Din aceeași categorie
1. Introduction
Schizophrenia is one of the most severe mental disorders, but its essential nature remains to be clarified. The global age-standardized point prevalence of schizophrenia in 2016 was estimated to be 0.28%(1). Unfortunately, there are important barriers to achieving optimal outcomes in this disease due to patient-related factors (males, single status, uneducated, unemployed, frequently or ever-used drugs or alcohol)(2), as well as treatment-related factors associated with favorable outcomes (antipsychotic medication adherence)(3). Insight refers to a multidimensional construct with distinct but interrelated subdomains: awareness of illness, consequences of illness, need for treatment and illness attribution(4). Interestingly, schizophrenia patients with severely impaired insight report a more positive perception regarding quality of life and illness severity, as opposed to physician evaluation or those with unimpaired insight(5). Psychopharmacological and psychological treatments can significantly reduce symptoms of schizophrenia but, overall, the disease is associated with a high number of hospital admissions and long-term use of psychiatric services(6,7). Our research aimed to identify the psychometric predictive factors involved in the clinical outcome of schizophrenia.
Method
We conducted a prospective cross‑sectional study including 80 patients admitted to the Psychiatry Clinic of the County Emergency Clinical Hospital Cluj-Napoca, diagnosed according to the 10th edition of the International Classification of Diseases (ICD-10) with schizophrenia, aged between 19 and 73 years old, with an educational level of minimum eight years of study (grades), and voluntary research inform consent singed. The exclusion criteria consisted in the presence of mental retardation, organic cerebral syndromes, alcohol and psychoactive substances abuse, epileptic seizures, and pregnancy or lactation. The patients were evaluated twice during their hospitalization, first after the admission and, secondly, before releasing from hospital.
A semi-structured interview collected the socio-demographical data. The psychotic symptoms were evaluated using the Positive and Negative Syndrome Scale (PANSS)(8), and the severity of the disease was assessed using the Clinical Global Impression (CGI). Insight was measured using the Scale for the Assessment of Unawareness of Mental Disorder (SUMD)(9-11), the Schedule for Assessment of Insight-Expanded Version (SAI-E)(12), and the Beck Cognitive Insight Scale (BCIS)(13). For the descriptive analysis, we used indicators of central tendency (arithmetic mean) and the dispersion (standard deviation, minimum and maximum values). Where the sample was analyzed simultaneously in terms of two categorical variables (e.g., gender × adherence to treatment), charts were constructed based on contingency tables and the association of variables was tested using the nonparametric chi-square correlation coefficient for categorical variables. The materiality of these values was tested at a significance level less than 0.05. To test the effectiveness of intervention on psychotic symptoms and the level of insight, from admission to discharge, we choose the dependent samples t test – i.e., repeated measurements. For statistically significant differences, we also calculated the size of intervention effect based on Cohen’s d coefficient.
Results
Our group consisted of 55% male patients and 45% female patients, the mean age was 35.76 ± 10.95 years old (range: 19-73 years old), the average age of illness onset was 27.24 ± 8.81 (range: 16-57 years old) and the average number of hospitalizations was 5.13 ± 5.64 (range: 0-35 hospitalizations). Our psychometric evaluation revealed the following results.
The progression of psychotic symptoms and insight from admission to discharge
CGI and PANSS progression
Table 1 presents the descriptive indicators of psychotic symptoms measured by PANSS and for the severity of symptoms measured by the CGI scale.
The results show a significant reduction in the severity of psychotic symptoms at discharge compared with its level on admission, with a large effect size, according to the Cohen’s d indicator (d=2.59). Also, there were significant reductions in psychotic symptoms measured by PANSS, for both positive and negative symptoms’ subscales, both with large effect sizes (positive d=1.68; negative d=1.16; total d=2.04).
SUMD progression
We note that for this scale high scores mean low levels of insight. Table 2 presents descriptive indicators of scores’ progression for insight, and wrong attribution of current and past symptoms.
As it can be seen in Table 2, following insight average column at admission and discharge, for every SUMD subscale and, as well, for total score, there is a downward trend indicating a decrease of non-insight (high SUMD scores show a high level of non-insight). T test with repeated measurements performed for each of these variables showed that these decreases are statistically significant for both insight and wrong attribution of current symptoms (SUMD C.a total for insight, and SUMD C.b for the attribution of current symptoms), and also for insight and wrong attribution of past symptoms (SUMD P.a for insight, and SUMD P.b for the attribution). By using Cohen’s d indicator, we calculated the size of these effects and we found average values for current symptomatology (d current insight = 0.43; d current attribution = 0.46) and reduced values for past symptoms (d past insight = 0.20; d past attribution = 0.25).
SAI-E progression
We note that for this scale high scores mean high levels of insight. Table 3 presents descriptive indicators of scores’ progression for insight and treatment compliance scales and for total score.
Dependent samples t-test performed in each of these cases confirms the statistically significance of these trends for subscales of insight, treatment compliance, and for total insight score. As the sizes of these effects highlighted by the Cohen’s d indicator, they are medium to low (d insight = 0.41; d compliance = 0.36; d SAI-E total = 0.43).
