Examination of categorical approach to symptom assessment: cross-validation of foulds’ Delusions-Symptoms-States Inventory with Korean non-patient and patient groups

Rich variations in psychopathology of schizophrenia have always presented challenges for clinicians who have predominantly relied on descriptive psychopathology for diagnosing and evaluating schizophrenia. Ironically, most traditional conceptualization and nomenclature of schizophrenia have yielded little room to adequately account for the evolving and shifting phenomenology of the illness. For example, given a history of an 'active phase’ of psychotic symptomatology, a diagnosis of schizophrenia precedes other considerations even in the absence of apparent psychotic features [1]. While subtypes, such as 'in remission’ or 'residual’ could be assigned, their diagnostic reliability, validity, and clinical utility have shown to be largely inadequate and hence have been eliminated in DSM-5 [2].

Characterization of the schizophrenia as either/or dichotomy following the disease entity model prevails in both clinical practice and research. However, a notion of continuum from normality to psychotic illness has been articulated for over a century [3], though it has not been rigorously pursued except by a group of British researchers [1]. The severity of illness can be conceived and is embedded in the hierarchical path along which psychiatric diagnose is usually made - in the ascending order of anxiety and other neuroses, depressive neurosis, manic and paranoid psychoses, schizophrenia, and then organic psychosis [4]. Perceivably, such conceptualization of illness is more likely consider the shifting nature of the illness rather than diagnosis, which may have some important implications for treatment and prevention. Individuals with characteristic attenuated psychotic syndrome are several 100-times more likely to develop psychotic disorder in the next year than the general population and present current mood and anxiety symptoms [5]. Nonetheless, a significant proportion of such individuals are identified and provided with interventions much later to prevent substantial damage [6, 7].

On the other hand, categorizing patients according to symptom clusters that reflect the level of illness severity along the hierarchical path may hold distinct advantages over other approaches. For example, dimensional approaches to evaluate the presence and severity of illness may better reflect the heterogeneity of the patients but they present difficulties in establishing the range of normality and inadequacies in accounting for the individual differences in 'symptom complexes’ and their prognostic value [8].

Overall, a taxonomic system that better reflects heterogeneity and fluctuating nature of illness within a continuum may deserve more in-depth examination for its validity and applicability in both clinical practice and research. Accordingly, a self-report measure developed to assess the degree of adverse change in personal functioning and severity of clinical disorder in a hierarchical arrangement has been proposed previously by Foulds and Bedford [9]. The central notion is that starting with those without significant level of illness, or not personally ill (NPI), patients could be categorized according to a four-class hierarchical model of the severity of mental illness as follows: dysthymic states (DS) consisting of manifestations of states of anxiety, depression, and/or elation; neurotic symptoms (NS) involving conversion, dissociation, phobia, compulsion, and rumination; integrated delusion (ID) with delusions of persecution, grandeur, and/or contrition; and delusions of disintegration (DD) representing a single syndromal group corresponding to schizophrenia with considerable extent of personal disintegration as an agent of his/her own feelings, thoughts and actions [10, 11]. The central hypothesis of this model is inclusive, non-reflexive relationship among the classes, which simply posits that the membership in the higher class also entails membership in the lower ones and that the converse does not hold. Hence, for example, every patient in ID class has both neurotic symptoms and dysthymic states but not delusions of disintegration [12].

The rate of concordance to the model has been found to reach above 90% in the majority of studies [13], with only a few studies that included narrow diagnostic class, such as chronic inpatients [14] and manic and hypomanic patients [15], showing low percentages of about 73%. The concordance rate of one-month retest by the original authors of the DSSI reached above 90% [16]. Despite such robust findings regarding the validity of the hierarchical model, cross-cultural validation and feasibility of the model have not been actively examined.

