Schizophrenia
Research,
10 1992 Elsevier
SCHIZO
6 (1992)
75
75-85
Science Publishers
B.V. All rights reserved
0920-9964/92/$05.00
00 I86
The assessment of schizotypal features over two points in time Elizabeth Department
of Medico1
Squires-Wheeler,
Genetics.
Columbia
Andrew
University,
College
722 West 168th Street, (Received
20 August
E. Skodol of Physicians
and Surgeons,
New York, NY
1990, revised received
5 February
and L. Erlenmeyer-Kimling and New York State Psychiatric
Institute,
10032, U.S.A. 1991, accepted
11 February
1991)
The expression of schizotypal personality traits was assessed in mid-adolescence and again in young adulthood for three groups of offspring defined by the psychiatric diagnosis of their parents. Parental diagnoses included schizophrenic disorder (47 offspring), affective disorder (39 offspring), and ‘no psychiatric disorder’, or normal controls (82 offspring). Initially, schizotypal traits were assessed from video-taped semi-structured psychiatric interviews, subsequently rated by trained psychiatrists blind to the parental psychiatric status of the subjects, and/or direct clinical interviews (Schedule for Affective Disorders-Lifetime Version @ADS-L)). The second assessment was conducted by trained social workers and psychologists by means of a semi-structured interview specifically for DSM-III-R personality disorders (Personality Disorder Examination) and sections of the SADS-L where indicated. These interviewers were blind to the parental status and to previous psychiatric assessments of the offspring. The rates of stability of features or the rates of progression to axis I psychotic disorders (Schizophrenia, Schizoaffective Disorder, and Unspecified Functional Psychosis) were evaluated. Concordance of assessments over time is reported as a function of threshold for expression of traits at initial evaluation, i.e., two or more, three or more, or four or more features present. Concordance increases as the threshold for expression increases, as expected. The effect of comorbid clinical status, e.g., the coexistence of schizotypal traits and anxiety and/or depressive features on the concordance pattern, is also examined by parental diagnostic group status. The offspring of affective disorder parents exhibited higher rates of anxiety and/ or depressive features at both points in time, exhibited higher concordance for anxiety and/or depressive features, and exhibited higher rates of ‘transformation’ of initial schizotypal features to anxiety and/or depressive features at the second assessment. Key words:
Schizotypal
personality
traits;
Longitudinal
assessment
INTRODUCTION
The identification of a nonpsychotic diagnostic category, exhibiting specificity and sensitivity to biological relatives with schizophrenia would enhance the power of studies of pathophysiological mechanisms and etiological components of schizophrenic disorders. For example, linkage analysis Correspondence to: E. Squires-Wheeler, Department of Medical Genetics, Columbia University, College of Physicians and Surgeons, and New York State Psychiatric Institute, 722 West 168th Street, New York, NY 10032, U.S.A.
of schizophrenic disorder pedigrees will have increased power to detect linkage if a schizophrenic spectrum disorder is included as part of an ‘affected phenotype’ model. If the clinical boundaries of the spectrum are uncertain, however, the possible errors introduced reduce the expected increase in power and indeed may obscure a real underlying linkage result. Historically, procedures used to identify specific and sensitive spectrum categories have varied in rigor and methodologic safeguards, although all approaches have used family aggregation of clinical features as empirical observations to guide identification of spectrum candidates. The
76
criteria for schizotypal personality disorder (STPD), one spectrum diagnostic category appearing in DSM-III (1980) were derived from an examination of the clinical features that characterized borderline schizophrenic subjects who clustered in the first- and second-degree relatives of index adoptees in the Danish Adoption Study (Kety et al., 1968, 1975; Kety, 1983, 1985, 1987, 1988). Validity criteria for such proposed schizophrenic spectrum categories have been suggested by Shields et al. (1975) and recently restated by Kendler (1985). Criteria for spectrum categories should exhibit sensitivity and specificity in discriminating relatives of schizophrenic index cases. The distribution of DSM-III-R spectrum features STPD, has been previously examined in relatives of schizophrenic probands compared to normal controls by several investigators including Lowing et al. (1983), Baron et al. (1985), Frangos et al. (1985) and Kendler (1984, 1988a,b). In these studies the relative risks ranged from 2.4 to 9.1, thus supporting the validity criteria. It is desirable, however, to contrast the rates of spectrum features in the relatives of schizophrenic probands with the rates of such features in the relatives of probands with other psychiatric disorders, in addition to relatives of normal controls, in order to establish the specificity of association of the features to schizophrenic disorder. When probands with affective disorder have been used as psychiatric contrast groups, the rates of spectrum features in their relatives have not consistently differed from the rates of these features in the relatives of schizophrenic probands (Endicott et al., 1985; Ingraham, 1987, 1988; Squires-Wheeler et al., 1988). For example, in our own recent report (Squires-Wheeler et al., 1989) although the rates of schizotypal features were elevated in the relatives of probands with psychiatric disorder in contrast to the relatives of normal controls, the rates did not differ significantly between relatives of schizophrenic disorder probands and the relatives of affective disorder probands, thus failing to support the expected diagnostic specificity. To further explore the nature of the apparent lack of specificity, we have employed a model proposed by Rice and associates ( 1987) concerning the relationship between true trait status and stability of trait expression. Rice’s approach defines
a ‘true case’ as one exhibiting stability of expression over time. Here a true case is assumed to be more likely to share a genetic continuity with schizophrenic disorder, so that relatives of schizophrenic disorder probands might be expected to express schizotypal features that are stable and/or progressive. In contrast, the expression of schizotypal features at a given point in time in relatives of affective disorder probands and normal controls may represent state determined factors and thus be less likely to be persistent and observed at a later time. That is, expression of schizotypal features at a single point in time may represent transient states, rather than personality traits. Rice et al.‘s (1987) stability model was developed for affective disorder. He demonstrated that stability of expression was maximized for cases with severe clinical features, initially. For example, the number of depressive symptoms expressed at intake was positively associated with stability, as might be expected. DSM-III-R criteria for prodromal schizophrenia and schizotypal personality disorder reflect similar perceptual. cognitive, and behavioral disabilities. Features of schizotypal personality disorder in DSM-III were derived from the Danish Adoption Study series in which biological relatives were assessed in mid-adult life. under the assumption that their clinical status represented enduring traits. Given the relative ‘youthfulness’ of our offspring sample, and the considerable overlap of criteria for schizotypal and prodromal clinical pictures, it may be expected that some individuals expressing features when first evaluated may progress at follow-up to an axis I psychotic disorder such as schizophrenia, schizoaffective disorder, or unspecified functional psychosis. To examine alternative natural history patterns, three groups of offspring of parents (schizophrenic disorder parents, affective disorder parents, and normal control parents) have been assessed for schizotypal personality features and disorder at two points in time. These subjects are drawn from a longitudinal family study and constitute a subset of Sample A of the New York High-Risk Project (Erlenmeyer-Kimling and Cornblatt, 1987). In summary, the purpose of the study we are reporting was to test the dual hypotheses that ‘true cases’ of schizotypal personality disorder, i.e., those that show a specific familial relationship to
schizophrenia, are stable over time and that stable cases are those that on initial, cross-sectional assessment are more severe. The rates of concordance of two assessments are reported as a function of initial diagnostic threshold by parental diagnostic groups. Additionally the expression of anxiety and/or depressive features was noted at time 1 and time 2 by parental diagnostic groups. The coexistence of such affective features (with spectrum traits) was evaluated by diagnostic group.
