Examining the latent structure of negative symptoms: Is there a distinct subtype of negative symptom schizophrenia?

Schizophrenia Research 77 (2005) 151 – 165 www.elsevier.com/locate/schres

Examining the latent structure of negative symptoms: Is there a distinct subtype of negative symptom schizophrenia? Jack J. Blancharda,*, William P. Horanb, Lindsay M. Collinsa a

Department of Psychology, University of Maryland, College Park, MD 20742-4411, United States b University of California, Los Angeles, United States

Received 17 November 2004; received in revised form 20 March 2005; accepted 24 March 2005 Available online 23 May 2005

Abstract Negative symptoms have emerged as a replicable factor of symptomatology within schizophrenia. Although rating scales provide assessments along dimensions of severity, categorization into a negative symptom subtype is typically conducted. A categorical view of negative symptoms is best reflected in the proposal that enduring, primary negative symptoms, or deficit symptoms, reflect a distinct subtype of schizophrenia [Carpenter, W.T., Heinrichs, D.W., Wagman, A.M.I., 1988. Deficit and nondeficit forms of schizophrenia: The concept. Am. J. Psychiatry 145, 578–583.]. Despite an accumulation of findings that support a categorical conceptualization, the data are also consistent with a dimensional-only model where negative symptom subtypologies simply reflect an extreme on a continuum of severity. Using taxometric statistical methods [Waller, N.G., Meehl, P.E., 1998. Multivariate Taxometric Procedures: Distinguishing Types From Continua. Sage, Newbury Park, CA.], the present study examined whether a taxonic, or latent class, model best describes negative symptoms in a sample of 238 schizophrenia patients. In order to obtain more stable estimates of symptoms, ratings on the Scale for the Assessment of Negative Symptoms [Andreasen, N.C., 1982. Negative symptoms in schizophrenia: Definition and reliability. Arch. Gen. Psychiatry 39, 784–788.] were averaged across two assessments over a 6-month period. Two taxometric methods, maximum covariance analysis (MAXCOV) and mean above minus below a cut (MAMBAC) identified a latent class or taxon with a base rate of approximately 28–36%. Members of the negative symptom taxon differed from the nontaxon class in that taxon members were more likely to be male and demonstrated poorer social functioning. Taxon and nontaxon schizophrenia patients did not differ in psychotic or affective symptoms. The findings converge to provide support for a categorical view of negative symptoms. Further research is required to replicate the present taxonic findings and to examine characteristics (including possible etiological factors) associated with this negative symptom taxon. D 2005 Elsevier B.V. All rights reserved. Keywords: Schizophrenia; Negative symptoms; Deficit syndrome; Taxometrics

* Corresponding author. Tel.: +1 301 405 8438; fax: +1 301 314 9566. E-mail address: [email protected] (J.J. Blanchard). 0920-9964/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.schres.2005.03.022

As schizophrenia researchers have grappled with the phenotypic heterogeneity of this disorder, great emphasis has been placed on determining whether

152

J.J. Blanchard et al. / Schizophrenia Research 77 (2005) 151–165

more homogeneous symptom clusters underlie the diverse clinical features of this disorder. These efforts have consistently demonstrated that the negative symptoms of schizophrenia cohere into a distinctive clinical domain that has important associations with other features of schizophrenia. Although there are some differences in the specific content of clinical assessment instruments, the negative symptoms of schizophrenia typically include anhedonia, flat or blunted affect, poverty of speech (alogia), avolition, and asociality (e.g., Andreasen, 1985; Kay et al., 1987). Negative symptoms are relatively common (Fenton and McGlashan, 1992) and are factorially independent from other symptom domains, including positive, disorganized, and affective symptoms (e.g., Andreasen et al., 1995; Bilder et al., 1985; Emsley et al., 2003; Fitzgerald et al., 2003; Lenzenweger et al., 1989; Smith et al., 1998). In addition, negative symptoms demonstrate unique associations with social functioning, neurocognition, and neurobiology (for a review see Earnst and Kring, 1997). While strong support for the construct validity of negative symptoms exists, it is not yet known whether these symptoms reflect purely dimensional individual differences across patients or instead are indicators of a discrete category or schizophrenia subtype. Phrased differently, are negative symptoms indicators of a latent class, or taxon, within schizophrenia? This distinction has practical implications for clinical assessment and is fundamental to how researchers proceed in developing and testing etiological models of negative symptoms (e.g., Meehl, 1992). For example, if the latent distribution of negative symptoms is truly continuous, dichotomous assessments of negative symptoms will needlessly increase measurement error and decrease statistical power, thereby diminishing researchers’ ability to identify symptom correlates or predict course and outcome. Knowledge about the latent structure of negative symptoms would also have different etiological implications and thereby provide guidance for optimal research designs. A latent discontinuity would imply that efforts to identify some discrete entity, structure, or event that distinguishes patient subgroups will be most sensible, whereas a continuous latent distribution would support efforts to identify etiological factors associated with varying levels of negative symptoms across all schizophrenia patients.

Although a variety of methods have been used to make inferences about the underlying structure of negative symptoms, existing research is limited in its ability to address this issue. For example, the above referenced factor analytic studies are often interpreted as evidence for a dimensional conceptualization of negative symptoms. However, these findings cannot be taken as evidence for a purely dimensional structure as indicators of a latent taxon must necessarily yield a large factor (Meehl, 1999). Alternatively, while clinical rating scales of negative symptoms (e.g., Andreasen, 1982; Kay et al., 1989) permit typological categorization, it may be that such cut-offs are merely arbitrary distinctions imposed on a continuum of symptom severity (Pogue-Geile and Keshavan, 1991). Indeed, if negative symptoms are correlated with other domains of interest such as clinical, neurocognitive, or psychophysiological functioning, the subsequent comparison of individuals high versus those low in negative symptoms will yield significant group differences regardless of whether or not the true latent structure of negative symptoms is purely dimensional (see Meehl, 1992, p. 124). In a refinement of the conceptualization of negative symptoms that explicitly adopts a categorical model, Carpenter et al. (Carpenter et al., 1988; Kirkpatrick et al., 2001) have proposed that enduring, primary negative symptoms, or deficit symptoms, reflect a distinct subtype of schizophrenia termed the bDeficit SyndromeQ. This putative subtype is thought to result from a unique pathophysiology that is not shared by other, nondeficit forms of schizophrenia (Buchanan and Carpenter, 1994). Several lines of evidence support the validity of the deficit syndrome. While deficit patients typically do not differ from nondeficit patients on levels of positive psychotic symptoms, they have been found to demonstrate greater anhedonia (Kirkpatrick and Buchanan, 1990) and less positive affectivity (Horan and Blanchard, 2003), poorer current and premorbid adjustment (e.g., Buchanan et al., 1990; Fenton and McGlashan, 1994; Horan and Blanchard, 2003), and greater neuropsychological impairment (Bryson et al., 2001; Buchanan et al., 1990; Horan and Blanchard, 2003; Ross et al., 1996). Neuroimaging findings indicate functional and structural differences between deficit and nondeficit schizophrenia patients (Heckers et al.,

