The study sample comprised a total of 155 outpatients with schizophrenia. Eligibility criteria for inclusion in the study were: age between 18 and 65 years (either sex), diagnosis of schizophrenia (psychiatric diagnosis was made by qualified psychiatrists using ICD-10 criteria), on stable maintenance treatment, and written informed consent to participate in the study. Stability was defined as patients who did not require any change in their current pharmacological treatment during the last three months. The same criterion was applied during the follow up.

Patients with intellectual developmental disorder, acquired brain injury, or who refused to participate in the study were excluded.

For each patient, demographic data and clinical information regarding diagnosis, medication, and drug use was obtained. Severity of the illness and the level of functioning were assessed. All testing was performed by the same clinician at baseline and 6-month follow-up.

Schizophrenia Objective Functioning Instrument (SOFI). The Schizophrenia Objective Functioning Instrument (SOFI) is an interviewer-administered scale designed to objectively assess the actual level of patient functioning. It is organized into four domains: 1) living situation, 2) instrumental activities of daily living, 3) productive activities, and 4) social functioning (Kleinman et al., 2009). Each domain is rated on a global score of 0 to 100, with anchors provided for each 10 point level. Higher scores indicate better functioning. Prior to completing the global rating, several dimensions are evaluated for each domain using a semi-structured interview format with patients or informants, and including any additional information. In addition, an overall global score is calculated by taking the mean of the four domain global scores. The SOFI was translated following the rules of the International Test Commission (Muñiz, Elosua, & Hambleton, 2013). The original instrument was first translated into Spanish by three independent groups, each one composed of three researchers (psychologists and psychiatrists) who were fluent in English. Each group agreed on a preliminary version. Then, the coordinators of each group approved the final version. Finally, the Spanish version was back-translated by a clinical psychologist. The Spanish version was then pilot tested, by administering it to ten patients to assess its understandability (see Appendix 1).

The Positive and Negative Syndrome Scale (PANSS) (Kay, Fiszbein, & Opler, 1987) was used to evaluate the severity of positive and negative symptoms and general psychopathology in schizophrenia. Each item is rated on a 7-point Likert scale (total scores ranging from 30 to 120). Higher scores indicate worse symptoms.

The Clinical Global Impression-Schizophrenia Scale (CGI-SCH) (Haro et al., 2003) was designed to evaluate positive, negative, depressive, and cognitive symptoms, and overall severity in schizophrenia. Each item is rated using a 7-point Likert scale of intensity (from normal to among the most severely ill patients).

The Personal and Social Performance scale (PSP) (García-Portilla et al., 2011) assessed patient functioning in socially useful activities, personal and social relationships, self-care, and disturbing and aggressive behaviors. Regarding these dimensions, higher scores indicate worse level of functioning. An overall global rating is provided, ranging from 0 to 100. Conversely, higher scores indicate better level of functioning.

The Global Assessment of Functioning (GAF) (American Psychiatric Association, APA, 1987) considers psychological, social, and occupational functioning on a hypothetical continuum of mental health-illness, excluding impairment due to physical or environmental factors. The ratings range from 1-10 (severely impaired) to 91-100 (superior functioning).

The analyses were performed using SPSS 17.0. The two-tailed level of significance used was .05. Demographic and clinical data were evaluated with descriptive statistics, using the mean and standard deviation for the total sample and by sex. Sp-SOFI domain scores were reported separately, and as a percentile rank. Differences relating to the measurements were assessed using Chi-square tests, Student's t-tests, and ANOVA (Duncan's post-hoc test). First of all, in order to determine the psychometric properties of the Sp-SOFI, the discrimination indexes of the items were calculated. The differential item functioning by sex was estimated by logistic regression (Gómez-Benito, Hidalgo, & Zumbo, 2013), and an exploratory factor analysis was carried out using the correlation matrix between Sp-SOFI domains. The unweighted least squares method was used because it showed the best fit of the data to the model. For determining the dimensionality the percentage of explained variance, the goodness-of-fit index (GFI) and the root-mean-square residual (RMSR) were taken into account. The internal consistency of the translated version was assessed using Cronbach's alpha coefficient. The test-retest reliability of Sp-SOFI scores was evaluated over an interval of 6 months in the entire sample. The effect of practice on retest performance on the Sp-SOFI was determined by calculating the effect sizes for changes in performance at month 6 (Cohen's d). Convergent validity was studied using the correlation matrix between Sp-SOFI domains and PSP dimensions. The canonical correlation was calculated between both sets of variables. On the other hand, the multiple correlation coefficient between the Sp-SOFI domains and GAF overall score was determined. The expectation was that a moderate to high r correlation would be found for both instruments, as they are functional outcome measures. Also, known-groups validity was calculated, as it was by Kleinman et al. (2009) in the SOFI development. Patients were stratified into two severity groups based on median split of PANSS scores, and Sp-SOFI global scores were compared by severity level using a t-test. The hypothesis was that there would be significant differences in functioning as measured by the Sp-SOFI when patients with more symptoms as measured by the PANSS were compared to patients with fewer symptoms (Kleinman et al., 2009). To determine discriminant validity, patients were classified into four groups based on CGI scores of the Positive, Negative, Cognitive, and Overall Severity Symptoms Scales: doubtful (1- 2), mild (3), moderate (4), or severe (≥ 5) (García-Portilla et al., 2013). To identify statistically significant differences in the Sp-SOFI domain scores according to severity groups in each CGI scale, an ANOVA with Duncan's post-hoc test was calculated. The expectation was that less severely ill patients would have higher Sp-SOFI scores.

