The extent and origin of discordance between self- and observer-rated depression in patients with psychosis

Abstract

It is assumed that patients with psychosis have difficulties indicating clinical symptoms accurately in self-reported measures. The present study investigated the ability of self-rating scales to detect symptoms of depression in patients with psychosis and aimed at identifying demographic, clinical and neurocognitive factors that predict the discordance between self-ratings and observer ratings. Inpatients and outpatients with psychosis (n=118) were assessed for depression by applying two observer rating and two self-rating scales. We found reasonable correlation scores between the ratings by patients and observers (range: r=0.50–0.57). In half of the patients (49.2%) the self-ratings corresponded well with the ratings of clinicians. Patients who rated their depressive symptoms as less severe than the clinicians demonstrated more negative symptoms such as blunted affect and poor affective rapport. Patients who rated their depression symptoms as being more severe were characterized by more self-reported general psychopathology. The concordance rates indicate that self-ratings of depression can be a valid additional tool in clinical assessment of patients with psychosis. However, clinicians should be attentive to the fact that some patients might have a general tendency to over-report symptoms and that patients with negative symptoms tend to be rated as more depressed in observer ratings compared with self-assessments.

Introduction

Symptoms of depression are common in patients with psychosis and can occur in every phase of the disorder (Heald et al., 2008). Reported prevalence rates range from 30% (Siris, 1991) to 70% in an acute phase of psychosis (Birchwood et al., 2000). Previous studies have demonstrated that depressive symptoms in patients with psychosis are associated with low subjective quality of life or well-being (Reine et al., 2003, Schennach-Wolff et al., 2011), in higher symptomatology (Chemerinski et al., 2008, Rocca et al., 2005), increased risk of relapse (Johnson, 1988) and increased suicidal ideation (Heilä et al., 1997).

It is therefore important to be able to identify depressive symptoms via a reliable and valid diagnostic procedure. In patients without psychosis, symptoms of depression are typically assessed via self-rating scales (e.g., Beck Depression Inventory (BDI); Beck et al., 1961). However, it is assumed that patients with psychosis have difficulties in describing their symptoms accurately that might be due to denial, shame, poor insight and information processing deficits in attention, concentration, memory, abstraction and concept formation (Hamera et al., 1996). As a consequence, observer-rated symptom scales are a common method of assessing the severity of depressive symptoms in patients with psychosis and most of the assessment research has focused on observer-rating scales (Addington et al., 1996, Collins et al., 1996, Kontaxakis et al., 2000, El Yazaji et al., 2002). However, observer ratings are time-consuming, require intensive training, are costly as well as prone to socially desirable answers and may be observer-biased. In contrast, self-ratings are efficient and could enhance therapeutic outcome and alliance (Liraud et al., 2004, McCabe et al., 2007) because they provide a means to assess and value the subjective well-being in therapy which might render improvement of depressive symptomology more likely (Schennach-Wolff et al., 2011).

So far, two studies have compared self-rating with observer-rating scales and found them to correlate moderately to highly (range from r =0.53 to r =0.84) (Addington et al., 1993, Kim et al., 2006). These studies have not, however, provided answers to the question of which factors might be responsible for the remaining discordance between clinicians’ and patients’ estimates of symptoms. Such factors have only been investigated in non-psychotic patients with depression (Domken et al., 1994, Enns et al., 2000). Both studies found several patient characteristics such as age, specific personality factors such as high neuroticism, and low self-esteem, to predict the disparities between self-ratings and observer ratings. With regard to patients with psychotic disorders, factors that explain discrepancies are likely to be found in the specific characteristics of these patients. Patients with psychosis might either under- or overestimate their symptoms due to lack of insight (Lincoln et al., 2007), higher levels of interpersonal distrust (Andreasen and Flaum, 1991), low self-esteem (Kesting et al., 2011), lack of cognitive flexibility (Beck et al., 2004) or cognitive impairment (Heinrichs and Zakzanis, 1998). On the other hand, the discrepancy could also be caused by biases on the side of the clinicians. For example, clinicians might underestimate symptoms of depression in patients with higher levels of positive symptoms, in particular those who appear agitated or aggressive. Similarly, clinicians might incorrectly interpret the diminished emotional expressions that are associated with negative symptoms or medication side effects (e.g., extrapyramidal effects; Müller et al., 2002) as symptoms of depression. However, it needs pointing out that the phenomenological similarity of negative and depressive symptoms makes it difficult to differentiate these syndromes. This makes it difficult to attribute an overlap of self- and observer-reported symptoms entirely to a rating bias (Chemerinski et al., 2008).

The present study therefore aims at identifying the clinical and neurocognitive variables as well as measurement-related factors that are likely to be responsible for self-observer discrepancies in the assessment of depression in patients with psychosis.