Forty-Year Study Reveals Patterns of Cognitive Decline in Schizophrenia
September 25, 2013. The most comprehensive longitudinal study of cognitive deficits in schizophrenia published to date, with follow-up assessments spanning nearly four decades, finds substantial neuropsychological decline between childhood and the post-onset period of adulthood. Led by Terrie Moffitt of Duke University in Durham, North Carolina, researchers examined both general IQ as well as specific components of cognition and found that deficits in mental functions such as processing speed display different patterns of progression across development. Their findings were published online September 13, 2013, in the American Journal of Psychiatry.
SRF advisor Jim Gold of the University of Maryland called the paper “a powerful confirmation of several observations that have been in the literature for many years.” In his comment (see below), he added, “It is the first time that these key findings have all been demonstrated in the same subjects followed over time in a population-based sample.”
Several studies have demonstrated that people who later develop schizophrenia show cognitive deficits early in life (Dickson et al., 2011). However, previous longitudinal studies that have assessed neuropsychological functioning both before and after illness onset have been limited by the lack of a comparison group, the use of different cognitive measures across time, and/or a focus solely on global measures of cognition. Moffitt’s study, part of an ambitious, ongoing project called the Dunedin Multidisciplinary Health and Development Study and not subject to the same pitfalls as previous research, provides the most detailed picture of cognition in schizophrenia across development yet.
The Dunedin study has examined nearly all aspects of the physical and mental health of 1,037 babies born at a single hospital in Dunedin, New Zealand, over one year between 1972 and 1973. Participants have been followed roughly every other year from ages three to 18 and every few years after that until the most recent assessment at age 38.
“What's really special about the Dunedin study is a combination of three things," Moffitt wrote in an email to SRF. "First, it began with all the babies born in one city, so it fully represented variation in the population, with over 1,000 children from all walks of life and all ability levels. Second, [nearly] 40 years later, 95 percent of those original babies are still taking part, which means that individuals with mental health problems have not dropped out along the way. Third, data collection involves a full day in the lab for each study member, which allows us to take an amazingly broad set of measurements.”
So far, the study has generated over 1,000 research publications about a wide range of topics including cardiovascular, dental, and mental health (see SRF related news story and SRF news story). A previous study from Moffitt’s group found premorbid cognitive deficits in the children who later developed schizophrenia, including early deficits that remained stable throughout childhood as well as some that declined over time (Reichenberg et al., 2010).
In the current study, first author Madeline Meier and colleagues extended the previous findings into adulthood and added data on IQ changes, as well as component mental processes, from the most recent study follow-up. General intellectual ability was assessed at ages seven, nine, 11, 13, and 38 by using IQ tests that were broken down into two components—verbal and performance IQ—and which index different mental functions. In addition, even more specific domains of cognition (learning and memory, processing speed, executive function, and motor function) were assessed with independent tests at ages 13 and 38.
Change across time
Compared to 517 healthy controls, the 31 participants who later developed schizophrenia started off with lower IQs, showing a nine-point IQ deficit in childhood. By adulthood, the difference was an even more substantial 15 points. Indeed, the schizophrenia group lost an average of six IQ points between childhood and adulthood, while no such loss was observed in the controls. Lower IQ was present at the earliest age tested (seven years old) and remained constant until age 13 before dropping sharply. Additional analyses indicated that the IQ changes in schizophrenia were not the result of treatment with antipsychotic medication or dependence on cannabis, alcohol, or hard drugs.
The IQ decline was not present in 185 subjects with persistent depression or in 120 people with mild cognitive impairment during childhood (although a three-point deficit in IQ was observed in the depressed subjects in both childhood and adulthood). A separate analysis that matched individual schizophrenia subjects with “at-risk” participants who shared similar childhood risk factors for the illness (such as low socioeconomic status) found that those with a later diagnosis of schizophrenia had a significantly greater IQ decline than those who did not. Taken together, these findings are important, said Meier, because “they show that neuropsychological decline is really quite specific to schizophrenia.”
