Factor structure of the Albanian version of the Clinical Assessment Interview for Negative Symptoms (CAINS): Associations with the Brief Symptoms Inventory (BSI)

INTRODUCTION

Measurement and assessment of symptoms across different types of psychotic disorders, such as schizophrenia, bipolar affective disorder, schizoaffective disorder, and psychotic depression is difficult due to heterogeneity in presentation and outcome. A misleading assessment may lead to ineffective treatments, which in turn influences functional decline, illness chronicity, and iatrogenic physical illness (Clark, Schubert, Olagunju, Lyrtzis, & Baune, 2018). As such, good quality mental health care for psychotic patients is characterized by a detailed and broad assessment of patient problems and needs. To facilitate this type of in-depth assessment of negative symptoms in psychotic disorders, an extraordinary range of measures have been produced (Martins, Carvalho, Castilho, Pereira, & Macedo, 2015). Among the most frequently used of these measures are the Positive and Negative Syndrome Scale (PANSS; Kay, Fiszbein, & Opler, 1987), the Scale for the Assessment of Negative Symptoms (SANS; Andreasen, 1982), and the Negative Symptom Assessment (NSA; Alphs, Summerfelt, Lann, & Mueller, 1989). However, these measures have many limitations (Daniel, 2013; Blanchard, Kring, Horan, & Gur, 2011; Kirkpatrick, Fenton, Carpenter, & Marder, 2006; Lincoln, Dollfus, & Lyne, 2017; Stahl & Buckley, 2007). Notably, these assessments don’t include all negative symptoms, since knowledge about schizophrenia and other psychotic disorders has progressed from the time these instruments were developed (Blanchard & Cohen, 2005; Montgomery & Zwieten-Boot, 2007). Another reported limitation is the focus on behavioural aspects and inclusion of cognitive functions, for example, ‘attention’ in SANS and ‘stereotyped thinking’ and ‘abstract thinking’ in PANSS (Harvey, Koren, Reichenberg, & Bowie, 2006). Thus, longevity, complexity, and inability to differentiate between ‘anticipatory’ and ‘consummatory’ anhedonia are additional criticisms of these commonly used measures (Blanchard et al., 2011; Kumari, Malik, Florival, Manalai, & Sonje, 2017).

To address these limitations, a new instrument called Clinical Assessment Interview for Negative Symptoms was developed (CAINS; Horan, Kring, Gur, Reise, & Blanchard, 2011; Kirkpatrick et al., 2006; Kring, Gur, Blanchard, Horan, & Reise, 2013). CAINS assesses five negative symptoms: asociality, avolition, anhedonia, blunted affect, and alogia. It provides two scales that are scored separately, Motivation and Pleasure (Social, Work & School, Recreation; nine items), and Expression (four items). Items are rated on a 5-point scale (0–4), where ‘0’ indicates that the symptom is absent while ‘4’ indicates that the symptom is severe. Importantly, items that measure anhedonia only take the past week and the next week into consideration (consummatory and anticipatory; Blanchard, Gur, Horan, & Kring, 2012). Regarding the psychometric properties, CAINS is considered to be a reliable measure: internal consistency scores for the total CAINS were satisfactory in the initial study (Forbes et al., 2010), while subsequent studies have shown that the scales have good internal consistency, test-retest reliability, and inter-rater agreement (Horan et al., 2011; Kring et al., 2013). Additionally, CAINS has been shown to be reliable across different cultures and languages (Chan et al., 2015; Engel, Fritzsche, & Lincoln, 2014; Hosakova, Viktorova, Lecbych, & Hosak, 2017; Rekhi et al., 2019).

When compared with other instruments that measure the same construct, CAINS was shown to have good convergent validity: both CAINS scales were correlated with the negative symptom subscale of Brief Symptom Rating Scale (BPRS; Overall & Gorham, 1962) and with the subscales of SANS (Kring et al., 2013). In support of convergent validity, both CAINS scales were highly correlated with the PANSS negative subscale (Engel, Fritzsche, & Lincoln, 2014). The discriminant validity of CAINS was demonstrated by analysing if it was correlated with scales with which it was not expected to be correlated. Empirical evidence showed good discriminant validity by insignificant correlations with PANSS positive subscale and Calgary Depression Scale for Schizophrenia (CDSS; Addington, Addington, & Schissel, 1990) total score (Blanchard et al., 2017; Rekhi et al., 2019).