METHODS Subjects
The subjects of this report are members of the initial sample (sample A) of the two-sample longitudinal New York High-Risk Project, which is described fully elsewhere (Erlenmeyer-Kimling and Cornblatt, 1987). This report is based on data obtained on subjects who were available for assessment for two measurement periods. Subjects included were evaluated (1) in mid-adolescence ~ average age of 16 years (time l), and (2) in young adulthood ~ average age of 25 years (time 2). The subjects are offspring of schizophrenic, affective disorder, and normal control parents. Ascertainment of parents was through the consecutive admission records of six New York State hospitals. Those patients who were married, white, Englishspeaking, passed exclusion diagnostic criteria, and had at least one offspring between the ages of seven and 12, were eligible for the study. Subsequent diagnostic assessments of parents were made using the Schedule for Affective Disorders and Schizophrenia-Lifetime Version (SADS-L) and Research Diagnostic Criteria (RDC) (Spitzer et al., 1978). Normal control families were secured through school systems, chosen to reflect the community characteristics from which the patient parents came. The offspring were free of psychiatric impairment at intake into the study (1971-1972). Table 1 displays the demographic characteristics of those sample A subjects present for the two measurement periods. Assessment
procedures:
The adolescent controls were
time I measurement
offspring of patient parents and interviewed, approximately eight
years after the initiation of the study, by a psychiatrist blind to parental diagnostic group using a semi-structured interview protocol, the Mental Health Assessment Form (MHAF) (Kestenbaum and Bird, 1977). The interviews were video-taped and edited to remove any reference to illness in the parents and were rated subsequently by at least two psychiatrists or psychiatric residents (also blind to the parental group status of the interviewee), who were trained in the assessment of MHAF indicators. These videotaped assessments preceded DSM-III by 2 years, and hence DSMIII schizotypal personality disorder was not an official diagnostic category at the time of the assessment. However, information relevant to the (subsequent) DSM-III criteria was elicited at the videotaped psychiatric interview (see SquiresWheeler et al. (1988) for a complete description of assessment and scoring). A subset of sample A study subjects was directly assessed with the SADS-L, and rated according to the RDC. The interviewers were psychiatric social workers trained in the use of the SADS-L by members of the Biometrics Research Department at New York State Psychiatric Institute. Axis I disorders were the focus of the interview and the presence of schizotypal features was noted at this assessment only if the interviewee exhibited another diagnosis. Notwithstanding the incomplete nature of the schizotypal assessment at this time, the data on schizotypal features have been combined with the systematic MHAF assessment, such that an individual is noted as expressing a given number of features at time 1 if he/she expresses features on the MHAF or on the SADS-L or both. Assessment
procedures:
time 2 measurement
Diagnostic assessments at time 2 were made employing a blind, direct diagnostic evaluation using the Personality Disorder Examination (PDE) (Loranger et al., 1987). The PDE is a standardized, semi-structured clinical interview for eliciting information relevant to the Diagnostic and Statistical Manual of Mental Disorders, third edition-revised (DSM-III-R) axis II personality disorders. The PDE was modified by insertion of selected SADSL components. If subjects responded positively to PDE screening items for axis I disorders (e.g., depressive or anxiety disorders), the appropriate SADS-L section was completed. The modified
78 TABLE
I
Demographic characteristics oJsample assessed at time I and time 2: sex, social class, and age by group Characteristic
Groups: o$ypring of‘ parents with Schizophrenic Disorder n=45
Affective Disorder n=39
n
(Xl
n
IX)
n
i”/oi
Sex Males Females
22 23
(48.9) (51.1)
14 25
(35.9) (64.1)
44 35
(55.7) (44.3)
Social class Hollingshead & Redlich Index I Highest 2 3 4 5 Lowest
5 I 6 30 3
(11.1)
0 0 II 24 4
(0.0) (0.0) (28.2) (61.5) (10.3)
16 17 19 25 2
(20.3) (21.5) (24.1) (31.6)
“No disorder
equals
normal
(2.2) (13.3) (66.7) (6.7)
No Disorder” n=79
(2.5)
control
PDE was administered to subjects by clinical psychologists and social workers after a period of orientation and training in the use of the schedule by means of group workshops and individual supervision. The interviewers were blind to parental diagnostic status and to results of previous clinical assessments of the offspring. See SquiresWheeler et al. (1989) for a full description of training and clinical consensus procedures. Table 2 shows the nine features of DSM-III-R criteria of Schizotypal Personality Disorder.
RESULTS
Table 3 shows the rates of expression of schizotypal features by varying thresholds for presence of features, e.g., two or more, three or more, or four or more schizotypal features present. The inclusion of two or more, three or more or four or more features reflects the range of diagnostic thresholds employed in the literature (Baron et al., 1980, 1983a,b, 1985; Risch et al., 1984). The denominators of the proportions reflect the number of offspring who were assessed at both time 1 and time 2. The MHAF clinical assessment was conducted by psychiatrists and psychiatric residents
TABLE
2
DSM-III-R ,features,from schizotypal personality disorder 1. Ideas of reference (excluding delusions of reference) 2. Excessive social anxiety, e.g., extreme discomfort in social situations involving unfamiliar people 3. Odd beliefs or magical thinking, influencing behaviour and inconsistent with subcultural norms, e.g.. superstitiousness, belief in clairvoyance, telepathy, or ‘sixth sense’, ‘others can feel my feelings’ (in children and adolescents, bizarre fantasies or preoccupations) experiences, e.g., illusions. sensing the 4. Unusual perceptual presence of a force or person not actually present (e.g., ‘I felt as if my dead mother were in the room with me’) or appearance, e.g.. unkempt, 5. Odd or eccentric behaviour unusual mannerisms, talks to self (or only one) other than 6. No close friends or confidants first-degree relatives Odd speech (without loosening of associations or incoherence), e.g., speech that is impoverished, digressive, vague, or inappropriately abstract Inappropriate or constricted affect, e.g., silly. aloof, rarely reciprocates gestures or facial expressions, such as smiles or nods Suspiciousness or paranoid ideation
who were blind to parental clinical status and to subjects’ prior clinical and treatment status. Therefore schizotypal features were rated present if exhibited by the subjects irrespective of the subjects’ axis I status. Given this lack of an hierarchi-
79 TABLE
3
Parental
group proportions
Spectrum
exhibiting
indicated
number
Group proportion
offeatures
with disorder
or outcome al time I or time 2 present
criteria
Schziphrenic
AfSective
Normal
parental
disorder
control
group
paren tat
parenlal
group n=39”
group
difference’ A vs. B
A vs. C
B vs. C
n=37b
n=82” n=81b
(Al
iBi
ici
0.28 (0.21)
0.36 (0.32)
0.18 (0.17)
0.557
0.307
0.058
0.26 (0.19)
0.26 (0.22)
0.09 (0.07)
1.000
0.018
0.024
0.21 (0.14)
0.15 (0.11)
0.07 (0.06)
0.674
0.042
0.288
0.21 (0.14)
0.13 (0.08)
0.06 (0.05)
0.448
0.020
0.360
n=47” n=43b
Two or more schizotypal features [time I) Three or more schizotypal features (time 1) Four or more schizotypal features (time 1) Time 2 total spectrum outcorned
Sign$cance of proportion
disorder
aSubjects present at both time 1 and time %ubjects present at both time 1 and time ‘Significance levels associated with Fleiss’s dTotal spectrum includes axis I disorders and/or four or more schizotypal features.
