Late-Onset Schizophrenia and Other Related Psychoses

Dilip V. Jeste, Jane S. Paulsen, and M. Jackuelyn Harris

The goal of this chapter is to discuss (a) the nomenclature of late-onset schizophrenia (LOS) and related psychoses, (b) diagnostic boundaries among these conditions, (c) neurobiologic findings in late-life schizophrenia, (d) the relationship of LOS to early-onset schizophrenia (EOS), and (e) therapeutic considerations in the treatment of late-life schizophrenia. In general, much less attention has been given to late-onset psychoses than to psychoses with onset during adolescence and early adulthood. One of the first steps toward meaningful research in this area is defining and distinguishing among different psychoses with late onset, especially schizophrenia, delusional disorder, psychotic mood disorder, and so-called "organic" psychoses. Furthermore, LOS needs to be contrasted with EOS in terms of clinical characteristics, course, treatment, and pathogenesis.

Most theories of the pathophysiology of schizophrenia are based upon its onset being restricted to adolescence or early adulthood. Yet it should be apparent that comprehensive theories of schizophrenia need to address LOS. We will discuss the available literature on LOS and related psychoses and consider the implications for improving our understanding of the neurobiology of schizophrenia (and other psychoses) in general (see The Neurobiology of Treatement- Resistant Mood Disorders).

At the turn of the century, Kraepelin (35) used the term "dementia praecox" to refer to a disorder we now know as schizophrenia. "Praecox" suggested the onset of the disorder in youth, and "dementia" referred to a deterioration of function in the "emotional and volitional spheres of mental life." Kraepelin (37) himself later questioned the appropriateness of the term "dementia" because the disorder was not always accompanied by a permanent deterioration; remissions did occur in some cases. In addition, he noted that not all of the patients first presented in youth. There was a subset of patients with onset of symptoms well into the fifth, sixth, or seventh decade of life. Kraepelin used the term "paraphrenia" to describe those patients with a relatively late onset of delusions and hallucinations, characterized by a predominance of paranoid symptoms and relatively preserved personality (36). Follow-up studies of Kraepelin's paraphrenic patients, however, indicated that a substantial proportion of them had clinical features (including course) similar to those of dementia praecox (44). Mayer (44) concluded that paraphrenia and dementia praecox were more similar than dissimilar. Subsequently, the label "paraphrenia" fell out of use in favor of the broader category of "schizophrenic syndrome" introduced by Bleuler (4). Later the term "paraphrenia" resurfaced (59), but its definitions were often inconsistent.

Although the literature on LOS and related psychoses dates back to the early 1900s, there are several confounding problems with its interpretation. The most confusing feature of the controversy has been the bewildering array of terms and definitions used to refer to psychotic symptoms beginning in late life. Variously referred to as paraphrenia, late paraphrenia, paranoia, and involutional paranoid disorder, there has been no consensus about the terminology. Although most investigators agree that patients do present with schizophrenia-like psychosis in late life, consensus regarding the grouping and labeling of such patients has been less forthcoming. The term "paraphrenia" has been popular in the European literature, whereas LOS has gained acceptance in the recent American literature. One problem with the term "paraphrenia" has been the wide variation in the diagnostic criteria used for paraphrenia (18). Some investigators have used that term synonymously with paranoid schizophrenia, some refer to it as a combination of all "nonorganic" paranoid psychoses in late life, while still others even include patients with obvious metabolic or other organic etiology of late-onset paranoid psychoses under the category of paraphrenia.

It is notable that the term "paraphrenia" is rarely used for early-onset paranoid psychoses. When a young patient develops a paranoid psychosis, considerable attention is paid to distinguishing among schizophrenia, mood disorder with psychotic features, delusional disorder, and so on. Yet, there is sometimes a tendency to lump different types of late-onset paranoid psychoses under the broad, ill-defined category of paraphrenia. We believe that the diagnostic rigor used for differentiating among various late-onset paranoid psychoses should be no less than that used in younger patients. The term "paraphrenia" is not included in the ICD-10 (66), the DSM-III, the DSM-III-R, or the DSM-IV Draft Criteria. Some authors avoid the problem of nomenclature altogether by the cautious description of such patient groups under "persistent persecutory states of the elderly" (55).