When the researchers “unpacked” IQ into specific mental functions, they found a decline in performance IQ, learning, processing speed, executive function, and motor function over time in schizophrenia subjects. Only the verbal IQ score and delayed memory performance of schizophrenia subjects were stable across time, though performance was still worse than that of the controls. Although a decline in cognition was present across several domains of function, the specific pattern of alterations varied across mental functions. For example, performance on the digit symbol test, an index of processing speed, was normal at age seven, but a deficit emerged slowly between ages seven and 38, reflecting “a gradual, progressive process of slowed growth ... that begins in childhood and continues beyond the early teen years,” according to Meier. In contrast, difficulties on the similarities test, a measure of verbal ability, were present by age seven and remained relatively stable through midlife.
In broad terms, "fluid" cognitive abilities such as processing speed, learning, and executive function showed the most substantial decline over time, while deficits in "crystallized" abilities such as verbal IQ emerged early and remained stable across time. This suggests that the two types of cognitive deficits may be the result of different pathophysiological mechanisms in the illness, say the authors.
The findings are “a particularly clear example of a moderate, generalized deficit in cognitive ability well before the onset of illness that, after onset, leads to further declines in fluid, but not crystalized intelligence,” wrote the University of Minnesota’s Angus MacDonald in a comment submitted to SRF (see below).
Sure of schizophrenia?
As any clinician can attest, diagnosing schizophrenia can be extremely complicated. It’s even trickier in population-based studies such as this one, which, unlike investigations of clinical cohorts, examine participants regardless of whether or not they have symptoms. In the Moffitt study, diagnoses of schizophrenia were made using the Diagnostic Interview Schedule (DIS). A lack of insight into one's illness can make relying on such self-reported information problematic, and prior attempts to use similar personal interviews, conducted by trained lay interviewers, have been unable to reliably diagnose schizophrenia (Kendler et al., 1996).
To circumvent these problems, Meier and colleagues used a clinician-administered DIS and relied on several additional pieces of information to make best-estimate diagnoses. They required the presence of hallucinations along with at least two other positive symptoms, a stricter criterion that can help to reduce overdiagnosis. Inclusion in the study also required evidence, obtained from a variety of sources, of objective impairments due to psychosis. Symptoms directly observed by the study’s clinicians, reports of mental illness symptoms and treatment from both parents and "informants" who knew the subjects closely, medication lists, and administrative hospital records were compiled into case files accumulated over assessments from ages 21 to 38. These dossiers were used to achieve 100 percent consensus diagnoses by four clinicians.
Using this approach, the prevalence rate of schizophrenia was higher than in many other studies, which could mean that not all members of the current group really had schizophrenia. By age 38, 2 percent of the original birth cohort met the study’s diagnostic criteria for schizophrenia and had previously been hospitalized for schizophrenia or received antipsychotic medications. Another 1.7 percent also met the criteria but had not specifically been treated for psychosis (though the majority had been treated for another mental health problem). The two groups were combined in all analyses.
However, the nature of the sample—a birth cohort with a very high participation rate—suggests that the study included some individuals who would be missed in other types of studies, say the authors. In addition, the findings that the premorbid and post-onset IQ scores for the schizophrenia group were very close to those recently identified in meta-analyses increases the researchers' confidence in the diagnostic criteria used (Mesholam-Gately et al., 2009). Moreover, both the subjects treated for psychosis (who received a formal diagnosis) and those who were not were nearly equivalent in adult IQ, substance abuse, and receipt of government benefits, as well as other variables. The authors argue that the similarities between the two groups suggest that both are representative of schizophrenia.—Allison A. Curley.
Meier MH, Caspi A, Reichenberg A, Keefe RS, Fisher HL, Harrington H, Houts R, Poulton R, Moffitt TE. Neuropsychological Decline in Schizophrenia From the Premorbid to the Postonset Period: Evidence From a Population-Representative Longitudinal Study. Am J Psychiatry. 2013 Sep 13. Abstract
Comments on News and Primary Papers
Comment by: Angus MacDonald, SRF Advisor
Submitted 23 September 2013
Posted 23 September 2013
The Dunedin study is not only a rich and rare resource for testing developmental hypotheses, but it has also been mined with ingenuity and resourcefulness over the years by Avshalom Caspi, Terrie Moffitt, and their colleagues to provide a number of provocative findings. In this case, they use the continuity of the sample and its multiple-informant design to test a number of useful hypotheses about the development of cognitive impairments in schizophrenia. Their population-based cohort of over 1,000 children yielded 31 cases of tightly defined schizophrenia by age 38. (The fact that this is over 3 percent of the sample, the authors argue, is explained by the comprehensiveness of their methods, suggesting that lower epidemiological estimates may underrepresent lifetime population risks.)