2 assessment periods. 2 assessment periods without onset of Axis I disorder between time 1 and time 2. (1981) z statistic for proportion difference. (Schizophrenic Disorder, Schizoaffective Disorder and Unspecified Functional Psychosis)
cal assessment procedure, schizotypal features noted in the interview could represent prodromal or residual schizophrenic signs. There were four subjects from the schizophrenic disorder parental group (SDPG) who were diagnosed as schizophrenic, schizoaffective, or unspecified functional psychosis at the SADS-L interview conducted after the MHAF clinical assessment (considered here as part of the time 1 assessment) and for whom the schizotypal assessment most likely represented prodromal or residual schizophrenia. There were two such subjects from the affective disorder parental group (ADPG) and one such subject from the normal control parental group (NCPG). The proportions in parenthesis on Table 3 represent the rates for offspring from the respective parental groups, excluding these axis I subjects. The pair-wise group contrasts reaching less than 0.05 levels of significance were (1) time 1 three or more schizotypal features in offspring from the SDPG versus offspring from the NCPG; (2) time 1 three or more schizotypal features for offspring from the ADPG versus offspring from the NCPG; and (3) time 2 total spectrum outcome for offspring from the SDPG versus offspring from the NCPG. If the significance level is corrected for the number
of contrasts employed, i.e., corrected level of < 0.004 for 12 contrasts, there are no significant differences observed, although the pattern of results is generally consistent with expectations. Table 4 shows the association between assessment at time 1 and assessment at time 2 (total spectrum outcome) in terms of positive predictive values, negative predictive values, sensitivity, and specificity, for parental diagnostic groups. In this measurement context, time 1 assessment may be considered a ‘predictor’, while time 2 assessment represents the ‘outcome’. The positive predictive value represents the probability of expressing outcome at time 2 given expression at time 1. The higher this probability, the more predictive, or stable time 1 assessment is. The negative predictive value represents the probability of a normal assessment at time 2 (no features or disorder exhibited) given the absence of (a given number of features) expressed at time 1. Sensitivity represents the probability of expressing threshold features at time 1 given expression of total spectrum at time 2. (The complement of sensitivity reflects the proportion of subjects expressing total spectrum at time 2 who experienced a recent onset (i.e., onset of total spectrum some time after time 1, but before time
80 TABLE
4
Associations between rime I and time 2 assessments
Two or more schizoiypalfeatures
Three or more schizotypalfeatures (time I) Schizophrenic disorder parental group (n = 47) Affective disorder parental group (n = 39) Normal control parental group (n = 82) Four or more schizotypal features (time I) Schizophrenic disorder parental group (n=47) Affective disorder parental group (n = 39) Normal control parental group (n = 82) Predictive
NPV-
SEN”
SPE”
0.538
0.912
0.700
0.838
0.286
0.960
0.800
0.700
0.200
0.970
0.600
0.844
0.583
0.914
0.700
0.865
0.400
0.966
0.800
0.824
0.286
0.960
0.400
0.935
0.700
0.919
0.700
0.919
0.667
0.970
0.800
0.941
0.333
0.961
0.400
0.948
(time I)
Schizophrenic disorder parental group (n = 47) Affective disorder parental group (n = 39) Normal control parental group (n = 82)
“PPV = Positive
PPV
Value; NPV = Negative
Predictive
Value: SEN = Sensitivity;
2.) Specificity refers to the probability of a normal assessment at time 1 given a normal assessment at time 2. For the offspring from the two psychiatric parental groups, the positive predictive values, the negative predictive values, and the specificities increase as the diagnostic thresholds for affected status increase. For the offspring from the NCPG, the positive predictive values and the specificities increase as the diagnostic thresholds for affected status increase, although the positive predictive values remain low even for the highest threshold. As noted above, the positive predictive values suggest the percent of subjects who exhibit a stable, persistent, or progressive disorder. Employing Rice’s assumption that a stable case is more likely to reflect a true case, we note that increasing numbers of features at time 1 are associated with increasing probability of identifying a stable or ‘true case’. However, our expectation that such stability would characterize only the offspring of the SDPG was not observed. Offspring of the ADPG exhibited rates of stability, as quantified
SPE = Specificity
by positive predictive values, similar to those of offspring of the SDPG. In the light of the failure to demonstrate specificity of stability for probable schizotypal personality disorder, we asked the following exploratory question: Is the expression of schizotypal features over time different for the offspring groups in terms of comorbidity patterns? For example, do the offspring of the ADPG express schizotypal features in conjunction with affective features, at time 1, or time 2? A subset of offspring from the SDPG (94%) the ADPG (92%), and the NCPG (96%) was assessed for anxiety disorders (anxiety disorders are defined by Research Diagnostic Criteria definite thresholds for generalized anxiety. phobic disorder, or panic disorder) and affective disorders (depressive disorders are defined by Research Diagnostic Criteria definite thresholds for major depression or Intermittent Depression) at the SADSL assessment at time 1. The rates of depression and/or anxiety at time 1 for the SDPG, the ADPG and the NCPG offspring are 45%, 68%, and 37%,
81
respectively. At time 2, the rates of depression and/or anxiety for the SDPG, the ADPG, and the NCPG offspring are 13.6%, 34%, and 17% respectively. The rates of concordance of the two assessments for the groups, SDPG, ADPG and NCPG are 3.6%, 36%, and 8% respectively. That is, the persistence of depression and/or anxiety for subjects from all groups is low, with the ADPG showing the highest stability of expression. Over one-third of the subjects from the ADPG expressed stable depression and/or anxiety disorder (or frequent recurrences identified at both periods). Table 5 shows the proportion of total spectrum outcomes as a function of comorbid clinical features expressed at time 1. Among the SDPG offspring, 18 (41 “h) expressed neither anxiety, depression or schizotypal features at time 1 and only one (5.5%) of these subjects expressed total spectrum outcome at time 2. Of 16 SDPG subjects (36%) exhibiting depression or anxiety but not four or more schizotypal features at time 1, one (6.25%) subsequently showed total spectrum outcome at time 2. Five (83%) of the six SDPG subjects (14%) who expressed four or more schizotypal features but not depression or anxiety at time 1 demonstrated total spectrum outcome at time 2, and two (50%) of the four SDPG subjects (9%) exhibiting four or more schizotypal features
TABLE