There has been no general agreement on the definition of late onset. Some studies chose 40 years of age as the cutoff, whereas others defined late onset as onset after 45, 60, or 65 years of age (18). Furthermore, it is often difficult to determine the age of onset of schizophrenia, especially in older subjects. Elderly patients may not remember and significant others may have died, making corroboration of the patient's history difficult. The presence of premorbid paranoid or schizoid personality traits may further confuse the issue, and older patients with psychotic symptoms may be thought to have organic mental syndromes, mood disorders, or simple sensory deficits (17). The earlier versions of the Diagnostic and Statistical Manuals (DSMs) did not have an upper age limit for the diagnosis of schizophrenia. It was not until the DSM-III (1) that it was stipulated that the onset of symptoms for schizophrenia had to be before age 45. Subsequently, the DSM-III-R (2) allowed an onset of schizophrenic symptoms after age 45, and it used the term LOS for these individuals. The DSM-IV Draft Criteria (3) do not specify an upper age limit, nor do they include a separate categorization of LOS.

To diagnose LOS, the patient should meet the DSM-III-R (2) criteria for schizophrenia (including duration of at least 6 months), with the additional requirement that the onset of symptoms (including the prodrome) be at or after age 45. The prototypical patient is a middle-aged or elderly person who functioned moderately well through early adulthood (despite some premorbid schizoid or paranoid personality traits) and who exhibits persecutory delusions and auditory hallucinations and shows some improvement in positive symptoms with low-dose neuroleptic therapy, yet has a chronic course.

Although debate regarding the utility of diagnosis without etiologic distinction has been present for years, the current atheoretical system (namely, DSM) is usually considered to be reasonably valid for clinical and research purposes. Spitzer and Williams (61) have discussed certain methods for assessing the validity of a classification of mental disorders and its component categories. Face validity requires expert consensus regarding the description of a diagnostic group. Studies of LOS using DSM-III-R criteria support the face validity of LOS (29). Descriptive validity is present when the characteristics of a particular disorder are unique to that condition. Although the diagnostic specificity of psychiatric symptoms, in themselves, is poor, the DSM-III-R criteria for LOS adequately describe (a) its similarity with EOS and (b) its differences from delusional and mood disorders. Predictive validity requires that course, complications, and treatment response be homogeneous in diagnostically similar patients. From the available data the treatment response, course, and complications of LOS are generally similar to those of EOS, although there is heterogeneity among both groups. Construct validity is the extent to which evidence supports a hypothesis that is helpful in understanding the etiology or pathophysiology of the disorder. At the present time, we can address the face validity, descriptive validity, and predictive validity of LOS. We still do not know enough about the etiopathology of schizophrenia in general to assess the construct validity of LOS.

Late-onset schizophrenia is often characterized by bizarre delusions, which have a predominantly persecutory flavor. Auditory hallucinations are the second most prominent psychotic symptom. Systematized delusions of physical or mental influence are seen in many of the patients. Grandiose, erotic, or somatic delusions may occur in some cases. Schneiderian first-rank symptoms, such as thought broadcasting or two voices arguing with each other, are less common but are not rare. Depressive symptoms are reported by a number of these patients. In contrast, looseness of association and inappropriateness of affect are less common than in younger schizophrenic patients (29).

Whenever an older patient presents with psychotic symptoms, organic pathology must first be ruled out. A complete history, followed by a careful neurologic evaluation, other physical examinations, and appropriate laboratory tests (including tests of thyroid function, toxicology screening, and serologic tests for syphilis), is usually part of the assessment. Computed tomography (CT) or magnetic resonance imaging (MRI) may be needed to identify cases where structural brain abnormalities are suspected (48).