Their findings provide a particularly clear example of a moderate, generalized deficit in cognitive ability well before the onset of illness that, after onset, leads to further declines in fluid, but not crystalized intelligence. The example is clear because it addresses the diagnostic specificity of the deficit—the pattern was different for children later diagnosed with depression or mild cognitive impairments—and it is corroborated by the reports from others’ throughout their lives. The findings reinforce efforts by the U.S. NIH and FDA to target cognitive impairments as a symptom of interest for patients with schizophrenia.
The findings hold important methodological lessons for schizophrenia researchers, too. As pointed out by Paul Meehl in 1971, psychopathologists who co-vary or control for factors influenced by the illness may make a systematic mistake. This kind of control variable sets up false equivalences by comparing the most able patients to the least able controls.
View all comments by Angus MacDonaldComment by: James Gold, SRF Advisor
Submitted 25 September 2013
Posted 25 September 2013
The recent paper by Meier et al. is a powerful confirmation of several observations that have been in the literature for many years. First, cognitive impairment is evident from early in development in people who go on to develop schizophrenia. Second, there is a loss of intellectual function that occurs in those people at risk who later become ill. Third, this "illness-associated" impairment appears to maximally impact more "fluid" intellectual functions and largely spares "crystallized" forms of verbal knowledge (however, those functions are not fully spared, as there is evidence of subnormal performance levels from early in development).
Thus, what Meier et al. have shown is not new, but it is the first time that these key findings have all been demonstrated in the same subjects followed over time in a population-based sample. The inclusion of a group of depressed patients as well as a mild cognitive impairment control group are innovations that enhance confidence that this is an effect related to schizophrenia in particular rather than psychopathology or cognitive limitations in general. Unexpectedly, this study also found a very sizeable number of people meeting diagnostic criteria for schizophrenia who appear to have been untreated with antipsychotics for extended time periods but did receive treatment for other mental health problems. It will be interesting to learn more about the life course and treatment history of this unusual group of people.
View all comments by James Gold
Comments on Related News
Related News: Meta-analysis Supports Case for Cannabis in Etiology of PsychosisComment by: Jim van Os
Submitted 8 August 2007
Posted 8 August 2007
This excellent review confirms the previous meta-analysis by Henquet et al. (2005) and as such does not add anything new. The importance lies in the UK context: previously the Lancet has been mostly skeptical with regard to this issue. The fact that the leading UK medical journal now also allows these findings to see daylight is a significant event and helps stimulate further funding for the effort that several groups worldwide have started working on over the last five years: the search for the mechanism explaining the link.
View all comments by Jim van Os
Related News: Meta-analysis Supports Case for Cannabis in Etiology of Psychosis
Comment by: John McGrath, SRF Advisor
Submitted 9 August 2007
Posted 10 August 2007
I recommend the Primary Papers
It is reassuring to see that the results of the latest meta-analysis (Moore et al., 2007) are consistent with previous meta-analyses, and that the various meta-analyses are broadly consistent with the now much-tortured primary data. Despite the meta-analysis fatigue, the results are too important to ignore.
When thinking about the impact of cannabis on schizophrenia frequency measures, it is important to remember that cannabis use may translate to an increase in the prevalence of active psychosis via two mechanisms. The data suggest that as the prevalence of cannabis use increases in a population, the incidence of schizophrenia should also increase (Hickman et al., 2007). Furthermore, in those with established schizophrenia, cannabis use is associated with poorer outcomes (i.e., reduced remission rates). Thus, from a modeling perspective, increased cannabis use could lead to an increase in the prevalence of active psychosis via two mechanisms (i.e., increased “inflow” and decreased “outflow”) (McGrath and Saha, 2007).
The prevalence of active psychosis in the community may be “under the influence” of cannabis from more than one perspective.