J.J. Blanchard et al. / Schizophrenia Research 77 (2005) 151–165

1999). Sex differences have also frequently been reported, with males having a higher prevalence rate for the deficit syndrome than females (Kirkpatrick et al., 2000; Roy et al., 2001). Finally, there is some evidence that the deficit syndrome is heritable (Fouldrin et al., 2001; Kirkpatrick et al., 2000; Ross et al., 2000). Findings concerning the validity of the deficit syndrome may be viewed as consistent with conjectures that enduring, primary negative symptoms reflect a unique disease process; however, as with the broader negative symptom literature, the results may simply reflect dimensional differences in severity rather than a true categorical distinction. Quantitative procedures may serve to clarify the true latent structure of negative symptoms. As noted by Kirpatrick et al. (2001), taxometric statistical techniques (Meehl, 1995, 1999; Waller and Meehl, 1998) should be useful for testing the categorical model. Taxometric statistical methods allow one to quantitatively address the question of whether a latent variable is best understood as dimensional-only or if taxonicity is also a feature (Waller and Meehl, 1998). These methods search for orderly statistical relationships between indicators that are indicative of a qualitative boundary between two groups. The detection of such a boundary provides the basis for inferring that a latent class variable must exist to explain this pattern of observable relationships. The superiority of taxometric methods over other approaches that have been used to study the structure of negative symptoms or alternative classification procedures has been extensively documented (Waller and Meehl, 1998). For example, latent class analysis can yield a categorical fit when purely dimensional models also fit (e.g., Reise and Gomel, 1995), and cluster analysis often does not effectively recover true taxa from real data and may detect spurious clusters (Golden and Meehl, 1980). In contrast, taxometric techniques have been shown to effectively detect real latent class structures known to exist and to not spuriously detect a taxon when none exists (Meehl, 1995; Meehl and Yonce, 1996). Direct comparisons between taxometric methods and alternative procedures (e.g., admixture analysis and cluster analysis) have shown that taxometric methods are more sensitive to taxonic results (Cleland et al., 2000). In particular, recent findings indicate that the taxometric

153

procedure known as maximum covariance analysis (MAXCOV; Meehl, 1973; Meehl and Golden, 1982) is sensitive for detecting taxa using modest sample sizes and performs better than cluster analysis in classifying cases under conditions of reduced effect size, a limited number of indicators, smaller base rates, and increased nuisance covariance (Beauchaine and Beauchaine, 2002)—conditions often present in studying clinical disorders. Taxometric techniques have now been applied to a wide range of issues pertaining to psychiatric nosology (see Haslam and Kim, 2002, for a comprehensive review), including the latent structure of schizotypal traits (e.g., Blanchard et al., 2000; Horan et al., 2004; Lenzenweger, 1999; Tyrka et al., 1995a,b). To determine whether the latent structure of negative symptoms is taxonic, taxometric methods were applied to global scores from the Scale for the Assessment of Negative Symptoms (Andreasen, 1982), which were rated in a relatively large, well-characterized sample of schizophrenia patients who participated in a multi-site NIMH collaborative study, Treatment Strategies for Schizophrenia (TSS; Mueser et al., 2001; Schooler et al., 1997). An advantage of this sample was the availability of negative symptom ratings over two assessment periods to allow for more reliable symptom measurement. In addition to providing evidence about the existence of a taxon, taxometric methods provide base rate estimates of the conjectured latent class. Prior estimates of a putative negative symptom syndrome have ranged widely, from 6% to 40% (e.g., Fenton and McGlashan, 1992), with more restrictive definitions based on deficit syndrome criteria estimating prevalence rates of 25–30% in chronic schizophrenia samples (Kirkpatrick et al., 2001). The latter estimate would be applicable to the chronic outpatient sample studied herein, and we predicted a negative symptom taxon with a base rate approximating 30%. In addition, the TSS data set enabled us to go beyond a test of the structural component of the latent class model of negative symptoms to also evaluate the external component of the latent class model (i.e., external validity; Gangestad and Snyder, 1985; Loevinger, 1957). This was addressed with demographic, symptom, and social functioning data. Should a negative symptom taxon exist, we sought to determine whether taxon members demonstrate

154

J.J. Blanchard et al. / Schizophrenia Research 77 (2005) 151–165

characteristics found in prior studies employing clinically derived categorical ratings (such as the deficit syndrome, reviewed above), including a preponderance of males and poorer social functioning than patients not assigned to a negative symptom taxon. Alternatively, we could examine if taxon members were similar to other individuals with schizophrenia in levels of conceptually unrelated symptoms such as psychosis or depression.

1. Methods 1.1. Participants Participants were patients with schizophrenia, schizoaffective disorder, or schizophreniform disorder (Diagnostic and Statistical Manual of Mental Disorders, 3rd ed., rev.; DSM-III-R; American Psychiatric Association, 1987) who were involved in TSS (Mueser et al., 2001; Schooler et al., 1997). Treatment within TSS involved supportive family management or applied family management and three different neuroleptic dosages (standard, low, and targeted). Inclusion criteria for participants in the TSS study included (1) DSM-III-R diagnosis of schizophrenia, schizoaffective disorder, or schizophreniform disorder determined by the Structured Clinical Interview for DSM-III-R (Spitzer et al., 1990), (2) aged between 18 and 55 years, (3) willing to take fluphenazine decanoate injections and not receive other neuroleptic, antidepressant, or mood stabilizing medications, (4) in contact with family of origin or legal guardian at least 4 h/week, (5) patient and relative willing and able to give written informed consent to participate in the study, and (6) psychiatric hospitalization or symptom relapse in the past 3 months. Exclusion criteria included (1) current physical dependence on alcohol, stimulants, barbiturates, or narcotics, (2) current hospitalization precipitated by substance abuse, (3) current pregnancy, (4) epilepsy or organic brain syndrome, and (5) unequivocal liver damage. The TSS study design involved (a) an index hospitalization, (b) a 6-month stabilization period, followed by (c) a 24-month double-blind treatment phase with assessments every 6 months. The current study used symptom assessments obtained at

the start of the double-blind treatment phase as this assessment period occurred following stabilization (approximately 6 months following index hospitalization) and was expected to be associated with symptom assessments that reflected typical functioning. The participants included in the current study were 238 patients who had completed negative symptom evaluations at the start of the doubleblind treatment phase and at the first 6-month follow-up. The mean age of participants was 29.77 (SD = 7.28), and the sample consisted of relatively more males, N = 153 (64.3%), than females, N = 85 (35.7%). The sample was racially diverse and comprised of African-American, N = 127 (53.4%), White, N = 91 (38.2%), Hispanic, N = 7 (2.9%), Asian, N = 2 (0.8%), and unspecified race, N = 11 (4.6%), participants. 1.2. Instruments 1.2.1. Negative symptoms The Scale for the Assessment of Negative Symptoms (SANS; Andreasen, 1982) was used to evaluate negative symptoms. The SANS was modified for use in the TSS study (see Sayers et al., 1996, for a full discussion) including making ratings based solely on interview and observations during the interview, using standardized probe questions, making ratings on 5point Likert-scales (1 to 5; not at all, mild, moderate, marked, or severe), eliminating the binappropriate affectQ item from the Blunted Affect subscale, and evaluating symptoms over a 1 week period rather than over a 1 month period as in some previous studies. Ratings were conducted by psychiatrists. Prior results indicate that the SANS used in TSS demonstrated acceptable interrater agreement and internal consistency (Mueser et al., 1994). Four SANS subscale scores were used in the current study: Anhedonia–Asociality, Alogia, Blunted Affect, and Avolition–Apathy. The SANS Attention scale was not included in the current analyses given findings suggesting that this scale is not conceptually related to the negative symptom complex (e.g., Andreasen et al., 1995; Earnst and Kring, 1997). Each scale is based on the sum of the constituent items. In order to obtain more stable estimates of negative symptoms, SANS subscale ratings were averaged across assessments from the start of the