In a recent study of 367 geriatric inpatients in four different hospitals, 50% of patients had paranoid symptoms (16). Given the relatively high incidence of paranoia in geriatric patients, diagnostic specificity becomes imperative. Two important conditions in the differential diagnosis of LOS (other than the "organic mental syndromes") are mood disorders with psychotic features and delusional disorder. These diagnoses are more likely to have onset during middle age or old age than during early adulthood. Mood disorders with psychotic features may present for the first time after age 45, and they can be confused with LOS. The predominance of affective symptoms and periodicity of the illness should make the clinician consider a mood disorder. A diagnosis of schizophrenia is made when the total duration of all mood symptoms has been brief relative to that of the primary psychotic symptoms. Delusional disorder may mimic LOS, but the latter diagnosis is more likely in the presence of bizarre delusions or prominent auditory hallucinations, Schneiderian first-rank symptoms, deteriorated functioning, and flattening of affect (67). Delusional disorder is distinguished from LOS by the presence of nonbizarre delusions and the absence of prominent auditory or visual hallucinations, disorganized speech, negative symptoms, and functional impairment outside the area of delusions (2).

Similar to EOS, there is an insidious deterioration of personal and social adjustment. A sizable proportion of LOS patients have abnormal premorbid personality traits of a paranoid or schizoid nature (23, 33). Some patients have never been married and have been previously considered by acquaintances to be eccentric, reserved, and suspicious. In our recent study of clinical characteristics in LOS, the LOS and EOS groups were both impaired (compared to normal controls) in childhood adjustment. These findings suggest that the LOS patients might have had a perinatal predisposition similar to that in the EOS group. Nevertheless, when compared to EOS subjects, patients with LOS were more likely to have been married, to have held a job, and to have had better adjustment during adolescence and early adulthood (28).

Caution is required in determining the age of onset of schizophrenia strictly on the basis of a first psychiatric hospitalization for psychosis. It is prudent to obtain a detailed history of the disease from both the patient and his or her family or friends because some patients might have had the prodromal symptoms for some time prior to the first hospitalization. In patients with suspected LOS, it is important to establish an absence of prodromal symptoms before age 45 to exclude the diagnosis of EOS. Prodrome is characterized by a clear deterioration in functioning before the active phase of psychosis, and it includes symptoms such as the following: marked social isolation; marked impairment in role functioning; peculiar behavior; marked impairment in personal hygiene; blunted or inappropriate affect; digressive, vague, over-elaborate, or circumstantial speech; odd beliefs or magical thinking influencing behavior and inconsistent with cultural norms; unusual perceptual experience; and marked lack of initiative (2). The prodrome differs from premorbid personality disorder in that it requires a clear deterioration in a previous level of functioning. Also, some patients with an apparently late onset of psychosis may have had a more benign illness which never required hospitalization until age 45.

LOS patients show a female predominance not found among EOS cases (8, 18, 54). Most studies of LOS show a 2-10 times higher proportion of women than men. In a review by Pearlson and Rabins (52), the authors concluded that there seemed to be an age-related fall in the number of dopamine D2 receptors but men lost them at a faster rate with increasing age, leaving older women with a relative excess. It has also been suggested that estrogen in premenopausal women might have a protective role. The contribution of these and other factors to the later onset of schizophrenia in women is unclear.

Many schizophrenic patients survive into old age, yet comparatively little is known about the long-term course of LOS. A review of the literature on the course of schizophrenia in general suggests that a majority of patients either undergo remission or are left with mild symptoms over the long term (45). The more positive, dramatic symptoms of schizophrenia seem to lessen in severity with the passage of time. In a recent study of a 3-year follow-up of older schizophrenic patients, negative symptoms were prevalent, moderately severe, and quite stable over time (24). Studies of the course of LOS have generally found it to be chronic (23, 33, 67). Hafner et al. (32) found a higher frequency of negative symptoms and a more extended course of negative symptoms in women over the age of 35 with a diagnosis of schizophrenia, again supporting a tendency towards chronicity.