Moore THM, Zammit S, Lingford-Hughes A, Barnes TRE, Jones PB, Burke M, Lewis G. Cannabis use and risk of psychotic or affective mental health outcomes: A systematic review. Lancet. 2007 July 28; 370:319-328. Abstract
McGrath J, Saha S. Thought experiments on the incidence and prevalence of schizophrenia “under the influence” of cannabis. Addictions 2007 Apr;102(4):514-5. Abstract
Hickman M, Vickerman P, Macleod J, Kirkbride J, Jones PB. Cannabis and schizophrenia: model projections of the impact of the rise in cannabis use on historical and future trends in schizophrenia in England and
Wales. Addiction. 2007 Apr;102(4):597-606. Abstract
View all comments by John McGrath
Related News: Meta-analysis Supports Case for Cannabis in Etiology of Psychosis
Comment by: Dana March, Ezra Susser (SRF Advisor)
Submitted 20 August 2007
Posted 20 August 2007
The recent meta-analysis in the Lancet (Moore et al., 2007) regarding cannabis use and psychotic or affective mental health outcomes is, indeed, a necessary contribution. It is the first systematic review restricted to longitudinal studies of cannabis use and mental health outcomes. For this addition to the contours of the literature, Zammit and colleagues are to be commended.
We may be more optimistic than the authors, however, about the potential for future longitudinal studies to shed further light on the question of causality, and perhaps more cautious about the present state of the evidence. Given the public health and policy implications, we propose a concerted effort to complete observational studies that are designed to rule out the main alternative explanations for the association (e.g., genetic or social factors that independently influence both cannabis use and psychosis). The Swedish conscript study (Zammit et al., 2002) is a fine example of one such study. We should also be considering natural experiments and designs based on instrumental variables enabled by in order to complement this work. For instance, we might capitalize on situations created by policy changes that affect the availability—and therefore use—of cannabis in order to examine the impact on the development of psychosis. Whether at the individual or the population level, both creativity and rigor are required.
Moore TH, Zammit S, Lingford-Hughes A, Barnes TR, Jones PB, Burke M, Lewis G. Cannabis use and risk of psychotic or affective mental health outcomes: a systematic review. Lancet. 2007 Jul 28;370(9584):
Zammit S, Allebeck P, Andreasson S, Lundberg I, Lewis G. Self reported cannabis use as a risk factor for schizophrenia in Swedish conscripts of 1969: historical cohort study. BMJ. 2002 Nov 23;325 (7374):1199. Abstract
View all comments by Dana March
View all comments by Ezra Susser
Related News: Meta-analysis Supports Case for Cannabis in Etiology of Psychosis
Comment by: Amresh Shrivastava
Submitted 20 October 2007
Posted 24 October 2007
Current interest in cannabis and the onset of psychosis is laudable. The Lancet paper no doubt establishes a causal link based upon what has been known in the literature (Raphael et al., 2005; Roberts et al., 2007; Rey et al., 2004; Wittchen et al., 2007). The authors need to be congratulated for taking extreme care to incorporate most of the studies and also for making conclusions with a sense of skepticism. That is where further questions arise.
1. Cannabis is used only in certain cultures and known to be involved in a maximum 50 percent of cases of psychosis, schizophrenia, and schizophreniform psychosis (Gregg et al., 2007). In that sense, are there two different phenotypes of schizophrenia, a) where exposure to cannabis is necessarily a factor and b) where a different set of potentiating or precipitating factors work, not cannabis?
2. Even if we focus only on the first possibility, there are few unanswered questions such as, what are the concurrent clinical conditions along with cannabis abuse? Do these patients have cognitive dysfunction? Is that reflective of broader brain mechanism changes?
3. There seems to be no reliable biological explanation as to why exposure to cannabis should precipitate psychosis. Cannabis is one of the most commonly used illicit drugs. Its active compound “cannabidols” has 64 active isomers, each having differing effects on health and behavior. There is strong support for a link between cannabis and development—exacerbation of psychosis as well as other mental health conditions (e.g., anxiety, depression). Further research is needed to determine the underlying neurochemical processes and their possible contributions to etiology, as well as the social factors that contribute to the increasing use of cannabis by young people.