J.J. Blanchard et al. / Schizophrenia Research 77 (2005) 151–165

double-blind treatment phase and at 6-month followup. Descriptive statistics for the SANS subscales across assessments are provided in Table 1. Paired ttests indicated that ratings over the two assessments were unchanged for Alogia, Avolition–Apathy and Anhedonia–Asociality (all pVs N 0.05). Blunted Affect ratings did show a decline over time, t(236) = 3.07, p b 0.005; however, this was a modest change with the mean difference between assessments of 0.97. As shown in Table 1, each SANS subscale demonstrated significant test–retest correlations suggesting some stability in ratings over time. Intercorrelations between the averaged SANS subscales are presented in Table 2 and indicate shared variance (18–46%) across the subscales; this moderate level of correlation among the subscales is compatible with the existence of a latent taxon (Meehl, 1999). 1.2.2. Symptomatology Symptomatology during the 1 week prior to the baseline and follow-up SANS assessments was evaluated by psychiatrists with an anchored version of the Brief Psychiatric Rating Scale (BPRS; Overall and Gorham, 1962; Woerner et al., 1988). BPRS ratings were completed by the same assessor conducting the SANS ratings. The BPRS consists of 18 items which are rated on a 7-point scale (1 to 7; not reported or not observed, very mild, mild, moderate, moderately severe, severe, and very severe). Based on findings concerning the factor structure of the BPRS (Mueser et al., 1997), we utilized four scales including Thought Disturbance (grandiosity, suspiciousness, hallucinatory behavior, and unusual thought content), Anergia (emotional withdrawal, motor retardation, uncooperativeness, and blunted

Table 1 Descriptive statistics for the Scale for the Assessment of Negative Symptoms (SANS) subscales at initial assessment and six-month follow-up (N = 238) SANS subscale

Initial assessment M (SD)

Six-month follow-up M (SD)

Test–retest r

Anhedonia–Asociality Blunted affect Avolition–Apathy Alogia

9.33 11.63 6.24 5.59

8.94 10.66 6.12 5.57

0.49* 0.52* 0.40* 0.50*

* p b 0.001.

(4.16) (4.96) (2.49) (1.81)

(4.02) (4.74) (2.57) (1.98)

155

Table 2 Intercorrelations of averaged Scale for the Assessment of Negative Symptoms (SANS) Subscales (N = 238) SANS subscale

Anhedonia– Blunted Avolition– Alogia Asociality affect Apathy

Anhedonia–Asociality Blunted affect Avolition–Apathy Alogia

– 0.55* 0.68* 0.45*

– 0.42* 0.66*

– 0.44*

–

* p b 0.001.

affect), Affect (somatic concern, anxiety, guilt, depressive mood, and hostility), and Disorganization (conceptual disorganization, tension, and mannerisms/ posturing). Each subscale score was based on the average of the item scores comprising the subscale. Interrater reliability (using independent raters) for these subscales has been shown to be acceptable (Blanchard et al., 2004). 1.2.3. Social functioning Subjects’ functioning in the community was evaluated with an abbreviated Social Adjustment Scale II, patient version (SAS; Schooler et al., 1979). Patient SAS global ratings (1 = excellent adjustment to 7 = severe maladjustment) were completed for Instrumental Role Performance (consistency and effectiveness of work performance), Household Integration (mutual support, affection, and participation with principal family members and others in household), Relationship with Family (support, affection, and participation with extended family), Social/Leisure (level and quality of activities and meaningfulness of relationships), and General Adjustment (performance in all roles and interpersonal relationships). Assessments of social adjustment were obtained approximately 1 month after hospitalized patients re-entered the community during the stabilization phase, which was approximately 6 months prior to the negative symptom ratings obtained at the start of the double-blind treatment phase. Research assistants administered the SAS. Although interrater agreement was not formally assessed, monthly conference calls were conducted to review issues related to SAS ratings with research assistants from all participating sites. One or more social adjustment items were missing for fourteen subjects, leaving an N of 224 for analyses of this measure.

156

J.J. Blanchard et al. / Schizophrenia Research 77 (2005) 151–165

2. Results 2.1. Structural component of the latent class model In the current study we utilized two taxometric procedures: maximum covariance analysis (MAXCOV; Meehl, 1973; Meehl and Golden, 1982) and mean above minus below a cut (MAMBAC; Meehl, 1995; Meehl and Yonce, 1994). The utilization of MAXCOV and MAMBAC allows for a determination of the consistency of findings across two mathematically independent procedures. The replicability of results across different techniques is a crucial aspect of taxometric procedures. Within and across the MAXCOV and MAMBAC procedures, if true taxa underlie the indicators of a putative taxon, (a) graphs plotting the statistical relationships among the indicators of the putative taxon should demonstrate consistent and predictable function forms, and (b) all base rate estimates should be consistent across procedures. Consistent base rate estimates strongly corroborate a taxonic model (e.g., Meehl, 1995). Both procedures also provide estimates of the latent means of the taxon and nontaxon classes, as well as standardized validity estimates that correspond to Cohen’s d (1988) (i.e., the effect size of the mean difference between latent classes). All taxometric analyses were conducted in R with program code available from Ruscio (2004). 2.2. MAXCOV MAXCOV (Meehl, 1973; Meehl and Golden, 1982) requires at least 3 indicators. One indicator is selected as an binput variableQ. Equal-interval cuts (i.e., slabs) are made along the input variable to create subsamples. Within the subsamples, covariances between a pair of other indicators (boutput variablesQ) are computed. When indicators covary due to their ability to discriminate two taxa, the covariance between the output variables will vary as a function of intervals on the input variable. Specifically, the maximum covariance should exist in the interval most closely approximating an equal mix of the two latent taxa (the HITMAX interval); small covariances should exist in intervals of near-pure samples of members of the taxon or nontaxon class (see Meehl, 1973, 1992; Meehl and Golden, 1982; Meehl and Yonce, 1996). For taxonic distributions, covariance

curves take on a distinctive peak at the point of maximum covariance. For purely dimensional data, covariance peaks are not identifiable and covariance plots are either flat or inconsistently peaked. Because all possible indicators can be used as an input variable, MAXCOV can be applied to multiple combinations of input and output variables, which should corroborate one another. For our analyses, each of the four SANS indicators served as input variables, with three possible pairs of output variables associated with each input variable. In total, 12 covariance curves were computed. Each input variable was standardized and slab width was 0.5 SD. To maintain stability of the estimations of covariance, a minimum subsample size of 25 was set for the extreme interval. Fig. 1 shows the averaged covariance curves for each of the four SANS indicators used as an input variable (each graph represents the average of the three covariance curves computed for the input variable). As can be seen, in all plots the covariance curve has an identifiable peak, a result consistent with a taxon. Table 3 presents the averaged base rate estimates yielded by the covariance curves for each indicator. Base rates ranged from 0.24 to 0.31 and, hence, are all quite similar. Their mean value, 0.28, represents a best approximation of the base rate of the taxon associated with negative symptoms. Standardized validity estimates were calculated from the MAXCOV results using pooled variances that correspond to a Cohen’s (1988) d effect size. Validity estimates for the SANS subscales are presented in Table 3. Validities ranged from 1.71 to 2.21 standard deviations between the latent means with an average of 1.93. These results suggest robust effect sizes, strongly supporting the validity of the taxon indicators. 2.3. MAMBAC As noted above, we utilized the taxometric procedure of MAMBAC (Meehl, 1995; Meehl and Yonce, 1994) as a consistency test for MAXCOV. In MAMBAC indicator pairs are used with each serving as an input indicator. Cases are sorted based on scores from the input variable and a sliding cut is then moved across this input variable. Differences on the second (output) indicator between the mean of cases above the cut and the mean of cases below the cut are computed, plotted graphically, and examined to deter-