and depression or anxiety at time 1 had total spectrum outcome at time 2. Of 36 offspring from the ADPG entering these assessments, 10 (28%) were free of depression, anxiety and four or more schizotypal features at time 1, and none of the subjects expressed total spectrum outcome at time 2. 20 (55.6%) subjects showed only depression or anxiety at time 1, and only one (5%) such subject exhibited total spectrum outcome at time 2. Both of the two ADPG subjects (5.5%) subjects who expressed four or more schizotypal features only at time 1 showed total spectrum outcome at time 2. Of the four ADPG subjects (11.1%) exhibiting both four or more schizotypal features and depression or anxiety at time 1, two (50%) expressed total spectrum outcome at time 2. Among the offspring from the NCPG entering these assessments, 45 (57%) showed none of the conditions considered at time 1; one (2.2%) of these subjects had total spectrum outcome at time 2. Of 28 NCPG subjects (35.5%) showing depression or anxiety but not four or more schizotypal features at time 1, two (7.1%) expressed total spectrum outcome at time 2. Two (50%) of the four NCPG subjects (5%) showing schizotypal features but not depression or anxiety at time 1 expressed total spectrum outcome at time 2,
5
Proportion of subjects(for whom affective features were evaluated) expressing time 2 total spectrum and parental diagnostic group Time I clinical status
Subject expressed. no spectrum or affective features only affective/anxiety features only spectrum features both spectrum and affective features Total time 2 spectrum outcome
Oflspring
outcome hJ time
I clinical status
From
Schizophrenic disorder parental
Aflective disorder purentul
Norm01 control parental
group (SDPC) n=44”
group (ADPG) n=36”
group (NCPC) n=19a
l/18 (0.055) I,!16 (0.0625) S/6 (0.833) 2/4 (0.500) 9b/44” (0.20)
OjlO (0.000) l/20 (0.05) 2!‘2 (1.0) 2/4 (0.50) 5bi36” (0.14)
1.45 (0.022) 2128 (0.072) 2.4 (0.50) 0,;2 (0.00) 5”/79” (0.063)
Subjects assessed for affective features at time 1. %ubjects assessed for affective features at time 1 and expressing total spectrum outcome at time 2. Only significant difference in Table 5 (p < 0.05) is contrast between SDPG and NCPG for total time 2 spectrum
outcome.
82
whereas neither of the two NCPG subjects (2.5%) showing four or more schizotypal features plus depression or anxiety at time 1 developed total spectrum outcome by time 2. A logistic regression was employed to assess the probability of expressing total spectrum outcome at time 2 as a function of the presence of schizotypal features at time 1 (four or more features present) controlling for the comorbid expression of anxiety and/or depression at time 1. The exponentiated coefficient for the time 1 schizotypal feature independent variable (see Table 6) can be interpreted as an odds ratio. This ratio is the odds of expressing time 2 total spectrum outcome given expression of time 1 schizotypal features over the odds of expressing time 2 total spectrum outcome given absence of schizotypal features at time 1. For example, for the offspring from the SDPG, the predicted probability of expressing total spectrum outcome at time 2 given expression of features at time 1 of 70%. The odds of expressing total spectrum outcome at time 2 given expression of features at time 1 is 2.33-I. The odds ratio adjusted for time 1 expression of comorbid anxiety and/or depression is 38.8, seen in Table 6. The coefficient indexing the prediction of time 1 schizotypal features controlling for comorbid clinical expression is statistically significant for offspring from all parental diagnostic groups except the offspring from the NCPG without axis I at time 1. While the logistic regression addressed the question of stability and/or progression as a function of clinical status at time 1, the question of comorbid clinical profile at time 2 was not examined. It may be hypothesized that the expression of spectrum features at time 1 among offspring from the ADPG may predict (i) an affective disorder at time 2 or (ii) a comorbid clinical profile reflecting both spectrum and affective features. Table 7 shows the rates at time 2 of(i) no follow-up, (ii) stability or progression of schizotypal features, (iii) comorbid expression of schizotypal features and depression and/or anxiety, (iv) depression and/or anxiety, and (v) no disorder.
CONCLUSIONS, IMPLICATIONS
LIMITATIONS
AND
A large percentage of subjects from groups expressed schizotypal features
all parental or disorder
likely to be state phenomena as inferred from instability of (failure to express) features over time. 28% of SDPG offspring, 34% of ADPG offspring, and 66% of NCPG offspring exhibiting at least four or more schizotypal features at time 1 were free of total spectrum disorder at time 2. The longitudinal picture of feature expression is clearly important and lack of a follow-up assessment obscures the distinction between state and trait expression. An informant who is able to provide such a longitudinal view may be necessary when direct follow-up assessment is not possible. We may tentatively conclude that rates of stability and/or progression of schizotypal features do not provide a means of differentiating the expression of features between the SDPG and the ADPG offspring as proposed. It is seen that ADPG offspring are the only group with ‘transformation’ of spectrum features (at time 1) to depression and/ or anxiety at time 2. 17% of ADPG offspring exhibited such a pattern. A further 17% of these subjects also expressed a comorbid profile consisting of spectrum and depression and/or anxiety at time 2. These proportions are not statistically significantly different between parental groups owing to the very small number of subjects expressing features and disorder at time 1. However, the pattern is consistent with the expectation that a subset of ADPG offspring express a potentially differentiable clinical picture longitudinally. Limitations of this study and the foregoing tentative interpretations of the results include (1) the small sample size and the ‘youthful’ age of the subjects, and (2) the confounding of assessment of stability (or progression) with measurement inadequacies at the respective time periods. The subjects evaluated at both assessment periods constitute 82.3% of the total sample A individuals ascertained through parental clinical status in 1971. These subjects ranged in age between 15 and 19 at time 1 and ranged in age between 20 and 29 at time 2. Baron has reported age of onset distributions for schizotypal features as estimated from onsets of individuals in 90 extended families. Baron et al. (1983a) reported a mean ‘age of onset’ before age 20 (18-19 years of age) with a standard deviation of around 4 years for probable (two or more schizotypal features) or definite (four or more schizotypal features) schizotypal personality disorder.
83 TABLE
6
Logistic
regression:
Independent
results for
time 2 outcome
on time I spectrum
variable
expression
controlling,for
diagnostic
parental
and/or
anxiety
The probability
The odds of
Odds
expression
ratio..
of time 2
qf time 2
exp.
standard
total spectrum
total spectrum
icoeff. I
error
given:
given: presence
of
time I
Offspring ,from parental
depression
qf expression
presence
Schizophrenic disorder All subjects (n = 47) Subjects without axis I time I (n =43)
comorbid
Coeff.
of
time I
spectrum
spectrum
,features
,features
0.700
2.33
38.8
3.5*
0.500
I .oo
15.8
2.5*
0.666
2.00
61.8
3.0*
0.500
1.00
19.0
2.1*
0.333
0.50
11.8
2.4*
0.200
0.25
5.8
groups.
group
Affective disorder parental group All subjects (n = 39) Subjects without axis I time 1 (n = 37) Normal control parental group All subjects (n = 82) Subjects without axis I time 1 (n=81)
1.4**
*p
Research,
10 1992 Elsevier
SCHIZO
6 (1992)
75
75-85
Science Publishers
B.V. All rights reserved
0920-9964/92/$05.00
00 I86
The assessment of schizotypal features over two points in time Elizabeth Department
of Medico1
Squires-Wheeler,
Genetics.
