Individual Differences in Cognitive Decline in Schizophrenia

There is now a general consensus that schizophrenia does not have a progressive degenerative course (1). A possible exception, however, is a subtype described by Keefe et al. (2), accounting for about 15% of the schizophrenic population, that may be degenerative in course. This proposed subtype has become known as “very poor outcome” or “Kraepelinian” schizophrenia. Thus, despite the general finding that aging in schizophrenia is normal, the subgroup of patients with Kraepelinian schizophrenia may exhibit abnormal age correlations of the type seen in progressive dementia or pervasive developmental disorders.

Correlational studies are useful in plotting the trajectory of ability over age. Several classes of trajectory may be noted, including 1) gradually progressive diminution of ability levels as one ages normally, 2) attenuation of a correlation between age and ability level seen in individuals with acute illness such as stroke (3), 3) a normal pattern until time of onset of illness and then an increased correlation between ability and age, as in progressive dementia, or 4) a zero-order correlation reflecting cognitive abnormality from early life, as in pervasive developmental disorders. If schizophrenia is a heterogeneous condition, different correlational patterns may be found in different subtypes.

Using cluster analysis, we identified a subgroup of patients who were not elderly but who performed on cognitive tests at levels comparable to what is found in patients with progressive dementia; we also identified another subgroup of patients who had evidence of only minimal cognitive impairment (4). We hypothesized that the group with relatively normal test performance would generate normal-level correlations between a summary index of neuropsychological test impairment and age but that the group with severely abnormal cognitive function would show an abnormal pattern of correlations. We hypothesized that the latter group would have high correlations suggestive of progressive dementia or very low correlations in an order of magnitude comparable to that found in patients with histories of acute brain damage or disease or pervasive developmental disorder. The purpose of the present study was to determine these correlations.

METHOD

Data were abstracted from the records of 228 male patients with schizophrenia reported on in a previous study (4). Diagnoses were made retrospectively for 136 patients; the remaining patients were recruited and diagnosed in a prospective study that used the Structured Clinical Interview for DSM-III-R administered by a trained interviewer with established high reliability, followed by a case conference with consultation by a board-certified psychiatrist. We found no demographic or cluster analytic differences between the two groups of subjects, so they were combined. For the present study, we used 77 of the 228 subjects in two clusters characterized by cognitive function. One cluster of 51 patients exhibited near-normal functioning; their mean age was 36.79 years (SD=9.65), and their mean level of education was 13.16 years (SD=2.2). The second cluster consisted of 26 patients with severely impaired function; their mean age was 43.77 years (SD=9.41), and their mean level of education was 9.85 years (SD=3.63). Two additional reference groups were drawn from an extensive database of patients who had received the Halstead-Reitan battery and detailed neurological and psychiatric diagnostic evaluations. The first reference group consisted of 299 patients with no evidence of structural brain damage. These patients had a variety of psychiatric and general medical disorders, but schizophrenia was carefully ruled out. Their mean age was 42.24 (SD=12.10), and their mean education level was 11.57 years (SD=2.98). The second group consisted of 21 patients meeting DSM-III-R criteria for progressive dementia, mainly of the Alzheimer type. Their mean age was 56.00 (SD=11.05), and their mean education level was 11.81 years (SD=4.14).

Subjects received the Wechsler Adult Intelligence Scale and the Halstead-Reitan Neuropsychological Test Battery (5). The dependent measure used for correlation with age was the Average Impairment Rating, a summary index based on the entire battery of tests (6). The Average Impairment Rating ranges from 0.0 to 5.0; higher scores reflect greater impairment. Pearson r correlations between age and the Average Impairment Rating were computed for each group.

RESULTS

The 51 patients with schizophrenia and near-normal cognitive function were characterized by normal to mild neuropsychological deficit (mean Average Impairment Rating=1.54, SD=0.41) and average intelligence (mean IQ=103.01, SD=11.01). The 26 patients with schizophrenia and severely impaired cognitive function obtained scores reflecting severe neuropsychological impairment (mean Average Impairment Rating=3.12, SD=0.41) and a low level of intelligence (mean IQ=79.55, SD=8.88). These two groups of patients differed significantly in age, education, and length of hospitalization (the severely impaired patients had longer hospitalizations). The groups did not differ in age at onset of schizophrenia, length of illness, and number of hospitalizations. The mean Average Impairment Rating for the comparison group of 299 patients without schizophrenia was 1.41 (SD=0.41), and their mean IQ was 103.01 (SD=11.01). For the 21 patients with progressive dementia, the mean Average Impairment Rating was 3.44 (SD=0.73) and the mean IQ was 79.38 (SD=17.45).

Because of the difference in educational level between patients with schizophrenia and near-normal or severely impaired cognitive function, partial correlations were computed with education as the covariate. For the patients with near-normal cognitive function, r=0.38 (df=49, p<0.001) and partial r²=0.46 (df=49, p<0.001). For the patients with severely impaired cognitive function, r=0.05 (df=24, n.s.) and partial r²=0.04 (df=24, n.s.). For the patients with dementia, r=0.48 (df=19, p<0.05), and for the patients in the group without schizophrenia, r=0.41 (df=297, p<0.001). All of the correlations except those for the patients with schizophrenia and severely impaired cognitive function were significantly different from zero (two-tailed probabilities were applied). As hypothesized, the correlations revealed that Average Impairment Rating increased with advancing age.

DISCUSSION

The patients with schizophrenia who had near-normal levels of cognitive function generated approximately the same correlations between the Average Impairment Rating and age as did nonschizophrenic patients with no brain damage. Patients with schizophrenia who had severe cognitive impairment produced zero-order correlations with age. Correlations between age and performance on neuropsychological tests are greatly attenuated in individuals with static brain damage (3). The substantial impact of the brain damage obliterates those correlations, and age effects become masked by the consequences of the brain lesion. We also found that nonschizophrenic patients with frank progressive dementia produced a correlation between age and the Average Impairment Rating that was quite high (r=0.48). Thus, severe impairment or age alone cannot account for absence of a significant correlation with age. Partial correlation analyses indicated that education did not greatly influence the differences in correlation between the two clusters of patients with schizophrenia. It is unlikely that these results are attributable to differences in range of scores among groups. Specifically, the Average Impairment Rating ranged from 1.50 for the cognitively intact schizophrenic group to 1.67 for the cognitively impaired group.

We conclude that patients with schizophrenia who have normal or mildly impaired cognitive function show normal age-related changes in cognitive ability but that there is a small subgroup of patients with schizophrenia who do not show the usual pattern of age-related change. These individuals may have been cognitively impaired essentially from birth or may have experienced a decline during the early years of their illness. Although this matter is best studied longitudinally, these findings tend to favor the early life theory, which also receives some support from the lower educational level in the severely impaired patients as well as from our unreported finding that the patients with schizophrenia who had severely impaired cognitive function had a significantly lower mean score on the reading subtest of the Wide Range Achievement Test than did patients in the near-normal cluster. This subtest is thought to provide a good estimate of premorbid ability in individuals with schizophrenia (7).

Received June 24, 1997; revised Nov. 6, 1997, and Jan. 20, 1998; accepted March 24, 1998. From the VA Pittsburgh Healthcare System, Highland Drive Division; and the University of Pittsburgh. Address reprint requests to Dr. Goldstein, VA Pittsburgh Healthcare System, Highland Drive Division (151R), 7180 Highland Dr., Pittsburgh, PA 15206-1287. Supported by the U.S. Department of Veterans Affairs.