4. There is a theory that preexisting cognitive dysfunction is a core feature of schizophrenia. Accepting this, there are no studies to show “causal relationship” between cannabis and cognitive dysfunction.
The current levels of information and understanding, though collected over last 25-30 years of research, are far from adequate to establish any direct relationship except “mere association.” It is hoped that more precise biological, imaging, and neuropsychological studies would be able to throw fresh light on this important area of research.
Acute cannabis administration can induce memory impairments, sometimes persisting months following abstinence. There is no evidence that residual effects on cognition remain after years of abstinence. The scarce literature on neuroimaging, mainly done in non-psychotic populations, shows little evidence that cannabis has effects on brain anatomy. Acute effects of cannabis include increases of cerebral blood flow, whereas long-term effects of cannabis include attenuation of cerebral blood flow. In animals Δ9-tetrahydrocannabinol enhances dopaminergic neurotransmission in brain regions known to be implicated in psychosis. Studies in humans show that genetic vulnerability may add to increased risk of developing psychosis and cognitive impairments following cannabis consumption. Δ9-tetrahydrocannabinol induces psychotic-like states and memory impairments in healthy volunteers (Linszen et al., 2007).
On the basis of six studies, it is concluded that there was insufficient evidence to prove conclusively that long-term cannabis use causes or does not cause residual abnormalities. The results of several reviews were also inconclusive as to whether cannabis use during adolescence may have a lasting effect on cognitive functioning and brain structure. However, it could not rule out that a) certain cognitive and cerebral abnormalities existed in patients before cannabis use began and b) that patients were suffering from subacute effects of cannabis (Weeda et al., 2006).
Continued cannabis use by persons with schizophrenia predicts a small increase in psychotic symptom severity but not vice versa (Degenhardt et al., 2007). Currently, there is a lot of interest in cannabis use as a risk factor for the development of schizophrenia. Cognitive dysfunction associated with long-term or heavy cannabis use is similar in many respects to the cognitive endophenotypes that have been proposed as vulnerability markers of schizophrenia. In this situation, we need to examine the similarities between these in the context of the neurobiology underlying cognitive dysfunction, particularly implicating the endogenous cannabinoid system, which plays a significant role in attention, learning, and memory, and in general, inhibitory regulatory mechanisms in the brain. Closer examination of the cognitive deficits associated with specific parameters of cannabis use and interactions with neurodevelopmental stages and neural substrates will better inform our understanding of the nature of the association between cannabis use and psychosis. The theoretical and clinical significance of further research in this field is enhancing our understanding of underlying pathophysiology and improving the provision of treatments for substance use and mental illness (Solowij et al., 2007). Many studies now show a robust and consistent association between cannabis consumption and the ulterior development of psychosis. Furthermore, our better understanding of cannabis biology allows the proposal of a plausible hypothetical model, based notably on possible interactions between cannabis and dopaminergic neurotransmission (Jockers-Scherubl, 2006). Do they suffer from other disorders, which are underlying or may be causal or comorbid, and do these comorbid conditions also have neurocognitive changes, e.g., psychosis, ADHD, LD, Tourette disorder, and other movement disorders, or depression? Is there an interrelationship among these factors to cause abuse and degree of cannabis consumption?
Raphael B, Wooding S, Stevens G, Connor J. Comorbidity: cannabis and complexity. J Psychiatr Pract. 2005 May; 11(3): 161-7.
Roberts RE, Roberts CR, Xing Y. Comorbidity of substance use disorders and other psychiatric disorders among adolescents: Evidence from an epidemiologic survey. Drug Alcohol Depend. 2007 Apr;88 Suppl 1:S4-13. Epub 2007 Feb 1.
Rey JM, Martin A, Krabman P. Is the party over? Cannabis and juvenile psychiatric disorder: the past 10 years. J Am Acad Child Adolesc Psychiatry. 2004 Oct; 43(10): 1194-205. Abstract
Wittchen HU, Frohlich C, Behrendt S. Cannabis use and cannabis use disorders and their relationship to mental disorders: A 10-year prospective-longitudinal community study in adolescents. Drug Alcohol Depend. 2007 Apr;88 Suppl 1:S60-70. Epub 2007 Jan 25.