Columbia
Andrew
University,
College
722 West 168th Street, (Received
20 August
E. Skodol of Physicians
and Surgeons,
New York, NY
1990, revised received
5 February
and L. Erlenmeyer-Kimling and New York State Psychiatric
Institute,
10032, U.S.A. 1991, accepted
11 February
1991)
The expression of schizotypal personality traits was assessed in mid-adolescence and again in young adulthood for three groups of offspring defined by the psychiatric diagnosis of their parents. Parental diagnoses included schizophrenic disorder (47 offspring), affective disorder (39 offspring), and ‘no psychiatric disorder’, or normal controls (82 offspring). Initially, schizotypal traits were assessed from video-taped semi-structured psychiatric interviews, subsequently rated by trained psychiatrists blind to the parental psychiatric status of the subjects, and/or direct clinical interviews (Schedule for Affective Disorders-Lifetime Version @ADS-L)). The second assessment was conducted by trained social workers and psychologists by means of a semi-structured interview specifically for DSM-III-R personality disorders (Personality Disorder Examination) and sections of the SADS-L where indicated. These interviewers were blind to the parental status and to previous psychiatric assessments of the offspring. The rates of stability of features or the rates of progression to axis I psychotic disorders (Schizophrenia, Schizoaffective Disorder, and Unspecified Functional Psychosis) were evaluated. Concordance of assessments over time is reported as a function of threshold for expression of traits at initial evaluation, i.e., two or more, three or more, or four or more features present. Concordance increases as the threshold for expression increases, as expected. The effect of comorbid clinical status, e.g., the coexistence of schizotypal traits and anxiety and/or depressive features on the concordance pattern, is also examined by parental diagnostic group status. The offspring of affective disorder parents exhibited higher rates of anxiety and/ or depressive features at both points in time, exhibited higher concordance for anxiety and/or depressive features, and exhibited higher rates of ‘transformation’ of initial schizotypal features to anxiety and/or depressive features at the second assessment. Key words:
Schizotypal
personality
traits;
Longitudinal
assessment
INTRODUCTION
The identification of a nonpsychotic diagnostic category, exhibiting specificity and sensitivity to biological relatives with schizophrenia would enhance the power of studies of pathophysiological mechanisms and etiological components of schizophrenic disorders. For example, linkage analysis Correspondence to: E. Squires-Wheeler, Department of Medical Genetics, Columbia University, College of Physicians and Surgeons, and New York State Psychiatric Institute, 722 West 168th Street, New York, NY 10032, U.S.A.
of schizophrenic disorder pedigrees will have increased power to detect linkage if a schizophrenic spectrum disorder is included as part of an ‘affected phenotype’ model. If the clinical boundaries of the spectrum are uncertain, however, the possible errors introduced reduce the expected increase in power and indeed may obscure a real underlying linkage result. Historically, procedures used to identify specific and sensitive spectrum categories have varied in rigor and methodologic safeguards, although all approaches have used family aggregation of clinical features as empirical observations to guide identification of spectrum candidates. The
76
criteria for schizotypal personality disorder (STPD), one spectrum diagnostic category appearing in DSM-III (1980) were derived from an examination of the clinical features that characterized borderline schizophrenic subjects who clustered in the first- and second-degree relatives of index adoptees in the Danish Adoption Study (Kety et al., 1968, 1975; Kety, 1983, 1985, 1987, 1988). Validity criteria for such proposed schizophrenic spectrum categories have been suggested by Shields et al. (1975) and recently restated by Kendler (1985). Criteria for spectrum categories should exhibit sensitivity and specificity in discriminating relatives of schizophrenic index cases. The distribution of DSM-III-R spectrum features STPD, has been previously examined in relatives of schizophrenic probands compared to normal controls by several investigators including Lowing et al. (1983), Baron et al. (1985), Frangos et al. (1985) and Kendler (1984, 1988a,b). In these studies the relative risks ranged from 2.4 to 9.1, thus supporting the validity criteria. It is desirable, however, to contrast the rates of spectrum features in the relatives of schizophrenic probands with the rates of such features in the relatives of probands with other psychiatric disorders, in addition to relatives of normal controls, in order to establish the specificity of association of the features to schizophrenic disorder. When probands with affective disorder have been used as psychiatric contrast groups, the rates of spectrum features in their relatives have not consistently differed from the rates of these features in the relatives of schizophrenic probands (Endicott et al., 1985; Ingraham, 1987, 1988; Squires-Wheeler et al., 1988). For example, in our own recent report (Squires-Wheeler et al., 1989) although the rates of schizotypal features were elevated in the relatives of probands with psychiatric disorder in contrast to the relatives of normal controls, the rates did not differ significantly between relatives of schizophrenic disorder probands and the relatives of affective disorder probands, thus failing to support the expected diagnostic specificity. To further explore the nature of the apparent lack of specificity, we have employed a model proposed by Rice and associates ( 1987) concerning the relationship between true trait status and stability of trait expression. Rice’s approach defines
a ‘true case’ as one exhibiting stability of expression over time. Here a true case is assumed to be more likely to share a genetic continuity with schizophrenic disorder, so that relatives of schizophrenic disorder probands might be expected to express schizotypal features that are stable and/or progressive. In contrast, the expression of schizotypal features at a given point in time in relatives of affective disorder probands and normal controls may represent state determined factors and thus be less likely to be persistent and observed at a later time. That is, expression of schizotypal features at a single point in time may represent transient states, rather than personality traits. Rice et al.‘s (1987) stability model was developed for affective disorder. He demonstrated that stability of expression was maximized for cases with severe clinical features, initially. For example, the number of depressive symptoms expressed at intake was positively associated with stability, as might be expected. DSM-III-R criteria for prodromal schizophrenia and schizotypal personality disorder reflect similar perceptual. cognitive, and behavioral disabilities. Features of schizotypal personality disorder in DSM-III were derived from the Danish Adoption Study series in which biological relatives were assessed in mid-adult life. under the assumption that their clinical status represented enduring traits. Given the relative ‘youthfulness’ of our offspring sample, and the considerable overlap of criteria for schizotypal and prodromal clinical pictures, it may be expected that some individuals expressing features when first evaluated may progress at follow-up to an axis I psychotic disorder such as schizophrenia, schizoaffective disorder, or unspecified functional psychosis. To examine alternative natural history patterns, three groups of offspring of parents (schizophrenic disorder parents, affective disorder parents, and normal control parents) have been assessed for schizotypal personality features and disorder at two points in time. These subjects are drawn from a longitudinal family study and constitute a subset of Sample A of the New York High-Risk Project (Erlenmeyer-Kimling and Cornblatt, 1987). In summary, the purpose of the study we are reporting was to test the dual hypotheses that ‘true cases’ of schizotypal personality disorder, i.e., those that show a specific familial relationship to
schizophrenia, are stable over time and that stable cases are those that on initial, cross-sectional assessment are more severe. The rates of concordance of two assessments are reported as a function of initial diagnostic threshold by parental diagnostic groups. Additionally the expression of anxiety and/or depressive features was noted at time 1 and time 2 by parental diagnostic groups. The coexistence of such affective features (with spectrum traits) was evaluated by diagnostic group.