J.J. Blanchard et al. / Schizophrenia Research 77 (2005) 151–165

157

0.6

0.6

0.5

0.5

Covariance

Covariance

0.4 0.4 0.3 0.2

0.3 0.2 0.1

0.1

0 -0.1

0 -1.00

-0.50

0.00

0.50

1.00

1.50

-1.00

2.00

-0.50

Blunted Affect

0.00

0.50

1.00

1.50

2.00

Anhedonia 0.6

0.6 0.5

Covariance

Covariance

0.5 0.4 0.3

0.4

0.3

0.2

0.2

0.1

0.1

0

0 -1.00

-0.50

0.00

0.50

1.00

1.50

2.00

Avolition

-0.50

0.00

0.50

1.00

1.50

2.00

Alogia

Fig. 1. Covariance curves for Scale for the Assessment of Negative Symptoms (SANS) subscales (N = 238). Each point represents averaged covariance from three pairings of SANS subscales not used as input.

mine whether predictable taxonic functions are evident in the plots. Two MAMBAC curves were computed for each pair of indicators with each serving as an Table 3 MAXCOV and MAMBAC estimates of base rates and indicator validities for the Scale for the Assessment of Negative Symptoms (SANS) Subscales (N = 238) SANS subscale

MAXCOV

MAMBAC

Estimated Indicator Estimated Indicator base rate validity base rate validity Anhedonia–Asociality Blunted affect Avolition–Apathy Alogia Mean (SD)

0.24 0.29 0.31 0.29 0.28 (0.03)

2.21 1.82 1.98 1.71 1.93 (0.22)

0.42 0.32 0.37 0.33 0.36 (0.05)

2.10 1.85 1.84 1.74 1.88 (0.15)

MAXCOV=maximum covariance; MAMBAC=mean above minus below a cut. Base rate estimates for each SANS subscale reflect the average of three MAXCOV covariance curves or, in the case of MAMBAC, two curves. Indicator validities are expressed as standard deviation units.

input indicator, resulting in a total of 12 difference curves. Taxonic results show a clear convex upward appearance or a steep rightward sweeping peak in lower base rate conditions, whereas non-taxonic results demonstrate a concave or dish shape curve (Meehl, 1995). As can be seen in Fig. 2, MAMBAC results demonstrated rightward sweeping peaks consistent with a taxonic structure. Base rate estimates derived from MAMBAC are presented in Table 3 with an average base rate of 0.36. Validity estimates derived from MAMBAC are also presented in Table 3 with a mean estimated indicator validity of 1.88 standard deviations. In general, these MAMBAC results corroborate the MAXCOV findings. 2.4. Classifying cases into the SANS taxon Based on each individual’s pattern of scoring at or above versus below the HITMAX score on the four indicators, we computed Bayesian probabilities of

150

200

100

150

200

50

100

150

200

100

150

200

1.4 1.2 1.0

150

200

1.0 0.8

50

100

150

200

Input Indicator 3 (Cases)

1.1

Input Indicator 2 (Cases)

50

100

0.6

Mean Difference, Indicator 1

50

Input Indicator 3 (Cases)

0.4

Mean Difference, Indicator 2

1.3 1.1 0.8

Mean Difference, Indicator 3

200

0.8

1.2 0.8

50

Input Indicator 4 (Cases)

150

0.5

200

100

Mean Difference, Indicator 3

100

150

Input Indicator 4 (Cases) Mean Difference, Indicator 2

1.6 1.2 0.8

Mean Difference, Indicator 4

50

Input Indicator 2 (Cases)

100

50

Input Indicator 1 (Cases)

0.4

1.3 1.0 0.7

50

Input Indicator 1 (Cases)

Mean Difference, Indicator 3

200

200

1.0

150

150

0.8

100

100

0.6

1.1 0.9

50

50

Input Indicator 2 (Cases)

Mean Difference, Indicator 4

200

Mean Difference, Indicator 1

150

0.8

100

0.4

50

Input Indicator 1 (Cases)

Mean Difference, Indicator 1

0.8

1.2

J.J. Blanchard et al. / Schizophrenia Research 77 (2005) 151–165

0.7

Mean Difference, Indicator 4

Mean Difference, Indicator 2

158

Input Indicator 3 (Cases)

50

100

150

200

Input Indicator 4 (Cases)

Fig. 2. Individual MAMBAC (mean above minus below a cut) curves for SANS subscales showing curves computed for each pair of subscales (Input Indicator 1 = Anhedonia–Asociality, Input Indicator 2 = Blunted Affect, Input Indicator 3 = Avolition–Apathy, Input Indicator 4 = Alogia). Each curve represents 50 evenly-spaced cuts.

individuals’ belonging to the taxon (Waller and Meehl, 1998). Schizophrenia patients were assigned to the taxon or the nontaxon class on the basis of the taxon-membership probability score. Patients with probability scores greater than or equal to 0.50 were assigned to the taxon class, and those with probability scores less than 0.50 were assigned to the nontaxon group. Analyses were conducted to examine demographic, social functioning, and symptom differences between the groups. 2.5. External component of the latent class model 2.5.1. Demographics Demographic characteristics of the groups are presented in Table 4. The two groups did not differ in age, t(236) = 0.33, p N 0.05. The chi-square analysis for gender was significant, v 2 = 5.24, p b 0.05, indicating that taxon members had a higher proportion of

males (75.8%) than nontaxon members (59.9%). The groups did not differ in racial composition, v 2 = 2.31, p N 0.05. With regard to marital status, the chi-square Table 4 Demographic characteristics of nontaxon (N = 172) and negative symptom taxon schizophrenia patients (N = 66)

Age, M (SD) Sex (% male) Race (%) White Black Asian Hispanic Marital (%) Married Divorced, widowed or separated Never married

Nontaxon (N = 172)

Taxon (N = 66)

29.87 (7.47) 59.9

29.52 (6.82) 75.8

36.0 55.2 0.6 3.5

43.9 48.5 1.5 1.5

2.9 19.2

6.1 6.1

77.9

87.9

J.J. Blanchard et al. / Schizophrenia Research 77 (2005) 151–165

was significant, v 2 = 7.13, p b 0.05. Nontaxon members had higher rates of ever being married compared to taxon members (22.3% versus 12.2%). Given the treatment conditions of TSS (family interventions and pharmacotherapy), we also examined possible group differences in treatment assignment to ensure that observed group differences were not associated with differential treatment. The groups did not differ in assignment to family management (applied or supportive), v 2 = 0.25, p N 0.05, or assignment to double-blind medication (standard, low dose, or targeted), v 2 = 0.67, p N 0.05. 2.5.2. Symptoms A repeated-measures MANOVA was conducted on the BPRS subscales comparing taxon and nontaxon schizophrenia patients over the two assessments (see Table 5). Given gender differences in the groups, we included gender as a factor in these analyses. There were significant main effects for Time, F(4,229) = 3.27, p b 0.05, and Group, F(4,229) = 20.38, p b 0.001. There were no significant findings for the main effect of Gender, F(4,229) = 2.29, p N 0.05, nor were there any significant interactions between Gender, Group, and Time: Gender  Group interaction, F(4,229) = 0.96, p N 0.05, Group  Time interaction, F(4,229) = 0.84, p N 0.05, Gender  Time interaction, F(4,229) = 0.59, p N 0.05, Gender  Group  Time, F(4,229) = 0.37, p N 0.05. Univariate tests examining the within-subjects effect of Time were significant for Disorganization, F(1,232) = 10.66, p b 0.005, but not for any other BPRS scale (all p’s N 0.05). An examination of the means for Disorganization indicated an increase in this scale rating over the two assessments.