Studies have reported the frequency of LOS among consecutive admissions to inpatient psychiatric facilities to range from 3-4% (13, 67) to as high as 8-10% (41, 59). These studies have utilized varying diagnostic criteria, however. A review by Harris and Jeste (18) estimated that 13% of the hospitalized schizophrenic patients were reported to have had onset of psychosis in their forties, 7% had onset in their fifties, and only 3% first presented after age 60. These percentages cannot be considered definitive, however, given the methodological issues in terms of diagnosis of schizophrenia and determination of age of onset, as well as policies and practices for hospitalizing patients.

The following discussion of neurobiology is limited by the problems in clinical methodology considered above. Nonetheless, the more recent studies in this area have had considerably greater diagnostic rigor than the older ones.

A structural brain finding in LOS reported by several groups of investigators in younger schizophrenic adults has been increased ventricle-to-brain ratio (VBR) seen on CT or MRI. Rabins (57) reported that 25 of 29 LOS patients had VBRs that were higher than the mean for an age-matched cohort; however, the mean VBR was smaller than that in an age-matched group of patients with probable Alzheimer's disease who had concurrent psychotic symptoms. In a more recent study of patients with LOS (N = 11), Alzheimer's disease (N = 12), and age- and gender-matched normal controls (N = 18), only one of three VBR measures was found to be increased in LOS (40). There were significant increases in third ventricular volume, lateral VBR, and total percentage CSF in Alzheimer's disease; by comparison, only third ventricular volume was significantly increased compared to controls in LOS. In one report directly comparing EOS and LOS patients on visual analog measures of atrophy, Pearlson et al. (53) showed similar ratings of atrophy in these two patient groups. In one of the few longitudinal studies to date, Naguib and Levy (49) reported that VBR was not associated with clinical course or outcome (26).

In addition to the quantification of ventricular and sulcal size, several reports have documented large areas of white-matter hyperintensities in patients with LOS and related psychoses compared to age-matched normal comparison subjects (6, 10, 38, 39, 47, 48). In addition, some reports have noted an increased number of vascular lesions in such patients (6, 14).

Only a few studies of functional brain imaging have been done in LOS patients. Single photon emission computed tomography (SPECT) studies by Miller et al. (48) revealed evidence of blood flow changes suggestive of cerebrovascular disease in some late-onset psychotic patients. Eighty-three percent of their late-life psychotic subjects and 27% of normals had one or more small temporal or frontal areas of hyperfusion (40). Similarly, Dupont et al. (12) found lower global cortical uptake (particularly in the left posterior frontal region and bilateral inferior temporal regions) in late-life schizophrenic patients (N = 11) than in normal comparison subjects (N = 11). The uptake did not correlate with age of onset, duration of illness, current daily neuroleptic dose, severity of psychopathology, or global cognitive impairment. In the only available published study of positron emission tomography (PET) in LOS, Pearlson et al. (53) reported elevated Bmax (receptor density) values for dopamine D2 receptors in 13 neuroleptic-naive LOS patients compared to age and gender norms. An important question arises as to whether brain abnormalities associated with LOS are of pathogenetic significance, and how they compare with those in EOS.

There is presently a considerable amount of published research on neuropsychological functioning in schizophrenia. Despite notable heterogeneity in the behavioral presentation of schizophrenia, there is general agreement regarding the presence of neuropsychological impairments that accompany the disorder. Relatively little is known, however, about the potential interaction between the cognitive deficits associated with schizophrenia and aging (19, 22).

A number of long-term follow-up studies of schizophrenia patients have shown that the clinical prognosis of schizophrenia is better than that implied in Kraepelin's use of the term "dementia praecox." Bleuler, Huber, and Ciompi all came to the same conclusion--that is, that schizophrenia was in no sense a "basically" or even "predominantly" unfavorable disease process that ran an inexorably deteriorating course (9). The proportion of patients with favorable outcome (either recovery or mild end state) was 53% in Bleuler's study, 57% in the study by Huber et al., and 59% in that by Ciompi (9). Only 18% of Ciompi's patients had "severe" end stage. These findings were corroborated in the 10 North American long-term follow-up studies of schizophrenia reviewed by McGlashan (46), who concluded that the schizophrenia process appeared to plateau after 5-10 years of manifest illness. Neuropsychological abilities were not addressed in these studies, however, so the nature and degree of cognitive impairment in late-life schizophrenia remain unclear.

