258

by Leslie C. Morey and Roger K. Blashfield

A Symptom Analysis of the DSM-III Definition of Schizophrenia

Abstract With the recent introduction of DSM-III has come a renewed interest in psychiatric classification. The development of specific criteria for diagnosis was designed to objectify heretofore suspect diagnostic procedures. However, this potential objectivity may be offset by shortcomings in the specific symptoms that define the syndrome. The experiment reported examined the DSM-III definition of schizophrenia by analyzing the characteristics of the individual symptoms. Certain problem areas are noted, and recommendations are made as to how the definition might be strengthened. The definition of schizophrenia has been a source of controversy since the concept was popularized by Kraepelin (1919) and Bleuler (1950). Each of these writers offered different definitions of the disorder, as have numerous contemporaries (e.g., Schneider 1959; Meehl 1962). These definitional issues have gained renewed interest recently because of the introduction of DSM-HI (American Psychiatric Association 1980). DSM-III represents a radical departure from earlier editions of this diagnostic manual in that it mandates the use of specific diagnostic criteria for decisionmaking. In doing so, it defines disorders by specifying the presence (or absence) of various signs and symptoms. Previous studies that used similar types of criteria succeeded in improving diagnostic reliability (e.g., Feighner et al. 1972). The improvement was accomplished by creating a relatively explicit definition of a syndrome. That DSM-III

is more explicit than its predecessor, DSM-II (American Psychiatric Association 1968), can be seen by examining the respective definitions of schizophrenia presented in table 1. When the two definitions are compared, DSM-HI appears clearer in establishing the boundaries of schizophrenia. The increase in specificity allows an evaluation of DSM-III in ways that were not possible with earlier editions. Until now, most research in the classification of psychopathology has focused on either diagnostic reliability across syndromes, or has attempted to establish new classifications based on factor or cluster analysis. The vague definitions of disorders in DSM-II made it difficult to study the classification system itself, so the focus was primarily on diagnostic agreement. Reliability was frequently poor, and most researchers attributed this problem to inadequacies in the diagnostic system (e.g., Beck et al. 1962). The developers of DSM-III have attempted to remove this source of inaccuracy by specifying certain signs and symptoms that must be present to make a diagnosis. Thus, a basic and implicit assumption in DSM-III is that these symptoms are reliable and clearly defined. In effect, once a clinician has decided on the presence of symptoms, DSM-III makes the diagnostic process automatic. As a result, it is crucial that these symptoms be both reliable and valid. The present study is novel because it attempted to evaluate some of the properties of the DSM-III by examining the compoReprint requests should be sent to L.C. Morey at 4402 Newcome, San Antonio, TX 78229.

VOL. 7, NO. 2, 1981

259

Table 1a. DSM-III diagnostic criteria for schizophrenia A. Characteristic schizophrenic symptoms: At least one symptom from any of the following 10 symptoms was present during an active phase of tHe illness (because a single symptom is given such diagnostic significance, its presence should be clearly established): Characteristic delusions: (1) Delusions of being controlled: Experiences his thoughts, actions, or feelings as imposed on him by some external force [SA1 ] (2) Thought broadcasting: Experiences his thoughts, as they occur, as being broadcast from his head into the external world so that others can hear them [SA2] (3) Thought insertion: Experiences thoughts, which are not his own, being inserted into his mind (other than by . God) [SA3] (4) Thought withdrawal: Belief that thoughts have been removed from his head, resulting in a diminished number of thoughts remaining [SA4] (5) Other bizarre delusions (patently absurd, fantastic, or implausible) [SA5] (6) Somatic [SA6a], grandiose [SA6b], religious [SA6c], nihilistic [SA6d], or other delusions without persecutory [SA6ex] or jealous content [SA6ey] (7) Delusions of any type if accompanied by hallucinations of any type [SA7] Characteristic hallucinations: (8) Auditory hallucinations in which either a voice keeps up a running commentary on the individual's behaviors or thoughts as they occur [SA8a], or two or more voices converse with each other [SA8b] (9) Auditory hallucinations on several occasions with content having no relation to depression or elation, and not limited to one or two words [SA9] Other characteristic symptoms: (10) Either incoherence [SA10a], derailment (loosening of associations) [SA10b], marked illogicality [SA10c], or marked poverty of content of speech [SA10d]—if accompanied by either blunted [SA10e], flat [SA10f], or inappropriate affect [SA10g], delusions [SA10h], or hallucinations [SA10i], or behavior that is grossly disorganized [SA10J], or catatonic [SA10k] B. During the active phase of the illness, the symptoms in A have been associated with significant impairment in two or more areas of routine daily functioning, e.g. work, social relations, self-care [SB] C. Chronicity: Signs of the illness have lasted continuously for at least 6 months at some time during the person's life, and the period must include an active phase during which there were symptoms from A with or without a prodromal or residual phase, as defined below [SCx] Prodromal phase: A clear deterioration in functioning not due to a primary disturbance in mood or to substance abuse, and involving at least two of the symptoms noted below. Residual phase: Following the active phase of the illness, at least two of the symptoms noted below, not due to a primary disturbance in mood or to substance abuse. Prodromal or residual symptoms: (1) Social isolation [SC1a] or withdrawal [SCib] (2) Marked impairment in role functioning as wage-earner, homemaker, student [SC2] (3) Markedly eccentric, odd, or peculiar behavior (e.g., collecting garbage, talking to self in cornfield or subway, hoarding food) [SC3]