In the present study, we examined the validity of the DSSI hierarchical class model using both Korean non-patient and patient (schizophrenia and depression) groups by conducting cross-validations with other clinical measures. We tested the hypothesis of inclusive non-reflexive relationships among the classes, i.e. hierarchical organization according to the severity of illness, assessed the power of the DSSI to detect presence of symptoms via cross-validation with other clinical measures, and examined the differences in the clinical features among the DSSI classes using the Minnesota Multiphasic Personality Inventory (MMPI-2).

In this study, the high rate of model conformity across the samples and cross-validation with other criterion measures provided empirical support for the DSSI hierarchical class organization and demonstrated its potential usefulness as a self-report measure for assessment of presence and severity of symptoms.

Our finding on the DSSI model conformity based on all subjects (91.1%) was consistent with the majority of the past studies [13]. The model conformity of the schizophrenia group (84.3%) was also comparable to those of previous studies including chronic inpatients which ranged from 73.8% to 85.0% [14, 25, 26]. On the other hand, the conformity rate of our depression group (82.5%) was lower than that of one other study reported on depression group involving only inpatients (92.3%) [27]. Among the nonconforming patients, 71% had duration of illness of longer than 5 years and this seemed to well correspond to the overall decline in the rate of model conformity with chronicity of illness. The nonconforming depression patients were all categorized within DI and there may be a distinct possibility that some of them may have major depression with psychotic features. Such patients have been found to have longer duration of episodes and greater recurrence than those with nonpsychotic depression [28]. Nonetheless, further examination of the relationship between the duration of illness and model conformity is warranted for the depression patients due to a relatively small sample size in our study.

A closer examination of the non-conforming patterns yielded an interesting observation that of 26 non-conforming participants, the overwhelming majority (23 participants including 5 of the NS class) did not report the symptoms of the adjacent subordinate class. Angelopoulos and Economous also reported a comparably high proportion (32 out of 47 non-conforming participants including 7 of the NS class) that did not report the symptoms of the adjacent subordinate class among the inpatients with schizophrenia [11]. Such results suggested that it may be more common for symptoms of one class to overshadow those of the adjacent subordinate class than of any other subordinated classes, which may be an important addition to Foulds’ suggestion that symptomatology from the higher class may mask, confuse, override, or distract from that in the lower classes [29].

The majority of past studies on the distribution of the DSSI classes have been conducted on general psychiatric patients and those involving specific psychiatric groups all dealt exclusively with inpatients except for one study that included day-patients [30]. This was in keeping with the notion that the hierarchical model is applicable only to acute patients [13, 26]. Hence, no other study has included a large number of outpatients with chronic schizophrenia, not to mention non-patients – apart from that by the original authors of the DSSI [10]. The results of our study however suggested that the DSSI can be applied as effectively to both subclinical stabilized chronic illness group and non-patients as a screening measure for presence of active symptoms. The sensitivity and specificity of the DSSI in detecting the presence of symptoms among schizophrenia and depression patients was comparable to those of MMPI-2 clinical scales in light of symptom remission criteria of HAMD-17 and PANSS, all of which require more time and trained staff.

Lastly, there are a number of important issues that should be raised prior to generalizing the results of our study. First, our results need to be considered in light of the Berkson’s fallacy: since our non-patient group has been screened with SCID-I, whilst none had been excluded by this process, there is a possibility that the composition of our non-patient group may actually under-represent the individuals with more severe levels of illness [31]. Also, the clinical group was not screened for comorbidity of psychiatric illness, not to mention alcohol dependence, with DSM-IV by the psychiatrists. Furthermore, inclusion of other diagnostic groups, such as anxiety or personality disorders, was not considered in our study design. Since such factors may have significant affect on the conformity of the model, future studies using broader range of diagnose and more refined study design are warranted to lend further support for the validation of the DSSI.

In summary, the results of the present study provided further support for the utility of the DSSI in covering a relatively wide range of psychiatric symptom spectrum. It may be usefully applied for not only the detection of presence of psychiatric symptoms among non-patient groups but also assessment of alleviation or aggravation of illness in patients. The DSSI can thereby provide complementary information for diagnosis, determination of the treatment efficacy, and decisions concerning maintenance of treatment strategy.