In a review addressing this issue, Heaton and Drexler (22) examined 100 cross-sectional studies and 10 longitudinal studies of neuropsychological functioning in schizophrenia. The results generally argued against the possibility of a progressive neuropsychological impairment, and in particular suggested that schizophrenia patients probably did not have more rapid neuropsychological decline associated with the aging process. Heaton and Drexler (22) noted numerous methodological limitations of the available literature, however, qualifying the conclusions made. Despite an obvious need, there have been very few neuropsychological studies of LOS.

Heaton et al. (20) recently investigated neuropsychological performance in patients with EOS, LOS, and Alzheimer's disease and in normal comparison subjects. All the schizophrenia groups were worse than the normal comparison group on all the ability areas except for memory; no schizophrenia group showed impairment on the memory deficit score. By contrast, the Alzheimer's disease group was worse than the normal comparison group on memory as well as on all the other areas. Discriminant function analyses further indicated that the neuropsychological discriminant functions rarely misclassified a schizophrenia subject as having Alzheimer's disease, or vice versa; moreover, there was no tendency for this type of error to be made more often with LOS subjects than with EOS patients.

It can be argued that some of the neuropsychological deficits in schizophrenic patients may be due to the effects of medication. There is evidence that anticholinergic drugs can interfere with cognitive functioning, especially learning and attention (21, 62). Typically, learning impairment is associated with higher anticholinergic dosage or acute change in anticholinergic medication regimen. In terms of the reported effects of neuroleptic drugs on cognitive and psychomotor functions in patients and normal controls, there has been some variability and inconsistency in the literature (34). In general, sedative phenothiazines have been found to depress psychomotor function and sustained attention, but higher cognitive functions are relatively unaffected. In the majority of studies of schizophrenic patients, both cognitive function and attention improved with neuroleptic treatment, in parallel with clinical recovery. In general, the studies of neuropsychological effects of neuroleptic therapy have not been addressed specifically in older schizophrenic patients.

An additional concern with regard to cognitive functions associated with late-life psychosis is whether tardive dyskinesia (TD) affects neuropsychological performance. It is well known that the risk of TD increases with age (27, 31). Thus, neuroleptic-treated patients with psychosis in late life are more likely to develop persistent TD than their younger counterparts. Less is known, however, about the interaction of cognition and TD. In a review of 31 studies we found that TD was generally associated with greater cognitive impairment (51). In a recent study of neuropsychological abilities in older schizophrenia patients we showed that severity of TD was highly associated with severity of neuropsychological impairment. Findings demonstrated that schizophrenia patients with TD had greater global cognitive deficits and greater deficits in learning than did age-, education- and subtype-matched schizophrenia patients without TD. These results suggest that research on neuropsychological abilities in schizophrenia needs to address the presence and severity of TD.

This brief summary of the neuropsychology of schizophrenia and aging emphasizes the need for further longitudinal research of cognitive functioning over the course of the schizophrenic illness. There are a number of methodological issues raised by the current research in this area which limit the conclusions made (22, 42, 62). Some of the methodological limitations include: (a) the methods for identification and selection of schizophrenic subjects; (b) the variable course of schizophrenic illness such that clinical state must be considered to determine which cognitive deficits are state- or trait-related requiring repeated assessments over time; (c) the issue of treatment effects (i.e., medication) and the cognitive pattern of patients with tardive dyskinesia; and (d) the evaluation of generalized and specific neuropsychological deficits. Future research in this area needs to control for these clinical and treatment factors as well as consider more specific analyses of cognition in addition to traditional global indices of performance.