Table 5 Brief Psychiatric Rating Scale (BPRS) scores in nontaxon (N = 171) and taxon (N = 65) schizophrenia patients at start of Double-Blind (DB) and 6-month Follow-up (FU) BPRS subscale Nontaxon M (SD) DB Thought disturbance Anergia Affect Disorganization

FU

Taxon M (SD) DB

FU

1.59 (0.85) 1.61 (0.95) 1.70 (0.88) 1.81 (0.98) 1.69 (0.61) 1.60 (0.59) 2.52 (0.86) 2.46 (0.92) 1.70 (0.73) 1.67 (0.78) 1.89 (0.88) 1.86 (0.75) 1.50 (0.61) 1.59 (0.62) 1.78 (0.80) 2.01 (0.80)

159

Univariate tests examining the main effect of Group were significant for Anergia, F(1,232) = 71.22, p b 0.001, and Disorganization, F(1,232) = 13.87, p b 0.001, but there were no group differences for Thought Disturbance, F(1,232) = 1.51, p N 0.05, or Affect, F(1,232) = 2.77, p N 0.05. These findings indicate that taxon and nontaxon members did not differ in psychotic symptoms or in affective symptoms. The negative symptom taxon group had higher ratings on both Anergia and Disorganization. This group difference in Anergia likely reflects this scale’s inclusion of the negative symptom of blunted affect and other symptoms conceptually related to negative or deficit symptoms (e.g., emotional withdrawal). Although modest, the higher scores of Disorganization within the taxon group were unexpected. Given these results, we sought to examine what, if any, contribution Disorganization made to the obtained taxometric findings. That is, does Disorganization serve as an indicator of the identified taxon? In order to address this question, we conducted MAMBAC analyses pairing Disorganization with each SANS subscale rating. As with the SANS scales, for these taxometric analyses we created an averaged rating of Disorganization using the two assessment ratings. The test–retest correlation, r = 0.60 ( p b 0.001), indicated that Disorganization ratings were relatively stable and supported the use of such an averaged rating. Prior to conducting MAMBAC analyses, we examined the zero-order correlations between Disorganization and the SANS scales. If Disorganization and the SANS scales jointly identify a taxon, one would expect at least modest correlations between these measures in the full mixed sample (Meehl, 1999). Disorganization had nonsignificant, near-zero correlations with Anhedonia-Asociality (r = 0.01), Blunted Affect (r = 0.02), Avolition-Apathy (r = 0.01), and Alogia (r = 0.03), all p’s N 0.05. Results from MAMBAC with Disorganization and each SANS scale failed to indicate curves associated with taxonicity—curves were dish shaped or failed to show clear convexity.1 Additionally, for each SANS subscale pairing with Disorganization, base rate estimates were inconsistent, ranging widely from 0 to 1.0,

1 Full graphical results of MAMBAC analyses with Disorganization are available from the authors.

160

J.J. Blanchard et al. / Schizophrenia Research 77 (2005) 151–165

and averaging around 0.50—findings consistent with dimensional-only structure (Meehl and Yonce, 1994). These MAMBAC results indicate that Disorganization is not an indicator of the taxon identified by negative symptoms. 2.5.3. Social functioning Comparisons on global scores for the SAS are presented in Table 6. t-Tests indicated that taxon members had significantly higher scores (reflecting poorer functioning) than nontaxon members on each of the SAS scales ( p’s b 0.05) with the exception of Relationship with Family. These results indicate that negative symptom taxon members demonstrated generally poorer social functioning than did other patients with schizophrenia. Given the greater representation of males in the deficit group and the association of worse social functioning in men compared to women with schizophrenia (e.g., Mueser et al., 1990; Shtasel et al., 1992), we considered the possibility that group differences in social functioning might be attributable in part to sex differences. To examine this possibility, we conducted stepwise regression analyses on each of the SAS scales, first stepping in sex followed by group membership (taxon vs. nontaxon) to determine whether taxon status explained variance in functioning above and beyond that accounted for by sex. Results are summarized in Table 7. Replicating the above group comparisons, taxon membership contributed a significant increment in explained variance for each of the

Table 6 Social functioning in nontaxon (N = 165) and taxon (N = 59) schizophrenia patientsa Social adjustment scale

Nontaxon M (SD)

Taxon M (SD)

Instrumental role performance Household integration Relationship with familya Social/leisure General adjustment

4.54 (1.16)

5.10 (1.05)

3.91 4.01 4.70 4.55

4.54 4.30 5.18 5.06

(1.13) (1.07) (1.23) (0.94)

(1.04) (0.92) (1.13) (0.91)

t 3.45* 4.03** ns 2.81* 3.93**

Higher scores reflect poorer functioning. a Ratings for the Relationship with Family item were missing for 21 participants, resulting in reduced Ns for the nontaxon (N = 149) and taxon (N = 54) comparison. * p b 0.005. ** p b 0.001.

Table 7 Social functioning: contributions of sex and taxon membership Social adjustment scale

Instrumental role performance Household integration Relationship with familya Social/leisure General adjustment

Step 1 Sex

Step 2 Taxon membership

R2

F

DR 2

DF

0.02

4.88*

0.06

13.51***

0.03 0.02 0.01 0.04

7.36** 3.24 1.99 8.21**

0.05 0.01 0.02 0.04

11.54** 2.42 4.27* 8.69***

*p b 0.05, **p b 0.01, ***p b 0.005. a Ratings for the Relationship with family item were missing for 21 participants, resulting in reduced N’s for the nontaxon (N = 149) and taxon (N = 54) comparison.

SAS scale scores, with the exception of Relationship with Family (which was unrelated to either sex or group membership). It is interesting to note that in each significant comparison, the explained variance accounted for by taxon membership met or exceeded that related to sex.

3. Discussion Negative symptoms constitute a clinically and theoretically important domain of symptoms in schizophrenia. However, previous efforts to determine whether negative symptoms differ across affected individuals either quantitatively or qualitatively have relied on procedures that are limited in their ability to resolve taxonic structure. Using taxometric statistical procedures (Meehl, 1995, 1999; Waller and Meehl, 1998) that were specifically developed to address this issue, the latent structure of negative symptoms was examined in a relatively large sample of well-characterized schizophrenia patients who were also assessed for demographics, symptoms, and social functioning. Results within and across taxometric procedures were consistent with a taxonic latent structure, supporting the existence of a discrete class of individuals with elevated negative symptom levels and a base rate of 28–36%. Furthermore, the external validity of the identified taxon was supported by findings that taxon members included a higher proportion of males and demonstrated worse social functioning than nontaxon members, while the groups did not differ in terms of conceptually unrelated symptoms.

