Schizophrenia remains the most common form of psychotic illness, affecting 1% o f the population (Andreasen, 1999). The standard compartmentalisation of schizophrenia, as with most illnesses (DSM), implies that the illness is categorically delineated. Recent evidence from large non-clinical samples, however, suggested that the disorder is not easily and/or consistently distinguishable from schizotypal and schizoid personality disorders (Siever, Kalus, & Keefe, 1993) and similarly overlaps with bipolar disorders (Craddock & Owen, 2007). These shared characteristics have led several researchers to question the conceptualisation of schizophrenia as a discrete illness entity (Claridge, 1994; Crow et al., 1995; Johns & van Os, 2001; McGovern &
Turkington, 2001; Strauss, 1969). Indeed, although this approach has only relatively recently gained ground in psychiatry, it has been widely applied throughout medicine.
Rose (1992) recognised that virtually all pathophysiological factors examined were continuously distributed throughout the population, and as a result advocated prevention strategies that target the population as a whole, an idea that remains influential (Manuel et al., 2006).
Evidence in support o f the argument for a spectrum of schizophrenic disorders comes from findings suggesting that certain common cognitive deficits (e.g., deficits in attention, abstract reasoning, cognitive inhibition, verbal working memory, recognition memory, and general intellectual functioning) and neural differences (e.g., the total volume of the left dominant posterior superior temporal gyrus [STG] relating to delusion scores and grey matter reduction in the left posterior STG relating to inverse thought disorder scores) exist in individuals with schizotypal personality disorder to a moderate extent but in schizophrenia to a greater extent (Cadenhead et
al., 1999; Menon et al., 1995; Shenton et al., 1992). Further research has focused on similarities between the known correlates of clinical symptoms, and those of their subclinical counterparts.
A. Similar correlates fo r subclinical and clinical symptoms
a. Demographics
One assumption of the continuum account is that subclinical psychotic symptoms should associate with known correlates of clinical symptoms (e.g., demographics). Indeed, positive subclinical symptoms (i.e., symptoms people do not usually experience: delusions, hallucinations and thought disorder) have been found to be associated with negative subclinical symptoms (i.e., the lack of normal traits:
e.g., flat affect or avolition) (van Os et al., 2000) and both have also been associated with depressive symptoms (Stefanis et al., 2002), thus reflecting the dimensions found within schizophrenia. In addition, a recent large meta-analysis revealed that demographic factors relating to schizophrenia (e.g., males, unmarried, unemployed, ethnic minorities) also relate to subclinical symptoms, with the exception of age, where the results are difficult to interpret (van Os et al., 2009). Moreover, the meta
analysis also found that other known risk factors, such as urbanicity, trauma and cannabis use, were associated with higher levels o f subclinical psychosis (Henquet et al., 2005; Krabbendam & van Os, 2005; Read et al., 2005).
As might be expected following a continuum account, measures of clinical and subclinical experiences have different degrees of association with demographic variables (van Os et al., 2000) and also different increased risks for developing the full-blown clinical disorder (Hanssen et al., 2005). Furthermore, van Os et al. (2001)
have demonstrated that the levels of psychotic experiences observed in the general population can predict the prevalence of disorder. Utilising the association between psychotic disorder and urbanicity, they used five samples grouped by degree of urbanicity to show that as the rate of psychotic disorder increases with urbanicity, the levels o f reported psychotic experiences also increased in a dose-response manner.
b. Genetics
Another way o f looking at the relationship between clinical and subclinical symptoms is by examining the genetic risk factors. Several studies of twins in the general population strongly support genetic links, implying that both genetic and environmental factors play roles in the presentation of psychotic symptoms (Kendler
& Hewitt, 1992; Linney et al., 2003; MacDonald et al., 2001). In terms of cognitive deficits, children of patients with schizophrenia are often found to have impaired verbal memory and deficits with other cognitive tasks (Owens & Johnstone, 2006).
Furthermore, a recent meta-analysis by Sitskoom et al. (2004) showed that first- degree relatives of patients with psychotic disorders had minor difficulties with verbal memory, executive functioning and to some degree with attention.
Similar results have been found for positive symptoms. Kendler et al. (1993) found that psychosis phenotypes (clinical and subclinical expression) tended to cluster in families. In the general population, Hanssen et al. (2006) used both self-report and interview measures to assess positive and negative subclinical psychosis within families, and found familial clustering for both dimensions. Similarly, the types of symptoms reported by patients often predict the expression of subclinical experiences in their relatives (Fanous et al., 2001), and the positive symptom scores of relatives of patients with psychotic disorders typically relate to their genetic risk (Vollema et al., 2002).