260

SCHIZOPHRENIA BULLETIN

Table 1a.—continued (4) Impairment in personal hygiene and grooming [SC4] (5) Blunted, flat, or inappropriate affect [see SA10e, SA10f, & SA10g] (6) Speech that is tangential [SC6a], digressive [SC6b], vague [SA6c], overelaborate [SC6d], circumstantial [SC6e], or metaphorical [SC6f] (7) Odd or bizarre ideation [SC7a], magical thinking, e.g., clairvoyance, superstitiousness, telepathy [SC7b], "sixth sense," "others can feel my feelings," overvalued ideas [SC7c], ideas of reference [SC7d], or suspected delusions [SC7e] (8) Unusual perceptual experiences, e.g., recurrent illusions, sensing the presence of a force or person not actually present, suspected hallucinations [SC8] Examples: Six months of prodromal symptoms with 1 week of symptoms from A; no prodromal symptoms with 6 months of symptoms from A; no prodromal symptoms with 2 weeks of symptoms from A and 6 months of residual symptoms; 6 months of symptoms from A, apparently followed by several years of complete remission, with 1 week of symptoms in A in current episode D. The full depressive or manic syndrome (criteria A and B of Depressive or Manic Episode) is either not present [SDa], or if present, developed after any psychotic symptoms [SDb] E. Not due to any Organic Mental Disorder [SE] Source: Diagnostic and Statistical Manual of Mental Disorders, Third Edition (1/15/1978 draft, pp. C 10-12).

nent bits of information that are used to make a diagnosis. Specifically, clinicians were asked to rate the extent to which DSM-HI defined symptoms that contribute information toward making a diagnosis. Because a systematic examination of every bit of information contained in DSM-HI was not considered feasible, three diagnoses were selected for intensive study.

Two of these diagnoses, mania and dementia, were disorders whose DSM-HI definitions had been singled out as having particularly good descriptive validity (Spitzer and Williams 1980). The third diagnosis was schizophrenia, which was thought to be a possible differential diagnosis from either of the first two. These three diagnoses were cho-

sen to represent the characteristics of DSM-HI with respect to some fundamental properties of classification systems. Blashfield and Draguns (1976) have suggested general considerations that are pertinent to the classification of mental disorders. The novel structure of DSM-IH creates a need for new classificatory concepts by which to evaluate its validity.