J.J. Blanchard et al. / Schizophrenia Research 77 (2005) 151–165

Pending replication, the results of this first taxometric study of negative symptoms have important implications for optimal assessment of negative symptoms and for selection of research strategies in studies of their underlying causes. Several characteristics of the negative symptom taxon members identified in this study bear considerable resemblance to previous studies of individuals with elevated negative symptoms, including those meeting formal clinical criteria for the deficit syndrome (Kirkpatrick et al., 2001). The estimated base rate of the taxon (28–36%) falls within the broad range of prior estimates for a putative negative symptom syndrome (ranging from 6% to 40%; Fenton and McGlashan, 1992). In addition, this base rate is very similar to prevalence estimates of 25–30% using a more restrictive definition based on deficit syndrome criteria in chronic schizophrenia samples (Kirkpatrick et al., 2001). Converging evidence thus suggests that about one-third of schizophrenia patients in typical, chronically ill samples may be negative symptom taxon members. The external component of the latent class model indicated that negative symptom taxon members were characterized by a sensible pattern of demographic, symptom, and social functioning characteristics, and helped rule-out some alternative explanations for the taxonic structural findings. While taxon members did not differ from non-taxon members in age, they did include a significantly higher proportion of males (76% versus 60%). Deficit syndrome samples have repeatedly been found to include higher proportions of males than females (Kirkpatrick et al., 2000; Roy et al., 2001). In addition, there were no group differences in assignment to family management or medication treatment conditions, indicating that the structural findings are not simply attributable to study design features. Taxon and nontaxon members did not differ in psychotic or mood symptoms, suggesting that the identified negative symptom taxon is not merely a secondary consequence of elevated levels of positive symptoms or mood-related symptoms such as depression or anxiety (e.g., Carpenter et al., 1988). Taxon members did have greater BPRS Anergia scores, most likely reflecting overlap with the negative features assessed by this subscale (e.g., blunted affect). Unexpectedly, taxon members showed an elevation (albeit

161

modest) in BPRS ratings of Disorganization compared to nontaxon members. However, supplementary analyses indicated that Disorganization was minimally correlated with the SANS ratings and did not serve as an indicator of the identified negative symptom taxon. Taxon members also demonstrated poorer community functioning than those patients who were not in the taxon. Specifically, clinician ratings indicated that taxon members had worse functioning across the domains of vocational functioning, family support, social networks, and leisure activities. These pervasive group differences held even after controlling for sex differences, and are consistent with findings of social dysfunction from other categorical studies of negative symptoms (Buchanan et al., 1990; Fenton and McGlashan, 1994; Horan and Blanchard, 2003). The results of this preliminary taxometric studyraise the question, bWhat is the practical significance of finding that negative symptoms demonstrate a latent taxonic structure?Q The answer to this question has several facets (Meehl, 1992). First, this finding is theoretically important as it corroborates models that posit the existence of a latent class of negative symptom schizophrenia and supports the construct validity of this typological entity. Second, information about the latent structure of negative symptoms has important implications for optimal assessment and classification. Assessment of negative symptoms in a manner that is consistent with their latent structure will increase validity, reliability, and statistical power to detect important relationships with other variables (see Ruscio and Ruscio, 2002, 2004). In the case of a taxonic latent structure, the critical assessment issue is to maximize the accuracy of classification of individuals to a category (negative symptom taxon versus nontaxon classes), rather than determining where individuals fall with respect to an underlying continuum. For example, taxometrics can be used to identify the most valid available indicators of a negative symptom taxon (including an evaluation of putative indicators from across multiple levels of analysis such as clinical interview, self-report, cognitive, and psychophysiological; Meehl, 1995). In addition, base rate estimates derived from taxometric analyses may be substituted into Bayes’ theorem and used to assign individual cases a probability of taxon membership. This information provides a basis for

162

J.J. Blanchard et al. / Schizophrenia Research 77 (2005) 151–165

classifying individual cases as taxon vs. nontaxon members, and establishing optimal cutting scores on available indicators that minimize classification errors (Ruscio and Ruscio, 2002, 2004). These quantitative methods may help minimize error associated with cutpoints that are based on professional fiat (Waller et al., 1996). The third implication of a latent taxonic structure concerns the guidance it provides in selecting optimal research strategies for identifying causal factors. The major etiological implication of a taxonic finding is that it supports the search for some dichotomous causal agent to explain the observed discontinuity (Waller and Meehl, 1998; see further discussion below). Insights into the relationship between negative symptoms and other causally-related variables of interest will be facilitated by studies comparing individuals with schizophrenia who are members of the taxon to those who are not (Ruscio and Ruscio, 2002, 2004), as done in the present study. Finally, classifying individuals into classes that may have differential causal factors may be relevant clinically (Meehl, 1992, 1995). Such a classification may be useful in understanding prognosis (e.g., taxon members in the current study were found to have poorer social functioning). Treatments may also be developed to target those causal factors that are ultimately found to be associated with taxon membership and thus enhance treatment matching within the heterogenous disorder of schizophrenia. While taxometric analyses provide a powerful tool for examining the latent structure of constructs in psychopathology and guiding research strategies, it is important to emphasize that taxometrics are not self-interpreting (Meehl, 2001). The present results cannot address the reason for the observed latent discontinuity of negative symptoms. Although genetically-based pathophysiological factors may be relevant (e.g., Carpenter et al., 1988; Kirkpatrick et al., 2001), taxonicity may result from the presence of a discontinuously distributed environmental factor (Meehl, 1973) and social processes cannot be ruled out (see Meehl, 1992). Additional research will be required to more closely examine possible etiological factors associated with the identified negative symptom taxon. Some additional limitations must also be acknowledged. The TSS data set provided a unique opportu-

nity to examine both the structural and external components of a latent class model of negative symptoms using data collected across multiple assessments. Despite these significant advantages and the relatively large size of this clinical sample, the sample size falls at the low end of recommendations for taxometric analyses, which typically include at least 300 cases (Meehl, 1995). It is noteworthy that a recent Monte Carlo examination of taxometric procedures found that MAXCOV was successful in identifying taxa and classifying cases as taxon versus nontaxon members using samples of 100 and 200 cases, even under the unfavorable circumstances that are often encountered in applied settings (e.g., high nuisance covariance, small number of indicators; Beauchaine and Beauchaine, 2002). However, the results of the current study must be considered preliminary and replication with larger samples and alternative negative symptom indicators is required. Although the present findings appear consistent in many respects with prior research on the deficit syndrome, a putative schizophrenia subtype (Kirkpatrick et al., 2001), deficit symptoms were not directly assessed. The taxometric analyses in this study were based on SANS ratings and deficit syndrome ratings as developed by Carpenter et al. (1988) were not available for comparison. It is noteworthy that SANS ratings have shown reasonably good convergence with deficit syndrome ratings in prior research (Fenton and McGlashan, 1992). A potential advantage of the SANS is that this instrument does not implicitly require a categorical perspective as is found in ratings of the deficit syndrome, nor were the symptom ratings within TSS originally driven by any a priori hypotheses regarding the latent structure of negative symptoms. These features of the SANS may be important as Beauchaine and Waters (2003) have shown that pseudotaxonicity may occur in situations were raters hold expectations of a categorical model. A final concern is that it is not possible to determine what, if any, effect medications may have had on the current findings. The taxon and nontaxon groups did not differ in assignment to family management (applied or supportive) or to double-blind medication (standard, low dose, or targeted dose of fluphenazine), suggesting that differential medication exposure did not substantially influence the structural findings or

J.J. Blanchard et al. / Schizophrenia Research 77 (2005) 151–165

the group differences on external variables. Nonetheless, it will be important to determine whether the current findings, based on a chronically ill sample that was stabilized on antipsychotic medications, generalize to other patient samples. In summary, the present findings are consistent with a latent discontinuity in negative symptoms within schizophrenia. The negative symptom taxon was comprised of approximately 28–36% of the sample and was characterized by higher representation of males and poorer social functioning compared to those schizophrenia patients not in the taxon. These results are consistent with a categorical model of negative symptoms and demonstrate the utility of taxometric statistical methods for the further exploration of this class of schizophrenia patients. However, caution is warranted in the interpretation of these preliminary results and replication in other diverse samples and using other indicators of the negative symptom taxon will be an important next step in corroborating these findings.

