The objective of this study was to develop and validate a short version of the SFS. METHODS: Data from 445 patients with schizophrenia who came from two separate studies, one longitudinal (n=250) and one cross-sectional (n=195), were used to produce and validate the short form of the SFS. The two samples were combined and then randomly split into two subsamples. In the first subsample (n = 223), items were eliminated using classical, modern (item response theory) psychometric criteria, as well as clinimetric criteria. The short version was independently validated using data from the other subsample (n = 222), by comparing the level of association (correlation and Area Under the ROC Curve [AUC]) with the EQ-5D VAS and the Clinical Global Impression (CGI), with the original scale. RESULTS: The original 78 items were reduced initially to 19 and 13 items (respectively using classic and modern psychometric criteria) and then to 15, since 2 items related with the employment capacity were added for clinometric criteria. The short form of the SFS had a Cronbach's alpha of 0.76. Spearman correlation coefficients with the EQ-5D VAS and with the CGI score (0.46 and 0.42, respectively) were similar or even higher for the short version than for the original version. The AUC of the SFS and the dichotomous CGI were practically the same for both the original (AUC: 0.74) and the short (AUC: 0.73) versions. CONCLUSIONS: The short version of the SFS proved to be reliable and valid. It could be adequate for use in clinical practice.
INTRODUCTION: The Social Functioning Scale (SFS) was designed to evaluate social functioning essential to schizophrenicpatients. Its length may be difficult to use in clinical practice. The objective of this study was to develop and validate a short version of the SFS. METHODS: Data from 445 patients with schizophrenia who came from two separate studies, one longitudinal (n=250) and one cross-sectional (n=195), were used to produce and validate the short form of the SFS. The two samples were combined and then randomly split into two subsamples. In the first subsample (n = 223), items were eliminated using classical, modern (item response theory) psychometric criteria, as well as clinimetric criteria. The short version was independently validated using data from the other subsample (n = 222), by comparing the level of association (correlation and Area Under the ROC Curve [AUC]) with the EQ-5D VAS and the Clinical Global Impression (CGI), with the original scale. RESULTS: The original 78 items were reduced initially to 19 and 13 items (respectively using classic and modern psychometric criteria) and then to 15, since 2 items related with the employment capacity were added for clinometric criteria. The short form of the SFS had a Cronbach's alpha of 0.76. Spearman correlation coefficients with the EQ-5D VAS and with the CGI score (0.46 and 0.42, respectively) were similar or even higher for the short version than for the original version. The AUC of the SFS and the dichotomous CGI were practically the same for both the original (AUC: 0.74) and the short (AUC: 0.73) versions. CONCLUSIONS: The short version of the SFS proved to be reliable and valid. It could be adequate for use in clinical practice.