B. The presence o f similar beliefs and experiences in non-clinical populations Clinical psychosis remains comparatively rare, with one recent study estimating lifetime prevalence of broadly defined psychosis at 3.48% (Perala et al., 2007). By comparison, the continuum account suggests that “the core symptoms of psychosis, delusions and hallucinations, are much more prevalent in the general populations than their clinical counterparts” (Krabbendam et al., 2004, p.411). This hypothesis is particularly interesting given the lack of a definitive demarcation in diagnosis, e.g., determining where schizotypal disorder becomes schizophrenia. To avoid these limitations inherent in diagnosing syndromes, some researchers adopt a cognitive neuropsychiatric perspective, which focuses on symptoms rather than medically or psychiatrically labelled syndromes.
The difficulties with drawing absolute boundary distinctions apply to all fields of clinical practice, however. Even when addressing a single symptom, diagnosis remains a dichotomous choice, while the presenting symptom exists to varying degrees across the population. Indeed, this is independent of the presentation of the illness, and holds whether symptoms are predominantly physical (e.g., obesity) or predominantly psychological, e.g., autism spectrum disorders (Newschaffer et al., 2007).
Indeed, whilst the definition o f a delusion in DSM-IV-TR (APA, 2000) supports a clear categorical distinction, ‘A false belief based on incorrect inference about external reality ’, the glossary now states that ‘delusional conviction occurs on a continuum and can sometimes be inferredfrom an individual’s behaviour' (p. 821). In fact, even Jaspers, whose work is often cited as support for the distinction between normal beliefs and delusions, appeared to consider at least a subset of delusions as continuous in some ways, suggesting that a ‘jealous man can develop into a man with
delusion-like jealousy ’ and 'a suspicious person into someone with delusion-like ideas o f persecution' (p. 640). Moreover, despite the historical dominance of categorisation, criticisms o f this approach are not new - Bleuler (1911) made similar arguments against assuming categorical divisions between ‘healthy’ and ‘ill’ individuals.
As such, a growing number of researchers agree that individual psychiatric symptoms (including beliefs) lie on a continuum where only a small number located at some (arbitrarily defined but clinically agreed) extreme endpoint become clinically relevant (i.e., delusions) and where much o f the distribution is not necessarily associated with any significant disability (Johns & van Os, 2001; McGovern &
Turkington, 2001; Rutten et al., 2008; Strauss, 1969; van Os et al., 2009). It is worth noting that such a continuum is not simply due to variation within a single factor (e.g., conviction, as suggested by the DSM description). People may differ in terms of the frequency, intensity and number of symptoms they present as they vary over the continua. Moreover, research has indicated that the conviction with which a delusional belief is held fluctuates over time and between contexts (Myin-Germeys et al., 2001). As Claridge (1997) points out, however, two people may have the same psychotic symptoms but one may require care and the other may not, as they may use a different coping strategy; for example, non-patients may be more likely to perceive hallucinated voices as predominantly positive (Honig et al., 1998).
Strong evidence in favour of the continuum hypothesis comes from general population studies, which estimate lifetime prevalence for delusions at around 15%
(Rutten et al., 2008, p. 53) and an average annual prevalence rate of 5% (van Os et al., 2009). The consistent finding that many non-clinical participants endorse questions relating to both delusions and hallucinations implies that “experiencing symptoms of psychosis such as delusions and hallucinations is not inevitably associated with the
presence o f disorder” (van Os et al., 2009, p. 1). Clearly other factors, such as intrusiveness, psychopathological co-morbidities, illness behaviour, societal tolerance, coping and distress, play a significant role in the clinical relevance of delusional beliefs, and this continuum approach underpins much of modem cognitive therapy for psychosis (Johns & van Os, 2001).
To date, most of the single-symptom studies have focused on hallucinatory experiences (e.g., Johns et al., 2002; van Os et al., 2000). Nonetheless, as delusions are often assumed to be a critical aspect of psychosis and a pathological form of belief (Davies et al., 2001; Langdon & Coltheart, 2000; van Os, 2003), research has begun to open up this rich vein o f study by examining “delusional ideation” or what might be best described as “delusion-like beliefs” in the non-clinical population (e.g., Lincoln, 2007; Peters, Joseph & Garety, 1999; Peters et al., 2004; Verdoux et al., 1998). These studies provide compelling evidence that delusion-like beliefs (i.e., beliefs that have a similar content to delusions but are not associated with the significant behavioural and/or psychological consequences found with delusions) are more commonly present in non-clinical populations than previously expected.