Table 1b. DSM-II definition of schizophrenia This large category includes a group of disorders manifested by characteristic disturbances of thinking, mood, and behavior. Disturbances in thinking are marked by alterations of concept formation which may lead to misinterpretation of reality and sometimes to delusions and hallucinations, which frequently appear psychologically selfprotective. Corollary mood changes include ambivalent, constricted, and inappropriate emotional responsiveness and loss of empathy with others. Behavior may be withdrawn, regressive, and bizarre. The schizophrenias, in which the mental status is attributable primarily to a thought disorder, are to be distinguished from the Major affective illnesses (q.v.) which are dominated by a mood disorder. The Paranoid states (q.v.) are distinguished from schizophrenia by the narrowness of their distortions of reality and by the absence of other psychotic symptoms. Source: Diagnostic and Statistical Manual of Mental Disorders, Second Edition, p. 33.

VOL. 7, NO. 2, 1981

Three such concepts were chosen for investigation here. These are seen as especially important given the approach that DSM-1I1 has taken to diagnosis and classification. The first concept examined was the diagnostic information contained

in a symptom definition. Symptoms from DSM-III criteria for schizophrenia should receive high ratings on diagnostic informativeness for schizophrenia. Conversely, symptoms from other diagnoses should be rated as containing little diagnostic informativeness toward a diagnosis of schizophrenia. In this manner, it is possible to examine whether DSM-III has grouped some nearly pathognomic signs with some relatively uninformative symptoms in constructing a definition of schizophrenia. If this is the case, then a particular symptom may act as a "weak link" in the diagnosis and, as such, allow an individual to be diagnosed as schizophrenic on the basis of a symptom that is of relatively little use in identifying schizophrenia. The second area of investigation was the descriptive validity of the

diagnosis. Robert L. Spitzer, Chair of the DSM-III Task Force, has referred to this concept as follows: Descriptive validity is the extent to which the characteristic features of a particular mental disorder are unique to that category, relative to other mental disorders and conditions. . . . The presence of descriptive validity justifies the assumption that a category represents a relatively distinct benavioral syndrome or pattern, rather than a random collection of clinical features. Historically, disorders such as mania and dementia were among the first mental disorders to be identified (face validity)

261

because they have considerable descriptive validity. [Spitzer and Williams 1980, p. 1037]

fects (Morey 1980). Hence, clinician variables were ignored in the subsequent analyses.

This means that a symptom should be seen as more informative for the diagnosis that it defines than for any other diagnosis. The third area of interest—the consistency with which clinicians interpret symptoms—was examined by measuring the extent to which clinicians agreed on the informativeness of a symptom. Some variability would be expected to arise from differences in training, but it was thought possible that certain symptoms might be interpreted quite differently because of ambiguity in the concept or the wording of a symptom. Ambiguity in interpretation can only result in an unreliable diagnosis; hence, it is important that clinicians be fairly consistent in deciding how informative a symptom is.

Stimuli. A card-sorting technique was used to obtain the clinicians' ratings. The stimuli were index cards with one symptom typed on the front. A symptom was defined as the smallest subunit of a DSMIII criterion that could be retained without losing its meaningfulness. Thus, a list of symptoms joined by an "or" was separated into individual symptoms. For instance, one criterion in the DSM-III definition of schizophrenia was defined as follows (criterion A-6):

Method

Subjects. The subjects for the experiments were 30 clinicians (15 clinical psychologists and 15 psychiatrists). They ranged in level of clinical experience from 1.5 to 44 years (mean = 13.6 years). Of the 30 clinicians, 6 were female (20 percent). All clinicians were from Gainesville, Florida, but their work settings were diverse. Sixteen held full-time academic appointments, while nine worked primarily in the community. The remaining five were trainees (psychiatry residents or psychology interns). Symptom ratings were analyzed first for the effect of clinician variables. Neither profession nor years of experience nor their interaction were found to have significant ef-

Somatic, grandiose, religious, nihilistic or other delusions without persecutory or jealous content. In this study, the above criterion was divided into six symptoms: Somatic delusions (SA6a) Delusions of grandiosity (SA6b) Religious delusions (SA6c) Nihilistic delusions (SA6d) Delusions without persecutory content (SA6ex) Delusions without jealous content (SA6ey) Hereafter, each symptom will be referred to by a code such as SA6c, which can be translated as follows: 5 —Schizophrenia A—Part A of the criterion 6 —Sixth criterion under A c —Third symptom found under A-6 Table 1 lists the DSM-III (15 January 1978 version) definition of schizophrenia. Inserted in the prose definition are the codes which denote the individual symptoms. When this study began, the 15 January 1978 version of the DSM-