Neurochemical investigation has played a major role in the search for the pathophysiology of schizophrenia. Historically, there has been an emphasis on the dopamine neurotransmitter system and its role in the etiology of schizophrenia. More recently, increased evidence for the role of other neurotransmitters, coupled with limitations in the dopamine hypothesis of schizophrenia, has extended the scope of clinical neurochemical studies. The neurotransmitters that have come under increasing scrutiny include serotonin, norepinephrine, glutamate, and related excitatory amino acids, as well as the neuropeptides cholecystokinin and neurotensin. To our knowledge, no large-scale systematic study has specifically evaluated the plasma or cerebrospinal fluid of patients with LOS or related psychoses for neurochemical or neuroendocrine abnormalities. The ability to predict neuroleptic responsiveness and psychotic relapse in older schizophrenic patients using neuropeptide, neurohormone, and neurochemical parameters may provide insight into the role of limbic/mesolimbic pathogenic mechanisms in the schizophrenic disease process.

Studies to evaluate the prevalence of schizophrenia in families of LOS patients have been hindered by methodologic problems. For example, not all family members have been followed into old age to ensure that every case of LOS is detected. Physical illness, geographic relocation, and untimely death of relatives make it difficult to conduct such studies. The published studies suggest that the prevalence of schizophrenia (whether early or late in onset) is approximately 7% in siblings and 3% in parents of all probands with LOS. Rokhlina (58) reported that the overall prevalence of schizophrenia in the first-degree relatives of LOS patients was lower than that in the families of EOS patients.

Funding (15) and some other authors conclude that the mode of inheritance of LOS, similar to that of EOS, is probably polygenic, with many nongenetic factors being influential. In a more recent study of HLA antigens, however, Naguib and McGuffin (50) postulated that the syndrome of LOS might be genetically distinct from EOS. Despite the previously reported association between HLA-A9 and paranoid schizophrenia, the investigators were unable to replicate this finding when 31 LOS patients were HLA-typed. Although their sample size was too small to refute confidently an association with A9, they did not even detect a trend in this direction. The authors concluded that the possibility of LOS being genetically distinct from EOS required further study.

Several studies have reported an association between hearing impairment and late-life schizophrenia. Cooper and Porter (11) noted an increased incidence of cataracts in a population of paranoid patients compared to affective controls and also noted an association between bilateral conductive hearing loss and paranoid illness. Prager and Jeste (56) recently reviewed 27 published studies of a possible association between sensory (visual or hearing) impairment and late-life psychosis with paranoid features. A majority of these investigations supported the postulated association between hearing impairment and LOS/paranoid disorder. Many of the published studies, however, had important methodologic limitations. Also, a causal relationship between sensory deficits and LOS remains to be established. In a case-control study, Prager and Jeste (56) assessed visual and hearing impairments in 87 middle-aged and elderly subjects, including 16 with LOS, 25 with EOS, 20 with mood disorder, and 26 normal comparison subjects. Uncorrected and corrected (with eyeglasses and hearing aids) visual and hearing impairments were assessed in a "blind" manner, using standardized quantitative assessments. The two schizophrenia groups as well as the mood disorder group had greater impairments in most variables of corrected visual acuity and in self-reported hearing deficit, but not in uncorrected (constitutional) visual acuity or in pure-tone audiometry, compared to normal subjects. This suggests that schizophrenia (LOS and EOS) and mood disorder patients may have equivalent sensory deficits but inadequate correction of their sensory impairments compared to normal comparison subjects. Hence a specific contribution of sensory deficits to the etiopathology of LOS remains uncertain.

Late-onset psychotic disorders may be fundamentally similar to their early-onset counterparts in the underlying neurobiologic predisposition. Certain specific protective factors may, however, prevent an earlier breakdown, whereas other aging-related precipitants may be responsible for onset of symptoms during later life. Specificity of neurobiologic substrates for different types of late-onset psychoses is yet to be established.

Attempts at an integration of neuroimaging, neuropsychology, neurochemical, and genetic research in schizophrenia have recently begun. Similar efforts must now be made to link clinical and basic research in late-onset psychoses with those findings already established in younger cohorts. Only with unified nomenclature, hypotheses, and research can the study of late-onset psychosis progress further.

According to the neurodevelopmental model, which is primarily based on studies of schizophrenia with onset during adolescence or early adulthood, the brain lesions putatively related to the pathogenesis of schizophrenia are of developmental origin (65). Furthermore, long-term treatment with neuroleptics and other environmental factors are not the principal causes of cognitive deficits and brain abnormalities reported in chronic schizophrenic patients (20, 65).