Acknowledgments Preparation of this article was supported in part by National Institute of Mental Health grant MH51240 to Dr. Blanchard. The authors thank Dr. Nina R. Schooler for kindly providing access to data from the Treatment Strategies in Schizophrenia (TSS) Cooperative Agreement Program and for her comments on an earlier draft of this manuscript. The TSS Cooperative Agreement Program was a multicenter clinical trial carried out by five research teams in collaboration with the Division of Clinical Research of the National Institute of Mental Health (NIMH). The NIMH Prinicipal Collaborators were Nina R. Schooler, Samuel J. Keith, Joanne B. Severe, and Susan M. Mathews. The principal investigors at the five sites and grant numbers were Albert Einstein School of Medicine and Hillside Hospital, Glen Oaks, New York, UO1MH39992 (John M. Kane), Medical College of Pennsylvania and Eastern Pennsylvania Psychiatric Institute, Philadelphia, UO1-MH39998 (Alan S. Bellack); Cornell University Medical College and Payne Whitney Clinic, New York, UO1-MH40007 (Ira D. Glick); University of California at San Francisco

163

General Hospital, UO1-MH40042 (William A. Hargreaves); Emory University and Grady Memorial Hospital, Atlanta, Georgia, UO1-MH40597 (Philip T. Ninan).

References American Psychiatric Association, 1987. Diagnostic and Statistical Manual of Mental Disorders, third ed. Author, Washington, DC (rev.) Andreasen, N.C., 1982. Negative symptoms in schizophrenia: definition and reliability. Arch. Gen. Psychiatry 39, 784 – 788. Andreasen, N.C., 1985. Positive vs. negative schizophrenia: a critical evaluation. Schizophr. Bull. 11, 380 – 389. Andreasen, N.C., Arndt, S., Alliger, R., Miller, D., Flaum, M., 1995. Symptoms of schizophrenia: methods, meanings, and mechanisms. Arch. Gen. Psychiatry 52, 341 – 351. Beauchaine, T.P., Beauchaine III, R.J., 2002. A comparison of maximum covariance and k-means cluster analysis in classifying cases into known taxon groups. Psychol. Methods 7, 245 – 261. Beauchaine, T.P., Waters, E., 2003. Pseudotaxonicity in MAMBAC and MAXCOV analyses of rating-scale data: turning continua into classes by manipulating observers expectations. Psychol. Methods 8, 3 – 15. Bilder, R.M., Mukherjee, S., Rieder, R.O., Pandurangi, A.K., 1985. Symptomatic and neuropsychological components of defect states. Schizophr. Bull. 11, 409 – 419. Blanchard, J.J., Gangestad, S.W., Brown, S.A., Horan, W.P., 2000. Hedonic capacity and schizotypy revisited: a taxometric analysis of social anhedonia. J. Abnorm. Psychology 109, 87 – 95. Blanchard, J.J., Sayers, S.L., Collins, L.M., Bellack, A.S., 2004. Affectivity in the problem-solving interactions of schizophrenia patients and their family members. Schizophr. Res. 69, 105 – 117. Bryson, G., Whelahan, H.A., Bell, M., 2001. Memory and executive function impairments in deficit syndrome schizophrenia. Psychiatry Res. 102, 29 – 37. Buchanan, R.W., Carpenter, W.T., 1994. Domains of Psychopathology. An approach to the reduction of heterogeneity in schizophrenia. J. Nerv. Ment. Dis. 182, 193 – 204. Buchanan, R.W., Kirkpatrick, B., Heinrichs, D.W., Carpenter, W.T., 1990. Clinical correlates of the deficit syndrome of schizophrenia. Am. J. Psychiatry 147, 290 – 294. Carpenter, W.T., Heinrichs, D.W., Wagman, A.M.I., 1988. Deficit and nondeficit forms of schizophrenia: the concept. Am. J. Psychiatry 145, 578 – 583. Cleland, C.M., Rothschild, L., Haslam, N., 2000. Detecting latent taxa: Monte Carlo comparison of taxometric, mixture model, and clustering procedures. Psychol. Rep. 87, 37 – 47. Cohen, J.D., 1988. Statistical Power Analysis for the Behavioral Sciences, second ed. Erlbaum, Hissdale, NJ. Earnst, K.S., Kring, A.M., 1997. Construct validity of negative symptoms: an empirical and conceptual review. Clin. Psychol. Rev. 17, 167 – 189.

164

J.J. Blanchard et al. / Schizophrenia Research 77 (2005) 151–165

Emsley, R., Rabinowitz, J., Torreman, M., 2003. The factor structure for the Positive and Negative Syndrome Scale (PANSS) in recent-onset psychosis. Schizophr. Res. 61, 47 – 57. Fenton, W.S., McGlashan, T.H., 1992. Testing systems for assessment of negative symptoms in schizophrenia. Arch. Gen. Psychiatry 49, 179 – 184. Fenton, W.S., McGlashan, T.H., 1994. Antecedents, symptom progression, and long-term outcome of the deficit syndrome in schizophrenia. Am. J. Psychiatry 151, 351 – 356. Fitzgerald, P.B., de Castella, A.R., Brewer, K., Filia, K., Collins, J., Davey, P., et al., 2003. A confirmatory factor analytic evaluation of the pentagonal PANSS model. Schizophr. Res. 61, 97 – 104. Fouldrin, G., Jay, M., Bonnet-Brilhault, F., Menard, J.-F., Petit, M., Thibaut, F., 2001. Concordance of deficit and non-deficit subtypes in siblings affected with schizophrenia. Psychiatry Res. 102, 59 – 64. Gangestad, S., Snyder, M., 1985. bTo carve nature at its jointsQ: on the existence of discrete classes in personality. Psychol. Rev. 92, 317 – 349. Golden, R., Meehl, P.E., 1980. Detection of biological sex: an empirical test of cluster methods. Multivariate Behav. Res. 15, 475 – 496. Haslam, N., Kim, H.C., 2002. Categories and continua: a review of taxometric research. Genet. Soc. Gen. Psychol. Monogr. 128, 271 – 320. Heckers, S., Goff, D., Schacter, D.L., Savage, C.R., Fischman, A.J., Alpert, N.M., et al., 1999. Functional imaging of memory retrieval in deficit vs. nondeficit schizophrenia. Arch. Gen. Psychiatry 56, 1117 – 1123. Horan, W.P., Blanchard, J.J., 2003. Neurocognitive, social, and emotional dysfunction in deficit syndrome schizophrenia. Schizophr. Res. 65, 125 – 137. Horan, W.P., Blanchard, J.J., Gangestad, S.W., Kwapil, T., 2004. The psychometric detection of schizotypy: do putative schizotypy indicators identify the same latent class? J. Abnorm. Psychology 113, 339 – 357. Kay, S.R., Fiszbein, A., Opler, L.A., 1987. The Positive and Negative Syndrome Scale (PANSS) for schizophrenia. Schizophr. Bull. 13, 261 – 276. Kay, S.R., Opler, L.A., Lindenmayer, J.P., 1989. The Positive and Negative Syndrome Scale (PANSS): rationale and standardization. Br. J. Psychiatry 155 (Suppl. 7), 59 – 65. Kirkpatrick, B., Buchanan, R.W., 1990. Anhedonia and the deficit syndrome of schizophrenia. Psychiatry Res. 31, 25 – 30. Kirkpatrick, B., Ross, D.E., Walsh, D., Karkowski, L., Kendler, K.S., 2000. Family characteristics of deficit and nondeficit schizophrenia in the Roscommon family study. Schizophr. Res. 45, 57 – 64. Kirkpatrick, B., Buchanan, R.W., Ross, D.E., Carpenter, W.T., 2001. A separate disease within the syndrome of schizophrenia. Arch. Gen. Psychiatry 58, 165 – 171. Lenzenweger, M.F., 1999. Deeper into the schizotypy taxon: on the robust nature of maximum covariance analysis. J. Abnorm. Psychology 108, 182 – 187. Lenzenweger, M.F., Dworkin, R.H., Wethington, E., 1989. Models of positive and negative symptoms in schizophrenia: an empiri-