SCHIZOPHRENIA BULLETIN

262

Figure 1. Implicit information structure of the DSM-IU schizophrenia criteria (1/15/78 version) 4.667 4.773 4.774 4.50

© Decision node ("or")

SA1 SA2 SA3 SA4 SA5

4.133

3.4266

)

3.433 3.433 3.70 3.167 3.40

Criterion A

SA6a SA6b SA6c SA6d

3.367

4.231 •

•

——r

'

3.867 4.550

SA7

4.033

> SA9

4.567 4.533

SA8a SA8b )

Criterion C (part 1)

•-

•

3.533

3.60

•SB

SCx

3.434

3.40 A

> SC2

3.20 3.60 3.00

2.933 Criterion C (part 2)

*

3.336

3.467

, ) (

3.433 2.867 2.867 2.433 2.833 3.167 3.667 2.60

,

3.833 2.933

Criterion D

•-

Criterion E

#-

1.884

2.566

-SE.ME

SC1a SC1b

SC3

4.067

3.767 1.769 2.00

3.533 4.167

SC4

)

3.420

3.433 4.00 3.60 3.267 3.533 3.733 3.867 3.50 3.70

3.667

Criterion B

3.433

/

sea •SDa •SDb

SC6a SC6b SC6c SC6d SC6e SC6f SC7a SC7b SC7c SC7d SC7e

SA6ex SA6ey

SA10a SA10b SA10c SA10d SA10e SA10f SA10g SA10h SA10i SA10J SA10k

263

VOL. 7, NO. 2, 1981

/// was the contemporary version. It is this version that was used in the extensive field trials of DSMIII. A later version (2 January 1979) added two new schizophrenic symptoms: NSA3—Delusions with persecutory or jealous content, accomplished by hallucinations of any type NSE —Onset of illness before the age of 45 Twenty-two of the 30 clinicians performed the task on both sets of symptoms. In total, there were 47 schizophrenic symptoms, as well as 20 dementia symptoms, 22 mania symptoms, and 15 personality disorder symptoms in the total cardsorting task. A complete list of all the symptoms is available from the first author. Procedure. A high rate of all clinicians contacted agreed to participate in the study (94 percent). Each clinician was given a shuffled stack of index cards and asked to sort the cards into five piles that ranged from the lowest information/importance for a particular diagnosis (value = 1) to the highest information (value = 5). The clinician was not told the diagnoses in advance. The diagnostic categories used were (in random order): mania, schizophrenia, and dementia resulting from an organic brain syndrome in which the specific etiology is unknown. Results Symptom Informativeness. The implicit information structure for the DSM-III definition of schizo-

phrenia is presented in figure 1, which uses a branching tree diagram to represent the diagnostic process specified in the DSM-III criteria. For instance, criterion A from the DSM-III definition of schizophrenia is subdivided into 10 parts, any one of which is sufficient to fulfill that criterion. In addition, each of these 10 may be subdivided further into specific symptoms (e.g., as was noted earlier, criterion SA6 can be broken down into six individual symptoms). It was decided to interpret any symptoms with an average informativeness rating of 3.0 or less as having insufficient information value. The value of 3.0 is pertinent for two reasons: (1) 3.0 was the midpoint of the rating scale; and (2) it was almost exactly one standard deviation below the grand mean of ratings of all schizophrenic symptoms (mean = .620, SD = .639). An analysis of figure 1 shows that all symptoms from criteria A and B are above 3.0. However, some of the symptoms from criterion C (prodromal/residual syndrome) and from criteria D and E (exclusion symptoms) are below this minimal informativeness level. For instance, symptom SDa "full depressive or manic syndrome is not present" only had a mean rating of 1.769. Of the total 47 symptoms, 9 (19 percent) had mean informativeness ratings below 3.0. Descriptive Validity. In order to assess descriptive validity, the mean ratings for each symptom in the three diagnostic categories of schizophrenia, mania, and dementia were compared using Duncan's tests. Of the 47 symptoms, 9 were found to have higher informative-