Two initial studies that analysed CAINS found a two-factor structure (Horan et al., 2011; Kring et al., 2013). The two modestly correlated scales were named: (1) Motivation and Pleasure and (2) Expression. The two-factor solution was also validated in a German population (n = 53; Engel, Fritzsche, & Lincoln, 2014), a Chinese population (n = 68; Chan et al., 2015), and a Spanish population (n = 100; Valiente-Gomez et al., 2015). Conversely, Rekhi and colleagues (2019) reported a four-component model in their study, in which they re-named the scales motivation-pleasure (MAP) social, MAP vocational, MAP recreational, and EXP (expression). The authors justified this result by considering the study of Ahmed and colleagues (2018), which indicated that two-factor models do not fit the data sufficiently well. However, further research is needed to clarify this discrepancy in results.

Although there is growing evidence suggesting that CAINS captures individual variation among psychotic patients in different countries (Rekhi et al., 2019; Hosakova, Viktorova, Lecbych, & Hosak, 2017), translating and using the measure in other countries without examining its psychometric properties is not recommended and might bias the interpretation of results. Therefore, the first goal of this study was to identify the factor structure of the Albanian version of CAINS. We expected to find the two-factor structure as in previous studies (Horan et al., 2011; Kring et al., 2013). The second goal was to extend previous research and examine the relations between CAINS and Brief Symptom Inventory (BSI, Derogatis, 1993) that covers psychological distress and psychopathology. Based on the previous studies, we expected to find a positive relation between individual factors measured by CAINS and BSI scales.

METHODS

RESULTS

Table 1 presents the number and percentage of patients with each diagnosis type.

Table 2 displays descriptive statistics of the BSI subscales. All sample-wide subscale means were greater than 1 point, except for Hostility, which has a mean of 0.72.

As Table 3 shows, all values of Cronbach’s alpha coefficient were greater than 0.700. Therefore, all BSI scales could be considered internally consistent.

According to Kaiser-Guttman’s criterion (i.e., eigen-value >1), four components were extracted. They accounted for 74.69% of manifest data variance. Additionally, Promax rotation was applied, allowing components to correlate with each other.

Cattell’s scree plot could give additional justification for the four-factor solution. It is apparent that the most pronounced elbow of the plot resides at four components (see Figure 1).

Figure 1

In Table 5, all factor loadings of CAINS items on primary components are greater than 0.35 and less than this value on other components. Hence, the obtained component (factor) solution seems interpretable and satisfactory. The first extracted component was Expression and includes items 10 to 13. The second one was Social Motivation and Pleasure (MAP), operationalized by items 1 to 4. The next one was Recreational Motivation and Pleasure (items 7 to 9) and the last extracted component was Work and School MAP (items 5 and 6). In accordance with the results of PCA, it turned out that CAINS could be regarded as a valid measure of motivation and pleasure in various life domains (social, recreational, and work/school domains).

Results shown in Table 6 reflect a very good internal consistency of not only each of the CAINS scales, but also of CAINS inventory itself (α = 0.860).

In Table 7, descriptive statistical values of CAINS subscales are presented. In addition, Table 8 contains CAINS scales’ mutual correlations. The strongest relationship was that of Social Motivation and Pleasure with Work & School MAP (r = 0.498, p < 0.001). All coefficients are statistically significant, indicating that all CAINS scales correlate with one another.

DISCUSSION

The purpose of this study was to investigate the psychometric properties of Clinical Assessment Interview for Negative Symptoms (CAINS) in the Albanian language. Moreover, the current study was aimed to identify the relationships between factors derived from CAINS and several dimensions of Brief Symptoms Inventory (BSI). We expected to find the two-factor structure, as previously reported by Chan et al. (2015), Valiente-Gomez et al. (2015) and Engel et al. (2014); however, our results did not support such a hypothesis. In contrast to previous studies that found Expression and Motivation and Pleasure as two major factors, our results revealed four components: expression (four items), social motivation and pleasure (four items), recreational motivation and pleasure (three items), and work and school motivation and pleasure (two items). Our findings replicated the component of expression; however, the items on motivation and pleasure were grouped clearly into three different components. The internal consistency of these four subscales and of the overall measure was very good, ranging from 0.78 to 0.90.

Our finding of the four-factor structure can be explained by the nature of reporting for CAINS assessment of symptoms. For example, the dimension of expression has a greater number of items observed by clinicians, whereas anhedonia and asociality tend to be patient-reported, and can be grouped into more than one domain (Elis, 2013; Rekhi et al., 2019). Motivation and pleasure may be different for different types of activities, although this is speculative at this stage. For example, people diagnosed with psychotic disorders may indicate different levels of motivation and pleasure for work and school compared with the level of the motivation and pleasure they may experience in recreational activities.

