One hundred and two subjects participated in the study that included a cohort of 59 patients with BD type I, and a cohort of 43 patients with schizophrenia, 15 years or older, consecutively enrolled from April 2008 to November 2009. The participants were recruited from Roozbeh Hospital, a referral teaching hospital in Tehran, the capital of Iran, at the time of discharge after hospitalization due to an acute episode of mania, mixed, or psychosis. All subjects met the Diagnostic and Statistical Manual of Mental Disorders, Fourth edition (DSM-IV) criteria for bipolar I disorder or schizophrenia. Exclusion criteria included an inability to obtain informed consent or comply with study requirements.
To measure QOL as the main outcome variable, the Farsi (Persian) version of the World Health Organization's QOL Instrument-Short Version (WHOQOL-BREF) [19
] was used. The WHOQOL-BREF is a self-administered, multidimensional scale used to assess physical, psychological, social, and environmental aspects of QOL. Also, we used the Young Mania Rating Scale (YMRS) [21
] and the Hamilton Depression Rating Scale (HDRS) [22
] to assess manic and depressive symptoms, respectively. The Global Assessment of Functioning Scale (GAF), which is a clinician-rated composite measure of functioning and symptoms, was used to assess overall psychological, social, and occupational functioning in the past month [24
]. Additionally, we used the Clinical Global Impression-Severity (CGI-S) [26
]. The CGI-S, a serial assessment, is also a clinician-rated scale to assess each patient's mental illness severity using a 1–7 response scale with higher scores corresponding to more severity. The assessment protocol was repeated 4-, 8-, and 12-months after hospital discharge. Inter-rater reliability on the YMRS, HDRS, GAF, and CGI-S scores was ascertained and intra-class correlation coefficients were between 0.7 and 0.85 across the above-mentioned scales.
Analyses were carried out in SPSS 16.0. Continuous and categorical variables were compared using the independent sample t-test (or Mann-Whitney test) and χ2 test, respectively. For comparison of two groups in each assessment period, the ratings of WHOQOL-BREF were treated as continuous data and were analyzed by using a two-tailed, repeated measures analysis of variance (ANOVA), with the diagnosis as a between-subject grouping factor, the WHOQOL-BREF domain as within-subject grouping factor, and other variables including age, marital status, educational level, work status, and GAF as covariates. To check the assumptions for repeated measures ANOVA, we used Mauchly's Test of Sphericity, Levene's Test of Equality of Error Variances, and Box's Test of Equality of Covariance. As four WHOQOL-BREF domain scores in each assessment were correlated more than what we would expect by chance, we had to account for this correlation. We can say that the four domain scores come from the same “subject”, or that this measure was “repeated”. Therefore, we considered the WHOQOL-BREF domain as within-subjects factor. If a significant overall ANOVA was observed it would be followed by individual pair-wise comparisons that used a two-tailed Student's t-test corrected by the Bonferroni method for multiple comparisons. Within bipolar patients, we analyzed the association of clinical and functional variables with the domains of QOL at the baseline, 4-, 8-, 12-month assessments through multiple linear regressions. In order to check the assumptions for linear regression analysis, we used the lack of fit test, the Shapiro-Wilk test, and the Durbin-Watson statistic to test the assumption of linearity, of normality, and of independence of errors, respectively. Adjusted analyses were performed for each domain of the WHOQOL-BREF separately using stepwise linear regression models. The outcome variable in each model was the domain of the WHOQOL-BREF, treated as a separate numeric variable. The numerical YMRS, HDRS, GAF, and CGI scores were included as independent variables. In the adjusted analyses, the effect of the score of the WHOQOL-BREF was controlled for sex and age. Spearman correlation coefficients were calculated between each item of the HDRS and the WHOQOL-BREF domains. The adjustment of multiple comparisons was applied and a P value of less than 0.05 was considered as statistically significant. The Ethics Committee of Tehran University of Medical Sciences gave ethical clearance. Written informed consent was collected before entering the study.