ness ratings for dementia than schizophrenia (e.g., SC4, "impairment of personal hygiene and grooming"). In addition, three were rated as more important for mania than for schizophrenia (e.g., SC6d, "overelaborate speech"). The Duncan's test, based on a one-way analysis of variance, revealed that 32 of the 47 symptoms were significantly more informative for schizophrenia than for either mania or dementia. Most of the 26 symptoms from criterion SA had significant descriptive validity except SA6b, SAlOa, SAlOd, and SAlOj. The majority of the problems in the area concerned prodromal/residual symptoms, i.e., the SC symptoms. Ten of these 18 symptoms failed to have significant descriptive validity (SCx, SC2, SC3, SC4, SC6b, SC6c, SC6d, SC6e, SC7c, and SC7e). Other symptoms that did not achieve significance were SB and SDa. A comparison of the mean ratings across diagnostic categories can be found in table 2. Consistency of Ratings. The consistency of the clinicians' ratings was assessed in two ways. First, the overall variance estimate for the ratings of each symptom was examined. Second, the consistency of the ratings for each symptom with respect to the pattern of ratings for the entire syndrome was measured. The variance estimates for the symptom ratings are presented in table 1. In general, the symptoms were rated fairly consistently, but some symptoms stand out as being rated quite variably. For instance, symptom SE ("no apparent organic mental disorder") had a variance of 2.19 on a 5-point scale, indicating a great deal of interrater

SCHIZOPHRENIA BULLETIN

264

Table 2. Comparison of mean rating scores for schizophrenic symptoms across diagnostic categories Symptom