In addition to the results of our study, there is also an increasing body of evidence for more than two factors in the structure of CAINS in other cultures. For example, very similar results were reported in the factor structure of CAINS in Singapore (Rekhi at al., 2019), where the same components were named as expression, vocational MAP, recreational MAP, and social MAP. Thus, it may be that culture plays a substantial role in factor structure of CAINS, since most of the exploratory studies of CAINS so far come from Western cultures. Findings from other cultures might be related to different appraisal of emotional situations, which are influenced by different socio-cultural practices (Rekhi, et al., 2019). As an example, Kosovo might be considered as a collectivist society, which tends to assess emotions in terms of social worth, taking external factors into account rather than just one’s inner world. In Kosovo and in other similarly collectivistic societies, emotions are evaluated more often in terms of self-other relationships rather confined to subjectivity (Mesquite, 2001). This type of emotional evaluation, however, highlights one of CAINS main limitations, which is the tendency to evaluate functioning rather than negative symptoms (Garcia-Portilla et al., 2015).

Further support for our findings comes from studies that critically analyse the two-dimensional model of assessments that measure negative symptoms (Ahmed et al., 2018; Blanchard & Cohen, 2005). These studies not only suggest a multidimensional structure, but conclude that the two-factor solution has been summarized too early, before rigorous, complete analyses could be done. Moreover, evidence for the two-factor solution comes from studies that have used only exploratory factor analysis (EFA), which is known for the limitations, especially the fact that researcher has a priori assumption that any indicator may be associated with any factor. Ahmed et al. (2018) and Rekhi et al. (2019) aimed to improve this limitation and found five- and four-factor structures, respectively. It is important to mention that the study by Ahmed and colleagues (2018) was conducted with samples from different cultures and languages.

Strauss and colleagues (2018) carried out a confirmatory factor analysis (CFA) in order to, among other things, check the construct validity of CAINS. Their analysis confirmed the existence of the following five first-order factors: Anhedonia, Asociality, Avolition, Blunted affect, and Alogia, corresponding to the DSM-5 conceptualization of negative symptoms. The second-order factors were MAP and EXP. However, the discrepancy between our results and those of Strauss and colleagues can be explained by several factors. First, the authors assumed that the latent structure of the CAINS could be best described by the five aforementioned dimensions (despite the authors of CAINS structuring the items of the instrument in a different way). Second, one of the factors (Alogia) saturated only one item (which indicates that it is a so-called trivial factor). Third, the authors did not test the four-factor model fit (although CAINS actually consists of four parts). Fourth, the sociodemographic background of their participants was quite different compared with the sample in our study. Finally, their sample was comprised of patients diagnosed with schizophrenia, whereas our sample included patients with psychotic disorders (which is a broader category).

In addition to the factor structure, the convergent validity of CAINS was shown by positive association of two factors (Social MAP and Vocational MAP) with all subscales of BSI. Surprisingly, we did not find correlations between expression and recreational MAP with the BSI subscales. This could be taken as indication that expression and recreational MAP can distinguish negative symptoms from other symptoms that might relate to specific disorders or symptoms of depression or anxiety.

This study has several limitations. First, the sample of this study could be considered small, which decreases the statistical power, and might have affected and the factor structure results in turn. Thus, replication with larger samples is necessary to generalize the present findings. Another limitation is the lack of other negative symptoms’ assessments, which could be used to assess its convergent validity. In addition to that, we did not use the interrater reliability, due to limitations in time and resources. As such, it will be important for future studies to address these limitations and further explore the factor structure of negative symptoms among psychotic patients.

This study provides relevant implications for research, policy, and practice. The study confirms that CAINS is a valid measure of motivation and pleasure in various life domains (social, recreational, and work/school) in the Albanian language. As such, this measure can be used in studies that aim to investigate the relationships between negative symptoms and other important mental health outcomes in Kosovo and across other Albanian speaking countries, where mental health research is still scarce. Replication of the current findings with larger samples would represent an important step in confirming the CAINS factor structure in the Albanian language. Future studies using CAINS can be used to understand the factors that influence negative symptoms among people diagnosed with psychotic disorders, which in turn may result in better strategies and intervention programs. By using CAINS to obtain data about negative symptoms, clinicians can develop and support structured and clinician-supervised activities for psychotic patients, aiming to decrease their negative symptoms and enhance their quality of life. Our findings on the four-factor structure of CAINS could be considered as a detailed and broad assessment of psychotic patients, which may lead to effective and more specific psychiatric treatment and psychotherapy for psychotic patients.