Table 1 shows the demographic and clinical patient characteristics (total sample and by sex). No significant differences were found according to sex except for marital status (Ӽ2=13.3; p=.004) and work status (Ӽ2=7; p=.029).

Table 2 shows the characteristics of the Sp-SOFI scores. Statistically significant differences were found in all domains according to sex, except for number 2 (instrumental activities of daily living), with females scoring higher. Table 2 also presents percentile ranks of the Sp-SOFI for the total sample and by sex. It is worth noting here that lower scores are indicative of poorer adjustment. The Sp-SOFI score corresponding to the 25th percentile was 44.2. Scores ranging between 62.5 and 77 corresponded to the 50th and 75th percentiles.

The Sp-SOFI was composed of items with discrimination indexes ranging from .21 to .77. No items presented differential item functioning (DIF) by sex (p<.01; R2<.035). Table 3 shows the correlation matrix between Sp-SOFI domain scores, which were positively and strongly correlated.

The results obtained in the exploratory factor analysis showed Kaiser-Meyer-Olkin (KMO) indexes above .80 (Table 4), as well as a statistically significant Bartlett's sphericity index (p<.001). All the factor loadings were in the range of .72 to .92. As can be seen in Table 4, the GFI was above .95, the RMSR was under .08, and the percentage of variance explained by the factor analysis is 76.22%.

Sp-SOFI reliability judged by internal consistency (Cronbach's alpha) was .93. Test-retest stability between Sp-SOFI domains was also adequate with a Pearson correlation of .80 to .82 (p<.001) after 6 months. For the Sp-SOFI total score, r was .87 (p<.001). Practice effects on the Sp-SOFI were very small (Cohen's d=0.09).

The Pearson correlation matrix between Sp-SOFI domains and PSP dimensions are shown in Table 5. All the scores were strongly and negatively correlated (p<.001). Note that higher scores in PSP dimensions mean worse level of functioning. Canonical correlation between the two sets of variables is shown in Table 6. As can be seen, the canonical correlation was .83. The Sp-SOFI domain with the highest relevance was “living situation” (domain 1). The multiple correlation coefficient between Sp-SOFI domains and GAF total score was .84, so 71.22% of GAF total variance was explained by Sp-SOFI domains.

On the other hand, known-groups validity was determined separately for all PANSS subscales (median Positive scale=13; median Negative scale=18; median General scale=30.5). Mean Sp-SOFI total scores were significantly different for the PANSS-Positive based groups (p=.007) and for each domain (p<.005), except for domain 3 (productive activities). Groups based on the PANSS-Negative scale demonstrated statistically significant mean differences for all the Sp-SOFI domains and total score (all p<.001). Likewise, all Sp-SOFI scores also were significantly different between high and low scores on the PANSS-General Psychopathology scale (all p<.001).

All Sp-SOFI measures demonstrate an ability to discriminate among patients with doubtful (CGI=1–2), mild (CGI=3), moderate (CGI=4), and severe (CGI≥5) schizophrenia disorder according to CGI-SCH Positive, Negative, Cognitive, and Overall Severity Symptoms scales (all p<.001). As expected, less severe patients have higher Sp-SOFI scores. Table 7 shows the mean Sp-SOFI domains and total scores for each severity group on each CGI-SCH scale.