Schizophrenia

Dementia

Mania

Duncans

Variance

SA1 SA2 SA3 SA4 SA5 SA6a SA6b SA6c SA6d SA6ex SA6ey

4.667 4.773 4.773 4.50 4.133 3.433 3.433 3.70 3.167 3.433 3.367

1.667 1.467 1.567 1.733 2.167 2.567 1.967 1.90 2.067 2.233 2.267

1.667 1.667 1.833 1.333 2.467 1.80 4.70 2.70 1.60 2.633 2.60

S>both S>both S>both S>D>M S>both S>D>M *M>S>D S>both S>both S>both S>both

.299 .271 .271 .397 .947 .943 1.151 1.045 1.454 1.151 1.205

SA7 SA8a SA8b SA9 SA10a SA10b SA10c SA10d SA10e

3.867 4.567 4.533 4.033 3.433 4.00 3.60 3.267 3.533

2.267 1.50 1.633 2.167 3.667 2.467 3.233 3.30 2.733

2.167 1.80 1.533 1.50 2.567 3.033 2.60 1.30 1.333

S>both S>both S>both S>D>M *both>M S>M>D *both>M *both>M S>D>M

1.361 .530 .740 1.275 1.495 .897 1.283 .823 1.637

SA10f SA10g SA10h SA10i SA1Oj SA10k

3.733 3.867 3.50 3.70 3.533 4.767

2.433 3.20 2.333 2.333 3.933 1.867

1.367 3.167 2.733 2.20 2.50 1.20

S>D>M S>both S>both S>both *both>M S>D>M

1.375 1.155 1.293 1.183 .740 1.040

SB SCx SC1a SC1b SC2 SC3 SC4

3.533 3.60 3.40 3.467 3.20 3.60 3.00

3.80 3.10 2.867 2.767 4.067 3.233 3.667

3.60 2.067 1.20 1.20 2.967 1.933 2.433

*NS *both>M S>D>M S>D>M *D>both *both>M *D>S>M

1.085 1.352 .938 .947 .786 .938 .966

SC6a SC6b SC6c SC6d SC6e SC6f

3.433 2.867 2.867 2.433 2.833 3.767

2.767 3.067 3.067 1.833 3.20 1.633

2.933 2.90 2.167 3.30 2.833 2.50

S>both 'NS *both>M *M>S>D *A/S S>M>D

1.288 1.154 1.223 1.151 1.178 1.523

SC7a SC7b SC7c SC7d

4.067 3.667 2.60 3.833

2.50 1.70 1.833 2.067

2.50 2.40 3.633 2.533

S>both S>M>D *M>S>D S>both

.892 1.195 .731 1.040

265

VOL. 7, NO. 2, 1981

Table 2 . --continued Symptom

Schizophrenia

Dementia

Mania

Duncans

Variance

SC7e

2.933

2.067

2.633

*both>D

.961

SC8 SDa SDb SE (ME)1

3.767 1.769 2.00 2.566

2.60 7.833 1.417

2.00

S>D>M 'NS S>D NS

1.151 1.359 .909 2.190

NSA3 NSE

4.227 3.455

2.045 1.636

S>both S>M>D

1.041 1.117

2.50 1.909 2.364

Notes: All Duncans tests based on a = .05. Largest mean score for each symptom is represented in italics. Starred (*) Duncans test indicates results contrary to DSM-III assumptions. 'Also a Manic sign.

variability. Other symptoms with high rating variability include SAlOe, SC6f, and SAlOa. The second approach to assessing rater variability is analogous to the notion of internal consistency estimates of item reliability in psychological test theory (Lord and Novick 1968). Here, the correlation of the rating of each symptom with the sum of all schizophrenic symptom ratings was examined. This estimate of consistency determines if items are measuring the same psychological trait or syndrome. Most of the correlations between an individual symptom and the composite schizophrenic rating ranged between .50 and .70. This range of internal consistency reliability estimates is not impressive. Only 18 of the 49 schizophrenic symptoms had correlations greater than .70. On the other hand, eight symptoms (SA1, SA2, SA4, SAlOd, SC3, SC6d, SC7c, and SC7d) had internal consistency correlations below .50. Because these internal consistency correlations were low, it was inferred that the symptoms did not "band together" and that the clinicians did not perceive the 49 symptoms as being reasonably homogeneous in terms of the construct

they represented. Thus, a decision was made to perform a factor analysis on the symptoms with complete data. The factoring method was a principal components technique followed by a varimax rotation with two factors rotated. (See table 3.) The first factor could be interpreted as the "primary schizophrenic symptom factor," which included Schneider's first-rank symptoms (SA1, SA2, SA3, and SA5) and most delusions and hallucinations (SA6, SA7, and SA8).

Generally the symptoms loading on the first factor were symptoms from criterion A ("characteristic symptoms"). The symptoms in this factor tended to have high informativeness ratings (mean = 4.02). The next factor could be called the "secondary schizophrenia symptom factor." Many of the symptoms loading on this factor were from the list of prodromal/ residual symptoms. In addition, most of the SA10 symptoms were highly correlated with this factor.

Table 3. Rotated factor matrix for schizophrenia symptoms Symptom

Factor 1 loading

Factor 2 loading

SA1 SA2 SA3 SA4 SA5 SA6a SA6b SA6c SA6d SA6ex SA6ey

-.229 -.394 .107 .171 .248 .245 .381 540* .260 .473 .508

.789* .560* .731* .296 .742* .481 .763* .518 .604* .599* .664*

SA7 SA8a SA8b

.205 -.090 .113

.813* .839* .750*

266

The symptoms associated with this factor had relatively low informativeness ratings (mean = 3.29).

Discussion As can be noted from table 1, DSM-I1I recognizes five subdivisions of schizophrenic symptoms:

SCHIZOPHRENIA BULLETIN

Table 3.—continued Symptom

Factor 1 loading

SA9 SA10a SA10b SA10c SA10d SA10e

.073 .668*

Factor 2 loading

.413 .532* .527*

.750* .337 .493 .480 .039

.811*

.127

SA10f SA10g SA10h SA10i SA1OJ SA10k

.743* .740* .346 .389 .724* .342

.285 .291 .788* .695*

Of these groupings, A and C contain the majority of the symptoms.