METHODS Subjects
The subjects of this report are members of the initial sample (sample A) of the two-sample longitudinal New York High-Risk Project, which is described fully elsewhere (Erlenmeyer-Kimling and Cornblatt, 1987). This report is based on data obtained on subjects who were available for assessment for two measurement periods. Subjects included were evaluated (1) in mid-adolescence ~ average age of 16 years (time l), and (2) in young adulthood ~ average age of 25 years (time 2). The subjects are offspring of schizophrenic, affective disorder, and normal control parents. Ascertainment of parents was through the consecutive admission records of six New York State hospitals. Those patients who were married, white, Englishspeaking, passed exclusion diagnostic criteria, and had at least one offspring between the ages of seven and 12, were eligible for the study. Subsequent diagnostic assessments of parents were made using the Schedule for Affective Disorders and Schizophrenia-Lifetime Version (SADS-L) and Research Diagnostic Criteria (RDC) (Spitzer et al., 1978). Normal control families were secured through school systems, chosen to reflect the community characteristics from which the patient parents came. The offspring were free of psychiatric impairment at intake into the study (1971-1972). Table 1 displays the demographic characteristics of those sample A subjects present for the two measurement periods. Assessment
procedures:
The adolescent controls were
time I measurement
offspring of patient parents and interviewed, approximately eight
years after the initiation of the study, by a psychiatrist blind to parental diagnostic group using a semi-structured interview protocol, the Mental Health Assessment Form (MHAF) (Kestenbaum and Bird, 1977). The interviews were video-taped and edited to remove any reference to illness in the parents and were rated subsequently by at least two psychiatrists or psychiatric residents (also blind to the parental group status of the interviewee), who were trained in the assessment of MHAF indicators. These videotaped assessments preceded DSM-III by 2 years, and hence DSMIII schizotypal personality disorder was not an official diagnostic category at the time of the assessment. However, information relevant to the (subsequent) DSM-III criteria was elicited at the videotaped psychiatric interview (see SquiresWheeler et al. (1988) for a complete description of assessment and scoring). A subset of sample A study subjects was directly assessed with the SADS-L, and rated according to the RDC. The interviewers were psychiatric social workers trained in the use of the SADS-L by members of the Biometrics Research Department at New York State Psychiatric Institute. Axis I disorders were the focus of the interview and the presence of schizotypal features was noted at this assessment only if the interviewee exhibited another diagnosis. Notwithstanding the incomplete nature of the schizotypal assessment at this time, the data on schizotypal features have been combined with the systematic MHAF assessment, such that an individual is noted as expressing a given number of features at time 1 if he/she expresses features on the MHAF or on the SADS-L or both. Assessment
procedures:
time 2 measurement
Diagnostic assessments at time 2 were made employing a blind, direct diagnostic evaluation using the Personality Disorder Examination (PDE) (Loranger et al., 1987). The PDE is a standardized, semi-structured clinical interview for eliciting information relevant to the Diagnostic and Statistical Manual of Mental Disorders, third edition-revised (DSM-III-R) axis II personality disorders. The PDE was modified by insertion of selected SADSL components. If subjects responded positively to PDE screening items for axis I disorders (e.g., depressive or anxiety disorders), the appropriate SADS-L section was completed. The modified
78 TABLE
I
Demographic characteristics oJsample assessed at time I and time 2: sex, social class, and age by group Characteristic
Groups: o$ypring of‘ parents with Schizophrenic Disorder n=45
Affective Disorder n=39
n
(Xl
n
IX)
n
i”/oi
Sex Males Females
22 23
(48.9) (51.1)
14 25
(35.9) (64.1)
44 35
(55.7) (44.3)
Social class Hollingshead & Redlich Index I Highest 2 3 4 5 Lowest
5 I 6 30 3
(11.1)
0 0 II 24 4
(0.0) (0.0) (28.2) (61.5) (10.3)
16 17 19 25 2
(20.3) (21.5) (24.1) (31.6)
“No disorder
equals
normal
(2.2) (13.3) (66.7) (6.7)
No Disorder” n=79
(2.5)
control
PDE was administered to subjects by clinical psychologists and social workers after a period of orientation and training in the use of the schedule by means of group workshops and individual supervision. The interviewers were blind to parental diagnostic status and to results of previous clinical assessments of the offspring. See SquiresWheeler et al. (1989) for a full description of training and clinical consensus procedures. Table 2 shows the nine features of DSM-III-R criteria of Schizotypal Personality Disorder.
RESULTS
Table 3 shows the rates of expression of schizotypal features by varying thresholds for presence of features, e.g., two or more, three or more, or four or more schizotypal features present. The inclusion of two or more, three or more or four or more features reflects the range of diagnostic thresholds employed in the literature (Baron et al., 1980, 1983a,b, 1985; Risch et al., 1984). The denominators of the proportions reflect the number of offspring who were assessed at both time 1 and time 2. The MHAF clinical assessment was conducted by psychiatrists and psychiatric residents
TABLE
2
DSM-III-R ,features,from schizotypal personality disorder 1. Ideas of reference (excluding delusions of reference) 2. Excessive social anxiety, e.g., extreme discomfort in social situations involving unfamiliar people 3. Odd beliefs or magical thinking, influencing behaviour and inconsistent with subcultural norms, e.g.. superstitiousness, belief in clairvoyance, telepathy, or ‘sixth sense’, ‘others can feel my feelings’ (in children and adolescents, bizarre fantasies or preoccupations) experiences, e.g., illusions. sensing the 4. Unusual perceptual presence of a force or person not actually present (e.g., ‘I felt as if my dead mother were in the room with me’) or appearance, e.g.. unkempt, 5. Odd or eccentric behaviour unusual mannerisms, talks to self (or only one) other than 6. No close friends or confidants first-degree relatives Odd speech (without loosening of associations or incoherence), e.g., speech that is impoverished, digressive, vague, or inappropriately abstract Inappropriate or constricted affect, e.g., silly. aloof, rarely reciprocates gestures or facial expressions, such as smiles or nods Suspiciousness or paranoid ideation
who were blind to parental clinical status and to subjects’ prior clinical and treatment status. Therefore schizotypal features were rated present if exhibited by the subjects irrespective of the subjects’ axis I status. Given this lack of an hierarchi-
79 TABLE
3
Parental
group proportions
Spectrum
exhibiting
indicated
number
Group proportion
offeatures
with disorder
or outcome al time I or time 2 present
criteria
Schziphrenic
AfSective
Normal
parental
disorder
control
group
paren tat
parenlal
group n=39”
group
difference’ A vs. B
A vs. C
B vs. C
n=37b
n=82” n=81b
(Al
iBi
ici
0.28 (0.21)
0.36 (0.32)
0.18 (0.17)
0.557
0.307
0.058
0.26 (0.19)
0.26 (0.22)
0.09 (0.07)
1.000
0.018
0.024
0.21 (0.14)
0.15 (0.11)
0.07 (0.06)
0.674
0.042
0.288
0.21 (0.14)
0.13 (0.08)
0.06 (0.05)
0.448
0.020
0.360
n=47” n=43b
Two or more schizotypal features [time I) Three or more schizotypal features (time 1) Four or more schizotypal features (time 1) Time 2 total spectrum outcorned
Sign$cance of proportion
disorder
aSubjects present at both time 1 and time %ubjects present at both time 1 and time ‘Significance levels associated with Fleiss’s dTotal spectrum includes axis I disorders and/or four or more schizotypal features.