Gregg L, Barrowclough C, Haddock G. Reasons for increased substance use in psychosis. Clin Psychol Rev. 2007 May;27(4):494-510. Epub 2007 Jan 19. Abstract
Linszen D, van Amelsvoort T. Cannabis and psychosis: an update on course and biological plausible mechanisms. Curr Opin Psychiatry. 2007 Mar; 20(2): 116-20. Abstract
Weeda MR, Peters BD, De Haan L, Linszen DH. Residual neuropsychological, structural and functional brain abnormalities after long-term cannabis use] Tijdschr Psychiatr. 2006; 48(3): 185-93. Abstract
Degenhardt L, Tennant C, Gilmour S, Schofield D, Nash L, Hall W,McKay D. The temporal dynamics of relationships between cannabis, psychosis and depression among young adults with psychotic disorders: findings from a 10-month prospective study. Psychol Med. 2007 Feb 9; 1-8.
Solowij N, Michie PT. Cannabis and cognitive dysfunction: parallels with endophenotypes of schizophrenia? J Psychiatry Neurosci. 2007 Jan; 32(1): 30-52. Abstract
Jockers-Scherubl MC. [Schizophrenia and cannabis consumption: epidemiology and clinical symptoms] Prax Kinderpsychol Kinderpsychiatr. 2006; 55(7): 533-43. Abstract
Curtis L, Rey-Bellet P, Merlo MC. [Cannabis and psychosis] Rev Med Suisse. 2006 Sep 20; 2(79): 2099-100, 2102-3.
Costentin J. [Neurobiology of cannabis--recent data enlightening driving disturbances] Ann Pharm Fr. 2006 May; 64(3): 148-59. Abstract
View all comments by Amresh Shrivastava
Related News: Children With Early Psychotic Symptoms Lag in Cognition
Comment by: Philip Harvey
Submitted 25 February 2014
Posted 25 February 2014
The Gurs have done it again. First, they developed the first truly remotely deliverable cognitive assessment with any validity data that would stand the test of peer review. Now they have passed another hurdle: general population screening. For years, those of us interested in this topic have said: "Sure. Those kids look different, but can you go into the community, screen the whole group, and find the outliers who may be impaired?" So these studies are really important because they constitute a true, largely epidemiologically interesting sample of the population. Then the researchers relate psychosis and cognition and find a link. As Ruben said, it may happen that someday we can have high-throughput cognitive (or functional capacity, to give our work a plug) testing that can be administered at a routine clinical visit to a doctor. Any findings could lead to a referral. This could lead to targeted early interventions. The only weak link is the family who never goes to the doctor at all. Are they the ones really at risk? This study does not have to answer that question. The results stand on their own.
View all comments by Philip Harvey
Related News: Children With Early Psychotic Symptoms Lag in Cognition
Comment by: Michael F. Green
Submitted 26 February 2014
Posted 26 February 2014
The very impressive article by Gur et al. on neurocognitive growth charting was already summarized by Michele Solis, so I will just make a couple of additional comments.
As mentioned in the summary, one of the remarkable contributions of this paper is to extend the link between cognitive impairment and psychosis down to the age of eight. This is uncharted territory, and, despite some variations, the lags in complex cognition and social cognition are relatively consistent from age eight to 20. Some of us would have expected a gradually increasing gap with increasing age, but that did not happen. That is encouraging in itself.
The authors were forward thinking and included a social cognitive domain in the battery. Sometimes people assume that social cognitive tests must be more inherently complex than mundane, non-social cognitive tests (after all, they are social), but that is not always the case. Indeed, the social cognitive tests in this battery can be considered low-level tests that place minimal demands on social inference. That means they are tapping only a subset of social cognitive processes, but they are useful here because they are fully appropriate for children.
What does a 1.5-year delay in social cognition mean for the kids? At these ages—it probably means a lot. The kids with psychosis spectrum are over a year behind in identifying the emotional expressions of their classmates, siblings, and teachers. It is not hard to imagine the social and functional consequences that might accompany the delay. Fortunately, such skills can be trained. Based on this study, we may need to develop training programs for younger participants.
View all comments by Michael F. Green