SB SCx

.590* .299

SC1a

Characteristic Symptoms. This subdivision contains 26 of the 47 symptoms in the 15 January 1978 version of DSM-III. Any 1 of the first 15 symptoms or any 2 of the last 11 are sufficient for the identification of the schizophrenic syndrome. In general, the symptoms from this set had adequate ratings on informativeness and discriminant validity. The most problematic symptoms were from SA10 ("other characteristic symptoms"). These symptoms belonged to the "secondary schizophrenic symptom factor" which had relatively low informativeness ratings. Four of the 11 SA10 symptoms failed to achieve significant discriminant validity. These symptoms appear to be least trustworthy as a basis for the diagnosis of schizophrenia. There was also some problem associated with SA6 symptoms (various forms of delusions). Although these symptoms had relatively low informativeness ratings (mean = 3.4266), they were loaded on the primary schizophrenic symptoms factor and had adequate

SC1b

.580* .817* .633*

(A) Characteristic symptoms (B) Impairment of daily routine (C) Prodromal/residual symptoms (D & E) Exclusionary symptoms

SC2 SC3

SC4

.591* .642*

.051 .600* .146 .062 .179 .021 .108 .488 .150

SC6a SC6b SC6c SC6d SC6e SC6f

.752* .800* .701* .426

.601* .780*

.476 .336 .335 .119 .359 .264

SC7a SC7b SC7c SC7d SC7e

.553 .438 .577* .170 .709*

.652* .607* .059 .476 .372

SC8 NSA3 NSE

.186 .321 .637*

.543* .892* .123

Notes: Starred (*) figures indicate primary factor loading above .500. Eigenvalues equal to 19.60, 5.37; accounting for 53 percent of total variance.

discriminant validity (except SA6b). Prodromal/Residual Symptoms. The characteristic symptoms are used to identify schizophrenia during the active phase of the disorder. The class of 18 symptoms

listed as prodromal/residual should be manifest during nonactive phases of the disorder. The factor analysis results suggested that these symptoms were not perceived by the clinicians as having the same meaning as the primary schizophrenic symptoms. In addi-

267

VOL. 7, NO. 2, 1981

tion, they lacked significant discriminant validity. Based on these results, it appears that the prodromal/residual syndrome should be deleted from the diagnostic criteria for schizophrenia in DSM-IH and that the diagnosis of this disorder can only be safely made during the active phase of illness. Exclusionary and Other Symptoms. Included in this group are all symptoms from SB, SD, and SE. All but one (SDb) of these four symptoms failed to achieve discriminant validity. The exclusionary symptoms were not rated by the clinicians as being important in diagnosing schizophrenia. Given the fact that DSM-III permits the use of multiple diagnoses of clinical syndromes, it is not clear why exclusionary symptoms are mentioned at all. Implications for the Classification of Psychopathology. The DSM-IH represents an advance in psychiatric classification by its attempts to objectify the practice of psychiatric diagnosis. Hempel (1966, p. 144), in viewing the components of a classification as extensions of scientific concepts, claimed that these components must "possess clear criteria of application that can be stated in terms of publically ascertainable characteristics." Relative to DSM-II, DSM-lll has made significant strides in the direction of clarity. With this clarity, however, come some potential dangers. A major danger involves the illusion of precision. The DSM-III presents symptoms in a quasi-objective format in which interpretation would be expected to be uniform, but the results reported here

suggest that this expectation is not met. For example, although most symptoms were rated as being associated with schizophrenia, some were rated as virtually pathognomic while others were only suggestive of the diagnosis. This results in a disturbing situation: DSM-III mandates a "minimum" amount of information necessary to make a diagnosis, but there is considerable variability in this minimum. That is, the options that the DSM presently allows the diagnostician permit him to diagnose an individual as schizophrenic on the basis of relatively uninformative symptoms. For example, both SA2 and SA6d may be used to fulfill criterion A, despite a difference of over 2.5 standard deviations in their rated informativeness for schizophrenia (SA2 = 4.773; SA6d = 3.167). Even though DSM-III allows the option of selecting either of these symptoms to fulfill the criterion, it is obvious that they are not equally informative. There is a need to provide a definition of schizophrenia in which a consistently high amount of information is required to make the diagnosis. Certainly the prodromal/residual symptoms are characteristic of schizophrenia, but the symptoms that are to be regarded as definitive should be selected with care (Frake 1968). Definitive characteristics are those that are necessary to distinguish between related disorders. Characteristics that are common to the category but are not useful for discrimination are less informative and, hence, less useful in definition. For example, a characteristic of schizophrenics is that members of this category are human beings. But such a characteristic is also