At the present state of our knowledge, we may hypothesize that the neurodevelopmental model applies to LOS too. Differences in severity and specific locations or nature of these "lesions" may be at least partly responsible for a delay in the onset of schizophrenia (e.g., lesser severity of the nonprogressive brain lesions that presumably occurred during the developmental period may explain the relatively better social, heterosexual, and occupational functioning during early adulthood in the LOS group). Another possibility is that the peak in dopaminergic activity related to the schizophrenic breakdown, as per the neurodevelopmental theory (65), may be delayed until later in life in the patients with LOS. For obvious reasons, this discussion must be considered speculative. Nonetheless, our suggestions may offer some interesting leads to pursue in this relatively unstudied entity.

Neuroleptics have been shown to be the most effective treatment modality for schizophrenia, although pharmacotherapy in the older patient is complicated by alterations in both the pharmacokinetic and pharmacodynamic responses. Unfortunately, the published literature on neuroleptic treatment of late-life psychosis is extremely sparse, although the available studies suggest that a significant number of late-life schizophrenic patients improve symptomatically with neuroleptic treatment (30). Most of the published reports do not, however, distinguish between LOS patients and elderly subjects with EOS.

In a 24-week, double-blind, placebo-controlled study involving 13 hospitals and 308 schizophrenic men aged 54-74 years, acetophenazine and trifluoperazine were both more effective than placebo in motor disturbances, conceptual disorganization, manifest psychosis, and a lack of personal neatness (25). In another study of 50 psychogeriatric (mostly chronic schizophrenic) patients, Tsuang et al. (64) reported a significant decrease in psychopathology for both haloperidol and thioridazine treatments in a 12-week double-blind study. Finally, Branchey et al. (5) demonstrated efficacy of orally administered fluphenazine and thioridazine in chronic schizophrenic patients (mean age 67 years) in a double-blind, crossover study.

With the possible exception of clozapine, the available data suggest that the commonly prescribed antipsychotic medications are equally efficacious. Therefore, selection of an antipsychotic for use in the elderly should be based primarily on the following: (a) the side effect profile of the particular drug; (b) the potential adverse consequences of the additional antipsychotic to preexisting medication regimen or a concomitant physical illness; and (c) a history of a patient's previous therapeutic response to a specific neuroleptic.

An important consideration is the higher incidence of side effects seen in elderly patients, as compared to younger patients, upon administration of a given amount of neuroleptic. Some studies have shown positive correlations between age and serum level of neuroleptic when equivalent doses of neuroleptic are administered (63). Usually, small doses of neuroleptics are sufficient for improvement in older patients. The side effect profiles of individual neuroleptics differ considerably, and such differences may be important in prescribing a particular medication to a patient for whom the occurrence of a particular side effect might prove dangerous (43). For example, in patients with preexisting parkinsonian symptoms, high-potency antipsychotics (e.g., haloperidol) may worsen tremor and rigidity. On the other hand, high-potency neuroleptics (especially haloperidol) are reported to have lower cardiovascular toxicity than low-potency neuroleptics. Hence, haloperidol may be a better choice than low-potency neuroleptics among patients with preexisting cardiovascular disorders. Low-potency neuroleptics (e.g., thioridazine) have marked anticholinergic activity and should be avoided in patients with prostatic hypertrophy or in those who are already on other anticholinergics. Finally, an agent that previously produced a positive response in that patient or in a blood relative may be tried first, whereas an agent that led to an unfavorable reaction may be avoided.

A number of cross-sectional investigations have found a significantly greater prevalence of neuroleptic-induced tardive dyskinesia (TD) in older patients than in younger ones (27). Saltz et al. (60) reported a 31% incidence of TD after 43 weeks of cumulative treatment with neuroleptics in a group of 160 elderly subjects (mean age 77 years), predominantly female, with a sizable proportion being institutionalized.