2 assessment periods. 2 assessment periods without onset of Axis I disorder between time 1 and time 2. (1981) z statistic for proportion difference. (Schizophrenic Disorder, Schizoaffective Disorder and Unspecified Functional Psychosis)
cal assessment procedure, schizotypal features noted in the interview could represent prodromal or residual schizophrenic signs. There were four subjects from the schizophrenic disorder parental group (SDPG) who were diagnosed as schizophrenic, schizoaffective, or unspecified functional psychosis at the SADS-L interview conducted after the MHAF clinical assessment (considered here as part of the time 1 assessment) and for whom the schizotypal assessment most likely represented prodromal or residual schizophrenia. There were two such subjects from the affective disorder parental group (ADPG) and one such subject from the normal control parental group (NCPG). The proportions in parenthesis on Table 3 represent the rates for offspring from the respective parental groups, excluding these axis I subjects. The pair-wise group contrasts reaching less than 0.05 levels of significance were (1) time 1 three or more schizotypal features in offspring from the SDPG versus offspring from the NCPG; (2) time 1 three or more schizotypal features for offspring from the ADPG versus offspring from the NCPG; and (3) time 2 total spectrum outcome for offspring from the SDPG versus offspring from the NCPG. If the significance level is corrected for the number
of contrasts employed, i.e., corrected level of < 0.004 for 12 contrasts, there are no significant differences observed, although the pattern of results is generally consistent with expectations. Table 4 shows the association between assessment at time 1 and assessment at time 2 (total spectrum outcome) in terms of positive predictive values, negative predictive values, sensitivity, and specificity, for parental diagnostic groups. In this measurement context, time 1 assessment may be considered a ‘predictor’, while time 2 assessment represents the ‘outcome’. The positive predictive value represents the probability of expressing outcome at time 2 given expression at time 1. The higher this probability, the more predictive, or stable time 1 assessment is. The negative predictive value represents the probability of a normal assessment at time 2 (no features or disorder exhibited) given the absence of (a given number of features) expressed at time 1. Sensitivity represents the probability of expressing threshold features at time 1 given expression of total spectrum at time 2. (The complement of sensitivity reflects the proportion of subjects expressing total spectrum at time 2 who experienced a recent onset (i.e., onset of total spectrum some time after time 1, but before time
80 TABLE
4
Associations between rime I and time 2 assessments
Two or more schizoiypalfeatures
Three or more schizotypalfeatures (time I) Schizophrenic disorder parental group (n = 47) Affective disorder parental group (n = 39) Normal control parental group (n = 82) Four or more schizotypal features (time I) Schizophrenic disorder parental group (n=47) Affective disorder parental group (n = 39) Normal control parental group (n = 82) Predictive
NPV-
SEN”
SPE”
0.538
0.912
0.700
0.838
0.286
0.960
0.800
0.700
0.200
0.970
0.600
0.844
0.583
0.914
0.700
0.865
0.400
0.966
0.800
0.824
0.286
0.960
0.400
0.935
0.700
0.919
0.700
0.919
0.667
0.970
0.800
0.941
0.333
0.961
0.400
0.948
(time I)
Schizophrenic disorder parental group (n = 47) Affective disorder parental group (n = 39) Normal control parental group (n = 82)
“PPV = Positive
PPV
Value; NPV = Negative
Predictive
Value: SEN = Sensitivity;
2.) Specificity refers to the probability of a normal assessment at time 1 given a normal assessment at time 2. For the offspring from the two psychiatric parental groups, the positive predictive values, the negative predictive values, and the specificities increase as the diagnostic thresholds for affected status increase. For the offspring from the NCPG, the positive predictive values and the specificities increase as the diagnostic thresholds for affected status increase, although the positive predictive values remain low even for the highest threshold. As noted above, the positive predictive values suggest the percent of subjects who exhibit a stable, persistent, or progressive disorder. Employing Rice’s assumption that a stable case is more likely to reflect a true case, we note that increasing numbers of features at time 1 are associated with increasing probability of identifying a stable or ‘true case’. However, our expectation that such stability would characterize only the offspring of the SDPG was not observed. Offspring of the ADPG exhibited rates of stability, as quantified
SPE = Specificity
by positive predictive values, similar to those of offspring of the SDPG. In the light of the failure to demonstrate specificity of stability for probable schizotypal personality disorder, we asked the following exploratory question: Is the expression of schizotypal features over time different for the offspring groups in terms of comorbidity patterns? For example, do the offspring of the ADPG express schizotypal features in conjunction with affective features, at time 1, or time 2? A subset of offspring from the SDPG (94%) the ADPG (92%), and the NCPG (96%) was assessed for anxiety disorders (anxiety disorders are defined by Research Diagnostic Criteria definite thresholds for generalized anxiety. phobic disorder, or panic disorder) and affective disorders (depressive disorders are defined by Research Diagnostic Criteria definite thresholds for major depression or Intermittent Depression) at the SADSL assessment at time 1. The rates of depression and/or anxiety at time 1 for the SDPG, the ADPG and the NCPG offspring are 45%, 68%, and 37%,
81
respectively. At time 2, the rates of depression and/or anxiety for the SDPG, the ADPG, and the NCPG offspring are 13.6%, 34%, and 17% respectively. The rates of concordance of the two assessments for the groups, SDPG, ADPG and NCPG are 3.6%, 36%, and 8% respectively. That is, the persistence of depression and/or anxiety for subjects from all groups is low, with the ADPG showing the highest stability of expression. Over one-third of the subjects from the ADPG expressed stable depression and/or anxiety disorder (or frequent recurrences identified at both periods). Table 5 shows the proportion of total spectrum outcomes as a function of comorbid clinical features expressed at time 1. Among the SDPG offspring, 18 (41 “h) expressed neither anxiety, depression or schizotypal features at time 1 and only one (5.5%) of these subjects expressed total spectrum outcome at time 2. Of 16 SDPG subjects (36%) exhibiting depression or anxiety but not four or more schizotypal features at time 1, one (6.25%) subsequently showed total spectrum outcome at time 2. Five (83%) of the six SDPG subjects (14%) who expressed four or more schizotypal features but not depression or anxiety at time 1 demonstrated total spectrum outcome at time 2, and two (50%) of the four SDPG subjects (9%) exhibiting four or more schizotypal features
TABLE