cal evaluation of latent structures. J. Abnorm. Psychology 98, 62 – 70. Loevinger, J., 1957. Objective tests as instruments of psychological theory. Psychol. Rep. 3, 635 – 694. Meehl, P.E., 1973. MAXCOV-HITMAX: a taxonomic search method for loose genetic syndromes. In: Meehl, P.E. (Ed.), Psychodiagnosis: Selected Papers. University of Minnesota Press, Minneapolis, pp. 200 – 224. Meehl, P.E., 1992. Factors and taxa, traits and types, differences of degree and differences in kind. J. Personal. 60, 117 – 174. Meehl, P.E., 1995. Bootstraps taxometrics: solving the classification problem in psychopathology. Am. Psychol. 50, 266 – 275. Meehl, P.E., 1999. Clarifications about taxometric method. Appl. Prev. Psychol. 8, 165 – 174. Meehl, P.E., 2001. Primary and secondary hypohedonia. J. Abnorm. Psychology 110, 188 – 193. Meehl, P.E., Golden, R., 1982. Taxometric methods. In: Kendall, P., Butcher, J. (Eds.), Handbook of Research Methods in Clinical Psychology. Wiley, New York, pp. 127 – 181. Meehl, P.E., Yonce, L.J., 1994. Taxometric analysis: I. Detecting taxonicity with two quantitative indicators using means above and below a sliding cut (MAMBAC procedure). Psychol. Rep. 74, 1059 – 1274. Meehl, P.E., Yonce, L.J., 1996. Taxometric analysis: II. Detecting taxonicity using covariance of two quantitative indicators in successive intervals of a third indicator (MAXCOV procedure). Psychol. Rep. 78, 1091 – 1227. Mueser, K.T., Bellack, A.S., Morrison, R.L., Wade, J.H., 1990. Gender, social competence, and symptomatology in schizophrenia: a longitudinal analysis. J. Abnorm. Psychology 99, 138 – 147. Mueser, K.T., Sayers, S.L., Schooler, N.R., Mance, R.M., Haas, G.L., 1994. A multisite investigation of the reliability of the Scale for the Assessment of Negative Symptoms. Am. J. Psychiatry 151, 1453 – 1462. Mueser, K.T., Curran, P.J., McHugo, G.J., 1997. The factor structure of the Brief Psychiatric Rating Scale in schizophrenia. Psychol. Assess. 9, 196 – 204. Mueser, K.T., Sengupta, A., Schooler, N.R., Bellack, A.S., Xie, H., Glick, I.D., et al., 2001. Family treatment and medication dosage reduction in schizophrenia: effects on patient social functioning, family attitudes, and burden. J. Consult. Clin. Psychol. 69, 3 – 12. Overall, J.E., Gorham, D.R., 1962. The Brief Psychiatric Rating Scale. Psychol. Rep. 10, 799 – 812. Pogue-Geile, M.F., Keshavan, M.S., 1991. Negative symptomatology in schizophrenia: syndrome and subtype status. In: Greden, J.F., Tandon, R. (Eds.), Negative Schizophrenic Symptoms: Pathophysiology and Clinical Implications. American Psychiatric Press, Washington, DC, pp. 41 – 59. Reise, S.P., Gomel, J.N., 1995. Modeling quantitative variation within latent trait dimensions: application of mixed-measurement to personality assessment. Multivariate Behav. Res. 30, 341 – 358. Ross, D.E., Thaker, G.K., Buchanan, R.W., Lahti, A.C., Medoff, D., Bartko, J.J, et al., 1996. Association of abnormal smooth pursuit

J.J. Blanchard et al. / Schizophrenia Research 77 (2005) 151–165 eye movements with the deficit syndrome in schizophrenic patients. Am. J. Psychiatry 153, 1158 – 1165. Ross, D.E., Kirkpatrick, B., Karkowski, L.M., Straub, R.E., McLean, C.J., O’Neill, F.A., et al., 2000. Sibling correlation of deficit syndrome in the Irish study of high density schizophrenia families. Am. J. Psychiatry 157, 1071 – 1076. Roy, M.-A., Maziade, M., Labbe, A., Merette, C., 2001. Male gender is associated with deficit schizophrenia: a meta-analysis. Schizophr. Res. 47, 141 – 147. Ruscio, J., 2004. Program Code and Documentation Files for Performing Taxometric Analyses and Simulating Comparison Data. Elizabethtown College, Department of Psychology, Elizabethtown, PA. www.etown.edu/psychology/faculty/ruscio.htm. Ruscio, J., Ruscio, A.M., 2002. A structure-based approach to psychological measurement: matching measurement models to latent structure. Assessment 9, 4 – 16. Ruscio, J., Ruscio, A.M., 2004. Clarifying boundary issues in psychopathology: the role of taxometrics in a comprehensive program of structural research. J. Abnorm. Psychology 113, 24 – 38. Sayers, S.L., Curran, P.J., Mueser, K.T., 1996. Factor structure and construct validity of the Scale for the Assessment of Negative Symptoms. Psychol. Assess. 8, 269 – 280. Schooler, N.R., Hogarty, G.E., Weissman, M.M., 1979. Social Adjustment Scale II. In: Hargraves, W.A., Atkinson, C.C., Sorenson, J.E. (Eds.), Resource Materials for Community Mental Health Program Evaluators: DHEW Publication 79-328. US Government Printing Office, Washington, DC, pp. 290 – 303. Schooler, N.R., Keith, S.J., Severe, J.B., Matthews, S.M., Bellack, A.S., Glick, I.D., et al., 1997. Relapse and rehospitalizaiton during maintenance treatment of schizophrenia: the effects of

165

dose reduction and family treatment. Arch. Gen. Psychiatry 54, 453 – 463. Shtasel, D., Gur, R.E., Gallacher, F., Heimberg, C., Gur, R.C., 1992. Gender differences in the clinical expression of schizophrenia. Schizophr. Res. 7, 225 – 231. Smith, D.A., Mar, C.M., Turoff, B.K., 1998. The structure of schizophrenic symptoms: a meta-analytic confirmatory factor analysis. Schizophr. Res. 31, 57 – 70. Spitzer, R.L., Williams, J.B.W., Gibbon, M., First, M.B., 1990. User’s Guide for the Structured Clinical Interview for DSMIII-R: SCID. American Psychiatric Association, Washington, DC. Tyrka, A., Cannon, T.D., Haslam, N., Mednick, S.A., Schulsinger, F., Schulsinger, H., et al., 1995a. The latent structure of schizotypy: I. Premorbid indicators of a taxon of individuals at risk for schizophrenia-spectrum disorders. J. Abnorm. Psychology 104, 173 – 183. Tyrka, A., Haslam, N., Cannon, T.D., 1995b. Detection of a longitudinally-stable taxon of individuals at risk for schizophrenia spectrum disorders. In: Raine, A., Lencz, T., Mednick, S.A. (Eds.), Schizotypal Personality Disorder. Cambridge University Press, New York, pp. 168 – 191. Waller, N.G., Meehl, P.E., 1998. Multivariate Taxometric Procedures: Distinguishing Types From Continua. Sage, Newbury Park, CA. Waller, N.G., Putnam, F.W., Carlson, E.B., 1996. Types of dissociation and dissociative types: a taxometric analysis of dissociative experiences. Psychol. Methods 1, 300 – 321. Woerner, M., Mannuzza, S., Kane, J.M., 1988. Anchoring the BPRS: an aid to improved reliability. Psychopharmacol. Bull. 24, 112 – 117.