In a recent prospective, longitudinal study of TD in older persons, Jeste et al. (28) evaluated the overall cumulative incidence of TD (i.e., the total number of new cases of TD) over a 1-year period in a sample of 236 middle-aged and elderly outpatients with a mean age of 66 years. The annual incidence was 29.87% for schizophrenia patients (N = 48), 30.73% for the patients with Alzheimer's disease (N = 55), 33.34% for patients with mood disorders (N = 33), and 33.96% for patients with miscellaneous organic diagnoses (N = 100). These data suggest that the risk of TD over a 1-year period is no different in schizophrenic outpatients than in other diagnostic groups among middle-aged and elderly subjects. The notably high incidence of TD among all the patient groups over age 45 years points to a need for caution in prescribing neuroleptics in this population. There have been few, if any, systematic studies of the incidence of TD with "atypical" neuroleptics such as clozapine and risperidone in elderly schizophrenic patients.

Although there have been a number of studies of neuroleptic dosage in young adults with schizophrenia, little is known about the neuroleptic dose requirement in older schizophrenic patients. We recently reported the results of cross-sectional data on associations of neuroleptic dose with selected demographic, clinical, and neuropsychological variables in a group of 64 schizophrenic outpatients over the age of 45 (30). The neuroleptic dose correlated significantly with current age, age at onset of illness, severity of negative symptoms, and impairment on a variety of neuropsychological test measures such as the Halstead Reitan Average Impairment Rating, Story and Figure Learning, and psychomotor speed. An inverse correlation between age and neuroleptic dose is consistent with studies reporting pharmacokinetic (high blood levels) and pharmacodynamic (increased sensitivity to drug response) changes associated with aging (63). Age-related pharmacokinetic alterations in drug disposition may produce higher plasma concentrations of specific neuroleptic medications in older patients than in younger ones. Age-related pharmacodynamic alterations at the receptor or neurotransmitter level may increase the intensity of drug response in the elderly. The association of later age at onset of schizophrenia with lower dosages may also be consistent with the suggestion that LOS has a better prognosis than EOS (7, 18). Well-controlled, prospective studies are required before one can interpret the correlation that we observed between neuroleptic dose and either the severity of negative symptoms or that of the neuropsychological impairment.

The results of controlled research on psychological treatment suggest that intervention may improve the outcome of schizophrenia, although many patients require long-term treatment due to the chronic nature of the illness. A widely studied psychological intervention for schizophrenia is social skills training, although cognitive retraining and didactic family counseling have also shown positive effects. However, despite more than two decades of psychological treatment research in young schizophrenic patients, no published study to our knowledge has evaluated the efficacy of these treatment strategies in older schizophrenic patients or in those with late-onset psychosis. Given the numerous psychological issues that become paramount in later life (e.g., retirement, grief over losses, financial limitations), future research needs to evaluate the efficacy of various psychological treatment strategies in patients with LOS and other late-onset psychoses.

In the past, most of the research interest in schizophrenia focused on younger adults. With the aging of the general population, there is now an increase in the attention being given to older schizophrenic patients, including those with a late onset of the disorder. Other late-onset psychotic disorders are also beginning to be studied more comprehensively. The literature in this area is still very limited, however, and is also fraught with a number of methodologic problems. Nonetheless, the following "conclusions" can be made with at least some degree of certainty at this time:

There is a need for longitudinal, multicenter, collaborative studies of late-onset psychosis using well-defined diagnostic criteria and reliable, validated instruments for assessment. Apart from clinical characteristics (including course and treatment-response), the other aspects that should be investigated include: neuropsychological performance, structural and functional brain imaging, neurochemical and neuroendocrine parameters, and neuropathological variables.

Comparisons of early-onset and late-onset psychosis patients using age-corrected measures are warranted. Also, the utility of the current diagnostic differentiation of late-onset psychosis (i.e., schizophrenia, delusional disorder, psychotic depression, etc.) needs to be established from both clinical and neurobiologic perspectives.

This work was supported, in part, by NIMH grants MH45131, MH43693, and MH49671-01 and by the Department of Veterans Affairs.

published 2000