Abstract

۱٫ Introduction

۲٫ Methods

۳٫ Results

۴٫ Discussion

Funding statement

References

Abstract

Background
     A challenge for clinicians working with individuals diagnosed with schizophrenia is distinguishing depressive symptoms from negative symptoms of schizophrenia. The Calgary Depression Scale for Schizophrenia (CDSS) was developed for this purpose. No review has previously explored its reliability across multiple studies using advanced statistical means.

Objectives
     This meta-analysis aimed to quantify the CDSS’ internal consistency, inter-rater reliability (IRR) and test-retest reliability.

Method
     A systematic literature search was conducted to find articles reporting on the CDSS’ reliability. Articles were screened against the inclusion and exclusion criteria, with data extracted from 40 studies. Overall meta-analytic effects were calculated, and for internal consistency and IRR coefficients subsequent analyses explored between-study variation. The small test-retest reliability dataset limited analysis.

Findings
     The internal consistency meta-analytic effect was 0.83 (95% CI:0.82–۰٫۸۴). Higgins I2 indicated an acceptable level of variation between studies’ alpha estimates. This suggests all items in the CDSS are measuring the same construct (i.e. symptoms of depression). The IRR meta-analytic effect was 0.88 (95% CI:0.86–۰٫۹۱), with Higgins I2 indicating high levels of heterogeneity. This was not deemed problematic variance as it is within levels expected for psychometric measures and, therefore, considered acceptable for this literature. This reflects high level of agreement between different raters when using the CDSS on the same client.

Conclusions
     This review suggests the CDSS has good internal consistency and excellent IRR. Further research will help understand its test-retest reliability.

Introduction

     Schizophrenia is associated with a range of symptoms, typically separated into positive and negative. Positive symptoms are associated with the individual’s perception or interpretation of stimuli being different from others, alongside difficulties distinguishing their thoughts and ideas from reality. Negative symptoms include loss of motivation, apathy, impaired concentration, flattening of emotions and reduced speech (Cuesta et al., 2009).

     It is widely accepted that mood disturbances are often observed alongside a diagnosis of schizophrenia (Rector et al., 2005; van Os et al., 2000). This includes mood disturbances experienced concurrently and independently from the psychotic symptoms (Birchwood et al., 2000). Negative symptoms, however, overlap with symptoms of depression, posing a clinical challenge of distinguishing negative symptoms of schizophrenia (i.e. difficulties with motivational state) from depression (i.e. difficulties with pervasive low mood).

Results

     Results 3.1. Study characteristics Table 2 presents the characteristics of the 39 studies included in the meta-analysis. Of the 40 studies, 27 were included for the CDSS’ internal consistency and 28 were included for the IRR of the CDSS. Eight of these studies also reported test-retest reliability data, however, due to limitations within this dataset the analysis was restricted (further information in Sections 3 and 4).

۳٫۲٫ Risk of bias of individual studies Findings of meta-analyses can be impacted by including poor quality studies (Higgins et al., 2011). Quality captures how appropriate the study is for answering its research question, considering design, delivery and analysis. There are various tools for assessing risk of bias. Higgins et al. (2011) advocate using a set of criteria specific to methodological issues pertinent to the literature and question under review. Assessment of risk of bias, therefore, was completed using a framework developed for this review (Table 3). Existing tools and information on types of bias guided the framework’s development (Higgins et al., 2011; Smith and Noble, 2014).

Risk of bias ratings were primarily made by the lead author (LP), who also completed the process of data extraction. When there was ambiguity within the data which complicated the decision as to whether there was a low, unclear or high risk of bias, a discussion was had between the authors (LP, CJ and AF) to enable the final decision to be made by consensus.