and depression or anxiety at time 1 had total spectrum outcome at time 2. Of 36 offspring from the ADPG entering these assessments, 10 (28%) were free of depression, anxiety and four or more schizotypal features at time 1, and none of the subjects expressed total spectrum outcome at time 2. 20 (55.6%) subjects showed only depression or anxiety at time 1, and only one (5%) such subject exhibited total spectrum outcome at time 2. Both of the two ADPG subjects (5.5%) subjects who expressed four or more schizotypal features only at time 1 showed total spectrum outcome at time 2. Of the four ADPG subjects (11.1%) exhibiting both four or more schizotypal features and depression or anxiety at time 1, two (50%) expressed total spectrum outcome at time 2. Among the offspring from the NCPG entering these assessments, 45 (57%) showed none of the conditions considered at time 1; one (2.2%) of these subjects had total spectrum outcome at time 2. Of 28 NCPG subjects (35.5%) showing depression or anxiety but not four or more schizotypal features at time 1, two (7.1%) expressed total spectrum outcome at time 2. Two (50%) of the four NCPG subjects (5%) showing schizotypal features but not depression or anxiety at time 1 expressed total spectrum outcome at time 2,
5
Proportion of subjects(for whom affective features were evaluated) expressing time 2 total spectrum and parental diagnostic group Time I clinical status
Subject expressed. no spectrum or affective features only affective/anxiety features only spectrum features both spectrum and affective features Total time 2 spectrum outcome
Oflspring
outcome hJ time
I clinical status
From
Schizophrenic disorder parental
Aflective disorder purentul
Norm01 control parental
group (SDPC) n=44”
group (ADPG) n=36”
group (NCPC) n=19a
l/18 (0.055) I,!16 (0.0625) S/6 (0.833) 2/4 (0.500) 9b/44” (0.20)
OjlO (0.000) l/20 (0.05) 2!‘2 (1.0) 2/4 (0.50) 5bi36” (0.14)
1.45 (0.022) 2128 (0.072) 2.4 (0.50) 0,;2 (0.00) 5”/79” (0.063)
Subjects assessed for affective features at time 1. %ubjects assessed for affective features at time 1 and expressing total spectrum outcome at time 2. Only significant difference in Table 5 (p < 0.05) is contrast between SDPG and NCPG for total time 2 spectrum
outcome.
82
whereas neither of the two NCPG subjects (2.5%) showing four or more schizotypal features plus depression or anxiety at time 1 developed total spectrum outcome by time 2. A logistic regression was employed to assess the probability of expressing total spectrum outcome at time 2 as a function of the presence of schizotypal features at time 1 (four or more features present) controlling for the comorbid expression of anxiety and/or depression at time 1. The exponentiated coefficient for the time 1 schizotypal feature independent variable (see Table 6) can be interpreted as an odds ratio. This ratio is the odds of expressing time 2 total spectrum outcome given expression of time 1 schizotypal features over the odds of expressing time 2 total spectrum outcome given absence of schizotypal features at time 1. For example, for the offspring from the SDPG, the predicted probability of expressing total spectrum outcome at time 2 given expression of features at time 1 of 70%. The odds of expressing total spectrum outcome at time 2 given expression of features at time 1 is 2.33-I. The odds ratio adjusted for time 1 expression of comorbid anxiety and/or depression is 38.8, seen in Table 6. The coefficient indexing the prediction of time 1 schizotypal features controlling for comorbid clinical expression is statistically significant for offspring from all parental diagnostic groups except the offspring from the NCPG without axis I at time 1. While the logistic regression addressed the question of stability and/or progression as a function of clinical status at time 1, the question of comorbid clinical profile at time 2 was not examined. It may be hypothesized that the expression of spectrum features at time 1 among offspring from the ADPG may predict (i) an affective disorder at time 2 or (ii) a comorbid clinical profile reflecting both spectrum and affective features. Table 7 shows the rates at time 2 of(i) no follow-up, (ii) stability or progression of schizotypal features, (iii) comorbid expression of schizotypal features and depression and/or anxiety, (iv) depression and/or anxiety, and (v) no disorder.
CONCLUSIONS, IMPLICATIONS
LIMITATIONS
AND
A large percentage of subjects from groups expressed schizotypal features
all parental or disorder
likely to be state phenomena as inferred from instability of (failure to express) features over time. 28% of SDPG offspring, 34% of ADPG offspring, and 66% of NCPG offspring exhibiting at least four or more schizotypal features at time 1 were free of total spectrum disorder at time 2. The longitudinal picture of feature expression is clearly important and lack of a follow-up assessment obscures the distinction between state and trait expression. An informant who is able to provide such a longitudinal view may be necessary when direct follow-up assessment is not possible. We may tentatively conclude that rates of stability and/or progression of schizotypal features do not provide a means of differentiating the expression of features between the SDPG and the ADPG offspring as proposed. It is seen that ADPG offspring are the only group with ‘transformation’ of spectrum features (at time 1) to depression and/ or anxiety at time 2. 17% of ADPG offspring exhibited such a pattern. A further 17% of these subjects also expressed a comorbid profile consisting of spectrum and depression and/or anxiety at time 2. These proportions are not statistically significantly different between parental groups owing to the very small number of subjects expressing features and disorder at time 1. However, the pattern is consistent with the expectation that a subset of ADPG offspring express a potentially differentiable clinical picture longitudinally. Limitations of this study and the foregoing tentative interpretations of the results include (1) the small sample size and the ‘youthful’ age of the subjects, and (2) the confounding of assessment of stability (or progression) with measurement inadequacies at the respective time periods. The subjects evaluated at both assessment periods constitute 82.3% of the total sample A individuals ascertained through parental clinical status in 1971. These subjects ranged in age between 15 and 19 at time 1 and ranged in age between 20 and 29 at time 2. Baron has reported age of onset distributions for schizotypal features as estimated from onsets of individuals in 90 extended families. Baron et al. (1983a) reported a mean ‘age of onset’ before age 20 (18-19 years of age) with a standard deviation of around 4 years for probable (two or more schizotypal features) or definite (four or more schizotypal features) schizotypal personality disorder.
83 TABLE
6
Logistic
regression:
Independent
results for
time 2 outcome
on time I spectrum
variable
expression
controlling,for
diagnostic
parental
and/or
anxiety
The probability
The odds of
Odds
expression
ratio..
of time 2
qf time 2
exp.
standard
total spectrum
total spectrum
icoeff. I
error
given:
given: presence
of
time I
Offspring ,from parental
depression
qf expression
presence
Schizophrenic disorder All subjects (n = 47) Subjects without axis I time I (n =43)
comorbid
Coeff.
of
time I
spectrum
spectrum
,features
,features
0.700
2.33
38.8
3.5*
0.500
I .oo
15.8
2.5*
0.666
2.00
61.8
3.0*
0.500
1.00
19.0
2.1*
0.333
0.50
11.8
2.4*
0.200
0.25
5.8
groups.
group
Affective disorder parental group All subjects (n = 39) Subjects without axis I time 1 (n = 37) Normal control parental group All subjects (n = 82) Subjects without axis I time 1 (n=81)
1.4**
*p
