The Longitudinal Course of Thought Disorder in Geriatric Patients With Chronic Schizophrenia

As part of a longitudinal study of chronic schizophrenia described elsewhere (5, 6), we selected schizophrenia patients who had follow-up assessments on the measures described.

We identified 799 schizophrenia patients at baseline and 220 with follow-up assessments. Patients were lost to follow-up because of uncooperativeness of the patient (N=87), death of the patient (N=290), or discharge to a noncooperative treatment site (N=202). The mean follow-up interval was 842 days (SD=489, range=362–2,264). Among the patients with follow-up data, 127 (57.7%) were women; the group had a mean age of 74.9 (SD=9.4, range=49–97) at baseline, a mean of 10.2 years (SD=2.7) of education, and a mean age at first hospitalization of 27.6 (SD=10.9). One hundred forty-three were residents of a psychiatric hospital for chronic illness, and 57 were residents of a nursing home who had been discharged from a psychiatric hospital for chronic illness. Exactly 77.7% of the patients were Caucasian, 16.4% were African American, and 5.9% were Hispanic. All were English speaking, without sensory deficits, and cooperative with the assessment procedures.

In this study, we report on the results of the patients’ performance on the Mini-Mental State Examination (MMSE) (7) and the interviewers’ ratings of thought disorder on the Thought, Language, and Communication Scale (8, 9). The Thought, Language, and Communication Scale ratings were based on a symptom interview and unscripted conversation with the patients at the time of the assessments. The Thought, Language, and Communication Scale provides definitions and criteria for 18 subtypes of abnormal language. For the present report, we were interested in two previously confirmed (1) subtypes of thought disorder, disconnection and verbal underproductivity. The disconnection subscale is the sum of the following Thought, Language, and Communication Scale items: poverty of content of speech, derailment, tangentiality, loss of goal, circumstantiality, and incoherence (1). Verbal productivity was examined with the poverty of speech item. The aspects of speech performance reported here were found in a previous study to have suitable interrater reliability (all intraclass correlation coefficients >0.56, range=0.56–0.91) in older patients with schizophrenia with our raters, who were blind to the present hypotheses (5).

We used repeated-measures analyses of variance (ANOVAS) to examine changes in the language and cognitive variables over time. To examine changes in disconnection, we selected only patients who were verbally productive (severity of poverty of speech <3) and entered changes in poverty of speech as a covariate to determine if changes in disconnection were independent of changes in verbal underproductivity. We examined the association between verbal underproductivity, disconnected speech, MMSE scores, and gender and compared changes in verbal productivity, disconnection, and MMSE scores in younger and older patients. For this final analysis, we used a median split at age 74 as the between-subject factor. These two groups were equivalent in time between assessments (t=–1.23, df=180, p=0.22).

A multivariate ANOVA to examine baseline differences in cognition, poverty of speech, disconnection, age, and age at first hospitalization between those who were reevaluated and those who were lost to follow-up was not significant (Pillai’s trace F=1.75, df=5, 550, p=0.12).

The patients displayed worsening in verbal underproductivity over time, with mean poverty of speech ratings of 1.9 (SD=1.4) at baseline and 2.3 (SD=1.4) at follow-up (Pillai’s trace F=13.3, df=1, 219, p<0.001, Wilks’s lambda=0.94). Only 5% of the patients displayed improvements in poverty of speech; the majority (74%) who improved did so by 1 point. Verbally productive patients demonstrated improved disconnection from baseline (mean=4.8, SD=4.7) to follow-up (mean=3.7, SD=4.4) (Pillai’s trace F=4.9, df=1, 124, p<0.03, Wilks’s lambda=0.96). However, when entering change in poverty of speech as a covariate, this finding was no longer significant (Pillai’s trace F=0.64, df=1, 124, p=0.43, Wilks’s lambda=0.99). MMSE scores were stable for the group as a whole from baseline (mean=11.4, SD=9.8) to follow-up (mean=12.1, SD=10.2) (Pillai’s trace F=1.3, df=1, 188, p=0.26, Wilks’s lambda=0.99).

Poverty of speech was associated with MMSE scores at baseline (r=–0.60, p<0.001) and follow-up (r=–0.58, p<0.001) and severity scores on disconnection at baseline (r=–0.45, p<0.001) and follow-up (r=–0.35, p<0.001). Worsening in poverty of speech was correlated with decreases in MMSE scores (r=–0.30, p<0.03) and greater age at baseline (r=–0.15, p<0.03) but not years of education, medication status (none, typical or atypical antipsychotic), or age at first hospitalization. Disconnection was not significantly associated with MMSE scores at baseline (r=0.03, p=0.71) or follow-up (r=0.04, p=0.60). Gender was not significantly associated with disconnection at baseline (r_{point biserial}=0.03, p=0.40) or endpoint (r_{point biserial}=0.05, p=0.51).

In the repeated-measures ANOVA examining changes in poverty of speech by age, the main effect of change in poverty of speech was significant (Pillai’s trace F=22.1, df=1, 180, p<0.001, Wilks’s lambda=0.89). The interaction between poverty of speech and age was significant (Pillai’s trace F=4.04, df=1, 180, p=0.04, Wilks’s lambda=0.97). As demonstrated in Figure 1, this interaction effect suggests that the older group had a greater worsening in verbal underproductivity over time. The repeated-measures ANOVA examining changes in disconnection by age was not significant (Pillai’s trace F=2.9, df=1, 180, p=0.09, Wilks’s lambda=0.98). The repeated-measures ANOVA examining changes in MMSE by age was also not significant (Pillai’s trace F=1.4, df=1, 180, p=0.25, Wilks’s lambda=0.99).

Older schizophrenia patients with a chronic course of illness show decreases in verbal productivity over time, with this change predicted by worsening in MMSE scores. Improvement in verbal productivity over time was rare and minimal. Scores on the disconnection scale indicated an improvement over time; however, this change is an artifact of decreases in output. It may be more accurate to refer to this “improvement” as a shift in the behavioral topography of communication disorders, from disconnected to unproductive speech, as patients age. Viewed this way, thought disorder does not appear to “burn out” in chronic schizophrenia.

Age-related differences in thought disorder were noted in the cross-sectional study (5), and the present longitudinal study confirms a worsening in poverty of speech over time in the truly geriatric patients. The reduction in verbal output over time complements previous work that has suggested a decline in cognitive and functional skills during this age range (10), particularly in frontal “executive” functions (11), as poverty of speech is associated with dorsolateral prefrontal cortex dysfunction (12). Reduced verbal output also suggests that findings of improvement in positive and disorganized symptoms with aging in schizophrenia should be carefully considered in the context of possible reductions in verbal productivity. Decreased verbal output may affect one’s ability to report hallucinations and delusions, as well as reducing the tendency to produce aberrant language. Future studies may address some of the limitations of this study, such as the influence of physical illnesses on thought disorder in late life.

Received March 1, 2004; revision received May 31, 2004; accepted June 10, 2004. From the Department of Psychiatry, Mount Sinai School of Medicine. Address correspondence and reprint requests to Dr. Bowie, Department of Psychiatry, Mount Sinai School of Medicine, 1425 Madison Ave, 4th Floor, Box 1230, New York, NY 10029; [email protected] (e-mail). Funded by the Mount Sinai Mental Health Clinical Research Center on Late-Life Schizophrenia (K.L. Davis, principal investigator), the Silvio Conte Neuroscience Center (NIMH grant MH-36692; K.L. Davis, principal investigator), NIMH grant 63116 (Dr. Harvey, principal investigator), and the VA Veterans Integrated Service Network 3 Mental Illness Research, Education, and Clinical Center.