BCIS progression
The third scale of insight analysis is BCIS cognitive insight scale. Table 4 presents the descriptive indicators of progression from admission to discharge.
Following the progression of cognitive insight average levels for every subscale and for total scores, we find that these values record an upward trend from admission to discharge. Dependent sample t-test revealed that this ascendent trend is statistically significant only for self-reflectiveness scale and for cognitive insight total score. For self-certainty scale, this trend is not statistically significant, and Cohen’s d indicator showed low and very low values for the effect at BCIS scales (d self-reflectiveness = 0.26; d self-certainty = 0.05; d total BCIS = 0.16).
Predictors of symptoms progression measured
by PANSS
Further on, we are interested to highlight the significant predictors for psychotic symptoms’ progression from admission to discharge. The analysis was set up based on some steps.
Step one – We defined the variable “∆ PANSS” as the difference between PANSS scores at admission and discharge, and tested the correlation of this variable according to baseline PANSS score. If there was a strong correlation, that means that the initial level of symptoms explained a significant proportion of variance of disease progression.
Step two – We tested the correlation of all the various potential predictors of change in magnitude of change (Δ PANSS), controlling the initial level of symptoms (PANSS score at baseline) using partial correlation.
Step three – We introduced the significant partial correlates of change as predictors in multiple linear regression equation to compare their predictive values for the magnitude of change in PANSS (progression of the disease from admission to discharge) and to determine the optimal predictive model.
In this sequence of the research, we tested the correlation between baseline PANSS and magnitude of change of this variable. The statistical analysis showed a high correlation (r=0.74), significant for a threshold p<0.01 which means that the initial level of PANSS explained 54% (square of correlation coefficient) of the variance for this variable magnitude of change from admission to discharge. Therefore, in step two, all potential predictors of change correlations with Δ PANSS were tested at the initial level statistical control PANSS (partial correlation). In the next step we analyzed the partial correlation (at baseline – first controlled PANSS) of PANSS symptoms progress with each of its potential predictors. Table 5 shows the results of our analysis.
What can be seen from Table 5 is that the potential predictors of psychotic symptoms change, with significance related to change, appear to be only those measured by the level of insight SUMD scale, presence of family history, and belonging to urban areas. So, as the initial level of insight is higher, there will be an even more substantial improvement of symptoms. Raising the predictive value coefficient to square, we obtain the coefficient of determination corresponding to this predictor. Therefore, the level of insight, as one predictor, with PANSS initially controlled, explained 16% of variance of improving psychotic symptoms during hospitalization. Also, patients from urban areas seem to have a slight tendency towards higher magnitudes of change, as the coefficient of determination indicates that urban membership explained 5% of the variance when symptoms improve. Also, it seems that the family history is positively associated with magnitude of change of symptoms, but the level of this relationship is also reduced.
In this sequence, we are interested to identify the optimal predictive model of disease progression, based on available significant correlates of this dynamic. In order to reach this, we used hierarchical regression analysis in which the predictive model introduces all predictors, one by one, in order of correlation with progression of the disease, aiming to identify the model with the fewest predictors but most explanatory value. Table 6 presents the results of hierarchical regression.
From Table 6, we can see that, from initial level of composed model of variables, PANSS and total SUMD explained 60% of variance when disease progression (adjusted R²=0.60), in a statistically significant extent (p<0.01). Adding to this model the “area of residence” variable explained the increase by a further 1% for proportion of variance (R² Change=0.01), but this addition is not statistically significant (F change=2.97; p>0.05). Thus, the best predictive model of disease progression is composed of variables SUMD total and PANSS total on admission.
Discussion
Our study shows that the clinical outcome in schizophrenia is complex and involves many patient-related factors. Our results revealed that the severity of symptoms measured by PANSS explained 54% of the variance for this variable magnitude of change from admission to discharge. Also, both PANSS total and SUMD explained 60% of variance of the disease clinical evolution. From our date knowledge, there is no research in which the PANSS and SUMD scale was used to predict the variance of improving psychotic symptoms during hospitalization. There was a significant improvement in insight measured by SUMD and BCI scales. These results are similar to those found in other research(14). The change in scores on the PANSS scale over the three-week or four-week period predicted the improvement in insight, results supported by several studies(15). Our study showed that only the level of insight measured with the SUMD scale, the positive family history, and the environment (urban area) correlated statistically significantly with the change in psychotic symptomatology. A higher level of insight at admission was associated with a more substantial improvement in symptoms. Our results were in agreement with some previous studies (Mintz et al., 2004(16); Gharabawi et al., 2006(17); Weiler et al., 2000(18); Chen, 1998(19,20)). The limitations of the current study are represented by the research design, in which the prediction of psychotic symptoms and insight was made under the specific hospitalization conditions and psychopharmacological treatment for each patient, each of these being factors that modulate the outcome.
Conclusions
Our findings offer an additional knowledge of using psychometric evaluation to predict clinical outcome in schizophrenia evolution.
Conflict of interest: none declared
Financial support: none declared
This work is permanently accessible online free of charge and published under the CC-BY.