true for manics, demented patients, etc. Being human is not a definitive characteristic. Thus, although many of the schizophrenic symptoms in DSM-III are indeed descriptive of schizophrenia, they have relatively little utility in discriminating among other psychotic disorders. The definition of schizophrenia could be made much less complex, yet more rigorous, if only those symptoms labeled primary symptoms in this study were kept as criteria required for diagnosis. Someday scientists may isolate factors that will definitely indicate the presence or absence of any of the mental disorders. In the meantime, however, society must rely on the clinician, who has had the most contact with these disorders, to make diagnoses. The present study attempted to study the clinician in evaluating DSM-III as a classification system, with a focus on informativeness, descriptive validity, and internal consistency. Although limited in the number of diagnoses and the range of clinicians studied, this study presented a new framework for examining a classification of psychopathology, as well as a paradigm for validating any such system. References

American Psychiatric Association. DSM-H: Diagnostic and Statistical Manual of Mental Disorders. 2nd ed.

Washington, DC: APA, 1968. American Psychiatric Association. DSM-IH: Diagnostic and Statistical Manual of Mental Disorders. 3rd ed.

Washington, DC: APA, 1980. Beck, A.T.; Ward, C.H.; Mendelson, M.; Mock, J.E.; and Erbaugh, J.K. Reliability of psychiatric diagnosis: 2. A study of consistency of

268

clinical judgments and ratings. American Journal of Psychiatry, 119:351-357, 1962. Blashfield, R.K., and Draguns, J.G. Evaluative criteria for psychiatric classification. Journal of Abnormal Psychology, 85:140-150, 1976. Bleuler, E. Dementia Praecox, or The Group of Schizophrenias. (1911) Translated by J. Zinkin. New York: International Universities Press, 1950. Feighner, J.P.; Robins, E.; Guze, S.B.; Woodruff, R.A.; Winokur, G.; and Munoz, R. Diagnostic criteria for use in psychiatric research. Archives of General Psychiatry, 26:57-63, 1972. Frake, C O . The ethnographic study of cognitive systems. In: Fishman, J.A., ed. Readings in the Sociology of Language. The Hague: Mouton, 1968. pp. 434-466. Hempel, C.G. Aspects of Scientific Explanation. New York: Free Press, 1966.

SCHIZOPHRENIA BULLETIN

Kraepelin, E. Dementia Praecox. Translated by R.M. Barclay. Edinburgh: E. & S. Livingstone, 1919. Lord, F.M., and Novick, M.R. Statistical Theories of Mental Test Scores. Reading, MA: AddisonWesley Publishing Company, 1968.

and DSM-III. In: Kaplan, H.I.; Freedman, A.M.; and Sadock, B.J., eds. Comprehensive Textbook of Psychiatry. Vol. III. Baltimore: The Williams & Wilkins Company, 1980. pp. 1035-1071. Acknowledgment

Meehl, P.E. Schizotaxia, schizotypy, schizophrenia. American Psychologist, 17:827-838, 1962.

This research represents a section of the first author's thesis leading to the Master of Science degree. The authors wish to thank Drs. Morey, L.C. Differences between psychologists and psychiatrists in John Kuldau and Paul Satz for use of DSM-11I. American Journal of their ideas and criticism. Psychiatry, 137:1123-1124, 1980. Schneider, K. Clinical Psychopathology. Translated by M.W. Hamilton. New York: Grune & Stratton, 1959. Spitzer, R.L., and Fleiss, J.L. A reanalysis of the reliability of psychiatric diagnosis. British Journal of Psychiatry, 125:341-347, 1974. Spitzer, R.L., and Williams, J.B.W. Classification of mental disorders

The Authors Leslie C. Morey, M.S., is Resident in Clinical Psychology, Department of Psychiatry, University of Texas Health Sciences Center, San Antonio, Tx. Roger K. Blashfield, Ph.D., is Associate Professor, Department of Psychiatry, University of Florida, Gainesville, FL.